Split (1)

train · ~5.33M rows (showing the first 168k)

id string	text string	source string	added string	created string	metadata dict
PMC1180428	Background ========== Ukrain (NSC-631570) is a semi-synthetic compound derived from the common weed, greater celandine (Chelidonium majusL.). This plant contains a range of alkaloids, most notably chelidonine, also known as benzophenanthridine alkaloid. A leaflet distributed to patients at the Bristol Cancer Help Centre, United Kingdom, describes Ukrain as \" the only known product, which at present does not also destroy healthy cells, and which reduces tumors and boosts the immune system\...\" \[[@B1]\]. Ukrain is most commonly administered intravenously and consists of one molecule thiophosphoric acid conjugated to three molecules of chelidonine. It has drug licenses in several states of the former Soviet Union. Research on Ukrain started about 20 years ago. Meanwhile, numerous in-vitro studies \[[@B2]-[@B37]\] animal experiments \[[@B38]-[@B83]\], case reports \[[@B84]-[@B97]\], and case series \[[@B98]-[@B108]\] have emerged. Collectively, these data suggest that Ukrain has anticancer activity in a wide range of cell lines, which could be of clinical value. Whether or not this translates into clinical effectiveness and whether or not Ukrain does indeed cure some type of cancer or improves their prognosis can best be decided on the basis of randomised clinical trials (RCTs). This systematic review is aimed at summarising and critically evaluating all such studies. Methods ======= Electronic literature searches were conducted in the following databases: MEDLINE (1966 to date, via Pubmed), EMBASE (1974 to date), CINAHL (Cumulative Index to Nursing and Allied Health Literature, 1982 to date), AMED (Allied and Complementary Medicine Database, 1985 to date), PsycINFO (1987 to date), DIMDI (Deutsches Institut für Medizinische Dokumentation und Information) and The Cochrane Central Register of Controlled Trials (CENTRAL). The following search terms were used: \'Ukrain\', \'chelidonium\', \'greater celandine\', \'cancer\', \'neoplasm\' or \'tumour\'. Further handsearches were performed in our unit\'s own files as well as in the reference lists of all located articles. The producer of Ukrain was also contacted. No restrictions regarding the language of publication were imposed. We included all RCTs of Ukrain as a treatment for any type of human cancer. Ukrain could be used as a sole treatment or as an adjunct to conventional therapy. Any type of intervention was permitted in the control groups. The clinical endpoints had to be survival or parameters indicative of tumour burden. Non-randomised studies or RCTs that did not quantify clinical endpoints were excluded \[e.g. \[[@B109]-[@B117]\]\], as were duplicates \[[@B118]\]. All articles were read in full by both authors and data relating to design, diagnosis, number of subjects, treatments for experimental and control groups, outcome measures and results were extracted independently by both authors. The methodological quality of each trial was assessed using the Jadad score, unless the study was only available in abstract form \[[@B119]\]. It evaluates methodological quality using three items assessing random allocation, double-blinding and the reports of withdrawals and drop-outs and a maximum of 5 points can be given if all criteria are met. The authors agreed to a consensus on the assessed data and cases of discrepancy would be settled by discussion. Because of overt clinical heterogeneity, a meta-analysis was deemed unreasonable. Descriptive summaries of the data are presented in the following text. Results ======= Our search strategy identified 7 RCTs \[[@B120]-[@B126]\]. The majority of these studies was published in two different journals between 1995 and 2002 by 4 different groups of authors from the Belarus and Germany. Key data from these studies are summarised in Table 1 and will be discussed below. Susak et al published an RCT in which 108 colorectal cancer patients received either Ukrain as a monotherapy or 5-fluororacil for an unspecified time duration \[[@B126]\]. The results suggest that this was followed by non-progression of the malignancy in 88.8% of the patients in the experimental group compared to 27.7% in the control group. This study is only reported in abstract form. Numerous methodological details are therefore not accessible and its methodological quality cannot be reliably assessed. One year later, the same research group published a similar clinical trial, this time including 96 colorectal cancer patients \[[@B120]\]. Forty-eight patients received Ukrain as a monotherapy and 48 patients received 5-fluorouracil and radiation. The survival rate differed substantially between the two groups. Two-year survival was 78.6% in the experimental group compared to 33.3% in the control group. This study was not blinded but applied an appropriate method of randomisation. Bondar et al treated 48 histologically verified rectal cancer patients either with X-ray radiotherapy, chemotherapy and surgery (control group) or with Ukrain and surgery (experimental group) \[[@B121]\]. Before and after these treatments, the authors measured 19 different laboratory parameters including two tumour markers. In addition, the Karnofsky Index, tumour dimensions, and recurrences were monitored. All of these variables strongly favoured Ukrain therapy over conventional treatment. This study has, however, numerous limitations. For instance, the method of randomisation was not explained; the authors merely stated that \"all patients were subdivided into two randomised groups\". Moreover, \"tumour dimensions\" were mentioned as an outcome measure but neither the methodology of measurement nor the results were provided. The recurrence rates are expressed as percentage figures and no test statistics seem to have been applied. Uglyanitsa et al conducted a study with 28 patients suffering from bladder cancer \[[@B116]\] aiming \"to evaluate the efficacy of Ukrain\". Patients were allocated to three groups treated with a total dose of 100 (group 1), 200 (group 2), or 300 mg Ukrain (group 3). Two weeks later tumour regression was verified through cytoscopy and ultrasound. Complete and partial regression was noted in 0/4 patients of group 1, 1/4 patients of group 2, and 2/6 patients of group 3. This study lacks many characteristics of a rigorous trial; its stated aims (to evaluate efficacy) cannot be achieved with the study design, which essentially was that of an equivalence or dose-finding study. Zemskov and colleagues published a \"pilot study\" with 42 patients suffering from pancreas cancer who had refused chemotherapy \[[@B122]\]. They were randomised to receive either Vitamin C alone or with Ukrain (total dose 100 mg/patient). The primary endpoint (survival) strongly favoured the Ukrain group. The analysis seems to include 4 protocol violations (the description is unclear). Even though the randomisation procedure is mentioned (\'closed envelopes\') it seems unusual that precisely 21 patients ended up in both groups. The results are surprisingly good -- much better than with any other treatment for that condition. Uglyanitsa et al randomised (\"by lottery\") 75 breast cancer patients into three groups of 25 patients each \[[@B123]\]. They received either no specific treatment, a total dose of 50 or 100 mg Ukrain 5--7 days before mastectomy. The authors note that Ukrain rendered the primary tumour and the affected regional lymph nodes larger, harder and \"more clearly defined\". They interpret this as Ukrain-induced tumour sclerosis. According to the investigators\' judgement, these changes facilitated surgery and the operative success. In addition, Ukrain was associated with remarkable symptomatic improvements, e.g. better appetite, more sleep, less weakness. The report is unclear in several respects. For instance, no details about statistical analyses are provided, the outcome measures seem subjective, no information regarding investigator blinding is given, and the randomisation procedure seems suspect. Zemskov and colleagues randomised 42 patients with pancreatic cancer who had refused conventional therapy \[[@B124]\]. They received either Ukrain (total dose 100 mg/patient) with Vitamin C or Vitamin C alone. The results confirmed this group\'s earlier findings \[[@B122]\]. Survival was remarkable in the Ukrain treated patients and symptoms responded well to this treatment. There are, however, numerous puzzling details. Why do the authors call their second study a \"pilot study\"? Why did their ethics committee consent to this \"placebo\"-controlled trial in the knowledge of the surprisingly positive earlier results? How could a proper randomisation again result in two equally sized groups of 21? In the discussion, the authors describe their earlier results as though this trial was conducted against 5-FU which, in fact, is not the case \[[@B122]\]. Gansauge et al reported a study of 90 patients with pancreatic cancer treated either with 1000 mg gemcitabine/m^2^or 100 mg Ukrain or the combination of both regimens \[[@B125]\]. Survival rates suggested that Ukrain was superior to gemcitabine alone. A direct comparison of the 12 month survival rates revealed large differences compared to the data from Zemskov et al \[[@B124]\] (29% vs 76% in the Ukrain-treated groups). The randomisation procedure was not explained and, again, the equal group sizes are remarkable. Conclusion ========== Collectively, these RCTs seem to suggest that Ukrain is an effective therapy for a range of cancers. In conjunction with the numerous encouraging case reports \[[@B84]-[@B97]\] case series \[[@B98]-[@B108]\], and non-randomised clinical trials \[[@B109]-[@B121]\] these data look impressive at first glance. Yet several important caveats need to be considered. None of the RCTs in this systematic review is without serious methodological limitations. The Jadad score \[[@B119]\] of most RCTs was low. Their sample size was usually small, and a sample size calculation to define the number of patients required was lacking in most cases. Even though most RCTs were non-inferiority studies by design and purpose, their statistical approach was that of a superiority trial. The majority of RCTs were conducted in Ukrainian research institutes and published in only two different journals. In several trials, there are clear signs of involvement of the manufacturer of Ukrain. Most RCTs have generally been poorly evaluated and reported, which possibly reflects the poverty of clinical science in Eastern Europe. Independent replications are not available. The only German study \[[@B125]\] has also been heavily criticised: its sample size (30 patients in each group) is minute, the report lacks statistical detail and there is an inequality of treatment cycles between groups \[[@B127]\]. It was also noted that this study (the only RCT not published in the same two journals as all the other RCTs) was published in a journal for which the senior author served as editor \[[@B127]\]. No RCTs were found showing negative or near neutral results; this might suggest the existence of publication bias for which we did, however, find no definite proof. Greater celandine (Chelidonium majusL), which forms the basis of Ukrain, was traditionally used for liver and gallbladder complaints, loss of appetite and gastroenteritis. None of these indications is supported by trial evidence. The main alkaloid from this plant, chelidonine, has antispasmodic, weak central analgesic and papaverine-like effects. In animal experiments, an alcoholic extract of greater celandine increased bile flow, caused non-specific immune stimulation and acted as a hepatoprotectant \[[@B128]\]. The oral administration of greater celandine in humans has been associated with several cases of toxic hepatitis \[[@B129]\]. The mechanism of action of Ukrain as an anticancer drug (if any) remains elusive. Collectively, the preclinical studies are suggestive of antineoplastic and immunomodulatory effects. It has been postulated that the antineoplastic effect is due to the alkaloids interfering with the metabolism of cancer cells, diminished synthesis of DNA, RNA and proteins, the inhibition of cellular oxygen consumption, and the induction of programmed cell death in malignant cells \[[@B130]\]. Several reports of adverse reactions after greater celandine have been published. Most notably, toxic hepatitis has been associated with its oral use \[[@B129],[@B131],[@B132]\]. No case reports of adverse events have emerged of intravenous Ukrain therapy for cancer. The clinical trial data suggest that Ukrain might cause the following adverse effects: an increase in patients\' body temperature (n = 26) \[[@B120],[@B123],[@B125]\], general burning sensations (n = 3) \[[@B123]\] and bleeding (n = 4) \[[@B125]\]. Levels between 0--2 according to World Health Organisation toxicity criteria were noted in two trials \[[@B122],[@B124]\] and toxicity criteria between 0--3 were observed in one trial \[[@B125]\]. The costs of Ukrain therapy are high; one course costs € 700 for the medication alone, and the total treatment costs have been estimated at € 3000 per week \[[@B133]\]. In conclusion, Ukrain is a plant-based anticancer drug that is supported by clinical and pre-clinical evidence in a range of malignancies. The data are, however, not free from problems. Before positive recommendations can be issued, independent replications with definite trials and larger sample sizes seem mandatory. Competing interests =================== The author(s) have no competing interests to declare. Authors\' contributions ======================= EE conceived of the review, participated in its design and coordination, the data extraction and helped to draft the manuscript. KS carried out the data extraction and helped drafting the manuscript. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2407/5/69/prepub>	PubMed Central	2024-06-05T03:55:59.426268	2005-7-1	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180428/", "journal": "BMC Cancer. 2005 Jul 1; 5:69", "authors": [ { "first": "E", "last": "Ernst" }, { "first": "K", "last": "Schmidt" } ] }
PMC1180429	Background ========== Traditional methods for studying signal transduction cascades have been based solely on biochemical analysis of whole cell populations and homogenized tissues (e.g. radioassays, western blots, etc.). In addition, in vitrostudies have required the use of radioactive isotopes for biochemical characterization of kinases. These methods are time consuming, produce large quantities of radioactive waste, and do not allow for the study of real-time kinase dynamics. Recently, sensors for studying signal molecules in vitro, as well as cascade dynamics in single cells, have been developed utilizing fluorescent proteins and the phenomenon of fluorescence resonance energy transfer (FRET). FRET is a phenomenon by which energy is transferred from one fluorescent molecule to another by way of dipole-dipole interactions during excitation of the donor molecule. FRET efficiency is given by the equation: ![](1472-6769-5-1-i1.gif) where R is the donor-acceptor radius and R~o~is the radius at which FRET efficiency is 50% (Förster radius). Small changes in R (1--2Å) and small orientation changes between the donor and acceptor fluorophores can dramatically affect the efficiency of FRET, making very small changes in structure easily detectable. The limitation, however, is that FRET is effective only between 10--100Å, and therefore the donor and acceptor must be maintained in close proximity. Genetically encoded fluorescent proteins fused by linker peptide sequences have been utilized to address proximity limitations, as well as facilitate the delivery of the sensor into live cells. Some of the earliest work done with these sensors includes genetically encoded sensors of caspase activity, in which caspase-3 and caspase-8 sensitive linker peptides were fused between the different variants of Enhanced Green Fluorescent Protein (EGFP) \[[@B1]\]. These studies allowed the characterization of apoptosis in single living cells with spatio-temporal resolution. Other types of FRET signaling sensors have since evolved, including sensors for calcium signaling \[[@B2]\], receptor tyrosine kinases \[[@B3],[@B4]\], intracellular kinases \[[@B4]-[@B7]\], and histone methylation \[[@B8]\]. These sensors have been shown to be useful in in vitroassays, as well as allow the study of signaling events in a single cell, in real time \[[@B4]\]. The Ras/ERK cascade controls differentiation and proliferation in many different cell types and organisms. In this signal transduction pathway, activated Ras (Ras-GTP) binds directly to Raf-1 and recruits it to the membrane where Raf becomes activated. Raf then phosphorylates and activates Mek-1 and Mek-2 (the MAPK kinase), which in turn phosphorylate the MAPKs ERK-1 and ERK-2. Activated ERKs translocate to the nucleus and directly phosphorylate transcriptional regulatory proteins (including members of the Ets family of transcription factors, and the bZIP factors Fos and Jun) (reviewed in \[[@B9]\]). Kinetics and in vivodynamics of ERK MAP kinase activity are not completely understood. Traditionally, the level of ERK activation has been thought to be relative to the strength of the upstream signal. Some data suggest, however, that ERK activation is \"switch-like,\" requiring a threshold of activation, above which all ERK molecules within the cell become activated \[[@B10]\]. Computational models of MAPK signaling mechanisms can fit both possibilities \[[@B11],[@B12]\]. Therefore, the development of new tools to study these dynamics is essential to determine the biochemical nature of ERK signaling. In this manuscript, we describe the development of several FRET-based sensors of MAP kinase activity which we have called ERK Activity Sensors (EAS). Peptide linker sequences taken from the Ets family transcription family members Ets1 and Elk1, both ERK targets, were fused between cyan and yellow variants of EGFP (ECFP and EYFP, respectively). Several of our constructs respond to phosphorylation by activated ERK with changes in FRET, and at least one of them, EAS-3, is specific for active Extracellular Regulated Kinase (ERK) as opposed to two other MAP kinases, p38 and SAPK/JNK. Therefore, EAS-3 is a viable, non-radioactive sensor of ERK MAPK activity in vitro. Results and discussion ====================== Design and model of EAS ----------------------- EAS was designed to have a short peptide linker sequence sensitive to phosphorylation by ERK fused between ECFP and EYFP. ECFP and EYFP were chosen as a FRET pair because of the good spectral overlap between the emission of ECFP and the excitation of EYFP. We hypothesized that phosphorylation would result in changes in secondary structure of the linker peptide that would alter FRET efficiency for the protein (Figure [1A](#F1){ref-type="fig"}). Since it was unclear which factors would contribute to forming a suitable FRET sensor, we prepared several different constructs with peptide linker sequences containing ERK phosphorylation sites taken from Ets1 and Elk1 transcription factors, which are physiological substrates of the ERK MAPKs (Figure [1B](#F1){ref-type="fig"}). EAS-2, derived from mouse Ets1, has a rather large linker peptide, due to the discovery that the MAP Kinase binding site (D-domain) is approximately 90 amino acids downstream of the target serine \[[@B13]\]. The linker sequences for EAS 3--5 were taken from human Elk1, which has been shown to have the DEF domain (FXFP amino-acid motif), a known MAPK binding site, near the C-terminus. These peptides also contain two Serine-Proline (SP) motifs (consensus MAPK phosphorylation sites) corresponding to serines 383 and 389 of Elk1, which have been shown to be targets of MAP kinases. Phosphorylation of these serines has also been shown to be directed specifically by the DEF domain \[[@B14]\]. EAS-Neg is a negative control construct with a short peptide linker derived from Ets1, containing a consensus MAPK phosphorylation site, but no MAPK binding site. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### A model for EAS activation and construct design. (A) Our inferred model indicates that upon pERK2 phosphorylation a conformational change in the linker peptide of EAS decreases the efficiency of FRET. The red area in the linker peptide indicates the relative position of either D-domain (EAS-2) or DEF domain (EAS-3, -4, -5) and the SP motifs denote the consensus phosphorylation sites. (B) The gene constructs include EAS-Neg and EAS-2, which are derived from mouse Ets1. EAS 3--5 contain peptide linkers derived from human Elk1. The composition of each linker peptide is indicated by the primary sequence positions derived from Ets1 and Elk1, respectively. ::: ![](1472-6769-5-1-1) ::: EAS FRET changes in the presence of active ERK2 ----------------------------------------------- We expected a change in energy transfer between ECFP and EYFP upon incubation with active ERK2 (pERK2) for EAS to be effective as a real time sensor. This would indicate that a phosphorylation event had affected the distance and orientation between the two fluorophores. Each EAS construct was incubated with pERK2 in a fluorimeter (30°C) and readings were taken prior to addition of ATP. Upon the addition of ATP, all EAS constructs, with the exception of EAS-Neg, showed a decrease of energy transfer indicated by an increase in donor (ECFP) emission (475 nm) with a concomitant decrease in acceptor (EYFP) emission (525 nm) after 40 minutes (Figure [2A](#F2){ref-type="fig"}). Therefore, a decrease in FRET efficiency correlates with phosphorylation of EAS by pERK2. The resultant change in ECFP/EYFP emission ratio is comparable to or exceeds that of similar FRET-based signal transduction sensors \[[@B4]-[@B6]\]. Under the same conditions, EAS-Neg showed no change in FRET over the course of 30 minutes, confirming the specificity of ERK2 for EAS 2--5 and the requirement of a phosphorylation event to induce a change in FRET. Figure [2B](#F2){ref-type="fig"} shows the absolute change in ECFP/EYFP ratio over a 40 minute time course for each construct. EAS-3 and EAS-4 show the greatest absolute decrease in EYFP/ECFP emission ratio, with EAS-5 having an intermediate change in emission ratio. EAS-2 showed very little change in FRET. Phosphorylation was also confirmed by an in vitrokinase ^32^P phosphorylation assay using recombinant pERK2 (Figure [2C](#F2){ref-type="fig"}). EAS 2--5 sensors were efficiently phosphorylated as compared to the non-specific control, EAS-Neg, and Myelin Basic Protein (MBP), a common target of activated MAPKs for in vitrostudies. Based on the robust FRET signal change and radioassay results, further studies focused on EAS-3. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### EAS proteins are targets for ERK2 and exhibit decreased FRETefficiency upon phosphorylation. (A) Emission spectra indicate the effective FRET change for each EAS protein sensor. Time course fluorimetry with EAS-Neg shows that pERK2 induces no FRET change 30 minutes after ATP addition. All other EAS proteins show varying gains in ECFP emission (475 nm) and losses in EYFP emission (525 nm) 40 minutes after ATP addition. (B) Absolute change in the ratio of EYFP/ECFP emission for EAS-Neg is zero over a 30 minute time course, whereas a decrease in ratio for EAS-3, EAS-4, and EAS-5 is readily detectable at 2 minutes, and continues to decay exponentially over the time course. EAS-2 has a detectable change in emission ratio, but the change is minimal as compared to other EAS proteins. Error analysis was determined from three independent experiments. (C) EAS proteins (EAS-2, -3, -4, and -5) are phosphorylated by pERK2 in the presence of γ-\[^32^P\]ATP. This confirms that EAS proteins are targets of pERK2 as compared with MBP, a known ERK2 substrate. As expected, EAS-Neg is not phosphorylated by pERK2. ::: ![](1472-6769-5-1-2) ::: EAS changes in FRET in response to pERK2 require phosphorylation and the ERK binding site ----------------------------------------------------------------------------------------- Mutants of EAS-3 were utilized to validate the structural features of EAS essential for decreased FRET efficiency upon incubation with pERK2. We refined EAS-3 by replacing the bulky N-terminal GST-purification tag (Glutathione-S-Transferase) with a Histidine-10 tag on the C-terminus. This reduced the possibility that the large purification tag would interfere with FRET efficiency changes. In addition, we mutated alanine 207 and alanine 487 to lysine. As previously reported, this prevents EGFP dimerization \[[@B15]\]. This latter modification reduced co-purification of truncation products with full-length EAS (data not shown). To further characterize EAS-3, we generated different constructs targeting critical residues, and analyzed their ability to serve as substrates for pERK2. Mutants EAS3-S286A, EAS3-S292A, and the double mutant EAS3-S(286,292)A eliminated consensus phosphorylation sites (Figure [2A](#F2){ref-type="fig"}). The dead binding domain mutant (EAS3-DBD) eliminated the DEF domain by replacing the two key phenylalanines with alanines, creating an AXAP motif (Figure [3A](#F3){ref-type="fig"}). As Figure [3B](#F3){ref-type="fig"} demonstrates, all five mutants have decreased phosphorylation compared to the wild type sensor. As expected, the EAS3-S(286,292)A mutant had the lowest level of phosphate incorporation due to absence of both serine-proline consensus phosphorylation sites required by MAP Kinases. A decrease in phosphorylation of EAS-DBD iss also consistent with the requirement of MAPKs to bind targets for efficient phosphorylation (reviewed in \[[@B16]\]). Differences between wild-type EAS-3 and mutants are also reflected in change of FRET efficiency when treated with pERK2 in fluorimeter experiments (Figure [3C](#F3){ref-type="fig"}). These relative changes in FRET efficiency are consistent with the radioassay data. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Mutation of key residues diminishes EAS-3 phosphorylation by pERK2. (A) Primary sequences of EAS-3 and EAS-3 mutant linker peptides are shown with consensus phosphorylation sites (highlighted in green) and ERK binding sites (underlined). Residue mutations to alanine are highlighted in red. (B) Analysis of mutant EAS-3 proteins by γ-\[^32^P\]ATP phosphorylation shows complete loss of pERK2 phosphorylation of the S(286,292)A mutant as compared to wild-type EAS-3. Decreased ^32^P-phosphorylation of site-specific mutants also indicates the involvement of both Ser286 and Ser292 for pERK2 activity. Requirement of the ERK binding domain for efficient phosphorylation is evident from decreased phosphorylation of the EAS3-DBD mutant. Band intensities were quantitated by densitometry and are shown graphically. Equal protein loading is shown by Coomassie staining of EAS-3 and EAS-3 mutants and phosphorylation assay was terminated after 15 minutes at 30°C. (C) Fluorimetry data for EAS-3 and EAS-3 mutants reveals that the FRET efficiency change is reduced with various mutations, consistent with the phosphorylation assay in B. ::: ![](1472-6769-5-1-3) ::: EAS-3 is not phosphorylated by pSAPK or pp38 -------------------------------------------- Distinguishing the activation of different MAP kinases within the cell is essential since each MAPK pathway is activated by multiple mitogens and external environmental factors to varying degrees (reviewed in \[[@B17]\]). Furthermore, there is extensive cross-talk between the different MAP kinase pathways. Detection methods must effectively isolate the signal of the target kinase from other family members to elucidate the contributions of these different pathways to a given cellular process. The target peptide linkers were designed to impart specificity for ERK. Elk1 and Ets1 were chosen because of their seemingly specific interaction with ERK relative to p38 or SAPK/JNK (reviewed in \[[@B18]\]). To determine whether EAS-3 acted as a specific substrate for ERK we performed in vitrophosphorylation assays to quantify the ability of pERK2, pp38, and pSAPK/JNK to phosphorylate EAS-3 (Figure [4](#F4){ref-type="fig"}). As shown in Figure [4A](#F4){ref-type="fig"}, EAS-3 is an approximately 2000-fold more specific target for pERK2 than pp38, and 50-fold more specific for pERK2 than pSAPK/JNK. Activities of pERK2, pp38 and pSAPK/JNK were confirmed by using known substrates specific for each kinase (Figure [4B](#F4){ref-type="fig"}). MBP was used in conjuction with pERK2, GST-ATF2 with pp38, and GST-c-Jun with pSAPK/JNK. Specificity of these kinases for EAS-3 was also confirmed by FRET in a fluorimeter (Figure [4C](#F4){ref-type="fig"}). It is clear that the change in FRET was dramatic when EAS-3 was incubated with pERK2, but showed little or no response when incubated with pp38 or pSAPK/JNK. This is consistent with the results found in 4A. Given the similarity of these kinases, this also suggests that EAS-3 would be specifically acted upon by pERK2 even in the presence of other activated intracellular kinases. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Determination of ERK2 specificity for EAS-3. (A) Phosphorylation of EAS-3 is efficient with pERK2, but not pp38 or pSAPK. pp38 and pSAPK incorporates little or no ^32^P into EAS-3 at the same relative specific activity as pERK2. Band intensities were quantitated by densitometry and are shown graphically. Equal protein loading of EAS-3 is shown by Coomassie staining. (B) Specific targets for pERK2 (MBP), pp38 (GST-ATF2), and pSAPK (GST-cJun) are efficiently phosphorylated in the presence of γ-\[^32^P\]ATP. The amounts of active kinase used in A were based on the relative incorporation of ^32^P into each cognate substrate by the respective kinase. (C) Fluorimetry of EAS-3 with MAPKs shows that the FRET efficiency change is significantly diminished in the presence of pp38 or pSAPK as compared to pERK2. This is consistent with the phosphorylation assay in A. ::: ![](1472-6769-5-1-4) ::: The demonstrated specificity of pERK2 for EAS-3 phosphorylation suggests that this sensor is a candidate for in vivostudies of ERK signaling. However, our preliminary experiments in NIH-3T3 cells indicated that EAS-3 was susceptible to intracellular phosphatase activity (data not shown). We surmise that this is due to the absence of a protective phospho-specific binding domain within the EAS-3 construct. Such binding domains have been crucial in the development of other FRET-based signaling sensors of kinase activity \[[@B2]-[@B4],[@B6],[@B7]\]. These binding domains, however, are taken from naturally occurring domains that are specific for each target sequence. Unfortunately, no known phospho-specific binding domain exists for Elk1 in the region of Ser383/389. Therefore, we are using a semi-rational approach to develop a phospho-specific binding domain that acts to protect the phosphorylated EAS-3 linker from intracellular phosphatases. This will enable us to adapt the EAS-3 sensor for use in live cells. Conclusion ========== These results show that our novel ERK Activity Sensors provide real-time in vitrodetection of MAP kinase activity. This method can be applied to studying kinetics of ERK activity in real time, as well as detection of ERK activity in unknown cell lysate fractions. There is also the potential to use these sensors for high-throughput screening of ERK kinase activity with fluorescence plate readers. This method is more direct and convenient to monitor ERK activation in vitrothan conventional assays that either use radioactivity for detection or rely on indirect detection using phospho-specific antibodies for MAPK targets. Methods ======= EAS constructs -------------- DNA coding for peptide linkers was amplified by PCR with primers designed with Bgl II and BamHI sites at the 5\' and 3\' ends, respectively. The products were cloned into pECFP-C1 (Clontech), and EYFP from pEYFP-N1 was subsequently cloned in frame to create EAS constructs. EAS\'s were cloned into pGEX-2T (Amersham) in frame with GST for expression and purification purposes. These GST-tagged constructs were used for initial fluorimetry experiment of EAS constructs. Other versions of EAS\'s were constructed using ECFP and EYFP with mutation A(207,487)K, which eliminates dimerization of GFPs, and cloned into pET21a (Novagen) with a Histidine-10 tag. These His10-tagged constructs were used for mutant and specificity experiments. EAS-3 linker mutants were made by Quickchange Site Directed Mutagenesis (Stratagene) to make either single or double point mutation within the peptide linker. For expression in NIH-3T3 cells, EAS\'s were cloned into the pECFP vector backbone without tags. EAS and active kinase expression -------------------------------- BL21(DE3) cells were transformed with EAS constructs and plated on LB agar supplemented with 100 μg/ml ampicillin. Colonies were picked into 4 ml LB-Amp and shaken overnight at 30°C. 250 ml LB was inoculated with starter culture, grown to A~600~of 0.6, and induced with 0.1 mM IPTG for 18 hours. Cells were spun down, lysed by french press, and cell fragments were spun down at 18,000 rpm in a Beckman JA-20 rotor for 30 min. Proteins were purified from lysates over Ni-NTA Superflow (Qiagen) or Glutathione Sepharose 4B (Amersham). Activated kinases were purified as described \[[@B19]\]. pERK2 was either purchased (NEB) or purified from bacteria transformed with a plasmid containing both constitutively active MEK1\* and His-tagged ERK2. Phosphorylation of ERK2 by MEK1\* occurred in bacteria prior to lysis. BL21(DE3) cells were electroporated with MEK1\/ERK2 construct and grown overnight at 37°C on LB-Amp agar plates. Several colonies were picked and incubated in 4 ml of TB supplemented with 100 μg/ml carbenicillin. The starter culture was added to 1L of TB, grown to A~600~of 0.35, and then induced with 0.25 mM IPTG for 12 hours at 30°C. Cells were harvested and lysed as above, and purified over Ni-NTA Superflow. Activated p38 and SAPK/JNK were purified from bacteria electroporated with two plasmids, one coding for constitutively active MEK kinase 4 (MEKK4\) and another coding for MEK4 and either His-tagged p38 or His-tagged SAPK/JNK. These constructs were grown and purified as above, except with both carbenicillin and kanamycin (50 μg/ml). Protein concentration and buffer exchange performed with Centriplus, Centricon, and Microcon ultrafiltration membranes (Millipore). Kinase assays and fluorimetry ----------------------------- Radioactive kinase assays were performed in 25 μl in 12.5 mM MOPS pH 7.5, 12.5 mM β-glycerophosphate pH 7.3, 7.5 mM MgCl~2~, 500 μM NaOrthovanadate, 500 μM NaF, and 9.7 nM DTT. Amounts of kinase and substrate added to reactions are indicated in figure legends. Finally, ATP supplemented with 0.2 nmol \[γ-^32^P\]ATP (6000 mCi/mmol, Molecular Bioproducts), was added to a final concentration of 1 mM. Reactions were incubated at 30°C for 15 minutes, 8.3 μl of 4× protein sample buffer (SB) were added, and samples were boiled for 5 minutes. Samples were analyzed with 10% or 12% SDS-PAGE (BioRad Mini-Protean II) and transferred to a nitrocellulose membrane. The membrane exposed to a phosphoscreen and scanned on a Storm 860 (Molecular Dynamics). Quantitative densitometry of bands was performed using ImageQuant 5.0 (Molecular Dynamics). Fluorimetry performed in a Shimadzu Spectrofluorophotometer RF-5301PC. Assays were performed in same buffer as above in a volume of 2.5 ml. The final concentration of EAS constructs was 250 nM and final concentration of pERK2 was 50 nM. The reaction was incubated in fluorimeter and warmed to 30°C. Excitation was set at 425 nm to excite ECFP and avoid excitation of EYFP and readings were taken in 0.2 nm increments. An initial reading was taken prior to ATP addition. To initiate reaction, ATP was added to a final concentration of 1 μM, a spinner was activated for 1 minute, and first reading taken at 2 minutes. EYFP to ECFP ratios were calculated by dividing EYFP peak emission by ECFP peak emission. Peak emissions were defined as an average intensity of EYFP between 524.6--525.4 nm and of ECFP between 474.6--475.4 nm. Authors\' contributions ======================= HMG performed all experimental work. JAI provided advice, funding and supervision for the work. Both authors have read and approved the final manuscript. Acknowledgements ================ We would like to thank Dr. Kornfeld for providing us with GST-Elk1 cDNA and Dr. Sharrocks for providing us with Ets1 cDNA, from which EAS linkers were derived. We also thank Dr. Richard Roberts for use of his fluorimetry equipment. This work was supported by the NIH (AI45072) and the Keck Foundation.	PubMed Central	2024-06-05T03:55:59.428160	2005-7-5	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180429/", "journal": "BMC Chem Biol. 2005 Jul 5; 5:1", "authors": [ { "first": "Harry M", "last": "Green" }, { "first": "José", "last": "Alberola-Ila" } ] }
PMC1180430	Background ========== Adenoid cystic carcinomas (cylindromas) of the trachea are very rare slow growing tumors and cause symptoms by virtue of their location and size. They are usually treated for many years as asthma unless some enthusiastic surgeon advises CT scan or performs endoscopy and surprisingly finds an intraluminal tracheal mass responsible for the symptoms. They usually become symptomatic when they start obstructing more than 75% of the tracheal lumen. The lower the mass is present, the more complicated is the anesthesia and the surgery for successful and safe removal of the mass. Here we are presenting a case of a 31 yrs. old female presenting with difficulty in speaking and breathing and found to have an intratracheal mass situated in thoracic trachea about 3 cms above the carina, occupying about 90% of the lumen. The case was successfully managed with femorofemoral cardiopulmonary bypass, sternotomy, tracheotomy and excision of mass. Case presentation ================= Case report ----------- A 31 yrs. old female was referred to us with the complaints of difficulty in speaking and progressively increasing difficulty in breathing since 3 yrs. The patient was being treated as asthmatic at some other hospital (with i.v. steroids, antibiotics, nebulizations and bed rest) since the onset of symptoms which relieved the symptoms but recurrence occurred after stopping the treatment. On examination, patient had a weak voice and was having stridor of grade II which increased with extension of neck or in fully supine position. She had mild tachycardia and tachypnoea. Indirect laryngoscopic examination was normal. She was maintaining adequate oxygen saturation on room air with conservative treatment. Fibreoptic tracheoscopy showed an intraluminal pedunculated tracheal mass which was occupying about 90% of the lumen (Figure [1](#F1){ref-type="fig"}). Biopsy from the mass was inconclusive. CT scan of neck and chest revealed a homogenous enhancing pedunculated intraluminal mass, about 2.4 × 1.5 × 1.1 cms in size attached to left lateral wall of the trachea. The peduncle was situated about 3 cms above the carina. About 90% of tracheal lumen was found to be occupied by the mass (Figure [2](#F2){ref-type="fig"}). Patient was taken for surgery and a femorofemoral cardiopulmonary bypass was established first for oxygenation of blood. Right femoral artery and vein were accessed and cannulated under local anesthesia and were connected to the cardiopulmonary bypass machine. General anesthesia was then induced and median sternotomy was done. Trachea was exposed up to the carina. There was no extratracheal extension of the mass. The trachea was opened in midline by about 3 cms vertical incision, starting at 2 cms above the carina and the mass was exposed. It was a soft pinkish pedunculated mass attached to left lateral wall of the thoracic trachea. Mass was removed along with its attachment and adjoining tracheal wall. The tracheal wall surrounding the attachment was normal looking. The resultant small defect in the tracheal wall was repaired with 000 vicryl and sealed with N-butyl cyanoacrylate. The patient was intubated with a cuffed endotracheal tube passing beyond the tracheal incision. Then the patient was taken on ventilator and inhalational anesthetics after weaning off the bypass. Tracheal incision was sutured & sealed with a thin layer of N-butyl cyanoacrylate. Sternotomy was closed in layers. Right inguinal incision was closed in layers after decannulating the femoral artery and vein and repairing the defects in the vessel walls. Patient was kept on ventilator in ICU and extubated after 24 hrs. The histopathology report of the excised mass revealed it to be adenoid cystic carcinoma also known as cylindroma. The patient was referred for radiotherapy. Repeat CT scan after one month (Figure [3](#F3){ref-type="fig"}) showed normal tracheal walls and lumen. Discussion ========== Primary tracheal masses are very rare and mostly malignant, occurring in 0.2 per 1, 00,000 persons per year \[[@B1]\] and among these squamous cell carcinomas form the main bulk. Next common primary tracheal malignancy is adenoid cystic carcinoma (cylindroma). Adenoid cystic carcinoma is a slow growing tumor with relatively good prognosis. It mainly involves the salivary glands. Its incidence in the lower airways is low (0.1% of all broncho-pulmonary neoplasms) \[[@B2]\]. The anesthesia in these cases is very challenging as the compromised airway is to be shared by both surgeon and anesthetist. For such cases, cardiopulmonary bypass standby after femoral artery and vein cannulation and then intravenous/inhalational induction while oxygenating the patient with the oxygen inhalation has been considered to be a reasonable approach \[[@B3]\]. In \>90% tracheal obstruction cases, IV induction is the preferred method after patient been taken on pump by which anesthesia can be maintained. We decided to go directly for femorofemoral cardiopulmonary bypass for induction and oxygenation as patient was not tolerating extension of the neck due to degree of tracheal obstruction, so even slight intraoperative neck or tracheal manipulation was anticipated to be hazardous. Byrne JG et al (2004) advocated planned use of CPB to facilitate complete resection of thoracic malignancies after careful patient selection \[[@B4]\]. The availability of CPB also provides a safety net in the event of injury to vascular structures during tumor resection. It has been advocated as safe and preferred method in difficult airway or near total airway obstruction situations \[[@B5]-[@B8]\]. Despite chances of excessive bleeding during surgery due to use of anticoagulants in CPB, it seems to be the safest way to provide anesthesia in such cases. Bleeding was not troublesome in our case as we used only partial heparinization (1 mg/kg body weight) instead of full heparinization (3 mg/kg body weight). Further to enhance the safety, heparin coated cannulae can also be used. The femoral vein cannula goes up to the junction of inferior vena cava and right atrium, while femoral artery cannula goes up to abdominal aorta. Care was taken to maintain pump compatibility with the beating heart to avoid fall in perfusion, especially in the peripheral organs. So bypass was used only as a supportive mean for oxygenation of the blood. Circulation of the blood was maintained by the beating heart and complete emptying of the heart was avoided carefully, as done in ECMO (Extra Corporeal Membrane Oxygenation). Tracheostomy or endotracheal intubation was not possible due to site and size of lesion and the amount of airway obstruction caused by it. In using high frequency jet ventilation with a small bore catheter, there was a risk of distal dislodgement of any broken piece of mass or increase in airway obstruction by movement of the mass itself, due to pedunculated nature of the mass, and we had nothing safe in our hands to manage that situation. Further in jet ventilation, there was a fair chance of compromising the oxygenation by collapse of the tracheal lumen distal to the mass during expiration or by development of pneumothorax. Endoscopic and laser excision were not opted due to problems of airway control and anesthesia during the procedure which may provide life threatening acute airway obstruction due to bleeding, secretions, edema or from the bronchoscope itself. These complications were more probable in a pedunculated mass due to possible movement of the mass during manipulation. Furthermore endoscopic laser excision is a better choice for palliation in advanced malignancies \[[@B3]\]. Resection and reconstruction followed by adjuvant radiotherapy provides best chance of cure in tracheal cylindromas \[[@B1],[@B8]\]. For a benign looking, pedunculated, slow growing mass without any obvious wall infiltration or surrounding structure involvement, it was decided to excise only the tumor attachment instead of going for complete resection of a segment of trachea and anastomosis. Donna E. Maziak et al (1996) found that there was only a small difference in survival between patients having complete and incomplete resection for tracheal cylindroma \[[@B8]\]. Conclusion ========== It can be concluded that it is always prudent to go for endoscopic or radiological evaluation of airways in patients with long standing or recurrent respiratory symptoms. Also even benign looking tracheal masses can come out to be malignant and should be sought with suspicion. Furthermore femorofemoral cardiopulmonary bypass is an effective, safe and convenient method for anesthesia for intraluminal tracheal masses in thoracic trachea with significant obstruction, allowing the surgeon enough freedom and comfort during surgery. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= All five authors 1\) Have made substantial contributions in management of this case and in conception, design, analysis and interpretation of results of this case report 2\) Have been involved in drafting the article or revising it critically for important intellectual content. 3\) Have given final approval to the version to be published Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6815/5/4/prepub> Acknowledgements ================ Written informed consent has been taken from patient and she has no objection in publication of this case report and its photographs. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Fibreoptic endoscopic image showing a pedunculated mass causing significant lumen obstruction. ::: ![](1472-6815-5-4-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Reconstructed sagittal CT image showing mass into the thoracic trachea. ::: ![](1472-6815-5-4-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### One month postoperative CT chest (axial view) showing normal tracheal walls and lumen. ::: ![](1472-6815-5-4-3) :::	PubMed Central	2024-06-05T03:55:59.429957	2005-6-7	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180430/", "journal": "BMC Ear Nose Throat Disord. 2005 Jun 7; 5:4", "authors": [ { "first": "Amit", "last": "Goyal" }, { "first": "Isha", "last": "Tyagi" }, { "first": "Prabhat", "last": "Tewari" }, { "first": "Surendra K", "last": "Agarwal" }, { "first": "Rajan", "last": "Syal" } ] }
PMC1180431	Background ========== The increasing incidence of and mortality from prostate cancer has led to widespread calls for prostate cancer screening. However, the subject is controversial \[[@B1]-[@B3]\]. PSA testing has high false positive rates and there is the potential for over-diagnosis of slow growing prostate cancers that may never present as a problem. Furthermore there is no strong evidence that treatments for localised cancer are effective, and no good evidence at present that screening would result in a reduction of mortality. Screening could therefore result in more harm than good \[[@B1]\]. Randomised trials are currently underway in Europe and the US to assess the impact of screening \[[@B4]\]. Definitive information from these trials will not be available until later this decade. Recent falls in mortality in the US \[[@B5]\], where there is widespread PSA testing, are often cited as being indicative of successful screening. However, interpretation of this data is complex, with factors such as lead-time bias (where PSA testing has prolonged the length of time a patient is identified as having the disease without prolonging their life) and length bias (where PSA testing has resulted in increased diagnosis of slow-growing or non-progressing tumours with a good prognosis) contributing to this outcome \[[@B6]\]. In the UK, the National Screening Committee has recommended that a prostate cancer screening programme should not be introduced at this time \[[@B7]\]. However, in response to growing public concern about prostate cancer, in 2001 the Department of Health introduced a Prostate Cancer Risk Management Programme (PCRMP) \[[@B8]\], which provides men with access to the PSA test, provided they have been given full information regarding its\' possible benefits and limitations. While rates of PSA testing in England and Wales have been assessed previously \[[@B9]\], there is very little evidence to date regarding general practitioners\' (GPs) current practice in testing for prostate cancer. Furthermore, while surveys conducted in Australia, NZ and USA have reported GPs\' knowledge and opinions of PSA testing for prostate cancer \[[@B10]-[@B14]\], little is known about GPs\' current views towards PSA testing in the UK. One survey conducted in 1998 with GPs in the North Staffordshire district of England found that 81% (134/167) GPs agreed there should not be a population screening programme at that time \[[@B15]\]. However, 57% (93/163) felt that if patients are educated about the benefits and limitations then they should be allowed to make their own decision about being tested. Only 9% (15/168) of GPs in this survey reported testing asymptomatic men, although nearly two-thirds of respondents reported that they or their staff were asked about prostate cancer screening by their patients. The main aims of this study were to determine GPs\' current reported practice regarding PSA testing, and to assess their views regarding current policy on informed decision-making and PSA testing. Methods ======= A questionnaire was developed and piloted with a sample of 30 GPs from the Oxfordshire GP research consortium and GP contacts in Wales and Hertfordshire in September 2003. The survey was hosted by MEDIX, a web-based internet service provider for doctors <http://www.medix-uk.com>. MEDIX has a user group of 12,000 doctors, with 4,700 GPs registered from a wide demographic base and with a wide distribution in terms of decade qualified. Male GPs are over-represented in the database (79%). No remuneration or incentives were offered for participating in this survey (although GPs registered with MEDIX are offered incentives for participating in certain types of online research). A random sample of 421 GPs from the MEDIX database were e-mailed an invitation to take part in the survey. The e-mail contained a link to the questionnaire which addressed GPs\' reported practice and views using a combination of direct questions and 5 consultation vignettes. The first two vignettes described men aged 55 and 70 years respectively who presented with mild lower urinary tract symptoms (LUTS). There was no mention in the vignette that the men had raised the issue of prostate cancer or PSA testing. The third man presented with concerns about his family history of prostate cancer (his father died aged 75 of prostate cancer and his brother had been diagnosed last week) and asked to be tested for prostate cancer. The fourth and fifth men were asymptomatic (aged 55 and 70 years respectively), but had both recently lost a friend to prostate cancer, and asked to be tested for prostate cancer. The survey was posted on 24/11/03. All responses were received within 30 hours. Results ======= The participants ---------------- The response rate was 95% (400/421). Of these, 94% (375) were partners in a GP practice, 4% (16) were GP assistants, and 2% (9) were GP registrars. The majority were full time (86%, 343), and male (82%, 327). There was a wide coverage by region across the UK (for example 15% (59) from the South-East, 9% (36) from the South-West, 10% (39) from the North West, 12% (49) from Scotland and 5% (21) from Wales). There was also a broad range from decade qualified, with 3% (14) qualified in the 1960s, 31% (125) qualified in the 1970s, 43% (173 qualified in the 1980s, 21% (85) qualified in the 1990s, and 1% (3) qualified in the 2000s. When compared with national statistics for GPs \[[@B16]\], female GPs, as expected, were significantly under-represented in this sample (18% vs 36%), as were non-partner GPs (6% vs 15%). General Practitioners\' reported practice ----------------------------------------- Table [1](#T1){ref-type="table"} below reports the tests which GPs reported they would elect to conduct for each of the vignettes. GPs were significantly more likely to report they would do a PSA test for a man who presents with concerns about his family history and requests a test, than for men who present with either mild lower urinary tract symptoms (LUTS) (p \< 0.0001) or asymptomatic men who request a test (p \< 0.0001). GPs were also significantly more likely to report they would do a PSA test for a 70 year old than for a 55 year old man either presenting with LUTS, or requesting a test in the absence of symptoms. GPs also frequently reported they would do a digital rectal examination (DRE) either if the man requested a test because of his family history, or presented with mild lower urinary tract symptoms. Less than half of the GPs reported they would do a DRE if the man was asymptomatic, but were significantly more likely to do one for a 70 year old than a 55 year old asymptomatic man in this situation. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Reported use of tests by general practitioners (N = 400) ::: Vignette Mild lower urinary tract symptoms (55 years) Mild lower urinary tract symptoms (70 years) Family history, asymptomatic, requests test (55 yrs) Asymptomatic, requests test (55 yrs) Asymptomatic, requests test (70 yrs) -------------------------------------- -------------------------------------------------- -------------------------------------------------- ---------------------------------------------------------- ------------------------------------------ ------------------------------------------ Prostate Specific Antigen (PSA) 60% (55,64)239 78% (73,82)311 91% (87,93)363 60% (55,65)239 71% (66,75)282 Digital rectal examination (DRE) 69% (64,73)274 78% (73,81)310 67% (62,72)268 37% (32,42)147 47% (42,52)189 Trans-rectal ultrasound 2% (1,4)9 7% (4,10)27 10% (7,13)41 1% (0,2)2 2% (0,4)7 No tests 6% (4,9)25 3% (2,5)13 3% (1,5)11 22% (18,26)88 14% (11,18)58 Other\* 47% (42,52)188 41% (36,46)165 14% (11,18)56 17% (14,21)69 13% (10,16)51 \Other mainly includes mid-stream urine, other urinary checks and discussion / counselling*. ::: When asked approximately how many asymptomatic men they had discussed PSA testing for prostate cancer with in the past 3 months, 12% (48) reported none, 65% (259) reported discussions with between 1 and 5 men, 17% (68) reported discussions with between 6 and 10 men, and 6% (25) reported discussions with more than 10 men. Twenty-four percent (94) of GPs said they had not conducted any PSA tests for asymptomatic men in the past three months, 63% (252) reported they had tested between 1 and 5 asymptomatic men, 9% (37) reported between 6--10 men, and 4% (17) reported more than 10 men. Regarding the role of practice nurses in the PSA testing process, 76% (304) said nurses were not involved, 16% (63) said the nurse conducts PSA testing following counselling from the GP, and 4% (16) reported that nurses conduct both counselling for the test and taking the sample. General practitioners\' views towards PSA testing ------------------------------------------------- 67% GPs responded that they supported the current policy. When specifically asked to indicate their preference for a national PSA testing policy, six percent (22) GPs believed PSA testing should not be available to asymptomatic men and 33% (132) GPs would prefer selective screening of high risk men. 8% (31) GPs said they would like to see the introduction of a population screening programme. The remainder supported patient or GP initiated screening, or a combination of both. No associations were found between years practising as a GP or gender of GP and the views expressed above. Informed decision-making ------------------------ When asked how GPs would prefer men to make their decision regarding whether or not to have the PSA test, 49% (196) said the man should make the final decision after seriously considering his GP\'s opinion, 34% (135) said the GP and the man should share the responsibility, 13% (51) said the man should make the final decision completely by themselves, 4% (17) said the GP should make the final decision, after seriously considering the man\'s opinion, and 1% (2) said the GP should make the final decision. GPs also indicated how they would prefer to conduct a consultation regarding PSA testing -- 55% (219) of GPs reported they would prefer to discuss the benefits and limitations of PSA, provide the man with written information, and request that the man makes another appointment if they decide to be tested; 35% (140) of GPs reported they would provide the counselling and then perform the PSA test within the same consultation if the man still wishes to be tested; 5% (20) said they would conduct the test without counselling, but explain why they are conducting the test, and 3% (13) would reassure the man that he did not need a PSA test. Discussion ========== This study of UK GPs indicates that PSA testing of asymptomatic men is a common occurrence, and that consultations which involve a discussion of PSA testing often result in the test being performed. Most GPs in the study reported they would perform a PSA test for men concerned about their family history of prostate cancer. Most reported they would also test men who present with mild LUTS, or who request a \'screening\' test in the absence of any symptoms -- particularly older men. An Australian study has also suggested that age and LUTS may provide \'cues\' to action for PSA testing \[[@B16]\]. The distinction between a \'screening\' and \'diagnostic\' PSA test tends to become blurred as men get older and the presence of some degree of urinary symptoms becomes increasingly common. Even if urinary symptoms are present there is disagreement amongst experts as to whether the PSA test should be routine or whether men should \'opt in\' to having the test \[[@B17]\]. Urinary symptoms are most often caused by benign prostatic hypertrophy (BPH), and there is a very weak evidence base for the primary care diagnosis of prostate cancer in men with LUTS \[[@B18]\]. Early, localised prostate cancer usually does not produce symptoms and by the time prostate cancer is causing urinary symptoms it is likely to have reached an advanced stage, where treatment options are reduced. Some men will have a co-existing early prostate cancer and BPH, and when a man seeks advice about LUTS this can set in train investigations which diagnose a coincidental prostate cancer. Coping with a positive PSA test performed with no prior warning could be distressing for some men, and research is needed to explore the level of counselling occurring in men with LUTS prior to PSA testing in primary care. Although the current policy recommendation in the UK is that a DRE is not warranted in asymptomatic men \[[@B19]\], almost half of the GPs in the survey indicated they would perform a DRE for an asymptomatic 70 year old man concerned about prostate cancer, and over a third for a younger man. Previous studies have also reported endorsement of use of both the PSA and DRE in asymptomatic men \[[@B20],[@B21]\], and it has been suggested that this may be because the DRE is seen as a routine component of a physical examination for men \[[@B12]\]. The American Cancer Society has also recommended the use of both \[[@B22]\], although the US Preventive Services Task Force concludes that the evidence is insufficient to recommend for or against routine screening for prostate cancer using PSA or DRE \[[@B23]\]. In line with the reported practice of GPs, the survey found that whilst GPs do not support the introduction of a national PSA screening programme, there is support for the current policy of providing men with access to the test, provided they have been given prior information on the benefits and limitations. Most respondents in this survey favoured both the GP and the man sharing in the decision-making process and the largest proportion indicated they would prefer to counsel men first and then take the blood sample during a separate consultation. Facilitation of informed decision-making is a key element of policy, but there is currently little evidence about how best to organise services to achieve an informed decision. One study reported that conducting prenatal screening in the same appointment, as opposed to requiring a return visit, may lead to higher levels of informed choice \[[@B24]\]. Another study assessing cystic fibrosis screening found patient choice was based on more knowledge when testing was part of a separate visit \[[@B25]\]. Separating information provision from PSA testing would require more consultation time, and would be less convenient for a man who knows he wants to be tested. Additionally, there may be a perception that because testing was not offered at the time the GP does not really support testing. Conversely, with same-day testing the patient may be more influenced by the GP\'s opinions and may not have had sufficient time to fully understand and consider the benefits and limitations. Research is needed to assess the most effective way of achieving and measuring informed decision making for PSA testing. In addition, attempts to study the informed decision-making process need to take into account the different influencing factors. As well as knowledge and the way testing is organised, personal attitudes and values, the attitudes of health professionals, and social and media influences can all affect the decision. This study has some limitations. The sample was drawn from a database of GPs registered with one particular internet service provider and whilst previous studies have demonstrated the acceptability of the internet as a resource for medical research in primary care \[[@B26]\], there are also recognised drawbacks \[[@B27]\]. The main limitation of the study is the potential for selection bias. As we expected from the profile of GPs registered with MEDIX, in comparison with national GP statistics \[[@B28]\] female doctors are under-represented in our study. We do not, however, believe this to be a major issue for this particular topic as anecdotal evidence from many GPs (both male and female) would indicate that the majority of consultations which include a discussion of prostate issues occur with male GPs. In support of this, some recent pilot work we have conducted found that 90% of GP responders to a survey sent out following a request for a PSA test were male. The sample is, however, also likely to have over-represented internet literate GPs who may not be representative of the wider GP community. A further important limitation of the study is the use of consultation vignettes to seek GP\'s reported behaviour. The vignettes can only provide minimal information about the patient / consultation, sometimes making expected behaviour difficult to predict. We also acknowledge the potential difference between reported and actual behaviour with respondents more likely to report what they perceive to be \'appropriate\' behaviour than what they may actually do in practice. It is important that future studies in this area address the actual behaviour of GPs. Overall PSA testing rates in the UK are rising \[[@B29]\]. Although the exact proportion of asymptomatic men being tested, and the extent to which this is patient driven, is unknown, a recent study reports that the rate of asymptomatic testing in general practice is at least 1.6% per annum and higher if private testing is included \[[@B9]\]. With the increasing media attention being given to prostate issues it is expected that these rates will continue to rise and it is therefore important there is adequate consideration of the implications for both primary and secondary care. The Prostate Cancer Risk Management Programme has circulated an information pack on this topic to GPs [http://http:\\\\www.cancerscreening.nhs.uk/prostate/informationpack.html](http://http:\\www.cancerscreening.nhs.uk/prostate/informationpack.html), but it is likely that a multi-faceted approach is required for the successful dissemination of this information to primary care \[[@B30]\]. Conclusion ========== This study indicates that PSA testing in asymptomatic men is a regular occurrence in the UK, and that there is general support from GPs for the current policy of making PSA tests available to \'informed\' men who are concerned about prostate cancer. While most GPs in the study indicated they would discuss the benefits and limitations prior to PSA testing, and most GPs favoured a shared approach to decision making, it is not known to what extent men are actually being informed. Research is required to determine the most effective approach to assisting men in making an informed decision about whether or not to have a PSA test. Competing interests =================== The author(s) declare that they have no competing interests Authors\' contributions ======================= JB participated in the design and conduct of the study, and co-wrote the manuscript; EW participated in the study design and co-wrote the manuscript; PH participated in the study design and writing the manuscript; CB participated in the study design and writing the manuscript; AE participated in the study design and writing the manuscript; GE participated in writing the manuscript; JA participated in the study design and writing the manuscript. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2296/6/24/prepub> Acknowledgements ================ We would like to thank the NHS Cancer Screening Programmes and Cancer Research UK for funding this survey. We would also like to thank Dr Rhodri Evans and Dr Mike Kirby for their input into this study, and to Prof Pat Wright and Dr Peter Rose for their helpful comments on the draft manuscript. We are also grateful to the GPs from the GP Oxfordshire Research consortium, and other GPs from Oxford, Wales and Hertfordshire for their assistance in piloting the questionnaire for this survey.	PubMed Central	2024-06-05T03:55:59.431341	2005-6-9	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180431/", "journal": "BMC Fam Pract. 2005 Jun 9; 6:24", "authors": [ { "first": "Jo", "last": "Brett" }, { "first": "Eila", "last": "Watson" }, { "first": "Paul", "last": "Hewitson" }, { "first": "Colleen", "last": "Bukach" }, { "first": "Adrian", "last": "Edwards" }, { "first": "Glyn", "last": "Elwyn" }, { "first": "Joan", "last": "Austoker" } ] }
PMC1180432	Background ========== Low back pain (LBP) is a common and costly condition in primary care practices\[[@B1]\]. Managing patients with LBP is complicated by many factors, including the inability to identify a pathological cause for the majority of patients\[[@B2],[@B3]\]. The fact that a specific diagnosis based on pathoanatomy can be made in only 10--20% of LBP sufferers seen in primary care leaves a large group of patients often given a nominal diagnosis such as \"non-specific\", or \"mechanical\" LBP. Many experts agree that sub-groups exist within the large category of patients diagnosed with \"non-specific\" LBP\[[@B4],[@B5]\]. The difficulty in identifying pathoanatomical causes in most patients combined with the high false positive rates of imaging studies\[[@B6]\] have led many to further conclude that meaningful sub-groups should be based on a patient\'s symptoms and clinical presentation \[[@B7]-[@B9]\]. The identification of sub-groups could improve the outcomes of clinical care by establishing more accurate prognoses, efficiently directing patients to therapies most likely to benefit their particular sub-group \[[@B10]-[@B12]\]. One proposed sub-group among patients with otherwise non-specific LBP consists of those who are likely to respond to spinal manipulation\[[@B13]\] There is evidence supporting the effectiveness of manipulation for patients with LBP,\[[@B14]\] however common sense, as well as research evidence,\[[@B15],[@B16]\] recognizes that not all patients with LBP should be expected to respond to a manipulation intervention. The efficiency of primary care management of patients with LBP could be improved if a pragmatic tool existed to identify those patients with LBP who are likely to respond to this approach. Flynn et al\[[@B16]\] examined factors from the clinical presentation of patients with LBP that predicted a successful response to two sessions of a manipulation intervention delivered by a physical therapist. Five clinical factors were identified as forming the most parsimonious set of predictors for identifying patients who achieved at least a 50% improvement in disability within one week with a maximum of two manipulation interventions (Table [1](#T1){ref-type="table"})\[[@B16]\]. The positive likelihood ratio for patients with at least 4 of these 5 criteria present was 24.4, which corresponded to a 95% likelihood of success with spinal manipulation among this sub-group of patients. Childs et al\[[@B17]\] sought to validate the effectiveness of these criteria for predicting a successful outcome from manipulation by randomizing patients to either a manipulation or strengthening/stabilization exercise intervention and examining outcomes based on the previously established criteria. The study found that patients with at least 4 out of 5 criteria present who received the manipulation intervention had better short-term (one- and four-week) and long-term (six-month) outcomes then patients receiving the manipulation who did not have at least 4 of 5 criteria. Furthermore, patients with at least 4 out of 5 criteria receiving manipulation had better short- and long-term outcomes than patients with 4 of 5 criteria who were randomized to receive the exercise intervention\[[@B17]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Original criteria for predicting success with a manipulation intervention.17 ::: ORIGINAL CRITERIA -- (at least 4 out of 5 must be present) ---------------------------------------------------------------- ------------------------------------------------------------------- Criterion Definition of positive 1\. Duration of current episode of low back pain \< 16 days 2\. Extent of distal symptoms Not having symptoms distal to the knee 3\. Fear-Avoidance Beliefs Questionnaire Work subscale score \< 19 points 4\. Segmental mobility testing At least one hypomobile segment in the lumbar spine 5\. Hip internal rotation range of motion At least one hip with \> 35° of internal rotation range of motion PRAGMATIC CRITERIA -- (both criteria must be present) Criterion Definition of positive 1\. Duration of current episode of low back pain \< 16 days 2\. Extent of distal symptoms Not having symptoms distal to the knee ::: The results of these studies support the validity of these five criteria for identifying patients with LBP likely to benefit both immediately and in the long-term, from spinal manipulation. The pragmatism of applying these criteria for routine use in primary care, however, may be compromised by the requirements that the patient complete a questionnaire (the Fear Avoidance Beliefs Questionnaire\[[@B18]\]), and the clinician perform an examination of hip range of motion and spinal mobility. Therefore, the purpose of this report was to use data collected from previously published studies to examine the association between two pragmatic criteria (duration of symptoms and distribution of symptoms) and response to a manipulation intervention performed by a physical therapist. Methods ======= Subjects -------- Subjects for this analysis were 141 patients with LBP who were participants in one of two previous studies. Seventy-one subjects were participants in the study used to develop the criteria for predicting success with manipulation,\[[@B16]\] and 70 subjects randomized to receive the manipulation intervention were taken from the study that validated the criteria\[[@B17]\]. These subjects underwent the same inclusion criteria, examination procedures, and treatment protocol and therefore were used to examine the validity of two criteria (duration of symptoms and distribution of symptoms) for predicting success with manipulation. Each study protocol was approved by the participating institutions and subjects all provided informed consent for participation. Subjects were between 18--60 years of age with a primary complaint of LBP with or without referral into the lower extremity, and an Oswestry disability score of at least 30%. Patients with \"red flags\" for a serious spinal condition (e.g., tumor, compression fracture, infection, etc.), signs consistent with nerve root compression (i.e., positive straight leg raise \<45°, or diminished reflexes, sensation, or lower extremity strength), current pregnancy, or prior surgery to the lumbar spine or buttock were excluded. Baseline examination -------------------- All subjects completed a series of self-report questionnaires and underwent a standardized physical examination prior to treatment. Subjects were asked to identify the date of onset of their LBP. Subjects also completed a body diagram to indicate the anatomical distribution of symptoms,\[[@B19]\] a numeric pain rating to indicate the current intensity of pain on a scale from 0--10\[[@B20]\], and the Fear Avoidance Beliefs Questionnaire to indicate the subject\'s fear of pain and beliefs about avoiding activity\[[@B18]\]. The modified Oswestry questionnaire (OSW)\[[@B21]\] was completed by each subject to quantify the level of disability related to LBP. The OSW is a well-validated tool for measuring disability in patients with LBP\[[@B22],[@B23]\]. The version used in this study has been shown to possess high levels of reliability and responsiveness\[[@B24]\]. The OSW was completed at baseline, and again after one week of treatment. Treatment --------- After completing the baseline examination, all subjects received the same manipulation intervention protocol. Subjects were sent to physical therapy and received 1--2 treatment sessions within one week before re-examination. Subjects received the manipulation intervention at each treatment session. The same manipulation technique was used for all subjects. The subject was supine. The physical therapist positioned the subject into side-bending, and then rotated the subject in the opposite direction. A quick thrust to the pelvis was delivered by the therapist in a posterior and inferior direction (Figure [1](#F1){ref-type="fig"}). If a cavitation (i.e. a \"pop\") occurred, the physical therapist proceeded to instruct the patient in the range of motion (ROM) exercise. If no cavitation was produced, the manipulation was attempted again. A maximum of two attempts per side was permitted. The ROM exercise was performed supine by rocking the pelvis back and forth. Subjects were instructed to perform 10 repetitions of the ROM exercise in the clinic and 10 repetitions 3--4 times daily on the days they did not attend physical therapy. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Manipulation technique. ::: ![](1471-2296-6-29-1) ::: Data analysis ------------- All subjects were categorized on two criteria from the baseline examination. The body diagram was used to determine if the subject\'s symptoms extended distal to the knee or not. The categorization of symptom distribution using body diagrams has been found to be highly reliable\[[@B25]\]. The subject\'s report was used to determine if the symptom duration was less than 16 days. Subjects with both criteria present (i.e., no symptoms distal to the knee andsymptom duration less than 16 days) were categorized as likely to have a good prognosis following the manipulation intervention. Subjects with 1 or 0 criteria present were categorized as likely to have a poor prognosis. We first examined the relationship between the two criteria rule and the original five-criteria rule for predicting success. Subjects with at least 4 out of 5 of the original criteria present were categorized as having a good prognosis with manipulation, while those with three or fewer were categorized as having a poor prognosis. Using the five-criteria rule as the reference standard, we calculated the accuracy of the two criteria as the percentage of times the two-criteria categorization agreed with the five-criteria categorization. We also calculated the sensitivity, specificity, and likelihood ratios with associated 95% confidence interval (CI). The outcome of the manipulation intervention was determined from the change on the OSW occurring between the baseline and post-treatment examinations. The percent improvement on the OSW was calculated as (Initial ~OSW~- Final ~OSW~) /Initial ~OSW~\* 100%. If the percent improvement was ≥ 50%, the intervention was considered a success. The association between the two criteria and manipulation outcome was examined by calculating the sensitivity, specificity, and positive likelihood ratio statistics with corresponding 95% CI. The sensitivity was calculated as the percentage of subjects with a successful outcome who had a good prognosis based on the presence of both criteria (i.e., the true positive rate). Specificity was calculated as the percentage of subjects with a non-successful outcome who were categorized with a poor prognosis based on the presence of only 1 or 0 criterion (i.e., the true negative rate). The positive likelihood ratio was calculated as (sensitivity / (1-specificity)) and represents the increase in odds favoring a successful outcome when both criteria are present\[[@B26]\]. Results ======= The baseline characteristics of the 141 subjects are listed in table [2](#T2){ref-type="table"}. At baseline, 105 subjects (74%) did not have symptoms distal to the knee, and 50 (36%) had a duration of symptoms \< 16 days. Overall, 41 subjects (29%) had both criteria present and were categorized as having a good prognosis. One hundred subjects (71%) were categorized as having a poor prognosis; 73 had one, and 27 had zero criterion present. Characteristics of subjects with or without both criteria present are presented in table [3](#T3){ref-type="table"}. Patients with both criteria present had higher baseline OSW scores (mean difference 5.6 points, 95% CI: 1.7, 9.5). The accuracy of the two criteria rule as compared with the original five criteria rule was high, with a percent agreement of 83.7%. The sensitivity and specificity were 0.67 and 0.93 respectively, resulting in a positive likelihood ratio of 8.9 and negative likelihood ratio of 0.36 (table [4](#T4){ref-type="table"}). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Baseline subject characteristics (n = 141). ::: Characteristic ------------------------------------------------------- ------------------------------------------ Age 35.5 (± 11.1) years Sex 49% female Duration of Symptoms Median = 22 days (range: 1 -- 2775 days) Distribution of Symptoms 26% symptoms distal to the knee Pain Intensity Rating 5.3 (± 2.0) Fear-Avoidance Beliefs Questionnaire -- Work Subscale 14.7 (± 10.3) Oswestry Disability Questionnaire 41.9 (± 10.9) (Numbers represent mean (± standard deviation) unless otherwise indicated) ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Baseline characteristics of subjects with or without both criteria for success with manipulation present ::: Characteristic Both criteria present (n = 41) Both criteria not present (n = 100) ------------------------------------------------------- ------------------------------------ ----------------------------------------- Age 36.9 (± 11.4) years 35.2 (± 10.4) years Sex 39% female 53% female Pain Intensity Rating 5.6 (± 2.2) 5.1 (± 1.9) Fear-Avoidance Beliefs Questionnaire -- Work Subscale 14.0 (± 11.0) 15.0 (± 10.1) Oswestry Disability Questionnaire\* 45.9 (± 13.7) 40.3 (± 9.1) (Numbers represent mean (± standard deviation) unless otherwise indicated, \significant difference between groups (p = 0.02)) ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Accuracy of two criteria rule relative to the five criteria rule for predicting response to manipulation ::: Likely responder\ Likely non-responder\* ------------------------------------------------------- ----------------------------------------- ----------------------------------------- Both criteria present 32 7 Both criteria not present** 16 86 Sensitivity = 0.67 (95% CI: 0.53, 0.78) Specificity = 0.93 (95% CI: 0.85, 0.96) Positive Likelihood Ratio = 8.9 (95% CI: 4.2, 18.6) Negative Likelihood Ratio = 0.36 (95% CI: 0.24, 0.54) \* The reference standard was the categorization of the subject based on the original five criteria prediction rule. ::: The mean baseline and one-week OSW scores were 41.9 (± 10.9) and 24.1 (± 14.2) respectively. The mean percent change on the OSW was 41.2% (± 33.9%). Sixty-three subjects (45%) experienced 50% or greater improvement on the OSW and were categorized as successful with the treatment (mean change 73.8% (± 15.4%)), and 55% were categorized as unsuccessful (mean change 14.8% (± 17.9%)). Subjects with both criteria present experienced significantly greater change on the OSW, and were more likely to be categorized as a successful manipulation outcome than subjects with one or zero criteria present (Table [5](#T5){ref-type="table"}). ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Manipulation outcomes based on the number of criteria present ::: Number of criteria present Number of subjects Number (%) with successful outcome Percent change in oswestry (mean, (sd)) -------------------------------- ------------------------ ---------------------------------------- --------------------------------------------- 0 27 2 (7.4%) 16.9% (23.4%) 1 73 26 (34.4%) 37.0% (32.7%) 2 41 35 (85.4%) 64.6% (27.2%) ::: The distribution of outcome with the manipulation intervention based on the presence of both criteria is displayed in table [6](#T6){ref-type="table"}. The sensitivity associated with the presence of both criteria was 0.56 (95% CI: 0.43, 0.67), and the specificity was 0.93 (95% CI: 0.84, 0.96), with a positive likelihood ratio of 7.2 (95% CI: 3.2, 16.1). ::: {#T6 .table-wrap} Table 6 ::: {.caption} ###### Accuracy of two criteria rule for success with manipulation for predicting clinical outcome ::: Success\* Non-success\* ----------------------------------------------------- ----------------------------------------- ----------------------------------------- Both criteria present 35 6 Both criteria not present 28 72 Sensitivity = 0.56 (95% CI: 0.43, 0.67) Specificity = 0.92 (95% CI: 0.84, 0.96) Positive Likelihood Ratio = 7.2 (95% CI: 3.2, 16.1) \*The reference standard was ≥50% reduction in Oswestry disability score. ::: Discussion ========== The results of this study demonstrate a strong association between the two pragmatic criteria and response to a manipulation intervention delivered by a physical therapist. These two criteria (duration of symptoms less than 16 days and no symptoms extending distal to the knee) can be easily assessed in a primary care setting. The two pragmatic criteria also showed a high level of accuracy in relation to judgments based on the original five criteria (84% agreement, 67% sensitivity, 93% specificity). Judgments from the two criteria showed a strong predictive relationship with the clinical outcome. Although the two criteria rule did sacrifice some accuracy related to the original five criteria rule, it appears that sufficient accuracy is maintained with a substantial increase in ease of use. The original five criteria rule has been validated in a randomized clinical trial\[[@B17]\]. This two criteria rule will also require further validation through randomized trials to confirm that patients with both factors respond best to an intervention involving manipulation versus an alternative intervention, or no intervention at all. Current clinical practice guidelines and expert recommendations generally support a \"stepped care\" approach for management of patients with LBP\[[@B27]\] in which treatment is initially limited to providing positive information and advice based on the understanding that the majority of patients will recover within 4--6 weeks. Referral to physical therapy may therefore not be initiated (ie, \"stepped up\") until patients fail to demonstrate the anticipated recovery. However, the results of this analysis, .along with our previous analyses,\[[@B16],[@B17]\] suggest that patients with LBP seen in primary care whose duration of symptoms is less than 16 days without any symptoms extending distal to the knee, may benefit from an immediate referral to physical therapy for a couple of sessions of a manipulation intervention along with range of motion exercise. This analysis only examined results immediately after one week of treatment (up to 2 sessions), however the randomized trial by Childs et al\[[@B17]\] show that the benefit of early manipulation with range of motion exercise in patients fitting the criteria for success persisted up to six months after the treatment period. In addition, several studies have demonstrated that patients with LBP whose disability persists beyond 4--6 weeks are at significantly increased risk of developing chronic disability, persistent work restrictions, and increased health care utilization \[[@B28]-[@B31]\]. In this study, 29% of subjects fit the two criteria and were designated as having a good prognosis with manipulation, suggesting this sub-group of patients may not be inconsequential. The high specificity (0.92) and positive likelihood ratio (7.2) indicate that patients with both criteria present should be referred for a manipulation intervention based on the high likelihood of rapid success. The sensitivity (0.56) and negative likelihood ratio (0.48) associated with this two-criteria rule were only moderate, indicating a relatively high potential for false negative results (i.e., subjects designated as likely non-responders who ultimately experienced success with manipulation). Given the safety of manipulation in the lumbar spine\[[@B32]\], this finding suggests that referral of patients who do not have both criteria present may be appropriate in some cases. The criteria for treatment outcome with manipulation was determined after 1--2 treatment sessions, suggesting that the effectiveness of this treatment can be determined quickly in referred patients. Those patients who do not respond quickly may then be rapidly directed towards an alternative treatment approach. There is preliminary evidence to suggest that the addition of a strength and stabilization program after the manipulation intervention may help to reduce the likelihood of recurrence,\[[@B33]\] however more research is needed. The exclusion criteria used in this study should be taken into account when considering the clinical implications of the results. In particular it is important to note that patients with signs of nerve root compression, low levels of disability, or prior surgery to the low back were excluded. The likely response of these patients to a manipulation intervention cannot be determined from these results. This study used one manipulation technique. It cannot be ascertained if the same criteria would apply to a different technique, however the lack of specificity associated with manipulation techniques\[[@B34],[@B35]\] suggests that the choice of a particular technique may not be as important as the choice of the patient on whom spinal manipulation is to be used. Conclusion ========== Individuals with \"non-specific\" LBP are not a homogenous group, and different sub-groups of patients are likely to preferentially respond to different therapeutic management strategies. One sub-group consists of those patients with a good prognosis following spinal manipulation intervention. The results of this study demonstrate an association between two factors; symptom duration of less than 16 days, and no symptoms extending distal to the knee, and outcome of a manipulation intervention. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= JMF participated in the design and statistical analysis of this study. JDC participated in the design and performance of this study. TWF participated in the conceptual development and performance of this study. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2296/6/29/prepub> Acknowledgements ================ Supported in part by a grant from the Foundation for Physical Therapy and the Wilford Hall Medical Center Commander\'s Intramural Research Program. The opinions and assertions contained herein are the private views of the author (JDC) and are not to be construed as official or as reflecting the views of the Department of the Air Force or the Department of Defense.	PubMed Central	2024-06-05T03:55:59.433125	2005-7-14	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180432/", "journal": "BMC Fam Pract. 2005 Jul 14; 6:29", "authors": [ { "first": "Julie M", "last": "Fritz" }, { "first": "John D", "last": "Childs" }, { "first": "Timothy W", "last": "Flynn" } ] }
PMC1180433	Background ========== Africa is inhabited by populations that show high levels of genetic diversity compared to most other continental populations today and it is thought to be the ancestral home of modern humans. African populations have the largest number of population specific autosomal, X-chromosomal and mitochondrial DNA haplotypes with non-African populations having only a subset of the genetic diversity present in Africa \[[@B1]\]. Estimates of F~ST~(the classic measure of population subdivision) from mitochondrial DNA are much higher in Africa than other populations, as summarized by Tishkoff et al \[[@B1]\]. In addition, analyses from studies based on autosomal SNPs, STRPs or Alu elements show higher F~ST~values for African populations \[[@B2]-[@B4]\]. Recent studies of world populations based on large genomic data also reported significant population structure among the African groups \[[@B5],[@B6]\]. However, given the cultural and linguistic diversity of African populations (with over 2000 distinct ethnic groups and languages), these studies have typically included only a handful of African populations indicating that most African populations have not been studied. As previously noted, most existing genetic data on African populations have come from a few countries that are relatively economically developed and/or with key research or medical centers \[[@B1]\]. Availability of more genetic data from sub Saharan Africa will clearly be useful in our understanding of population structure, demographic history and the efforts to map disease-causing genes. Several genetic epidemiologic studies mapping complex disease-causing genes have been designed to take advantage of the population genetic characteristics of contemporary African populations for fine mapping of informative genomic regions. These characteristics include lower linkage disequilibrium values \[[@B5]-[@B9]\] and smaller haplotype block sizes \[[@B10],[@B11]\]. On the other hand, African populations have more divergent patterns of LD and more complex pattern of population substructure or stratification \[[@B12]-[@B17]\]. Population stratification refers to differences in allele frequencies between cases and controls due to systematic differences in ancestry rather than association of genes with disease and it can have a major impact on the ability of genetic epidemiologic studies to detect valid associations between a putative risk allele and a disease or trait. We investigated population structure or stratification in four ethnic groups in two countries in West Africa (Akanand Gaa-Adangbefrom Ghana, Yorubaand Igbofrom Nigeria) using data from 372 autosomal microsatellite loci \[see [Additional file 1](#S1){ref-type="supplementary-material"}\] typed in 493 unrelated persons (986 chromosomes). Firstly, we used a clustering algorithm to infer population structure in the whole sample while ignoring ethnic group information and compare our findings to reported ethnic grouping. Next, we used analysis of molecular variance (AMOVA) models on the same data. Finally, we estimate FST and allele sharing distances between all population pairs. Results ======= The estimates of the logarithms of the probability of the data under the models and assumptions regarding independence of allele frequencies are shown in Table [1](#T1){ref-type="table"}. Under the admixture model, the smallest probability is associated with a prior K of 1 and little of the posterior probability is associated with higher K values. The distribution of members of the sample to inferred clusters is consistent with this observation. The proportion of individuals assigned to each cluster is approximately the same with little variation between ethnic groups (Table [2](#T2){ref-type="table"}). This symmetry is strongly suggestive of the absence of population structure in the AADM study sample. This is so because real population structure is associated with individuals being strongly assigned to one inferred cluster or another with the proportions assigned to each ethnic group showing asymmetry. The posterior probability under the no-admixture model also favours a K of 1. Examination of the distribution of individuals sampled to inferred clusters also shows the same strong symmetry. These consistent displays of symmetry suggest that a K of 1 is the most parsimonious model. The same conclusion was reached by examining the membership coefficients (Q). Irrespective of the value of K between the range of 2 and 6, Q is similar across the whole sample as illustrated by the bar plots in Figure [2](#F2){ref-type="fig"}. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Bar plots of estimates of membership coefficient (Q) for each individual by ethnic group. Legend for population groups: 0 = Akan, 1 = Gaa-Adangbe, 7 = Yoruba, 8 = Igbo. Analyzed under admixture model, assuming correlated allele frequencies. ::: ![](1471-2156-6-38-2) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Estimates of log probability of data under various assumptions for K = 1--6 ::: K No-admixture model Admixture model --- ------------------------ --------------------- --------- ---------------- 1 -642431 \~1.0 -642486 \~0.99 2 -646015 0 -642606 5.6 × 10^-57^ 3 -649140 0 -642800 2.9 × 10^-137^ 4 -649168 0 -644022 0 5 -647275 0 -645623 0 6 -652040 0 -647265 0 ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Proportion of membership of each ethnic group in inferred clusters for K= 2 to 6 under admixture model with correlated allele frequencies ::: Inferred cluster ----------- ------------- ---------------------- ------ ------ ------ ------ ------ K = 2 Akan 0.52 0.48 Gaa-Adangbe 0.51 0.49 Yoruba 0.46 0.54 Igbo 0.44 0.56 K = 3 Akan 0.33 0.33 0.34 Gaa-Adangbe 0.33 0.33 0.34 Yoruba 0.33 0.34 0.31 Igbo 0.35 0.35 0.30 K = 4 Akan 0.25 0.29 0.21 0.25 Gaa-Adangbe 0.25 0.28 0.22 0.25 Yoruba 0.23 0.24 0.28 0.25 Igbo 0.23 0.23 0.30 0.24 K = 5 Akan 0.20 0.20 0.20 0.24 0.16 Gaa-Adangbe 0.20 0.21 0.20 0.22 0.17 Yoruba 0.20 0.21 0.18 0.19 0.22 Igbo 0.19 0.21 0.18 0.18 0.24 K = 6 Akan 0.22 0.17 0.17 0.14 0.16 0.14 Gaa-Adangbe 0.20 0.17 0.16 0.14 0.18 0.15 Yoruba 0.15 0.16 0.16 0.18 0.17 0.18 Igbo 0.13 0.16 0.15 0.20 0.17 0.19 ::: Analysis of molecular variance (AMOVA) shows that most of the variance in the sample is attributable to within-ethnic groupvariation (99.91% of the variance) and between-ethnic groupvariation is only 0.09% (Table [3](#T3){ref-type="table"}). Locus-by-locus AMOVA shows that this pattern of partitioning of the variance between within-population and between-population variation is consistent across all loci and can be observed on single locus analysis \[see [Additional file 2](#S2){ref-type="supplementary-material"}\]. An AMOVA model that includes \"country\" as well as \"ethnic group\" in the model shows that the variance attributable to between-countryvariation was 0.13%, that due to between-ethnic groupvariation was 0.01% and that due to within-ethnic groupvariation was 99.86% (Table [3](#T3){ref-type="table"}). The between-countrygenetic variance in this model was significant, suggesting that the two groups from one country can be distinguished from the groups from the other country. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Analysis of Molecular Variance (AMOVA) results: AADM Study ::: Source of variation d.f. Sum of squares Variance components % variation -------------------------------------- ---------- -------------------- ------------------------- ----------------- Model A: Among ethnic groups 3 494.426 0.126 (V~a~) 0.09 Within ethnic group 982 132287.848 134.713 (V~b~) 99.91 Total 985 132782.274 134.839 Model B: Among countries 1 220.117 0.172 (V~a~) 0.13 Among ethnic groups within countries 2 274.309 0.012 (V~b~) 0.01 Within ethnic group 982 132287.848 134.713(V~c~) 99.86 Total 985 132782.274 134.895 ::: Pair-wise genetic distance measures show that there is little difference between the four ethnic groups (Table [4](#T4){ref-type="table"}). The fact that all calculated pair-wise F~ST~values were low suggests little evidence for genetic differentiation between the ethnic groups. The fixation index for the entire sample as estimated by F~ST~is 0.00093. Allele-sharing distances are also similar between the groups (Table [3](#T3){ref-type="table"}). Plotting these distances on an unrooted radial tree using a neighbour-joining algorithm (Figure [3](#F3){ref-type="fig"}) suggests that the two Ghanaian groups can be distinguished from the two Nigerian groups. This observation is consistent with the findings of the hierarchical AMOVA model in Table [3](#T3){ref-type="table"}. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Unrooted radial neighbour-joining tree showing the genetic relationships of the four populations groups studied. ::: ![](1471-2156-6-38-3) ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Pairwise genetic distances between the ethnic groups studied ::: Group Akan Gaa-Adangbe Yoruba Igbo --------------- --------- --------------- ------------- ------------- Akan \* 0.11833 0.10410 0.10798 Gaa-Adangbe 0.00013 \* 0.12470 0.12793 Yoruba 0.00099 0.00072 \* 0.09508 Igbo 0.00177 0.00162 0.00005 \* Notes: Above diagonal: Allele-sharing distance (Bowcock et al 1994) Below diagonal: F~ST~(Weir and Cockerham, 1984) ::: Discussion ========== Using data from 372 microsatellite loci typed in 493 unrelated persons from four major ethnic groups in Nigeria and Ghana, we sought for evidence of population structure using several methods. Our results did not show any significant population substructure and no ethnic group corresponded to inferred clusters. This finding has been reported by others \[[@B5]\]. Although Rosenberg et al observed significant population structure among six African groups (Bantu-Kenya, Mandenka, Yoruba, San, MbutiPygmy and BlakaPygmy), they reported that inferred clusters for some of the African populations did not correspond to predefined groups, unlike groups from America, Oceania and Eurasia \[[@B5]\]. The within-population component of genetic variation accounts for most of the diversity in the sample. This is consistent with previous findings \[[@B5]\] showing that the within-population component of genetic variance among six African populations studied was 96.9%; we estimated an even higher value of 99.9% in this study. The higher value of the within-population variance in this study is likely due to the smaller geographic area from which the samples were derived. The maximum distance between any two sites in this study is less than 700 miles and there are no major natural barriers e.g., mountains, between the regions inhabited by the groups. In addition, these four ethnic groups have a long history of trade and other interactions and they all speak languages belonging to the Niger-Kordofanian group. As noted by Cavalli-Sforza et al \[[@B18]\] the genetic relationships observed in West Africa indicate that major migrations and admixtures occurred within the region in earlier times It is important to point out that despite the small amount of genetic differentiation in the sample as a whole, it was possible to distinguish between the groups from each country using a hierarchical AMOVA model and a dendrogram algorithm. Thus, the absence of significant population structure between the four groups did not mean that the groups could not be distinguished from each other. Rather, the data in Table [4](#T4){ref-type="table"} show that enough differences exist to separate the two populations from Nigeria from those from Ghana. From the disease-mapping point of view, population stratification is important in the analysis of association genetic data, especially when that data is being used to infer the contribution of genetics to a disease. The presence of undetected population structure can mimic association (leading to more false positives) or mimic lack of association (leading to false negatives) \[[@B19]\]. While there has been much debate about the impact of population stratification on association studies, there are limited data that quantify the magnitude of this effect. The largest study to quantify this effect analyzed data from 11 case-control and case-cohort association studies \[[@B20]\] and showed that there was no statistically significant evidence for stratification. However, most of the studies evaluated above used limited number of markers making it difficult to completely rule out moderate levels of stratification that could lead to the finding of false positive associations. Typically, efforts are made to minimize the effect of stratification during study design and data analysis, including a careful selection of cases and controls (e.g., matching) and by conducting family-based association tests. However, for the size of study needed to detect typical genetic effects in common diseases, even modest levels of population structure within population groups cannot be safely ignored \[[@B19]\]. Given this, we have searched for evidence of population stratification in this genetic epidemiologic study, the first of its kind for T2DM in West Africa. Noting that the number of markers needed to assess stratification depends on the magnitude of genetic effects under study \[[@B19]\], we have used a large number of markers, rather than just a few dozen as in many studies. The number of markers we have used (372) can bring the conservative 95^th^percentile upper bound on the level of stratification to within 10% of the true value \[[@B20]\]. Conclusion ========== In summary, there was little evidence for significant population substructure in the four major West African ethnic groups represented in the AADM study sample. Classification of individuals into clusters showed symmetry, with roughly the same proportion of each ethnic group assigned to each cluster(s). Ethnicity apparently did not introduce differential allele frequencies that may affect analysis and interpretation of linkage and association studies. These findings, although not entirely surprising given the geographical proximity of these groups, provide important insights into the genetic relationships between the ethnic groups studied and confirm previous results that showed close genetic relationship between most studied West African groups. Methods ======= The AADM study is an affected sibling pair (ASP) design with enrolment of available spouses as controls. Recruitment strategies and eligibility criteria for the families enrolled in this report have been described in a previous publication \[[@B21]\]. The three centers in Nigeria (Enugu, Ibadan and Lagos) enrolled 2 major ethnic groups -- Igbos(28%) and Yorubas(28%); the two centers in Ghana (Accra and Kumasi -- see figure [1](#F1){ref-type="fig"}) enrolled two major ethnic groups -- Akan(25%) and Gaa-Adangbe(11%). For this analysis, 493 unrelated persons were studied, comprising 147 Akan, 61 Gaa-Adangbe, 129 Yorubaand 156 Igboparticipants. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Map of Africa showing the AADM field sites in the two countries. ::: ![](1471-2156-6-38-1) ::: Marker set ---------- Genotyping was done at the Center for Inherited Disease Research (CIDR). The CIDR marker set is composed primarily of trinucleotide and tetranucleotide repeats and consists of 392 primer pairs with average spacing of 8.9 cM throughout the genome. There are no gaps in the map larger than 18 cM. The average marker heterozygosity is 0.76. Approximately 10% of the marker loci are different between the current CIDR marker set and the Marshfield Genetics screening set version 8. Almost all reverse primer sequences have been modified from the version 8 sequences in order to reduce \'+A\' artifacts. The resulting PCR products are sized using a capillary sequencing platform. Data for the markers are generated with 218 PCR reactions (41 triplex reactions, 92 duplex reactions and 85 single reactions). Each primer pair has undergone extensive optimization to improve performance and reliability. Error rate was 0.1% per genotype. Inconsistency rate was 0.11%. Extensive quality checks were carried out to verify consistency of marker genotyping as previously described \[[@B22]\]. For this analysis, all 372 typed autosomal microsatellite markers were included. The markers comprised 272 (73%) tetranucleotide, 46 (12%) trinucleotide and 54 (15%) dinucleotide microsatellites. The markers and their characteristics are provided \[see [Additional file 1](#S1){ref-type="supplementary-material"}\]. The raw genotype data can be obtained by contacting the authors (<aadeyemo@howard.edu> or <crotimi@howard.edu>.) Analysis -------- We used a model-based clustering method for inferring population using genotype data consisting of unlinked markers as implemented in the structureprogram version 2.1 \[[@B23]\]. The model assumes there are Kpopulations (where Kmay be unknown), each of which is characterized by a set of allele frequencies at each locus. Individuals in the sample are assigned probabilistically to populations, or jointly to two or more populations if their genotypes indicate they are admixed. It is assumed that within populations, the loci are at HWE and linkage equilibrium. This method has the advantage that it does not assume any particular mutation model and it can be applied to microsatellite, SNP and RFLP data. The data was analyzed under an admixture model, assuming correlated allele frequencies between populations as previously described \[[@B24]\]; these assumptions have the advantage of being able to detect recent population divergence and recent admixture, thus giving better performance on difficult problems, although at the potential cost of overestimating K \[[@B23]\]. The analysis was then repeated under a no-admixture model, assuming independence of allele frequencies. Each run was done for K = 1 to 6 after 100,000 burn-in iterations and 10^6^estimation iterations (admixture model) or 2 × 10^6^estimation iterations (non-admixture model). Each run was carried out several times to ensure consistency of the results. Posterior probabilities for each K were computed for each set of runs. Analysis of molecular variance (AMOVA) was done using data from all 372 loci as implemented in Arlequin 2000\[[@B25]\]. AMOVA enables the partition of genetic variance at a locus or several loci into variation withinpopulations and variation betweenpopulations. In addition, AMOVA can be used for a hierarchical analysis of three genetic-variance components -- those due to genetic differences (i) between individuals within groups, (ii) between populations within groups, and (iii) between groups. We conducted AMOVA analyses on the study sample using two models (a) a model in that partitioned the genetic variance into that within each ethnic group and that between ethnic groups, (b) a hierarchical model with the country as the first level and the ethnic group within each country as the second level. Additional locus-by-locus AMOVA analysis was done (see [Additional file 2](#S2){ref-type="supplementary-material"}). Significance of the AMOVA values was estimated by used of 10,000 permutations. F~ST~, the fixation index or coancestry coefficient \[[@B26]\], was also computed as a measure of the effect of population division. F~ST~ranges from 0 (no population subdivision, random mating occurrence, no genetic divergence within the population) to 1 (complete isolation or extreme division), and F~ST~values of up to 0.05 represents negligible genetic differentiation. Allele-sharing genetic distances \[[@B14]\] were also computed between each pair of ethnic groups. Authors\' contributions ======================= AA and CR conceived and designed the study; AA did the statistical genetic analyses; AA and CR drafted the manuscript. GC and YC contributed to the interpretation of the results and development of the manuscript. Supplementary Material ====================== ::: {.caption} ###### Additional File 1 The 372 microsatellite markers on the 22 autosomes studied ::: ::: {.caption} ###### Click here for file ::: ::: {.caption} ###### Additional File 2 Single locus AMOVA for all 372 loci studied ::: ::: {.caption} ###### Click here for file ::: Acknowledgements ================ This project is also supported in part by multiple NIH institutes including NCMHD, NCRR, NHGRI, NIGMS and NIDDK. The AADM investigators and physicians made this study possible and are hereby acknowledged. Genotyping was done by the Center for Inherited Disease Research (CIDR) and detailed information on laboratory methods and markers can be found at the CIDR web site \[[@B27]\].	PubMed Central	2024-06-05T03:55:59.435345	2005-6-24	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180433/", "journal": "BMC Genet. 2005 Jun 24; 6:38", "authors": [ { "first": "Adebowale A", "last": "Adeyemo" }, { "first": "Guanjie", "last": "Chen" }, { "first": "Yuanxiu", "last": "Chen" }, { "first": "Charles", "last": "Rotimi" } ] }
PMC1180434	Background ========== Like the chicken, the Japanese quail belongs to the Phasianidae family \[[@B1]\]. In the wild, it is a small migratory bird originating from the Far East which was first raised in bird cages in Japan and China around the 15^th^century because of its singing abilities, and later became to be known in Japan as a good egg-laying bird for human consumption \[[@B1]\]. Exported to North America in the 1930s, it started to be used for pilot studies in genetics and physiology and in avian research \[[@B2]\]. The Japanese quail is now a well established animal model in biology and a bird used for intensive egg and meat production mainly in Asia and Europe, but also in the Middle East and America \[[@B3]\]. Until recently, however, only a few linkage groups have been known for the Japanese quail, and few genes have been mapped \[[@B4]\] despite the large variety of traits that have been studied in the Japanese quail \[[@B3]\]. The development of the first microsatellite \[[@B5]\] and AFLP \[[@B6]\] panels and of the corresponding maps \[[@B6],[@B7]\] have made it possible to localize new genes \[[@B8],[@B9]\], which should further enhance the interest of the quail in biological research and open the way to more comparative genetic studies with chickens. Maps can also be used to do a genome-wide search for quantitative trait loci (QTL) by locating QTL nearby anonymous AFLP or microsatellite markers in segregating backcross or F2 populations. Moreover, the existence of markers that cross-amplify quail and chicken DNA \[[@B5],[@B8]\] makes it possible to connect both maps and to compare the QTL results for these two species, thereby getting further insight into their evolutionary relationship. The objective of the present work was to produce the first set of QTL detected in Japanese quail for growth, egg production and quality, feed consumption and other traits. For that purpose, an F2 experiment was carried out with two quail lines selected for early egg production \[[@B10]\] and for high duration of tonic immobility \[[@B11]\]. The microsatellite map \[[@B7]\] used for the detection of QTL in the present work was developed earlier from the same F2 individuals. Results ======= Overall performances -------------------- Table [1](#T1){ref-type="table"} shows the number of observations, the mean, the standard deviation and the range of values for the traits recorded on the 434 F2 females of the experiment. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Elementary statistics on the characters measured for the 434 F2 female Japanese quail of the genome scan ::: Trait^1^ N Mean Standard deviation Minimum Maximum ----------- ----- ------- -------------------- --------- --------- TI (s) 434 177.8 86.3 17.0 300.0 BT (C) 434 41.67 0.36 40.70 43.20 BW1 (g) 434 189.0 16.8 146.6 238.8 FI (g/d) 321 27.4 2.8 18.9 35.8 RFI (g/d) 321 0.0 1.8 -4.2 7.5 EW (g) 323 11.9 1.1 9.1 15.3 YW (g) 323 3.0 0.4 2.1 5.2 SW (g) 323 0.88 0.09 0.66 1.25 Y/A (%) 323 41 5 31 78 AFE (d) 429 44.2 5.2 37 87 EN 363 322.4 72.8 18 415 CL (d) 363 7.9 4.1 1.3 47.9 BW2 (g) 352 271.0 30.4 185.0 361.0 ^1^: TI = tonic immobility at 8--9 days of age; BT = rectal temperature at 5 weeks of age; BW1 = 5-wk body weight; FI = feed intake at 30 weeks of age; RFI = residual feed intake at 30 weeks of age; EW = egg weight at 30 weeks of age; YW = yolk weight; SW = eggshell weight; Y/A = ratio of yolk weight over albumen weight; AFE = age at first egg; EN = number of eggs laid until the age of 69 weeks; CL = clutch length; BW2 = 70-wk body weight. ::: QTL analysis ------------ Table [2](#T2){ref-type="table"} shows the location of the significant QTL, their position on the chromosome, the maximum F value obtained at this position, their genetic effects, the reduction of the residual variance obtained by fitting a QTL at this location, and the corresponding chromosome-wide and genome-wide significance levels. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Chromosomal location, test statistic (F), genetic effects and significance of the QTL detected in the Japanese quail ::: Chromosome (map length cM) Trait^1^ Position (cM) Flanking markers F Additive effect ± SE Dominance effect ± SE Reduction of σ^2^(%) Chromosome-Wide Probability Genome-Wide ---------------------------- ---------- --------------- ------------------- ------ ---------------------- ----------------------- ---------------------- ----------------------------- ------------- ------------------ CJA01 (206) CL 0 GUJ0055 9.9 -0.93 ± 0.36 -1.81 ± 0.58 4.9 0.0006 0.002 very significant BW2 18 GUJ0055-GUJ0052 11.6 -1.29 ± 0.27 ns^2^ 5.7 0.0002 0.0005 very significant BW1 19 GUJ0052 5.9 -4.36 ± 1.38 ns 2.3 0.034 0.09 suggestive FI 19 GUJ0052 8.3 -1.03 ± 0.26 ns 4.4 0.004 0.01 significant TI 91 ADL0037 6.2 -20.3 ± 6.3 ns 2.5 0.034 0.09 suggestive BT 180 GUJ0062 5.7 -0.088 ± 0.026 ns 2.5 0.047 0.12 ns SW 191 GUJ0062-GUJ0068 7.2 -0.027 ± 0.009 -0.030 ± 0.0134 3.4 0.016 0.04 significant CJA02 (61) BW1 54 GUJ0063-GUJ0027 4.6 ns 7.42 ± 2.75 1.7 0.038 0.29 ns CJA03 (38) BT 1 GUJ0099 6.1 -0.090 ± 0.029 ns 2.5 0.008 0.11 ns CJA05 (21) SW 12 GUJ0049 5.5 -0.026 ± 0.008 ns 2.3 0.012 0.27 ns CJA06 (74) EW 0 GUJ0021 5.4 0.330 ± 0.100 ns 2.7 0.023 0.16 ns EN 32 GUJ0087 7.0 21.6 ± 6.1 ns 3.3 0.006 0.04 significant AFE 34 GUJ0087-GUJ0054 6.2 -1.52 ± 0.46 ns 2.9 0.012 0.09 suggestive CJA09 (25) BW1 25 GUJ0071 4.3 3.34 ± 1.22 ns 1.6 0.028 0.46 ns CJA14 (8) BW2 7 GUJ0023-GUJ0097 6.2 ns -1.32 ± 0.38 2.9 0.004 0.24 ns CJA20 (25) FI 2 GUJ0065-GUJ0083 8.5 -0.883 ± 0.231 ns 4.5 0.001 0.02 significant SW 21 GUJ0065-GUJ0083 6.7 -0.026 ± 0.008 0.024 ± 0.013 3.4 0.003 0.06 suggestive RFI 25 GUJ0083 7.2 -0.525 ± 0.142 ns 3.8 0.002 0.04 significant ^1^: CL = clutch length (d); BW2 = 70-wk body weight (g); BW1 = 5-wk body weight (g); FI = feed intake at 30 weeks of age (g/d); TI = tonic immobility at 8--9 days of age (s); BT = rectal temperature at 5 weeks of age(C); SW = eggshell weight at 30 weeks of age (g); EW = egg weight (g); EN = egg number until the age of 69 weeks; AFE = age at first egg (d); RFI = residual feed intake at 30 weeks of age (g/d). ^2^: ns = not significant. ::: Out of the 18 QTL found to be chromosome-wide significant (p~c~\< 0.05) on 8 of the 12 autosomes scanned in this work, 7 were genome-wide significant (p~g~\< 0.05), and 4 were genome-wide suggestive (p~g~\< 0.10). Genome-wide significant QTL for early (BW1) and late (BW2) measures of body weight were only found on CJA01, but there was a chromosome-wide significant QTL for BW1 on CJA02 and on CJA09, and for BW2 on CJA14. Significant QTL were found for feed intake (FI) on CJA01 and for FI and residual feed intake (RFI) on CJA20. Genome-wide significant or suggestive QTL for egg production characters were obtained, on CJA01 for clutch length (CL), and on CJA06 for total egg number (EN) and age at first egg (AFE). There were QTL for eggshell weight (SW) on CJA01 and CJA20, and some evidence for another one on CJA05. One genome-wide suggestive QTL was found for tonic immobility (TI) on CJA01, and there was marginal evidence of QTL for egg weight (EW) on CJA06, and for body temperature (BT) on CJA01 and CJA03. In most cases, the genetic effect was additive and negative, indicating that the QTL allele with the lowest additive effect originated from Line DD. The few exceptions were for QTL on CJA06 and CJA09. Among genome-wide suggestive and significant QTL, dominance was only significant for CL and SW. The reduction of the residual variance due to fitting a QTL was moderate and did not exceed 5.7%. Chromosome CJA01 ---------------- Interval mapping results for CJA01 are shown in Table [2](#T2){ref-type="table"} and Figure [1](#F1){ref-type="fig"}. For all traits, the allele received from the Line DD ancestry decreased the additive genotypic value. The QTL for BW1, BW2 and FI were found at very close positions, the QTL for TI was obtained in the middle of the interval near the chicken microsatellite ADL0037, and the other QTL were placed at both ends of the chromosome. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Interval mapping of QTL on chromosome CJA01.Only traits with a QTL detected on CJA01 are shown. Significance thresholds varied little between traits, so unique genome-wide suggestive (p \< 0.10), significant (p \< 0.05) and very significant (p \< 0.01) thresholds are indicated by horizontal dashed or continuous lines. The positions of the markers are indicated by red stars. ::: ![](1471-2164-6-87-1) ::: Chromosome CJA06 ---------------- Interval mapping results for CJA06 are shown in Table [2](#T2){ref-type="table"} and Figure [2](#F2){ref-type="fig"}. The QTL for EN and AFE were found in the middle of the chromosome at very close positions. Both alleles received from the Line DD were favourable for egg production, with a positive additive effect on EN and a negative one on AFE. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Interval mapping of QTL on chromosome CJA06.Only traits with a QTL detected on CJA06 are shown. Significance thresholds varied little between traits, so unique genome-wide suggestive (p \< 0.10) and significant (p \< 0.05) thresholds are indicated by horizontal dashed lines. The positions of the markers are indicated by red stars. ::: ![](1471-2164-6-87-2) ::: Chromosome CJA20 ---------------- Interval mapping results for CJA20 are shown in Table [2](#T2){ref-type="table"}. In this short linkage group, the QTL for FI and RFI were found at opposite ends but, for both traits, the allele decreasing the additive value was transmitted by the Line DD. Imprinting and two-QTL alternative ---------------------------------- Imprinting (parent-of-origin effect) was estimated for genome-wide significant QTL. It was only found to be significant for BW1 (F~s~= 9.0 \> F~.01\ \[1,415\]~= 6.7) and FI (F~s~= 7.6 \> F~.01\ \[1,318\]~= 6.7), on CJA01, with a larger effect (5.7 ± 1.4 g and 0.73 ± 0.25 g, respectively) for sire-transmitted alleles. Similarly, the two-QTL hypothesis was tested, but there was no evidence for two different QTL located on the same linkage group and acting simultaneously on the same trait. Discussion ========== Method ------ The line-cross model used for QTL detection assumes that different QTL were fixed in the Lines DD and LTI \[[@B12]\]. This hypothesis could not be checked, but bandsharing (BS) between the two quail origins was low (0.19) \[[@B13]\] and similar to that between highly divergent poultry lines of broilers and layers \[[@B14]\], suggesting that the departure from the ideal situation of fixation in the parental lines was limited. Moreover, the main consequence of starting from outbred F0 lines would only be to underestimate the QTL effects \[[@B15]\]. This might partly explain the relatively small reduction of residual variance observed (see Table [2](#T2){ref-type="table"}) for significant QTL under the full model. Using a more general approach with full/half sib models \[[@B16]\] was also possible, but it would have required larger full-sib families \[[@B17]\] in order to accurately estimate the larger number of parameters needed with those models. QTL effects ----------- The additive effects listed in Table [2](#T2){ref-type="table"} estimated the differences between Line DD and Line LTI alleles. For body weight and feed consumption QTL, the negative sign of the additive effects indicated that favourable alleles (increasing body weights and decreasing FI and RFI) originated from Line DD, in agreement with the results of the zootechnical comparison between the two lines at the F0 generation of the present experiment \[[@B18]\]. The situation was similar for the QTL found for EN (positive difference) AFE (negative difference) and SW (negative difference), since Line DD was a better laying line but with lower egg (and eggshell) weight. Surprisingly, however, the best allele for the very significant QTL found for CL was transmitted by Line LTI, although its average clutch length was 1.1 d shorter than that of Line DD in the F0 generation \[[@B18]\]. Of course, an inferior line may possess superior alleles as first evidenced in early tomato QTL studies \[[@B19]\]. Also, this QTL had a larger dominance effect (see Table [2](#T2){ref-type="table"}), which could not be forecasted from information on the pure lines. It is also possible, however, that its localisation at one end of the linkage group and its effect were some artefact related to the significant segregation distortion observed at this position. The QTL allele originating from the LTI line selected for extreme duration of tonic immobility \[[@B11]\] also increased TI in the F2 population. Its position on CJA01 confirmed the result obtained for TI in another F2 experiment on the Japanese quail carried out to detect stress-related traits using AFLP markers \[[@B9]\]. To our knowledge, the location of a significant QTL for residual feed intake in layers is the first one in a poultry species. The QTL was located on a microchromosome with few markers, however, and it will need to be confirmed in independent genome scans, but it is already an interesting starting point for further work. Finally, it was not surprising that the QTL for BT were only chromosome-wide significant in this medium-sized experiment because body temperature is highly regulated and shows very little variation (see Table [1](#T1){ref-type="table"}), but quantitative genetic variation for BT has already been reported in the chicken \[[@B20]\], and confirmation of these QTL would represent a very significant finding. Co-location of QTL ------------------ It is likely that the QTL for BW2, BW1 and FI, located at 18 and 19 cM on CJA01, represent a single gene. Indeed, body weight and feed intake are known to be highly correlated genetically (0.5 to 0.9) in poultry \[[@B21]\], and the rank correlation between BW1 and FI was 0.5 (p \< 0.01) in the F2. It is worth noting, however, that the suggestive QTL found for TI on the same chromosome was not placed near any other significant QTL. This observation was in good agreement with the absence of association between TI and all the other traits reported from factorial correspondence analysis of the same F2 data \[[@B22]\]. Therefore, the association between stress susceptibility and production in the quail, if it exists, is not mediated significantly by major genes involved in the determinism of the duration of tonic immobility measured at a young age. The only other instance of close location of two QTL was for EN and AFE at 32 and 34 cM on CJA06, but the rank correlation between them was only -0.1 (p \< 0.05) in the F2, which does not strongly suggest that there might be also a single pleiotropic QTL at this position. Unfortunately, pleiotropy could not be tested with the software presently available. Comparison with chicken QTL --------------------------- Several recent reports of genome-wide scans in the chicken have described QTL for body weight \[[@B23]-[@B30]\], feed intake \[[@B23],[@B25],[@B27]\], egg production traits \[[@B25],[@B27],[@B28]\] and tonic immobility \[[@B31]\], but none was found for body temperature and residual feed intake in layers. A QTL was found for RFI in broilers on GGA01, GGA04 and GGA05, however, but in studies \[[@B32],[@B33]\] which were focused on some specific chromosomal regions. Significant QTL for body weight in the chicken were only reported repeatedly on GGA01 \[[@B23],[@B24],[@B26],[@B27],[@B30]\], homologous to CJA01, and on GGA04 \[[@B25],[@B28],[@B32]\]. Moreover, in two out of the three chicken studies where both BW and FI were measured, QTL were found at the same position for body weight and feed intake on GGA01 \[[@B23],[@B27]\], as on CJA01 in the present work. In the third report \[[@B25]\], despite the varied nature of the traits which were studied, no QTL was found at all on GGA01, but QTL which affected BW and TI were detected in the same region of GGA04. In our work, however, no QTL was found on CJA04. Overall, then, the largest chicken and quail chromosomes might contain an important and homologous QTL for growth. Regarding QTL for egg production in layers, none were found on GGA06 which is homologous to CJA06 on which QTL for EN and AFE have been detected in this work. But egg production is a complex trait because egg laying evolves with time, and it was only measured over a limited period in the works on the chicken, so the characters were not directly comparable. The location of a QTL for tonic immobility on two homologous chromosomes in chicken \[[@B31]\] and quail suggests that the duration of tonic immobility, a trait associated to fear response \[[@B11]\], is, at least partly, under a similar genetic determinism in the two species. The associations found between QTL for TI and QTL for growth and egg weight on GGA01 in the red jungle fowl layer intercross \[[@B31]\], however, contradict the lack of association reported in our quail QTL experiment in the present paper, and reported in previous phenotypic analyses \[[@B18],[@B22]\]. Some parent-of-origin effects were recently reported in the chicken \[[@B34]\]. They were detected mainly on GGA01, and some were found for QTL for body weight and feed intake, as in the present study. It is rather unlikely that these converging results obtained independently in two species from the family Phasianidae were only coincidental. Consequently, they should raise some interest in looking further for this type of effect in avian species (where reciprocal effects are common), and, more generally, for possible imprinting-like mechanisms in Birds \[[@B34]\]. Possible candidate genes ------------------------ Of course, only chromosomal regions could be detected in this study, but it is interesting to note that in the chicken, the gene for the insulin-like growth factor IA associated with growth \[[@B35]\], and the gene for a receptor of serotonin which might be implicated in behaviour \[[@B36]\] are on GGA01 <http://www.ncbi.nlm.nih.gov/mapview/>, whereas we found a QTL for body weight and one for tonic immobility on the homologous quail chromosome CJA01. In the same way, the agouti gene is linked to feed intake \[[@B37]\], and a QTL for feed intake was detected on CJA20 where an agouti-like locus was mapped in the Japanese quail \[[@B8]\]. These chromosomal co-locations are only indicative, but they might deserve to be explored further. Conclusion ========== For the first time with the Japanese quail, a partial genome scan using microsatellites has revealed genome-wide significant QTL for major production-related traits (BW, FI, CL, EN), and QTL for other traits with wider potential interest (TI, RFI, BT) were also detected. Few results for traits measured on both the quail and chicken are available, and comparisons between the two species are only at the beginning, but the common co-location (CJA01 and GGA01) of QTL acting on BW and FI, and of QTL for TI are already indicative of the detailed genetic similarity between the two species that future comparative work should further explore. Of course, only chromosomal regions could be detected in this study, but it is interesting to note that in the chicken, the gene for the insulin-like growth factor associated with growth and the gene for the serotonin 1F receptor implicated in behavioral traits are on GGA01, whereas a QTL for body weight and one for tonic immobility were found on CJA01. In the same way, the agouti gene is linked to feed intake, and a QTL for feed intake was detected on CJA20 where an agouti-like locus was just mapped in the Japanese quail. Methods ======= Birds and Husbandry ------------------- Two lines, LTI and DD, with different origins \[[@B13]\] and selected respectively for high duration of tonic immobility \[[@B11]\] and for early egg production \[[@B10]\] were crossed reciprocally (12 single-pair matings) to produce F1 quail. All LTI breeders had a maximum duration (300 s) of experimental tonic immobility, and a maximum value for TI was 55 s in Line DD. Ten males and 30 egg-laying females were drawn randomly across F1 families to produce the F2 generation. The duration of TI in the F1 breeding group varied between 31 and 300 s. Each F1 sire was mated to three full sisters from another F1 family, and 30 F2 full-sib families were produced in three consecutive hatches to obtain the 434 female quail studied in the present work, with 39 to 45 F2 birds per sire family and 12 to 19 quail per full-sib F2 family. F2 quail were successively given standard starter and commercial layer diets \[[@B10]\]. At 5 wk of age they were assigned at random to individual cages of a 4-tier battery maintained at 25°C, and in which they remained until the end of the experiment, with free access to feed and drinking water. Traits ------ The traits will be described chronologically. The duration of TI was measured for up to 5 min at 8 or 9 days of age. TI was the length of time during which a chick remained immobile after the freezing reaction had been induced by keeping it gently on its back for 10 s. Body weight (BW1) and rectal body temperature (BT) were measured at 5 weeks of age. Egg production was recorded daily and individually from 5 to 69 weeks of age. Age at first egg (AFE), total egg number (EN) and clutch length (CL: average number of consecutive days with an egg) were obtained from the egg laying data. A 24 day feed trial was conducted on 30-wk females fed ad libitum. Individual daily intake was recorded, and total residual feed intake was estimated for each bird as the residual of the multiple regression of total feed intake on mean metabolic body weight, body weight gain and egg mass produced on test \[[@B22]\]. Observed and residual intake values were then divided by 24 to obtain FI and RFI. RFI is a relative value which represents the daily amount of feed used for non-productive traits (activity, basal metabolism, heat increment). It is an important character for production purposes but also for more fundamental metabolic studies. Egg weight (EW) and gross composition (YW: yolk weight, SW: eggshell weight; Y/A: ratio of yolk weight over albumen weight) were obtained from three consecutive eggs per quail collected around 30 weeks of age for each quail. Finally, body weight was measured again at 70 weeks of age (BW2). Statistical analyses were run with untransformed data for all characters, except for tonic immobility which was also analysed as log(1+TI), and produced the same results as for TI. Genotyping ---------- All 24 F0, 40 F1 and 434 F2 quail in the present experiment belonged to the resource family set up to establish the first microsatellite quail map \[[@B7]\], and they had all been typed previously for the microsatellites listed in the augmented microsatellite panel \[[@B5]\]. Genotyping was carried out at the Gifu University, and genotypic data were arranged, validated and stored in the MAPGENA database. Mean heterozygosity of the 10 F1 sires was 61%. The map had been built from 72 microsatellite loci, and 58 of them had been resolved into 13 linkage groups, including a Z chromosome group, for a total map distance of 576 cM with a 10 cM average spacing between loci \[[@B7]\]. At that time, only seven of the linkage groups could be assigned to chromosomes CJA01, CJA02, CJA05, CJA06, CJA14, CJA27 and CJAZ through comparative mapping with the chicken. Since then, however, all other linkage groups in the microsatellite map were assigned to other chromosomes \[Kayang et al., unpublished data; \[[@B8]\]\]. Consequently, the remaining linkage groups Q03, Q04, Q08, Q09, Q10 and Q11 in the original article \[[@B7]\] have been renamed respectively, CJA03, CJA13, CJA09, CJA04, CJA20 and CJA10 in the present one. QTL analyses ------------ QTL detection was performed using the line cross method \[[@B12]\] implemented in the QTL Express software \[[@B38]\]. Briefly, for each trait independently, an analysis of regression was carried out along each autosome to test for the effect of a putative QTL. In the present study, the linear model also included a hatch effect (3 classes) for all traits, and a technician effect (2 classes) for TI, so an analysis of covariance was carried out. The most likely position of the QTL corresponded to that found with the maximum F value. Five thousand permutations \[[@B39]\] were then carried out to set significance levels (p~c~) for the most likely chromosome-wide QTL. Finally, genome-wide significance (p~g~) for a QTL detected on a given chromosome was obtained from p~c~by: p~g~= 1-(1-p~c~)^1/r^, where r was the ratio of the length of this chromosome over the total chromosome length (545 cM) spanned by the present study. Chromosome-wide significance was set up conservatively at p~c~= 0.05. Genome-wide significant and suggestive thresholds were set up respectively at p~g~= 0.05 and 0.10. A significant (p \< 0.05) segregation distortion was found only near the 0 position on CJA01. Authors\' contributions ======================= FM designed and coordinated the study, and wrote the paper. BBK, SI, MIM and MM designed, organized, and conducted all the microsatellite work. AV led the mapping part. AN was responsible for the data bank, and DG and JLM supervised and carried out the data collection. Acknowledgements ================ The authors dedicate this article to the memory of Dr Andrew Mills who developed the LTI Japanese quail line with JM Faure at INRA in Nouzilly (France). His determination and devotion to research have produced a unique bird model that we dearly thank him for. The authors wish to thank JM Faure (INRA-SRA, Nouzilly) for kindly providing F0 birds of Line LTI. Expert care of experimental F0, F1 and F2 quail, and data collection by C Moussu and the team at the INRA Unité Expérimentale de Génétique Avicole (Nouzilly, France) are gratefully acknowledged. The linguistic revision of the paper was kindly carried out by W Brand-Williams.	PubMed Central	2024-06-05T03:55:59.438654	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180434/", "journal": "BMC Genomics. 2005 Jun 8; 6:87", "authors": [ { "first": "Francis", "last": "Minvielle" }, { "first": "Boniface B", "last": "Kayang" }, { "first": "Miho", "last": "Inoue-Murayama" }, { "first": "Mitsuru", "last": "Miwa" }, { "first": "Alain", "last": "Vignal" }, { "first": "David", "last": "Gourichon" }, { "first": "André", "last": "Neau" }, { "first": "Jean-Louis", "last": "Monvoisin" }, { "first": "Shin'ichi", "last": "Ito" } ] }
PMC1180435	Background ========== MicroRNAs (miRNAs) are an abundant evolutionarily conserved class of small (\~22 nts) RNAs which play a substantial gene regulatory role in plants and animals \[[@B1]\]. The first miRNA discovered, lin-4, was identified in a genetic screen focused on identifying genes involved in the heterochronic pathway in C. elegans \[[@B2]\]. The 22 nt lin-4transcript temporally negatively regulates translation of lin-14, apparently through antisense RNA-RNA interaction between the lin-4transcript and multiple regions in the lin-143\' UTR. Seven years later a second small RNA, let-7, was found, and it too acts in the heterochronic pathway in C. elegans \[[@B3]\]. let-7represses translation of lin-41in a temporally dependent manner, also through targeting complementary regions in the 3\' UTR of the regulated gene \[[@B4]\]. let-7transcripts are found in all bilaterians tested \[[@B5]\]. This discovery led to the understanding that miRNA-mediated regulation may be a general phenomenon. Several hundred miRNAs have since been identified in a variety of plants and animals through cloning and computational methods, including 78 miRNAs in Drosophila \[[@B6]\]. Many of these miRNAs are expressed in a temporal or tissue-specific dependent manner \[[@B1]\]. miRNAs in animals usually act to repress translation of their target genes through imperfect hybridization to complementary sites in target 3\' UTRs \[[@B2],[@B7],[@B8]\]. This translational repression occurs post-initiation, since miRNA-induced gene silencing does not change the abundance or polysome profile of target mRNA, at least in the examples tested \[[@B9],[@B10]\]. This is in contrast to RNAi, in which short RNAs called siRNAs are usually perfectly complementary to their target mRNA and result in its degradation \[[@B11]-[@B14]\]. A miRNA directed against a perfectly complementary 3\' UTR target site also results in mRNA degradation \[[@B15],[@B16]\], indicating that a miRNA can function in the RNAi pathway given a perfectly complementary target site. miRNAs are produced from a larger transcript through stepwise processing by ribonuclease III-like endonucleases in the nucleus and cytoplasm \[[@B17]-[@B19]\]. Following maturation, miRNAs reside in a miRNA ribonucleoprotein complex (miRNP) which shares many similarities to the RNA-induced silencing complex (RISC) involved in RNAi \[[@B1],[@B16],[@B20]-[@B22]\]. While many animal miRNAs have been identified, only a few have a known function or target \[[@B2],[@B23]-[@B28]\]. Incomplete base pairing of miRNA to target causes inherent difficulty in the prediction of miRNA targets due to the high levels of noise involved in any simple alignment of miRNAs to 3\' UTRs. In addition, the very few experimentally derived miRNA/target pairs provide limited biological information needed to define the necessary and sufficient characteristics for a miRNA/target pair. Therefore, miRNA target prediction programs for which the selection parameters can easily be adjusted based on current interpretation of miRNA/target constraints, and on newly discovered rules governing miRNA/target interactions, are a valuable resource to the research community. Implementation ============== Our bioinformatics approach to identifying miRNA targets includes two steps: the creation of a database of potential targets, and screening all possible miRNA/target pairs for adherence to constraints suggested by analysis of the known miRNA/target interactions. Potential miRNA target database ------------------------------- The selection of sequences for the database was guided by progress in understanding the actions and features of miRNAs and their targets. All known animal miRNAs appear to target regions in the 3\' UTRs of mRNAs \[[@B2],[@B23]-[@B28]\], and so the database was limited to 3\' UTR sequences. miRNAs are highly conserved, particularly in closely related species \[[@B1],[@B5]\], and 68 of the 78 known D. melanogaster miRNAs are identical to their predicted counterparts from D. pseudoobscura. The constraint on miRNA evolution is thought to be a consequence of their interaction with multiple targets, thereby restricting the rate of change of both the miRNA and its targets \[[@B29]\]. In addition, miRNA targets that have been experimentally derived are very highly conserved in other species \[[@B1],[@B5],[@B7],[@B24],[@B27]\]. Therefore, the database was further limited to highly conserved regions of 3\' UTRs. Finally, the majority of experimentally derived miRNA targets contain multiple predicted target sites\[[@B1],[@B2],[@B23]-[@B28]\]. Our approach to allow detection of multiple targets in a single 3\' UTR was to fragment the database sequences into segments no longer than 50 nt, each of which is tested for target sites. For conserved 3\' UTR segments longer than 50 nt, overlapping 50 nt segments with end points differing by 5 nt were added to the database (i.e. a 100 nt sequence would be fragmented into 11 overlapping 50 nt segments). We chose 50 nt for the maximum target site length since all predicted target sites of known miRNA targets are less than this size \[[@B2],[@B23]-[@B28]\]. The Berkeley Genome Pipeline \[[@B30],[@B31]\] was accessed to obtain all 3\' UTR sequences which are at least 80% conserved between D. melanogaster and D. pseudoobscura, and are at least 12 nt long \[the smallest predicted target site of known miRNA targets that we are aware of is 13 nts in length \[[@B2]\]\]. The D. pseudoobscura genome sequence is largely known, but the last stages of sequence finishing (to allow assembly of the final sequence from the large number of shorter contigs) and annotation are not complete. Thus some regions of the D. pseudoobscura sequence that are highly conserved with D. melanogaster 3\' UTRs have not been definitively linked to a coding region. Nevertheless, these conserved sequences were included to ensure that the largest fraction of potential miRNA target sites would be evaluated in the prediction process. Using D. melanogaster genome annotation release 3.1 (BDGP) we obtained 14,287 annotated 3\' UTRs. Of these, 6702 contained unique segments, at least 12 nt in length and 80% conserved with D. pseudoobscura DNA, that were included in the database. The 6702 database entries correspond to 6399 different genes, with some genes represented more than once because of alternate splicing. Biological miRNA target constraints ----------------------------------- The MovingTargets algorithm applies a set of five biological constraints to all possible alignments of each miRNA with the miRNA target database sequences, producing a set of predicted targets. The user sets values for the constraints. This adjustable algorithm facilitates focused searches with individual mRNAs, where experimental evidence may suggest that miRNA-dependent regulation exists. 1\) Number of target sites in the mRNA. For most of the known miRNA/target mRNA pairs the miRNA is predicted or known to interact with multiple sites within the 3\' UTR \[[@B23]-[@B28]\]. Furthermore, there is experimental evidence of a synergistic effect between multiple miRNP complexes associated with a single mRNA \[[@B21]\], suggesting that multiple target sites may allow for rapid translational control \[[@B32]\]. 2\) Strength of miRNA-mRNA hybridization. The specific interaction of a miRNA with a target mRNA involves base pairing \[[@B4]\], and it is reasonable to assume that target site occupancy will be positively correlated with the strength of the base pairing \[[@B22]-[@B28]\]. We therefore rank potential miRNA/target interactions according to the strength of hybridization between the miRNA and its target site, as measured by the predicted free energy of binding. These predictions are made using M. Zuker\'s DINAMelt Server software \[[@B33]\] which was expressly designed for evaluating the interactions of short RNAs and thus offers advantages over the commonly used alternatives, mFold \[[@B34]\] and RNAfold \[[@B35]\]. 3\) Number of consecutive base pairs involving the 5\' part of the miRNA. There is suggestive evidence that miRNA/target interactions require a series of consecutive base pairs between the 5\' part of the miRNA and the target \[[@B7],[@B23]-[@B28],[@B36]-[@B38]\]. Of the experimentally validated animal miRNA targets \[[@B2],[@B23]-[@B28]\], 19 of the 24 predicted miRNA/mRNA interactions have 6 or more consecutive base pairs within the first 8 nucleotides of the miRNA; 10 of these interactions have perfect complementarity in this region. This is contrasted with only 5 of the 24 predicted miRNA/mRNA interactions having 6 or more consecutive base pairs at the miRNA 3\' end. Note that for almost all examples of mRNAs known to be regulated by miRNAs, the specific target sites in the mRNA (identified as regions with significant complementarity to the miRNA) have not been individually tested and verified. Additional evidence comes from mutational analysis of miRNAs and their targets. The ability of miR-30 to repress translation of an artificial target in cultured human cells is eliminated by a mutation in the target mRNA that disrupts a single base pair in the middle of the 5\' region of the miRNA, while a mutation in the target mRNA disrupting base pairing in the 3\' part of the miRNA retains about 60% of the repressive activity \[[@B22]\]. We used the DINAMelt Server to predict the effect of both mutations on hybridization strength, and found that the inactivating mutation had a more modest effect than the weak mutation. These results argue that the important aspect of the interaction disrupted by the first mutation was the consecutive series of base pairs at the 5\' end of the miRNA, rather than the strength of the interaction as measured by thermodynamic stability considerations alone. A mutation in the 5\' region of the let-7miRNA eliminates repression of lin-41mRNA in vivo, but also reduces the level of the mature miRNA, making it difficult to conclude why it is ineffective \[[@B3]\]. 4\) Total number of miRNA 5\' nucleotides involved in base pairing to the target. For mRNAs shown to be miRNA targets, all of the 24 predicted miRNA/mRNA interactions have 6 or more total base pairs within the first 8 nucleotides of the miRNA; 21 of these interactions have at least 7 base pairs. This is contrasted with only 11 of the 24 predicted miRNA/mRNA interactions having 6 or more total base pairs at the miRNA 3\' end \[[@B23]-[@B28]\]. In addition, there is more stringent sequence conservation in the 5\' end of homologous miRNAs than in the 3\' end \[[@B36]\]. 5\) Number of nucleotides in the miRNA 5\' region involved in G:U base pairs. Predicted miRNA/target interactions of known miRNA targets have at most one (6 out of the 24 predicted miRNA/mRNA interactions) and usually no G:U base pairs in the miRNA 5\' region. In contrast, despite having fewer overall base pairs in the miRNA 3\' region, 9 out of the 24 predicted miRNA/target interactions have more than 1 G:U base pair in the miRNA 3\' region, and 8 of the 24 have 1 G:U base pair in this region \[[@B23]-[@B28]\]. Thus, canonical base pairing appears to be favored over G:U base pairing in the miRNA 5\' region. Subsequent to development of the MovingTargets algorithm an extensive study of the rules of miRNA/target interactions was published \[[@B39]\]. The results emphasize the importance of the latter three constraints described above. Methods ======= DNA constructs -------------- Reporter plasmids were constructed by cloning the 3\'UTR of each target gene into the BamHI/XbaI site of luk-ttkUTR\[[@B40]\]. luc/tramtrackis the luk-ttkUTRplasmid. For luc/CrebA, the CrebA3\'UTR was amplified by PCR from genomic DNA (all sequence coordinates are from Release 3 from FlyBase, <http://www.flybase.net>): 3L:15500154-15502103. For luc/ab, the ab3\'UTR was PCR amplified from genomic DNA: 2L:11248260-11249979. For luc/Eip74EF, the Eip74EF3\'UTR beginning at position 29 was cloned from pBSE74AcDNA, a gift from Carl Thummel. The control plasmid for monitoring transfection efficiency is MT-RLuc\[[@B40]\]. miRNA plasmids were constructed by cloning a DNA segment containing the predicted primary transcript of each miRNA into the BamHI/EcoRI site of ActMSI\[[@B40]\]. Each primary transcript was amplified by PCR from genomic DNA to generate fragments with the following sequence coordinates: let-7: 2L:18450072-18450291; mir-92b: 3R:21466427-21466673; mir-312:2R:15647675-15647897; mir-34:3R:5926642-5926792. In each case, BamHI and EcoRI sites were introduced at the 5\' and 3\' ends, respectively, by PCR. Targeting Assay --------------- S2 cells were transfected with 2 μg of microRNA plasmid (for non-control samples), 50 ng reporter plasmid, and 10 ng control plasmid. For each transfection, 0.5 mL Schneider\'s Drosophila Medium (Gibco) containing indicated plasmids and 0.5 mL Schneider\'s Drosophila Medium containing 5 μL Cellfectin (Invitrogen) were mixed gently and incubated at RT for 15--45 minutes. 2 × 10^6^cells were centrifuged for 5 minutes at 1000 g, aspirated, resuspended in the DNA-lipid mix described above, and transferred to a 35 mm well of a 6-well plate. Cells were incubated at 25°C for the remainder of the transfection. After 4--5 hours, 0.5 mL of Schneider\'s Drosophila Medium containing 30% Fetal Bovine Serum (Gibco) was added to each well. The following day, 2 mL of complete growth medium \[Schneider\'s Drosophila Medium with 10% FBS and 1% Pen/Strep (Gibco)\] was added to each well. Between 38 and 45 hours after transfection, the reporter and control plasmids were induced by adding 3.5 μL of 700 mM CuSO4. After 6--6.5 hrs, cells were harvested, lysed, and assayed for reporter and control luciferase expression using the Dual-Luciferase Reporter Assay system (Promega). Each transfection was carried out at least 5 times. Potential miRNA Target Database ------------------------------- D. melanogaster 3\'UTRs were obtained from the Berkeley Drosophila Genome Project <http://www.fruitfly.org>, Drosophila Release 3.1 Annotations. Conserved D. pseudoobscura sequences were obtained from the Berkeley Genome Pipeline using the following parameters: minimum conservation width = 1; calculation window = 20; minimum conservation = 80%. We used the 7-8-03 AVID alignment for determining conserved regions \[[@B30],[@B31]\] miRNA sequences --------------- D. melanogaster miRNA sequences were obtained from The microRNA Registry \[[@B6]\]. Results ======= In this initial use of MovingTargets we set stringent values for all of the adjustable biological constraints to produce a high likelihood set of miRNA target predictions. The following values were used: minimum of 3 target sites; maximum free energy of hybridization of -15 kcal/mole at room temperature (22°C) for each target site; minimum 7 out of 8 consecutive 5\' miRNA nt matches; maximum of 1 G:U base pair in the miRNA 5\' region. The high-likelihood set of miRNA target predictions corresponding to these strict biological constraints, generated from analysis of all 78 miRNAs and the full database of 6399 potential targets containing conserved 3\' UTR sequences, is given in Table [1](#T1){ref-type="table"}. Given the strict constraints, this group will not contain all miRNA/target pairs (and includes no predictions for a subset of the known miRNAs). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### microRNA/target pairs predicted by MovingTargets using the following strict targeting criteria: minimum 3 target sites, maximum dG of microRNA/target hybridization of -15 kcal/mole at RT for each target site, minimum 7 out of 8 consecutive base pairs in microRNA 5\' end, maximum 1 G:U base pair in microRNA 5\' region. Transcription factor targets are listed first followed by neural targets. Gene function and biological process are as given by FlyBase <http://www.flybase.net>, March 2004. ::: ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Target miRNA \# target sites dG of miRNA-mRNA hybrid (kcal/mole) Molecular Function Biological process ---------------- --------------- ----------------- ------------------------------------- ------------------------------------------------------------------------------------------------------ ------------------------------------------------------------------------------- CrebA mir-92b\ 3\ -16, -24, -19\ transcription factor salivary gland development mir-312 4 -16, -20, -17, -22 fkh mir-315 3 -18, -16, -17 transcription factor salivary gland morphogenesis Eip74EF mir-34 3 -25, -27, -21 transcription factor mesoderm development Eip93F^A^ mir-280^A^ 3 -21, -15, -16 transcription factor autophagy, induction of apoptosis by hormones pros mir-34 3 -18, -17, -18 transcription factor ectoderm development zfh1 mir-5 3 -17, -15, -15 transcription factor ectoderm development, mesoderm development zfh2 mir-276a-3 3 -16, -15, -17 transcription factor, RNA binding ectoderm development SoxN mir-34\ 3\ -23, -24, -16\ transcription factor, DNA bending ectoderm development, visual perception mir-309 3 -19, -16, -20 HLHm5 mir-7 3 -24, -24, -26 transcription factor ectoderm development, cell proliferation CG32527^B^ mir-34 3 -25, -24, -21 multiple (including transcription factor) unknown ab let-7 5 -20, -25, -17, -17, -19 transcription factor transmission of nerve impulse, sex determination sbb mir-33 3 -16, -17, -19 transcription factor axon guidance, axon target recognition, larval walking behavior nerfin-1 mir-279\ 3\ -19, -17, -24\ transcription factor neuronal lineage restriction mir-286 3 -19, -16, -27 Syn mir-92b 3 -19, -20, -15 unknown neurotransmitter secretion, synaptic vesicle exocytois synaptogyrin mir-313 3 -17, -16, -15 unknown synaptic vesicle exocytosis, regulation of calcium ion dependent exocytosis Pkc98E mir-210 3 -15, -15, -15 multiple multiple (including neural processes) nAcRbeta-96A mir-210 3 -19, -18, -23 nicotinic acetylcholine-activated cation-selective channel activity, acetylcholine receptor activity multiple (including neural processes) W^A^ bantam\ 4\ -21, -16, -27, -17\ unknown induction of apoptosis, programmed cell death mir-280^A^ 3 -17, -19, -15 kel mir-310\ 3\ -20, -19, -22\ actin binding apoptosis, ovarian ring canal formation mir-311\ 4\ -17, -21, -19, -15\ mir-312 4 -16, -18, -20, -17 RhoGEF2 mir-9c 3 -16, -16, -18 Rho guanyl-nucleotide exchange factor activity, diacylglycerol binding multiple adat mir-3\ 3\ -28, -17, -19\ tRNA specific adenosine deaminase activity purine base metabolism mir-309\ 3\ -23, -16, -15\ mir-318 3 -26, -15, -20 bru-2 mir-9a\ 3\ -19, -17, -16\ RNA binding mRNA processing, protein metabolism mir-9c 3 -19, -15, -16 CG32062^A^ mir-12\ 4\ -19, -17, -23, -15\ RNA binding unknown mir-280^A^ 3 -16, -16, -19 fus mir-303 3 -16, -15, -16 RNA binding epidermal growth factor receptor signaling pathway mbl^A^ mir-280^A^ 3 -17, -17, -21 RNA binding, DNA binding mesoderm development Asph mir-9b 3 -21, -19, -16 peptide-aspartate beta-dioxygenase activity transmembrane receptor protein tyrosine kinase signaling pathway CG32429 mir-33 3 -17, -17, -18 unknown unknown lmg mir-34 3 -27, -20, -22 unknown mitotic anaphase wb mir-34 4 -22, -22, -19, -23 binding, structural molecule activity cell-cell adhesion, cell-matrix adhesion, signal transduction Cbl mir-34 3 -22, -17, -22 ligase activity cellular defense response Pdi mir-34\ 4\ -26, -23, -30, -20\ protein disulfide isomerase activity protein folding, protein modification mir-263b\ 3\ -18, -15, -17\ mir-305\ 4\ -24, -18, -19, -22\ mir-316 3 -21, -16, -21 CG31637 mir-92b 3 -17, -18, -17 sulfotransferase activity carbohydrate metabolism CG3689 mir-210 3 -20, -15, -17 pre-mRNA splicing factor activity mRNA cleavage, nuclear mRNA splicing, via spliceosome CG8475 mir-263b 3 -18, -17, -21 kinase activator activity, phosphorylase kinase regulator activity glycogen metabolism didum mir-276a-3 3 -18, -16, -15 multiple multiple Ggamma1 mir-277 3 -19, -16, -15 heterotrimeric G-protein GTPase activity G-protein coupled receptor protein signaling pathway CG1441 mir-278 3 -30, -22, -26 oxidoreductase activity unknown Rh6 mir-278 3 -19, -22, -20 G-protein coupled photoreceptor activity phototransduction, visual perception, sensory perception CG31163^A^ mir-289^A^ 3 -17, -15, -17 SH3/SH2 adaptor protein activity unknown CG18854 mir-306-3 4 -27, -19, -24, -21 inositol-triphosphate 3-kinase activity unknown CG7908 mir-309 3 -17, -19, -22 zinc ion binding, metalloendopeptidase avtivity cell surface receptor linked signal transduction, proteolysis and petidolysis CG14507 mir-276a-5\ 3\ -24, -20, -21\ phospholipase A2 activity unknown mir-276b-5 3 -24, -20, -21 CG33085^A,B^ mir-184-3\ 3\ -21, -22, -16\ argininosuccinate lyase activity unknown mir-284^A^ 3 -23, -25, -17 CG32316^B^ mir-184-3\ 3\ -20, -22, -16\ oxoglutarate dehydrogenase (succinyl-transferring) activity tricarboxylic acid cycle mir-278 3 -18, -20, -23 CG32912^B^ mir-279 3 -15, -21, -16 peptidoglycan recognition activity immune response CG33047^A,B^ mir-133\ 3\ -19, -17, -19\ alpha-L-fucosidase activity O-glycoside catabolism, fucose metabolism mir-284^A^ 3 -27, -17, -15 CG33075^B^ mir-306-3 3 -26, -20, -27 carrier activity transport CG32956^B^ mir-9a\ 5\ -17, -19, -18, -21, -18\ multiple multiple mir-9b\ 5\ -17, -20, -16, -20, -17\ mir-9c\ 5\ -17, -22, -18, -20, -18\ mir-33 3 -19, -23, -18 CG33038^B^ mir-9a\ 4\ -20, -17, -16, -18\ multiple heparan sulfate proteoglycan biosynthesis, proton transport mir-9b\ 5\ -18, -19, -17, -16, -16\ mir-9c 4 -21, -17, -17, -19 CG32791 mir-31a 3 -15, -17, -15 unknown multiple tinc mir-34 3 -22, -23, -33 unknown unknown CG9932 mir-263b 4 -16, -21, -20, -21 unknown unknown CG30389 mir-14\ 3\ -17, -16, -25\ unknown unknown mir-34 3 -28, -21, -24 CG3975 mir-33 3 -17, -19, -22 unknown unknown CG8963 mir-6 3 -16, -17, -17 unknown unknown CG3638 mir-184-3 3 -27, -17, -22 unknown unknown CG33006^B^ mir-278 3 -20, -15, -23 unknown unknown CG31305^B^ mir-278 4 -16, -32, -29, -25 multiple unknown CG32206^A^ mir-289^A^ 3 -15, -15, -16 unknown unknown CG12071 mir-305 3 -17, -21, -19 unknown unknown ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ^A^: exact 5\' and 3\' ends for mir-263a, mir-274, mir-280, mir-282, mir-284, mir-289 have not yet been determined (6); target predictions for these microRNAs are likely to change once the precise microRNA sequences are known ^B^: gene annotated as a \"gene cassette\"; molecular function and biological process listed is the union of molecular functions and biological processes for each ORF in the putative dicistronic transcript ::: A striking feature of the set of predicted miRNA targets is the disproportionate fraction of mRNAs that encode transcription factors (13 of 41 genes with known function) or have assigned roles in neural processes (7 of 41). \[[@B41]\] noted a similar enrichment for transcription factors and neural genes in predicted miRNA targets in Drosophila. Because the different predictions identify groups of genes that are not fully overlapping, the bias is even more striking. The emphasis on transcription factors was also observed for predicted miRNA targets in plants \[[@B42]\] and mammalian cells \[[@B43]\]. Validation of predicted targets ------------------------------- We chose a subset of three predicted miRNA-target pairs for validation in a cultured cell assay. The assay is similar to that used by Zeng and Cullen \[[@B22]\]. A reporter plasmid expresses, under control of the inducible metallothionein promoter \[[@B44]\], a hybrid mRNA in which the firefly luciferase coding region is fused to the candidate target 3\' UTR (luc/target). The miRNA plasmid expresses, under control of the constitutive actin promoter, a genomic DNA segment that contains the miRNA primary transcript. Finally, a control plasmid expresses Renilla luciferase under control of the inducible metallothionein promoter, and serves to monitor transfection efficiency. In parallel transfections one population of S2 cells receives all of the plasmids, while a second population receives the reporter and control, but not the miRNA plasmid. Approximately 1.5 days after transfection (to allow the miRNA to accumulate), transcription of the reporter and control mRNA is induced by addition of CuSO4 to the growth medium. After an additional 6 hours, the cells are harvested and the levels of firefly and Renilla luciferase are measured. The let-7miRNA is predicted to target the abrupt(ab) mRNA at five positions in the 3\' UTR (Table [1](#T1){ref-type="table"}, Fig. [1A](#F1){ref-type="fig"}). The level of luciferase activity from the luc/abmRNA in the absence of exogenous miRNA provides the standard for measurement of the effect of the miRNA. When let-7is coexpressed with the luc/abreporter, the level of luciferase activity is substantially reduced (Fig. [1B](#F1){ref-type="fig"}). To confirm that expression of the luciferase reporter in this system is not simply sensitive to the coexpresssion of any miRNA, the experiment was also performed using miR-92binstead of let-7. Using the parameters noted above, MovingTargets does not predict mir-92bto target ab. Exogenous miR-92bcan repress expression in the assay system of a reporter mRNA bearing the 3\' UTR of a predicted miR-92btarget (below), but has only a small effect on luc/abexpression, much less than the effect of let-7(Fig. [1B](#F1){ref-type="fig"}). We conclude that let-7is specifically targeting the ab3\' UTR. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Drosophila let-7miRNA targets the abrupt3\' UTR. A. Predicted sites of let-7interaction with the ab3\' UTR. The schematic at top shows the relative positions of the sites as vertical bars in the ab3\' UTR. The predicted pairings are shown below, with the free energies (in kcal/mol) and exact positions in the ab3\' UTR indicated. B. Luciferase expression in transfected S2 cells. Expression from the luciferase/abmRNA with no added miRNA is shown at left, set to a relative value of 1. The other bars indicate the results of coexpression with let-7or miR-92b. In this figure and in Figure 2, all values represent the average luciferase expression from at least 5 experiments, and error bars represent standard deviation. ::: ![](1471-2164-6-88-1) ::: Two additional predicted targets were tested for miRNA-dependent regulation. The CrebAmRNA is predicted to have three target sites for miR-92band mir-312(Table [1](#T1){ref-type="table"}); these miRNAs are closely related, sharing the same nts in positions 2--9 and 15--21. Both miRNAs repress expression of the luc/CrebAreporter (Figure [2B](#F2){ref-type="fig"}). The Eip74EFmRNA is predicted to have three target sites for miR-34(Table [1](#T1){ref-type="table"}; Figure [2A](#F2){ref-type="fig"}), and miR-34represses expression of the luc/Eip74EFreporter (Figure [2B](#F2){ref-type="fig"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### miRNA-dependent regulation of CrebA, Eip74and ttk. Each panel shows at top the distribution of the miRNA target sites in the CrebA(A), Eip74(B) and ttk(C) 3\' UTRs. For panels A and C, the vertical bars above the line indicate sites for miR-92b, while vertical bars below the line indicate sites for miR-312. Luciferase expression data are presented as in Fig. 1. ::: ![](1471-2164-6-88-2) ::: Flexible MovingTargets search to identify potential miRNAs targeting specific genes ----------------------------------------------------------------------------------- In addition to validating a subset of our predicted targets, we wondered if we could use the flexibility of MovingTargets to identify a miRNA that regulates a gene known to be under another form of translational control. One such gene is tramtrack(ttk), which encodes a transcription factor that determines non-neuronal identity in developing sensory organ cells, and which has been shown to be translationally repressed by the RNA-binding protein Musashi \[[@B40]\]. ttkis not a predicted miRNA target using the strict biological constraints described above. However, by relaxing these constraints to require a maximum free energy of miRNA/target hybridization of -12 kcal/mole at room temperature and a minimum of 6 out of 8 consecutive base pairs in the miRNA 5\' end, MovingTargets predicts that mir-9c, mir-92a, mir-92b, and mir-312(the latter three are closely related miRNAs) target ttk. We tested mir-92band mir-312in the S2 cell assay, and both miRNAs repress expression of the luc/ttkreporter (Figure [2C](#F2){ref-type="fig"}). Discussion ========== The nature of the interaction between a miRNA and its target -- incomplete and interrupted base pairing -- creates a substantial challenge for the prediction of candidate miRNA targets. Furthermore, very few mRNAs have been shown to be under miRNA regulation, limiting the number of examples from which the basic rules governing miRNA/target interactions can be determined. The latter problem is particularly acute, since even in mRNAs known to be regulated, the actual target sites are usually only inferred from their partial complementarity to the miRNA. Thus a precise description of these rules is a moving target, and will undoubtedly be refined as additional targets are identified by methods not biased by current prediction strategies. Our approach to predicting miRNA targets addresses both of these difficulties. The combination of a conserved 3\' UTR database and the MovingTargets algorithm allowed us to predict 83 miRNA targets that meet stringent biological constraints based on features of the probable or proven interactions between individual miRNAs and the mRNAs under their control. Each of three target predictions chosen for testing was verified in the S2 cell transfection assay. Thus the algorithm succeeds in predicting miRNA targets. At present we have restricted the database of potential target sequences to those from 3\' UTRs, but this could be expanded to include entire mRNAs given evidence that miRNAs bind to other regions of animal mRNAs. The biological relevance of a predicted miRNA/target interaction depends on whether the miRNA and target are expressed at appropriate concentrations in a particular cell type. Thus, this and all other prediction methods represent only the first step, albeit an important one, in identifying bona fide miRNA targets. The MovingTargets software allows individual researchers to specify which constraints the software should enforce. Dramatically different predictions will result by adjusting the parameter values used here. This capacity of the software has two notable benefits. First, it provides the means to adjust the parameter values as the rules of miRNA/target interactions become better understood. Second, the adjustability of the algorithm facilitates less constrained searches that focus on a particular mRNA or miRNA. For example, if experimental evidence suggests that an mRNA is regulated by miRNAs, yet it is not among the group predicted using the stringent screening parameters, then relaxing different parameters one by one would produce a set of candidate miRNA regulators. Here the drawback of increased sensitivity -- an increased number of predictions with an unknown fraction presumed to be false -- would be acceptable. We were able to identify miRNAs targeting ttkusing this strategy. What is the minimal quality of miRNA/target site interaction sufficient for regulation? --------------------------------------------------------------------------------------- In the initial examples of miRNA-dependent regulation, the miRNAs very efficiently blocked accumulation of the proteins encoded by the target mRNAs. These dramatic effects could be the normal mode of miRNA-dependent regulation: when some quality of miRNA/mRNA target interaction occurs, then protein accumulation is blocked. Alternatively, the degree of regulation could be directly correlated with the quality of the individual interactions (or their sum), and several mutation studies have shown that the level of translational repression of a miRNA target varies in relation to the quality of miRNA targeting, as defined by such characteristics as free energy of hybridization, consecutive 5\' base pairing, and number of target sites \[[@B3],[@B22],[@B39]\]. In our experiments to validate the predicted miRNA targets we tested a miRNA, miR-92b, not predicted to interact with the ab3\' UTR. Although miR-92bdid not produce the strong inhibitory effect of let-7, it did cause some inhibition. One interpretation is that miR-92bweakly interacts with the ab3\' UTR, and that the weak interaction is sufficient for a weak regulatory effect. When the \'number of target sites\' parameter of MovingTargets was relaxed to two, rather than three, we found that miR-92bis predicted to target abat two sites that satisfy the remaining strict biological constraints. Thus our results are also consistent with the notion that there is a correlation between the overall strength of miRNA/mRNA interactions and the degree of regulation. An implication of this conclusion is that even weak interactions may have consequences, and that many or even most of all cellular mRNAs may be regulated, but that the degree of regulation may vary substantially. Comparison of different prediction results ------------------------------------------ The high-likelihood miRNA targets predicted in this paper using strict biological constraints produce very different results in comparison to the other Drosophila miRNA target prediction algorithms. The simplest forms of comparisons are not possible, because the other predictions produce a ranked list of predicted targets for each miRNA, while our method identifies a group of miRNA targets that adhere to specific constraints. Nevertheless, when considering our group of 83 predicted miRNA targets, only 11 are included among the top 20 targets for any individual miRNA predicted by Enright et al. \[[@B41]\], only 13 appear in the top 50 targets for individual miRNAs predicted by Stark et al. \[[@B45]\], and only 4 of the 83 are predicted by Rehmsmeier et al. \[[@B46]\]. There are several possible reasons for the differences in the predictions by the different algorithms. First, only our approach goes beyond thermodynamic stability considerations in imposing a penalty on G:U base pairs involving the 5\' part of the miRNA. Second, the value of multiple target sites is treated differently (there is experimental evidence of synergy between multiple target sites and most known miRNA targets have multiple predicted target sites \[[@B2],[@B21],[@B23]-[@B28]\]. For example, Enright et al. reward for multiple sites by summing a score for all complementary sites in the target 3\' UTR, whereas our algorithm requires a miRNA/target pair to have a user-specified absolute number of target sites each meeting a user-defined set of biological constraints. A third difference centers on the importance of extensive base pairing in the miRNA 5\' region, for which there is both experimental evidence and the precedent of the predicted interactions between known target mRNAs and their miRNAs \[[@B2],[@B23]-[@B28]\]. Our algorithm requires a miRNA-target interaction to have a user-specified minimum absolute number of consecutive and total base pairs in the miRNA 5\' region. In contrast, Enright et. al. appear not to heavily weight this feature, since many of their top-rated miRNA/target interactions have significant gaps in the miRNA 5\' region. Five other miRNA target prediction methods for animals have been published, but the predictions cannot be compared directly to ours since four of the five examined mammalian mRNAs and the other tested only a small number of Drosophila genes \[[@B38],[@B43],[@B47]-[@B49]\]. None of these approaches is identical to ours, and so if used with Drosophila mRNAs and miRNAs, each would be expected to provide results not identical with ours, just as for the published examples. Conclusion ========== Prediction of animal miRNA targets is a challenging task due to the incomplete and interrupted base pairing between a miRNA and its target. We developed the MovingTargets software program to provide a tool for the accurate and flexible prediction of miRNA targets in Drosophila. Using this tool, we identified a set of 83 high-likelihood miRNA targets. We tested and verified 3 of these predictions, including a target for the Drosophila let-7homolog. MovingTargets provides flexibility in describing the characteristics defining a miRNA target. Thus, as the rules governing miRNA-target interactions are better elucidated, these constraints can be enforced through MovingTargets to produce more refined sets of miRNA target predictions. We used this flexibility to relax the constraints placed on a miRNA-target interaction to predict and validate miRNAs targeting tramtrack. MovingTargets is freely available on DVD by request. Availability and Requirements ============================= The MovingTargets software is available on DVD by request. It can be used on any Perl platform, such as the Macintosh \'Terminal\' utility. Usage of the software requires only minimal computer skills. The researcher can specify the biological constraints for a miRNA target search through the user interface. In addition, the researcher can specify a single target for focused searches with individual mRNAs. For a miRNA target search of the entire target database, the program runs in about 2 hours on an earlier generation Macintosh computer (466 MHz G4); focused searches for miRNAs targeting an individual mRNA are much faster and take about 20 minutes. List of abbreviations ===================== miRNA microRNA UTR untranslated region RNAi RNA interference RISC RNA-induced silencing complex PCR polymerase chain reaction DNA deoxyribonucleic acid RT room temperature FBS fetal bovine serum D. melanogaster Drosophila melanogaster D. pseudoobscura Drosophila pseudoobscura BDGP Berkeley Drosophila Genome Project mRNA messenger RNA Authors\' contributions ======================= CB designed and implemented the software, performed all of the experiments and drafted the manuscript. CB and PMM together conceived of the study, and PMM helped to draft the manuscript Acknowledgements ================ We thank the Texas Advanced Computing Center for use of computing resources, Takao Imai and Carl Thummel for plasmids, John Sisson for S2 cells and advice about their culture and use, and Robin Gutell and Rick Russell for discussions. Supported by NIH grant GM54409.	PubMed Central	2024-06-05T03:55:59.442304	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180435/", "journal": "BMC Genomics. 2005 Jun 8; 6:88", "authors": [ { "first": "Craig", "last": "Burgler" }, { "first": "Paul M", "last": "Macdonald" } ] }
PMC1180436	Background ========== The prognosis of patients with acquired immune deficiency syndrome (AIDS) was so limited until recently, that the medical interest in other long term health problems was irrelevant. The potency and sustained efficacy of the highly active antiretroviral therapy (HAART) for treating these patients brought a profound positive impact on their life expectancy reducing their mortality rates from AIDS \[[@B1]\]. Several reports \[[@B1]-[@B10]\] described the worsening of coronary heart disease (CHD) and vascular atherosclerotic complications in HIV+ patients after HAART therapy. Recently, the DAD study (Data Collection on Adverse Events of Anti-HIV Drugs) showed an increase in the risk of myocardial infarction (MI) from 0.30% in patients with no antiretroviral therapy to 1.07% in patients receiving these therapies, over a 3 year period \[[@B10]\]. Dyslipidemia is a major risk factor for the development of CHD. It has also been reported that the AIDS infection itself is capable of inducing dyslipidemia \[[@B11]-[@B14]\]. Hypertriglyceridemia was the first finding to be reported in HIV-infected patients, but other lipid abnormalities have also been described such as hypocholesterolemia, hypobetalipoproteinemia, hypoalphalipoproteinemia and, more rarely, hypercholesterolemia \[[@B11],[@B13],[@B15]-[@B17]\]. Brazil is the epicenter of the epidemic in South America and accounts for three-fifths of reported AIDS cases and 57% in Latin America and Caribbean. Among the population of high risk the prevalence is 42% \[[@B18]\]. Up to now two local studies \[[@B14],[@B19]\] explored the dyslipidemia of Brazilian HIV+ patients but both in a small number of cases. Therefore the aim of this study was to determine the prevalence and severity of different types of dyslipidemia in a large local HIV+ Brazilian population using antiretroviral therapy. Serum lipids, lipoproteins and apolipoproteins were measured and the effects on them of the viral load, CD4 counting and duration of therapy were evaluated. Methods ======= This study was approved by the Medical Ethics Committee of the Medical Sciences Faculty of the University of Campinas. Written consent was obtained from the patients or their relative for publication of study. Two hundred and fifty seven HIV+ patients were enrolled in the protocol. They were attended in the Infectious Diseases Clinic at the University of Campinas. Sixty-two percent were men and 38% were women, with an average age of 35 ± 8 years, body weight average 67 ± 13 Kg and body mass indexes (BMI) 24 ± 4 Kg/m^2^. Two hundred and thirty-eight (93%) were submitted to antiretroviral therapy (224 treated with protease inhibitors plus nucleoside reverse transcriptase inhibitors, 14 treated only with the latter, 12 naïve and 7 had no records of treatment). The average time on drug treatment with protease inhibitors was 20 months (range 2 to 47 months). None of the patients was under lipid lowering drugs and any other disease was described in their records. The measurements of fasting serum cholesterol (Chol), HDL-cholesterol (HDL-chol) and triglycerides (TG) were obtained by enzymatic colorimetric methods (automated Mega-Bayer system). The LDL-cholesterol (LDL-chol) and VLDL-cholesterol (VLDL-chol) were estimated by Friedewald\'s equation. In patients with triglyceride levels above 400 mg/dL (n = 16), Friedwald\'s equation was not used. The apolipoproteins A1 (Apo A1), B100 (Apo B) and lipoprotein \"a\" \[Lp (a)\] were measured by nephelometric methods (semi-automated Beckman Array-360 system) \[[@B20]\]. Free fatty acids (FFA) and phospholipids (PL) were determined by enzymatic colorimetric methods (Wako Chemicals GmbH, Japan). The plasma lipoproteins were fractioned by agarose gel electrophoresis (Paragon Electrophoresis System -- Beckman, Ca, USA). The bands were carefully analyzed by visual inspection \[[@B21]\]. Apo E genotyping was performed by PCR assay \[[@B21]\] in 3 cases suspected of type III phenotype. In a sub-group of patients presenting Chol and/or TG, respectively equal to or above 200 and 150 mg/dL (n = 141), the criterion for describing the dyslipidemia was in accord with WHO/ Fredrickson\'s classification \[[@B22]\]. The lipid profiles were analyzed using the values recommended by the National Cholesterol Education Program (NCEP, 2001) \[[@B23]\]. The desirable Chol/HDL-chol ratio was lower than 3.5 \[[@B24]\] and was used as an CHD risk index. The reference intervals for apolipoproteins, Lp (a), PL and FFA were obtained from the manufacturers\' recommendations. Apo A1: 94--178 mg/dL and 101--199 mg/dL; Apo B100: 52--109 mg/dL and 40--103 mg/dL for men and women, respectively; Lp(a) below 30 mg/dL; PL: 150--250 mg/dL and FFA: 0.1--0.6 mEq/L. Immunologic evaluation included measurements of lymphocyte sub-populations, CD4 and CD8, by flow citometry and the determination of the viral load by PCR. Kruskal-Wallis with post test Dunn was employed to compare the lipidic parameter averages between the 2 groups of antiretroviral therapy users and between the treated groups and naïve patients. Ancova and Pearson\'s tests were employed to correlate lipid profiles and CD4 cells, viral load and time of use of protease inhibitors. BMI, age and gender were chosen as covariates. Differences were considered significant when p ≤ 0.05. Results ======= The patients were classified according to the Center for Disease Control (CDC) -1993 \[[@B25]\] as follows: stage A, 39.3%; stage B, 12.4%; stage C, 45.6% (44.0% belonging to the clinical C3 category) and 2.7% were not clinically defined. There was an inversion of CD4 counting (average 337 cells/mm^3^) and CD8 counting (average 947 cells/mm^3^) that may be explained by the fact that approximately 44% of the patients belonged to the C3 category. Forty-four percent of the patients who had their viral load measured presented less than 400 copies/mL, 29% between 400--10,0000 copies/mL and 27% above 10,000 copies/mL. Fifty-five percent of total HIV+ population (141 patients) was selected by their high Chol and/or TG levels for further serum electrophoresis analysis. In eighteen patients the electrophoresis were not done and in 20 patients the results were inconclusive. The most frequent patterns according to Fredrickson\'s classification were type IV (41%) and IIb (51%) (Table [1](#T1){ref-type="table"}). One patient was type III, another one type V and seven IIa. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Percentual distribution of HIV+ patients according to WHO/Fredrickson classification in total HIV+ patients. ::: Type Percent of dyslipidemic patients\* (n = 103) Percent of the total HIV+ population (n = 257) ------ ---------------------------------------------- ------------------------------------------------ IIa 7 3 IIb 51 20 III 1 0,4 IV 41 16 V 1 0,4 \Patients with chol and/or TG respectively, \>200 and \>150 mg/dL, were classified based on Fredrickson\'s classification ::: Patients\' lipid, lipoprotein and apolipoprotein profiles are reported in Table [2](#T2){ref-type="table"}. Figure [1](#F1){ref-type="fig"} shows the percent distribution of HIV+ patients according to the NCEP (2001) classification \[[@B23]\]. Thirty-three percent of total HIV+ patients had Chol levels ≥ 200 mg/dL, 65% had HDL-chol below 40 mg/dL, 61% had LDL-chol ≥ 100 mg/dL; 46% had triglycerides ≥ 150 mg/dL and 86% had the cholesterol/HDL-chol ratio ≥ 3.5. Seventy-nine percent of the patients presented FFA values above the maximum reference value of 0.6 mEq/L. The average serum concentration of PL and Lp(a) were within the reference limits. However, 22% of the patients had Lp(a) above 30 mg/dL and 9% had PL above 250 mg/dL. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Serum lipid, lipoprotein and apolipoprotein profiles in HIV+ patients (n = 257) ::: Variables^a^ Mean ± SD Minimum Maximum --------------- ------------ --------- --------- Chol 185 ± 55 68 672 TG 192 ± 179 36 1512 HDL-chol 36 ± 12 12 73 LDL-chol\ 112 ± 38 10 227 VLDL-chol\* 31 ± 16 7 79 PL 192 ± 72 94 702 FFA 1.2 ± 0.75 0.10 6.3 Apo A1 121 ± 28 29 291 Apo B 88 ± 26 37 173 Lp(a) 21 ± 24 1 124 Chol/HDL-chol 5.6 ± 3 2 39.5 ^a^Data expressed as mg/dL and FFA as mEq/L. \* n = 241 patients ::: ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Lipid frequency distribution patterns in HIV+ patients (n = 236 to 257) according to NCEP\'s recommendations. ::: ![](1471-2334-5-47-1) ::: The mixed dyslipidemia in HIV+ patients are widely distributed and the frequency varied from 13% for TG or Chol + LDL-chol + HDL-chol to 38% for Chol+TG+HDL-chol. The treatment with protease inhibitors plus nucleoside reverse transcriptase inhibitors increased significantly Chol and TG levels (p \< 0.05) in these patients when compared to naïve HIV+ patients (187 ± 55 vs150 ± 21 mg/dL and 199 ± 187 vs106 ± 51 mg/dL for Chol and TG, respectively), no differences were found between the two drugs groups. There was a significant negative correlation (Figure [2](#F2){ref-type="fig"}) between the time of protease inhibitors (PI) therapy and HDL-chol (r = -0.214) and Apo A1 (r = -0.154) levels (p = 0.01 and p = 0.02, respectively). Also a significant negative association (p = 0.03) was found between Chol levels and the viral load. Total cholesterol levels have a trend to be associate with protease inhibitor therapy (p = 0.09). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Correlations between time on protease inhibitor therapy and HDL-chol, (panel A) and Apo A 1 (panel B). ::: ![](1471-2334-5-47-2) ::: Discussion ========== Dyslipidemia using the WHO/Fredrickson classification in a Brazilian HIV+ patients is examined here for the first time. Schmidt et al \[[@B25]\] described dyslipidemia in 57% of 98 HIV+ individuals treated with protease inhibitors and found a different prevalence from our group: phenotypes IV and V were more frequent than IIb and IIa. Maus et al \[[@B26]\] just recently presented this classification in 187 treated German HIV+ individuals. Some curious differences were found between the phenotyping of the 2 populations: types IIb and IV was equally frequent in our study, but the German group type IV was the predominant one. The data presented here are alarming. A very high prevalence of different types of dyslipidemia was found in the population of 257 HIV+ patients: 33% showed increased cholesterol, 46% increased triglycerides, decreased HDL-cholesterol level in 65%, cholesterol/HDL-chol ratio ≥ 3.5 in 86%, increased LDL-cholesterol in 61%, increased Lp(a) in 22%, increased free fatty acids in 79% and increased phospholipids in 9%. All of these profiles are strongly related to CHD risk. In this study we showed that in patients using antiretroviral therapy there was an augment in the concentrations of plasma cholesterol and triglycerides when compared to naïve patients. We also found negative associations between the time on PI therapy and HDL-chol and Apo A1 levels and a trend to a positive correlation between the viral load and cholesterol levels. Although HIV infection itself has been associated with altered lipid metabolism, substantial evidence indicates a role for some protease inhibitors and reverse transcriptase inhibitors in causing metabolic complications \[[@B27]-[@B32]\]. Many adverse reactions can be attributed to protease inhibitors, however; because several treatments use drug association regimens, it is difficult to find out the exact causal connection \[[@B33]\]. Several reports have associated hypertriglyceridemia with the use of antiretroviral therapy \[[@B6],[@B8],[@B30],[@B33]-[@B37]\]; however, the increase in triglyceride levels was also described in HIV+ patients before the utilization of HAART \[[@B11],[@B38]\]. Hypertriglyceridemia was highly prevalent in our and other studies \[[@B6],[@B11]-[@B13],[@B36],[@B38]-[@B43]\]. This may also be secondary to cytokine production. Cytokines have been shown to increase hepatic lipid synthesis and/or decrease levels of lipoprotein lipase, which results in slower clearance of circulating triglyceride-rich particles \[[@B44]\]. Grunfeld et al \[[@B44]\] found a highly significant positive correlation between interferon-α and triglyceride levels. In our study, hypertriglyceridemia was present in 46% percent of the HIV+ patients. Thirty-six percent of the patient\'s hyperlipoproteinaemia was due to types IV and IIb (Table 3). The origin of the hypertrygliceridemia could be an increased hepatic VLDL-TG secretion rate, with high availability of FFA to the liver, secondary to insulin resistance, common in these patients \[[@B45],[@B46]\]. Seventy nine percent of our patients presented high plasma levels of FFA, corroborating with this metabolic state. We did not find significant correlations between triglycerides and viral load \[[@B16]\], CD4 levels \[[@B13],[@B40]\] and the type of antiretroviral therapy \[[@B35]\], but we did find negative correlation between the duration of treatment with PI and HDL-chol (p = 0.01). Low HDL-Chol was one of the major findings in this Brazilian HIV+ population (65%). These findings are in accordance with several authors \[[@B13],[@B35],[@B42]\]. Several metabolic processes may contribute to these low HDL levels: decrease in free cholesterol removal from cells, low cholesterol esterification rate and high cholesteryl ester transfer from HDL to apo-B containing lipoproteins \[[@B47]\]. Hypercholesterolemia was very prevalent in the dyslipidemic individuals described here: 58% are types IIa and IIb. Grunfeld et al.\[[@B44]\] didn\'t find significant differences in the cholesterol levels of patients affected by HIV: HIV+; HIV+ with AIDS or HIV negative, but Law et al. found raised cholesterol and triglyceride levels in patients receiving HAART, compared with patients not receiving HAART \[[@B10]\]. Our results are in agreement with Law\'s paper. Grunfeld et al \[[@B39]\] showed a significant decrease in cholesterol levels in the HIV+ and HIV+ with AIDS patients, when compared to HIV- subjects. Shor-Posner et al \[[@B48]\] observed hypocholesterolemia (chol \<150 mg/dL) in 41% of HIV+ and in 17% of HIV- patients. Christeff et al \[[@B40]\] reported decreased concentrations of cholesterol and phospholipids in HIV+ patients, except in those with low CD4 counting (400--150 cells/mm^3^). In this study, 29% of the patients presented hypocholesterolemia. We found an inverse correlation between plasma cholesterol levels and the viral load. The observed high frequency of hypercholesterolemia and hyperbetalipoproteinemia in HIV+ Brazilian patients were never reported previously (Table [1](#T1){ref-type="table"}, Figure [1](#F1){ref-type="fig"}). Also analyzing the various patterns of the combined hyperlipidemia the most prevalent were alterations in TG plus HDL-chol and LDL-chol plus HDL-chol, respectively 35 and 32% ; total cholesterol plus TG and HDL-chol in 38% and 11% of patients had four different lipidic parameters altered. Conclusion ========== The high frequency of phenotypes IIb and IV found in the Brazilian HIV positive patients and the severity of the disturbances such as: hypercholesterolemia up to 672 mg/dL, hyperbetalipoproteinemia up to 227 mg/dL, hypertriglyceridemia up to 1512 mg/dL and hypoalphalipoproteinemia below 13 mg/dL, indicate the urgent need for their early diagnosis, the identification of the presence of other risk factors for CHD and, when needed, the prompt intervention on their lifestyle and/or drug treatment \[[@B49],[@B50]\]. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= EMVA performed the biochemical analyses, data calculations and interpretation of results and helped to write the manuscript. ECF helped in the clinical trials and to write the manuscript. HCFO helped to write the manuscript. DOM participated in the clinical trial of HIV patients. LNC planned and coordinated the whole study, and helped to write the manuscript. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2334/5/47/prepub> Acknowledgements ================ Financial support for the study was provided by Coordenadoria de Aperfeiçoamento de Pessoal de Ensino Superior -- CAPES -- and Fundo de Apoio ao Ensino e Pesquisa -- FAEP/FCM. The authors are gratefull to Dr. Joyce M. Annichino-Bizzacchi -- Director of Hemocentro -- UNICAMP and Cleide Aparecida Moreira Silva (Statistical Research Commission of UNICAMP) for their excellent technical assistance.	PubMed Central	2024-06-05T03:55:59.446529	2005-6-14	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180436/", "journal": "BMC Infect Dis. 2005 Jun 14; 5:47", "authors": [ { "first": "Edilma MV", "last": "Albuquerque" }, { "first": "Eliana C", "last": "de Faria" }, { "first": "Helena CF", "last": "Oliveira" }, { "first": "Daniela O", "last": "Magro" }, { "first": "Lucia N", "last": "Castilho" } ] }
PMC1180437	Background ========== The immune system uses several strategies to modify genetic material to generate various types of high affinity antibodies \[[@B1]\]. These strategies enable production of multiple antibody variants to a wide range of different antigens \[[@B2]\]. Initially, antigen receptors are generated by a site-specific recombination process called V(D)Jrecombination occurring in the bone marrow \[[@B3]\]. However, this is not sufficient to assure an adequate immune response. Mature B-lymphocytes migrate to the secondary lymphoid organs where they encounter antigens. Upon activation by antigens, mature B-lymphocytes begin to proliferate and form germinal centers, where immunoglobulin genes undergo additional modifications: class switch recombination (CSR), immunoglobulin gene conversion (IGC) and somatic hypermutation (SHM) \[[@B4]\]. SHM, IGC and CSR, all require active transcription \[[@B5]\] and generate diversity of antibodies, that is followed by selection leading to the production of high affinity antibodies \[[@B6]\]. The frequency of mutations during this process is up to six orders of magnitude higher than in other genes \[[@B6]\]. Most of the mutations are base pair substitutions, occurring with a similar frequency at G-C and A-T base pairs. Statistically preferred hotspots for mutations at G-C pairs are R[G]{.underline}YW/WR[C]{.underline}Y motifs (mutating G-C are underlined, R stands for purine base, Y stands for pyrimidine base and W stands for A or T) \[[@B7]\], or recently refined D[G]{.underline}YW/WR[C]{.underline}H motifs (D stands for G, T or A) \[[@B8]\]. Hotspots of mutations at A-T pairs are in W[A]{.underline} /T[W]{.underline} motifs (mutating A-T are underlined) \[[@B9]\]. A major breakthrough in understanding the mechanisms of CSR, IGC and SHM was the discovery that they all depend on activation-induced cytidine deaminase, AID \[[@B10]-[@B16]\]. Patients with defective AID have giant germinal centers and elevated levels of only one type of low-affinity antibodies, IgM. They suffer from recurrent bacterial infections in the respiratory tract \[[@B17]\] due to the lack of efficient antibody responses that depend on several crucial steps of B cell terminal differentiation including CSR and SHM. SHM is targeted to specific DNA regions in specialized tissues. Defects in this targeting may result in genome-wide mutagenesis and cancer. B-cell lymphomas possess translocations that bring proto-oncogenes into immunoglobulin loci (see \[[@B18]\]). Constitutive expression of AID in mice leads to an increase of tumor incidence \[[@B19]\]. When discovered, AID was thought to act in mutagenesis and recombination in immunity by RNA editing \[[@B10],[@B11],[@B20]\]. It was proposed that AID edits pre-mRNA encoding a nicking endonuclease that initiates SHM, IGC and CSR \[[@B5]\]. This model is called \"RNA-editing\" \[[@B20]\]. The AID is homologous to the known RNA-editing enzyme APOBEC1, which deaminates cytosine at position 6666 in ApoB100 mRNA and seemingly has no role in immunity. AID possesses the ability to deaminate cytidine, and shuttles between the nucleus and cytoplasm similar to APOBEC1 \[[@B4],[@B5],[@B21],[@B22]\]. A different hypothesis, called \"DNA deamination\", suggests AID deaminates cytosine directly and that uracil generated in this reaction triggers downstream reactions leading to genetic instability \[[@B23]-[@B26]\] (see \[[@B27]-[@B33]\] for reviews). Experimental evidence is accumulating in favor of the DNA deamination hypothesis of AID function \[[@B29],[@B31]-[@B34]\]. AID is able to induce SHM and CSR in hybridomas and in fibroblasts, suggesting that it is the only B-cell specific component required for induction of both genetic events \[[@B13],[@B14],[@B35]\]. AID can also induce mutations when expressed in E. coli\[[@B24]\]. These mutations occur in the same DNA sequence motifs as mutations during SHM \[[@B8],[@B36]\]. Therefore, eukaryotic cell-specific components are not necessary for mutagenesis. This mutator effect is enhanced in uracil-DNA glycosylase-deficient ung1strains, which are unable to repair uracil in DNA \[[@B37]\], suggesting that the deamination of cytosine to uracil in DNA is the cause of these mutations \[[@B24]\]. It was found that the expression of two homologous deaminases, APOBEC1 and APOBEC3G, is highly mutagenic in bacteria \[[@B38]\]. Almost all mutations arising upon expression of these deaminases were G-A to A-T transitions, consistent with the DNA deamination model. AID deaminates single-stranded and supercoiled double-stranded DNA \[[@B39]-[@B44]\] (see also review \[[@B31]\]). AID exhibits clear DNA sequence context specificity, which resembles the specificity of G-C to the A-T component of SHM mutagenesis ([G]{.underline}YW/WR[C]{.underline} motifs, see\[[@B8],[@B40],[@B44]-[@B46]\]). The specificity of induced mutations in bacteria is consistent with predominant deamination of the non-transcribed DNA strand \[[@B36],[@B45]\], which is thought to be single-stranded during transcription (reviewed in \[[@B31]\]). During SHM, however, both DNA strands are targeted for mutagenesis \[[@B7]\], \[[@B47]\]. This discrepancy between the parameters of SHM in vertebrates and deaminase-induced mutagenesis in prokaryotes still needs to be resolved. To characterize the initial steps of AID-induced mutations, we examined the specificity of the mutator effect of human AID expressed in yeast. We constructed a yeast vector with an artificially synthesized human AIDgene insert using codons common to highly expressed yeast genes. We found that expression of the artificial hAIDScgene was moderately mutagenic in the wild-type strain and highly mutagenic in the ung1strain, similar to expression of unmodified human AID \[[@B48]\]. This is consistent with the uracil DNA deamination model of mutagenesis. We identified a spectrum of mutations in the CAN1gene occurring in wild-type and ung1strains expressing hAIDSc. We compared the sequence context of AID-induced mutations in yeast at G-C bases with somatic mutations in immunoglobulin genes. These comparisons revealed significantly similar properties and further support the hypothesis that AID is a primary cause of mutations at G-C pairs in immunoglobulin genes during SHM. Results ======= hAIDScexpression and its mutator effect ----------------------------------------- Codon usage is different in yeast and humans. To improve our system of expression of human AIDover work published earlier \[[@B48]\], we constructed a new yeast expression vector with the human AIDgene recoded to use the same codons utilized by highly expressed yeast genes and with a galactose-inducible promoter. Appropriate transformants were grown in galactose-containing medium and the AID protein was readily detected in yeast extracts by Western blot (Fig. [1](#F1){ref-type="fig"}, lane 3). The expression of the hAIDScdid not result in any profound growth inhibition; the cell titer usually reached 5 × 10^7^, which is typical for galactose-containing minimal medium (data not shown). Mutation rates were analyzed by fluctuation analysis (Table [1](#T1){ref-type="table"}). Our strain permits the detection of various classes of genetic events (see Materials and Methods, and also \[[@B48],[@B49]\]). Using this strain we can obtain the express information about the specificity of the mutagenic effect. The expression of the hAIDScdid not induce frameshift mutations to His^+^(last column of Table [1](#T1){ref-type="table"}). In Ung1^+^strains, hAIDScexpression leads to a 7.6 fold increase in Can^r^forward mutations and a 3 -- 6 fold increase in nonsense mutation reversion (Ade^+^, Trp^+^). The ung1mutation per seled to a 5 -- 10 fold increase of mutation rates as shown in rows one and four. When the hAIDScwas expressed in the ung1strain, the mutator effect was multiplicative for Can^r^forward mutations (82 fold increase over the wild-type strain) and synergistic for nonsense mutation reversion (a 404 -- 1290 fold increase over wild-type). TAG and TAA nonsense mutations cannot revert by true back-mutations via G-C to A-T transitions. We have previously shown by genetic analysis and sequencing of revertants that reversion is caused by dominant suppressors and most likely represent mutations in the anticodon of tRNA genes, which could be G-C to A-Ts \[[@B48]\]. The high response of ade5-1and trp1-289markers to hAIDScmay reflect the role of transcription in AID-induced hypermutation in yeast, since tRNA genes are transcribed differently from metabolic genes. The ura3-29allele reversion was stimulated only weakly. It is known, that the allele reverts via various changes at G-C pair in \"TCT\" DNA sequence context \[[@B50]\], which is different from hotspots of AID deaminations. The results suggest that uracil DNA deamination is the primary source of mutation induced by the hAIDScin yeast and are consistent with our previous studies \[[@B48]\]. Optimized codon usage did not lead to increased mutagenesis under conditions of constant induction of galactose promoter since the mutagenic potential of expression of the hAIDScwas comparable with expression of native human AID\[[@B48]\]. Mutagenic specificity of hAIDSc --------------------------------- We studied the specificity of mutations in the CAN1gene induced by the expression of the hAIDSc. Independent Can^r^mutants were obtained under conditions of hAIDScexpression in the wild-type and the ung1strain. Results of sequencing of mutants are summarized in the Tables [2](#T2){ref-type="table"}, [3](#T3){ref-type="table"} and \[see [Additional file 1](#S1){ref-type="supplementary-material"}\]. Most mutations (64 out of 70 in the wild-type and 62 out of 66 in the ung1strain) were at G-C base pairs. Transversions comprised 12% of the mutations at the G-C pairs in wild-type and 1.6% in the ung1strain (Table [4](#T4){ref-type="table"}). The decreased proportion of transversions in the ung1strain is consistent with the data obtained earlier in chicken and mice \[[@B26],[@B51]\]. We compared these spectra with the spectra of spontaneous mutations in CAN1in the wild-type strains obtained by Rattrey and coauthors \[[@B52]\], Table [5](#T5){ref-type="table"}. The major property of these mutation spectra was a high frequency of frameshift mutations (\>20%) \[[@B52]\]. Another feature of the spontaneous mutations is a high frequency of mutations in A-T bases (\>50%) and a higher frequency of transversions compared to transitions (\>50%) (see also the breakdown of the types of spontaneous mutations obtained previously by other groups \[[@B53]-[@B55]\]). These features of CAN1spontaneous mutations are similar to the properties of mutations observed in the yeast SUP4-ogene \[[@B56]\]. Thus, the spectra of mutations induced by the expression of the hAIDScare different from spontaneous mutations in yeast genes. This result indicates that spontaneous mutations constitute a minor fraction (if any) of the mutations induced by the expression of the hAIDSc. G-C mutations may arise by putative deamination on the transcribed or non-transcribed DNA strand. Mutations in the ung1strain, representing deamination proclivity without of uracil repair, occur at a higher rate on the transcribed strand (Table [4](#T4){ref-type="table"}). This is different from the effect of AIDexpression observed in the most E. coliselective systems \[[@B31]\]. In the wild-type strain, there is some prevalence of mutations due to putative non-transcribed strand deaminations (Table [4](#T4){ref-type="table"}) suggesting the possibility that in our system the repair of uracil in the transcribed DNA strand is more efficient than in non-transcribed strand. Clearly, hAIDScis targeted to both DNA strands in yeast, similar to somatic mutations in G-C bases during SHM \[[@B6],[@B7],[@B9],[@B57]-[@B59]\]. It is important to mention that, under normal circumstances, there are no differences in DNA strand preferences between mutation spectra from the wild-type and ung1strains \[[@B60]\]. Next, we examined whether the DNA context of mutations induced by AID in yeast is similar to SHM mutations in mammals or in E. coliexpressing hAID(Table [5](#T5){ref-type="table"}). D[G]{.underline}YW and [G]{.underline}YW motifs \[[@B7],[@B8],[@B40]\] were under-represented in mutations occurring spontaneous in wild-type or ung1strains (Table [5](#T5){ref-type="table"}, row 1--2) and were 2 -- 5 fold over-represented in mutations induced by AID in yeast (Table [5](#T5){ref-type="table"}, rows 3--4). Lists of mutation hotspots are shown in Table [6](#T6){ref-type="table"}. Distributions of AID-induced mutation hotspots in the wild-type and ung1strains are significantly different (Table [6](#T6){ref-type="table"}, P = 0.003). The specificity of mutations in yeast correlates better with the hotspot motifs for SHM in mice than does the specificity of AID induced mutations in E. coli. Indeed, out of four comparisons, the indices of preference for mutation hotspot motifs in yeast were higher than in E. coli(compare rows 3--4 with rows 5--8; rows 10--11 with rows 5--8, Table [5](#T5){ref-type="table"}). Some properties of mutations in yeast resemble the in vitroAID induced mutation spectrum \[[@B40]\]. For example, one C[G]{.underline}YW/WR[C]{.underline}G sequence which is not mutable in SHM \[[@B8]\] had a high mutability in the ung1mutation spectrum (Table [6](#T6){ref-type="table"}). It has been suggested that a mammalian DNA repair enzyme, perhaps the uracil-DNA glycosylase, efficiently repairs the lesion of CpG dinucleotides and thus eliminates mutations from C[G]{.underline}YW/WR[C]{.underline}G motifs in vivo\[[@B8]\]. We have found 5 G[G]{.underline}YW and 3 T[G]{.underline}YW hotspots (Table [6](#T6){ref-type="table"}). Interestingly, no hotspots were found in A[G]{.underline}YW motifs (Table [6](#T6){ref-type="table"}), which are the most frequent hotspot motifs in mammalian immunoglobulin genes \[[@B47]\]. The lack of mutation hotspots at A[G]{.underline}YW could not be attributed to its lesser prevalence because the number of AGYW, G[G]{.underline}YW, and T[G]{.underline}YW motifs in CAN1was similar (results not shown). However, a general pattern of mutations in yeast is similar to the targeting of somatic mutation in D[G]{.underline}YW/WR[C]{.underline}H mutable motifs, which are highly specific for SHM in mammals. In control, no significant targeting of mutations to D[G]{.underline}YW/WR[C]{.underline}H mutable motifs was found for spontaneous mutations in wild type or ung1yeast (P~W≤Wrandom~\> 0.05, Table [5](#T5){ref-type="table"}, rows 1--2). We did not find a substantial number of A-T mutations, which typically comprise one-half of all SHM \[[@B47],[@B58],[@B61]\]. This result corresponds with earlier published results on the expression of AID in E. coli\[[@B24],[@B36]\], in yeast \[[@B48]\], in murine fibroblasts \[[@B13]\], and in human hybridomas \[[@B35]\]. Apparently, additional components are required to model the full spectrum of SHM under conditions of AID expression in heterologous systems or in non-B cell tissues. Discussion ========== Mutator efficiency and specificity of expression of hAIDSc ------------------------------------------------------------ Yeast is a well-studied model eukaryotic organism used for various genetic studies. Yeast was used in this study to characterize the mutator effects of ectopic expression of human AID. The CAN1reporter gene has been chosen because of numerous mutational studies \[[@B52],[@B54]\] and a well-characterized transcription pattern \[[@B62]-[@B64]\]. The results are different from studies of AID effects in prokaryotic models and in vitroexperiments. We observed mutations arising due to deamination occurring in both DNA strands. In E.coli, transcription enhances deamination of the non-transcribed DNA strand, which is exposed as single-stranded DNA during the elongation reaction, but not mutation of the transcribed DNA strand, which is likely to be protected by E. coliRNA polymerase \[[@B42],[@B43],[@B65]\]. The observed DNA strand targeting of mutations in ung1yeast closely resembles targeting of somatic mutations in vertebrate immunoglobulin genes (Table [5](#T5){ref-type="table"}). Interestingly, there was a significant strand bias of mutations in wild-type yeast toward the non-transcribed strand (Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}). A more efficient repair of the transcribed DNA strand is one possible explanation of this asymmetry. Preferential nucleotide excision repair of the transcribed strand is a well-known phenomenon \[[@B66]\]. The possibility of transcription-coupled repair of uracil bases in DNA has not yet been thoroughly studied. Interestingly, a strand bias toward the non-transcribed DNA strand was found in Ung^-/-^Msh2^-/-^mouse (Table [5](#T5){ref-type="table"}, row 8) \[[@B67]\]. The difference between the number of mutations in D[G]{.underline}YW/WR[C]{.underline}H sites versus all other G:C sites in wild-type and Ung^-/-^Msh2^-/-^strains was statistically significant (Fisher exact test, P = 0.04). This may indicate that AID has a preference to the non-transcribed DNA strand as suggested earlier (see review \[[@B32]\]). An excessive DNA deamination of the non-transcribed DNA strand may be compensated by more efficient repair of this strand during the SHM phase 2 \[[@B67]\] causing approximately equal frequencies of mutations in both DNA strands (Table [5](#T5){ref-type="table"}). More efficient repair of the non-transcribed strand is consistent with the idea of preferential targeting of the DNA polymerase η to the non-transcribed strand during SHM \[[@B9],[@B25]\]. In general, the strand specificity of SHM in Ung^-/-^Msh2^-/-^mouse is similar to AID-induced mutations in wild-type strains of yeast and E. coli(Table [5](#T5){ref-type="table"}). Substantial differences between the observed targeting of AID to the mutable motifs in Ung^-/-^Msh2^-/-^and wild-type mouse (Table [5](#T5){ref-type="table"}) are not consistent with a hypothesis that mutagenesis during the A:T-focused phase is nearly exclusively targeted to A:T bases \[[@B67],[@B68]\]. It is possible that mutagenesis during this phase is targeted to both A:T and G:C bases with a preference to A:T bases and no preference to D[G]{.underline}YW/WR[C]{.underline}H mutable motifs, this is consistent with the observed mutational and context specificity of the DNA polymerase η in vitro\[[@B9],[@B25],[@B69]\], D[G]{.underline}YW/WR[C]{.underline}H-independent mutagenesis of G:C bases will cause erosion of a high initial D[G]{.underline}YW/WR[C]{.underline}H motif specificity observed in Ung^-/-^Msh2^-/-^mouse (Table [5](#T5){ref-type="table"}). There are also differences between strand specificity of Ung^-/-^Msh2^-/-^mouse and AID-induced mutations in human fibroblasts (Table [5](#T5){ref-type="table"}), this might be explained by some differences in AID targeting or transcription-associated repair of uracil between B-lymphocytes and fibroblasts. All these results suggested that a weak strand bias is an intrinsic property of SHM. A significant difference between in vitrosystems and our experiments was observed. AID catalyses multiple deaminations in vitro\[[@B40]\]. We detected 11 clones with multiple mutations (10 clones with two mutations and one clone with three mutations) and checked the number of mutations in the first and second halves of CAN1. If multiple mutations emerge as a result of independent events, half of the clones are expected to have mutations in different halves of CAN1. In six out of 11 clones mutations were located in different parts of CAN1, thus independent mutation events is the most likely explanation. In general, the specificity and distribution of mutations in yeast did not exhibit a pattern of multiple mutations that would have been consistent with postulated processive action of AID \[[@B40]\]. These results are consistent with a high frequency of rearranged immunoglobulin V genes with one somatic mutation (for example, \[[@B70]\]). Apparent non-processive action of AID in vivomay be explained by a competition for binding to the CAN1DNA sequence between AID and other proteins participating in transcription, replication and/or repair. For example, it is known that replication factor A stimulates AID \[[@B71]\], while the specificity of AID in vitrowas studied on DNA without any additional factors. Clearly, this requires additional investigation. Mechanisms of mutagenesis by AID -------------------------------- The mechanism of SHM initiated by AID may be as follows (see \[[@B23],[@B24],[@B27],[@B28],[@B67],[@B72]\]). Deamination of cytosine in DNA leads to the formation of a mismatched U-G base pair. If left unrepaired, further rounds of replication of uracil-containing DNA will generate only transition type mutations, G-C to AT. Uracil removal by uracil-DNA glycosylase leads to an apyrimidinic (AP) site. The AP site may be bypassed by a specialized DNA polymerase and, being a non-coding lesion, could lead to a transition or transversion mutation. The AP site may also be incised by AP-endonuclease and then repaired by the short patch base excision repair (BER) with involvement of error prone DNA polymerases with generation of transitions and transversions (e.g., see \[[@B72]\]). This mechanism generates mutations at G-C pairs. In order to explain the high frequency of mutations during the short patch BER reaction, it should be postulated that the relatively accurate DNA polymerase β is substituted in B-cells by an error-prone polymerase. The candidate is DNA polymerase ι, which is expressed in Burkitt\'s lymphoma cell line BL-2 \[[@B73]\] and whose inactivation suppresses SHM in this line \[[@B74]\]. However, 129-derived strains of mice, lacking active polymerase η, are fully proficient in SHM \[[@B75]\]. The reason for this discrepancy in not established yet. Another type of mutation, which comprise about 50% of all mutations during SHM, is a change at the A-T base pairs \[[@B47],[@B58],[@B61]\]. The explanation of the mutation origins in A-T base pair is based on several observations. It is known that mutations at A-T base pairs depend on mismatch repair components MSH2, MSH6, EXO1and error-prone DNA polymerase η \[[@B9],[@B29],[@B67],[@B68],[@B70],[@B72]\]. They are thought to be the result of error-prone bypass or repair of abasic site by error-prone polymerases, in particular, DNA polymerases η, ι and ζ \[[@B9],[@B25],[@B70],[@B73],[@B74],[@B76],[@B77]\] (reviewed in \[[@B28],[@B72],[@B78],[@B79]\]). It is possible that they are generated in the following way. Initiation of mismatch repair of a G-U base pair leads to a gap. Gaps may also be generated by long patch BER. Repair of gaps with the involvement of error-prone DNA polymerases may lead to mutations distal to initial G-U pair \[[@B25],[@B68]\]. Again, it should be postulated that gap repair is unusual in B-cell being inaccurate, since normally it is performed by highly accurate replicative DNA polymerases. The final feature of the current model of AID-initiated genetic modifications is that nicks and gaps, arising during DNA repair, stimulate recombination \[[@B16],[@B48]\]. SHM in the Ung1^-/-^mice is greatly biased in favour of transitions, since the pathway via apyrimidinic sites is blocked \[[@B26]\]. Mutations at A-T base pairs are absent in Msh2^-/-^Ung1^-/-^mice \[[@B67]\]. Is it important to notice that Ung1 is not a major enzyme involved in the overall repair of G:U mismatches in mice, as suggested by small mutator phenotypes in the Ung1^-/-^mice and the existence of the robust Smug1 glycosylase \[[@B80]\]. In B-cells, however, the Ung1 alone appears to be crucial for all genetic diversification processes \[[@B67],[@B81]\]. Mutations at A-T base are not observed when AID is expressed in prokaryotes or in yeast \[[@B31],[@B36],[@B48]\], and this work. Therefore, current model systems only partially reconstruct SHM. Delicate balance of mismatch repair and activity of error-prone polymerases, specific for B cells, might be required for the full spectrum of SHM mutations \[[@B68]\]. Changes in the chromatin structure are necessary for SHM \[[@B82]\] and this additional level of regulation should be taken into account when considering different SHM models. Conclusion ========== In the present study, we have shown that expression of human AIDis mutagenic in yeast and the mutagenic effect is one-two orders of magnitude higher in the ung1strain. This observation suggests that the cause of the mutator effects is AID-driven DNA deamination. DNA sequence contexts of mutation hotspots coincide with D[G]{.underline}YW/WR[C]{.underline}H mutable motifs of somatic hypermutation, which is consistent with the DNA deamination model of SHM, suggesting that the intrinsic substrate specificity of AID itself is a primary determinant of mutational hotspots at G-C base pairs during SHM. Methods ======= Construction of the expression vector ------------------------------------- A new hAIDgene was constructed using codons characteristic to highly expressed yeast genes. The DNA Builder program <http://cbi.swmed.edu/computation/DNABuilder/dnabuilder.html> and yeast codon usage data \[[@B83],[@B84]\] was used to construct a DNA sequence encoding human AID, with the preferable yeast codons. The DNA corresponding to this sequence and encoding for the c-myc Tag at the C-terminus (hAIDSc) was custom-synthesized and cloned into BamHI-SalI cut pESC-LEU (Stratagene) expression vector by the McLab Company (San Francisco). In this construct, the deaminase genes were placed downstream of the strong, galactose-inducible GAL1promoter. DNA sequencing analysis confirmed the complete sequence of the insert. Protein production was demonstrated by Western blot as described earlier \[[@B85]\], with one modification -- the Western Breeze Kit (Invitrogen) was used for detection of the protein in yeast extracts. Yeast strains ------------- For our experiments with the yeast vector expressing the deaminase genes we used yeast strain CG379-3-29RL (MAT α ura3Δleu2-3,112 trp1-289 bik1::ura3-29RL his7-2 ade5-1 lys2-Tn5-13) \[[@B48],[@B86],[@B87]\]. This strain allows concomitant measurement of mutation rates at several loci. These include a) the forward mutation rate at the CAN1locus, where mutations reflect a variety of substitution, frameshift and more complex events; b) the rate of reversion of nonsense mutations: the trp1-289(TAG \[[@B88]\]) and ade5-1(TAA, \[[@B89]\]), where mutations reflect base substitutions in the nonsense codon as well as in suppressor genes encoding tRNAs; c) reversion of the ura3-29missense mutation TCT which occurs via C-G to T-A transitions, C-G to C-A and C-G to G-C transversions \[[@B50]\]; d) reversion of the his7-2mutant allele which occurs mainly via + 1 frameshifts in a homopolymeric AT run \[[@B49],[@B90]\]. Measurement of mutations rates ------------------------------ Mutation rates were analysed by fluctuation analysis \[[@B49],[@B90]\]. Independent transformants of the wild-type and ung1derivatives of our basic strain were grown in a complete minimal medium lacking leucine to select for the plasmid, and containing galactose instead of glucose, to induce the hAIDScexpression. Isolation and sequencing of can1mutants ----------------------------------------- Yeast transformants patches originating from single colonies (64 per one plate) were replica-plated onto galactose-containing medium without leucine. After two days, they were replica-plated onto canavanine-containing medium to select for can1mutants. After five days of incubation, one Can^r^colony was picked from each streak and streaked onto canavanine-containing medium. Chromosomal DNA from cells originating from one colony of these can1mutants was isolated using a Yeast DNA Extraction Kit (Epicentre). Subsequent PCR amplification and sequencing was performed as described earlier \[[@B91]\]. In vivo and in vitro mutation spectra ------------------------------------- Five in vivoand one in vitromutation spectra, which have been described before \[[@B13],[@B26],[@B36],[@B40],[@B47]\] were used in this study. We consider that these large mutation spectra reflect intrinsic bias in mutation process. The compilation of somatic mutations in the VkOx transgene includes data derived from transgenic light chains with multiple copies of the transgene and from cells selected in gut Peyer\'s Patches (PP). The multiple copies are targeted in the same cell even when the light chain they encode is not part of the antigen binding antibody molecule. This implies that the majority of the mutations accumulated are unselected. In the case of PP derived cells, the selective pressure is multiple, therefore again, the common denominator of the biases observed would reflect the intrinsic biases \[[@B92],[@B93]\]. Statistical analyses -------------------- The Fisher exact test was used to compare frequencies of transitions and transversions. This test was also used to compare the number of mutations in D[G]{.underline}YW/WR[C]{.underline}H sites versus all other G-C sites in wild-type and Ung^-/-^Msh2^-/-^strains of mouse. A Monte Carlo modification of the Pearson χ^2^test of spectra homogeneity \[[@B94]\] was used to compare mutation distributions along hotspot positions of the CAN1sequence. Calculations were done using the COLLAPSE program <ftp://ftp.bionet.nsc.ru/pub/biology/dbms/>\[[@B95]\]. Mutations in the CAN1gene were detected using the phenotypic assay described above, however the full list of detectable positions in this gene is not known. We predicted these positions using the SIFT program with default parameters \[[@B96]\]. Mutation hotspots were defined using a threshold for the number of mutations at a site. The threshold is established by analyzing the frequency distribution derived from a mutation spectrum using the CLUSTERM program <http://www.itb.cnr.it/webmutation/>\[[@B97]\]. Briefly, this program decomposes a mutation spectrum into several homogeneous classes of sites, with each class approximated by a binomial distribution. Variations in mutation frequencies among sites of the same class are random by definition (mutation probability is the same for all sites within a class), but differences between classes are statistically significant. Each site has a probability P(C) to be assigned to class C. A class with the highest mutation frequency is called hotspot class. Sites with of P(C~hotspot~) ≥ 0.95 of being assigned to the hotspot class C~hotspot~are defined as hotspot sites. This approach ensures that the assignment is statistically significant and robust (see Rogozin et al. \[[@B98]\] for detailed discussion of this approach and problems associated with its application). Nucleotide sequence features can be correlated with a mutation spectrum and the correlation can be tested for statistical significance. The significance of correlations between the distribution of mutable motifs and mutations along a target sequence was measured by a Monte Carlo procedure (the CONSEN program) \[[@B7]\]. This approach takes into account frequencies of substitutions for each nucleotide, the possibility of multiple mutations in a site, and context of the mutating sites. The Monte Carlo simulation was run with weighted sites, with the weight of a site defined as: ![](1471-2172-6-10-i1.gif) where M~j~is the number of mutations in site j. W~j~weights were summed for all sites in the analyzed sequence resulting in the total weight W. A distribution of total weights W~random~was calculated for 10,000 target sequences with randomly shuffled mutation spectra. Each of the resulting random mutation spectra contained the same number of mutations as the observed spectrum with the same distribution of mutations over randomly chosen sites. The distribution of W~random~was used to calculate probability P~W≤Wrandom~. This probability is equal to the fraction of random spectra in which W~random~is the same or greater than W. Small probability values (P~W≤Wrandom~≤ 0.05) indicate a significant correlation between a mutable motif and the mutation frequency \[[@B7],[@B99]\]. Authors\' contributions ======================= This work was designed by YIP, TAK and IBR. YIP and CF were involved in vector construction, Western blot analysis of proteins, mutation rates measurements and isolation of can1mutants. VIM and LA performed PCR and DNA sequencing and were involved in analysis of mutations. IBR did statistical analysis of mutation spectra. Supplementary Material ====================== ::: {.caption} ###### Additional File 1 *Spectra of can1mutations in yeast. Types of nucleotides changes found are shown above the DNA sequence of the non-transcribed strand. A. ung1strain, B. wild-type strain. Most of the mutations were single base pair substitutions. The multiple mutations found in the CAN1are listed below. wild-type strain 452,C-\>T; 1728,T-\>G 123,A-\>G; 1392,G-\>A 123,A-\>G; 412,C-\>T 589,T-\>A; 1151,C-\>G 1485,G-\>T; 1486,C-\>A ung1strain 553,C-\>A; 1633,C-\>T 1018,G-\>A; 1728,T-\>G 299,G-\>A; 1475,C-\>T 424,G-\>A; 980,G-\>A 572,T-\>C; 573,C-\>T 509,G-\>A; 516,G-\>A; 806,G-\>A ::: ::: {.caption} ###### Click here for file ::: Acknowledgements ================ This work was supported in part by MedCen Foundation and RFBR. We are grateful to Barry Gold, Eugene Koonin, Elena Vasunina, Robert Lahue and Tadayoshi Bessho for fruitful discussions. We thank Dr. Polina Shcherbakova for critical reading of the manuscript. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Western blot analysis of hAIDScexpression in yeast. Yeast strain CG379-3-29RL transformed by expression vector pESC-LEU2 or pESC-LEU-hAIDScwere grown to logarithmic phase in a complete minimal medium without leucine. Then cells were washed and transferred into similar medium but containing galactose instead of glucose. Yeast protein extracts were prepared from approximately 200 mg of cells by the glass beads cell disruption method as described in \[85\]. Proteins were separated using 4--12% gradient PAA NuPage gel (Invitrogen). Transfer to PVDF membrane and reaction with primary antibodies (mouse anti c-myc) and then secondary antibodies (goat antimouse) was accomplished as suggested by the vendor (Western Breeze kit, Invitrogen). Lane 1 -- Molecular weight markers (Benchmark, His-Tagged) were detected with antiHisx6 antibodies. Lane 2 -- extract of yeast strain containing vector pESC-LEU. Lane 3 -- extract of yeast strain containing pESC-LEU-hAIDSc. ::: ![](1471-2172-6-10-1) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Mutagenic effect of the hAIDScexpression in yeast. ::: Strain Plasmid Mutation rates\ -------------- ---------- ------------------ ---------------- ---------------- --------------- ---------------- Can^r^× 10^-7^ Ade^+^× 10^-8^ Trp^+^× 10^-8^ Ura+ × 10^-8^ His^+^× 10^-8^ wild-type vector 2.5 24 4.1 4.0 2.4 1.2--6.5 21--34 1.3--14 2.2--6.6 1.9--2.9 hAIDSc 19 72 24 4.0 2.3 14--25 60--124 21--38 2.8--7.8 1.1--4.2 ung1::hygB vector 13 210 134 23 3.0 10--31 190--290 110--170 20--35 1.9--4.5 hAIDSc 205 9700 5300 52 4.0 170--220 7500--12600 4400--6600 43--77 2---7.1 \Median mutation rates determined in 9--18 cultures. 95% confidence limits are shown below. ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### DNA sequences changes in can1mutants induced by expression of hAIDSc. ::: Total Observed sequence change ----------- ------- -------------------------- --- --- wild-type 67 68 1 2 ung1* 59 64 1 0 ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Types of base of substitutions found in can1mutants induced by expression of hAIDSc. ::: Substitution wild-type ung1 -------------- ----------- -------- G→A 19 37 C→T 37 24 G→T 2 0 C→A 4 1 G→C 0 0 C→G 2 0 A→G 2 2 T→C 1 1 A→C 0 0 T→G 1 1 A→T 0 0 T→A 2 0 ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Differences in occurrence of transitions/transversion and mutations in two DNA strands of the CAN1gene in wild-type and ung1strains. ::: Spectra compared wild-type ung1 P~fisher~ ------------------------------ ----------- -------- ------------- Transitions at G-C bases 56 61 Transversions at G-C bases 8 1 0.032 Transcribed strand (G→A) 19 37 Non-transcribed strand (C→T) 37 24 0.005 P~fisher~is the probability that a 2 × 2 contingency table is homogeneous as calculated using Fisher exact test. ::: ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Mutations in different mutable motifs in different spectra. ::: Spectrum Sequence motif (mutable positions are underlined) Reference ----------------------------------------------------- --------------------------------------------------- ----------------- ----------- Spontaneous can1mutations in the wild-type yeast 0.5 / 0.5 0.7/0.5 \[52\] Spontaneous SUP4-omutations in ung1yeast strain 0.9/1.1 0.9/1.1 \[60\] hAIDScin the CAN1in wild-type yeast *1.2 / 5.2* *1.5 / 4.8* This work hAIDScin the CAN1in ung1yeast *2.0 / 1.8* *3.2 / 1.6* This work SHM in V~κ~Ox1 in mouse *3.5 / 2.5* *3.4 / 2.5* \[47\] SHM in J~H~4 region in wild-type mouse *4.4 / 3.2* *4.4 / 3.2* \[26\] SHM in J~H~4 region in Ung1^-/-^mouse *4.5 / 3.6* *4.5 / 3.6* \[26\] SHM in J~H~4 region in Ung1^-/-^Msh2^-/-^mouse *3.3 / 5.0* *3.2 / 5.0* \[67\] AID in GFPin human fibroblasts *7.6 / 4.8* *7.1 / 4.1* \[13\] AID in sacBin E.coli *2.3 / 3.7* *1.6 / 4.6* \[36\] AID in lacZ in vitro *\-- / 2.1* *\-- / 2.2* \[40\] The values listed represent the fold increase in occurrence of mutations at mutable motifs above the average occurrence of mutations at other G-C sites. Number of mutations in mutable motifs was calculated for the underlined bases. Bold italicized values represent a statistically significant correlation (P ≤ 0.05) between a mutable motif and the distribution of mutations, as revealed by using a Monte Carlo procedure \[7\]. ::: ::: {#T6 .table-wrap} Table 6 ::: {.caption} ###### Base substitution hotspots and mutable motifs. ::: Number of substitutions ------------------------- ----------------------------- ----------- -------- ------------------------------------------------------------- Position Sequence Wild-type ung1 N[G]{.underline}YW/WR[C]{.underline}N variant (N = A/T/C/G) 238 GTA [C]{.underline}AGA 4 2 T[G]{.underline}YW 268 AAG [C]{.underline}AAA 7 2 T[G]{.underline}YW 299 GTG [G]{.underline} TAC 1 4 G[G]{.underline}YW 424 GGT [G]{.underline} AAA 0 3 \- 612 ATG [G]{.underline} AAT 0 3 \- 896 AAG [G]{.underline} TAC 3 1 G[G]{.underline}YW 980 TCC [G]{.underline} TAT 2 6 C[G]{.underline}YW 1166 CTG [C]{.underline}CGC 4 0 G[G]{.underline}YW 1392 ATG [G]{.underline}TTA 3 2 G[G]{.underline}YW 1426 ATG [C]{.underline}AAG 1 4 T[G]{.underline}YW 1486 ATG [C]{.underline}CCG 3 0 G[G]{.underline}YW The CLUSTERM program \[97\] was used for hotspot analysis. The hotspot threshold value is three mutations for both spectra. Hotspot motifs are shown in Bold and enlarged font. The hotspot base is underlined. :::	PubMed Central	2024-06-05T03:55:59.449132	2005-6-10	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180437/", "journal": "BMC Immunol. 2005 Jun 10; 6:10", "authors": [ { "first": "Vladimir I", "last": "Mayorov" }, { "first": "Igor B", "last": "Rogozin" }, { "first": "Linda R", "last": "Adkison" }, { "first": "Christin", "last": "Frahm" }, { "first": "Thomas A", "last": "Kunkel" }, { "first": "Youri I", "last": "Pavlov" } ] }
PMC1180438	Background ========== Although problem-based learning has been at the forefront of reforms to medical curricula since its inception more than thirty years ago, conclusive evidence about its effectiveness as an educational approach remains elusive. In addition to research-related issues, such as non-randomised groups and small sample sizes, are variables inherent in the implementation of PBL curricula. These include variability in selection criteria for prospective students and the precise model of PBL employed \[[@B1]\]. An associated issue for which definitive answers have yet to be found relates to the qualifications and backgrounds of staff employed as facilitators in PBL tutorials. The function of the tutor in PBL differs considerably from that of the tutor in conventional tutorials in which the tutor assumes a comparatively didactic role. A major feature of PBL is that learning is student-centred in that students take responsibility for identifying and addressing their own learning needs; tutors are required to facilitate this rather than adopt the position of content expert. Facilitation requires understanding of the learning process and primarily involves monitoring of student learning and promotion of effective group function. The student-centred learning approach of PBL means that for tutors, content knowledge should be subordinate to proficiency in group facilitation \[[@B2]\]. Thus, effective tutors promote student learning by creating a supportive environment which encourages active participation by all members of the group, by monitoring the quality of learning through questions and feedback and by encouraging the development of students\' metacognitive skills \[[@B3]\]. Most of the studies \[[@B4]-[@B8]\] that have looked at the characteristics of skilled PBL tutors have compared the effectiveness of content experts with that of non-experts as measured either by student satisfaction or academic achievement. (In medical curricula, the problems that students address are presented as integrated clinical scenarios. Thus, content experts are seen as those having relevant subject matter knowledge. This refers not only to clinical skills such as history taking and physical examination but also to knowledge of basic science, public health and ethico-medico-legal issues as required by the problem. Thus, although non-clinically qualified academic staff may be seen as having expertise in the teaching/facilitation process and/or particular aspects of the curriculum, when it comes to subject matter knowledge, it is clinicians who are considered as experts.) Although findings from these studies are inconclusive, overall there is a slight preponderance in favour of content expertise over group facilitation skill with regard to both academic achievement and student satisfaction \[[@B4],[@B6]-[@B8]\]. This is not a surprising outcome given the subjective nature of student perceptions of tutor effectiveness, possible conflict between students\' perception of effective tutors and those characteristics essential to PBL (such as student-centredness, a non-didactic approach, emphasis on group function and self-directed learning) and the large number of factors that influence student academic achievement in addition to the skill of their PBL tutor. Findings from research into the relative effectiveness of staff versus student tutors are also inconclusive \[[@B9]-[@B11]\]. Additionally, Schmidt and Moust \[[@B12]\] in a study of PBL in a series of six-week health science courses looked at the influence of tutor style on student learning behaviour and academic achievement. Their findings suggested that the most effective tutors, as judged by the students, were those with both content knowledge and the ability to empathise with their students\' circumstances. As far as we are aware, there have been no studies which have specifically explored the relationship between students\' perception of overall effectiveness, particular tutor behaviours, clinical qualifications and academic appointment. The medical program at the University of Queensland is a four-year, graduate entry program and features a PBL curriculum. The first year of the MBBS has an enrolment of approximately 270 students, divided into 26 PBL groups, and employs a range of PBL tutors over three teaching terms of about 11 weeks each. Although tutors may have medical, basic science or educational qualifications, the majority have expertise in one or other of the basic sciences, reflecting the dominant focus of the First Year curriculum. Tutors may be full-time academic staff or postgraduate students and others employed on a casual basis. All tutors are specifically trained in PBL before appointment to a student group and may teach up to three terms each year. As stated above, tutoring in PBL tutoring has two components: facilitation skill and content knowledge. It may be expected that students would consider the principal strength of clinically qualified tutors to be their greater relevant content knowledge. In contrast, the principal strength of non-clinically qualified academic staff to the PBL process would be the facilitation skills derived from (often extensive) teaching experience. This study therefore has two aims. First, to compare the tutoring style and overall effectiveness (as perceived by their students) of Year 1 PBL tutors based on academic qualifications and appointment category; and second, to determine which of six specific tutor styles contributed most to students\' perception of effective tutoring. The following hypotheses were tested: a) clinicians would be seen as more effective tutors than non-clinicians; b) full-time academic staff would be seen as more effective tutors than non-staff (casually-employed) tutors; and c) students would view subject matter knowledge as a more important determinant of effective tutoring than group facilitation skill. Thus, it is hoped that this study will provide a useful contribution to the PBL literature specifically by allowing both PBL tutors and their students to increase their understanding of the PBL process and how it may be optimally applied in medical education. Method ====== Subjects -------- Subjects were the PBL tutors, each facilitating one of 26 groups (ten students per group) for an 11-week term in first year of the MBBS Program at the University of Queensland. Forty-two tutors were employed over the three terms of the teaching year, with each teaching an average of 1.90 groups. All tutors were given the opportunity to participate in this study and 40 agreed to evaluation by their students. Of the tutors, 26 were basic scientists, eight were medically qualified and six were classified as \"other\" having a social sciences background, predominantly in education. Additionally, 14 held full-time academic staff appointments (two clinicians, seven basic scientists and five \"others\") and 26 held non-staff appointments of which 14 were employed on a casual basis (five clinicians, 8 basic scientists and one teacher) and 12 were basic science PhD students, one of whom was also medically qualified. Thus, tutors were classified on the basis of both qualifications (clinical versus non-clinical) and type of appointment (staff versus non-staff) Student participation was also voluntary. Students were allocated to PBL groups following enrolment and remained in the same group for the whole of First Year. Allocation was not random but was structured to ensure minimum variation between groups with regard to age, gender, academic background and nationality. Because the year consisted of three, eleven-week terms, each PBL group experienced three different tutors and thus provided three sets of data each. The total number of first year students was 270 allocated to 26 groups making the maximum number of student responses 810. However, because not all tutors agreed to participate, total of 702 student responses from 76 PBL tutor evaluations was received (overall response rate of 86.7%). Instrument ---------- The questionnaire used was developed and validated at Maastricht University \[[@B10]\]. This instrument was chosen because it looks at specific tutor characteristics important to the PBL tutorial process. It consists of 39 items which explore tutoring style with respect to Knowledge of Subject Matter and Skill in Facilitation using a Likert-type rating scale. Within each of these categories are three scales which assess corresponding sub-categories of behaviour. The resultant six scales are: ### (a) Knowledge of Subject Matter • [Use of expertise (UE):]{.underline} the degree to which the tutor uses his/her knowledge of relevant subject matter to help students (e.g. \"The tutor used his/her subject-matter knowledge to help us\"). • [Cognitive congruence (CC):]{.underline} the degree to which the tutor is able to understand, and express him/herself at the students\' level of knowledge (e.g. \"The tutor had no difficulty understanding the group\'s problems with the subject-matter\"). • [Test orientation (TO):]{.underline} the degree to which the tutor focuses on summative assessment to direct the students\' learning (e.g. \"The tutor mentioned subjects we certainly had to know for the assessment of this course\"). ### (b) Facilitation Skill • [Authority (AU)]{.underline}: the degree to which a tutor uses his/her authority to direct students\' activities within the group (e.g. \"S/He intervened in ways that disturbed the progress of the group discussion\"). • [Role congruence (RC)]{.underline}: the degree to which a tutor is able to empathise with, and relate to, students\' lives (e.g. \"The tutor understood the problems first-years have with their study\"). • [Cooperation orientation (CO)]{.underline}: the degree to which a tutor is interested in encouraging cooperation among members of the group (e.g. \"At regular intervals, the tutor evaluated with us the group\'s functioning\"). The items were randomised in the order in which they occurred in the questionnaire except for the last item which asked students to rate how well the tutor played his/her role overall. This last item was thus considered to be a measure of the students\' perception of their tutor\'s overalleffectiveness. Analysis -------- Participating tutors were evaluated by each of the students in their PBL groups at the end of each teaching term, resulting in a maximum of 20 evaluations per tutor per term depending on the number of PBL groups they facilitated. Completed responses were scanned and collated into a spreadsheet containing data on each tutor\'s qualifications and type of appointment. Although no effect due to tutor age and gender was expected, this information was also included to establish if age and gender exerted a confounding effect on the results. The data were analysed using the SPSS^®^program \[[@B13]\]. The various items for each outcome variable were described by Schmidt & Moust \[[@B12]\], However principal components analysis was used to confirm the correct loading onto each of the items into the appropriate outcome variable (use of expertise, cognitive congruence, test orientation, authority, role congruence, cooperation orientation and overall effectiveness, as perceived by the students). These outcome variables formed the basis of all subsequent data analysis. Of the explanatory variables, age was transformed into a categorical variable: 20--39 years and 40+ years. The other categorical variables were gender (M or F), qualifications (clinical, basic science or other) and appointment type (staff, student or casual). Initial analysis indicated that within the latter two variables, the only significant differences were between clinicians compared to non-clinicians and between staff compared to non-staff. Consequently, these variables were recoded (as clinician or other, and staff or non-staff) for all subsequent analyses. Finally, because students were allocated to PBL groups in a way that ensured that each group contained a similar mix of students with regard to age, gender, academic and cultural background, there was relatively little variation between PBL groups compared to the variation between students within each group. Hence, the PBL group for each term was regarded as the experimental unit and each group\'s mean scores rather than individual responses were used for all analyses. The reliability of each of the scales and of the items was calculated using Cronbach\'s alpha coefficient of internal consistency. For this type of analysis, an alpha coefficient \>0.70 indicates acceptable reliability. Univariate analysis was performed for each outcome variable using one-way ANOVA to assess the relationship between categorical and outcome variables. p-values \<0.05 were considered to indicate a statistically significant difference. Subsequently, general linear modelling was used to analyse the influence of characteristics which univariate analysis had shown to have a significant relationship with tutoring style. The final model was selected by backwards elimination until all remaining terms were significant. The model was then screened for interactions between these terms. As no interactions were significant, and adjustment for other terms in the final model was minimal, results are presented for the one-way ANOVA for each explanatory variable. Results ======= A summary of the analysis of the data is given in Table [1](#T1){ref-type="table"}. This indicates that the reliability of the instrument varied between scales from 0.55 (Authority) to 0.95 (Use of Expertise). Of the six scales measured, only Authority (0.55) and Role Congruence (0.57) fell below 0.70 while the reliability of the instrument, over all 39 items, was 0.92 which is acceptable. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Evaluation of the reliability, correlation with perceived overall effectiveness, mean and standard deviation of each of the measured scales ::: -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- UE\ CC\ TO\ AU\ RC\ CO\ Overall Effectiveness\ Max = 36 Max = 28 Max = 16 Max = 20 Max = 24 Max = 24 Max = 4 -------------------------------------------- -------------------- ------------ ------------ ------------ ---------- ------------ ------------ ---------------------------- Reliability 0.92 0.95 0.85 0.55 0.57 0.91 0.70 Correlation with Overall Effectiveness Correlation 0.79 0.67 0.52 -0.06 0.77 0.69 p-value \<0.01 \<0.01 \<0.01 0.63 \<0.01 \<0.01 Corrected Model^1^ Grand mean^2^ 24.9 21.2 11.2 13.4 18.0 20.8 3.5 Standard Deviation 3.1 2.7 1.4 2.3 2.4 2.4 0.6 -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ^1^Corrected for attenuation ^2^The mean of all scores in all groups ::: Correlation coefficients were calculated to determine the strength of the association between students\' perception of their tutor\'s overall effectiveness and each of the tutor behaviours assessed. With the exception of the scale measuring use of authority, all scales showed a significant, although variable, association with overall effectiveness score. Use of expertise and role congruence correlated most strongly. When the multivariate relationship between individual scales and overall score was analysed, there was a close association between five scales (Use of Expertise, Cognitive Congruence, Test Orientation, Role Congruence and Cooperation Orientation). A final regression model retained only Use of Expertise (β-coefficient = 0.53, p \<0.01), Authority (β-coefficient = -0.20, p \<0.01) and Role Congruence (β-coefficient = 0.39, p \<0.01). Table [2](#T2){ref-type="table"} summarises the results of the univariate analysis of tutor characteristics and compares the average scores for each scale based on gender, age academic qualifications and appointment type. These results indicate that there was no significant difference between male and female tutors on any of the scales despite a consistent trend for males to have higher scores. With regard to ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Comparison of scale mean scores on the basis of gender, age, academic qualifications and appointment type. ::: Outcome variable Scale UE CC TO AU RC CO Overall Effectiveness ---------------------- ----------------- ------------ -------- ---------- -------- ---------- ---------- --------------------------- Gender Female N = 57 23.3 20.5 10.8 13.0 17.4 20.2 3.4 Male N = 19 24.9 20.9 11.4 13.4 17.8 20.8 3.5 p-value 0.07 0.56 0.09 0.47 0.56 0.33 0.28 Age 20--39 N = 23 23.1 20.1 10.4 12.7 17.6 20.1 3.5 40+ N = 53 24.0 20.8 11.2 13.2 17.5 20.4 3.4 p-value 0.26 0.31 0.03 0.41 0.84 0.67 0.44 Qualifications Clinical N = 15 25.9 21.5 11.3 13.7 18.6 20.9 3.6 Other N = 61 23.2 20.4 10.8 12.9 17.2 20.2 3.4 p-value \<0.01 0.19 0.30 0.21 0.05 0.27 0.21 Appointment Casual N = 51 23.3 20.2 10.7 13.0 17.4 19.9 3.3 Staff N = 25 24.5 21.4 11.4 13.3 17.8 21.1 3.5 p-value 0.17 0.09 0.03 0.61 0.53 0.04 0.22 ::: • [Age]{.underline}: Older tutors tended to have higher scores on all scales except Role Congruence. However, this difference only reached statistical significance for Test Orientation (p = 0.03), which suggests that older tutors emphasise assessment as a means of encouraging group learning more than younger tutors. • [Qualifications]{.underline}: Clinicians tended to have higher scores on all scales than non-clinicians but these were only significant for Use of Expertise (p \< 0.01) and Role Congruence (p = 0.05). • [Appointment]{.underline}: There was a tendency for staff tutors to have higher scores on all scales than non-staff tutors but these only reached statistical significance for Test Orientation (p = 0.03) and Cooperation Orientation (p = 0.04). That is, staff tutors focussed more clearly than non-staff tutors on assessment to motivate learning and were better able to promote effective group function. Discussion and Conclusion ========================= Full-time academic staff are more likely than non-staff (casual and postgraduate student) tutors both to emphasise summative assessment as a factor in learning, and to display greater skill in establishing and maintaining a team approach among their students. These findings may be due to the considerably greater time that faculty spend in teaching and assessment tasks in general, as well as experience in dealing with student issues relative to casual tutors and postgraduate students. Although age was shown not to have a significant impact on any of these behaviours, it is worth noting that most (85.7%) staff tutors were in the 40+ age group, compared to 53.8% for non-staff tutors. Using the same instrument, Moust and Schmidt \[[@B10]\] investigated differences in tutoring style between staff and non-staff tutors. They found that staff tutors scored significantly lower than non-staff tutors in three of the six behaviours: cognitive congruence, test orientation and role congruence, with staff tutors scoring higher on the authority scale only. This is in contrast to our findings which found significant differences in only two of the scales (and those in favour of staff tutors), test orientation and cooperation orientation. The discrepancy between the two studies may be explained, at least in part, by the different composition of the non-staff group between the two studies. The non-staff group in our study was more heterogeneous, consisting of both casual and student tutors, thereby blurring the distinction between staff and non-staff, particularly with regard to age -- only 8.3% of student tutors were over 40 years compared with 85.7% of casual tutors. With regard to clinical versus non-clinical tutors, medically-qualified PBL tutors are significantly more likely to make use of their expertise in facilitating their groups\' learning and to empathise with their students\' lives as medical students. This is not surprising, given their own prior experience as medical students themselves and that they can be assumed to also have a large body of clinically-relevant knowledge which they are able to apply to addressing the problem being studied. None of these findings are reflected in the students\' ratings for the last item in the questionnaire: how well they perceived that their tutor played his/her role (overall effectiveness). This may be due to the relative crudity of the four-point scale for each item in the questionnaire. Such an effect would be particularly noticeable with regard to the item on overall effectiveness which has a maximum score of only four points, as well as a relatively higher standard deviation than those for the other scales. Additionally, when taken in conjunction with the finding of significant correlations between five of the six behaviours and perceived effectiveness, it may be that characteristics other than those measured in this study also contribute to the students\' perception of their tutor\'s effectiveness. Nevertheless, the lack of a significant difference between clinicians and non-clinicians in students\' ratings for overall effectiveness, given that they rate clinicians\' use of expertise significantly higher, is striking. This is especially so in view of the substantial correlation between use of expertise and overall effectiveness, and raises the question of what exactly is meant by the term \"expertise\" in relation to problem-based learning. The issue of whether content experts are more effective as PBL tutors than those without content expertise has been, and continues to be, a subject of debate in the literature with many studies finding in favour of the experts \[[@B4],[@B7]\]. In research into PBL in integrated medical curricula, the experts are assumed to be clinicians, with basic scientists and others designated as non-experts for the reasons discussed earlier. However, it is possible that, in First Year of the MBBS where the principal emphasis is on the basic sciences, expertise in the basic sciences is seen as an equally valuable characteristic in PBL tutors as clinical expertise. In contrast, the Second-Year curriculum has a much greater clinical focus and the majority of PBL tutors are medically qualified. A follow-up study of tutors in Year 2 is planned. An alternative explanation is that it is the way a tutor uses his/her expertise, rather than the degree of expertise, which is important in determining a tutor\'s score on this scale. That is, it is possible that clinicians adopt a tutor-centred, rather than student-centred approach to PBL tutoring but that this characteristic is not regarded by students as having a major impact on their tutor\'s effectiveness. Indeed, a recent study has shown that clinicians tend to ask questions directly of students, whereas non-clinicians are more likely to encourage students to ask questions of each other \[[@B6]\]. The same argument can be applied to the finding regarding staff versus non-staff tutors. That is, that although full-time staff are more test-oriented and better at establishing a collaborative environment for student learning, these differences are not enough in themselves to impact significantly on overall effectiveness, again suggesting that characteristics other than those measured at least partially influence this assessment. In summary, these findings suggest that, although clinicians and staff tutors consistently scored higher on each of the measured behaviours than did non-clinicians and non-staff, most of these differences are not statistically significant and do not have a substantial impact on students\' assessment of their effectiveness as PBL tutors. This conclusion has implications for the recruitment and training of PBL tutors in that it suggests that both subject matter knowledge and process facilitation skills are necessary but not individually sufficient characteristics of effective tutors. Developing a broad range of strategies to encourage optimal group functioning and to stimulate student learning should therefore be a major focus of tutor preparation. Competing Interests =================== The author(s) declare that they have no competing interests. Authors\' Contributions ======================= MG conceived the study, participated in its design and coordination, contributed to the analyses and drafted the manuscript. PR participated in the study design and coordination, carried out the statistical analysis and approved the final draft. PO\'R carried out the statistical analysis and approved the final draft of the manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6920/5/20/prepub>	PubMed Central	2024-06-05T03:55:59.452940	2005-6-7	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180438/", "journal": "BMC Med Educ. 2005 Jun 7; 5:20", "authors": [ { "first": "Michele", "last": "Groves" }, { "first": "Patricia", "last": "Régo" }, { "first": "Peter", "last": "O'Rourke" } ] }
PMC1180439	Background ========== The last ten years have seen a global change in the methods used for the education and training of medical students \[[@B1]-[@B3]\]. In part, this has been due to the concurrent change in the operations of the tertiary level hospital system, traditionally the key learning site for the medical student. Hospitals are increasingly becoming short stay institutions with a focus on same day treatments and early discharge, with only the more seriously sick patient staying for a longer time \[[@B3]-[@B5]\]. This change of priorities has meant that students in tertiary teaching hospitals no longer have access to a full range of educational experiences as many of the less severe illnesses and other chronic illness move to the primary health care system \[[@B5]-[@B7]\]. One result of these changes has been a large scale move towards increasing community based medical education within the medical schools\' curriculum, with students attached to a community based general / family practice \[[@B3],[@B6]-[@B8]\]. Although these are usually relatively short placements of approximately 8 to 12 weeks \[[@B4],[@B9]\], some medical schools have been providing a full year\'s study in the community \[[@B2],[@B5],[@B10],[@B11]\]. An additional benefit of placement in the community is the increased exposure to general / family practice. When the community is in a rural area, this has a positive influence on choice of rural practice as a career choice, an area of significant shortage in many countries \[[@B9],[@B12]-[@B14]\]. There are also positive benefits to established rural practitioners, with enhanced job satisfaction and consequently a positive impact on rural retention \[[@B7]\]. Initial concerns of both students and teachers experiencing long term placements in the community focused on the potential disadvantage to students because of the uncertainty of whether they would see a broad spectrum of clinical conditions through the community patient base to achieve their learning objectives \[[@B8],[@B10]\]. However, there are also advantages such as experiencing longitudinal care \[[@B5],[@B11]\], the opportunity to see early presentation of conditions, preventative medicine and an understanding of the psychosocial aspects of primary health care \[[@B5]\]. Community-based students also have the advantage of small group learning as they are generally working in groups of 2 to 4 students \[[@B2],[@B11]\]. This obviates concerns regrading peer-to-peer interaction \[[@B15]\]. Early concerns regarding access to resources have by and large been resolved with appropriate Internet technology and equipment being made available within the community teaching sites \[[@B16],[@B17]\]. Recent studies examining assessment results indicated that students in longitudinal community placement performed as well, if not better than their hospital-based counterparts in acquisition of clinical skills, including those associated with specialist disciplines \[[@B2],[@B11]\]. Alderson and Oswald reported that students recorded a large number of patient contacts with varied clinical conditions representing problems in all major specialty areas \[[@B8]\]. Hence, earlier studies have successfully demonstrated that medical students can achieve success in core clinical rotations with long term attachments in small groups to rural general / family practices. The aim of this pilot study was to assess the feasibility of successfully completing the five third year core clinical rotations while based alone without their peers in a relatively isolated rural community general / family practice, with success measured though academic achievement. Methods ======= Setting ------- The University of Queensland medical course is a four-year graduate entry program culminating in a Bachelor of Medicine and Bachelor of Surgery. The overall course plan is detailed in table [1](#T1){ref-type="table"}. Students spend the first two years at the university campus in the capital city and then elect to attend one of three clinical divisions. Two are located in the metropolitan area, and one, the Rural Clinical Division, is based over two regions in rural Queensland. Twenty five percent of the School\'s students complete at least one of the last two years of the MBBS course in the Rural Clinical Division, with almost all of these students being volunteers. All terms for a particular year are completed within the chosen Clinical Division. All of the School\'s students are placed in a rural area to complete a third year term in rural health. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Course structure ::: Year Location Subjects ------ --------------------------------------------- ---------------------------------- 1 University: main campus Introduction to Medical Practice 2 University: main campus Foundations of Medical Practice 3 Rural or Metropolitan Clinical School\* Medicine Surgery Mental Health General Practice Rural Health \* 4 centerRural or Metropolitan Clinical School Paediatrics OBGYN Surgical specialties Medical specialties \* Except Rural Health which all students do in rural areas ::: Problem Based Learning (PBL) is a key component of the program at all levels. Using a series of carefully planned patient-centered exercises, the students work cooperatively with their tutors to examine core issues of health and disease. In years three and four, students examine a new PBL case each week. The remainder of their clinical time is focused on supervised learning from their patients, supplemented by clinical teaching, tutorials, case demonstrations and lectures. Adult learning processes are encouraged with students\' teaching and learning driven by their own learning objectives and goals. Key features of the program include life long learning skills including an ability to search and critically analyze information. All students are learning under the same curriculum and participate in the same assessment process, regardless of the geographical location of their clinical experience. A summative mark is generated for each core clinical rotation in third year from an end-of-term test with or without an in-course assessment. From the rotation marks, an average score for the year is produced. The pilot program ----------------- The concept was initially developed at a stakeholder meeting of clinical and academic staff of the Central Queensland Rural Clinical Division held in early 2003. Several rural general practitioners (GP) who were clinical teaching staff at this Division expressed interest in developing and participating in the program. By late 2003, three rural GPs had agreed to participate. In August / September 2003, the second year student body was approached at a series of public meetings and offered participation in the new program. Those students who were interested were asked to provide a confidential written expression of interest, explaining their motivation to participate, although there were no formal criteria to evaluate this process. The three students who expressed interest attended a weekend orientation program visiting all three prospective locations, meeting with their prospective GP supervisors, as well as the rural teaching hospital. All three subsequently agreed to volunteer for the program and self selected their site for attachment for 2004. Program coordination was by a dedicated team based at the Division\'s rural teaching hospital. The attached academic staff also provided support both by telephone and regular visits. Each student was attached to a different private solo rural general practitioner / family doctor for one academic year. Each student\'s home base was their own consulting room located within the community based General Practice. The practices are located in relatively isolated and geographically separated communities, all a considerable distance from each other. They are between 170 and 270 km from the Rural Clinical Division teaching hospital. The demographics of these locations are detailed in table [2](#T2){ref-type="table"}. The locations are rated 5 or 6 on the RRMA scale (Rural, Remote, Metropolitan Classification: capital city = 1, other metropolitan = 2, large regional city = 3, most remote community = 7 \[[@B18]\]). The towns had respective populations of 500, 2000 and 10,000. One practice had a GP registrar and in one other location there were doctors outside the teaching practice but in the same town. All 3 supervisors had hospital admitting rights to a community hospital within their town. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Demographics of pilot study locations ::: Location RRMA \[18\] ARIA \[18\] Distance to (km): Public Hospital Facilities (2002 -3) \[24\] ---------------------------- ------------- ------------- ---------------------------- --------------------------------------------- ------------------------------ ---------------- ----------------------- -------------------------------- ----------------------------- Division Teaching Hospital Metropolitan Capital Nearest Pilot Study Location Available beds Medical staff \[FTE\] Visiting medical staff \[FTE\] Admitted patients (in 2002) A 5 4.03 172 640 78 10 1 0 463 B 5 4.79 210 564 78 10 1 0 291 C 6 5.06 271 896 289 37 5 3 2967 Division Teaching Hospital 3 1.49 N/A 648 172 203 73 16 21,395 RRMA: Rural, remote, metropolitan areas classification ARIA: Accessibility / remoteness index FTE: Full Time Equivalent ::: The educational process was similar to that of the teaching hospital, being based on supervised learning from patients, clinical teaching, tutorials and case demonstrations. In common with their peers at other geographical sites within the School of Medicine, the program was based on the adult learning model, requiring participating students to exercise a significant degree of self reliance and initiative. Although not essential, being a graduate rather than undergraduate program combined with an extensive use of the PBL mode of learning would have assisted this process. A notable difference was that only 25% of the clinical material was seen at the hospital interface. Each student participated in the care of a large variety of patients, experiencing in any given week problems covering all five core clinical rotations. They were also able to follow their patients over time, creating a longitudinal experience. However, for administrative reasons, the students operated within the five-term structure, with separate 8-week blocks for each of the five core rotations. Hence, the weekly PBL tutorials, activities and assessment tasks were aligned with the designated term, an essential process to keep the curriculum and assessment process uniform across all geographical locations within the School of Medicine. Although the GP principal in each of the practices was the primary teacher / mentor, the student\'s learning environment was enhanced by interactions with other locally based doctors, visiting specialists and allied health professionals. PBL tutorials were conducted by teleconference between the pilot study students and the designated tutor (either one of the GP Principals or a teacher from the Division\'s teaching hospital). A separate teleconference between the three students, but without supervision by the academic, teaching or administrative staff was conducted weekly to encourage peer-to-peer discussion and support. Each student also rotated to the Division\'s main teaching hospital or a second teaching hospital of the Division for 3--5 days per month to participate in either scheduled teaching activities with other students or one to one teaching with the hospital based specialists. This arrangement was flexible dependent on the needs of the students, in keeping with the philosophy of self-directed learning. Each student was provided with high-speed Internet access. This gave them the same access as students elsewhere to the Internet based resources of the university, where the PBL and other learning resources are located. Each student was issued with a hand held computer and associated software, providing access to information within the consultation process and for diarizing their learning experience. This personal learning tool, akin to a learning journal was for personal self learning and development, rather than program evaluation. At no time did academic staff have access to this information. This learning tool was not provided at other geographical locations. The summative assessment process -------------------------------- In order to ensure the comparability of the pilot study students\' marks with their rural and metropolitan counterparts, the assessment process was uniform across the school regardless of geographical location. Although the pilot study students were in essence in a longitudinal rather than term based program, their assessment followed the term process. In-course assessment was completed on a nominal term basis, in the sequence rural, general practice, mental health, surgery and medicine. In contrast, end of course assessment could be attempted in any order chosen by the student, as long as they completed the process on the same day as their peers elsewhere and completed exactly the same assessment. All students chose to sit the surgery assessment at the end of term 4, the medicine assessment at the end of term 5, but varied in their timing of the other 3 assessments. The summative assessment process was essentially similar across the third year curriculum, combining a series of in-course assessments with a final written test in each of the 5 terms. Learning and assessment programs included knowledge, skills and attributes in four broad areas titled \'Domains\': (1) basic and clinical sciences, (2) interpersonal and clinical skills, clinical reasoning and practice, (3) population and preventive health, and (4) ethics, personal and professional development. Hence over the course of the year, each student was extensively and repeatedly assessed in a variety of formats and settings, looking at a broad range of learning issues beyond merely acquisition of knowledge. Planning and implementation --------------------------- The entire program was funded under the Rural Clinical School initiative of the Australian Government Department of Health and Ageing \[[@B19]\] and managed by the School of Medicine, University of Queensland. The preceptors were financially supported for teaching through both the Rural Clinical School initiative and a separate Australian Government program supporting teaching by General Practitioners. The major concern during planning was to find appropriate preceptors as this program was heavily dependent on the active and dedicated participation of the preceptor in an intensive, one to one teaching and mentoring relationship. However, in the planning stages, supervisors who were essentially self-selected came forward and after the academic credentialing processes were appointed. Each supervisor was a dedicated rural doctor with greater than ten year\'s continuous service in that particular location. They all had extensive procedural experience although they utilized these skills to varying degrees. Each had been a supervisor of General practice trainees over many years, had supervised medical students in the past and were known to be effective teachers. Orientation and ongoing oversight of the supervisors was performed by the program\'s dedicated part time academic supervisor who made frequent contact with both the students and their preceptors, both by telephone and regular face to face meetings. He was supported by part time administrative support from the student affairs officer, finance officer and regional school manager. In order to physically accommodate students in the rural preceptors\' practice, structural changes were required. These were jointly funded by the University and the preceptors. Unfortunately, the building / renovation program was not completed until after the academic year began. Student accommodation was sourced from the local real estate agents and funded by the University. Analysis -------- The Statistical Package for the Social Sciences was used for analysing the results \[[@B20]\]. Each student was awarded a final score for each of the five core clinical rotations, the average being the third year\'s global score. Marks were compared across geographic locations by analysis of variance, with Tukey post-hoc analysis. Statistical significance was defined as p \< 0.05. Results ======= All three students in this pilot study completed the full year. The marks for each rotation are detailed in table [3](#T3){ref-type="table"}. All three students in the study passed each term and the year with marks well above the nominal pass mark of 60, as did all their colleagues at other rural teaching hospitals. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Academic performance \[for all third year students\] ::: n Medicine Surgery Mental Health General Practice Rural Health Global Score ----------------------- --------- ---------- --------- --------------- ------------------ -------------- -------------- ------- --------- ------- --------- ------- --------- Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Mean Std Dev Pilot study 3 70.33 2.52 76.90 2.95 75.63 5.98 81.47 2.15 78.00 1.32 76.47 2.02 Rural: hospital based 47 65.62 8.56 71.77 5.93 79.15 8.4 82.94 6.34 79.04 6.25 75.70 4.96 Metropolitan 174 68.25 9.42 73.44 5.25 76.02 9.83 81.50 7.1 79.15 8.4 75.65 5.04 Results are percentages. The assessment process is described in detail in the text. ::: There were no statistically significant differences in marks in any of the between the three students in this pilot study and those students in the rural teaching hospitals or metropolitan hospitals. Student evaluation data showed no variation by geographic location in satisfaction with the course. However, there were differences in experiences between the pilot study students and their hospital based counterparts. Discussions with the students and their supervisors suggested that while the overall range of clinical conditions seen was essentially similar to their hospital based counterparts, there was a distinct frequency bias towards community based chronic disease. Pilot study students had the opportunity to participate in continuity of care, following the progress of individual patients over a considerable period of time. They were also able to follow their patient across multi-professional boundaries, from the community to the local hospital and in some cases accompany them to the regional hospital and participate in the surgical care they received there. On completion of the year, all 3 (100%) students remained in the same geographical region and attended the rural clinical school. In comparison, only 7 of the 22 (32%) students attending the rural clinical school in this region in year 3, continued to stay for year 4. Discussion ========== This pilot study has demonstrated that it is possible for medical students to successfully complete the five core clinical rotations in an isolated rural general / family practice environment. However, addressing a number of apparent issues would assist further development of the program. The design of this pilot study exposed participating students to unusual levels of isolation, especially from their peers. Research into peer support has demonstrated that peer support helps to reduce student anxiety and stress and may improve students\' confidence \[[@B15],[@B21]\] and that students are more likely to turn to classmates for support and help when difficulties arise \[[@B22]\]. It is also well recognised that one of the key concerns for undergraduate students undertaking a rural placement rotation is the geographical distance from their teachers and peers, a barrier that can result in physical and social isolation as well as the loss of peer-based learning opportunities \[[@B16]\]. A variety of solutions to isolation were attempted in this pilot including Internet access, regular teleconferences and visits to the rural teaching hospital. The provision of Internet access has been shown to help relieve feelings of isolation during rural placement as a medical student \[[@B17]\], although there is a lack of robust information evaluating electronic student support groups and peer-to-peer communication \[[@B23]\]. Expansion of peer support using technology, eg real time multipoint video conferencing, may further decrease the sense of isolation. Unlike traditional teaching environments where a variety of clinical teachers contribute to teaching in any given subject, in this program a single general / family practitioner was given the onerous duty of providing the vast majority of the teaching. This was performed on a one to one basis in the context of isolated and usually very busy rural practice. Solo rural doctors could find the extra work associated with teaching to be unsustainable. However, unlike a traditional 4--6 week placement, the longitudinal nature of the program can deliver benefits to the practitioner\'s medical practice. Over time, the supervising GP develops a clear understanding of the strengths and weaknesses of the student and is able to allocate tasks for the student to perform under supervision that while facilitating learning are also service tasks. This can temper the workload of the supervisor, similar to the balance between service and teaching commonly seen in the post registration medical education environment. As one person provides most of the teaching in an environment with few if any other doctors, there is no real provision for backup if the primary teacher is incapacitated or absent. Although a temporary doctor may be found at short notice to provide patient services, it is unlikely this person would be equipped to seamlessly continue the teaching program. Utilizing group practices may be one solution to this dilemma, although these were generally unavailable in the pilot study region. Although the course was offered to 226 students, there were only three volunteers, a number quite acceptable for a pilot study of this nature. However, in a long term sustainable program of this nature, larger student numbers would be required to ensure a steady supply of students to interested practices, as there is the potential to gain or lose students at different points during the clinical years due to other factors, e.g. sickness, failure, changed family circumstances, etc. There may be economies of scale to be considered, considering the substantial infrastructure support required for the program, as well as enhanced opportunities for peer to peer interaction by distance with a larger cohort. Conclusion ========== This study has demonstrated the feasibility of teaching core clinical rotations in isolated rural areas. To maintain confidentiality, participant characteristics could not be described. Nonetheless, program effectiveness may be related to the characteristics of the self-selected participants such as gender and previous experience living and working in rural areas. Further studies, therefore, should be designed so that information about participant characteristics can be collected, reported and examined. These could also address the issues of student acceptance, teacher burnout and cost structure to assess the viability of expanding this program to a wider audience. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= SM shared executive responsibility for managing the project, analysed the results, prepared and reviewed the manuscript. LD was a clinical tutor, participated in the conception, shared executive responsibility for managing the project, prepared and reviewed the manuscript. VY analysed the results, prepared and reviewed the manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6920/5/22/prepub> Acknowledgements ================ The authors gratefully acknowledge Dr John Birks who established this project, the team of dedicated educators who contributed to this project and the students who volunteered to participate. The Department of Health and Ageing, Australian Government funded the study, through recurrent funding to the rural clinical school program.	PubMed Central	2024-06-05T03:55:59.456160	2005-6-27	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180439/", "journal": "BMC Med Educ. 2005 Jun 27; 5:22", "authors": [ { "first": "Stephen A", "last": "Margolis" }, { "first": "Llewellyn M", "last": "Davies" }, { "first": "Valmae", "last": "Ypinazar" } ] }
PMC1180440	Background ========== In the modern teaching environment, the small group tutorial is favoured over lectures. A well run tutorial is stimulating and involves frequent participation from all members within the class in a non-threatening environment. Perhaps the main advantage of a small class size is the ability to incorporate a number of active learning strategies. The inclusion of discussion methods can lead to improved student performance \[[@B1],[@B2]\], particularly when considering retention of knowledge, critical thinking and problem solving \[[@B3]\]. Unfortunately, it is a common observation that tutors tend to use the same teaching methods whether they are delivering lectures or running tutorials. This can result in dry and lifeless tutorials, devoid of any student participation and wasting the great potential for active learning implicit in the format. Several factors influence the dynamics of a tutorial. The style and attitude of the tutor is critical and those who seek to engage their group in a non-threatening manner are more likely to be rewarded by increased group participation. The larger the group, the more difficult will be the engagement. This is partly because of the difficulty in involving many people in one discussion, but also the reluctance of individuals to potentially be embarrassed in front of their peers. Students need to feel that they can trust the tutor and their fellow students before they will contribute \[[@B4]\]. Another consideration is the cultural background and gender of the students, which has a bearing on the nature of participation in groups \[[@B3],[@B5]\]. The mix of tutor and student personality as well as culture may have a considerable impact on the dynamics of any tutorial in a multicultural environment and even with the best endeavours of the tutor, getting contributions from all members of the group may be exceedingly difficult. Many strategies exist to try and improve individual participation. We have undertaken a study to assess the place of an electronic voting system in the dynamics of a tutorial. We have set out to observe the effectiveness of such a system during a tutorial, to record the opinions of the students on the value of the system and to measure what was learnt and understood during the tutorial. At its simplest, an electronic voting system (EVS) consists of a master control unit and a series of hand held devices, which communicate with the controller. Audience members, students in this case, push one or more of these buttons in response to a question posed by the tutor. A computer allows the tutor to examine and display the information gained from the hand held units. In this way, the tutor and students receive instant feedback about the student responses to the question. The hypotheses to be tested are that the introduction of an electronic voting system (EVS) into a class environment enhances enjoyment of the learning experience and improves understanding and retention of knowledge of the material presented using the system in both the short and long-term. Methods ======= 102 students enrolled in the study. All of the students were in the age range 21--23 and had completed three years of a six-year undergraduate medicine programme. This was their first year of clinical experience. All students in the study came from one academic year and represented 93% of the students in that year. The students were randomly allocated into one of six groups and stratified for gender and academic ability (i.e. previous academic performance in the medical course). The randomisation was carried out by the secretariat in the medical school, who had no connection with the study. In order to verify that this randomisation had been carried out effectively, a one-way ANOVA was used to look for differences between groups in the areas of gender and academic ability. Academic ability was measured using the previous years exam results. The study was carried out at six different times during the year as groups of 13--21 students rotated through the surgical component of their year. It consisted of a pre-test evaluation of the students\' prior knowledge and understanding of two topics followed by a tutorial on each topic. At the end of each tutorial there was a post-test evaluation and a questionnaire. Six weeks after the tutorials, the post-test evaluation was repeated. The tests for each topic had two components; a knowledge test of 11 multiple-choice questions and two multi-stage problem-solving questions designed to measure understanding. The tests did not contribute in any way to the overall student assessment for their degree. The two tutorials were on the topics of acute abdominal pain and gastrointestinal haemorrhage. A series of clinical simulations were constructed and used as the basis for each tutorial. The simulations were computer-based and delivered using a data projector, and a computer program, Medici, which has been described previously \[[@B6],[@B7]\]. Each tutorial was approximately 45 minutes long and covered three different clinical scenarios. The use of the computer material was similar for the two tutorial subjects, with or without the use of EVS. Due to technical issues, the EVS was always applied in the second half of the teaching session. Students were randomised into two groups. The randomisation was done independent of the study and stratified students for gender, ethnic origins and academic ability. One group had the topic of acute abdominal pain discussed in tutorial format alone followed by the second topic of gastrointestinal haemorrhage, which was a tutorial supplemented with an electronic voting system. In the other group, the topic of gastrointestinal haemorrhage was discussed in tutorial format with the acute abdominal pain topic tutorial following and supplemented with an electronic voting system. The electronic voting system was developed by one of the authors (DM) and has been used as part of a telemedicine system to aid postgraduate training in obstetrics \[[@B8]\]. Each member of the tutorial group was provided with a hand-held keypad and asked to select the number corresponding to their choice at various stages of the tutorial. Each keypad was connected by wires to a central control box, and the accumulated data fed into a computer. A software program collated and displayed in histogram form, the accumulated responses for each choice provided by the students. The case simulations in the Medici program provide a clinical scenario on which discussion develops. Several diagnostic or management options are displayed and the members of the tutorial group invited to use their keypads to make a selection. The combined group selections are then displayed with the EVS software. Further discussion promoted by the tutor then takes place, based on the histograms. The behaviour and interactions between students was formally monitored by one of the authors (EP) during the tutorial sessions. A second independent observer was used to check the results of the primary observer. The results of these observations were checked at the end of each session. Where differences were noted, the results of the observer closest to the interaction were used. The number of interactions between lecturer and students and the level of each interaction was recorded and classified according to the schema shown in Table [1](#T1){ref-type="table"}. A Level 2 interaction would be considered a baseline for a meaningful student-lecturer dialogue. During 5-minute \'snapshots, other student activity (writing notes, talking between themselves) was recorded. Also noted were the behaviour of the class as a whole and the attitudes of students and lecturer. At the end of each tutorial, the students\' attitudes towards the use of the EVS were measured. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Schema for assessing levels of interaction ::: --------- ----------------------------------------------------------------------------- Level 1 Simple Interaction eg a yes/no answer to a question from a student or tutor Level 2 A short answer (\<1 min) to a question from either student or tutor Level 3 An extended interaction of at least 1 minute creating further discussion --------- ----------------------------------------------------------------------------- ::: The instrument used to measure the student\'s opinion of the introduction of the EVS was a six-item questionnaire. Each item was scored on a five-point scale (strongly disagree, disagree, no opinion, agree and strongly agree). It measured the reaction of the student to the use of the EVS (was it worthwhile, did it aid participation, was it a distraction, did it aid understanding). The scale was computed by summing the ratings for the six items. The internal consistency of the scale was determined using the Cronbach\'s Alpha statistical test (SPSS) Comparisons of the results of observing student dynamics were carried out using the Mann-Whitney test. The results of the pre, post and long-term tests of knowledge and understanding were compared using a mixed model repeated measures analysis. Pairwise comparisons of the impact of EVS at each combination of time (pre-test, post test and 6 week test) and order (acute abdomen (no EVS), gastrointestinal haemorrhage (EVS) or gastrointestinal haemorrhage (no EVS), acute abdomen (EVS)) were made using Student\'s t-test. The impact of gender of students and their educational status (international student or Australian student) was assessed. Results ======= In the six groups, there were a total of 102 students. 99 students completed the questionnaire and 86 completed all three elements of the pre- and post-testing. The groups were statistically identical with regard to gender (p = 0.92) and academic ability (p = 0.54) as indicated by one-way ANOVA. Student dynamics ---------------- In the standard tutorial environment, the students began to participate slowly but engaged with the lecturer more often as it progressed. Students would often begin by writing, but this would diminish as the tutorial progressed. By the time the second case in the standard tutorial had begun, the class would be participating at its peak. Group responses to questions and small to medium scale student-student interactions were common. By the beginning of the third case, small groups were comfortable discussing issues, but the group itself seemed to tire and interactions with the lecturer lessened. The introduction of the EVS into the second tutorial occurred after a 10-minute break. Students came back refreshed and enthusiastic about using the teaching aid. The interaction with the lecturer ceased at this point and did not return. Students appeared to devote most of their attention to the screen and the EVS controller. Small groups, which had formed in the first tutorial continued to exist and new ones began as students discussed which option was most appropriate to manage the case presented on-screen. This behaviour was consistent throughout the entire second tutorial, although activity did decrease near the end of the last case as students began to tire. Note taking activities throughout this second tutorial were less than in the first. The dynamics of student interactions during the tutorials with and without the electronic voting system is shown in Table [2](#T2){ref-type="table"}. The introduction of an electronic voting system did not affect the length of teaching time required to present material of equivalent length. It did not affect the level of student-student interactions, but did lead to a reduction in the number of students note-taking and the time spent writing. The introduction of the EVS dramatically reduced the number of student-lecturer interactions. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Summary of student tutorial dynamics, shown as medians ::: No EVS EVS ------------------------------------- -------- ----- ------------- Time Spent per Case (minutes) 15 15 ns Students Participating/case 12 10 p = 0.02 Total Writing/case 10 2 p = 0.006 No. Different Students Writing/case 5 2 p = 0.006 Level 1 Interactions/case 4 1.5 p = 0.003 Level 2 Interactions/case 7.5 1 p \< 0.0001 Level 3 Interactions/case 1 0 p \< 0.0001 Student-Student Interactions/case 3 5 ns ::: Test evaluations ---------------- The gender and educational status had no confounding effect and were therefore excluded from the mixed model analysis. ### Multiple-choice questions There was a significant 3-way interaction between time, EVS usage and order (p \< 0.001) however it was not possible to find any effect consistent with the use of the EVS. The adjusted mean MCQ test scores for both tutorials are shown in Table [3](#T3){ref-type="table"}. There was a significant improvement in scores both immediately after and six weeks after the tutorial, irrespective of whether the tutorial included the EVS or not for both tutorial topics. At pre-test for the topic of the acute abdomen, the students who had EVS for this topic had significantly different pre-test scores from those students who did not. This result is likely to be a type II error, as the students had not used the EVS at the time of the pre-test. It is possible that the discrepancy may be due to the reliability of the series of multiple-choice questions used (pre, post and 6 week), which ranged between 0.40 and 0.53. There was a significant difference in Post-Test scores between EVS and no EVS for the gastrointestinal subject, but in light of the unexplained discrepancy at the time of pre-testing, the authors cannot conclude that this represents a real effect. All other comparisons between groups yielded no significant differences regardless of order, nor was there any significant difference in retention (Immediate Post-Test scores subtracted from six Week Post-Test score) between students who had tutorials with and without EVS for both topics. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Number of correct MCQs from 11 questions, shown as adjusted mean (standard error) ::: Subject: Acute Abdominal Pain --------------------------------------- ------------- --------------------- ------------------ ------------------ Tutorial Pre-Test Immediate Post-Test 6 week Post-Test Retention No EVS 3.88 (0.24) 5.74 (0.23) 6.22 (0.27) 0.48 (0.28) With EVS 3.91 (0.23) 5.59 (0.21) 5.89 (0.22) 0.30 (0.25) difference ns ns ns ns Subject: Gastrointestinal Haemorrhage Tutorial Pre-Test Immediate Post-Test 6 week Post-Test Retention No EVS 4.79 (0.23) 7.60 (0.21) 6.51 (0.25) -1.09 (0.27)\\ With EVS 4.20 (0.25) 6.94 (0.22) 6.20 (0.23) -0.74 (0.26)\* difference ns p= 0.03 ns ns \\p \< 0.0001, pairwise comparison for difference over time (t-test) \p \< 0.05, pairwise comparison for difference over time (t-test) ::: ### Multi-stage questions The interaction between time, EVS usage and order for the problem-solving questions approached significance (p = 0.07). The adjusted mean MSQ test scores for both tutorials are shown in Table [4](#T4){ref-type="table"}. As with the multiple-choice questions, there was a significant improvement in scores both immediately after and six weeks after the tutorial, irrespective of whether the tutorial included the EVS or not for both tutorial topics. There was a significant difference between EVS and no EVS for the post-test scores for gastrointestinal haemorrhage and in the 6-week post-test scores for the acute abdomen. All other comparisons between groups yielded no significant differences regardless of order. There were small but significant benefits observed in retention between students who had tutorials with EVS compared with those who had a tutorial alone. This was true for both tutorial topics. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Scores for multi-stage questions, shown as adjusted mean (standard error). Maximum score was 12 ::: Subject: Acute Abdominal Pain --------------------------------------- ------------- --------------------- ------------------ ------------------ Tutorial Pre-Test Immediate Post-Test 6 week Post-Test Retention No EVS 3.60 (0.25) 6.24 (0.17) 5.96 (0.20) -0.28 (0.23) With EVS 3.97 (0.26) 6.20 (0.20) 6.72 (0.22) 0.52 (0.26)\ difference ns ns p = 0.01 p = 0.01 Subject: Gastrointestinal Haemorrhage Tutorial Pre-Test Immediate Post-Test 6 week Post-Test Retention No EVS 3.88 (0.24) 6.79 (0.15) 5.92 (0.19) -0.87 (0.22)\\ With EVS 3.90 (0.28) 6.22 (0.21) 6.20 (0.23) -0.02 (0.27) difference ns p= 0.03 ns p = 0.02 \\p \< 0.0001, pairwise comparison for difference over time (t-test) \*p \< 0.05, pairwise comparison for difference over time (t-test) ::: Student opinion --------------- Most students found that the standard tutorial without the EVS was reasonably relaxed, found it stimulating, could easily interact and participate. Only a minority felt inhibited and did not want to join discussions. The large majority considered the tutorial well organised and provided useful and easily understandable information. The internal consistency of the scale used to measure the students\' attitudes towards the EVS was determined using Cronbach\'s Alpha, which yielded a value of 0.84 (n = 99). The average result from summing the items in the questionnaire was 24 ± 1, where the range of values was from 5 to 30 (30 indicates the maximum positive response to the use of the EVS). The majority of the students felt that addition of the electronic voting system to the tutorial increased the enjoyment and encouraged participation -- more so than with the tutorial alone. However one quarter of the tutorial group felt that the EVS made no difference to the tutorial and did not influence their ability to contribute. Discussion ========== This study evaluated the place of an electronic voting system in the structured learning process of an undergraduate medical curriculum. Results showed that when used with students at the beginning of their clinical experience, this teaching and learning aid fostered enthusiasm for the tutorial and encouraged group participation. Quantifiable benefits were limited to a small improvement in retention of understanding of the topics tested. Active participation in a tutorial has a positive bearing on the educational process. In trying to increase audience participation there are a number of problems to be overcome. These include class size, the cultural backgrounds of the participants and the ability and enthusiasm of the tutor to engage the audience. Whether or not some of these problems can be overcome by means of various teaching aids is debatable. Different types of teaching aids have been used in the classroom, including audio-visual aids such as slides, movies, PowerPoint presentations, and computers \[[@B9]-[@B11]\]. Paper-based tasks such as worksheets and partially completed notes for students to complete during a teaching session have also been used \[[@B12]\]. Studies of the effect of these interventions consistently indicate the lack of effectiveness of a standard didactic presentation and show student preference for some other form of teaching. Many of these interventions are time-consuming and costly to use and it can be difficult to justify their use if they are not shown to have some form of positive educational benefit. A previous study by the authors used Medici as a teaching aid for a tutorial (unpublished). This study showed that the addition of computer based material into the tutorial made no difference to long-term understanding of material covered in the program. It also showed inclusion of computer-based material increased the time students spent writing within the tutorial. Students perceived the teaching aid to be enjoyable, but many felt it to be too reliant on the ability of the tutor to make it worthwhile. Presentation of material in this form has not been shown to have any negative effect on performance or learning \[[@B13]\]. Electronic voting systems are frequently encountered at meetings and used to measure audience opinion. Whilst it is obviously a more accurate method of counting responses than relying on a show of hands, questions remain on the effectiveness of such systems in engaging the audience and increasing participation. These voting systems are used in student teaching, but little formal evaluation of the process has been undertaken \[[@B10],[@B14],[@B15]\]. The theoretical advantages of an electronic voting system include increased audience participation by allowing anonymous decision-making and by providing immediate feedback to both student and tutor allowing misconceptions to be addressed at the moment they occur. The participants can then see the collective response and as a result may feel less inhibited when contributing to any forthcoming discussion. This could be a two-way process, making contact easier for more diffident tutors. In principle, this may make it easier for a tutor to develop reciprocity and cooperation among students, use active learning techniques, provide prompt feedback and communicate high expectations. These are four of the seven principles of good practice in undergraduate education \[[@B16],[@B17]\] and as such, should be among the goals any tutor sets for a tutorial. Use of instant feedback devices in tutorials and large class settings in a tertiary environment has been reported in a variety of settings dating back to the late 1960s \[[@B10]\]. The most basic of these is a rudimentary discussion of the answers to multiple-choice questions \[[@B18]\]. A more complex arrangement obtaining more complete feedback from students relies on students holding up coloured pieces of paper to answer true false questions \[[@B19]\] or standard 5 stem MCQs \[[@B20]\]. Electronic response systems have been used in nursing and obstetrics \[[@B8],[@B14]\]. Tutors have reported increased participation from students in these studies and student reaction to these interventions has been consistently positive. Where measures of student performance have been made, the results tend towards a positive effect. While the students in the present study felt that the EVS fostered participation, the observers noted that the participation changed from student-tutor interaction, to student-student interaction, with relative isolation of the tutor. Although the students reported that the system encouraged them to participate, it reduced direct interaction with the tutor. We have shown that the tutor\'s role had become secondary to the technology itself and that the type of participation the students had was altered to be more inclusive of fellow students and less inclusive of the tutor. This change in interaction provides advantages for both student and tutor, especially if either party struggles to interact with the other. It is quite possible that the addition of an electronic voting system to a tutorial with a more didactic and passive tutor may have shown objective benefit in short and long term knowledge and understanding of the tutorial subjects by the students. Conversely, this lack of tutor interaction might be regarded as a negative factor for those tutors who enjoy interacting with classes. It could be that any perceived benefit from using the EVS may be a one-off effect, due to the novelty of the device. This study did not address this issue, but long-term evaluation of a non-electronic true/false feedback device indicates a growing acceptance of using that type of response method and an increase in the perceived benefit of the use of such a system \[[@B19]\]. Thus it is likely that the use of an EVS may provide longer-term benefits as students become more familiar with the use of the intervention. The observed reduction in note taking caused by the use of the EVS may have both positive and negative aspects. If notes form a pivotal part of the learning process, then the reduced time spent writing was a disadvantage. If less time taking notes meant more concentration on the tutorial and more participation, this could be seen as advantageous. In this study, where there was no requirement for the students to study for any of the assessments, the role of notes being required for revision did not arise. If the EVS is used in tutorials included in formal assessment, it may well be desirable for notes to be provided to students to allow for later revision. The authors have noted previously (unpublished) that the presence of computer aided learning tools in a tutorial has been reported to increase note taking, so the introduction of the EVS may be acting as a balancing effect. The results of the test measurements in this study demonstrated that the addition of the EVS to the tutorial made no difference to the short term knowledge and understanding imparted but that there was a small increase in long-term retention of problem-solving skills. The increase in long-term understanding is of potential importance, as this is an indicator of stimulus of deeper learning. It has been reported that students involved in problem based learning exercises retain long term understanding better than those who participate in traditional teaching environments \[[@B21]\]. A similar pattern has been observed in this study, where in many respects both the EVS and the case based computer material turned a conventional tutorial into a problem based discussion group. It has been reported that student attention dramatically decreases after 15 minutes in a lecture environment \[[@B22]\]. One might expect any method, which provides regular stimulus in a class will be of benefit in terms of increasing the attention span of students. A tutorial is intended to provide a continuing intellectual challenge and it may be that the effects of additional teaching aids are lessened due to the nature of the teaching environment. Greater objective benefit may be derived from teaching aids when the class size is larger, or when the tutor is less able to interact with the students. This study was conducted in small group setting because it was planned to study one complete academic year. These students never met as a large class and therefore the only practical way of running the study was when the students were undertaking their six week surgical attachment and could be studied then in a block -- albeit a relatively small one. In addition, there was only a limited amount of EVS equipment available -- sufficient for 30 students. These are common types of constraints for any form of educational research. In order to test the usefulness of any educational intervention in a real learning situation, compromises are often required. This study had a number of strengths. First, we were able to enrol the students from a single year of the clinical course and thus ensured an appropriate spectrum of academic ability from a single cohort of students. Second, the study was able to determine clear end-points and apply both quantitative and qualitative measurement in a \'real\' educational environment. A weak point -- which was recognised when the study was designed and which was unavoidable -- was that it would have to run over an entire academic year over six different groups of students and therefore required smaller groups than would be desirable in an ideal situation. The authors feel this may have reduced the impact of the introduction of the EVS. The risk of different groups of students starting with different baselines, depending on how far along the course they had progressed when they did the study was tested and found not to have any confounding effect on the results. Time constraints imposed on the study by the medical curriculum meant that the number of questions, which could be imposed in the testing phase were not quite optimal. Conclusion ========== Teaching aids such as computers and electronic voting systems can be expensive to purchase and maintain. Apart from the initial hardware cost, software must either be bought or constructed and appropriate content developed. If there are no tangible benefits from systems such as the one used in this study, funding might be better used on staff development. Despite this caveat, the future of immediate feedback response systems appears bright. This study has shown that electronic voting systems can provide a stimulating learning environment for students and in a small group tutorial may improve educational outcomes. The wired systems such as the one used in this study are rapidly being replaced by wireless equipment, which should be cheaper, easier to use and more versatile. The growing popularity of Personal Digital Assistants (PDAs) and the expanding capabilities of mobile phones combined with the relative ease of using existing web-based technologies mean that the need for a dedicated response system is waning. The future for this type of system is clearly personal, mobile and adaptable. Competing interests =================== One of the authors (DM) has developed the electronic voting system used in this study. One version of this product (not the version used in this study) is available for hire. At no stage did the author seek to influence the outcome of the study nor did he have any part in the analysis of results. Authors\' contributions ======================= Funding: EP, PGD, DM Study Design EP, PGD, DM Study management and implementation: PGD, NDeY, EP, DM Initial analysis of data: EP, NDeY First draft of manuscript: EP, NDeY, PGD Intellectual input into later drafts and final manuscript: EP, DM, PGD, NDeY Contributors ============ Statistical analysis was completed by Mr. Justin Lokhorst, Dept. of Public Health, University of Adelaide Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6920/5/24/prepub> Acknowledgements ================ This study was funded in part by a Teaching and Learning Grant from the University of Adelaide. The funding body had no role in any facet of the study. The researchers worked independently of the funding body at all times.	PubMed Central	2024-06-05T03:55:59.459615	2005-7-7	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180440/", "journal": "BMC Med Educ. 2005 Jul 7; 5:24", "authors": [ { "first": "Edward J", "last": "Palmer" }, { "first": "Peter G", "last": "Devitt" }, { "first": "Neville J", "last": "De Young" }, { "first": "David", "last": "Morris" } ] }
PMC1180441	Text ==== With curriculum reform, whether we admit it or not, the first cohort of students will be \'test-driving\' the new programme. Not only are they the pioneers of a new curriculum, but as they progress through their studies, they experience each year of the innovation for the first time. As curriculum designers, we learn from their experiences and their feedback to improve the programme content and delivery, invariably for subsequent cohorts. A considerable onus therefore rests with this pioneer group, and their contribution to curriculum design, evaluation and programme revision should be valued \[[@B1],[@B2]\]. From a personal perspective, based on experiences with our pioneers in an institution with a long traditional history, much advice can be offered to faculties considering implementing a curriculum such as problem-based learning (PBL). Perhaps of most importance is that pioneer students should not feel like experimental subjects or \'guinea pigs\', but rather as torch-bearers and ambassadors of the new programme whose constructive input is valued. They should, however, not be overloaded with evaluation, and they must benefit from theirinput into programme revision. Criticism, rumours and misconceptions are detrimental -- address them in the early stages of curriculum reform \[[@B3]\]. To ensure buy-in, involve staff and students in the planning and development of the new programme from the outset, and update them regularly on the progress. Advertising well defined outcomes and objectives by which competency is to be assessed will allay fears regarding quality issues. Also critical is a dean or senior faculty leader who is perceived to be a major player driving the reform \[[@B3],[@B4]\]. Curriculum developers and organisers need clear insight (i.e. long-term planning) into the entire PBL curriculum, particularly if it is of shorter duration than the traditional programme. If registration is not suspended for one year (highly recommended), bear in mind that there will be two cohorts of final year students. Students in both curricula will need reassurance of equitable support in this year. The year in which the pioneer students are studying must be well organised, with minimal disruption. Major modifications only serve to exacerbate uncertainty and provide sceptics with ammunition for criticism. Each year of the programme must be planned in advance as students [will]{.underline} ask for details of what they can expect the following year. As student learning is driven by the assessment and as their progress hinges on their success in the assessment, this aspect of the curriculum must be assigned priority, rather than leaving it to one of the last issues to be addressed in the reform \[[@B5]\]. Explain the format and rationale for the assessment, as it will (or should) be different from what they are accustomed to. Providing pioneers with sufficient support (e.g. meetings with curriculum developers; assistance with self-directed learning skill; counselling) is of critical importance, as these students have no senior PBL colleagues to consult. Ideally, faculty mentors would be of great benefit to these students, but since the human resource capacity is often limited, interaction between students of the two curricula should be encouraged. Our research suggests that students from different curricula draw on each other\'s strengths, forming extensive informal student support networks. Faculties should capitalise on these interactions by formalising cross-curriculum peer support groups. Students deserve the right to contribute to their curriculum. Engaging with pioneers as they progress through their studies provides not only valuable insight into their experiences as trail-blazers but also into those of their traditional curriculum colleagues. Just as pioneers need to be nurtured, traditional curriculum students deserve the same support during the phasing out of their curriculum. Documenting student perceptions during the simultaneous offering of dual curricula is certainly an area for future research as the handful of comparisons between the two curricula have, to date, concentrated on the outcomes (as opposed to the \'process\') of programmes. Competing interests =================== The author(s) declare that they have no competing interests. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6920/5/26/prepub> Acknowledgements ================ The author salutes the pioneer students of Curriculum 2001. You are truly trail-blazers from whom we have learnt much.	PubMed Central	2024-06-05T03:55:59.462616	2005-7-15	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180441/", "journal": "BMC Med Educ. 2005 Jul 15; 5:26", "authors": [ { "first": "Michelle", "last": "McLean" } ] }
PMC1180442	Background ========== It is not uncommon for health care professionals to clash with the family of the patients for whom they care over treatment decisions. Some patients will inevitably suffer the consequences of an error made during their care or hospitalization. Many people in need of diagnostic tests or surgical procedures are forced to wait months, and perhaps even years, to receive these services. These are just some examples of the kinds of ethical challenges that patients and their families may confront in the health care setting. Although these challenges have been discussed widely in the literature as isolated ethical issues in health care, no attempt has ever been made to collate and prioritize them. Ranking the top ethical challenges facing the public can be an effective and valuable way of bringing them to the public\'s attention. Moreover, efforts to address ethical challenges in health care vary significantly from one to another, with some receiving a great deal of attention from the media and from government, while others go largely unnoticed; it would be valuable to discover whether the attention given to these challenges is allocated appropriately. Therefore, the purpose of this study was to identify the top ethical challenges facing patients and families in health care, from the perspective of a panel of clinical bioethicsts. How the study was conducted --------------------------- A modified Delphi study was conducted with twelve clinical bioethicist members of the Clinical Ethics Group of the University of Toronto Joint Centre for Bioethics. The justification for using a panel of bioethicists rather than a panel of community members is that clinical bioethicists will have a greater familiarity with the overall range of challenges than community members due to the fact that the ethical challenges are highly concentrated in their day-to-day work. These clinical bioethicists work in a wide range of health care institutions, including quaternary-level institutions (for both adult and pediatric care), geriatrics/long-term care, rehabilitation, addiction and mental health, and community hospitals. In addition, the experience of the panel members covers both inpatient and outpatient health care. The Clinical Ethics Group at the Joint Centre for Bioethics is the largest institutionally-affiliated collection of clinical bioethicists in Canada, and perhaps in the world. Most of the panel members have several years of experience in clinical ethics, and the Clinical Ethics Group as a whole conduct more than 1200 consults per year. We believe that due to their extensive experience in ethics consultation and bioethics research, this group would be able to offer a uniquely informed perspective on the ethical challenges facing patients and their families. The twelve panel members chosen for the study represent a subset of the larger Clinical Ethics Group; although other members of the Clinical Ethics Group participated in various steps of the process, only the results of the twelve panel members who participated in all rounds were included in the results. In the first round of the process the bioethicists were provided with a list of 38 themes that summarized the themes discussed during the previous two years of case conference discussions at the Joint Centre for Bioethics. The bioethicists were asked to provide a list of what they believed to be the top ten ethical challenges facing the public, which they could pull from the list of 38 themes or provide additional themes in their own words. In this context the phrase \"ethical challenges facing the public\" was meant to imply issues, situations, or problems, which have ethical implications, and would impact or affect the public either directly or indirectly. Although there were no formal criteria for determining the relative impact of the various challenges, the panel members considered such factors as the prevalence of the challenge (how often it occurs and is likely to occur in the future), how many patients and families are and will be affected by the challenge, and the seriousness of the impact on the public. The panel members each responded by email with their list of ranked challenges, along with brief explanations as to why their chosen items were important and challenging. Two of the authors (SKM and JB) then clustered and reworded the themes as necessary to meet a desired level of specificity, and any themes from the original list not ranked by any panel member were dropped from the list. Following this, a list of 32 items was fed back to the panel in advance of a face-to-face meeting. The main purpose of this meeting was for the group to narrow the list further by grouping similar challenges together to make sure that all items were in fact distinct challenges. From this meeting a new list of 23 items was circulated for the second round of ranking, in which participants were again asked to rank their top ten items and give rationales for their rankings. This third round of ranking produced the final list of top ten challenges. The process was stopped after the third round because the list of challenges could not be specified or differentiated any further. The overall ranking was done as per the Delphi method, whereby the highest ranked scenario cited by a panel member was then assigned ten points, followed by the second highest ranked scenario receiving nine points, and so on until the tenth highest ranked scenario. Individual scores were summed up to create a total score for each scenario and a corresponding ranked list for the group. The maximum score that could be achieved by a single item was 120, which would result only if the same item was ranked as the top challenge by all twelve panel members. Discussion ========== Results of the study are listed in Table [1](#T1){ref-type="table"}. In addition to the rankings, the comments from the panel members made during both the face-to-face meeting and the third round of ranking were collected and serve as the basis for much of the content in the Discussion section. With a total of 113 out of a possible 120 points, the highest ranked ethical challenges facing the public in health care was disagreement between patients/families and health care providers over treatment decisions. According to the panel, these disagreements typically take one of two forms: health care professionals might push a treatment option (either for more or less treatment) that patients and families deep unacceptable, or conversely patients/families may push a treatment option (more or less treatment, or different treatment as in alternative or complementary treatments) that health care professionals deem unacceptable. We expand on this challenge in the \"Discussion\" section of the paper below. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Top 10 ethical challenges facing Canadians in health care ::: Rank Scenario Score ---------- ------------------------------------------------------------------------------------------------- ----------- 1 Disagreement between patients/families and health care professionals about treatment decisions 113 2 Waiting lists 102 3 Access to needed health care resources for the aged, chronically ill and mentally ill 89 4 Shortage of family physicians or primary care teams in both rural and urban settings 82 5 Medical error 76 6 Withholding/withdrawing life sustaining treatment in the context of terminal or serious illness 56 7 Achieving informed consent 43 8 Ethical issues related to subject participation in research 40 9 Substitute decision-making 38 10 The ethics of surgical innovation and incorporating new technologies for patient care 21 ::: The second highest ranked ethical challenge facing the public in health care, with 102 total points, was waiting lists. This has been a growing problem in Canadian health care as progressively increasing demand for health care services has put mounting pressure on the already strained Provincial health care systems in the country. According to the panel, waiting for needed care may in some cases compromise the health status and outcomes of patients, impede their ability to return to normal functioning at work and at home, and may also contribute to psychological distress. Waiting lists may also contribute to inappropriate use of scarce resources as is the case when acute care beds are used for long-term care patients, or ICU beds for chronic care patients. Waiting lists also raise the issue of geographical inequities among regions or various health centres. The third highest ranked challenge was issues related to access to needed health care services for the aged, chronically ill and mentally ill. There are two components to this set of issues: one, according to the panel, is the marginalization of populations such as the elderly and mentally ill due to negative attitudes of many citizens toward those populations. The other component is the historical lack of priority of the needs of these populations in the funding allocation schemes of Canadian health care: the bulk of the funding has traditionally gone toward acute, life-saving care, while long-term care, rehabilitation care, and mental health have been grossly under-funded. According to the panel, socially or economically disadvantaged or mentally ill patients require appropriate advocacy to ensure their needs are met. Lack of patient compliance or self-care is sometimes used as reasons to withdraw resources. According to the panel members, we have an ethical obligation to acknowledge and challenge discriminatory beliefs around age, culture, and mental illness that are culturally and socially constructed in order to reduce the risk of emotional and physical harms of the vulnerable in our hospitals and nursing homes. Often these issues emerge when resources are limited. The fourth ranked challenge was the shortage of family physicians or primary care teams in both rural and urban settings. According to a 2002 study published by the Canadian Institute for Health Information, the proportion of Canadian medical graduates starting practice as a general or family practitioner dropped sharply, from a high of 80% in 1992 to only 45% in 2000 \[[@B1]\]. This has become such a significant problem in Canadian health care that it was one of the major issues in the recent contract negotiations between the Ontario government and the Ontario Medical Association \[[@B2],[@B3]\]. Many Canadians living in rural areas simply do not have family physicians; in urban settings many patients have to wait so long to see their family physicians that some choose to seek care in emergency rooms as an alternative. This just puts added pressure on already stressed emergency rooms in major Canadian cities. The shortage of family physicians is of considerable concern for a country whose health care system is centred on universal and reasonable access to medically necessary health care services. The fifth ranked ethical challenge facing the public by the panel was the issue of medical error. Although errors have always been part of medicine, it wasn\'t until the 1999 report from the Institute of Medicine in the U.S., To Err is Human: Building a Safer Health System\[[@B4]\], that the public was made aware of how common medical errors actually are. Examples of such errors can include things that affect single patients, such as a patient receiving the wrong prescription or dosage of medication, or a patient having the wrong surgery performed, or things that impact a larger patient group, as when a hospital fails to properly sterilize surgical equipment. Although medical errors do not in themselves represent an ethical challenge per se, they do carry with them significant ethical implications. For instance, the prevalence of medical errors raises such ethical questions as if, under what circumstances, and how medical errors should be disclosed to patients and/or families. Sixth on the list, but well behind the issue of medical error in overall scoring, was the challenge associated the appropriate use of pain medication in the terminally or chronically ill, and the use of palliative care at the end of life. For example, health care providers sometimes struggle with how to use pain medication appropriately for terminally ill patients because treating the patient\'s pain sufficiently can potentially hasten the death of the patient. The panel has suggested that this is one of the contributing factors behind the widespread under-treatment of pain in the terminally and chronically ill. Another challenge that falls into this category surrounds the timing of palliative care, i.e., decision making around when is the appropriate time to shift from a curative to a palliative approach to care. Seventh on the list according to the panel was the challenge of obtaining informed consent in the health care setting. Research \[[@B5]\] and experience of the panel have shown that there is a huge gap between informed consent in theory and informed consent in practice: many patients do not or cannotread the consent forms they\'re asked to sign; consent discussions and capacity assessments are often superficial and rushed due to time constraints; and those same time constraints often contribute to staff not using interpreters with patients whose first language is other than English. The implication of this is that many patients may be subjected to medical interventions without providing properly informed consent. Since the ethical principle of respect for patient autonomy, on which the doctrine of informed consent is based, has become a central and foundational principle in modern Western health care, the implication of this challenge is troubling. The eighth top challenge was a family of issues associated with participant involvement in research. There are a wide range of ethical issues related to research in the health care setting, including obtaining informed consent, the balance between providing participants with fair compensation and the risk that the compensation will be a coercive influence, the challenge of balancing benefits and risks of the research, issues around patient privacy and confidentiality, and the ethical appropriateness of involving in research participants who are not capable of giving an informed consent. The ninth ranked challenge, finishing closely behind the challenges associated with research, was the challenge of substitute decision making. When a patient is incapable of making a particular health care decision, the health care team will turn to the substitute decision maker to make the decision. Depending on the particular jurisdiction there may be a legal hierarchy of decision makers, which typically places the patient\'s most intimate relationship at the top (spouse or partner) and other relatives toward the bottom of the hierarchy (many Canadian provinces and US states have such a hierarchy written into health care consent legislation). In the experience of the panel members, substitute decision makers often find this task to be a heavy burden, and struggle with the responsibility attached to making a potentially life-altering (and often life-ending) decision on behalf of their loved ones. This burden is experienced to the greatest degree when no guidance has been provided by the patient as to what his or her wishes would be in the current circumstances. When there is no guidance from the patient, conflict often ensues between the health care providers and the family/substitute decision makers as to what would be in the patient\'s best interests. Finally, the tenth ranked challenge was that of surgical innovation. This is a challenge that patients and families will only face indirectly, as the general public is likely unaware of what the issues are related to surgical innovation. Surgical innovation raises such questions as, should innovative surgical techniques be considered research and be required to go through research ethics approval? Since variation is often part of the routine process of perfecting surgical techniques, it becomes difficult to ascertain when a surgical innovation becomes an experiment that requires research ethics approval. Also, what protections should be in place to ensure that innovative techniques or procedures can be developed while the risks to patients are minimized? There are a number of benefits that can be realized with an exercise focused on ranking the top ten ethical issues the public may face. These include providing new contributions to knowledge, raising public awareness, and re-focusing attention on the top challenge. These benefits will be discussed in the discussion section below. Providing new contributions to knowledge ---------------------------------------- The issues described as top ethical challenges by the panel have all been discussed individually in the literature, some extensively. And there have been a few attempts in the past to elicit the views of particular groups on major ethical issues in specific areas. For example, Ersek at al. surveyed a group of oncology nurses to elicit the ethical issues determined to be most important to that group \[[@B6]\]. Along slightly different lines, Walker et al. interviewed a group of physicians and nurses to elicit their perception of \"ethics problems\" in the care of their patients \[[@B7]\]. However, these previous studies have typically focused on the views of a specific group of health care professionals on ethical issues in particular health care contexts. No attempt has ever been made to seek the opinion of clinical bioethicsts who are in a unique position to offer comment on the overall ethical issues in the health care system. Furthermore, despite extensive coverage of ethical issues in the healthcare literature, no systematic effort has been made to collate and rank these kinds of issues from the perspective of the impact on the public. Raising public awareness ------------------------ A second benefit of such an exercise is that it can be part of an effective strategy to bring these challenges to the public\'s attention. Another component of the public awareness strategy might include a press release or other form of media attention coordinated with the publication of the research paper. Even the paper itself can spark discussion and bring the issues to the public\'s attention. Not only would this help to inform the public about ethical challenges they may confront in the health care system to they can be better prepared for those challenges, but it can help garner the public\'s support in advocating for steps to be taken to address the top challenges. The challenges described by the panel will impact patients and their families in different ways and to varying degrees. For example, waiting lists (ranked 2^nd^) and the shortage of family physicians (ranked 4^th^) are challenges that will likely have an impact that is felt directly by a large percentage of the public. The same can be said of the third ranked challenge, access to needed health care resources for the aged, chronically ill, and mentally ill, and that challenge will impact an increasing number of patients and families in the future as our populations age and the number of elderly and chronically ill patients rise. This direct impact on the public, combined with the attention that issues like waiting lists do receive in the media, means that some of the challenges described by the panel are already at the forefront of the public\'s attention. On the other hand, the public is likely to be largely unaware of some of the other challenges mentioned by the panel. These are the challenges that tend to impact a smaller number of patients and families, such as issues related to participation in research (ranked 8^th^), or may impact patients and families more indirectly, such as the issues related to surgical innovation (ranked 10^th^). Re-focusing attention on the top challenge ------------------------------------------ The most interesting result of this study is that the ethical challenge ranked highest by the panel is a challenge that actually receives very little attention in the popular media and at the level of government, and a challenge of which most members of the public are likely completely unaware. It is not surprising, however, that a panel of clinical bioethicists ranked disagreements between patients/families and health care professionals over treatment decisions as the top ethical challenge facing the public in health care. It is not surprising because it is probably the most common reason for requests for ethics consultations, and an area in which many bioethicists focus their research activities. A 2001 study by DuVal et al. found that the most common trigger for ethics consultations among U.S. internists was a desire for help to resolve a conflict \[[@B8]\]. Although the most common arena in which these disagreements occur is the intensive care unit, they can and do occur between patients/families and health care professionals in virtually every health care context: palliative care, rehabilitation, mental health, surgery, general internal medicine, family medicine, and so on. These conflicts can be as serious as an emotionally charged fight over a decision to withdraw aggressive treatment from a terminally ill patient in the intensive care unit, or as mundane as a family physician refusing to acquiesce to a patient\'s request for antibiotics for a viral infection. According to the panel, it\'s the end-of-life critical care cases that tend to be the most emotionally charged, and the most intractable, because these are the cases in which the most is at stake -- they typically amount, literally, to conflicts over life and death. A paradigm example of what has become the most common scenario would involve a patient in the late stages of a terminal illness, such as cancer with multiple metastases, or an elderly patient with multiple co-morbidities, who is ventilated in the intensive care unit. The family would be demanding that \"everything\" be done to maintain the patient\'s life, while the team feel strongly that subjecting the patient to aggressive interventions would amount to torture. Emotions run high, conflict ensues, and communication inevitably breaks down. The above is a paradigm example of what is often referred to in the literature and by health care professionals in the clinical setting as a \"futility\" case. Although there are volumes of literature on the problems associated with the definition and use of the concept of futility, health care professionals know exactly what is meant when a colleague uses the concept: the likely harms of the aggressive intervention(s) outweigh the potential benefits to such a degree that subjecting the patient to the intervention(s) violates their professional (and sometimes personal) values. From the perspective of the health care professionals, the \"right\" decision is obvious and they cannot understand why the family doesn\'t see it the way they do. This often leads to these families being labelled as \"irrational\" or \"unreasonable\" by members of the health care team. The family, on the other hand, views the situation very differently. They will tend to focus on the positive, holding out hope that their loved one will beat the odds. If the physician tells them their loved one has a 90% chance of mortality, what they hear is that their loved one has a 10% chance of survival. They aren\'t guided by success rates or statistics or prognostics; they are guided by devotion and/or a sense of duty to their loved one, a protective instinct, and hope. They may also be guided by deeply held religious beliefs, which they claim are also held by the patient. From the family\'s perspective, the health care professionals are being insensitive and disrespectful, unwilling to listen to or accept what is important to them. Sometimes families will go so far as to accuse the health care team of wanting to withdraw treatment to save money or to give the resources to another patient. Many of the panel members reported having been involved in ethics consultations where family members have expressed these sentiments. What lies at the root of these conflicts is a clash of value systems. It is our value systems that influence the decisions we make, especially when we are faced with significant life-altering decisions in the health care setting. But it is not just patients and their families that are guided in their decisions by their values; health care professionals also come to their encounters with patients and families with their own value systems, both personal and professional. The fact that Canada is one of the most culturally diverse nations in the world means that clashes between the value systems of patients/families and health care professionals may be more common in Canadian health care institutions than in other countries. Addressing the top challenge ---------------------------- Compared to the attention given to many of the challenges listed in the top ten, it is remarkable how little attention has been given to the top challenge. It is especially remarkable given that these conflicts occur in health care institutions across the country on a daily basis. Below, we propose several steps to help address this top challenge. 1\. Educating health care professionals: Although most health care professionals are now taught communication skills, they are not taught the negotiation and mediation skills needed to address serious disagreements. The key is to make an attempt to understand the patient\'s perspective. We recommend that all health professional programs -- undergraduate, postgraduate and continuing -- takes steps to address this deficiency; 2\. Creating policies for health care institutions: Some institutions have developed policies on cases of disagreement, especially at the end of life, but there is no consistency in this area across institutions. Some national accreditation organizations, such as the Canadian Council on Health Services Accreditation <http://www.cchsa.ca> and the Joint Commission on Accreditation of Healthcare Organizations in the U.S. <http://www.jcaho.org>, require health care institutions to have systems in place to address ethical issues facing patients, family members, and staff. We recommend this requirement be sharpened to include mechanisms to resolve disagreements between the health care team and patents or their substitute decision makers. In addition, it may be worthwhile to explore the plausibility of approaching policy development in this area through a process of public consultation. Having stakeholders with diverse value systems come together to discuss the challenge may prove to be a more fruitful approach than applying the standard top-down approach; 3\. Examining the patient\'s perspective: Disagreements between patients or their substitute decision makers and health care teams present a difficult problem with no perfect solution. What is needed is a better understanding of the patient\'s perspective on this challenge. Some excellent work has been done in the attempt to shift the focus of end-of-life issues from the perspective of health care professionals and bioethicists to patients themselves \[[@B9],[@B10]\]. However, what is still needed is quality research that focuses specifically on the perspectives of patients toward disagreements over treatment decisions; 4\. Reporting to the public: Research studies like ours are only one part of a strategy to address the top challenge. A key part of the strategy would be a systematic effort to keep the public informed of such research and the attempts being made to address the challenge. National health councils or other similar bodies would be an excellent mechanism to pull together the diverse initiatives described above and to keep the public informed. Not only should the public be kept informed of steps to address the challenge but they should ideally be involved in the process itself. As mentioned above, one example of involving the public in the process would be to engage the public in the development of policies or guidelines to help address the top challenge. Limitations of the study ------------------------ The main limitations of this study relate to the generalizability of the results. First, the ranking of challenges may not be generalizable to contexts outside of Canada. Some of the challenges listed may be challenges that are particular to the Canadian context because of our Medicare system, such as the challenge of waiting lists or the shortage of family physicians. If this same study were conducted in other countries, it is possible that these challenges would be ranked much lower than in our study, or may not be ranked as a major challenge at all. However, we believe that on the whole our results are likely generalizable at least to other industrialized nations. The challenge of medical error, for example, is a universal challenge because medicine is, by its nature, a human endeavour. As long as humans remain imperfect, medical errors will occur. Moreover, we would predict that the top ranked challenge, disagreements between patients/families and health care providers over treatment decisions, would probably appear at or near the top of similar lists in other industrialized nations. Thus, although the panel was asked to report on the top ethical challenges facing Canadians in health care, we believe the results of this study would be of interest to other countries. Second, because the panel was made up of clinical bioethicists in Toronto, the ranking of challenges may not be representative of the challenges facing the entire Canadian public. Some of the challenges might be considered more or less significant or prevalent in other parts of Canada, especially since there are some very apparent differences between the health care systems of the different provinces. Nevertheless, for the same reasons as mentioned above in the context of generalizability to other nations, we believe the results are in general representative of the challenges facing the Canadian public. Third, since our panel was made up entirely of clinical bioethicists, we recognize that the results may not be generalizable to other groups. For example, if the panel was comprised of, or included, members of the public, hospital administrators, or clinicians, the results might have looked very different. Moreover, although the panel members do represent a wide range of health care instutitions, there were health care settings not represented amongst the group (e.g., home care or community family medicine). However, we believe that this is not a significant limitation of the study because the purpose was not to make a factual claim about what, objectively speaking, are the top ten ethical challenges facing the public. Rather, the purpose was to identify what those top ten challenges are from the perspective of a group of highly qualified and experienced clinical bioethicists who work in a variety of health care institutions. Finally, we recognize that the modified Delphi process that we have presented in this paper is not typical because of the face-to-face meeting of panel members that took place prior to the final round of ranking. One of the potential limitations of including a face-to-face meeting during a consensus process is that a member or members of the group could exert influence over others, thus skewing the process away from genuine consensus. Nevertheless, we believe this potential problem was mitigated by the fact that the face-to-face meeting was not actually part of the ranking process but was an intermediate step between ranking rounds for the purpose of clarifying and differentiating the items. Thus, the consensus process itself was not directly affected by the face-to-face meeting. Summary ======= Patients and their families face a number of ethical challenges in health care. Many of these challenges are no different from the kinds of challenges faced by patients and families in other industrialized nations. Other challenges on the list are more particular to our social context, with their roots in the very nature of the Canadian Medicare system. Waiting lists, access to needed care for the aged, chronically ill, and mentally ill, and the shortage of family physicians, are challenges that may impact Canadians to a greater or lesser degree than citizens of other nations. Interestingly, these three context-specific challenges were all ranked in the top four of the top ten ethical challenges facing Canadians. Moreover, some of the challenges have received far more public attention than others. Since so little attention has been given to the top ranked challenge, disagreements between patients/families and health care professionals over treatment decisions, we have suggested several steps to help address this top challenge. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= JMB contributed substantially to the drafting of the article and gave final approval for the version to be published. SKM contributed substantially to the conception and design, analysis and interpretation of data, drafting of the article and revising it critically and gave final approval for the version to be published. JB contributed substantially to the conception and design, analysis and interpretation of data, and gave final approval for the version to be published. PAS contributed substantially to the conception and design, analysis and interpretation of data, drafting of the article and revising it critically and gave final approval for the version to be published. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6939/6/5/prepub> Acknowledgements ================ Grant support: Dr. Singer is supported in part by a Distinguished Investigator award from the Canadian Institutes of Health Research.	PubMed Central	2024-06-05T03:55:59.463480	2005-6-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180442/", "journal": "BMC Med Ethics. 2005 Jun 26; 6:5", "authors": [ { "first": "Jonathan M", "last": "Breslin" }, { "first": "Susan K", "last": "MacRae" }, { "first": "Jennifer", "last": "Bell" }, { "first": "Peter A", "last": "Singer" } ] }
PMC1180443	Background ========== Infection of the gastric mucosa with H. pyloriresults in a number of disease outcomes including gastritis, which precedes the development of peptic ulcer disease, gastric cancer and lymphomas of the MALT \[[@B1]-[@B3]\]. These diseases caused by H. pyloriand their prevalence rates differ in different geographic countries and only a subset (10%) \[[@B4]\] of infected patients develop one of them. This raises the question as to why H. pyloricauses disease in a few individuals, but not in the great majority \[[@B5]\]. Many studies have demonstrated the involvement of bacterial virulence factors, host genetics and environmental factors in contributing to the development of disease. Bacterial virulence factors include proteins mediating establishment/colonization, persistence of infection and finally long-term damage to the host \[[@B6]\]. The cagpathogenicity-island (cagPAI) is the most noteworthy among these factors. PCR analyses have suggested that this island is not intact in many strains across the world \[[@B7]\] and the presence of an intact PAI although not always \[[@B8],[@B9]\] is indicative of a more severe outcome \[[@B10]\]. The expression of various products encoded in the cagPAI is known to be involved in inducing inflammation, ulceration and carcinogenesis \[[@B11]\]. However, the cagA is expressed by the majority of H. pyloristrains, irrespective of the geographic origin and clinical diagnosis \[[@B12]\]. The vacuolating cytotoxin antigen (VacA) is another virulence factor that is considered to constitute an increased risk for development of peptic ulcers and gastric cancer \[[@B13],[@B14]\]. Allelic variations in the vacA gene are found in the signal (s1, s2) and the middle region (m1a, m1b, m2) and the s1 type is associated with ulcer disease \[[@B14],[@B15]\]. More pronounced inflammation is associated with strains, which express the outer membrane protein OipA. OipA induces IL-8 secretion by epithelial cells. Active OipA protein production may be \'on\' or \'off\' depending on the number of CT repeats in the signal sequence of the oipA gene (HP0638). H. pyloristrains may also be grouped geographically based on oipA sequence pattern \[[@B16]\]. Specific adhesins viz., babA and babB mediate the adherence of the bacterium to specific human blood group Lewis antigens and are associated with various disease outcomes \[[@B17]\]. Similarly, a putative E. colirestriction enzyme NlaIII homologue, the iceA gene in H. pylori, which is activated on contact with the epithelium, is also shown to induce high levels of IL-8 \[[@B18]\]. Accordingly, strains with OipA \'on\' status, active forms of iceA and babA \[[@B18]\] and particularly strains which are cagA+ and vacA s1 have been shown to cause a more severe outcome \[[@B14],[@B15],[@B19]\], though not in all cases \[[@B20]\]. Many studies have pointed out a bio-geographical variation in virulence factors; for example, the sequences of vacA and cagA differ in strains from the United States and Europe from those in China and Japan \[[@B21]\]. Also, the prevalence and type of H. pyloriinfection varies with a very high rate of occurrence (up to 70%) in Asia and the Middle East \[[@B22]\], in contrast to only 30--50% in Europe and the United States \[[@B23]\]. Further, the infection is minimal in children in the west while in the rest of the world it affects both young and the old. Active infection with H. pyloriwas seen in about 7.5 million people in the general population of England and Wales. This although varied from one region to the other with the highest rates recorded in London \[[@B24]\]. Thus H. pyloriremains an important infection in the UK. H. pyloripopulation has been described as highly recombining, and therefore exhibits enormous strain diversity, part of it may be due to the presence of the plasticity zones \[[@B25]\]. Since this organism has also been shown to be transmitted within families, a greater number of epidemiological studies reveal that these strains not only show similar genotypic profiles when obtained from related patients but also show common profiles within isolates from specific countries \[[@B26]\]. Phylogeographic affinities were pronounced in case of European strains based on the multi locus sequence typing of seven housekeeping genes where the European strains and the Asian strains shared an ancestral relationship \[[@B27]\]. This observation was also recorded in other studies based on the evolutionarily conserved ERIC sequences that indicated close associations between the Irish, Spanish and the European strains and also clustering of the English strains with a few Asian strains \[[@B26]\]. However, the number of English strains used in that study was very small. It is noteworthy in this respect that comprehensive phylogenetic analyses in case of English strains have been rarely performed. In this study, we aimed at a comprehensive assessment of the prevalent genetic structure of H. pyloristrains infecting the English population in Nottingham, which is centrally located in the United Kingdom. The strains were analyzed to study a total of 45 different parameters pertaining to 28 informative loci including the virulence factors cagA, oipA, iceA and vacA in addition to other genes of the cagPAI and the motifs downstream to it. Composition of the plasticity region cluster including the putative gastritis (JHP0986) and gastric cancer (JHP0947) associated markers were also studied \[[@B25]\]. Phylogenetic analyses were performed using FAFLP markers, nucleotide sequences of the cagA, babB and the glmM genes and the repetitive sequences interspersed in the H. pylorigenome (ERIC and REP). According to our observations, phylogenetic placement of English strains shows affinities with East Asian and Amerindian strains. Results ======= Details of all the genotyping and phylogenetic analyses have been depicted in Figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"} and are summarized in Additional file [1](#S1){ref-type="supplementary-material"}. Macroscale analysis of the cag-PAI and the downstream motifs -------------------------------------------------------------- The status of cagA gene was assessed by using primers specific to sequences at both the 5\' end and the 3\' ends. PCRs for the 5\' end were positive in 41 strains (62.1 %), whereas only 35 strains (53%) were PCR positive for the cag3\' end. Twenty-one strains (31.8%) had both the ends detected by PCR indicating therefore the possible presence of a complete gene. Of the ten strains PCR negative for both cagA ends, the oipA gene frame status was \"on\" for 8 of them. Hence we showed no association between the presence of the cagA gene and the frame status of oipA. Only two strains out of the 10 completely PCR negative for cagA did not have any motif type on the right end of the gene PCR amplified. The most frequently detected gene by PCR in UK strains was cagT (83.3%) followed by cagE (71.2%). Upon analysis of the extreme right junction of the cagPAI (region extending from the cagA 3\' region to the glrgene), 54 strains out of 66 had either the type IIIa motif (28.8%) or the type I (Ia/Ib) motif (28.8%). The type IIIb motif was observed in a single isolate (N3), while 13.6% strains displayed the type II motif. The type IV motif was amplified in only 9% of the strains and among the three strains recovered from patients of Indo -- Pakistani origin, N105 showed a type I a signature, while N115 showed type II motif. The type III motif was also observed for strains from patients of Chinese (N 99) and Russian (N90) ethnicities settled in UK. The frame status of the oipA gene was \'off\' in 8 of the 12 strains that did not have successful amplification of any motif types on the right end of the cagPAI and between the 3\' end of the glrgene. The sequence of the 250 bp product amplified from the 3\' end of the cagA gene was determined for 16 English strains. Phylogenetic analyses of these sequences in comparison with others from Holland (n = 1), East Asia (n = 4), India (n = 2), Bangladesh (n = 2), South America (Peru, n = 2 and Guatemala, n = 2), South Africa (n = 1) and Gambia (n = 2)) (Figure [2E](#F2){ref-type="fig"}) revealed that the Asian strains carried a unique cagA gene sequence and formed a segregated cluster. Strains from Africa clustered close to the Indian ones whereas English strains showed no specific clustering. vacA and iceA statuses of the isolates ------------------------------------------ 87.8 % of the strains possessed the toxigenic type s1 vacA allele while the less toxigenic s2 allele was detected in \~ 6% strains. The vacA m2 genotype was present in 66.6% strains and only 21.2 % strains had the m1a genotype. The m1c subtype found in strains from India \[[@B28]\] was observed in 3 strains and none of the strains had a type m1b vacA allele. Therefore, the s1m2 type of vacA was most commonly (66.6%) found in these English strains. Only 38 of the 58 strains (65.5%) with vacA s1 allelic subtype had the oipA gene in frame. The iceA1 allele was present in 38 strains (57%) whereas the iceA2 allele was found in 24 strains (36.3%). Only two strains were positive for both the alleles (N22, N105). The likely explanation is that these \"strains\" were in fact mixture of two strains. Status of the Proinflammatory protein oipA gene (HP0638) ---------------------------------------------------------- Strains from UK mostly had the oipA frame status \'on\' (70.5%) with the CT dinucleotides repeated 6 times in 37.7% strains. This was followed by the repeat number of 7 observed in 18 % strains and 9 in 11.5% strains. 10 CT repeats were found in a single isolate (N52) and a single repeat was shown in three strains. These results and those for other loci studied are shown in the bar diagram \[Figure [1](#F1){ref-type="fig"}\]. Inventory of the plasticity region ORFs in English strains ---------------------------------------------------------- The ORF HP986 (referred to as the gastritis associated marker) \[[@B25]\] was PCR amplified in 31 strains (52.5%), while the gastric cancer associated ORF JHP947 \[[@B25]\], was amplified in only 10 strains (19.2%). Other ORFs from this region that were amplified included JHP912, which was seen in 93.5% strains. The ORF JHP926 was amplified in 32.6 % of the strains while a J99 specific ORF JHP931 thought to be involved in DNA replication \[[@B25]\] was found in 51.1% strains. ORFs JHP944 and JHP945 were amplified in 13% and 40% strains respectively. None of the strains showed any specific pattern of ORFs within the plasticity region. Phylogenetic placement and affinities with other genogroups ----------------------------------------------------------- The housekeeping gene, glmM, was present in all the strains (100%) and the adhesin babB was amplified in 51 strains (77.27%). The babB gene has been a marker of choice for tracing lineage in H. pyloriand recent studies employing this gene have postulated H. pylori\'s association with its human host to be approximately 11,000 years old \[[@B27],[@B29]\]. Hence phylogenies \[Figure [2C](#F2){ref-type="fig"} -- glmM tree and 2D -- babB tree\] were generated based on the sequences of these genes in representative strains. These phylogenies revealed that strains from England clustered with other European strains (Ireland-Ire and Spain-HupB), while some affinities between them and Peruvian strains could also be noticed. Individual branches representing geographically specific glmM sequences were observed for India (MS, L), Japan (Hu), and Africa (R). FAFLP patterns of English strains revealed about 130 fragments in the size range of 50--500 bp when the genomic DNA was digested with enzymes MseI+0/EcoRI+A. A binary table indicating the presence or absence of a particular amplicon in each strain was scored and these values were used to assess the genetic relatedness within English (abbreviated N) and with other European strains including those from Sardinia (SarD), Spain (HupB) and Ireland (Ire). These strains clustered in one group labeled in the figure as the European cluster. Another cluster obtained was the one which represented contribution from Asian- European gene pool and included strains from India (L, BJ & MS), Japan (Hu), Africa (R) and others from Europe (HupB, Ire, N) \[Figure [2B](#F2){ref-type="fig"}\]. A similar trend was observed with other fingerprinting techniques employing the Enterobacterial Repetitive Intergenic Consensus sequences (ERIC) and the Repetitive Extragenic Palindrome (REP) sequences. Based on the amplification patterns of genomic DNA between the REP sequences, the English strains grouped with the Irish ones in a European cluster and a segregated cluster constituting all Indian strains from Ladakh and a single strain from England (N114) was obtained \[Figure [2F](#F2){ref-type="fig"}\]. ERIC phylogeny for this set of strains in comparison with other strains from different world populations as reported earlier \[[@B26]\] indicated that these strains clustered closely with either the Spanish and Irish strains and more distantly with the Indian and African strains. Discussion ========== Evolution of infectious microorganisms is a consequence of the genetic polymorphisms they accumulate, which in turn is the result of the long term selection pressure exerted by the host immune system in case of chronic infection as well as the environment \[[@B9],[@B30]\]. This is more pronounced in case of bacteria like H. pylorithat cause multi decade long infections, wherein an acquiescent requirement to constantly keep their genomic content recasting is crucial. In H. pylorithis concern is resolved by the use of restriction -- modification systems and regulation at gene level by nucleotide substitution, insertion and deletion events \[[@B31]\]. Further, the presence of insertion elements, plasticity regions and the pathogenicity islands contribute considerably to its genetic diversity. We attempted to analyze genetic variation and structure of H. pyloripopulations infecting the native people of England. The UK today is more culturally diverse than ever before with the majority of the UK population being ethnic Europeans (92 %). The remaining 4.6 million (or 7.9 %) people represent other ethnic groups. South Asians are the largest of these groups, followed by Caribbean and Black Africans \[[@B32]\]. Such a multiethnic presence creates an interesting genetic conundrum when we attempt to analyze incidence and healthcare-impact of any pathogen that biases itself with respect to the host genetic makeup. Also, associations of the disease outcome with the virulence factors has thus far been enigmatic and since there were no comprehensive studies involving multiple loci for genotype-phenotype assessments, the current study was envisaged in combination to generate base line data relevant in molecular epidemiology of virulent strains in England. The cagPAI is a major virulence determinant in H. pyloriand strains lacking this island are akin to commensals rather than pathogens \[[@B1]\]. Reports suggest that the presence of a complete set of genes within the cagPAI ensures a 5-fold increased severity of disease outcome than the intermediate PAI \[[@B10]\]. We have earlier showed that a higher number of strains from Japan had an intact cagPAI \[[@B7]\], hence it may be thought as an important factor influencing the outcome of the infection as a higher rate of gastric cancers was observed in the Japanese patients. The possible role of cagA in oncogenic mechanisms is being worked upon \[[@B33]\]. Most English strains in our case retained the cagT and the cagE genes. Studies showed that strains lacking the cagT gene had a defective \'molecular syringe\' that is encoded by the PAI \[[@B34]\] reflecting thus on inability of the Type IV system to eject out the cagA protein. The cagE gene on the other hand is known to induce NFkB activation and IL8 secretion \[[@B8]\] in addition to mediating host-cell cytokine rearrangements in infected epithelial cells. About 25 % of the strains carried the type IIIa motif on the right end of the cagA gene. This observation also supports the hypothesis that some of the European strains share some features typical of the Asian genogroups. This is also supportive of an earlier observation \[[@B31]\] that a very small number of strains from the European countries also show the type III motif. English strains mainly showed a higher number of the toxigenic s1 type of vacA allele. Interestingly, the s1m1 combination was observed in less percentage in contrast to the s1m2 genotype that was seen in 60 % of the cases. Earlier studies in a subset of European population (Mid-Essex) by RFLP of the mid-region of the vacA gene in strains originating from dyspeptic patients demonstrated that 46% of these strains had the s1m2 genotype while 40% strains had the more toxigenic s1m1 combination \[[@B35]\]. The s1m2 genotype was also common in strains from North Wales among clarithromycin sensitive and resistant H. pylori\[[@B36]\]. It was interesting to find the m1c allele \[[@B28]\], which is known to be prominent in the East Asians, in UK strains. None of the strains had multiple vacA genotypes, which are common in China and other East Asian countries \[[@B37]\]. It has been reported earlier that the s1 allele was most frequently observed in the European population with the s1a allele predominant in Northern and Eastern Europe. Also, s1a and s1b alleles were observed in France and Italy while the Spanish and Portuguese strains had the s1b subtype \[[@B13]\]. The virulence spectrum of the English strains was also exemplified by the observation that 70% of these strains had the oipA gene in frame. Greater than 6 CT repeats in the upstream homopolymeric tail of the oipA gene is characteristic of European strains \[[@B16]\]. Our results indicate that English strains most commonly displayed 6 CT repeats. Strains with 9 CT repeats were reported to have the gene out of frame owing to the deletion of the CTAA sequence present immediately upstream of the CT repeats \[[@B16]\]. Similar results were found with English strains with 9CT repeats. From our analyses, the iceA1 allele was more common than the iceA2 allele. Although no association between iceA subtypes and clinical outcome has been reported, strains carrying iceA1produce higher levels of IL-8 in the gastric mucosa and are more often associated with DU in the North America and Dutch people than strains carrying iceA2\[[@B38]\]. The plasticity region genes are speculated to provide the strains with survival benefits in some hosts \[[@B25]\]. This region extends from the ORFs JHP914 to JHP951 in the sequenced strain J99 and is shown to be unstable since some genes within the zone are lost during subsequent infections or laboratory passages \[[@B25]\]. It is now evident from partial sequencing of a Peruvian isolate that this plasticity region might encode yet another type IV secretion system \[[@B39]\] and that the 2 sequenced genomes carry incomplete sets of genes corresponding to this cluster. All the English strains we looked at were similar to either strains 26695 and J99 in that none of them harbored a complete plasticity region cluster as shown recently for a Peruvian strain \[[@B39]\] (data not shown). However, any role of such a putative secretory system is still enigmatic due to lack of correlation of its intactness or abrogation with disease, although, some of the ORFs in the plasticity region have been shown to be associated with a particular disease outcome \[[@B25],[@B40]\]. For English strains, we analyzed seven ORFs from the plasticity region cluster, of which HP986 was strain 26695 specific, while others excluding HP912 were J99 specific. The ORF HP912, the predicted cell division protein/septum formation protein (ftsA) was PCR amplified in 93.5 % strains. JHP931, a predicted DNA topoisomerase I was amplified in 51%. None of the strains had the same regions of the plasticity zone deleted. Our phylogenetic analyses based on the FAFLP markers showed a star like distribution on an unrooted neighbor-joining network. Four clusters were evident and the largest cluster was populated mainly by English isolates. Further, affinity to Irish and Indo-European genogroups was also observed. These observations were also substantiated by other slow evolving markers such as ERIC and REP sequences. This denotes stable associations among the genogroups. Similarly, cluster analysis of babB and glmM genes also revealed close associations within the European strains and with the Indian strains. However, homologies with East Asian and Amerindian strains were most noteworthy and were comparable to those shown by Irish strains \[[@B41]\]. This reflects ancient genetic events and possible oriental influences on the evolution of H. pyloriin the English population. Such kinds of non-random genetic links of H. pylorimay be helpful in understanding evolution of this organism and its clinical consequences in different parts of the world. These findings are in accordance with a recent study that demonstrated that Indian and European H. pyloriisolates grouped in the same subpopulation and that East Asian and a subset of European isolates share an ancestral relationship and diverged from each other recently \[[@B27],[@B42]\]. The Asian strains, however, were distinctly separated from the European and western strains based on the cagA gene sequences except for a few strains that show remote similarity to the East Asian gene pool. We found only a single English strain (N115) that diverged significantly towards the Asian cluster and was recovered from a patient of Indo-Pakistani origin and thereby denotes contribution from the Asian gene pool. Conclusion ========== In summary, our study demonstrated certain distinctive genetic features of the H. pylorigene pool in England based on genotypes of virulence genes and neutral markers. Important among these features is the genetic affinity towards East Asian strains. This is also probably the first comprehensive study on detailed, multilocus and multi method genotyping of H. pylorifrom England or elsewhere. The genomic profiles generated in this study may be useful for electronic archiving and retrieval for inter-laboratory comparison and are suitable for storage in epidemiological databases for comparative analyses. However, it will be necessary to analyze additional representative strains, especially from other European populations. Also, our study has largely been an examination of a specific (peptic ulcer disease) group of patient isolates and may not be reflective of other patient isolates from different disease stages in England. Future studies are therefore clearly needed to involve other disease specific strain groups. Further characterization of associations of such informative loci as we examined, in the gene pool of diverse strain groups and with varied disease spectrum may lead to newer insights into the mechanisms of H. pyloricolonization, and virulence in different hosts. Methods ======= Bacterial DNA preparations -------------------------- Sixty-six DNA preparations using Nacl-CTAB method \[[@B43]\] from English H. pyloristrains were obtained from the strains corresponding to patients reporting at the Queen\'s Medical Centre, Nottingham, UK. These strains were recovered from patients diagnosed with ulcer disease, having either current ulceration, past ulceration, evidence of scarring at endoscopy or erosions at endoscopy. More than half of these patients were taking acid suppression therapy. These strains were mainly from ethnic English people, although a few were from people originally from Russia (N90), China (N99), South Asia (N105, N115, N131) and Italy (N106) who had settled in the UK. Strains from other countries were taken from our international collection of genomic DNAs provided by our collaborators. Among these are strains from Spain (HupB, n = 7), Ireland (Ire, n = 14), Japan (Hu, n = 10), Peru (Sjm, n = 6), Sardinia (Sard, n = 2), India (MS, n = 1; L, n = 10; BJ, n = 1), Bangladesh (n = 2), Holland (n = 2), and S. Africa (R, n = 8). Molecular genotyping and sequencing ----------------------------------- Amplification of candidate gene loci including oipA, babB, vacA middle region, cagA and glmM genes were carried out as described previously \[[@B44],[@B31],[@B41]\]. Purified PCR amplified products (QIAquick Gel extraction kit) were sequenced using the ABI Prism Big Dye Terminator Cycle Sequencing Reaction Kit (Applied Biosystems, Foster City, USA) in an ABI 3100 automated DNA sequencer. The iceA allele status was determined using oligonucleotide primers mentioned elsewhere \[[@B44]\]. The cagA, cagE and the cagT genes within the cagPAI were detected using 4 pairs of primers as mentioned earlier \[[@B7]\]. Analysis of rearrangements of the motifs at the right end of the cagA gene and towards the 3\' end of the glutamate racemase gene (glr) were performed with seven different sets of primers as described previously \[[@B31]\]. PCR primers and procedures used for evaluating the presence of the plasticity region ORFs HP 912, JHP 926, JHP 931, JHP 944, JHP 945, JHP 947 and HP 986 have been described elsewhere \[[@B25]\]. The annealing temperatures for the ORFs HP912, JHP 926 and JHP 944 were standardized to 59, 57 and 66°C respectively for 1 min, followed by an extension at 72°C for 1 minute. Consensus sequence for each sample was generated using Genedoc (version 2.6.002). Clustal X (version 1.81) was used to align these sequences and dendrograms representing the genetic relationships between strains were generated using Treeview (version 1.6.6). Frame status of the oipA gene was analyzed using Lasergene software (DNAStar Inc. USA). Genetic diversity of the cagA sequences of all the representative isolates tested from English patients were compared to other records from Genbank \[China47 (AJ252985), Hong Kong81 (AF198486), Hong Kong77 (AF198485), Japan 54 (AF198484), S. Africa19 (AF198470), Gambia 4659 (AF198468), Gambia4797 (AF198469), Peru24C (AF198473), Peru34B (AY18476), Guatemala88 (AF198472), India18A (AF202224), India19A (AF202225) DH102 (AY169292), DH200 (AY169294), Dutch107 (AJ252963).\]. These Genbank sequences were also used along with sequences from English strains for the phylogenetic tree construction. Whole genomic fingerprinting and genotyping ------------------------------------------- Whole genome fingerprinting based on FAFLP genotyping was done as described previously \[[@B43],[@B45]\]. Briefly, the profiling of whole genome micro-restriction fingerprints with EcoRI/MseI enzymes using fluorescence tagged primer pairs EcoRI+A/MseI+0 and EcoRI+G or A / MseI+0 was performed. The PCR amplified fragments for each of the strains were then subjected to electrophoretic separation on a 5% acrylamide gel and scoring of the fluorescent markers was done using an automated DNA analysis workstation (ABI Prism 3100 DNA sequencer). The PCR methods for the ERIC fingerprinting technique has been previously described \[[@B26]\]. The REP based typing procedure involved primers for amplifying unique DNA sequences between the two REP signatures \[[@B46]\]. All the gel images corresponding to ERIC and REP PCRs were analyzed using the Quantity 1.0 software in a gel documentation system (Bio-Rad, USA). These images were then uploaded into Diversity 2.2.0 database (Bio-Rad, USA). Band sizes, band attributes and standard molecular weights were assigned alongside the molecular weight markers. Cluster analysis of DNA profiles was conducted on the basis of fingerprint characteristics. Based on the data for the presence or absence of 3--15 different DNA fragments in the fingerprints of strains of H. pylori, a binary data matrix was created. Overall similarity between the pairs of strains was calculated from the binary data matrix using the simple matching dice coefficient. The resulting similarity matrix was used for cluster analysis by the unweighted paired group method with arithmetic averages (UPGMA) to generate trees. Data archiving and genome wide comparisons ------------------------------------------ All the data obtained through candidate gene sequencing and DNA profiling was deposited in the genoBASE pyloridatabase <http://www.cdfd.org.in/amplibase/HP>. The genoBASE pyloriserver was queried for genome wide comparisons. The cagPAI rearrangement profiles and cagA -- glrmotif types were also compared to existing records in the database. Authors\' contributions ======================= FK, MAH and IA did all the PCR based genotyping and sequencing. SS and MD carried out phylogenetic analyses. KRR and AAM provided informatics support and comparative analyses on AmpliBASE server. AAK provided isolates from India and provided some infrastructure support for microbiology work. LAS contributed to the editing of the manuscript and partly supervised the study as a collaborator under the H. pylorievolutionary genomics interest group. NA conceived, designed and supervised the study, compiled and edited the manuscript and provided overall supervision and leadership. Supplementary Material ====================== ::: {.caption} ###### Additional file 1 ::: ::: {.caption} ###### Click here for file ::: Acknowledgements ================ We are thankful to Seyed E. Hasnain, Director of the CDFD for his permission to carry out this study, his keen guidance and encouragement. Our special thanks are due to John Atherton who provided DNA of the English isolates and for critically reading the manuscript. Our thanks are also due to Douglas E. Berg and Asish Mukhopadhyay for their help in articulating this Indo-English collaboration. We are also thankful to our collaborators, especially Ian M. Carroll for other DNA samples used in this study. Funding support from the CDFD core grants to NA is gratefully acknowledged. Our special thanks are also due to the International Society for Genomic and Evolutionary Microbiology (ISOGEM) <http://www.isogem.org> for support. This work was carried out as a collaborative study under the ISOGEM Working Group on Genetics of Helicobacters <http://www.pathogen-evolution.org>. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Diagram showing the percent distribution of different genetic loci in the English isolates. The presence of the cagPAI with the motifs on its right end, the vacA and iceA genotypes and the presence of the glmM and babB house keeping genes and characteristics of the virulence gene oipA and the genes included in the plasticity region cluster are shown. The distribution was determined by PCR using primer sequences from the reference articles denoted by the number in superscript. ::: ![](1471-2180-5-32-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### \(A) shows a gel image of representative FAFLP patterns of English H. pyloristrains with the bands in red indicating the internal lane standards (Genescan Rox 500). (B) Representative phylogenetic tree indicating genetic relationships between English H. pyloristrains (n = 16) and those from other world populations (HupB, n = 3; Ire, n = 3; Italy (Sard), n = 2; R = 2; E. Asia (Hu), n = 2; India (L, MS, BJ), n = 5) based on the FAFLP profiles generated for the strains from different countries (described in the materials and methods section). The neighbour joining network tree created based on the level of similarity between the amplitypes shows that European strains form a distinct cluster although two other clusters showing associations with the Indian strains are also seen. (C) Cluster analysis of the 296 bp glmM was generated with 10 English strains, 2 strains each from India (L), E. Asia (Hu), Peru (Sjm) and S. Africa (R). 4 each from Ire and HupB and the sequenced HP26695. The rounded circle here shows the grouping of English islolate (N 17) with the Indian (L) and E. Asian (Hu) strains. (D) Tree constructed from the babB gene products from different strains (N, n = 8; Ire, n = 4; HupB and MS, L, n = 3 respectively, 2 each of Hu and Sjm with HP26695). (E) Genetic relationships of the English strains based on the 250 bp sequence of the 5\'end of the cagA gene (England, n = 7; East Asia, n = 4; S. Africa, n = 1; W. Africa, n = 2; S. America, n = 2; central America, n = 1; India, n = 2; Bangladesh, n = 2 and Holland, n = 1). A few strains from Nottingham showed sequence similarity with the Asian strains as indicated by the circles in green. (F) Phylogenetic tree based on the REP fingerprinting patterns generated by the English strains (n= 5) in comparison with 5 strains from Ire and 6 from India (L). ::: ![](1471-2180-5-32-2) :::	PubMed Central	2024-06-05T03:55:59.467186	2005-5-25	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180443/", "journal": "BMC Microbiol. 2005 May 25; 5:32", "authors": [ { "first": "Farhana", "last": "Kauser" }, { "first": "M Abid", "last": "Hussain" }, { "first": "Irshad", "last": "Ahmed" }, { "first": "Sriramula", "last": "Srinivas" }, { "first": "S Manjulata", "last": "Devi" }, { "first": "Ahmed A", "last": "Majeed" }, { "first": "K Rajender", "last": "Rao" }, { "first": "Aleem A", "last": "Khan" }, { "first": "Leonardo A", "last": "Sechi" }, { "first": "Niyaz", "last": "Ahmed" } ] }
PMC1180444	Background ========== The use of systematic literature review to inform evidence-based practice in diagnostics is rapidly expanding. Although the primary diagnostic literature is extensive, there remain a number of problems for systematic reviews of diagnostic tests. Appropriate methods for rigorous evaluation of diagnostic technologies have been well established \[[@B1]-[@B5]\]. However, available studies have generally been poorly designed and reported \[[@B6]-[@B8]\]. Similarly, although a number of quality checklists for diagnostic accuracy studies have been proposed \[[@B9]\] and there is growing evidence on the effects of bias in such studies \[[@B10]\], there has been no rigorously evaluated, evidence-based quality assessment tool for diagnostic studies. The objective of this study was to investigate the impact of quality on the results of a diagnostic meta-analysis, using regression analysis. A large diagnostic systematic review was required to enable the use of regression analysis to investigate the impact of components of quality upon results. We have recently completed a systematic review, which aimed to determine the most appropriate pathway for the diagnosis and further investigation of UTI in children \[[@B11]\]. It included an assessment of the accuracy of tests for three different clinical stages of UTI: the diagnosis UTI, localisation of infection, and further investigation of patients with confirmed UTI. The nature of the tests included in these three clinical sections of this review differed. Tests used to diagnose UTI were generally laboratory-based or near-patient methods, with relatively objective interpretation of results, e.g. dipstick tests and microscopy. By contrast, tests used to investigate confirmed UTI mainly utilised imaging technologies which are largely subjective in their interpretation, and where diagnostic thresholds are difficult to define. Tests used to localise infection spanned both categories. We hypothesised that the components of methodological quality affecting results were likely to differ between the three sections of the review. Such potential differences may indicate a need for topic-specific checklists for the assessment of quality in diagnostic studies. A secondary aim of this study was to contribute to the evaluation of QUADAS, an evidence-based tool for the assessment of the quality of diagnostic accuracy studies that was specifically developed for use in systematic reviews of diagnostic tests \[[@B12]\], by investigating the importance of specific QUADAS items. Methods ======= We used QUADAS \[[@B12]\] (Table [1](#T1){ref-type="table"}) to assess the quality of primary studies included in the review. Items were rated as \'yes\', \'no\', or \'unclear\'. We examined differences in the individual QUADAS items fulfilled, as well as their impact on test performance. The review divided the diagnosis and further investigation of UTI into the following three clinical stages: diagnosis of UTI, localisation of infection, and further investigation of the UTI. Each stage used different types of diagnostic test, which were considered to involve different quality concerns. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### QUADAS ::: Item \# Description ------------- -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 1. Was the spectrum of patients representative of the patients who will receive the test in practice? 2. Were selection criteria clearly described? 3. Is the reference standard likely to correctly classify the target condition? 4. Is the time period between reference standard and index test short enough to be reasonably sure that the target condition did not change between the two tests? (disease progression bias) 5. Did the whole sample or a random selection of the sample, receive verification using a reference standard of diagnosis? (partial verification bias) 6. Did patients receive the same reference standard regardless of the index test result? (differential verification bias) 7. Was the reference standard independent of the index test (i.e. the index test did not form part of the reference standard)? (incorporation bias) 8. Was the execution of the index test described in sufficient detail to permit replication of the test? 9. Was the execution of the reference standard described in sufficient detail to permit its replication? 10. Were the index test results interpreted without knowledge of the results of the reference standard? (test review bias) 11. Were the reference standard results interpreted without knowledge of the results of the index test? (diagnostic review bias) 12. Were the same clinical data available when test results were interpreted as would be available when the test is used in practice? (clinical review bias) 13. Were uninterpretable/ intermediate test results reported? 14. Were withdrawals from the study explained? ::: We analysed results grouped by clinical stage. Within these groups, we pooled studies of similar tests or test combinations where sufficient data were available and where pooling was clinically meaningful. (Table [2](#T2){ref-type="table"}) The minimum number of studies that we required for regression analysis was ten. This choice was made based on published guidance \[[@B13],[@B14]\]. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Tests included/excluded in the regression analysis ::: Tests included in the regression analysis (number of studies) Tests for which there were insufficient studies to permit regression analysis (number of studies) ------------------------------------------------------------------------------ ------------------------------------------------------------------------------------------------------------------------------------------------- Diagnosis Dipstick: nitrite (23), LE (14), nitrite or LE positive (15) Clinical history(6) Microscopy: pyuria (28), bacteriuria (22) Dipstick: nitrite and LE positive (9), glucose (4), protein (2), blood (1), protein and LE positive (1), combinations of 3 dipstick tests (5) Microscopy: pyuria or bacteriuria (8), pyuria and bacteriuria (8) Culture:standard (1), dipslide (1) Combinations of different tests(10) Localisation Ultrasound(20) Clinical history(5) Laboratory based tests(16) Imaging techniques:MCUG (7), MRI (1), CT (1), IVP (4), cystography (2), scintigraphy (3) Further investigation Detection of reflux:Ultrasound (28): standard (11), contrast enhanced (17) Detection of reflux:IVP (4), voiding radionuclide cystography (3), NAG/creatinine ratio (1), scintigraphy, (3), risk scoring system (1) Prediction of scarring:ultrasound (2), IVP (1), non-invasive indicators (1), MCUG (2) Detection of scarring: IVP (4), static scintigraphy (7), dynamic scintigraphy (2), MCUG (4), cystography (1), MRI (1), US and MCUG (1) ::: We estimated summary receiver operator characteristic (SROC) curves using the following equation \[[@B15]\]: ![](1471-2288-5-20-i1.gif) a and b were estimated by regressing D against S for each study: D = a + bS D = {logit (sensitivity) - logit (1-specificity)} = log diagnostic odds ratio (DOR) S = {logit (sensitivity) + logit (1-specificity)} We used both weighted and unweighted models. For the weighted model we weighted on sample size. We chose to weight on sample size rather than inverse variance, a method sometimes used in this type of analysis, as we believe that weighting on the inverse variance can produce biased results. The reason for this bias is that the DOR is associated with its variance and so large DORs will inevitably have large variances, which will be reflected in the weightings. We assessed between study heterogeneity through visual examination of forest plots and statistically using the Q statistic \[[@B16]\]. Where sufficient data were available, we used regression analysis to investigate whether individual QUADAS items and additional variables thought likely to be associated with diagnostic accuracy were associated with the DOR and hence whether differences in these items between the studies accounted for some of the observed heterogeneity. Where data were available, the following additional variables were investigated: • Patient age (\<2 years, \<5 years, \<12 years and \<18 years) was included to examine possible variation with age within the paediatric population. • The geographic region where studies were conducted was included to account for possible regional differences in test technology and infective agent. • Specific variations in index test technique were also included. For microscopy for pyuria and bacteriuria a variable on whether the sample was centrifuged was included, and for microscopy for bacteriuria a variable for Gram stain was included. For ultrasound for the detection of reflux a variable for whether or not the ultrasound involved a contrast agent was included. • The SROC model \[[@B15]\], was extended to include each of the 14 QUADAS items and each of the variables above as individual covariates \[[@B17]\]. As each QUADAS item can be scored as \"yes\", \"no\" or \"unclear\", we included QUADAS items as categorical variables with 3 possible outcomes, thus including the comparisons of \"yes vs no\", and \"yes vs unclear\". This allowed us to make some distinction between associations of aspects of methodological quality with test performance and associations of completeness of reporting with test performance. A number of QUADAS items only received two of the three possible scores (i.e. were scored either \"yes\" or \"no\", or \"yes\" or \"unclear\", or \"no\" or \"unclear\"). These items were therefore included as dichotomous variables. A multivariate linear regression analysis was conducted. Initially, we performed univariate analysis with all items included separately in the model. Items that showed moderate evidence of an association with D, defined as p \< 0.10, were investigated further using step-down regression analysis. All items found to show moderate evidence of an association in the univariate models were entered into the multivariate model, then dropped in a step-wise fashion with the item with the weakest evidence of an association (largest p-value) dropped first. For covariates with more than one level, evidence of an association of one indicator variable with test performance was considered sufficient for inclusion in the model. The final model was achieved when all items remaining showed strong evidence of an association with D, defined as p \< 0.05. Interaction terms were not included. Associations of covariates with D were expressed as relative diagnostic odds ratios (RDOR). The DOR is used as an overall measure of diagnostic accuracy. It is calculated as the odds of positivity among diseased persons, divided by the odds of positivity among non-diseased. When a test provides no diagnostic evidence then the DOR is 1.0. The RDOR is calculated as the DOR when the covariate is present divided by the DOR when the covariate is absent. It therefore provides an indicator of the overall impact on diagnostic accuracy of the presence of a given covariate. Results ======= Results of QUADAS assessment ---------------------------- The proportion of QUADAS items fulfilled by studies included in our systematic review was similar for each of the three clinical stages assessed in the review. Studies evaluating tests to diagnose UTI fulfilled a median of 8 (range 5--13) items, those evaluating tests used to localise infection also fulfilled a median of 8 (range 3--13) items, and those evaluating further investigations fulfilled a median of 7.5 (range 3--12) items. Figure [1](#F1){ref-type="fig"} illustrates the number of QUADAS items fulfilled by studies in each category. The similarity in numbers of QUADAS items fulfilled masks apparent differences in the individual items fulfilled. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Numbers of quality items fulfilled by studies in the three sections of the review. ::: ![](1471-2288-5-20-1) ::: Figure [2](#F2){ref-type="fig"} shows the proportion of studies that scored \"yes\", \"no\" and \"unclear\" for each of the QUADAS items, separately for the three sections of the review. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Proportion of studies rated as yes, no or unclear for each of the QUADAS items, separately for diagnosis of UTI, localisation of infection and investigation of confirmed UTI. ::: ![](1471-2288-5-20-2) ::: ### Tests for the diagnosis of UTI (n = 79 studies) \[[@B18]-[@B96]\] The use of an inappropriate spectrum of patients and inadequate reporting of inclusion criteria were problematic for studies in this category. The majority of studies provided insufficient details on how the reference standard was performed. Studies failed to report sufficient details on clinical review bias, diagnostic review bias and test review bias to judge whether these were avoided. Study withdrawals and handling of uninterpretable results were also poorly reported. ### Tests for the localisation of infection (n = 39 studies) \[[@B48],[@B69],[@B97]-[@B133]\] The time delay between the index test and reference standard was more of a problem with these studies than with those on the diagnosis of UTI. The use of an appropriate reference standard was also an issue in some of these studies. Spectrum composition and reporting of details of how children were selected for inclusion in the study was better in these studies than in the studies of the diagnosis of UTI. Only around half of studies provided sufficient details of how the index test and reference standard were performed to allow replication of these tests. More studies in this category, almost 40%, provided information indicating that test and diagnostic review bias had been avoided, in the remainder of studies this information was not reported. As with studies of the diagnosis of UTI, reporting of clinical review bias, handling of uninterpretable results, and withdrawals from the study was poor. ### Tests for the further investigation of confirmed UTI (n = 71 studies) \[[@B134]-[@B204]\] As with studies of the diagnosis of UTI, spectrum composition and reporting of inclusion criteria were poor in this group. The time delay between the index test and reference standard was also an issue in many of these studies. Around half of studies reported that diagnostic and test review bias had been avoided, the remaining studies did not report whether the index test and reference standard were interpreted blind to the results of each other. This was similar to the situation seen for studies on the localisation of infection. Reporting of the reference standard was poor. As in all previous groups, studies also provided very little information on whether appropriate clinical information was available when test results were interpreted, how uninterpretable results were handled, and whether there were any withdrawals from the study and if so whether all withdrawals were accounted for. Results of multivariate regression analysis ------------------------------------------- ### Tests for the diagnosis of UTI Tests involving dipstick or microscopy techniques were the only categories where enough studies were available to enable regression analysis. Table [3](#T3){ref-type="table"} summarises the results of the regression analysis for studies assessing dipstick tests. For dipstick to detect urinary nitrite (23 studies) \[[@B20],[@B26],[@B28],[@B34],[@B36],[@B40],[@B41],[@B43],[@B52],[@B54]-[@B57],[@B60],[@B63],[@B66],[@B72],[@B74],[@B84],[@B88],[@B93]-[@B95]\], the weighted analysis found that studies reporting that clinical review bias had been avoided had a DOR 4.7 (95% CI: 1.7, 12.7) times greater than those which did not report on whether clinical information was available to those interpreting the test results (p = 0.004). This is what would be expected, as the DOR is likely to be higher when those interpreting test results have access to appropriate clinical information similar to that, which would be available in practice. No studies reported the presence of clinical review bias. This was the only item investigated to show strong evidence of an association with test performance in the weighted multivariate analysis, although age and geographic region did show moderate evidence of an association in the univariate analysis. The unweighted analysis showed slightly different results. The same three items were found to show at least moderate evidence of an association in the univariate analysis. However, only country remained in the multivariate model, suggesting that studies conducted in North America showed higher accuracy than studies conducted in Europe or other areas (p \< 0.05). ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Results of the regression analysis for dipstick tests for the diagnosis of UTI ::: Univariate analysis Multivariate model ----------------------------------------------------------------------- ------------------------- ------------------------ ------------------- ------- Nitrite dipstick -- weighted (n = 23 studies) Clinical review bias avoided: yes vs unclear 4.7 (1.7, 12.7) 0.004 4.7 (1.7, 12.7) 0.004 Age \<5 years vs \<2 years 5.8 (0.3, 101.8) 0.213 Dropped^\$^ Age \<12 years vs \<2 years 2.2 (0.1, 35.8) 0.548 Dropped Age \<18 years vs \<2 years 3.6 (0.9, 14.9) 0.076 Dropped Europe vs North America 0.3 (0.1,0.9) 0.041 Dropped Other areas vs North America 0.3 (0.1, 1.0) 0.050 Dropped Nitrite dipstick -- unweighted (n = 23 studies) Clinical review bias avoided: yes vs unclear 3.4 (0.9, 13.7) 0.078 Dropped Age \<5 years vs \<2 years 7.3 (0.9, 61.8) 0.067 Dropped Age \<12 years vs \<2 years 2.9 (0.5, 18.3) 0.243 Dropped Age \<18 years vs \<2 years 3.8 (1.1, 13.4) 0.039 Dropped Europe vs North America 0.3 (0.1, 1.0) 0.044 0.3 (0.1, 1.0) 0.044 Other areas vs North America 0.3 (0.1, 1.2) 0.089 0.3 (0.1, 1.2) 0.089 LE dipstick -- weighted (n = 14 studies) No association at p \< 0.10 LE dipstick -- unweighted (n = 14 studies) Test details reported: yes vs no 19.0 (1.9, 192.2) 0.017 Dropped Age \<5 years vs \<2 years 5.4 (0.5, 64.2) 0.158 5.4 (0.5, 64.2) 0.158 Age \<12 years vs \<2 years 28.1 (2.3, 343.3) 0.015 28.1 (2.3, 343.3) 0.015 Age \<18 years vs \<2 years 1.3 (0.3, 4.7) 0.703 1.3 (0.3, 4.7) 0.703 Nitrite or leukocyte esterase dipstick -- weighted (n = 15 studies) Reference standard details reported 4.5 (0.9, 22.5) 0.064 Dropped North America vs Europe 5.0 (0.8, 10.5) 0.076 Dropped North America vs Other areas 1.1 (0.28, 5.0) 0.854 Dropped Nitrite or leukocyte esterase dipstick -- unweighted (n = 15 studies) No association at p \< 0.10 \Only items that showed moderate evidence (p \< 0.10) for an association with the DOR in the univariate analysis are included. ^\$^Items were dropped where there were too few studies in one category to allow a coefficient to be calculated. ::: For dipsticks measuring urinary leukocyte esterase (14 studies) \[[@B20],[@B28],[@B34],[@B36],[@B43],[@B56],[@B57],[@B60],[@B63],[@B66],[@B72],[@B84],[@B94],[@B95]\] and for dipsticks for the presence of either nitrite or leukocyte esterase (15 studies) \[[@B19]-[@B21],[@B28],[@B34],[@B56],[@B60],[@B63],[@B66],[@B84]-[@B86],[@B92],[@B94]-[@B96]\], no items showed strong evidence of an association with the DOR in the weighted analysis. However, for urinary leukocyte esterase, the unweighted analysis found strong evidence of an association between patient age and the DOR. There was strong evidence (p = 0.015) that the dipstick was more accurate in children aged \<12 years than in those aged \<2 years (RDOR = 28.1, 95% CI: 2.3, 343.3). There was no evidence of any difference in accuracy between children aged \<18 years and those aged \<2 years (p = 0.703), and very little evidence of any difference between children aged \<5 years and those aged \<2 years (p = 0.158). Table [4](#T4){ref-type="table"} summarises the results of the regression analysis for studies that assessed the accuracy of microscopy. In studies evaluating microscopy to detect pyuria three items showed a strong association with test performance in the weighted analysis (28 studies) \[[@B19]-[@B23],[@B28],[@B29],[@B34],[@B35],[@B41],[@B43],[@B46],[@B47],[@B49],[@B50],[@B58],[@B59],[@B63],[@B67],[@B70],[@B75],[@B77],[@B80],[@B81],[@B83],[@B85],[@B92]-[@B94]\]. The DOR was 1.3 (95% CI: 1.1, 1.6; p = 0.007) times higher in studies that adequately reported details of the reference standard execution. The DOR was lower, RDOR = 0.2 (95% CI: 0.1, 0.4; p \< 0.001) in studies that did not report on reasons for withdrawals compared to studies in which it was unclear whether there were any withdrawals, and 1.8 times higher (95 % CI: 1.0, 3.4; p = 0.056) in studies in which withdrawals were accounted for compared to those in which this was unclear. The DOR was lower, RDOR = 0.2 (95% CI: 0.1, 0.3; p \< 0.001), in studies where samples were centrifuged compared to studies in which samples were not centrifuged. In the unweighted analysis, only centrifugation showed any evidence of an association with test performance (p = 0.08). All of these items, with the exception of centrifugation, relate to the completeness of reporting. The association for centrifugation is counter-intuitive, as we would expect centrifugation of the sample to lead to improved test accuracy. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Results of the regression analysis for microscopy for the diagnosis of UTI ::: Univariate analysis* Multivariate model ------------------------------------------------------------- ------------------------- ------------------------ ----------------- --------- Microscopy for pyuria: weighted (n = 28 studies) Selection criteria reported: yes vs no 2.4 (1.0, 5.9) 0.057 Dropped^\$^ Reference standard details reported: yes vs no 3.6 (0.8, 16.1) 0.089 1.3 (1.1, 1.6) 0.007 Test review bias avoided: yes vs unclear 4.8 (1.6, 14.7) 0.008 Dropped Diagnostic review bias avoided: yes vs unclear 5.5 (1.8, 17.1) 0.005 Dropped Uninterpretable results reported: no vs unclear 0.3 (0.0, 4.1) 0.364 Dropped Uninterpretable results reported: yes vs unclear 2.9 (0.9, 9.1) 0.073 Dropped Withdrawals accounted for: no vs unclear 0.2 (0.1, 0.7) 0.012 0.2 (0.1, 0.4) \<0.001 Withdrawals accounted for: yes vs unclear 1.9 (0.7, 5.3) 0.200 1.8 (1.0, 3.4) 0.056 Europe vs North America 0.3 (0.1, 1.6) 0.190 Dropped Asia vs North America 0.6 (0.1, 6.6) 0.646 Dropped Other areas vs North America 0.2 (0.0, 1.3) 0.090 Dropped Age \<5 years vs \<2 years 0.1 (0.0, 0.4) 0.004 Dropped Age \<12 years vs \<2 years 0.1 (0.0, 0.3) 0.004 Dropped Age \<18 years vs \<2 years 0.2 (0.1, 0.5) 0.001 Dropped Sample centrifuged: yes vs no 0.3 (0.1, 0.6) 0.005 0.2 (0.1, 0.3) \<0.001 Microscopy for pyuria: unweighted (n = 28 studies) Age \<5 years vs \<2 years 0.3 (0.1, 1.1) 0.076 Dropped Age \<12 years vs \<2 years 0.5 (0.1, 2.1) 0.337 Dropped Age \<18 years vs \<2 years 0.3 (0.1, 1.0) 0.048 Dropped Sample centrifuged 0.5 (0.2, 1.1) 0.080 0.5 (0.2, 1.1) 0.080 Microscopy for bacteriuria: weighted (n = 22 studies) Incorporation bias avoided: no vs yes 24.5 (1.0, 604.4) 0.050 3.0 (1.6, 5.5) 0.001 Diagnostic review bias avoided: yes vs unclear 3.2 (0.8, 12.8) 0.092 Dropped Gram stain used: yes vs no 3.6 (1.3, 10.4) 0.018 5.3 (2.3, 12.0) 0.001 Microscopy for bacteriuria: unweighted (n = 22 studies) Disease progression bias: yes vs unclear 0.05 (0.0, 1.5) 0.083 Dropped Incorporation bias avoided: no vs yes 32.5 (1.2, 895.0) 0.041 3.2 (1.6, 6.4) 0.003 Uninterpretable results reported: yes vs unclear 7.1 (1.1, 46.9) 0.042 Dropped Withdrawals reported: no vs unclear 6.6 (1.0, 43.3) 0.049 Dropped Withdrawals reported: yes vs unclear 2.2 (0.3, 18.7) 0.447 Dropped Sample centrifuged: yes vs no 0.2 (0.0, 1.1) 0.058 Dropped Gram stain used: yes vs no 3.9 (0.9, 16.3) 0.062 6.5 (2.0, 21.2) 0.004 \Only items that showed moderate evidence (p \< 0.10) for an association with the DOR in the univariate analysis are included. ^\$^Items were dropped where there were too few studies in one category to allow a coefficient to be calculated. ::: Two items showed a strong evidence of an association with the DOR in the weighted analysis of studies evaluating microscopy to detect bacteriuria (22 studies) \[[@B20],[@B21],[@B23],[@B28],[@B34],[@B35],[@B41],[@B47],[@B50],[@B61]-[@B64],[@B67],[@B70],[@B76],[@B77],[@B80],[@B85],[@B90],[@B91],[@B94]\]. The DOR was 3.0 (95% CI: 1.6, 5.5, p = 001) times greater in studies in which incorporation bias was present compared to those in which it was avoided, and 5.3 (95% CI: 2.3, 12.0, p = 0.001) times greater if samples were Gram stained. We would expect both Gram staining and the presence of incorporation bias to increase test performance as found in the analysis. The unweighted analysis found very similar results. ### Tests for the localisation of infection Only the evaluation of ultrasound for the localisation of infection provided sufficient data to enable the conduct of regression analysis (20 studies) \[[@B48],[@B69],[@B97],[@B99]-[@B102],[@B109]-[@B111],[@B113]-[@B115],[@B117],[@B118],[@B121],[@B126],[@B128],[@B132],[@B133]\]. None of the QUADAS items, or other items investigated, showed moderate evidence of an association with the DOR in this analysis, using either the weighted or unweighted model. ### Tests for the further investigation of confirmed UTI Table [5](#T5){ref-type="table"} summarises the results of the regression analysis for studies assessing this clinical stage. The use of ultrasound to detect reflux was the only test in this category with sufficient data to support regression analysis (28 studies) \[[@B69],[@B135],[@B137],[@B140],[@B141],[@B150],[@B152],[@B153],[@B155],[@B164],[@B169],[@B170],[@B172],[@B176]-[@B178],[@B181],[@B185],[@B187],[@B189],[@B190],[@B195]-[@B198],[@B202]-[@B204]\]. Three items showed strong evidence of an association with the DOR in the weighted analysis. The DOR was 8.0 (95% CI: 2.9, 22.0; p \< 0.001) times greater in studies that used contrast enhanced ultrasound compared to those that used standard ultrasound. As this was also thought to be a clinically important variable it was included in all further analyses. The DOR was 1.4 (95% CI: 1.0, 1.9; p = 0.033) times higher in studies that reported that disease progression bias had been avoided compared to those in which this information was not reported. No studies reported sufficient information to determine that disease progression bias was present. Studies in which details were provided on reasons for withdrawals had DORs that were 2.8 times higher (95% CI: 1.1, 6.9, p = 0.033) than those in which it was unclear whether there had been any withdrawals. There was no evidence of any difference in the DOR between studies that did not report on reasons for withdrawals and studies in which it was unclear whether there were any withdrawals (p = 0.502). In the unweighted analysis, only two items showed a strong evidence of an association with the DOR. As in the weighted analysis there was very strong evidence that the DOR was higher in studies that used contrast enhanced ultrasound than those that used standard ultrasound (RDOR = 29.8, 95% CI: 13.5, 65.8, p \< 0.001). Studies in which partial verification bias was avoided had DORs 4.1 times higher (95% CI: 1.1, 14.8) than those that did not (p = 0.034). ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Results of the regression analysis for ultrasound for the diagnosis of reflux ::: Univariate analysis* Multivariate analyis --------------------------------------------------------------------------------------------------------------------- ------------------------- -------------------------- ------------------- --------- Ultrasound for the detection of reflux: weighted (n = 28 studies) Use of contrast enhanced ultrasound: yes vs no 23.9 (9.8, 58.8) \<0.001 8.0 (2.9, 22.0) \<0.001 Ultrasound for the detection of reflux, with ultrasound type forced into the model: weighted (n = 28 studies) Appropriate reference standard: yes vs unclear ^+^ 0.2 (0.0, 1.0) 0.047 Dropped^\$^ Disease progression bias avoided: yes vs unclear 3.5 (1.4, 9.2) 0.011 1.4 (1.0, 1.9) 0.033 Withdrawals accounted for: yes vs unclear 3.2 (1.2, 8.5) 0.020 2.8 (1.1, 6.9) 0.027 Withdrawals accounted for: no vs unclear (0.4, 0.1, 1.7) 0.175 0.6 (0.1, 2.8) 0.502 Ultrasound for the detection of reflux: unweighted (n = 28 studies) Use of contrast enhanced ultrasound: yes vs no 29.8 (13.5, 65.8) \<0.001 29.8 (13.5, 65.8) \<0.001 Ultrasound for the detection of reflux, with ultrasound type forced into the model: unweighted (n = 28 studies) Appropriate reference standard \: yes vs unclear 0.2 (0.0, 1.2) 0.075 Dropped Partial verification bias avoided: yes vs no 4.1 (1.1, 14.8) 0.034 4.1 (1.1, 14.8) 0.034 \Only items that showed moderate evidence (p \< 0.10) for an association with the DOR in the univariate analysis are included. ^+^Only 1 unclear ^\$^Items were dropped where there were too few studies in one category to allow a coefficient to be calculated. ::: Discussion ========== The methodological quality of primary studies remains a significant issue for systematic reviews of diagnostic tests \[[@B8],[@B205],[@B206]\]. The STARD initiative has provided clear guidance for the reporting of diagnostic accuracy studies \[[@B5]\]. This should have a positive impact on the quality of the diagnostic literature in the future. The QUADAS tool facilitates systematic evaluation of the quality of diagnostic accuracy studies, and was specifically developed for use in systematic reviews of diagnostic tests \[[@B12]\]. However, where studies are poorly reported the information that can be derived from quality assessment becomes limited. We cannot know whether an unreported QUADAS item reflects a true methodological flaw or poor reporting of a study that may be methodologically sound. Many of the studies included in our review were poorly reported. Our assessment of the impact of components of methodological quality on diagnostic accuracy may therefore partially reflect completeness of reporting. Whilst poor reporting remains a widespread problem, it is almost impossible to assess the impact of components of methodological quality on the results of diagnostic meta-analyses. The common practice of using summary quality scores in systematic reviews has been widely debated elsewhere \[[@B207]-[@B209]\]. Summary scores, when used to inform quality-based analyses, may mask important effects of individual quality components \[[@B210]\]. As we report, the numbers of QUADAS items that were adequately addressed by studies included in our review were similar between the three clinical stages assessed in the review. Had the number of QUADAS items fulfilled been used as a summary score, potentially important variations in the individual items fulfilled would have been hidden. We therefore advocate that components of quality assessment should be reported fully, and their impact on outcome measures analysed individually rather than as summary scores. Although ours was a large review, it included 187 studies reporting 487 data sets, our analysis of the impact of methodological quality on diagnostic accuracy was severely limited both by the diversity of the included studies (few tests were evaluated by sufficient studies to allow meaningful use of meta-analytic pooling and investigation of heterogeneity), and by incomplete reporting. All of the data sets used were sub-optimal, in that the numbers of observations were low in comparison to the number of variables investigated in the multivariate analyses\[[@B13]\]. Although different types of diagnostic tests were evaluated in the three clinical stages used by the review, generalisibility is limited in that all data concerned a single condition (UTI). A number of the items found to be associated with test performance related to specific test methodologies (e.g. Gram stain and contrast-enhanced ultrasound) and have no generalisability elsewhere. These items were found to show association in both the weighted and unweighted analyses. For the individual quality items there were some differences between the results of the weighted and unweighted analyses. In general, the results of weighted analyses showed more intuitive associations. Unweighted analyses more often produced results that were difficult to explain, for example, in leukocyte esterase dipstick tests the unweighted analysis found that the test was more accurate in the group of children aged \<12 years than in those aged \<2 years. This might be expected and would probably reflect a higher likelihood of sample contamination in younger children, however, no difference in accuracy was found between under 18\'s and children aged \<2 years. For both tests on the diagnosis of and further investigation of UTI weighted analyses showed an association between a number of variables relating to quality of reporting and diagnostic accuracy (well reported studies had higher DORs). We might expect this association to extend to diagnostic accuracy studies of all types of tests, but the present study is not adequate to demonstrate this. Weighted analysis of studies of ultrasound for the detection of reflux showed that the DOR was higher where studies reported information to determine that disease progression bias had been avoided. Disease progression bias is a particular issue for imaging studies of this type where follow-up examinations (used as the reference standard of diagnosis) may be scheduled some time after ultrasound (usually the initial examination). This association was not shown in the unweighted analysis. The information derived from these analyses is also limited by the use of the summary ROC approach to pool studies. This method takes the DOR as the dependent variable. The DOR is used as a single indicator of test performance and shows how much more frequently a positive test result occurs in a person with the condition of interest than in one without the condition, relative to how much more frequently a negative result occurs in a person without the condition than in one with the condition. Using the DOR to investigate heterogeneity means that we cannot assess whether the factors investigated are associated with paired measures of diagnostic accuracy, such as sensitivity and specificity, or positive and negative likelihood ratios. Often factors that lead to an increase in sensitivity will lead to a decrease in specificity and vice versa. Factors that lead to this pattern of change may have no effect on an overall measure such as the DOR. Using the DOR to investigate heterogeneity may thus miss relevant clinical associations. Recently a new method for pooling sensitivity and specificity has been developed. This method is known as the \"bivariate model\" \[[@B211]\]. It preserves the underlying two-dimensional nature of the data and produces direct pooled estimates of sensitivity and specificity, incorporating any correlation that might exist between these two measures. The model can be extended to include explanatory variables leading to separate effects on sensitivity and specificity. This method has two advantages over the standard methods: (1) the pooled estimates of sensitivity and specificity take into account the correlation between these two measures; (2) the effect of possible sources of heterogeneity on both sensitivity and specificity can be investigated in a single model rather than just looking at the effect of these variables on a single measure of test performance, the DOR. These methods may have potential applications in future studies of this type. Conclusion ========== Given the limitations we describe, the results of this study should be treated as hypothesis generating. Further work is needed to elucidate the influence of components of the methodological quality of primary studies on the results of diagnostic meta-analyses. Large data sets of well-reported primary studies are needed to address this question. Without significant improvements in the reporting of primary studies, progress in this area will be limited. The components of quality assessment should always be reported, and their impact on summary outcome measures be investigated, individually rather than as summary quality scores. Careful consideration should be given to the choice of weighting when conducting regression analyses. Weighting by sample size appears the most appropriate method for analyses of diagnostic accuracy studies, but this area requires further investigation. Competing interests =================== The author(s) declare that they have no competing interests Authors\' contributions ======================= All authors contributed towards the conception and design of the study and the interpretation of the data. They also read and approved the final manuscript. PW and MW participated in data extraction, the analysis of data, and drafted the article. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2288/5/20/prepub> Acknowledgements ================ We would like to thank Professor Martin Bland of the University of York and Mr Roger Harbord of the University of Bristol for their statistical advice. The work was done as part of a project commissioned and funded by the NHS R&D Health Technology Assessment Programme (project number 01/66/01). The views expressed in this review are those of the authors and not necessarily those of the Standing Group, the Commissioning Group, or the Department of Health.	PubMed Central	2024-06-05T03:55:59.471045	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180444/", "journal": "BMC Med Res Methodol. 2005 Jun 8; 5:20", "authors": [ { "first": "Marie E", "last": "Westwood" }, { "first": "Penny F", "last": "Whiting" }, { "first": "Jos", "last": "Kleijnen" } ] }
PMC1180445	Background ========== Prediction of cancer incidence is of great interest for both health authorities and the scientific community \[[@B1]\]. Estimates of the future cancer burden should indicate the appropriate amounts of resources that will be needed for diagnosis, treatment and rehabilitation. Since quantification of future cancer incidence is inherently uncertain \[[@B2]\], some measurement of the uncertainty would be useful. It has been suggested that, similar to the confidence intervals calculated in standard statistical modeling, prediction intervals should be presented with predictions of cancer incidence \[[@B3],[@B4]\]. When future cancer incidence is predicted in a statistical model, three sources of uncertainty are associated with the predicted numbers. The first is the variance of the parameters estimated in the model; the second is the random variation of the future number of cases; and the third is the adequacy of the model. The last includes the uncertainty in both the mathematical structure of the model and the choice of projected components, such as whether to assume that current trends will continue into the future. The first two sources of variance can be included in a statistical model. Intervals that are based only on the variance of the parameters estimated in the model are called \'confidence intervals\', while intervals that also encompass variation in the future number of cases are called \'prediction intervals\'. The third source of uncertainty, variations caused by deviations from the model assumptions, is difficult to formalize. Engeland et al. \[[@B5]\] argued that \"The large uncertainties associated with the specification of the models used in the predictions obviate the construction of confidence intervals.\" One way of investigating the extent of the uncertainty in specification of a model is to calculate prediction intervals from historical data and then calculate the proportion of those intervals that actually cover the observed number of cancer cases. If the proportion is close to the nominal level, e.g. 95 %, the prediction intervals could be taken to give a fair description of the range of likely values to be expected. A low proportion would indicate that there was great uncertainty in the model assumptions, and that prediction intervals should be interpreted with caution. The aim of this study was to evaluate the extent to which prediction intervals derived from incidence rates can be expected to cover the actual numbers observed 10 years later. Methods ======= Material -------- The material consisted of new cases of cancer reported to the cancer registries of Denmark, Finland, Iceland, Norway and Sweden between 1958 and 1997, and population figures covering the same calendar period from the central statistical offices in these countries. Sweden has the largest population, with 8.9 million inhabitants, followed by Denmark, Finland, Norway and Iceland with 5.3, 5.2, 4.5 and 0.3 million, respectively. The Nordic cancer registries receive reports from physicians, hospitals, pathological and cytological laboratories and (except in Sweden) death certificates \[[@B6]\]. Compulsory reporting and information from multiple sources ensure almost 100% completeness of all the cancer registries \[[@B7]\]. Table [1](#T1){ref-type="table"} lists the cancer sites included in the study. Detailed descriptions of the types of tumors at each site that are included was given by Engeland et al. \[[@B5]\]. As there were few cases of cancers of the lip and larynx among women and of the breast among men, these cancers were included in \'other sites\'. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Proportions of observed numbers of cases in 1993--97 covered by prediction intervals based on trends up to 1987, by site, sex, country and frequency\. ::: Number of intervals Coverage number (%) P-value for difference ---------------------------------- --------------------- --------------------- ------------------------ Site 0.54 Lip 5 2 (40) Tongue, oral cavity and pharynx 10 5 (50) Oesophagus 10 5 (50) Stomach 10 5 (50) Colon 10 7 (70) Rectum 10 5 (50) Pancreas 10 6 (60) Larynx 5 3 (60) Lung 10 4 (40) Breast 5 2 (40) Cervix uteri 5 2 (40) Corpus uteri 5 2 (40) Ovary 5 1 (20) Prostate 5 2 (40) Testis 5 4 (80) Kidney 10 7 (70) Urinary bladder 10 5 (50) Melanoma of the skin 10 3 (30) Thyroid 10 7 (70) Non-Hodgkin lymphomas 10 4 (40) Hodgkin disease 10 9 (90) Multiple myeloma 10 4 (40) Acute leukaemia 10 7 (70) Other sites 10 3 (30) Sex 0.67 Male 100 50 (50) Female 100 54 (54) Country \<0.001 Denmark 40 18 (45) Finland 40 21 (53) Iceland 40 35 (88) Norway 40 20 (50) Sweden 40 10 (25) Frequency \<0.001 ≤ 70 per year 50 42 (84) \> 70 and ≤ 230 per year 50 26 (52) \> 230 and ≤ 555 per year 50 23 (46) \> 555 per year 50 13 (26) \Average number of cases per year in the period 1983--87, grouped into four quartiles. ::: The data were tabulated in five-year age groups (0--4, 5--9, \..., 80--84, ≥85) and five-year calendar periods (1958--62, 1963--67, \..., 1993--97). Since there were 20 types of cancers for each sex and five countries in the study, we had 200 different combinations for which to make predictions. Statistical model ----------------- The age-period-cohort (APC) model \[[@B8]\] has been widely used for predicting cancer incidence and mortality \[[@B9]-[@B13]\]. This Poisson regression model is based on tables with five-year age groups and five-year calendar periods. Birth cohorts were constructed synthetically by subtracting age from period. The model can be written as R~ap~= exp(A~a~+ D·p+ P~p~+ C~c~), where R~ap~is the incidence rate in age group ain calendar period p, Dis the common drift parameter \[[@B14]\], A~a~is the age component for age group a, P~p~is the non-linear period component of period pand C~c~is the non-linear cohort component of cohort c, c= p- a. We used a slightly modified model \[[@B5]\], substituting a power link for the log link: R~ap~= (A~a~+ D·p+ P~p~+ C~c~)^5^, in order to level off the exponential growth in the multiplicative model. An empirical study of these two models showed that the power model gave predictions that were closer to the observed rates \[[@B2]\]. For some cancer sites for which there were only a few cases in Iceland, a model without the cohort component was used; see Engeland et al. \[[@B5]\] for details and on the lower limits of the age groups included for each site in all the countries. To ensure a reasonable fit of each data set to the model, the number of five-year periods on which the predictions should be based on was chosen. First, a model including the last six 5-year periods (1958--1987) was fitted. If the model was rejected by a test for goodness-of-fit (5% level), a model including the last five periods was fitted. If this model was also rejected, only the last four periods were used. Future non-linear effects of cohort and period were assumed to be equal to the last estimated effect in the model, and predictions were made by projecting the drift. Numbers of cancer cases were predicted by multiplying the predicted rates by the person-years at risk in a given age group and time period. Prediction intervals, coverage level and discrepancy ratio ---------------------------------------------------------- In an article on the precision of cancer incidence predictions, Hakulinen and Dyba \[[@B3]\] derived prediction intervals for Poisson distributed variables. Following their paper, let ![](1471-2288-5-21-i1.gif), where ais the age group, fis the future calendar period for which predictions are to be made (f= 8 which corresponds to the period 1993--97), R~af~is the incidence rate in age group ain period f, and k~af~and n~af~are the corresponding number of cases and person-years, respectively. Further, let the expected number of cases, E(k~af~), be defined as λ~af~= n~af~(A~a~+ D·P+ P~f~+ C~c~)^5^, and let ![](1471-2288-5-21-i2.gif) and ![](1471-2288-5-21-i3.gif). The variance of the future number of cases, var (k~f~), can be written as var (k~f~) = var (![](1471-2288-5-21-i4.gif)) + σ^2^E(![](1471-2288-5-21-i4.gif)). The first component, var (![](1471-2288-5-21-i4.gif)), reflects the uncertainty in estimating the parameters in the model. The second component, σ^2^E(![](1471-2288-5-21-i4.gif)), reflects the random variation of the future number of cases for a Poisson-distributed variable, allowing for extra-Poisson variation, i.e., σ^2^measures the degree of over-dispersion. An estimator for the variance of the future number of cases can be found by using Taylor series expansion of non-linear functions (see formulas in the appendix). A 95% prediction interval for the future number of cancer cases, k~f~, can then be calculated from an assumption of normality: ![](1471-2288-5-21-i5.gif) where k~f~was estimated by ![](1471-2288-5-21-i4.gif). On the basis of up to six 5-year periods between 1958 and 1987, prediction intervals for the numbers of cases in the period 1993--97 were calculated for all 200 combinations of 20 sites for each sex in each of the five countries. The coverage level was defined as the proportion of the prediction intervals that covered the observed number of cases in 1993--97. The coverage level only indicates whether the observed number of cases was inside the prediction interval or not. Additional information can be gained by looking at how far outside of the interval the observations fall. Discrepancy ratio was defined as the absolute distance between observed and predicted number of cases, measured in half prediction interval widths: ![](1471-2288-5-21-i6.gif) where ![](1471-2288-5-21-i7.gif) and k~f~are the predicted and observed number of cases, respectively, and ![](1471-2288-5-21-i8.gif) is the distance between predicted number of cases and the limit of the prediction interval. Figure [1](#F1){ref-type="fig"} illustrates the discrepancy ratio. When the discrepancy ratio is larger than 1, it measures how much wider the prediction interval had to be to cover the observed number of cases. In Figure [1](#F1){ref-type="fig"}, the observed number of cases is about twice as far from the predicted number compared to the lower limit of the prediction interval, giving a discrepancy ratio of 2. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### The discrepancy ratio. Illustration of the components of the discrepancy ratio. The discrepancy ratio compares the distance between predicted and observed number of cases with the distance between predicted number and the limit of the prediction interval. ::: ![](1471-2288-5-21-1) ::: Fisher\'s exact test was used to evaluate differences between sites, countries and quartiles of number of cases. A binomial regression model was used to study the effect of country and frequency simultaneously. Results ======= Prediction intervals were calculated from data up to 1987 for the 200 combinations of sex, site and country. After observing the number of cases 10 years later, in the period 1993--97, 104 (52%) of the observed numbers were covered by the prediction intervals. Coverage levels for specific sites varied from 20% to 90%, but only five or ten intervals were calculated for each site and the difference between the sites was not statistically significant (Table [1](#T1){ref-type="table"}). The coverage levels for the five Nordic countries varied widely. For the country with the smallest population, Iceland, the coverage level was 88%, which is relatively close to the nominal level of 95%. The levels for Denmark, Finland and Norway were around 50 %, while that for Sweden, the most populous country, was only 25%. When the 200 different predictions were ranked according to the annual number of cases in the period 1983--87 (frequency), the cut-offs for the four quartiles were 70, 230 and 555. Subdividing the predictions according to these quartiles, the coverage level decreased markedly with the annual number of cases, being 84% for the first quartile and 52%, 46% and 26% for the next three, respectively (Table [1](#T1){ref-type="table"}). Table [2](#T2){ref-type="table"} shows the associations between country and coverage level as odds ratios (ORs), where the odds of covering the observed number of cases with the prediction intervals in each country was calculated relative to the odds in Denmark. The crude numbers reflect the pattern seen in Table [1](#T1){ref-type="table"}, the chance of the observed number of cases being within the prediction interval being similar in Denmark, Finland and Norway, higher in Iceland and lower in Sweden. In the binomial regression model the logarithm of the annual number of cases in 1983--87 was used instead of the frequency itself or the quartiles of the numbers. The reason for this was that the fit to the model was worse when the frequency variable was entered on a linear rather than on a log-linear scale. Also, entering the variable as four categories instead of as a continuous variable did not further improve the fit. The OR for the logarithm of the frequency was 0.52, indicating that an increase of one unit on the log scale of the frequency reduced the chance of covering the observed number of cases by about 50%. When country and frequency were mutually adjusted for in a multiple binomial regression analysis, the effect of country declined and became non-significant, while the effect of frequency remained almost unchanged. In the univariate analysis, the OR for Iceland relative to Denmark was 8.5 (95 % confidence interval (CI): 2.8 -- 26), which was reduced to 1.7 (95 % CI: 0.4 -- 7.6) in the adjusted analysis. The coverage level for Sweden remained lower, with OR = 0.49 (95 % CI: 0.2 -- 1.3) for Sweden relative to Denmark. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Odds ratios (ORs) with corresponding 95% confidence intervals (CIs) for covering the observed number of cases in the period 1993--97 with the prediction intervals calculated from trends up to 1987. ::: [Crude]{.underline} [Adjusted\]{.underline} ------------------- --------------------- -------------------------- ---------- ------ -------------- ------ Country \< 0.001 0.39 Denmark 1.00 Reference 1.00 Reference Finland 1.35 0.56 -- 3.2 1.18 0.47 -- 2.9 Iceland 8.54 2.8 -- 26.3 1.72 0.39 -- 7.6 Norway 1.22 0.51 -- 2.9 1.08 0.43 -- 2.7 Sweden 0.41 0.16 -- 1.1 0.49 0.18 -- 1.3 Log(frequency)^§^ 0.52 0.41 -- 0.66 \< 0.001 0.60 0.44 -- 0.83 0.02 \Both variables included in the model. ^§^Logarithm of average number of cases per year in the period 1983--87. ::: The distribution of the discrepancy ratio for each country is plotted in Figure [2](#F2){ref-type="fig"}. For Iceland, most of the discrepancy ratios were below 1, corresponding to a coverage level of 88%. Of the 12 % with a higher discrepancy ratio, none was more than 50 % outside of the prediction interval. In Finland and Norway, observed numbers of cases fell up to tree times further from the predicted numbers compared to the limits of the prediction intervals, while in Denmark and Sweden some predictions were 5--6 times outside of the intervals. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Empirical distribution of discrepancy ratio. Empirical distribution of discrepancy ratio by country. Predictions for the 40 combinations of 20 sites for each sex constitute the distribution in each country. ::: ![](1471-2288-5-21-2) ::: Discussion ========== The coverage levels in the five Nordic countries differed mainly as a function of the numbers of cases that formed the basis for the predictions. In Iceland, where there were generally few cases, the coverage level calculated from 95% prediction intervals was 88% while that in the other Nordic countries, which had much more cases, was of only 25--53%. The problem associated with interpretation of prediction intervals is illustrated in Figure [3](#F3){ref-type="fig"}, where the observed age-standardized (World standard \[[@B15]\]) incidence rates for cancer of the lung in women in Iceland and Denmark are plotted against the predicted rates, with 95 % prediction intervals. Although the difference between the observed and predicted rates was smaller in Denmark than in Iceland, the wide prediction interval for Iceland meant that the observed rate in 1993--97 was covered by the interval constructed for Iceland, but the narrower interval for Denmark failed to cover the observed rate for that country. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Illustration of prediction interval. Age standardized (World population) incidence rates of lung cancer among women in Iceland and Denmark. Predicted rates based on observed rates up to 1987, with corresponding 95% prediction interval for the period 1993--97 for each country. ::: ![](1471-2288-5-21-3) ::: A statistical model is only a simplification of true underlying associations between variables. The finding that the coverage level is inversely proportional to the sample size can be explained by considering the difference between the modeled (simplified) and true (complex) relationship between the cancer rate and the explanatory variables age, period and cohort. When the sample size increases, deviations between the modeled and true relationship will dominate, and non-overage will become a problem. It is useful to distinguish between calculations of prediction intervals for values within the observed range of values of the explanatory variables (interpolation), and outside the range (extrapolation). The non-coverage problem is even larger for extrapolations, because they also relay on the assumption of continuation of current trends. Another difference between interpolation and extrapolation is that when the sample size increases, the possibility to improve the model increases. This would then reduce the non-coverage problem for interpolations, by reducing the distance between the true and the modeled relationship between the variables. Extrapolations, on the other hand, consist of making predictions for values of the covariates outside the range of observed values. Thus, we have to make assumptions that cannot be evaluated from the observed data and the problem with non-coverage for extrapolations are not necessarily reduced by improvements in the model. It could be argued that a low coverage level indicates that the model is not appropriate, rather than that calculation of predication intervals is per se misleading. With the possible exception of tobacco and lung cancer, few associations are strong enough to be modelled directly. Instead, we used calendar period and birth cohort as proxy variables for changes in underlying risk factors in the model. Møller et al. \[[@B2]\] found that the method used in this study performed fairly well in comparison with other methods currently in use for predicting cancer incidence, and that all the methods evaluated missed the observed number of cases by 10--15% on average for 10 year predictions. Prediction methods could therefore be improved to increase the overall coverage level, although it would be unreasonable to expect that the correct model, or close to it, could be specified for all cancer sites. As long as predictions are based on some type of extrapolation from a statistical model, the coverage level will generally decrease as a function of sample size. The problem is that at the time when the prediction intervals are calculated, the appropriateness of the model with respect to extrapolation into the future usually cannot be evaluated. Prediction intervals can thus be misleading, if they are interpreted as the range of likely values for the number of cases to be expected. The number of cases from which the predictions for the different cancer sites were made varied widely. Cancers at some sites are very common, like those of the prostate, lung and breast, while others occur less frequently. Because the coverage levels vary with frequency, we would also have expected them to vary by site. The differences were not, however, statistically significant, probably because of the small number of prediction intervals calculated for each site. A relatively large sample size was associated with a narrow prediction interval, as seen for lung cancer among women in Denmark (Figure [3](#F3){ref-type="fig"}). The prediction intervals are based on asymptotic theory, which can result in underestimates of variance. Bootstrapping is a suitable method for investigating this problem \[[@B16]\]. We constructed a 95% prediction interval by bootstrapping the data for lung cancer among women in Denmark, assuming that each cell followed a binomial distribution in which the incidence rate and the number of person-years at risk were used as probability of success and number of trials, respectively. We re-sampled the data 1000 times, calculating the predicted world-standardized incidence rate each time. The 95% bootstrap interval, calculated by selecting the 2.5 and 97.5 percentiles of the 1000 predictions, was 32.7 -- 36.7, which is fairly close to the asymptotic interval of 32.6--36.8. This indicates that the asymptotic intervals calculated in this study describe the uncertainty in the predicted number of cases well, given a correctly specified model. Population forecasts are needed for predicting the number of cancer cases. In this paper, we assumed that these numbers were known, but in reality they constitute a separate source of uncertainty. Population forecasts are themselves extrapolations, relying on assumptions about future migration patterns, birth rates and death rates. If our projections for 1993--97 had been calculated from population forecasts in 1987, the coverage levels would probably have been even lower. Most of the differences in coverage level among the five countries disappeared when the number of cases was controlled for. The remaining difference was not significant, but the probability of covering the rates with prediction intervals for Sweden continued to be lower after adjustment of sample size. Møller et al. \[[@B2]\] showed that the predictions for Sweden were more different from the observed number of cases in 1993--97 than those in the other countries, measured as the median of the absolute value of the relative difference between the predicted and observed numbers of cases. This explains the lower coverage level for Sweden, and indicates that the trends current in 1987 continued to a lesser extent in Sweden than in the other countries. Engeland and co-workers \[[@B5]\] were reluctant to include prediction intervals with their predictions of cancer incidence in the Nordic countries. A similar view was expressed both with regard to an update of predictions for the Nordic countries \[[@B17]\], and to a prediction of cancer incidence in New South Wales, Australia \[[@B18]\]. There are, however, some instances where prediction intervals can be of value. In cancer surveillance, inclusion of prediction intervals in a routine comparison of the latest observed rates with rates predicted from previous trends, can help to identify changes in the rates beyond random variation. The potential reasons for any discrepancy between the observed and predicted rates can then be studied, including changes in risk factors, diagnostic methods or interventions such as screening programs. In Finland, predicted values with prediction intervals for 1980 were calculated based on rates up to 1968 and compared to observed number of cases in 1980 \[[@B19]\]. Of 33 prediction intervals, 22 (67%) covered the observed values, and the authors discussed possible reasons for those cancers where the prediction intervals failed to cover the observed number of cases. Prediction intervals can also be used to identify highly uncertain predictions. For instance, 26 male cancer cases of the lip were predicted in Iceland in the period 1993--97, and the 95 % prediction interval was 6--46 cases. Conclusion ========== We do not recommend use of prediction intervals when the predictions are used for administrative purposes, like planning appropriate amounts of resources for diagnosis, treatment and rehabilitation, as the intervals can give a false impression of the precision of the predictions. When predictions are based on larger numbers of cases, the uncertainty in estimating the parameters of the model and the random variation of the future rates are decreased. Even relatively minor deviations in the assumptions can then result in observed rates in the future that are outside the range of likely values indicated by the prediction interval. Appendix ======== A 95 % prediction interval for the future number of cancer cases, k~f~, can be calculated on the basis of an assumption of normality: ![](1471-2288-5-21-i9.gif) were k~f~can be estimated by ![](1471-2288-5-21-i4.gif). The variance of k~f~can be written ![](1471-2288-5-21-i10.gif), and ![](1471-2288-5-21-i11.gif) and ![](1471-2288-5-21-i12.gif) can be estimated by Taylor series expansion of non-linear functions: ![](1471-2288-5-21-i13.gif) Further, the parameter for over-dispersion, σ^2^, can be estimated from the ratio between the residual deviance of the model and the corresponding number of degrees of freedom \[[@B20]\], and ![](1471-2288-5-21-i14.gif) can be estimated by ![](1471-2288-5-21-i15.gif). We used the statistical package R in the calculations, and the variances and covariances of the estimates of the parameters in the model were found in the covariance matrix of the glm-object. ![](1471-2288-5-21-i16.gif) Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= BM designed the study based on discussions with HWF and TH. BM and HWF collected and analyzed the data. BM drafted the initial manuscript. All authors contributed and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2288/5/21/prepub> Acknowledgements ================ The authors would like to thank Petter Laake for valuable comments on the article. The project is supported by the Nordic Cancer Union.	PubMed Central	2024-06-05T03:55:59.475134	2005-6-10	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180445/", "journal": "BMC Med Res Methodol. 2005 Jun 10; 5:21", "authors": [ { "first": "Bjørn", "last": "Møller" }, { "first": "Harald", "last": "Weedon-Fekjær" }, { "first": "Tor", "last": "Haldorsen" } ] }
PMC1180446	Background ========== Mechanical or chemical changes in degenerative intervertebral discs comprise a hypothesized cause of low back pain without sciatica or neurological deficits. This is commonly referred to as \"discogenic\" low back pain, and is distinct from a herniated disc causing sciatica. The diagnosis often leads to spinal fusion surgery and may partly explain the recent rapid increase in lumbar fusion operations in the United States (U.S.) \[[@B1],[@B2]\]. A few randomized studies have examined outcomes of spinal fusion surgery compared to non-surgical treatment for discogenic pain; these suggest little or no advantage of surgery over carefully designed rehabilitation therapy \[[@B3]-[@B6]\]. In one study, an early advantage of surgery was lost with longer follow-up \[[@B4],[@B7]\]. Major surgical complication rates have been as high as 19% \[[@B5],[@B8]\]. Furthermore, the only randomized trials have been conducted in Europe, where dramatically lower surgery rates suggest a more selective practice style than in the U.S. \[[@B9]\]. Thus, more data are needed on safety and outcomes in routine care for discogenic back pain in the U.S. Although the concept of a painful disc causing back pain has merit on theoretical grounds, mechanical and chemical changes develop in all vertebral discs with aging. Clinical studies have not firmly established diagnostic criteria that distinguish patients with painful discs from others with normal aging. The distinction is often based on lumbar discography, itself a controversial procedure troubled with high rates of false positive results \[[@B10]-[@B12]\]. Research also suggests that patients with psychological distress are more likely to report pain on discography than those without psychological distress \[[@B12],[@B13]\]. Furthermore, clinical studies of patients with chronic back pain provide strong evidence that psychological distress is an important risk factor for having poor outcomes after spine surgery \[[@B14]-[@B16]\]. Some patients with psychological distress, therefore, may be preferentially treated with fusion, and paradoxically have poor outcomes. Even when patients are meticulously selected on the basis of discography and clinical screening, fusion results are often poor \[[@B12]\]; this challenges the concept that discography correctly identifies the source of pain and that spinal fusion corrects this problem. Little is known about how patients undergoing lumbar fusion compare in preoperative physical and psychological function to patients who have degenerative discs, but receive only non-surgical care. Finally, useful models to predict which patients will have a good response to surgical therapy have not been developed. We implemented a prospective cohort study to compare patients with presumed discogenic pain who undergo lumbar fusion with those who have non-surgical care. The study is intended to address several questions. First, are there differences in preoperative physical and psychological function between those having surgical versus non-surgical treatment? For example, do patients undergoing fusion for discogenic pain have greater preoperative psychological distress than patients not undergoing fusion? Secondly, how do treatment outcomes, including symptoms, functional status, return to work and subsequent surgery, differ between the two groups? Do the results support the conventional wisdom that outcomes improve more after fusion than after non-surgical care? A final study objective is to identify predictors of favorable outcomes of both surgical and non-surgical care for discogenic pain. Are there characteristics that predict a good response to surgical therapy, but not to non-surgical therapy? Such knowledge could result in better patient selection for surgery. Potential predictor characteristics include demographics, work and disability compensation status, initial symptom severity and duration, imaging results, functional status, and psychological distress. Methods ======= We designed a multi-center prospective cohort study of patients with presumed discogenic back pain. We will use follow-up assessments over two years to compare health outcomes between those who receive a spinal fusion versus those who receive other treatment. Important secondary analyses will include comparisons of those who have elective surgery and those who do not in terms of preoperative physical and psychological function and identification of patient characteristics that predict outcomes. The study protocol was approved by the University of Washington (U.W.) Human Subjects Division and all participants provide written informed consent. Patients are recruited from five orthopaedic clinics in the Puget Sound region of Washington State: the U.W. affiliated practice sites (U.W. Medical Center and Harborview Medical Center), Orthopaedics International (affiliated with Providence Medical Center and Evergreen Hospital), and Proliance Surgeons, Inc., P.S. (associated with Orthopedics Physician Associates and Swedish Hospital in Seattle). These practices include eleven surgeons who have referred patients to the study. We identify patients with predominant low back pain as a symptom, one or two-level disc degeneration confirmed by imaging, and a normal neurological exam. Because surgery for discogenic pain is rarely considered as initial therapy, we required patients to have pain lasting for at least six months. We did not require discography for diagnosis. Additional inclusion and exclusion criteria are listed in Table [1](#T1){ref-type="table"}. These eligibility criteria mimic those of the European randomized trials of surgery for disc for discogenic pain \[[@B4]-[@B6]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Inclusion and exclusion criteria. ::: Inclusion Criteria: -------------------------------------------------------------------------------------------------- Age greater than 20 years and less than 65 years Back pain of greater than 6-month duration Has telephone and willing to complete telephone interviews Ability to communicate in English Physician diagnosis of discogenic back pain One or two level disc degeneration on MRI scan Exclusion Criteria: Leg pain greater than back pain Pregnant Instability on lumbar flexion-extension radiographs (if done) Motor deficit on physical examination Abnormal electrodiagnostic studies (if done) Lumbar spinal stenosis requiring a laminectomy. Prior lumbar fusion or multilevel laminectomy Spondylolisthesis of more than 25% (\>grade I) Inflammatory spondyloarthropathy Spinal malignancy or infection Severe co-morbid illness that would contraindicate surgery or result in major functional decline Developmental spine deformities or vertebral fractures Disc protrusion or extrusion on imaging with either: Nerve root compression or displacement, or pain radiating below the knee ::: Patients are classified as surgical or non-surgical based on the treatment they receive during the six months following study enrollment. We use both patient interview and medical record information to determine whether patients have had surgery. Surgical details, including the type of procedure, levels operated, and surgical implants, are recorded from operative notes in the medical records. These details are important because there are several variations in surgical technique for spinal fusion for discogenic pain. Non-surgical treatment of patients with chronic low back pain and degenerative discs may include physical therapy, exercise, cognitive-behavioral therapy, medication, injection, transcutaneous electrical nerve stimulation, alternative treatments such as acupuncture and spinal manipulation, and intradiscal electrothermal therapy (IDET). The treatments patients receive are recorded both at baseline and at follow-up interviews. We conduct the baseline assessment in person and obtain the follow-up measures in telephone interviews at six, nine, twelve, and twenty-four months after enrollment. Also, after three months, we re-administer the baseline measures. For patients who have surgery late in the initial 6-month interval, this provides a more recent pre-surgical baseline for analysis. We evaluate several dimensions of outcome, including pain, functional status, general well-being, and role disability. The measures have been previously validated and have been recommended for standardized outcome assessment in back pain trials \[[@B17],[@B18]\]. Furthermore, most of these measures have been recommended by the American Academy of Orthopaedic Surgeons and the North American Spine Society \[[@B19]\]. These instruments were chosen with an eye towards brevity, established reliability and validity, ease of administration by telephone, and inclusion of a range of relevant outcomes. The Modified Roland Scale is the primary outcome measure. The original Roland-Morris Disability Questionnaire \[[@B20]\] was derived from the 136-item Sickness Impact Profile (SIP) \[[@B21]\]. It was subsequently modified to select items from the SIP that would more likely detect change in patient status and to include attribution of limitations to leg pain as well as back pain \[[@B22]\]. This 23-item self-report measure of physical disability due to back and leg pain has established validity, reliability, and responsiveness to change \[[@B22]\]. Higher scores indicate a greater level of functional disability. The SF-36 Health Survey, Version 2 \[[@B23]\] consists of eight scales (general health, physical functioning, role limitations due to physical problems, role limitations due to emotional problems, bodily pain, social function, mental health, and vitality) scored on a scale of 0 (worst health) to 100 (ideal health). The earlier version has been used to assess general health status among patients with variety of health conditions, including back pain \[[@B17],[@B24],[@B25]\]. Study participants also complete the Symptom Check List-90 (SCL-90) 12-item somatization and 13-item depression scales \[[@B26]\]. Subjects indicate whether they have each symptom using a 5-point scale ranging from \"not at all\" to \"extremely\". Higher scores indicate greater somatization or depression symptom severity. The 13-item Pain Catastrophizing Scale \[[@B27]\] is used as both a predictor and a secondary outcome. The measure includes subscales that reflect three components of pain-related catastrophizing: helplessness, rumination, and magnification (e.g., fear that the pain will become worse). Previous research has consistently found substantial associations between pain-related catastrophizing and pain-related disability \[[@B28],[@B29]\]. We are interested in learning whether pain-related catastrophizing is a risk factor for poor outcomes in patient with low back pain. The baseline questionnaire also assesses demographics, pain (numerical rating scale, bothersomeness, other painful body sites), medical co-morbidity, back pain history, work status, and litigation/compensation issues. Problematic alcohol use is assessed by the first three items of the Alcohol Use Disorders Identification Test (AUDIT-C) \[[@B30]\]. The AUDIT-C has been shown to be a valid screening test for heavy drinking and/or active alcohol abuse or dependence \[[@B30]\]. Although imaging is required for enrollment, not all imaging studies are accessible to the researchers (e.g., those done at a distant site). However, a majority of study participants have magnetic resonance imaging (MRI) scans that are available to the investigators. These images are interpreted by a neuroradiologist without knowledge of the patient\'s clinical history. The key imaging findings were defined as in the Longitudinal Assessment of Imaging and Disability for the Back and include T2 signal, spondylolisthesis, endplate Modic changes, disc height loss, disc morphology, annular tears, facet changes, and central and foraminal stenosis \[[@B31]\]. All data are collected on paper forms, and then entered into a web-based data system requiring double entry to reduce transcription errors. Data checks to identify out-of-range answers, inconsistent responses, missing data, and response rates are performed on a monthly basis. Results ======= In the primary analysis, the modified Roland Disability scale at 24 months will be compared between the surgical and non-surgical treatment arms using regression techniques to adjust for important baseline characteristics. Secondary analysis will include a comparison of pain, SF-36 scales, and psychological measures for the two treatment arms. To characterize time trends in the primary and secondary outcomes, we will use linear mixed models (or Generalized Estimating Equations in the case of categorical variables) to analyze the repeated measures obtained at all follow-up interviews. Finally, we will examine potential predictors of outcome, including baseline disability, psychological factors, and image findings. The goal will be to identify subgroups of patients who respond well to surgery but not to non-surgical therapy, or to non-surgical treatment but not surgery. The Maine cohort study \[[@B25]\] provided crude estimates of possible differences between surgical and non-surgical patients on the Roland Scale, SF-36, and pain scales. Table [2](#T2){ref-type="table"} shows the estimated sample size needed to detect differences of various magnitudes between the surgical and non-surgical patients. Based on these estimates, our goal is to enroll 150 patients each in the spinal fusion arm and in the non-surgical treatment arm. Enrollment will proceed over two years. Based on estimates of eligible patient numbers, and our conservative estimate that 60% of eligible patients will enroll in the study, this will enable us to obtain the target enrollment. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Estimated sample sizes per group, based on different outcome measures. For function and symptoms, standard deviations are from the Maine Lumbar Spine study, with 3-month follow-up (roughly equal proportions treated surgically and non-surgically). Return to work proportion is based on a study by the Washington State Department of Labor and Industries. ::: Category Measure Hypothesized difference between groups N/group, 2-sided alpha = 0.05, power = 0.80 ------------------- ----------------------------------------------------------------------- ---------------------------------------- --------------------------------------------- Functional Status Difference in change in Roland score (0--23 scale, SD = 7.37) 2.5 points 136 3.0 points 95 Symptoms Improvement in pain (1 = gone, 5 = same, 7 = much worse) (SD = 1.841) 0.75 point 94 1.0 point 53 Return to Work Proportion currently employed, if lower proportion is 0.40 0.15 173 0.20 97 ::: Discussion ========== In designing the study, we considered a randomized controlled trial (RCT), but chose an observational design for several reasons. First, the surgical treatments are already approved and in wide use, and physicians and patients often have strong preferences for either surgical or non-surgical care. Thus, genuine equipoise is rare among U.S. surgeons and patients. In fact, we attempted to enroll subjects in a pilot randomized trial and had no success. Second, randomized trials comparing surgery with non-surgical treatment have several features that are distinctly different from drug trials and result in serious limitations. If drugs cause side effects, they can be stopped and most ill effects will resolve. Surgery, however, has many irreversible features. Efforts to blind the treatment allocation require a sham surgical procedure, raising patient anxieties and ethical concerns. Without a sham surgical control, blinding is impossible. Unlike pills, which are essentially identical, no two surgical procedures are exactly the same. These features constrain the validity of surgical randomized trials in comparison to drug trials. Finally, we were interested in treatment outcomes as they occur in routine practice, rather than within the narrow constraints typically imposed in randomized trials. Thus, we chose to study treatment effectiveness in routine care, rather than efficacy under ideal circumstances. Though an RCT would provide the most valid data on efficacy, the prospective cohort design seemed substantially superior to uncontrolled case series, which remain the predominant study design in the surgical literature. Furthermore, when carefully designed, the results of cohort studies sometimes approximate the results of randomized trials \[[@B32],[@B33]\]. The Maine Lumbar Spine Study \[[@B24],[@B25]\], for example, was a prospective cohort study that yielded results concordant with those from randomized trials of discectomy \[[@B34]\]. Conclusion ========== This study will contribute important new information on a highly controversial area of back pain treatment. Though it is not a randomized trial, we believe a rigorously designed and analyzed cohort study will improve our knowledge of both treatment effectiveness and safety in routine practice. The primary aim of this prospective cohort study is to better define the outcomes of lumbar fusion for discogenic back pain as it is practiced in the U.S. The results should help improve the selection criteria for surgical treatment, better define the prognosis after therapy, and improve our ability to match patients with optimal treatment approaches. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= RAD and SKM conceived of the study. RAD, BIM, SKM drafted the manuscript. PJH performed statistical analysis. All authors participated in the design of the study and coordination of the study. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2474/6/24/prepub> Acknowledgements ================ This study is funded by the National Institute of Arthritis, Musculoskeletal, and Skin Disease (NIAMS) (P60 AR48093) and is conducted by the University of Washington\'s Multidisciplinary Clinical Research Center located in Seattle, Washington, USA.	PubMed Central	2024-06-05T03:55:59.478232	2005-5-24	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180446/", "journal": "BMC Musculoskelet Disord. 2005 May 24; 6:24", "authors": [ { "first": "Richard A", "last": "Deyo" }, { "first": "Sohail K", "last": "Mirza" }, { "first": "Patrick J", "last": "Heagerty" }, { "first": "Judith A", "last": "Turner" }, { "first": "Brook I", "last": "Martin" } ] }
PMC1180447	Background ========== Posture reflects the relationship between spinal segments, and the influence of the environment on spinal segments \[[@B1]\]. Correct upright posture is considered to be an important indicator of musculoskeletal health. Costs associated with musculoskeletal impairments in health and loss of work, have contributed to a growing interest in optimizing posture, particularly in relation to sitting positions associated with the use of visual display units \[[@B2]\] and standing posture in children in relation to backpack use \[[@B3]\]. There is no standard approach to measuring posture. Photographic observations of ideal posture have been ranked visually or simple equipment such as a tape measure, penciled landmarks and a plumbline, have been used \[[@B4],[@B5]\]. The linking of body landmarks has given angular measurements, allowing a more quantitative assessment of posture \[[@B6]\]. Watson and Mac Donncha \[[@B6]\] reported 85% reliability when ten aspects of adolescent photographic posture were qualitatively categorized and rated. Straker and Mekhora \[[@B2]\] photographically evaluated sitting postures as a series of angles in adults working at visual screens. This method was reported in Straker et al \[[@B7]\] to have previously shown reliability (r^2^\>0.8) in adults. Grimmer et al \[[@B8]\] adapted the measurements used by Straker and Mekhora \[[@B2]\] to assess standing posture in adolescent high school students aged 12--18 years. The reliability of this photographic method of posture assessment however has not been tested in children. Indeed the measurement of posture in children has received scant attention in the literature and little is known about the variability of children\'s standing posture. There are many factors which may influence the reliability of photographic posture assessment in children. These include maturation and developmental factors such as age, gender, height and the development of postural control and co-ordination. The presence of pain and the testing environment may also have an effect. Factors associated with the measurement process including palpation of bony landmarks for marker placement and reproducibility of the digitization process, may also contribute to the reliability. Growth spurts occurring in 9--12 year olds may cause widespread alterations in body shape and dimensions and have an effect on muscle tightness and flexibility, all of which may influence posture in children \[[@B9]-[@B11]\]. Children have relatively larger heads and also a higher center of mass at about T12, compared to L5-S1 in adults. The combination of being shorter and having a higher center of mass may result in increased sway in children and difficulty in maintaining static balance \[[@B1]\]. The stage of development of postural responses may influence the ability of the child to maintain a relaxed standing posture. Postural control development occurs sequentially in a cephalo-caudad direction with head control first, followed by the trunk and then postural stability in standing \[[@B1]\]. The motor and sensory systems involved in postural stability go through a transition period at 4--6 years and reach adult maturity by 7--10 years \[[@B12]-[@B14]\]. Spinal pain is recognized as a significant influence on normal posture in adults, with an antalgic posture commonly taken up to avoid a painful position \[[@B15]-[@B17]\]. It has been proposed that pain may indirectly affect posture, through the alteration of somatosensory signals to the central nervous system \[[@B18]\]. There is no reason to suspect that this effect would be different for children. The aim of this study was to assess the variability of children\'s posture using repeated measures within the same hour. Methods ======= Study design ------------ A repeated measures observational study was conducted in December 2000. This study formed part of a larger epidemiological study conducted by the Centre for Allied Health Evidence, University of South Australia, where primary school students participated in a range of tests, to identify factors which may contribute to spinal health. In addition to posture, other measures included anthropometric, school bag weight, muscle endurance, coordination and a questionnaire about spinal symptoms and activity levels. There were six separate testing stations at which children were measured, with time taken per station for testing ranging from three minutes (anthropometric) to 15 minutes (muscle endurance). Sample selection ---------------- One class from each year level (Reception, aged 5 years to Year 7, aged 12 years) of a large suburban Adelaide primary school was chosen to be involved in the larger study. Ethical approval was gained from the Ethics Committee of the University of South Australia and the Department of Education and Children\'s Services, and written parental consent was obtained prior to the commencement of the study. All consenting children were included. Children arrived at the testing area in class level groups of varying sizes. As only a short time was allotted by the school for testing, the measurements at each test station were determined by the order that children approached the six test stations, the speed of testing at each station and the availability of places at the next test station. Before the arrival of the subsequent class group, children who had completed all the testing stations once, were re-tested, forming a sample of convenience for the posture reliability study. Equipment, preparation and testing procedure -------------------------------------------- Subjects were tested in the school gymnasium and efforts were made to control for temperature, noise and distractions. In an attempt to minimise data collection error, research assistants at all stations received comprehensive training in the use of study test protocols prior to commencement of the study. Strict protocols were used to ensure the correct placement of anatomical markers, positioning of the subject and camera placement \[[@B8]\]. To capture postural information on body segments, adhesive markers were placed over right-sided lateral landmarks including the lateral canthus of the eye, the tragus, the greater trochanter and the lateral malleolus. A small reflective ball was placed over the spinous process of C7, to ensure that this landmark would be detected on the scanned photographs (Figure [1](#F1){ref-type="fig"}). The markers that were placed over the shoulder, pelvis and knee were not included in the angle calculations used for this study, however they were used for other research purposes \[[@B8]\]. As the study aimed to assess variability in posture on repeated occasions of testing, the markers were left in place between tests. Removing and then replacing the markers would have introduced an additional element of reliability of the examiner in marker placement, which was not the aim of this investigation. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Adhesive marker placement and postural angles. a trunk angle; b neck angle; c gaze angle; d head on neck angle; e lower limb angle ::: ![](1471-2474-6-35-1) ::: Portrait -- format photographs were obtained using a Canon SLR camera (EOS 500) that was attached to a tripod and placed at a distance of 3.1 m and in a direct line from the subject. Spirit level adjustments placed on the top of the camera and the front of the lens confirmed horizontal and vertical alignments of the camera respectively. The tripod was secured in the correct position on the floor using masking tape. Floor markers were used to standardise subject placement and to ensure that the subject\'s right side was aligned perpendicular to the camera. A set-square and spirit level, attached to a perspex calibration board, were connected to a second tripod. The calibration board was placed in the field of view and aligned with the subject to allow referencing of horizontal and vertical axes from the photographs. The calibration board also displayed each subject\'s identification number. For positioning, the child was instructed to stand comfortably in a normal standing position \'as if waiting at the canteen\', and to look straight ahead at a pre-determined point on the wall. To allow for visualization of the greater trochanter marker, the subject was further instructed to move the elbows forward but still touching the body and with minimal shoulder movement. The position was checked by a researcher and by the photographer prior to the photograph being taken. After the first photograph, the subject was asked to move away from the testing station, walk around a small area and then return to the photographic position where the second photograph was taken. The anatomical markers were not moved between photographs, and their position was rechecked prior to the second photograph to ensure that they were securely in place. The letter \'R\' was recorded after the subject identification number to indicate the \'repeat\' photographs. This method of posture measurement is reported elsewhere by Grimmer et al \[[@B19]\]. Measurement of factors which potentially influenced posture was restricted to age, gender, height, weight, recent pain and gross motor control ability in relation to co-ordination, strength, balance and flexibility (using the Brace test) \[[@B20]\]. The Brace tests are validated gross motor skill tests that vary in complexity from very simple (eg walk heel to toe for 10 steps along a line) to very difficult (eg with the free foot, jump over the hand holding the opposite foot). The subject received a score of one (success) or zero (failure) for each test to a maximum total score of 20. Self-reported outcomes of age, gender and recent pain were obtained from a parent or carer, in conjunction with the child, and recorded on the study questionnaire. If pain was present on the day of testing or had occurred during the previous week, this was recorded by ticking boxes that corresponded to the correct body part (eg head, neck, upper back, elbow or knee). The pain variable used in the statistical analysis for this paper reflected any pain reported in the neck, upper or lower back. The child\'s weight (in kilograms) and height (in centimetres) was measured in accordance with standard anthropometric protocols used in previous research conducted by our group \[[@B19]\]. Digitising and synthesising posture data into angles ---------------------------------------------------- After testing was completed and films developed, a negative scanner (Nikon LS-2000) was used to convert developed negatives into electronic format files. Image analysis software (ImageTool UTHSCA Version 2.0, University of Texas Health Science Center, San Antonio, TX) was employed to digitise the x and y plane coordinates obtained from each anatomical landmark from the photographs. A strict pre-existing protocol for scanning and digitization was followed, which reduced repeated measure estimates of landmark coordinates from any photograph to less than 10 pixels difference (non-significant (p \> 0.05)). One researcher undertook all scanning and digitizing to eliminate inter-examiner error. The x, y coordinate values were imported into MS Excel spreadsheets for calculation of the five body angles used for data analysis (Figure [1](#F1){ref-type="fig"}). ### Trunk Angle The angle between the trunk (as indicated by a line drawn through anatomical markers at C7 and the greater trochanter) and a vertical line through the greater trochanter. ### Head and Neck on Trunk Angle (abbreviated to \"Neck Angle\" for this article) The angle between the head and neck segments, as indicated by a line drawn through anatomical markers at C7 and the tragus of the ear, and the trunk, as indicated by a line drawn through anatomical markers at C7 and the greater trochanter. ### Gaze Angle The angle formed by a line drawn through anatomical markers at the canthus of the eye and tragus of the ear and a horizontal line through the tragus of the ear. ### Head on Neck Angle The angle between the neck as indicated by a line drawn through anatomical markers at C7 and the tragus of the ear, and the head as indicated by a line drawn through the canthus of the eye and the tragus of the ear. ### Lower Limb Angle The angle between the lower limb as indicated by a line drawn through anatomical markers placed at the greater trochanter and the ankle, and the vertical, as indicated by a vertical line drawn through the greater trochanter. Trunk, neck, gaze and head on neck angles were adapted from Straker et al \[[@B7]\] and Straker and Mekhora \[[@B2]\], who used a mid-iliac marker in place of the greater trochanter and a marker over the external auditory meatus in place of the tragus marker. Straker et al \[[@B7]\] and Straker and Mekhora \[[@B2]\] developed angles to assess sitting posture particularly in relation to use of visual display units. These angles were modified for the current study to minimize error of marker placement by using accessible and specific anatomical landmarks. A fifth angle, the lower limb angle, was established for standing posture assessment, for the purposes of this study. Testing and statistical analysis -------------------------------- The testing approach is summarized in Figure [2](#F2){ref-type="fig"}. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### The testing approach ::: ![](1471-2474-6-35-2) ::: Four age groups were constructed, these being less than or equal to 6 years, 7--9 years, 10--11 years and greater than 11 years. These divisions were determined on age quartiles in the data, which provided appropriate groupings to reflect structural and functional changes in children related to their growth and maturation \[[@B21]\]. The mean values and standard deviations (SD) for each angle in each age group for the first set of measurements (Test 1) were determined. An understanding of the variability of the posture angles relative to the mean for the first set of measurements was determined for each age group, by dividing the standard deviation by the mean. To understand what may influence posture, the association between the first set of posture measures and potential predictors of children\'s postures (height, weight, motor control, spinal pain) were individually assessed using univariate linear regression models, or in the case of spinal pain, an ANOVA model. A pvalue of \< 0.01 was chosen to reduce the possibility of missing an important effect. The strength of the linear association was determined using criteria by Dawson and Trapp \[[@B22]\], where r^2^\<0.25 indicates a poor relationship, 0.25-0.5 indicates a fair relationship, 0.5--0.75 indicates a moderate to good relationship, r^2^\>0.75 indicates a strong relationship. The 95% confidence intervals (CI) were reported around the mean differences in angles within age groups, to provide information on usual variability of young people\'s posture for future studies. Confidence intervals that do not include zero indicate a statistically significant difference at the 5% level. Intraclass correlation coefficients (ICC ~1,3~) were used to identify the reliability between test and retest angle measure differences based on the within and between subject variances, obtained from output from the ANOVA models \[[@B23]\]. Multiple ANOVA models were used to examine the effects of potential predictors (age, gender, height, weight, motor control repeated testing and pain) on posture outcome measures. Scheffe\'s post hoc multiple comparisons test was used \[[@B23]\]. Post hoc power calculations were performed, using the effect sizes derived from this study to determine if the study sample had sufficient power to enable a true difference in estimates to be detected if they truly existed \[[@B23],[@B24]\]. Results ======= Descriptive data ---------------- There were 38 primary school students in the sample, whose demographic details are outlined in Table [1](#T1){ref-type="table"}. Mean (SD) for height, weight and motor control (Brace Test scores) are provided per age group in Table [2](#T2){ref-type="table"}. The height and weight of the sample did not differ remarkably from normative data for same aged children in the western world \[[@B25]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Gender of children by age group. ::: \< = 6 yrs (n = 9) 7--9 yrs (n = 9) 10--11 yrs (n = 12) \>11 yrs (n = 8) ------- -------------------- ------------------ --------------------- ------------------ Girls 3 3 7 2 Boys 6 6 5 6 ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Mean (SD) posture predictors by age group. ::: \< = 6 years n = 9 7--9 years n = 9 10--11 years n = 12 \>11 years n = 8 ----------------------------------------- -------------------- ------------------ --------------------- ------------------ Height (cms) 112.1 (3.7) 130.0 (7.8) 138.8 (5.8) 146.9 (5.4) Weight (kgs) 33.7 (9.2) 29.7 (7.2) 32.4 (3.8) 42.5 (6.2) Brace tests (number out of possible 20) 13 (4.0) 10 (3.0) 14 (3.0) 13 (3.0) ::: On the day of testing four children (11%) reported spinal pain (two each in the youngest and oldest age groups) and 14 children (37%) reported experiencing spinal pain (2 each in the youngest and oldest age groups, 4 in the 7--9 year olds, and 6 in the 10--11 year olds) in the week prior to testing. There were no gender differences within any age group for any of these measures. Variability in angle measurements --------------------------------- The mean (SD) angles (in degrees) for each of the five angles measured on Test 1 in each age group are presented in Table [3](#T3){ref-type="table"}. Negative angles were found in 97% of the first set of trunk angles, 4% of the gaze angles and 20% of the lower limb angles. The variability of the posture angles relative to the mean in Test 1, is illustrated in Figure [3](#F3){ref-type="fig"}. Similar relative variability was found for the youngest and oldest age groups for all angles except the lower limb angle, which seemed the most variable over all the age groups. The age groups with the greatest relative variability were the 7--9 and 10--11 years groups, particularly for the trunk, gaze and lower limb angles. The least variable angles over all the age groups were the neck angle and the head on neck angle. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Mean (SD) test 1 for each angle (in degrees) for each age group. ::: \<= 6 years 7--9 yrs 10--11 yrs \>11 years -------------------- ------------- ------------ ------------ ------------ Trunk angle -8.8 (2.5) -5.0 (2.6) -5.0 (4.3) -5.6 (2.6) Neck angle 61.4 (5.3) 58.5 (3.4) 55.7 (8.7) 51.6 (4.9) Gaze angle 12.3 (5.4) 11.1 (7.9) 10.3 (7.6) 13.1 (9.0) Head on neck angle 25.1 (7.5) 25.5 (7.3) 29.1 (7.4) 30.9 (9.6) Lower limb angle 3.3 (2.6) 0.05 (1.2) 2.1 (1.9) 2.9 (2.0) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Relative variability of the first set of measurement of the angles for each age group ::: ![](1471-2474-6-35-3) ::: Examining predictors of posture angles -------------------------------------- In this sample, age was strongly associated with height (r^2^= 0.84) and moderately associated with weight and motor control (r^2^= 0.67, 0.56 respectively). There was an unconvincing association between any of the postural angles, height, weight and motor control (see Table [4](#T4){ref-type="table"}), with the strongest associations observed for the trunk angle with height, weight and motor control. Spinal pain was not associated with age, or any of the postural angles (p \> 0.05). ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### The effect of height, weight and motor control (Brace Tests) on the five body angles. ::: Angle Height Motor control Weight -------------------- -------- --------------- -------- Trunk angle 0.22\* 0.18\* 0.22\* Neck angle 0.13 0.16 0.11 Gaze angle 0.02 0.003 0.004 Head on neck angle 0.03 0.01 0.004 Lower limb angle 0.06 0.03 0.003 (\* indicates significant findings p \< 0.01) ::: Differences between tests ------------------------- The mean differences (95%CI) (in degrees) between tests 1 and 2 for each angle in each age group are presented in Table [5](#T5){ref-type="table"}. The ICC~1,3~values for repeated measures of the five angles, ranged from 0.93 to 0.99, suggesting no significant effect of testing on any angle. ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Mean differences (95%CI) between test 1 and 2 for each angle (in degrees) by age group. ::: \< = 6 years n = 9 7--9 years n = 9 10--11 years n = 12 \>11 years n = 8 ------------------------- -------------------- -------------------- --------------------- --------------------- Trunk angle diff 0.2 (-2.6 to 2.9) 1.7 (-0.8 to 4.1) 2.04 (-1.0 to 5.0) 1.3 (-1.4 to 4.0) Neck angle diff -0.5 (-5.7 to 4.8) 1.5 (-2.7 to 5.6) -1.9 (-8.4 to 4.7) -1.4 (-7.6 to 4.8) Gaze angle diff -2.2 (-7.2 to 2.8) -2.5 (-9.5 to 4.5) 2.7 (-3.8 to 9.1) 2.5 (-5.3 to 10.3) Head on neck angle diff 2.5 (-4.6 to 9.5) -0.1 (-7.6 to 7.4) -2.8 (-9.2 to 3.5) -2.4 (-10.7 to 6.0) Lower limb angle diff -1.1 (-4.0 to 1.8) -0.7 (-2.2 to 0.8) -0.6 (-2.4 to 1.1) -0.6 (-2.6 to 1.3) ::: Examining predictors of differences between posture angles ---------------------------------------------------------- Multivariate ANOVA models found no effect of gender on the mean test difference for any angle. However, there was a significant age effect (p \< 0.01) on the difference between repeated measures of the trunk angle, the neck angle, the head on neck angle and the lower limb angle i.e. for all but the Gaze Angle. Height, weight, motor control and pain were found to have no influence on the difference between test -retest measurements (p \> 0.05). Scheffe\'s post-hoc test \[[@B23]\] however, found that only three percent of the observations for repeated test differences between any angle were significant. Post-hoc power calculation using the effect size directly derived from the results of the study \[[@B24]\] indicated that the sample was sufficiently powered at 0.80 and p \< 0.05 for the trunk angle and the lower limb angle, to detect the effect of repeated testing. Greater numbers were required for 80% power (p \< 0.05) to test the effect of repeated measures of the other three angles. Using a specific sample size electronic calculator \[[@B24]\], and based on the effect-size data from this study, 32% power for repeated testing of the neck angle was found (for 80% power, 146 subjects required), 25% power for repeated testing of the gaze angle (for 80% power, 360 subjects required) and 28% power for repeated testing of the head on neck angle (for 80% power, 312 subjects required). Discussion ========== This study provides rare information on the repeatability of relaxed standing posture in children aged 5--13 years, with posture expressed as five angles derived from x,y coordinates of anatomical points, calculated electronically from photographs. As the height and weight of our sample did not differ remarkably from normative data for same aged children in the western world \[[@B25]\] and as our sampling approach should not have incurred bias beyond chance, we believe that this paper presents useful externally generalisable information not only for clinical purposes, but also to inform further research on larger numbers of children, particularly to test reliability of neck angle, gaze angle and head on neck angle. Subject measurement ------------------- The students were tested in a limited time period under minimum stressed conditions (i.e. there was little opportunity for fatigue). They were moved to the testing stations in a variable order, as a position became available for testing, thus it is possible that the order of testing may have influenced the results. As testing was completed within a period of 40--60 minutes, fatigue or boredom bias is considered unlikely. The most likely reason for differences in measurements would be natural variability in subject responses to the test-retest situation \[[@B1]\]. As so little is known about young people\'s postural variability, a study with larger numbers is required to clarify this issue. The reliability of the posture measurements ------------------------------------------- This study suggested that children\'s standing posture (quantified by five whole body or segmental angles) did not change significantly on repeated testing. Further testing with larger subject numbers, is required however, to be certain of these findings for the neck, gaze and head on neck angles. The five posture angles are considered useful and easily attained postural outcome measures that may be appropriate for application to clinical studies. The trunk angle is a measure of the trunk position relative to the line of gravity. The negative values shown for the trunk angle in Table [3](#T3){ref-type="table"} indicate a relative backward lean of the trunk and this angle is much greater for the children 6 years and under. The digitized marker points on C7 and the greater trochanter are a long distance apart, thus the protocols which guided accurate marker placement and subject positioning were essential to minimize error in measurement. The mean neck angle across the age groups ranged between 61 and 51 degrees. This angle is a measure of the head and neck position in relation to the trunk and gives a measure of the forward head position, which is a useful clinical indicator of mid/lower cervical spine dysfunction. A lower measurement may indicate a more corrected head posture. The mean gaze angle ranged between 10.3 and 13.1 degrees in these young children during relaxed standing, indicating small changes in the line of sight. The head on neck angle ranged from 25 to 31 degrees and indicates change in the position of the head relative to the neck. A decrease in this angle is considered to result in a \'poking chin\' posture and may indicate stresses on the upper cervical spine. The lower limb angle gives an indication of the hip position (and therefore centre of gravity) over the base of support, as measured at the ankle. An increase in this angle may occur with increasing postural control related to positioning of the centre of gravity over the midfoot rather than the heel. Similar to the trunk angle, this lower limb angle uses landmarks a long distance apart, and thus our strict measurement protocols assisted in measurement accuracy. Effect of age on repeated posture measurements ---------------------------------------------- The mean difference between repeated testing of any angle (as shown in Table [5](#T5){ref-type="table"}) was less than three degrees. These figures need to be considered in relation to the 95% CI around the mean difference in each set of repeated measures. This data provides the first known information on \'usual\' postural variability expected in children of these ages. Table [5](#T5){ref-type="table"} illustrates the impact of age on postural performance during repeated testing. The lack of test -- retest effect can be seen in all instances (all angles, all age groups), where the 95% confidence intervals span zero. As suggested however, by the low power for detecting differences in three of the repeated measures (neck, gaze and head on neck angles), further testing with larger numbers is required. Multivariate ANOVA models confirmed that age had a significant influence at p \< 0.01, on repeated testing for four of the angles, with only the gaze angle not being influenced by age. This finding adds support to the importance of stable gaze for orientation in children \[[@B1]\]. Effect of age and other predictors on posture angles ---------------------------------------------------- The effect of age on measures of standing posture is clearly demonstrated in Table [3](#T3){ref-type="table"}, which shows the step-wise decrease in the neck angle and the stepwise increase in head-on-neck angle as children get older. The changes in these two angles suggest taller, more corrected upright posture and may reflect growing postural maturity with age \[[@B14]\]. Figure [3](#F3){ref-type="fig"} also demonstrates the low relative variability for the neck angle and the head-on-neck angles in each age group and the similarity of the variability across the age groups for these two angles. In comparison, there was high variability in general and quite different variability for each age group for the trunk angle and the lower limb angle. Excessive postural sway recognized in younger children, and measured in this study by movement of the greater trochanter over the base of support, may have influenced the variability demonstrated in these latter two angles \[[@B1]\]. Long levers were also associated with the trunk and lower limb angles. However, the comparative low variability of the neck angle, which also used a marker over the greater trochanter, may reflect a stable relationship between this hip point and markers in the head and neck region, in children of these ages. Under standardized positioning instructions, the normal high variability associated with movement of the eyes (as measured by the canthus, for determining the gaze angle), can be seen in the Figure [3](#F3){ref-type="fig"}. It was considered that the measure of age in this sample of subjects may have reflected anthropometric growth (measured as height and weight) as well as developments in motor control influencing the body\'s ability to balance against the forces of gravity \[[@B26]\]. Age was found to be strongly associated with height (r^2^= 0.84) and moderately associated with weight and motor control (r^2^= 0.67, 0.56 respectively). These findings concur with current literature on paediatric development \[[@B21]\]. Neither pain reported on the day of testing, nor pain in the previous week, was significantly associated with age, or with differences between repeated measures of the angles. However, as univariate predictors, motor control, height and weight were unconvincingly associated with four of the five posture angles (Table [4](#T4){ref-type="table"}). The strongest findings were found for the trunk angle, where approximately 20% of the univariate associations were explained by motor control, height or weight. This association may be explained by the nature of this angle, which involves multi-segments of the body, using landmarks on C7 and the greater trochanter, and thus requires motor control between the head and trunk. Conclusion ========== On testing of repeatability of five postural angles in children 5--13 years, there was no significant effect of repeated testing. Increasing age influenced four of the five postural angles, with the only angle not affected by age being the gaze angle. Height, weight and motor control explained approximately 20% of the variability in the trunk angle, but explained very little of the variability in the other four angles. While the subject numbers in this study are small, the findings provide useful information on which further studies in posture and its development in pre-adolescent children can be based. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= KG conceived of the study, participated in the design of the study and conducted the statistical analysis. MMc participated in the design of the study, collected the data and drafted the manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2474/6/35/prepub> Acknowledgements ================ The authors wish to acknowledge the contributions made to this manuscript by Steve Milanese, Andrea Bialocerkowski, Ian Fulton and Trisha Maddison.	PubMed Central	2024-06-05T03:55:59.480107	2005-6-29	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180447/", "journal": "BMC Musculoskelet Disord. 2005 Jun 29; 6:35", "authors": [ { "first": "Maureen P", "last": "McEvoy" }, { "first": "Karen", "last": "Grimmer" } ] }
PMC1180448	Background ========== Osteoporosis is the most common metabolic disease of bone which is known by deficit in bone mineral density and skeletal micro destruction that increases risk of bone fracture \[[@B1]-[@B3]\]. The importance of BMD is, in diagnosis of osteoporosis and prevention of bone fractures and its consequent disability \[[@B1],[@B2]\]. BMD depends on age, disease, genetic, mechanical factors, nutrition, and the body hormones effects\[[@B4]\] Studies showed that the prevalence of osteoporosis the age of over 70, it increases to 87% \[[@B5],[@B6]\]. In Thailand\'s over 70 yr women, prevalence of disease was 50% \[[@B6]\]. In the UK, 1/3 of women and 1/12 of men affected from osteoporosis \[[@B7]\]. The investigations showed that in the next 50 years, considering the population growth in the old people in Asia, South America and Africa, it is expected almost 75% of these fractures occur in progressing countries \[[@B1]\]. A study, that recently performed in Iran showed, osteoprotic fracture data was important public health problem in Iran \[[@B8]\]. For reliable interpretation of individual BMD data, however, they need to be expressed in relation to established normative data. Comparisons can be made either in terms of the age-matched standard deviation score, use of the Tscore, which indicates deviation from the mean BMD of a young normal population \[[@B9]\]. For this reason, comparison of Tscores yields the best available information on the extent of osteoporotic bone loss and the associated fracture risk. In clinical practice, individual BMD values are compared with a reference value. For diagnostic purposes, a panel convened by the WHO proposed to define osteoporosis on the basis of the T-score, According to this categories, a Tscore between -1 and -2.5 is indicative of osteopenia, while a Tscore -\'-2· 5 reflects osteoporosis \[[@B9]-[@B12]\]. Despite its limitations \[[@B13]-[@B15]\]; this definition is currently applied worldwide. However, the normal values provided by manufacturers may not be fully representative of specific local populations. In fact, BMD is influenced by several variables, including genetic and environmental factors \[[@B16],[@B17]\]. Thus, reference ranges may vary in different populations \[[@B18]-[@B23]\], and \[[@B24]\]. Early diagnosis of osteoporosis by assessment of bone density can prevent its complications, especially fractures. Bone density relates to many items like race, genetic, sex, environmental factors and nutrition. In order to define osteoporosis and osteopenia, knowledge of the reference data of bone mineral density (BMD) is important. As there is not any reference data on BMD and osteopenia and osteoporosis in Iran, we decided to perform a first national comprehensive study which it\'s aims of this study were to determine normal values of bone mineral density at lumbar vertebras and neck of femur and determine prevalence of osteopenia and osteoporosis for Iranian normal population. Methods ======= This was a cross sectional study. In duration of 6 months, 553 subjects were selected among the men and women of 20\_ 69 yr of Tehran. The individuals selected based on randomized clustered sampling from 50 blocks in Tehran. For selection of clusters, whole Tehrani\'s population on base of distribution of them, divided to many clusters. From all clusters 50 of them were randomly selected and after that, in each block, home\'s numbers whose numbers were twin were selected on based of exclusion and inclusion criteria, in each home 1 person were selected, until the individuals in each group reach to 24 persons. The exclusion criteria were selected from diseases and drugs that may have effects on metabolism of bone and Vitamin D. Smoking, alcohol, pregnancy, breast-feeding during the study, Professional sport, General conditions and Immobility also were on exclusion criteria. Healthy individual were selected if they did not have any problem that affect on bone or Vitamin D metabolism based on exclusion criteria. Previous and current diseases, drugs, and habits were determined by personal interview and were evaluated by nurses through the interview. 750 individuals were invited for this study, 533 individuals came for assessment. After 2 times recall, there is not any significant difference between the mean age and sex distribution of individuals who came and who did not come. 217 subjects refused to participate in this study. The study protocol was approved by research ethic committee of Endocrinology and Metabolism Research Centre (EMRC) and the data gathered by cluster random sampling. The subjects with osteoporosis were referred to the EMRC osteoporosis clinic in the Shariati Hospital for treatment. Having received the letters of consent, the related questionners were completed and clinical examinations such as height and weight were carried out. BMD was measured by DXA using Lunar DPX-MD device (Lunar Corporation, Madison, Wisconsin, 53713. USA). The DXA device was calibrated daily and weekly by using appropriated phantoms methods. To assess BMD, second to fourth lumbar spine and from the femur bone (neck, trochanter and the whole femur), bone density was calculated based on gr/cm^2^. SPSS (ver 11.5) was used for data analysis. To compare the mean, the student T test was used and for comparing frequency of variable between groups Chi-square was used. Results ======= 553 subjects (34% men, 66% women) between 20 to 69 yr (mean ± SD, 44.07 ± 12.68) participated in the study. Basic characteristics of the subjects showed that in Table [1](#T1){ref-type="table"}. There was no significant difference in the spinal BMD of women between the age groups 20--29 and 30--39 (Table [2](#T2){ref-type="table"}). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Basic characteristic of the subjects in each age and sex groups ::: Age groups Numbers Weight(Kg) Height(cm) BMI(Kg/m^2^) ---------------- ------------- ---------------- ---------------- ------------------ -------------- 20--29 Women 44 60.1 ± 11.4 160 ± 4.38 23.45 ± 4.1 Men 27 71.27 ± 12.7 173.58 ± 6.16 23.04 ± 3.95 30--39 Women 104 67.43 ± 11.34 156.46 ± 8.29 27.72 ± 5.7 Men 38 77.68 ± 15.99 171.51 ± 7.6 26.29 ± 4.45 40--49 Women 98 70.96 ± 17.58 157.2 ± 6.31 28.61 ± 6.01 Men 48 76.37 ± 10.85 168.11 ± 6.11 27.02 ± 3.55 50--59 Women 82 68.27 ± 11.68 154.04 ± 6.19 28.73 ± 4.74 Men 42 73.66 ± 12.99 166.07 ± 6.15 66.67 ± 4.33 60--69 Women 36 65.67 ± 10.09 152.97 ± 6.72 28.12 ± 4.45 Men 34 72.94 ± 11.73 16312 ± 6.07 27.41 ± 4.23 ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Mean BMDs at the Lumbar Spine and Femur for each Age and Sex groups ::: Measurement Site Mean BMD in femur (gr/cm^2^) Mean BMD in Lumbar Spine(gr/cm^2^) ---------------------- ---------------------------------- -------------------------------------------- -------------------- ---------------- Age group Women Men Women Men 20--29 0.962 ± 0.132 1.098 ± 0.15 \\1.198 ± 0.1132 1.209 ± 0.132 30--39 1.022 ± 0.122 1.042 ± 0.146 1.206 ± 0.1249 1.216 ± 0.1414 40--49 0.968 ± 0.120 1.009 ± 0.144 1.158 ± 0.148 1.202 ± 0.176 50--59 0.9179 ± 0.120 0.966 ± 0.206 1.024 ± 0.178 1.120 ± 0.129 60--69 0.833 ± 0.111 0.935 ± 0.105 0.982 ± 0.161 1.117 ± 0.155 \\ Mean ± SD ::: Comparing of BMD between different age decades showed that; spinal BMD in age group of 40--49, 50--59 and 60--69 in women, 3%, 1% and 4% was less than previous age decade respectively. In the same age groups, spinal BMD in men 1%, 6% and 0% was less than previous age decade. Also, femur BMD in age group of 40--49, 50--59 and 60--69 in women, 5%, 5% and 9%, was less than previous age decade, and in men 3%, 4% and 3% was less than previous age decade respectively. (Figure [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### BMD of lumbar spine within age decades and sex (g/cm^2^). ::: ![](1471-2474-6-38-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### BMD of total hip within age decades and sex (g/cm^2^). ::: ![](1471-2474-6-38-2) ::: A significant relationship was found between age and BMD (P \< 0.001 and P \< 0.001). There was a significant relationship (P = 0.025) between BMD and BMI among the men but there was no relationship of this kind among the women. On the other hand, there was a significant relationship between BMI and femur BMD among both the women and the men (P \< 0.001 and P \< 0.001 respectively). The loss of BMD among the women was seen more than after menopause so that during the first ten years after the menopause, BMD of the spinal column and femur bone was 1.16 and 2.2 less than each year respectively. In this study, the mean femur BMD among the women with menopause was 10.5 % lower than the women without menopause (P \< 0.001). The mean spinal BMD among the women with menopause was 16% lower than the women without menopause (P \< 0.001). In total, 7.4% of all cases in lumbar spine and 2.4% in femur bone had osteoporosis and 30.4% in the spinal column and 23.9% in femur bone had osteopenia. (Table [3](#T3){ref-type="table"}) show the prevalence of osteoporosis and osteopenia in both sexes and age groups. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Prevalence of osteoporosis osteopenia in each age and sex groups ::: Osteoporosis(%) Osteopenia(%) ------------ ------- ----- --------------------- ------------------- ------ ------ 20--29 Women 44 0 2.2 13 17.4 Men 27 3.8 0 23.1 15.4 30--39 Women 104 0 0.9 13.9 8.3 Men 38 2.7 2.7 27 24.3 40--49 Women 98 3.2 2.2 29 15.1 Men 48 4.3 2.1 31.9 31.9 50--59 Women 82 21.8 2.6 46.2 38 Men 42 5.3 5.3 42.1 36.8 60--69 Women 36 32.4 5.9 50 50 Men 34 9.4 3.1 50 46.9 ::: Discussion ========== In this study, the mean of BMD from spinal column and femur in all age groups of men were more than that of women. Most similar studies justify these results through comparing the fluctuations of androgen level with estrogen level, in men the level of androgen to estrogen does not reduce compared with women. On the other hand, the bone mass, physical activities and peak bone mass in men were more than women \[[@B25],[@B26]\]. These results also indicate that the peak bone mass among women were in age of 30--39 which is in accordance with other studies \[[@B4]-[@B7],[@B25]-[@B27]\]. The mean of spinal column BMD among women 30--39, was 5.6%lower than the American women, 3.9% more than the Japanese women 6.5% more than Filipino women,,0.06% more than Lebanese women \[[@B28],[@B29]\]. There was no significant difference between these two groups because maximum BMD of Japanese occurs in women of 40--49 \[[@B26]\]. Many studies have demonstrated that alteration in BMD depends on type of the bone, different function, menstruation condition, environmental factors, genetics effects and age \[[@B30],[@B31]\]. As the results achieved in other countries indicate different means and amounts, the information obtained through this study show a similar BMD pattern. The present study suggests that the maximum BMD of femur bone compared with spinal column occur later. This is justifiable considering the fact that the maximum BMD in cortical bone compared with trabecular bone occurs later \[[@B27],[@B30],[@B31]\]. The amount of BMD of spinal column was lower of 15.6% after ten years, and this reduction was 16.9% more than the Japanese women \[[@B33],[@B35]\]. The Peak bone mass of femur bone was 4.48% less than the American women \[[@B34]\]. The pattern of bone loss, both in femur and lumbar, depends on the age. The pace of loss in bone mass up to menopause period, in the women in question, is similar to the Canadian, British and American women. Although after that period, it is faster than the women of Belgium, United Kingdom, France and America whereas it is less than the Japanese women \[[@B32],[@B33]\]. On the other hand, the rate of bone loss compared with the other studies (Western Belgian, Japanese,) is either the same or more \[[@B32]-[@B35]\]. This trend, therefore, caused an increase in incidence of osteoporosis and osteopenia. In men, the peak bone mass in spinal column was 3.5% less than the American men \[[@B7]\]. The incidence of osteoporosis and osteopenia among women and men was 32.4% and 9.4% respectively, which is more than of Lebanese and Thai women and less than Hong Kong and US women, this is justifiable with regard to the above-mentioned explanations. Studies indicate that the peak bone mass plays an important role the incidence of osteoporosis which this peak bone mass depends on genetics, kind of diet, sport and the hormonal state. Genetics was the most important factor justifying low BMD in our study. As well, with respect to the deficiency of vitamin D in Iran, which is common among 80% of people in some areas, and also lack of enough activity, in particular, among young girls of 20--29 can cause the low level of bone mass \[[@B7]\]. Conclusion ========== It is notice-worthy that this paper represents the early results of the comprehensive plan for prevention, diagnosis and treatment of osteoporoses carried out in the EMRC of Tehran University of Medical Sciences, which is still being conducted and is not finished yet. With completing the project and data gathering and studying all patients fully, the final results could be achieved and the relationships could be analytically discussed. Broadly speaking, the present study indicates the high incidence of osteoporosis and osteopenia among the Tehran population, which requires our proper attention and planning for prevention. Also, the low amount of peak bone mass in the ages 20--39 is helpful to adopt an adequate strategy in this respect. There are many factors involved in this maximum BMD including genetic factors, body activity and providing enough vitamin D and calcium. Among the intervening factors, enough nutrition together with calcium and vitamin D could be enumerated. The results of this study indicate that there is an increase in bone mass in the first decade after menopause requiring a proper treatment during the years before and after menopause. The limitations of this study are the following points. This is the primary result of the national comprehensive study of osteoporosis in Iran. The individuals excluded from this study based on personal interview and their statement about their disease, not documented diagnostic diseases. In addition, the data of this study limited to Tehran and, our data assumed as an estimation of reference data in Iran. For demonstration of Iran\'s data, a bigger study is needed. List of abbreviations ===================== EMRC: Endocrinology and Metabolism Research Center, TUMS: Tehran University of Medical Sciences, BMD: Bone Mineral Density, BMI: Body Mass Index, DXA: Dual X-ray Absorption Competing interests =================== The author(s) declare that they have no competing interest. Authors\' contributions ======================= Conception and study design and coordination: BL, AH, Drafting manuscript and Data analysis: AH, AM, Participation in its sequence alignment: MP, AS, MHB, SZM, RD, Review of manuscript and Important intellectual content: BL, AH, AM, All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2474/6/38/prepub> Acknowledgements ================ The help of EMRC BMD unit especially Dr. Hamdi is gratefully acknowledged.	PubMed Central	2024-06-05T03:55:59.483406	2005-7-2	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180448/", "journal": "BMC Musculoskelet Disord. 2005 Jul 2; 6:38", "authors": [ { "first": "Bagher", "last": "Larijani" }, { "first": "Arash", "last": "Hossein-Nezhad" }, { "first": "Alireza", "last": "Mojtahedi" }, { "first": "Mohammad", "last": "Pajouhi" }, { "first": "Mohammad H", "last": "Bastanhagh" }, { "first": "Akbar", "last": "Soltani" }, { "first": "Seyed-Zahra", "last": "Mirfezi" }, { "first": "Roya", "last": "Dashti" } ] }
PMC1180449	Background ========== Headache complicates approximately 10 to 40% of dural punctures \[[@B1]\]. This postdural puncture headache (PDPH) is typically orthostatic; provoked or aggravated by a vertical or upright position and relieved by a horizontal position. PDPH is probably caused by cerebral spinal fluid leakage through the dural rent, into the epidural space. The leakage causes a decrease in CSF pressure and volume, leading to traction on pain-sensitive structures in an upright position. Besides of headache, the patient may complain of diplopia, tinnitus, dizziness, and myalgia. PDPH may occur immediately after spinal tap, but it starts within 48 hours after the procedure in more than 90% of the patients. PDPH and accompanying symptoms are self-limiting. They generally resolve within 7 days or less, in 80% of the cases. In a small minority of cases, the symptoms may persist for weeks or even months \[[@B2]\]. During an episode of PDPH the patient may be completely incapacitated and confined to bed. Obviously this has financial, social and psychological repercussions. Different prophylactic measures such as: small needle size, the use of Sprotte\'s needle, reinsertion of the stylet before withdrawing the needle, and direction of the brevel perpendicular to the dura, have all been shown to reduce the occurrence of PDPH \[[@B3]-[@B6]\]. If, despite the prophylactic measures, PDPH occurs, epidural blood patch (EDBP) may be a beneficial therapeutic intervention. EDBP has gained popularity as a therapeutic measure for PDPH. It involves the injection of 10--20 ml of autologous blood into the epidural space around the site of the spinal tap. Gormly introduced this technique in the 1960\'s \[[@B1]\]. He noticed that inadvertent bloody spinal taps were less often complicated by PDPH. He theorised that the epidural bleeding might lead to clot formation over the dural rent, preventing CSF leakage into the epidural space. He therefore continued to treat 6 subjects suffering from PDPH with EDBP, locating the epidural space with the hanging-drop or loss of resistance method. All 6 subjects were relieved of their complaints. Many observational studies followed; they reported success rates of the EDBP for PDPH between 70% and 90% \[[@B7]-[@B12]\]. Seven controlled trials concerning prophylactic treatment have been published \[[@B13]-[@B19]\]. One of these studies was not blinded \[[@B18]\], three were not randomised \[[@B14]-[@B16]\], one was only reported as an abstract \[[@B17]\]. One other study \[[@B13]\] compared prophylactic EDBP versus no blood patch among obstetric patients and reported a high success rate. In this study adverse effects were not mentioned, which prevents firm conclusions. The latest study \[[@B19]\], in which prophylactic EDBP was compared to a sham procedure in a double blind setting, showed no decrease in the incidence of PDPH between the two groups. The effectiveness of prophylactic treatment does not seem to have been established firmly. Only one randomised and blinded trial concerning the therapeutic effect of EDBP has been reported \[[@B20]\]. In this study 12 patients, suffering from PDPH for more than 4 days, despite conservative treatment following lumbar puncture, spinal anaesthesia or myelography, were randomly allocated to EDBP or sham treatment. In the placebo group none of the patients noted complete relief of pain. In the treatment group 5 of the 6 patients obtained immediate relief. Subsequently placebo group patients were also treated with an EDBP, resulting in complete relief of PDPH in all patients. The size of the study, the crossover effect, and the absence of any documentation regarding the effectiveness of blinding of the observers and patients, makes it difficult to draw firm conclusions from this study. A Cochrane review on prophylactic and therapeutic blood patching \[[@B21]\] argued that further randomised trials of epidural blood patching must be carried out before the balance of risks and benefits of this intervention can be properly assessed\". Therefore, we decided to conduct a randomised, controlled, clinical trial comparing the efficacy of the EDBP with conservative treatment, consisting of 24 hours of bed rest and adequate fluid intake for the treatment of PDPH. Methods ======= Inclusion criteria ------------------ To be included in the study, patients should have PDPH longer than 24 hours and not longer than 7 days after a diagnostic spinal tap. Furthermore patients should be aged 18 years or older. Written informed consent is required. Exclusion criteria ------------------ Excluded are patients with relative contra-indications for lumbar puncture: hemorrhagic diathesis and space-occupying intracranial lesions, and patients with a body temperature over 38° Celsius. Design ------ This is an observer-blind prospective, controlled and randomised parallel group study. Active treatment ---------------- Patients allocated to active treatment will receive an EDBP on the day of randomisation. The subject is placed in the lateral position, after which the back is flexed, sterilised and draped. Sterile gloves are used. A needle (Spinocan canule: 0.9 × 88 mm/206 × 3.5) is placed in the epidural space, using the loss of resistance technique. Subsequently, 20 cc of blood is than drawn from the antecubital vein, and injected slowly into the epidural space, after which the needle is removed. The subject is held in the supine position for a few minutes, after which there are no further restrictions. Control treatment ----------------- Conservative treatment consists of the advice to take 24 hours bed rest and drink at least 2.0 litres of fluid a day. The use of painkillers is not prohibited. Treatment with EDBP is not an option during the study period of 7 days, not even when conservative treatment fails. Outcomes -------- The primary outcome is whether or not headache is present at 24 hours after the start of treatment. Headache is classified on a 4-point scale (no, mild, moderate, severe). Mild headacheis defined as: postural headache slightly restricting daily activities. The patient is not confined to bed and there are no associated symptoms. Moderate headache:postural headache confining the patient to bed for part of the day. Associated symptoms are not necessarily present. Severe headache:postural headache where the patient is bedridden for the entire day and associated symptoms are always present. The associated symptoms are: nausea, vomiting, dizziness, hearing loss, hyperacusis, tinitus, photophobia, diplopia, stiffness of the neck and scapular pain \[[@B2]\]. Secondary outcome measures are the presence of headache at day seven after the start of treatment, and the number of days until headache subsides. Treatment complications ----------------------- A systematic assessment of complications of EDBP will be carried out. Back pain will be assessed systematically, and classified as no, mild moderate, severe. Statistical analysis -------------------- The number of patients with headache at 24 hours after start of treatment will be compared between the two treatment strategies. The effect of treatment on the occurrence of the headache will be expressed as an odds ratio with 95% confidence interval. Adjustment for the effect of potential confounders, such as age and sex, will be made by multiple logistic regression analysis. The number of days until headache relief will be compared with Kaplan-Meier survival analysis techniques; observations will be censored at the end of the study period, i.e. 7 days. Adjustments for the effect of potential confounders will be made with proportional hazards regression \[[@B22]\]. Sample size ----------- A trial with 20 patients in each treatment group will provide us with a power (1-β) of 80% to detect a relative reduction in number of patients with persisting PDPH after 24 hours of 50% (OR = 12) and a power of 99% to detect a relative risk reduction of 80%, (OR = 81, at the usual significance level α = 5%, taking into account that in approximately 10% of the patients the PDPH will have resolved spontaneously after one day. Recruitment of eligible patients -------------------------------- Patients receiving a diagnostic dural puncture are informed about the possibility of developing PDPH. They all receive written information about the study. A subject suffering from headache after a dural puncture will be advised to contact his physician. The physician evaluates whether the headache is a PDPH or not. When the conclusion is that the subject is suffering from PDPH, he or she is asked to participate. Randomisation is done by a telephone call, by the investigator. Randomisation procedure ----------------------- The randomisation procedure is carried out by one of the investigators (RO or FvK). During the telephone contact, while the patient is entered into the computer database the treatment allocation (based on a random number generator) is provided by the computer. The patient cannot be entered twice into the study, nor can the entry be erased. Ethical considerations and informed consent ------------------------------------------- The local medical ethics committee and review board approved this study. Written informed consent will be obtained from all patients, by asking them to return the form by mail or in person, after the randomisation by telephone. Baseline data ------------- At baseline, patient demographics age and gender will be registered. Both have been proven to be independent risk factors for the development of PDPH. Clinical characteristics, such as headache characteristics, other complaints, and use of analgesics will be noted. Information concerning needle type, size and re-insertment of the stylet before withdrawing is registered, since all three factors are independent risk factors for the development of PDPH \[[@B3]-[@B5]\]. Furthermore the severity of the PDPH is registered on a four-point scale. Follow-up and blinding ---------------------- The follow-up visits are carried out by telephone, by a research nurse at the trial office, 24 hours after randomisation, and at 1 week after randomisation. The research nurse is kept blind to the treatment allocation. At the beginning of the telephone interview the patients are instructed not to inform the observer of the treatment they received. The effectiveness of the blinding will be checked in a sample of 12 patients, by letting the research nurse fill in a forced choice item indicating the treatment allocation. Discussion ========== The EDBP, is widely accepted as the treatment of choice for PDPH. In a review of the literature on this subject however, we found only little evidence to justify its use in general practice \[[@B21]\]. Therefore, we decided to conduct an observer blinded, randomised, controlled, study comparing the therapeutic efficacy of the EDBP to the efficacy of conservative treatment. Our study design not only enables us measure the effectiveness of the EDBP, but also its efficiency. The number of days spent with headache after treatment in the two groups will be characterised as incapacitated-days. Immediate relief of headache, and reduction of the number of incapacitated-days is the ultimate goal of this treatment. By taking into account the adverse effects of EBPD, most notably lower back pain, we will be able to make a more realistic estimate of the overall effect of EPBD. The observer-blind method we have chosen enables us to compare the two treatment strategies of interest as if they are carried out in normal practise, without the artificial circumstances created by sham treatment, in futile attempts to maintain patient blinding. So a genuine comparison is made between the experimental treatment and normal clinical practise. Because patients are of course aware of the nature of the treatment, the study design limits our ability to assess the placebo effect of the epidural blood patch. One may argue that this may lead to overestimation of the effect of the blood patch. We think that for a treatment of a self/limiting condition this is not really a drawback. Another potential limitation is the small sample size of our study. Although our trial will turn out to be the largest randomised study of therapeutic EDBP ever, a sample size of 20 patients in each treatment group will provide us sufficient power (1-β = 80%) to detect a relative reduction in the risk of PDPH after 24 hours of 50%. Much depends therefore on our assumption that only in approximately 10% of the patients the PDPH will have resolved spontaneously after one day. Currently, the study is well underway, and we expect to be able to report its results in the end of 2005. Conclusion ========== EDBP is accepted as the treatment of choice for PDPH although randomised, controlled data is scarce. Our randomised, observer-blind clinical trial enables us to compare the efficacy of two clinically practiced methods of PDPH treatment; EDBP versus conventional treatment, as they are applied in clinical practise. List of abbreviations ===================== CSF: Cerebrospinal fluid EDBP: Epidural blood patch PDPH: Post dural punction headache Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= RO participated in the study design and acquisition of data, and drafted the manuscript. FK participated in data acquisition, coordination of the study, and helped to draft of the manuscript. SB participated in the conception of the study, and in data acquisition. DD participated in the conception of the study, data acquisition, and drafting of the manuscript. All authors have seen and approved the final manuscript. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Schematic section through the vertebral column, showing the cauda equina and its covering membranes with the dural leakage site before (A) and after (B) application of the epidural blood patch. ::: ![](1471-2377-5-12-1) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Therapeutic studies of EDBP for PDPH ::: Author Number of patients Type of study Blinding Time window to inclusion Follow-up Outcome Adverse effects ----------------------------- ----------------------------------------------------------------------------- --------------------------------------------------------- -------------- ------------------------------ --------------- --------------------------------------------------------- ------------------------- Safa-Tisseront 2001 \[8\] 527 (504 analyzed) Therapeutic, prospective, non-randomised, observational NA 4 days (1--53) 15 days 75% complete relief, 18% incomplete relief, 7% failure Fever in 3 patients Williams 1999 \[12\] 55 (7 prophylactic, 41 therapeutic EDBP, 7 conservative treatment) Retrospective, non-randomised, observational NA \<1 week ? 34% complete relief, 54% incomplete relief, 12% failure Back pain in 3 patients Banks 2001 \[9\] 100 (81 pdph7 prophylactic, 58 therapeutic EDBP, 23 conservative treatment) Prospective, non randomised, observational NA 0--3 days ? 67% complete relief, 28% incomplete relief, 5% failure 22% back pain Taivainen 1993 \[7\] 81 (55 patients 10 ml, 26 patients 10--15 ml EDBP) Prospective, partly randomised, observational NA 0--10 days 1 week Initial relief 91%, permanent relief 61% 25% back pain Vercauteren 1999 \[11\] 190 (186 EDBP, 4 atypical symptoms) Retrospective, non randomised NA \>24 h after symptoms 1 day Initial relief 99%, permanent relief 73% ? Stride 1993 \[10\] 34819 (461 PDPH, 137 EDBP attempted, 135 completed) Retrospective, non randomised NA 2 days ? Initial relief 90%, permanent Relief 64% ? Seebacher 1989 \[20\] 12 (6 EDBP, 6 sham treatment) Prospective, randomised Yes \>4 days 1 day EDBP 83% relief sham treatment 0% relief Back pain ::: Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2377/5/12/prepub> Acknowledgements ================ This study was supported by the Netherlands Headache Society. We thank Ron Meier, neurologist, for training us in delivery of the blood patch.	PubMed Central	2024-06-05T03:55:59.485362	2005-7-5	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180449/", "journal": "BMC Neurol. 2005 Jul 5; 5:12", "authors": [ { "first": "R", "last": "Oedit" }, { "first": "F", "last": "van Kooten" }, { "first": "SLM", "last": "Bakker" }, { "first": "DWJ", "last": "Dippel" } ] }
PMC1180450	Background ========== The dentate gyrus (Dt) may be known best for its input from the entorhinal cortex and its mossy fiber projection to the CA3 \[[@B1]-[@B3]\]. In addition, the Dt has extensive associational and commissural connections \[[@B4]-[@B8]\] considered crucial for the generation of gamma frequency oscillations and learning (e.g. \[[@B9]-[@B11]\]). Most attention among the dentate interneurons may recently have received the cells in the hilus region (DtHi) due to their vulnerability to various injuries \[[@B12]-[@B14]\] as well as their remarkable structural, histochemical and electrophysiological diversity \[[@B15]-[@B20]\]. A few hilar neurons also project to the cornu ammonis (CA) and the subiculum (Sbi) \[[@B15],[@B21]\]. Extrahippocampal projections from the dentate gyrus, however, are not known with the exception from some septal projecting neurons in the DtHi \[[@B22]-[@B25]\] usually considered displaced cells of CA3 \[[@B1],[@B21]\]. Continuing our attempts to elucidate the hippocampal and parahippocampal circuits in mammals with a poorly differentiated brain \[[@B26]-[@B29]\] the present study in the lesser hedgehog tenrec (Echinops telfairi, Et) will demonstrate some true dentate hilar cells projecting to a circumscribed region in the olfactory tubercle. The projection may represent an additional parallel pathway transferring cortico-limbic information to the ventral striatum \[[@B30]-[@B32]\]. Results ======= Axo-terminal labeling in the striatum following tracer injections into the dentate gyrus ---------------------------------------------------------------------------------------- Among the experiments with tracer injections into the hippocampus (Table [1](#T1){ref-type="table"}) two cases were injected with biotinylated dextran amine (BDA) into the dentate area exclusively. These tracer injections mainly involved the dentate molecular layer (DtMo) but there was some tracer uptake by the granule cells as seen from the weak to moderate mossy fiber projection to CA3 (Fig. [1](#F1){ref-type="fig"}). The sparse (Et01-36B, Fig. [1C](#F1){ref-type="fig"}) to moderate (Et01-21B, Fig. [1A](#F1){ref-type="fig"}) number of labeled cells in the dentate hilar region subjacent to the main injection site might be due to a retrogarde transport and/or a direct uptake of tracer substance. In Et00-11B and Et98-49W (Fig. [1B](#F1){ref-type="fig"}) all layers of Dt as well as the adjacent portions of CA3 were labeled directly with BDA and wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP), respectively (the letters B and W following the case number indicate the tracer injected). The Dt and portions of CA1 were injected with BDA in Et00-08B, while in the remaining experiments the injection involved the subiculum (Sbi) with (Et00-13W) or without (Et00-41B) an involvement of the Dt and CA1. The latter two cases revealed the well-known hippocampo-striatal projection pattern to the nucleus accumbens (Acb) and the olfactory tubercle (Tu) \[[@B29]\]. No hippocampo-striatal projection, on the other hand, was found in Et01-36B (Table [1](#T1){ref-type="table"}, Fig. [1G](#F1){ref-type="fig"}) injected with tracer almost exclusively into the DtMo. The remaining cases consistently showed some fiber labeling in the Tu (Fig. [1E](#F1){ref-type="fig"},[1F](#F1){ref-type="fig"},[1H](#F1){ref-type="fig"}) but did not reveal any projections to the Acb (Table [1](#T1){ref-type="table"}; Fig. [1D](#F1){ref-type="fig"}). The labeling in Tu involved especially the molecular layer in the medial quarter of Tu at levels rostral to the insula magna. In all cases but Et01-21B the hippocampo-tubercular projection was represented on both sides with a slight ipsilateral predominance (Fig. [1H](#F1){ref-type="fig"},[1I](#F1){ref-type="fig"},[1J](#F1){ref-type="fig"}). In Et01-21B the course of fibers towards Tu appeared to pass exclusively within the molecular layer of the hippocampal continuation (HCt; Fig. [1D](#F1){ref-type="fig"}; also note the difference to Fig. [1G](#F1){ref-type="fig"} of Et01-36B). In the other cases labeled fibers were seen in the HCt as well as the fornix. A fiber course across the fornix appeared most likely for the contralateral projection as far as the contralateral HCt was almost unlabeled and a fiber crossing was never observed for the HCt-fibers unlike the fornix fibers. For the subsequent experiments carried out in order to identify more accurately the origin of the hippocampo-tubercular projections, it should be mentioned that similar to other reports \[[@B33]\] there were also hippocampal projections to the diagonal band-medial septum complex (DgMS). Evidence for these projections was obtained from the BDA experiments which revealed an extensive terminal-like fiber pattern across the DgMS, but only few, if any labeled perikarya unlike the WGA-HRP experiments (Table [1](#T1){ref-type="table"}). Perikaryal labeling in the dentate gyrus following tracer injections into the striatum and septum ------------------------------------------------------------------------------------------------- There were two cases with WGA-HRP injections involving the rostromedial Tu including its molecular layer (Table [2](#T2){ref-type="table"}). The tracer injection in Et03-58W showed a relatively circumscribed injection site (Fig. [3A](#F3){ref-type="fig"}); a direct labeling of the DgMS, however, could not be excluded. In Et01-47W there was some loss of tracer along the electrode tract leading to a contamination of the third layer of the HCt. A control injection into the ventral septum including the DgMS (Fig. 10L in \[[@B26]\]) was available too as well as two tracer injections involving the Acb (Table [2](#T2){ref-type="table"}; Fig. [2A](#F2){ref-type="fig"}). Consistent with previous studies in subprimate species \[[@B34]-[@B38]\] the tracer injections into the Acb led to retrograde labeling of exclusively the Sbi and CA1 (Fig. [2B](#F2){ref-type="fig"}; regarding the involvement of CA3 in monkey see ref.\[[@B39]\]). The septal injection, in addition, labeled numerous neurons in CA3 (Fig. [2C](#F2){ref-type="fig"}, [2D](#F2){ref-type="fig"}) as well as a substantial number of cells in the DtHi adjacent to the heavily labeled CA3, particularly at caudal levels (Fig. [2D](#F2){ref-type="fig"}, [2E](#F2){ref-type="fig"}). In the cases injected with tracer into the Tu labeled cells were found in the HCt (Fig. [3B](#F3){ref-type="fig"}, [3C](#F3){ref-type="fig"}), the Sbi (Fig. [3F](#F3){ref-type="fig"}) and the DtHi (Figs. [3D](#F3){ref-type="fig"},[3E](#F3){ref-type="fig"},[3F](#F3){ref-type="fig"},[3G](#F3){ref-type="fig"},[3H](#F3){ref-type="fig"}). These cases, however, consistently failed to show any retrograde labeling in the CA3 (or the dentate granule cells layer). In Et03-58W the labeled neurons in the DtHi were restricted to its rostral third, while in Et01-47W they involved rostral as well as caudal portions of the hilus. The latter case also showed a very few labeled cells on the contralateral side. A particular distribution of the labeled cells within the DtHi was not obvious, and their poor dendritic labeling did not allow a further characterization of the cells. Discussion ========== Unlike the dentate projection to the HCt the projection to the olfactory tubercle cannot be explained by a retrograde-anterograde collateral transport of tracer \[[@B28]\]. Tubercular terminations are not seen in the case with a BDA injection largely confined to the dentate molecular layer (Fig. [1C](#F1){ref-type="fig"},[1G](#F1){ref-type="fig"}), and tubercular projections are noted in the cases with dentate injections of WGA-HRP (Fig. [1B](#F1){ref-type="fig"},[1E](#F1){ref-type="fig"},[1F](#F1){ref-type="fig"}), a tracer substance not known to be transported in a retrograde-anterograde collateral fashion \[[@B40]\]. Moreover, the dentate-hilar neurons are also filled by the retrograde axonal flow from the tracer deposits in the Tu. The projection to the olfactory tubercle differs from previously reported extrahippocampal projections from the dentate area as far as its cells of origin are confined to the DtHi while the septal \[[@B22],[@B23],[@B41]\] and hypothalamic \[[@B42]\] projections arise predominantly from the CA and the Sbi, only sparsely from the DtHi. Unlike the latter cells, therefore, the dentato-tubercular neurons can not be considered displaced cells of CA3. We have not characterized yet the particular morphology and histochemistry of these neurons \[[@B16],[@B26],[@B43]-[@B45]\] and do not know whether their projections are distinct ones or arise as collaterals \[[@B46]-[@B49]\]. Nevertheless, it is a projection which involves a circumscribed region in the olfactory tubercle and avoids the nucleus accumbens, the main striatal target area of the hippocampus \[[@B34],[@B35],[@B39]\]. These findings are in line with previous suggestions considering the medial Tu as a separate striopallidal subdivision \[[@B50]-[@B53]\] and fit with the concept of the parallel processing of cortico-striatal information \[[@B30],[@B31],[@B54]-[@B56]\]. The question remains as to why a dentato-tubercular projection has not been demonstrated so far in other species. The projection may be unique to tenrecs or mammals with little differentiated brain but its apparent absence in other mammals may as well be explained by the low strength of its projection and/or its circumscribed field of termination. Such a projection may easily be overlooked in anterograde tracer studies, especially in species with a large brain. Retrograde tracer studies, on the other hand, have only been done in the rat and the hamster \[[@B37],[@B50],[@B57],[@B58]\] and it is questionable, whether their small circumscribed dentato-tubercular target area, if present, has been injected with tracer. Rodent species, in addition, are not representative for studying the hippocampo-tubercular projections. In the rat, the Tu receives considerably less hippocampal afferents from the CA1/Sbi fields \[[@B35]\] than in the tenrec \[[@B29]\], cat \[[@B34]\] and monkey \[[@B39]\]. Similarly, the relative strength of the entorhinal projections to the Tu as compared to the Acb is much less in rodents \[[@B36],[@B59],[@B60]\] than in non-rodent species \[[@B29],[@B39],[@B61]\]. It may also be noted that the hippocampal continuation, the only other striatal input region projecting exclusively to the tenrec\'s rostromedial Tu \[[@B28]\], is poorly represented in the rat compared to other mammals \[[@B62]\]. One may speculate that the dentato-tubercular neurons in the DtHi represent an extension of the HCt into the dentate gyrus and are only present in species showing a well differentiated HCt as e.g. in tenrecs and primates \[[@B62]\]. Unfortunately we still know little about the precise connectivity of the circumscribed dentato-tubercular target area. It receives few if any direct projections from the olfactory bulb (personal reinvestigation of the material published previously \[[@B63]\]) but gets a distinct input from the anteromedial frontal cortex \[[@B29]\]. The dentato-tubercular target area is likely to be part of a cortico-basal ganglia-thalamo-cortical loop involving the mediodorsal nucleus \[[@B64]-[@B67]\] and the prefrontal cortex, at least in subprimate species (rat: \[[@B68]-[@B70]\]; cat: \[[@B71]\]; monkey: \[[@B72]-[@B74]\]). Notably, while the prefrontal and hippocampal afferents to the Acb terminate in a predominantly non-overlapping fashion \[[@B29],[@B32],[@B75]\], the cortico-striatal fibers appear to overlap considerably with the hippocampal afferents in the Tu. Conclusion ========== The present data show for the first time a projection from the dentate gyrus to the rostromedial olfactory tubercle. Its cells of orgin, located in the dentate hilus, cannot be considered displaced neurons of CA3 but represent true dentato-tubercular projection neurons. The projection is assumed to be present in other mammals, too. In the tenrec the dentato-tubercular projection overlaps with the subiculo-tubercular projection and the fiber terminations arising in the anteromedial frontal cortex. The circumscribed ventral striatal target area appears worth investigating in studies of the parallel processing of cortico-limbic information in tenrec as well as in cat and monkey. Methods ======= Animals ------- The lesser hedgehog tenrec (Echinops telfairi, Et) is a member of the tenrecomorpha classically considered an insectivoran suborder \[[@B76]\] but actually grouped within the superorder Afrotheria (order Afrosoricida e.g. \[[@B77],[@B78]\]). The animals were obtained from our breeding colony \[[@B79]\] built up to investigate their little differentiated forebrain and get an insight into the evolution of the mammalian brain \[[@B80],[@B81]\]. Tracer injections ----------------- Eleven tenrecs weighing between 80 and 140 gm were anesthetized with tribromoethanol (1.0 ml/100 gm, i.p.) and injected with tracer into the hippocampus (n = 7), the ventral striatum (n = 4) and the septum (n = 1) following the German laws on protection of animals (one animal received two injections at different locations). Wheat germ agglutinin conjugated to horseradish peroxidase (WGA-HRP; Sigma; n = 7) and biotinylated dextran amine (BDA; Sigma; n = 5) were used as tracer substances. Most cases have been described previously with regard to other connections \[[@B26],[@B28],[@B82]\]. The additional tracer injections (Et01-47W, Et03-58W) were done in the same fashion: The WGA-HRP (1.5--8 nl of a 2--5% solution in distilled water) was pressure injected through a glass micropipette (tip diameter 8--15 μm) attached to a Hamilton syringe driven by a micromanipulator. The BDA was injected iontophoretically (10 % in 0.01 M phosphate buffer, pH 7.25; current of 2--5 μA for 5--10 min). After a survival time of 2--7 days the animals were reanesthetized and perfused transcardially with saline. The fixation consisted of a phosphate-buffered solution containing 1% paraformaldehyde and 2.5% glutaraldehyde, followed by a phosphate-buffered solution containing 15% sucrose. Tissue processing ----------------- The brains were soaked overnight in a 30% phosphate-buffered sucrose solution, embedded in an albumin-gelatine mixture and cut on a freezing microtome at 40 μm in the frontal plane. The WGA-HRP staining was usually done on two out of four sections according to the standard tetramethylbenzidine technique \[[@B83]\] and a modified tungstate-stabilized tetramethylbenzidine technique \[[@B84]\]. Following a pretreatment in mercaptoethanol (2% in 0.05 M PBS at 37° for 5--15 min in order to suppress a staining of WGA-HRP) the BDA was visualized by incubating the sections in extravidin-peroxidase and developing the reaction product with nickel-enhanced diaminobenzidine and tungstate-stabilized tetramethylbenzidine (modified according to \[[@B84]\]) in an alternating fashion. The mounted sections were counterstained with neutral red (following treatment with diaminobenzidine and tetramethylbenzidine) and cresyl violet (following use of tungstate-stabilized tetramethylbenzidine). The tungstate-stabilized tetramethylbenzidine proved quite useful for visualizing faint projections but might result in large cristalline artefacts, particularly after overnight incubation. Data analysis ------------- Bright-field illumination and polarized light were used for the analysis. Images were captured on an Ilford 50 negative film or a Fujichrome 64T positive film (using a Zeiss axiophot microscope), scanned (Nikon Coolscan 5000) and transported into Adobe Photoshop (v. 7.0) and Corel Draw (v. 11.0). The sharpness, contrast and brightness were adjusted to reflect the appearance of the labeling seen through the microscope. For the sake of clarity the most disturbing crystalline artefacts were removed, particularly in darkfield images. In some darkfield micrographs the light reflecting embedding medium was also removed. Furthermore to facilitate the analysis all micrographs are shown with the injection site on the left. List of Abbreviations ===================== Acb -- nucleus accumbens BDA -- biotinylated dextran amine CA1/3 -- subfields of cornu ammonis cma -- commissura anterior DgMS -- diagonal band-medial septum complex Dt -- dentate area DtGr -- dentate granule cell layer DtHi -- dentate hilar region DtMo -- dentate molecular layer Et -- Echinops telfairi HCt -- hippocampal continuation Sbi -- subiculum Tu -- olfactory tubercle WGA-HRP -- wheat germ agglutinin conjugated to horseradish peroxidase Acknowledgements ================ The excellent technical assistance of Angelika Antonius, Amela Klaus, Antonja Nekic and Sigi Schaller is gratefully acknowledged. The work was supported by the Deutsche Forschungsgemeinschaft, grant Ku 624/3-2. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Anterograde labeling in the olfactory tubercle following tracer injections into the dentate gyrus.Photomicrographs taken by transmitted and polarized light showing anterograde projections to the olfactory tubercle (D, E, F, H, I, J; arrows point to labeling) and/or the anterior hippocampal continuation (D, G) following tracer injections into the dentate gyrus in Et01-21B (A, D), Et98-49W (B, E, F), Et01-36B (C, G) and Et00-11B (H -- J). The molecular layer of Tu was labeled in all cases (D-H, ipsilateral side; I and J, contralateral side) but Et01-36B (G) where the tracer injection scarcely involved the DtHi (note the low labeling density of the mossy fiber projection to CA3 in C \[arrow head\] as compared to A). The Acb (D, E) was not labeled following tracer injections into Dt unlike the cases injected with tracer into CA1 / Sbi. All BDA cases also clearly labeled the HCt (D, G) due to the retrograde-anterograde collateral transport of the tracer within the entorhino-hippocampal system described previously \[28\]. Open arrow heads in E, F and I, J respectively point to similar location. Scale bars in B (also applying to A and C) and G (also applying to D) 0.5 mm; in E and J = 0.4 mm, in H = 0.3 mm; in F and I = 0.2 mm. ::: ![](1471-2202-6-38-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Retrograde labeling in the hippocampus following tracer injections into nucleus accumbens and septum.Following WGA-HRP injection into the Acb (A) in Et02-65W there were numerous retrogradely labeled cells in Sbi and CA1 (B), but none in CA3 and Dt (B). The tracer injection into the septum of Et00-08W (see Fig. 10L in \[26\]) also labeled heavily the CA3 (C-E) as well as some cells in the adjacent DtHi (D, E), particularly at caudal levels (section D is taken 960 μm caudal to C and 160 μm rostral to E). Scale bars = 0.8 mm in A, 0.4 mm in B, 0.3 mm in E and 0.2 mm in D (also applies to C). ::: ![](1471-2202-6-38-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Retrograde labeling in DtHi and HCt following tracer injections into the olfactory tubercle.Tracer injections into the rostromedial Tu (A) consistently labeled a few dentate hilar neurons (D-H; arrows point to labeled cells), but failed to label the CA3 (D-F). A-E are from Et03-58W, F-H from Et01-47W. Some labeled neurons are also noted in the ipsilateral, anterior HCt (B, C) confirming previous anterograde data \[28\]. Remarkably, both the HCt and the DtHi project to the rostromedial Tu, but not to the Acb. Arrow heads point to similar location. OfB, olfactory bulb; PCx, paleocortex. Scale bars = 0.8 mm in A, 0.4 mm in B and D, 0.3 mm in F, 0.2 mm in C, E and H (as G). ::: ![](1471-2202-6-38-3) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Cases showing hippocampo-basal telencephalic projections with anterograde tracing techniques ::: Case Injection site Terminal projections ---------- --------------------- -------------------------- -------- ------------ Et01-36B DtMo\>\>DtGr~±DtHi~ \- \- \- Et01-21B DtMo\>DtGr~±DtHi~ ++ \- ± Et00-11B Dt\>CA3 \+ \- +++^b^ Et98-49W Dt\>CA3 \+ \- ++^c^ Et00-08B Dt, CA1 + \- ++^a^ Et00-13W Dt, CA1\<Sbi + ++ +++^c^ Et00-41B Sbi ++ ++ +++^a^ Letters B and W following the case numbers indicate the tracer injected, BDA and WGA-HRP respectively. The regions injected with tracer are written with large or small letters depending on their major or minor contribution to the projection pattern (± indicates questionable involvement). The symbols -, ±, +, ++, +++(none, questionable, weak, moderate, strong) refer to the overall intensity and extent of the projections. The letters superscripted in column DgMS refer to additional labeling of isolated (a), few (b) and numerous (c) cell bodies. ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Cases showing hippocampo-basal telencephalic projections with retrograde tracing techniques ::: Case Injection site Origin of projection ---------- -------------------- -------------------------- --------- --------- --------- Et03-58W TuM + \- \- ± Et01-47W TuM~+HCt~\* +± \- ± ++ Et02-63W Acb~+FrCx~ \- \- ++ +++ Et02-65W Acb \- \- ++ +++ Et00-08W SeV ++ +++ +++ +++ The areas injected with tracer and the density of labeled perikarya are judged as in Table 1. \* The contamination of the anterior HCt by the tracer is restricted to the third layer not connected with the hippocampus \[28\]. FrCx, frontal cortex; SeV, ventral septum. :::	PubMed Central	2024-06-05T03:55:59.487474	2005-5-31	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180450/", "journal": "BMC Neurosci. 2005 May 31; 6:38", "authors": [ { "first": "Heinz", "last": "Künzle" } ] }
PMC1180451	Background ========== The critical role of the suprachiasmatic nucleus (SCN) in the orchestration of circadian rhythms represents one of the most successful brain-behavior relationships established in mammalian neurobiology \[[@B1]\]. The cellular and molecular bases of circadian rhythmicity are similarly well characterized \[[@B2],[@B3]\]. Despite these advances, the practical manipulation of human rhythms remains unrealized. Most shift-workers, for instance, fail to reset their circadian pacemakers to effect an alignment of the work shift with the worker\'s circadian day \[[@B4]-[@B6]\]. This misalignment may be a contributing factor in the higher nighttime incidence of industrial accidents and the compromised health of shift-workers \[[@B7],[@B8]\]. In the laboratory under free-running conditions, humans and other mammals typically undergo the rhythmic alternation between physiological and behavioral states appropriate for day (subjective day) and for night (subjective night) \[[@B9]\]. Although individual rhythmic outputs (e.g., locomotor activity, birdsong etc) might be observed to be bimodally expressed, more proximate measures of clock function (e.g., SCN clock gene expression, light sensitivity of the pacemaker, melatonin secretion, etc) tend to alternate unimodally with a period near 24 h \[[@B10]-[@B12]\]. One approach to shift-work involves rapid phase-shiftingof the entire circadian program to an optimal compromise between the conflicting demands of work and home schedules \[[@B13]\]. A re-entrainment strategy that manipulates only circadian phase, however, can be easily undone by the entraining effects of bright light exposure in transit to and from work and by exposure to sunlight during the day, both of which serve to reinforce normal diurnal entrainment \[[@B6],[@B14]\]. Additionally, even effective re-entrainment of circadian phase may have costs in terms of the resulting coincidence of subjective night with social demands during the day. An alternative solution to the problem of shift-work might involve focusing on change in circadian waveforminstead of phase. Under appropriate lighting conditions, hamsters and mice readily adopt stable, multi-modal entrainment patterns that may meet the requirements of shift-workers for separate intervals of alertness during both the night and the day \[[@B15]-[@B19]\]. These entrainment patterns, furthermore, do not require the complete avoidance of daytime exposure to sunlight. Specifically, rodents maintained in 24 h light:dark:light:dark cycles (LDLD) may entrain their circadian pacemakers so that nocturnal locomotor activity is programmed during each of the two daily scotophases 12 h apart (i.e., in anti-phase). A series of behavioral and physiological studies has indicated that this reorganization reflects a global splitting of the circadian pacemaker into two oscillatory components that cycle out of phase \[[@B20]\] (unpublished data). The net result is that, in each 24 h period, the subject goes through two short subjective nights interrupted by two short subjective days. This entrainable split activity pattern occurs in each of three rodent species examined, although the stimulus requirements for maintaining stable entrainment apparently differ by species \[[@B18]\]. To date, two factors have been shown to influence the incidence of rhythm splitting in LDLD cycles. First, animals that engage in robust wheel-running in response to scheduled daily exercise are more likely to split their activity patterns than are animals that remain inactive \[[@B16],[@B21],[@B22]\], and splitting often coincides with a cage change that provokes wheel-running activity \[[@B17]\]. Second, the presence of dim illumination (e.g., 0.004 -- 0.10 lux versus complete darkness) during all dark periods markedly increases the proportion of animals that splits in LDLD \[[@B17],[@B19]\]. These two effects may be related as animals are sometimes more active in dim light than in complete darkness, although this cannot wholly account for the facilitative effects of dim light on splitting \[[@B23]\]. In the present experiments, we assessed directly whether the feedback from a running wheel was necessary for the induction and maintenance of rhythm splitting. Additionally, we examined how split rhythms changed as a function of the relative length of day and night. Identification of factors that influence rhythm splitting should serve to clarify underlying mechanisms and inform the design of protocols examining the feasibility of similar entrainment in humans. Finally, body temperature (T~b~) telemetry allowed assessment of circadian markers that may be mediated by mechanisms separate from those underlying locomotor activity \[[@B24]-[@B27]\]. Results ======= Experiment \#1. Is a wheel necessary for maintenance and induction of splitting in Syrian hamsters? --------------------------------------------------------------------------------------------------- Transfer of Syrian hamsters from standard lighting conditions (14 h light, 10 h dark per day; LD14:10) to LDLD7:5:7:5 and subsequently to LDLD9:3:9:3 yielded the patterns of circadian re-entrainment depicted in Figure [1](#F1){ref-type="fig"}. In the first two weeks of LDLD7:5:7:5, elevations of general locomotor (GL) activity and T~b~reliably occurred during each of two daily scotophases for several animals (Fig. [1A, C, D](#F1){ref-type="fig"}). In some cases, these rhythms permanently (Fig. [1A](#F1){ref-type="fig"}) or transiently (Fig. [1C](#F1){ref-type="fig"}) fused into a unimodal (i.e., unsplit) pattern following immobilization of the running wheel. Other hamsters adopted bimodal T~b~and GL rhythms in LDLD7:5:7:5 only after immobilization of the running wheel (Fig. [1B](#F1){ref-type="fig"}). Still other hamsters first exhibited split rhythms under LDLD9:3:9:3. In all cases, split rhythms fused into the unsplit pattern under constant conditions (Fig. [1A--D](#F1){ref-type="fig"}). Table [1](#T1){ref-type="table"} indicates the number of animals meeting objective criteria for split and unsplit rhythms in each of the experimental intervals. There were no discrepancies between the two measures. Because of limited battery life of the telemeters, the number of reported animals decreased as the experiment progressed. Is the T~b~rhythm distinct from that of activity? ------------------------------------------------- Daily mean T~b~and amplitude of the circadian T~b~rhythm varied across the experimental intervals (Fig [2A, B](#F2){ref-type="fig"}; repeated measures ANOVA, p \< 0.001 for both). Mean T~b~, but not T~b~amplitude, increased following initial transfer to running wheels and LDLD7:5:7:5 (p \< 0.005; p \> 0.05, respectively). Immobilization of wheels 2 weeks later reduced T~b~amplitude (p \< 0.001), whereas the reduction in mean T~b~just failed to meet statistical significance after correction for multiple comparisons (p = 0.025). In no interval did mean T~b~or T~b~amplitude differ for split versus unsplit subjects (data not shown). For each individual animal, T~b~was positively correlated with levels of general activity (p \< 0.001 in every case; data not shown). Because of likely masking effects of activity on T~b~, we assessed whether there existed an endogenous T~b~rhythm independent of locomotor activity using linear regression to partial out the influence of GL on T~b~rhythms. The residual T~b~not accounted for by GL activity exhibited a reliable daily rhythm in individual-and group-analyzed hamsters during exposure to LDLD9:3:9:3 (Fig. [3](#F3){ref-type="fig"}). Among split hamsters, there were two unambiguous and equal-sized peaks and troughs daily in the endogenous Tb rhythm (Fig. [3A, B](#F3){ref-type="fig"}). For unsplit hamsters, there was a substantial rise only during the subjective night and much smaller elevation in antiphase (i.e., during the second scotophase; Fig. [3C, D](#F3){ref-type="fig"}). Experiment \#2: Do Siberian hamsters require wheels for splitting? ------------------------------------------------------------------ The transfer of Siberian hamsters from LD14:10 to LDLD7:5:7:5 and subsequently to LDLD cycles with progressively shorter scotophases (i.e., LDLD8:4:8:4 etc) induced in individual animals the same split and unsplit rhythms described in Experiment \#1 (Figs. [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"}). A majority of animals with wheels demonstrated split activity bouts for at least one photoperiod condition, but not before LDLD8:4:8:4 (Fig. [4A, B](#F4){ref-type="fig"}; Table [2](#T2){ref-type="table"}). Animals lacking running wheels and instead monitored by passive infrared (PIR) motion detectors did not differ statistically in the incidence of splitting as compared to the wheel group (Fisher\'s exact test, p \> 0.21 in all photoperiods). The number of animals demonstrating split activity tended to increase or remain the same as the length of the scotophases was shortened with the exception of PIR animals in LDLD11:1:11:1. Split rhythms were restored upon return to LDLD9:3:9:3, however. Of the animals that displayed split rhythms during the second exposure to LDLD 9:3:9:3 (n = 12), 6 maintained a split-activity rhythm when transferred to \"skeleton photoperiods\" (Fig. [4C](#F4){ref-type="fig"}). Activity of four of these split animals \"phase-jumped\" into the two longer (i.e., 7 h) of the four daily scotophases (Fig. [4B](#F4){ref-type="fig"}, [5C](#F5){ref-type="fig"}), and one animal that was previously unsplit met the criterion for splitting during the skeleton photoperiod. The activity duration (α) of each activity component of split animals varied with the length of the scotophase regardless of whether hamsters had wheels or not (Fig [6, B](#F6){ref-type="fig"}; p \< 0.001 for each). Phase angle of entrainment for the nighttime, but not the daytime component, was positively correlated with scotophase duration for animals with wheels (Fig. [6C](#F6){ref-type="fig"}). This measure was negatively correlated with scotophase duration in animals lacking wheels (Fig. [6D](#F6){ref-type="fig"}). For unsplit hamsters, the expected negative correlation between phase angle of entrainment and scotophase duration was obtained only in animals monitored with motion detectors (y = -0.26x + 1.29; r = 0.50, p \< 0.005; y = 0.06x -- 0.32; r = 0.11, p \> 0.45 for PIRs and wheels, respectively). Discussion ========== As reported in previous studies, a substantial fraction of each of two hamster species readily adopts bimodal wheel-running rhythms under 24 h LDLD cycles \[[@B18],[@B19],[@B22],[@B23]\]. Three aspects of the present studies validate the use of measures besides wheel-running to identify these split circadian entrainment patterns. In Experiment \#1, the simultaneous recording of T~b~, general locomotor activity and wheel-running yielded essentially identical rhythms within individual animals. Second, separate groups of Siberian hamsters in Experiment \#2 monitored by PIR motion detectors and by wheels generated very similar constellations of activity patterns (i.e., split, unsplit, transiently and delayed splits). Finally, the entrainment patterns identified here with motion detectors also mirror wheel-running patterns reported in other studies \[[@B17],[@B18]\]. The persistence of the split entrainment state in some hamsters exposed to skeleton photoperiods is particularly informative as it discounts the possibility that split activity patterns derive from a) a single long subjective night rendered bimodal by negative masking by light or b) a single short subjective night plus a second non-circadian component positively masked by the second dark period. The formal basis of this split pattern has been considered in detail elsewhere \[[@B15],[@B17]\]. The use of measures other than wheel running to identify split rhythms permits testing of the hypothesis that a running wheel is necessary to maintain and/or induce this entrainment pattern. In previous studies, animals that failed to run in a wheel during scheduled daily exposure to novel wheels also failed to later exhibit split home-cage running patterns \[[@B21],[@B22]\]. Additionally, among animals that did not split immediately in LDLD7:5:7:5, we have repeatedly observed abrupt splitting following a cage change during the daytime dark phase \[[@B17],[@B23]\], which typically induces robust wheel-running activity. The present results clearly establish that a wheel is not necessary for splitting in either species: in Experiment \#1, split rhythms in LDLD persisted in several Syrian hamsters following withdrawal of wheel access, and other hamsters later adopted split rhythms in the absence of a functional wheel when the length of each of the daily photophases was lengthened. Wheel immobilization was not necessarily responsible for the loss of splitting in some hamsters (see Table [1](#T1){ref-type="table"}) because rejoining has been observed in the first weeks of chronic wheel access in similar experiments \[[@B17]\]. In Experiment \#2, Siberian hamsters exhibited split rhythms with or without access to a wheel at rates that did not differ between groups. Thus, wheels are necessary neither for maintenance nor for induction of split rhythms. These experiments, however, were not designed with sufficient statistical power to rule out a minor influence of a running wheel on split rhythms. Whereas it was expected that general locomotion and wheel-running -- both measures of activity -- would correspond closely, we hypothesized that the T~b~rhythm might differ from activity rhythms if it was controlled by a distinct oscillatory mechanism as suggested in several, but not all, studies \[[@B24]-[@B27]\]. As noted above, T~b~rhythms were split in all animals with split activity patterns. Because wheel running likely elevates T~b~\[[@B28],[@B29]\], this bimodal pattern might be expected on the basis of masking alone, and indeed, in this study T~b~rhythms were acutely influenced by access to running wheels. The wheel-running masking account can be dismissed, however, by the persistence of split T~b~patterns after the wheels were immobilized as has been reported with constant light-induced splitting \[[@B30]\]. Highly significant correlations between T~b~and GL values raise the vexing possibility that general locomotor activity, as well as wheel running, can have masking effects \[[@B31]\]. Adjustment of the T~b~data to control for the correlation between T~b~and GL diminished the amplitude but did not eliminate the rhythmicity of the endogenous T~b~rhythm, suggesting that an activity-independent T~b~rhythm is split in the same fashion as that of activity. Of course, the estimate of the activity-independent T~b~rhythm is only as good as the statistical model relating T~b~and GL. Use of a number of different statistical models (e.g., log linear regression; polynomial regression; use of various lag times between T~b~and GL and GL integration times \[[@B32]\]), did not alter our basic conclusion. As illustrated in Fig. [1](#F1){ref-type="fig"} and in previous studies \[[@B17]\], splitting of Syrian hamster rhythms under LDLD7:5:7:5 is not always stable nor was activity always equally distributed between scotophases. In Experiment \#1, an increase in splitting incidence was noted after transfer to LDLD9:3:9:3, and no split animal ever rejoined under this photoperiod suggesting that this lighting condition may promote stable splitting. This finding accords with other of our unpublished data from wheel-running hamsters. The same trend was apparent in Siberian hamsters of Experiment \#2 where splitting was numerically most common under nights shorter than 5 h. Under the shortest nights, however, split rhythms largely disappeared and were later recovered in LDLD9:3:9:3. These observations are consistent with a model of splitting in which compression of subjective night reduces the stability of the unsplit pacemaker \[[@B15],[@B23]\], and thus a split anti-phase entrainment pattern is preferred. The phase-jumping of split rhythms of several Siberian hamsters in skeleton photoperiods further suggests that the individual split components are not fully stable under 3 h scotophases. Further work will be necessary to determine whether a circadian parameter related to \"stability\" can be operationalized and convincingly distinguished from known parameters such as period, phase and phase response to light. The multiple oscillations that underlie splitting under LDLD do not share the same physiological basis as those observed after prolonged exposure to constant light \[[@B33]\]. Whereas the latter case is associated with anti-phase oscillations of the left and right SCN, the former appears to exhibit no such laterality \[[@B34]\](Meyer-Bernstein, unpublished observations). Whether the two LDLD oscillations correspond to the as yet unidentified dual oscillators proposed to underlie photoperiodism, however, remains an open question. At a formal level of analysis, distinct \"evening\" and \"morning\" oscillators nicely account for the photoperiodic regulation of activity duration (α) of unsplit animals \[[@B12],[@B35]\]. Because similar photoperiodic regulation of α is seen for each component of the split rhythms in Experiment \#2, each of the two split components may be similarly generated by a complex pacemaker comprising multiple oscillators. Supporting this inference, the relationship between phase angle of entrainment and photoperiod for PIR data in Experiment \#2 parallels that of unsplit pacemakers \[[@B36]\], albeit non-significantly in the case of the daytime component. Thus, the circadian system likely comprises not just two functional oscillators, but many, which may couple to generate a variety of entrainment states. Because all mammalian circadian pacemakers likely derive from multiple oscillators in the SCN, we suspect that this split entrainment pattern could be induced in other species under permissive conditions. The practicality and benefit of this entrainment regime to human subjects await evaluation. At least superficially, it may better meet the needs of shift-workers than present phase-shift protocols that attempt to schedule sleep during the real day without achieving full-scale phase shifts of subjective night. With a split circadian rhythm, shift-workers might schedule sleep in two equal intervals sandwiched between a late shift and daytime activity. If the human and hamster pacemakers respond similarly, bright light exposure before and after each sleep phase would reinforce the split pattern rather than compromise entrainment as is the case in some phase-shift protocols. It also remains to be determined whether this circadian arrangement entails any physiological costs, and if so, whether these outweigh potential benefits. From a translational perspective, we are encouraged that split rhythms can be observed under a broader range of conditions than initially identified. Human splitting protocols should consider the influence of factors identified in animal studies: nocturnal illumination, photoperiod and activity levels. Future animal work will assess the influence of age, endocrine status and photoperiodic history. Conclusion ========== Entrainment of novel circadian waveforms is possible with the use of exotic lighting conditions and may represent a productive new strategy for chronotherapeutics. While prior studies point to a role of locomotor activity in inducing split activity patterns, access to running wheels is not necessary to induce or to maintain these entrainment patterns. Instead, split entrainment is influenced by the relative length of light and dark phases of the 24 h LDLD cycle. Methods ======= General ------- All procedures were approved by the UCSD Institutional Animal Care and Use Committee. Laboratory bred Syrian hamsters, Mesocricetus auratus(of original Harlan stock; HsdHan: AURA, Harlan, Indianapolis, IN), and Siberian hamsters, Phodopus sungorus(from a colony maintained at UCSD), were housed at 22 ± 2°C with ad libitum access to water and Purina chow (St. Louis, MO). From birth, Syrian and Siberian hamsters were maintained in 14 h light, 10 h dark daily (i.e., LD14:10; lights on 0300 PST) with photophase illumination of 100 -- 150 lux and with no scotophase illumination. Beginning with and continuing throughout the experiments, green light-emitting diodes (LEDs) generated a mean scotophase illumination of 0.027 ± 0.007 lux on the floor of the cage. Photophase illumination remained 100 -- 150 lux. Cage changes, which are potent circadian zeitgebers that facilitate rhythm splitting, always occurred during the first 60 min of the daytime scotophase at intervals of 1--2 weeks. These cage changes defined intervals for statistical analysis, which excluded any data collected 24 h after the perturbation. Experiment \#1 -------------- ### Surgery At 12 weeks of age, sterile radio-telemeters (Mini-Mitter, Bend, OR) were implanted in the abdominal cavity of Syrian hamsters (16 male/2 female) under Nembutal anesthesia, and hamsters were transferred to individual cylindrical polypropylene cages (20.5 cm diam). #### Photoperiod and wheel manipulations At 1000 PST on the tenth day post-surgery, animals were transferred to rectangular polypropylene cages (27 × 20 × 15 cm) equipped with running wheels (17 cm diam) with plastic interleaved through the rungs of these wheels to increase wheel running coordination. Transfer coincided with the onset of the first scotophase of a new LDLD7:5:7:5 cycle (lights off 1000, lights on 1500, lights off 2200, lights on 0300 PST). Animals were divided between two light-tight ventilated chambers with space for 12 cages each. After 2 weeks in LDLD7:5:7:5, running wheels were permanently immobilized at 0700 PST with plastic zip ties that bound rungs to the top of the cage. After 2 weeks without wheel access in LDLD7:5:7:5, the photoperiod was changed to LDLD9:3:9:3 (lights off 1100, lights on 1400, lights off 2300, lights on 0200 PST) at the beginning of the evening photophase (1500 PST). After four weeks, hamsters were exposed to constant dim illumination (0.027 ± 0.0067 lux) for two additional weeks beginning with the daytime scotophase. Experiment \#2 -------------- At either 21--24 or 41--44 weeks of age, group housed male Siberian hamsters (n = 22) were transferred at 1000 PST to individual cages coinciding with the beginning of a 5 h scotophase of a new LDLD7:5:7:5. Half of the animals, with equal representation of each age cohort, were housed in cylindrical cages equipped with plastic-wrapped running wheels (17 cm diam). The remaining hamsters were housed in rectangular polycarbonate cages (27 × 20 × 15 cm) without a wheel, but equipped instead with a passive infrared motion detector (PIR; Coral Plus, Visonic, Bloomfield, CT) to detect locomotor activity. After two weeks the LDLD cycle was changed to LDLD8:4:8:4 by reducing the duration of both scotophases by 30 min at each end. At weekly intervals each scotophase cycle was successively shortened by one hour (to LDLD9:3:9:3, LDLD10:2:10:2 and LDLD11:1:11:1). After one week of the latter photocycle, hamsters were maintained for an additional four weeks in LDLD9:3:9:3. Subsequently, each 9 h photophase was replaced with two 1-h \"skeleton pulses\" to yield LDLDLDLD1:7:1:3:1:7:1:3 for two weeks. Data collection --------------- In both experiments, activity and temperature data were collected with Dataquest III hardware (Mini-mitter, Bend, OR) configured for 6 min bins. Prior to implantation, telemeters were calibrated at 36 and 38.5°C. Wheel-running revolutions triggered mechanical sensors that recorded a single count every half rotation. PIR motion detectors registered activity whenever 3 of 27 zones were crossed. Rhythms were plotted and analyzed in ClockLab (Actimetrics, Evanston, IL) with supplementary analyses with Microsoft Excel. Statistical analyses were performed with Statview 5.0 (SAS Institute, Cary, NC). Data analysis ------------- ### Incidence of splitting Different criteria for splitting were required for the different rhythms measured. For wheel running, rhythms were classified as split in any given interval if there were \> 15 wheel revolutions for 3 consecutive 6 min bins in both daily scotophases for at least 3 days. As in past studies \[[@B18],[@B23]\], there was never any ambiguity about the split versus unsplit status of wheel-running records. For PIR rhythms and for telemetered GL and T~b~data, values were smoothed over 30 min bins and the dataset reduced to 48 values per 24 h period (i.e., every 30 min). For classification as split or unsplit during each analysis interval, a 24 h histogram was produced for each hamster by averaging 5--6 days of this reduced dataset. The rhythm was considered split if each histogram scotophase was associated with elevated activity or T~b~levels defined as follows: values exceeding the daily mean by more than one standard deviation for 2 consecutive 30 min bins. In nearly all cases, these objective determinations corresponded with subjective judgments of the rhythms as essentially unimodal (unsplit) or bimodal (split). ### Analyses specific to Experiment \#1 #### T~b~mean and amplitude In each analysis interval, mean T~b~values were determined by averaging all values over 5--6 days prior to a cage change. T~b~amplitude was calculated as the difference between the maximum and minimum values of 5--6 day histogram calculated from 30 min averages as described above. Repeated measures ANOVA was performed for the 12 animals with uninterrupted telemeter function. #### Activity-independent component of T~b~rhythm For each animal with complete data in the second 6-day analysis interval of LDLD9:3:9:3 (n = 8 split, n = 4 unsplit hamsters), 30 min T~b~values were regressed against 30 min GL. Residual values representing the component of T~b~rhythm not accounted for by GL values were retained and averaged at each 30 min time point over 6 days. Alternative models to account for GL yielded very similar results on sample animals. These models included regression of T~b~against log GL values using 30 min values; and use of various lag times and integration intervals for the independent variable \[[@B32]\]. ### Analyses specific to Experiment \#2 #### Phase angle of entrainment In each photoperiod, the phase angle of entrainment for PIR data in Experiment \#2 was determined from 5--6 day averages of the 30 min smoothed data. For each component of the split activity rhythm, or for the single component of the unsplit rhythm, activity onset was defined as the earliest point near the L/D transition that exceeded the daily mean. Phase angle was expressed in relation to this transition (positive values indicated activity anticipates lights off). #### Activity duration For each scotophase, activity duration (α) was taken as the time difference between the first and last points exceeding the daily activity mean. Both dependent measures (phase angle and activity duration) were regressed against the length of each scotophase for split and unsplit hamsters separately. Because the number and identity of split hamsters changed across the experiment, the dataset was not a proper repeated measures nor were data points fully independent. Recognizing that treating data points as independent would have the effect of reducing statistical power, we opted to do so in linear regression analysis. Abbreviations ============= GL -- telemetered general locomotion; LD -- 24 h light:dark cycle; LDLD -- 24 h light:dark:light:dark cycle; PIR -- passive infrared motion detector; SCN -- suprachiasmatic nucleus; SEM -- standard error of the mean; T~b~-- body temperature Authors\' contributions ======================= SLR and MMV collected and analyzed data for Experiments \#1 and 2, respectively and contributed substantially to their design. They contributed equally to this work and are listed alphabetically. JAEv provided significant guidance and assistance to SLR and MMV in data collection and analysis and contributed to both experimental designs. JAEl contributed to experimental design and oversaw the collection of telemetry data. MRG contributed to experimental design, supervised data analysis and worked with SLR and MMV in preparation of the manuscript. All authors edited earlier drafts, and all approved the final manuscript. Acknowledgements ================ The authors are grateful to Antonio Mora and Robert Lanuza for outstanding animal care. This research was funded by grants IBN-0346391 from NSF and NICHD-36460 and NS-30235 from NIH. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Representative double-plotted general locomotion (GL), wheel-running and body temperature (T~b~) rhythms of Syrian hamsters transferred in Experiment \#1 from LD14:10 to LDLD7:5:7:5, to LDLD9:3:9:3 and to constant dim illumination (DD). On the left of each panel is illustrated the GL for the entire experiment. On the right of each panel is shown, respectively, 7 days of wheel-running, T~b~over the same interval, T~b~in LDLD 7:5:7:5 and LDLD9:3:9:3. Lines connect enlarged records on the right with the corresponding GL dates on the left. Times of darkness are shaded blue; GL actograms are scaled as percentile values (ClockLab); wheel running actograms are scaled from 0 to 150 counts/min, and T~b~plots are scaled from 36.0 to 40.0°C. See text for characterization of rhythms of representative animals. ::: ![](1471-2202-6-41-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Mean (± SEM) (A) and amplitude (± SEM) (B) of the T~b~rhythm over each of the intervals throughout the experiment. Illustrated are data from 12 hamsters with continuously functioning telemeters, analyzed with repeated measures ANOVA. In some cases (LDLD7:5:7:5 with wheels; LDLD9:3:9:3 without wheels) conditions persisted for several weeks allowing values to be calculated over two intervals separated by cage changes. Means of group pairs below lines with P values differ significantly by paired t-test after correction for multiple comparisons. ::: ![](1471-2202-6-41-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Estimates of the activity-independent T~b~rhythms in individual (A, C) and groups (B, D) of split (A, B) and unsplit (C, D) hamsters maintained in LDLD9:3:9:3. Error bars for individual hamsters represent SEM of 6 values (days) at each time point. Error bars for groups represent between-subjects SEM. Estimates were derived by regressing T~b~against GL and plotting residual values. In (B) and (D) the T~b~rhythm (not removing influence of activity) is plotted (gray lines) in relation to the residual values. ::: ![](1471-2202-6-41-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Representative double-plotted wheel-running actograms of Siberian hamsters transferred in Experiment \#2 from LD14:10 to LDLD7:5:7:5 and subsequently to other photocycles as noted on the left margins of actograms. Other conventions as in Figure 1. Animals in (A) and (B) demonstrated split behavior very early on in the experiment and remained split until skeleton photoperiods were introduced. The subject in (C) showed no signs of splitting until a second exposure to LDLD 9:3:9:3 and then maintained the split entrainment pattern under skeleton photoperiods. The hamster in (D) exhibited no signs of splitting at any point during the experiment. ::: ![](1471-2202-6-41-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Representative double-plotted actograms (left) and 6-day average activity profiles (± SEM) (right) of hamsters lacking running wheels and monitored by motion detectors. Conventions as in Figure 1. Rhythms of animals in (A) and (B) split early in the experiment but did not remain split under skeleton photoperiods. The hamster in (C) expressed a split rhythm only in LDLD9:3:9:3. ::: ![](1471-2202-6-41-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Scatterplots of activity duration (A, B) and phase angle of entrainment (C, D) as a function of entraining photocycle in split hamsters with running wheels (A, C) or without (B, D). Least squares regression lines for the daytime (open circles; dashed lines) and nighttime (filled triangles; solid lines) were calculated separately. ::: ![](1471-2202-6-41-6) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Fraction of Syrian Hamsters Displaying Split Rhythms in Experiment \#1† ::: LD 14:10 No Wheels LDLD 7:5:7:5 Wheels (1) LDLD 7:5:7:5 Wheels (2) LDLD 7:5:7:5 No Wheels LDLD 9:3:9:3 No Wheels (1) LDLD 9:3:9:3 No Wheels (2) ------------------------- -------------------- ------------------------- ------------------------- ------------------------ ---------------------------- ---------------------------- General Locomotion (GL) 0 1/17 13/18 6/18 8/17 10/14 Body Temperature (T~b~) 0 1/17 13/18 6/18 7/16 10/14 †Sample size varies with time due to telemeter malfunction; numbers in parentheses indicate 1st and 2nd intervals of a given photoperiod condition. ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Fraction of Siberian Hamsters Displaying Split Rhythms in Experiment \#2 ::: LDLD 7:5:7:5 LDLD 8:4:8:4 LDLD 9:3:9:3 LDLD 10:2:10:2 LDLD 11:1:11:1 LDLD 9:3:9:3 Skeletons† ------------------------- -------------- -------------- -------------- ---------------- ---------------- -------------- ------------ Wheels 0/11 1/11 2/11 4/11 4/11 6/11 5/11 Motion Detectors (PIRs) 3/11 3/11 5/11 5/11 1/11 6/11 2/11 †Skeleton photoperiods of LDLD9:3:9:3 (i.e., LDLDLDLD1:7:1:3:1:7:1:3) :::	PubMed Central	2024-06-05T03:55:59.489431	2005-6-20	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180451/", "journal": "BMC Neurosci. 2005 Jun 20; 6:41", "authors": [ { "first": "Sheila L", "last": "Rosenthal" }, { "first": "Martin M", "last": "Vakili" }, { "first": "Jennifer A", "last": "Evans" }, { "first": "Jeffrey A", "last": "Elliott" }, { "first": "Michael R", "last": "Gorman" } ] }
PMC1180452	Background ========== Neurotrophins have been considered to play roles in the differentiation, neurite outgrowth and survival of a certain group of neurons \[[@B1]-[@B4]\]. In addition to these well-known functions, most neurotrophins are involved in rapid changes in the function of neural circuits \[[@B5],[@B6]\]. In particular, brain-derived neurotrophic factor (BDNF) plays a role in activity-dependent changes in synaptic function \[[@B7]-[@B9]\]. To serve such a broad-ranging function, BDNF produced in the nucleus of neurons is sorted into a regulated secretory pathway through the trans-Golgi network, and transported to release sites in neurites \[[@B10],[@B11]\]. In axons it is suggested that BDNF is transported through the fast axonal flow and then released and transferred to postsynaptic neurons in an activity-dependent manner \[[@B10]-[@B15]\]. In dendrites or dendrite-like neurites also, the targeting of BDNF to distal parts and its release were suggested to occur in an activity-dependent manner \[[@B16]-[@B22]\]. Since these previous studies were carried out using immunohistochemical and/or in situhybridization technique after the fixation of neurons, or by observing the decrease in fluorescence intensity of neurons expressing BDNF tagged with green fluorescent protein (GFP), the actual dynamics of BDNF trafficking was not analyzed in dendrites. Thus, a question of whether the trafficking of BDNF is different between axons and dendrites of neurons is not answered yet. An answer to this question will give a clue that may resolve the controversial issue of whether BDNF acts retrogradely through dendrites on presynaptic neurons or anterogradely through axons on postsynaptic neurons. To address this question, it is desirable to perform a real-time analysis of movements of BDNF tagged with GFP in both axons and dendrites of living neurons. However, such an analysis of BDNF trafficking has not been successfully carried out except for two recent studies on its retrograde transport in dorsal root ganglion neurons \[[@B23]\] and bidirectional transport in cortical cell neurites \[[@B24]\]. However, the observation was restricted to axons in the former study, and axons and dendrites were not distinguished in the latter study. In the present study we carried out a real-time analysis of movements of BDNF tagged with GFP in both axons and dendrites of living cortical neurons using the method of direct injection of their plasmid cDNAs into the nucleus \[[@B14]\]. We found that most of BDNF-GFP moves smoothly in the anterograde direction in axons while it does not move in dendrites in most cases. Even if it moves in dendrites, its velocity is slower than that in axons. Parts of the present results were published in an abstract form \[[@B25]\]. Results ======= Plasmids encoding BDNF tagged with GFP or CFP at the COOH-terminus were injected into the nucleus of cultured cortical neurons through a micropipette under visual control, as reported previously \[[@B14]\]. GFP- or CFP-tagged BDNF resulting from these plasmids were confirmed to be biologically active and mimic the releasing properties of untagged BDNF \[[@B20]\]. In the present experiments, 20--30% of the neurons into which plasmids had been injected expressed fluorescent signals, as reported previously \[[@B14]\]. The expression of signals was already detected 16 h after the injection, but recordings were usually carried out about 24 h after the injection. In about two-thirds of the neurons that expressed fluorescent signals of BDNF-GFP, the signals were punctated in clusters along dendrites and axons (Fig. [1A](#F1){ref-type="fig"}). The soma and bases of dendritic trunks appeared to be filled with fluorescent signals because of the very dense expression of BDNF-GFP. To see whether the distribution of BDNF-GFP signal was similar to that of endogenous BDNF, neurons cultured in companion dishes (sister-cultured neurons) were stained immunocytochemically with antibody to BDNF. The distribution of endogenous BDNF was also punctated in neurites and diffuse in the soma (Fig. [1B](#F1){ref-type="fig"}), as reported previously (14, 19, 20). This distribution pattern was similar to that of granular BDNF-GFP, suggesting that BDNF-GFP was expressed in plasmid-injected neurons in a similar way to endogenous BDNF if the expression in neurites was granular. In about one-third of the neurons that expressed fluorescence, the signals were diffuse and smear-like even in dendrites. These neurons were not included in the analysis of BDNF-GFP trafficking. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Similar distribution of BDNF-GFP and endogenous BDNF, and expression of BDNF-GFP in GAD65-negative neurons and its localization in axons and somatodendritic regions. A, BDNF-GFP expressed in a neuron 24 h after plasmid cDNA encoding BDNF-GFP was injected into the nucleus. Scale bar in A and B indicates 50 μm B, Immunocytochemical image of another, non-injected neuron stained with anti-BDNF antibody. C, Image of neurons expressing and not-expressing BDNF-GFP. Scale bar indicates 50 μm and applies to D and E. D, Image of the neurons shown in C, immunocytochemically stained with anti-GAD65 antibody. E, Superimposed images of C and D. F, Distribution of BDNF-GFP expressed in another neuron. Arrowheads indicate a MAP2-negative neurite. Scale bar indicates 50 μm and applies to G and H. G, Image of the same neuron as shown in F, immunocytochemically stained with anti-MAP2 antibody. H, Superimposed images of F and G. I, Magnified image of the boxed area in H. Vesicular puncta of BDNF-GFP in the axon are seen more clearly than those in H. ::: ![](1471-2202-6-42-1) ::: In the present study we attempted to inject plasmids into pyramidal cell-like neurons which are excitatory. To confirm that neurons expressing fluorescent signals were excitatory, they were stained immunocytochemically with anti-glutamic acid decarboxylase 65 (GAD65) antibody. As shown in Figure [1C](#F1){ref-type="fig"} and [1D](#F1){ref-type="fig"}, a neuron that expressed BDNF-GFP was not stained with the antibody. This was confirmed by the superposition of the two images (Fig. [1E](#F1){ref-type="fig"}). This result suggests that the BDNF-GFP-injected cell was an excitatory neuron. All of the 12 BDNF-GFP-injected neurons that were immunocytochemically tested were negative to anti-GAD65 antibody, indicating that the neurons analyzed in the present study were most likely excitatory. We also stained neurons by immunocytochemistry using anti-microtubule-associated protein 2 (MAP2) antibody to differentiate axons from dendrites. As shown in Figure [1F--I](#F1){ref-type="fig"}, we could identify axons on the basis of their negativity to anti-MAP2 antibody (arrowheads in Fig. [1F, H](#F1){ref-type="fig"} and [1I](#F1){ref-type="fig"}). Fluorescent signals were detected also in MAP2-negative neurites (Fig. [1H](#F1){ref-type="fig"} and [1I](#F1){ref-type="fig"}), indicating that BDNF-GFP was expressed in axons as well as somatodendritic regions. In part of the experiments we injected plasmids encoding NGF-GFP to see whether BDNF-GFP is expressed in a manner specific to this neurotrophin. The granular expression of NGF-GFP was seen in a minor group of the cells that expressed fluorescent signal (see Fig. [2E](#F2){ref-type="fig"}). In most of the cells NGF-GFP was expressed in a smear-like manner even in dendrites. To exclude a possibility that such a difference in the expression pattern between the two neurotrophins tagged with GFP might reflect cell-to-cell variability, we simultaneously injected a combination of plasmids, BDNF-CFP and NGF-YFP into the nucleus of a single neuron. For this purpose CFP or YFP was used in place of GFP, because the fluorescent wavelength of GFP overlaps with that of CFP or YFP, but those of CFP and YFP do not substantially overlap with each other. As in the case of single injection, BDNF-CFP was expressed mostly in a granular manner while NGF-YFP was expressed in a smear-like manner in the same cell. An example for this finding is shown in Figure [2A--C](#F2){ref-type="fig"}. Many puncta of BDNF-CFP were seen along neurites while NGF-YFP was seen in a smear-like manner in the neurites of the same cell. Such a difference in expression pattern between the two neurotrophins was evident in the superimposed figure (Fig. [2C](#F2){ref-type="fig"}). The results obtained with simultaneous expression of the two neurotrophins are summarized in Figure [2D](#F2){ref-type="fig"}. As seen in this graph, about 60% of the cells expressed CFP-tagged BDNF in a vesicular manner while only about 20% of the cells expressed YFP-tagged NGF in such a manner. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Simultaneous expression of BDNF-CFP and NGF-YFP in the same neuron and proportion of neurons with granular expression of fluorescent puncta. A, Image of a neuron expressing BDNF-CFP. Scale bar indicates 50 μm and applies to B and C. B, Image of the same neuron as in A, expressing NGF-YFP. C, Superimposed images of A and B. D, Proportion of neurons with granular expression of fluorescent puncta. Numerators at the right end of bar indicate the number of neurons showing granular expression of each neurotrophin tagged with CFP or YFP, and denominators indicate the total number of cells which expressed both fluorescent signals. E, Mean proportions of neurons with granular expression of fluorescent puncta, obtained from all the data including those obtained with the single injection of plasmid cDNAs. Numbers at the right end of bar indicate the number of injection trials, in which 4--10 neurons expressed fluorescent signals. Horizontal lines at the end of bars indicate 2SEMs. ::: ![](1471-2202-6-42-2) ::: As mentioned before, the single injection of plasmid cDNAs encoding GFP-, CFP- or YFP-tagged neurotrophins also showed the difference in the expression pattern between BDNF and NGF (Fig. [2E](#F2){ref-type="fig"}). In this analysis the proportion of cells with vesicular expression was calculated for each session of injection trials and then the mean values were calculated from 21 and 20 sessions of trials for fluorescent protein-tagged BDNF and NGF, respectively. The means ± SEMs were 71.2 ± 1.5 and 29.2 ± 5.0%, respectively. The difference between these two values was statistically significant (P \<0.001, unpaired t-test). These results indicate that the granular expression of BDNF-GFP was not common to other neurotrophins, although we did not observe the expression of fluorescent protein-tagged neurotrophin-3 and neurotrophin-4/5 in the present study. We then found that most of the fluorescent puncta were moving in neurites of living neurons in the anterograde as well as retrograde directions (see Additional file [1](#S1){ref-type="supplementary-material"}). The movements of puncta were observed in about 80% of the neurons that showed a vesicular expression of BDNF-GFP. Neurons in which moving puncta were not detected were excluded from the analysis of BDNF-GFP trafficking. A detailed observation indicated that most of the puncta in axons moved smoothly in the anterograde direction, while most of the puncta in dendrites did not move or moved back and forth within a small distance in a fluttering manner. Figures [3B](#F3){ref-type="fig"} and [3C](#F3){ref-type="fig"} show an example of such observations. In an axon of a neuron shown in Figure [3B](#F3){ref-type="fig"}, a bright vesicle was observed to move in the anterograde direction at an almost constant velocity (punctum c) and another vesicle (punctum b) started to move about 10 s later (see also time-lapse pictures in Additional file [1](#S1){ref-type="supplementary-material"}). A small number of vesicles did not move during the observation time (punctum a). In dendrites, a substantial number of vesicles did not move (punctum c of Fig. [3C](#F3){ref-type="fig"}), but some puncta suddenly moved in the distal direction and then moved back to the original point (punctum a). We also found that a few puncta in dendrites moved initially, and then suddenly stopped and stayed in the same position during the observation time, as shown in Figure [3C](#F3){ref-type="fig"} (punctum b). These observations were carried out up to about 200 μm from the soma. Within this distance we did not detect a notable regional difference in the trafficking of puncta in dendrites nor in axons. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Movements of BDNF-GFP puncta in an axon and dendrites of a living neuron. A, Image of a cortical neuron which expressed BDNF-GFP in soma, axon and dendrites. Time-lapse images in each square along an axon and a dendrite are magnified and shown in B and C, respectively. Scale bar, 50 μm. B, Moving (b and c) and non-moving (a) puncta in the axon. Each image was taken at a time point indicated at the top of each figure. Right is the distal side of the axon. Scale bar in the bottom image indicates 5 μm and applies to all the images in B and C. C, Moving (a and b) and non-moving (c) puncta in the dendrite. Right is the distal side of the dendrite. Each image was taken at a time point indicated at the top of each figure. ::: ![](1471-2202-6-42-3) ::: To quantify these observations we classified fluorescence-positive puncta into four groups, i.e., puncta moving in the distal direction over a distance of 10 μm, those moving in the proximal direction over 10 μm, those not moving during the observation time of at least 15 min, and those fluttering within 10 μm. As shown in Figure [4A](#F4){ref-type="fig"}, 59% of the puncta in axons moved in the distal direction toward the terminals and only 23% of the puncta in axons moved in the proximal direction. On the other hand, 64% of the puncta in dendrites did not move at all or only fluttered within the short distance. Nevertheless, the rest of them in dendrites were observed to move in the distal or proximal direction. We then calculated the velocity of moving puncta in axons and dendrites, and found that the velocity was faster in axons on average than in dendrites. Figure [4B](#F4){ref-type="fig"} shows the distribution histograms of the velocity of moving puncta in axons or dendrites. The mean velocity of moving puncta in the distal (anterograde) direction was not markedly different from that in the proximal (retrograde) direction. Therefore, movements in both directions were combined in these histograms. The mean velocity in axons was 0.73 ± 0.26 (SD) μm/s, while that in dendrites was 0.47 ± 0.23 μm/s. The difference between these two distributions is evident in the cumulative probability curves of the values (Fig. [4C](#F4){ref-type="fig"}). The difference between the two curves was statistically significant (Kolmogorov-Smirnov test, P \< 0.001). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Properties of movement of BDNF-GFP puncta in axons and dendrites. A, A proportion of puncta which move in axons (filled columns) or dendrites (open columns) in the distal and proximal directions, do not move, and flutter within a small distance (10 μm). The numbers of puncta observed are given in parentheses. B, Distribution of moving puncta with different velocities. The bin width of the abscissa was 0.1 μm/s, and 0 indicates that the velocity was slower than 0.1 μm/s. The total numbers of vesicles analyzed were 132 in axons and 349 in dendrites. C, Cumulative plots of velocity distributions shown in the histograms of B. Dotted and continuous lines represent dendrites and axons, respectively. ::: ![](1471-2202-6-42-4) ::: Discussion ========== In the live-cell imaging of GFP- or CFP-tagged BDNF, we observed that most of fluorescent protein-tagged BDNF which were expressed as vesicular puncta in axons moved rapidly in the anterograde direction, as reported previously \[[@B14]\]. It is possible to argue that because of the direct injection of the plasmids into the nucleus of neurons, the anterograde movements of BDNF-GFP from the soma were dominant. This possibility seems unlikely, however, because the movements of BDNF-GFP vesicles in axons were already observed 16 h after the injection and the vesicles were calculated to move over a distance of 20 mm by the observation time (24 h after the injection) assuming that the mean velocity of 0.73 μm/s was maintained. Thus, BDNF-GFP is expected to have reached terminals of neurites long before the observation time. Indeed we observed that a substantial proportion of vesicles were moving in the retrograde direction in axons and dendrites, as shown in Figure [4A](#F4){ref-type="fig"}. BDNF-GFP vesicles were expressed in dendrites as well as in axons. The number of expressed vesicles in dendrites was much larger than that in axons. This difference seems to reflect the difference in areas between dendrites and axons. Thus, we did not find a strongly polarized expression of BDNF-GFP in either structural domain of neurons. In dendrites, however, the trafficking pattern of BDNF-GFP was markedly different from that in axons. Most of the BDNF-GFP vesicles in dendrites did not move at all or just fluttered within a small distance. Only a small number of puncta in dendrites moved continuously, while most of the puncta in axons moved smoothly over a long distance. There is a possibility that the directed movement of the fluorescent vesicles was not seen in dendrites, because BDNF-GFP had already arrived at dendritic sites where it was packed for release machinery. This possibility seems unlikely, since we did not observe a higher mobility of the puncta in dendrites at 16 h after the injection than at the standard observation time (24 h after the injection). We further found that the velocity of transport of a minor group of BDNF-GFP puncta in dendrites was slower than that of BDNF-GFP puncta in axons. The mean velocity in axons (0.73 ± 0.26 μm/s) is comparable to the reported values for the anterograde transport of a synaptic vesicle protein, synaptophysin tagged with GFP, in the axons of mouse dorsal root ganglion cells (0.69 ± 0.33 μm/s, \[[@B26]\]) and of ^125^I-labelled BDNF in the optic nerve of neonatal rats (0.47--0.92 μm/s, \[[@B15]\]). The present observations that the velocity and the pattern of movements of BDNF-GFP in dendrites were different from those in axons seem consistent with the previous results that dendrites have microtubules of mixed polarity orientation, while axonal microtubules have a uniform polarity orientation, with their plus ends toward the axonal terminal \[[@B27]\] and that motor proteins, microtubule-based kinesin superfamily proteins (KIFs), are differentiated in axons and dendrites \[[@B28]\]. The question of what kinds of KIFs or other motor proteins are responsible for the difference in the pattern of trafficking of BDNF-GFP between axons and dendrites will be clarified in future studies. In the present study using the plasmid cDNA injection method, there is a possibility that BDNF-GFP was expressed at much higher level than that of endogenous BDNF, and thus the sorting and trafficking characteristics would be different from those of endogenous BDNF. Although we cannot completely exclude this possibility, we believe that it is unlikely, based on the two observations. 1) The intracellular localization of expressed BDNF-GFP is similar to that of endogenous BDNF when BDNF-GFP is expressed in a punctated manner. 2) As discussed above, the mean velocity of trafficking of BDNF-GFP in axons is similar to the reported value for the anterograde transport of ^125^I-labelled BDNF in the optic nerve of neonatal rats. This suggests that the trafficking properties of BDNF-GFP are not significantly different from those of endogenous BDNF. Furthermore we observed that the expression pattern of BDNF-GFP or -CFP was different from that of NGF-GFP or -YFP in most cases. These results altogether suggest that the expression and trafficking of BDNF-GFP may not be substantially different from those of endogenous BDNF. This further suggests that the GFP tag may not notably affect sorting and trafficking of secretory proteins, as reported previously \[[@B26]\]. In sum, the present finding that the anterograde trafficking of BDNF-GFP in axons was dominant in cortical neurons seems consistent with the view that BDNF acts on postsynaptic neurons in the anterograde direction \[[@B13]-[@B15],[@B29]-[@B35]\]. Conclusion ========== The present results show that the pattern and velocity of the trafficking of BDNF are different between the two structural domains of neurons, axons and dendrites, and suggest that the anterograde transport in axons may be the dominant stream of BDNF to release sites. Methods ======= Culture preparation of neurons ------------------------------ Neonatal mice (C57/BL6, postnatal day 0--1) were anesthetized with ketamine (\> 30 mg/kg, i.p.), and then killed by cervical dislocation. The experimental procedures met the regulation of the Animal Care Committee of Osaka University Graduate School of Medicine. Neurons were cultured at a low density using conventional methods, as reported previously \[[@B14]\]. A piece of the visual cortex was removed from neonatal mice, enzymatically dissociated with papain (20 U/ml), and triturated with a fire-polished glass pipette. Neurons were plated on a previously prepared glial feeder layer, and were grown in a solution based on the Neurobasal A medium (GIBCO, Rockville, MD, USA) supplemented with 5% B27 (GIBCO). All experiments were carried out 14--24 days after plating. Injection of plasmid cDNA of BDNF or NGF tagged with fluorescent proteins ------------------------------------------------------------------------- The cDNAs of mouse BDNF or NGF tagged with GFP, CFP or YFP at the C-terminus were provided by Dr. Masami Kojima. Glass micropipettes were filled with Tris-ethylene-diamine-tetraacetic acid (EDTA) buffer (pH 8.0) which contained cDNAs of BDNF-GFP, or -CFP (0.5--1 μg/μl), or NGF-GFP, or -YFP (0.5 μg/μl). Then cultured neurons were placed on the stage of an inverted epifluorescence microscope (TE300, Nikon, Tokyo, Japan), and cDNAs were injected into the nucleus of a neuron through a micropipette under visual control, using a micromanipulator (MMO-202ND, Narishige, Tokyo, Japan), as described previously \[[@B14]\]. Live-cell imaging ----------------- Neurons which expressed fluorescent protein-tagged BDNF or NGF signals were observed using a 40 X /1.3 NA oil immersion objective (Plan Flour, Nikon) attached to the inverted epifluorescence microscope. The fluorescence of GFP. CFP and YFP excited by light at the wavelength of 488, 434 and 514 nm was measured using a cooled CCD camera (ORCA-ER, Hamamatsu Photonics, Hamamatsu, Japan). Time-lapse recordings were carried out about 24 h after the transfection, and sequential images were acquired using a cooled CCD camera at about 30°C. The exposure time was 1 or 2 s, the lapse of time was about 5 s, and the total time of recordings was 15--20 min. The trafficking of BDNF was analyzed by tracking the positions of GFP fluorescence-positive vesicles in neurites as a function of time. The analyses and animations of the data were made using the AQUA COSMOS software (Hamamatsu Photonics). Immunocytochemistry ------------------- For immunocytochemical staining, neurons were fixed with 4% paraformaldehyde (Sigma, St. Louis, MO, USA) and 4% sucrose in Dulbecco\'s phosphate-buffered saline (PBS) for 20 min at room temperature. The cells were incubated with PBS containing 0.2% Triton-X (Sigma) for 5 min and blocked by 10% goat serum in PBS for 1 h at 37°. Then, anti-MAP2 monoclonal antibody (isotype: IgG1, 1:250, Sigma), anti-BDNF polyclonal antibody (2 μg/μl, provided by Dr. Ritsuko Katoh-Semba), and anti-GAD65 monoclonal antibody (isotype IgG2a, 1:100, Chemicon, Temecula, CA, USA) were applied overnight at 4°. Endogenous BDNF was visualized by anti-rabbit secondary antibody conjugated with Alexa 546 (1:2000, Molecular Probe). MAP2 and GAD65 were visualized by anti-mouse secondary antibody conjugated with Alexa 647 or 350 (1:200, Chemicon), and by isotype-specific secondary antibody conjugated with Alexa 546 (1: 2000, Chemicon). Abbreviations ============= BDNF, brain-derived neurotrophic factor CFP, cyan fluorescent protein GAD65, glutamic acid decaroxylase 65 GFP, green fluorescent protein KIF, kinesin superfamily protein MAP2, microtubule-associated protein 2 NGF, nerve growth factor PBS, phosphate-buffered saline YFP, yellow fluorescent protein Authors\' contributions ======================= NA designed and carried out most of the experiments. KH helped with the design and analysis of experiments. TT planned the experiments and coordinated the work. Supplementary Material ====================== ::: {.caption} ###### Additional File 1 Supplementary movie of BDNF-GFP puncta in an axon and dendrites of a cortical neuron. An arrow at rest indicates an example of non-moving and fluttering puncta in dendrites, and another moving arrow that appeared 6 min after starting the movie indicates an example of fast-moving puncta in axons. ::: ::: {.caption} ###### Click here for file ::: Acknowledgements ================ We would like to thank Dr. Masami Kojima for kindly providing the cDNA of BDNF-GFP, BDNF-CFP, NGF-GFP and NGF-YFP, and Mr. Atsushi Maruyama for assisting in the preparations of cultured neurons. This study was supported in part by a Grant-in-Aid for Scientific Research on Priority Areas (12031230) to TT from the Ministry of Education, Science, Sports and Culture, Japan.	PubMed Central	2024-06-05T03:55:59.492492	2005-6-21	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180452/", "journal": "BMC Neurosci. 2005 Jun 21; 6:42", "authors": [ { "first": "Naoki", "last": "Adachi" }, { "first": "Keigo", "last": "Kohara" }, { "first": "Tadaharu", "last": "Tsumoto" } ] }
PMC1180453	Background ========== The combined use of morphological and functional information has been shown to improve diagnostic accuracy in many clinical disciplines \[[@B1]-[@B3]\]. In the management of patients with malignancies, coronary heart disease and diseases of the brain, fusion of different image modalities, such as CT, MRI, PET and various other nuclear diagnostic tests, is widely used \[[@B3]-[@B7]\]. Retinal diseases are also very complex, and often more than one diagnostic or imaging technique is used to assist in making a diagnosis. Cunha-Vaz and co-workers \[[@B8],[@B9]\] have shown the value of multimodal mapping systems for the macula, incorporating retinal imaging techniques such as confocal scanning laser ophthalmoscopy (SLO), the retinal leakage analyzer, the retinal thickness analyzer, automated perimetry and SLO angiography. However, inherent difficulties exist in the ability to correlate these various modalities, as these techniques lack common, precise reference points. Recently, a new modality in ocular imaging was introduced combining high depth resolution OCT with high transversal resolution confocal ophthalmoscopy \[[@B10]-[@B13]\]. The prototype OCT-Ophthalmoscope simultaneously produces en-faceOCT scans and pixel-to-pixel corresponding confocal images, so called OCT C-scans. The OCT C-scans are oriented in a transversal plane, perpendicular to the optical axis of the eye. A stack of such OCT C-scans allows for a reconstruction of the retina in three dimensions. Although this transversal plane is the more familiar plane when viewing the retina, compared to the conventional, longitudinal OCT images \[[@B14]-[@B16]\], single en-faceOCT scans may look confusing to the uninitiated, due to their high depth resolution (about 10 micron) and their transversal orientation through the retina \[[@B12],[@B17]\]. The confocal channel creates a fundoscopic image with a high transversal resolution but a low depth resolution (\~300 μm). There is seemingly little change in the confocal image when scanning through the retina along the Z-axis (depth). Therefore the confocal part of the OCT C-scan provides a high quality fundus image throughout the whole OCT stack in depth. As both the confocal and OCT image are acquired simultaneously with the same light source and at the same scanning rate, there is pixel-to-pixel correspondence between the OCT and the confocal image of the OCT C-scan \[[@B11]\]. The confocal image is used for general orientation and localization, whereas the OCT image shows detailed morphology, and subsequent pathological changes within the retina \[[@B10]\]. The high quality confocal image makes it a reliable reference image to overlay the OCT C-scans onto images generated with other diagnostic techniques acquired in the transversal plane. For example, with appropriate software to implement an affine transformation, conventional angiographic images can be spatially transformed and then superimposed over the confocal image. Given the pixel-to-pixel correspondence between the confocal and OCT channel, the converted angiographic image can be directly superimposed over the OCT image as well. The fusion of the more familiar angiographic information with the highly detailed morphological information from the OCT C-scan could improve our knowledge of retinal diseases as visualized by en-faceOCT. In this paper, we describe this novel procedure, and assess its ability to enhance our understanding of both fluorescein angiography and OCT C-scans through the combination of structural and functional information. Methods ======= At the Department of Ophthalmology of the Academic Medical Centre (University of Amsterdam, Amsterdam, The Netherlands) a prototype OCT-Ophthalmoscope (Ophthalmic Technologies Inc., Toronto, Canada) was used to evaluate patients with various retinal pathologies. Patients undergoing routine fluorescein angiography (FA), using Imagenet (Topcon, USA), were scanned with the OCT-Ophthalmoscope in the area of interest on the same day. To demonstrate the possibilities of this new technique, we selected some illustrative, though not necessarily common, pathological cases. The configuration of this OCT prototype has been described previously \[[@B10],[@B13]\]. In short, the system uses a super luminescent diode, with a central wavelength of 820 nm (20 nm bandwidth). The light beam is split, directing one part to the patient\'s eye (sample arm) and the other part to a reference arm (mirror). The returning light beams from both the patient\'s eye and the reference arm are collected through an interferometer to produce the OCT signal \[[@B11]\]. A fraction of the light returning from the patients\' eye is also directed towards another detector to produce a confocal signal \[[@B11]\]. The OCT-Ophthalmoscope produces a transversal OCT C-scan (size: 1042 × 512 pixels), currently at 2 frames/second, in the X-Y-plane at a fixed Z-coordinate, in which the OCT and confocal images are in pixel-to-pixel correspondence with each other \[[@B17]\]. ImagePro is an image-processing program working under Windows^©^. By using a custom made plug-in for this program, it is possible to spatially transform images, and to superimpose them over reference images. For the purpose of this study, the confocal part (size: 512 × 512 pixels) of the OCT C-scan was used as a reference image. Since every transformation inherently leads to loss of anatomical resolution in the transformed image, we chose to leave the OCT C-scan intact. The OCT C-scan covers a smaller area than the FA. Detailed information in the OCT scan would be lost due to a necessary volume transformation if the OCT C-scan would be transformed onto the FA. One or two OCT C-scans were chosen from the stack, depending on the depth at which the pathologies were best visible. The appropriate angiographic Imagenet frame was transformed onto, and superimposed over the confocal image. Alignment was then achieved by using evident landmarks as reference points. Inherent to the scanning procedure of the OCT-Ophthalmoscope, using a curved fan ray equivalent to the curvature of the eye, there is some distortion at the borders of both parts of OCT C-scan relative to the centre of the images. We found that it was best to choose reference points centrally within the vascular arcades. Four to six pairs of reference points were manually selected. This was done by zooming in on the area in which a landmark was localised. Landmarks were chosen at vascular crossings or vessel bifurcations that were clearly visible in both image modalities. These reference points were used to compute a linear transformation by the least-square method, compensating for translation, rotation, scaling, and shearing (full affine). The spatially transformed angiographic image was doubled and then superimposed over both the confocal image and the OCT scan. To further facilitate visualization, the images were combined into the different layers of an RGB image: the green channel of the RGB image was used for the spatially transformed angiographic image, and the other channels were used for both parts of the OCT C-scan. Images presented in this paper show on the left-hand side the confocal image, and on the right-hand side the OCT image with or without angiographic overlay. Figure [1](#F1){ref-type="fig"} shows the above described procedure for making an overlay image with the help of OCT C-scan and FA images of a normal fundus. The whole procedure (including acquisition and selection of the appropriate image frames) took no more than 15 minutes per patient. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Procedure used in overlay technique using images of a normal fundus. A) original OCT C-scan of normal fundus taken at midretina (large white area in confocal part (left) = corneal reflection); B) Confocal part of OCT C-scan (left) and midfase FA image (right), red markers represent reference points used for spatial transformation; C) Confocal part of OCT C-scan (left) and transformed FA image (right); D) duplicated converted FA image; E) overlay image: green = FA and red/blue = OCT C-scan, showing on the left side the achieved registration based on the confocal part and the FA image, and on the right side the fusion with the OCT part. ::: ![](1471-2415-5-12-1) ::: Results ======= Example A: Birdshot retinochoroiditis (figure [2](#F2){ref-type="fig"}) ----------------------------------------------------------------------- ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Birdshot retinochoroiditis. A) OCT C-scan taken pre-foveal. Right side (OCT part): arrow = temporal retinal thickening, \* = irregularly shaped retinal contour. Left side (confocal part): large patchy white area = corneal reflection, bright central white spot = reflection from system lenses; B) FA late phase image, arrow: temporal leakage; C) transformed FA image, \* = diffuse leakage nasally and along arcades; D) overlay image, red/blue = OCT C-scan, green = FA. ::: ![](1471-2415-5-12-2) ::: The OCT C-scan shows marked retinal swelling temporal to the fovea (figure [2a](#F2){ref-type="fig"}), with small cystoid changes. On the late FA image (figure [2b](#F2){ref-type="fig"}) leakage is seen along the vascular arcades and in the fovea. The spatially transformed FA image (figure [2c](#F2){ref-type="fig"}) is superimposed over the OCT C-scan (figure [2d](#F2){ref-type="fig"}). This overlay image shows that the temporal area of retinal thickening on the OCT exactly matches the temporal, parafoveal area of leakage on the angiographic image. Reviewing the overlay image more closely, smaller areas of extensive leakage at the nasal side of the fovea and along the vascular arcades on the angiographic image (figure [2c](#F2){ref-type="fig"}, asterisks) correspond to the somewhat irregularly shaped retinal contour on the OCT C-scan (figure [2a](#F2){ref-type="fig"}, asterisks). Generalized retinal thickening along the arcades corresponds to extensive intraretinal vascular leakage on the angiogram. There is correspondence between a physiologic parameter (leakage) and morphological changes (retinal thickening and cystic changes). Example B: high myopia with secondary neovascularization (figure [3](#F3){ref-type="fig"}) ------------------------------------------------------------------------------------------ ::: {#F3 .fig} Figure 3 ::: {.caption} ###### High myopia with secondary neovascularization. A) OCT C-scan taken at level of RPE, arrow (right) = neovascular membrane surrounded by a hyporeflective halo (left: patchy central white area in confocal part = corneal reflection); b) en-faceOCT B-scan through neovascular membrane, arrowheads indicate disruption of the double highly reflective layer at the level of RPE caused by the changed orientation of the photoreceptor layer; C) FA mid-phase image showing inside-out a central hyperfluorescence, a hypofluorescent halo (hyperpigmentation) and a halo of hyperfluorescence (window defect) in the area of the neovascularization; D) transformed FA image; E) overlay image showing that the central hyperfluorescence on FA is enclosed by the RPE detachment on OCT, and that the hyperfluorescent halo corresponds to the hyporeflective halo in the OCT. ::: ![](1471-2415-5-12-3) ::: Figure [3a](#F3){ref-type="fig"} shows an OCT C-scan at the level of the retinal pigment epithelium (RPE) in a highly myopic patient with a solitary neovascular membrane, located just outside the foveolar area. An irregularly shaped, hyperreflective circular line is seen, corresponding to a pigment epithelial detachment (PED). The content of this circular area is dense, suggesting that it contains a highly reflective tissue such as a neovascular membrane rather than serous fluid as might be seen in central serous retinopathy \[[@B18]\]. In addition, on the OCT C-scan, the distinct sharp border generated by the PED is surrounded by a dark halo. On the OCT B-scan, the inner band of the double highly reflective layer at the level of the RPE is interrupted just before the edges of the PED (figure [3b](#F3){ref-type="fig"}), corresponding to the dark halo on the OCT C-scan. The dark halo on the OCT C-scan and the interruption of this inner highly reflective layer in the OCT B-scan is most likely caused by the changed orientation of the photoreceptor outer segments. This changes the reflectivity properties of the photoreceptor layer and therefore the reflectivity profile in the OCT image. The FA image (figure [3c](#F3){ref-type="fig"}) shows very prominent leakage in the centre of the lesion, surrounded by a first halo of hypofluorescence, and a second halo of hyperfluorescence. Good spatial correspondence is seen in the overlay (figure [3e](#F3){ref-type="fig"}). The bright spot of prominent leakage on the FA is located well within the confinement of the PED as seen in the OCT C-scan. The hypofluorescent halo on the FA corresponds to the hyperreflective circle on OCT delineating the RPE surrounding the neovascular membrane. The outer hyperfluorescent zone on the FA corresponds to the hyporeflective halo on the OCT. This hyperfluorescent zone on the FA shows the characteristics of a window defect. The corresponding hyporeflective halo in the OCT C-scan surrounding the membrane can be a result of this loss of pigmentation, but is more likely caused by changed reflectivity of the photoreceptor layer due to its changed orientation, as supported by figure [3b](#F3){ref-type="fig"}. Both OCT B- and C-scan also clearly show that the membrane is located below the RPE, which makes it ineligible for current surgical techniques. Example C: Retinal Angiomatous Proliferation (RAP, figure [4](#F4){ref-type="fig"}) ----------------------------------------------------------------------------------- ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Retinal Angiomatous Proliferation (RAP). A) C-scan taken at midretina, small arrow = retinal vessel, arrowhead = large irregular pigment epithelial detachment, arrowhead placed within serous retinal detachment; B) FA image; C) transformed FA image; D) Overlay image showing leakage on FA to be located at the centre of the serous/epithelial detachments on OCT, though its vertical location within the 3-dimensional RAP cannot be specified; E-H) second overlay image created from a C-scan taken midretina, in which the bright spot on FA seems to correlate with the high reflective area in the OCT C-scan (arrows). This is most likely the crest of the irregularly shaped PED or of the retinal component of the angiomatous lesion. ::: ![](1471-2415-5-12-4) ::: This patient has a stage II RAP based on the FA image (figure [4b](#F4){ref-type="fig"}) \[[@B19]\]. Reviewing the FA image, a large area of hyperfluorescence is seen surrounding a bright, fluorescent central spot, presumably the RAP. The first OCT C-scan (figure [4a](#F4){ref-type="fig"}, right) is taken at the level of the retinal vessels. It shows a large serous neurosensory detachment and a large, very irregularly shaped PED. The overlay shows a discrepancy between the hyperfluorescence in the FA and the structural information in the OCT image (figure [4d](#F4){ref-type="fig"}). The hyperfluorescence seems mostly confined to a part of the PED seen on the OCT. The bright spot within the area of diffuse hyperfluorescence can be correlated with a highly reflective area in a second OCT C-scan taken above the large PED (figure [4e--h](#F4){ref-type="fig"}). This appeared to correspond to the top of the PED, but more likely relates to a highly reflective proliferation located between the neurosensory retina and the RPE (not shown). Discussion ========== The OCT-Ophthalmoscope provides clinicians with a novel imaging modality of the retina. The transversally oriented OCT C-scans allow software-assisted overlay with more traditional imaging techniques, such as FA or indocyanine green (ICG) angiography, that are also displayed in a transversal plane. The confocal channel in the OCT-Ophthalmoscope provides a fundoscopic image with a high transversal resolution which has a pixel-to-pixel correspondence with the OCT channel that possesses a very high depth resolution \[[@B10]\]. Therefore, the confocal image can easily be used as a reference image. Images produced by other diagnostic techniques can be mathematically transformed to fit snugly over the confocal image of the OCT C-scan, and thereby be directly superimposed over the OCT image. Our initial goal in making overlay images, was to enhance the interpretation of individual OCT C-scans. These are initially difficult to interpret. The high transversal resolution makes recognition of retinal landmarks difficult in the OCT C-scan images. Classically, pathologic processes are studied in longitudinal sections, so that their transversal appearance is rarely appreciated. By using the overlay technique, it is possible to correlate pathologic features imaged with familiar transversal imaging techniques, such FA, with the morphological information provided on OCT C-scans. We also found that the overlay technique provided additional insight not readily available with either modality alone. In our example of Birdshot retinochoroiditis (figure [2](#F2){ref-type="fig"}) the OCT C-scan showed an irregularly shaped retinal contour in the parafoveolar area, and along the arcades. On the FA image diffuse parafoveolar oedema was present as well as diffuse leakage along the major retinal vessels. Leakage correlated well in the overlay to the irregularly shaped retinal contour on the OCT. Thus, the patchy retinal thickening seen on OCT was caused by diffuse oedema, which if needed could be followed prospectively, without the need of the more invasive angiographic technique. In figure [3](#F3){ref-type="fig"}, OCT and FA concurred in the location of the neovascular membrane. Further, the OCT also showed that the neovascular membrane was situated below the RPE, making this membrane ineligible for current surgical procedures \[[@B20]\], while the retina above was intact and not involved in the process. Combining findings on the longitudinal and transversal OCT allowed one to better comprehend the FA manifestations observed at the edge of the neovascular lesion. Hypofluorescence at the edge of the PED is due to superposition of pigmented RPE cells, while the more peripheral hyperfluorescence is caused by a window defect. The OCT and FA images in the RAP example (figure [4](#F4){ref-type="fig"}) revealed the limitations of FA in defining the extent of involvement, in particular of the RPE \[[@B21],[@B22]\]. FA leakage was present in only a portion of the PED, probably due to the presence of blocked fluorescence, although the cause of this blocked fluorescence could not be ascertained with the OCT. Conventional OCT is a valuable tool in making a diagnosis and in monitoring the progress of macular diseases. The commercially available StratusOCT scans in a longitudinal orientation, providing high resolution, morphologic, cross-sectional OCT scans of the retina, but it does not provide a precise localization of the scanned lines within the macular region, because there is always a time delay between acquisition of the line and storage of the line with its associated fundus image. A retinal thickness map is based on the interpolation of the thickness measurements from six radial OCT scan lines. While these maps are useful to monitor patients, they cannot be combined with other transversal diagnostic techniques because both the map and the scan lines which it is made of, lack good reference points. For this and other reasons, conventional OCT has not yet been incorporated in multimodal mapping systems such as the one developed by Cunha-Vaz and co-workers \[[@B8]\]. Rosen and co-workers \[[@B23]\] were able to produce simultaneous ICG angiography and en-faceOCT in an adjusted, experimental OCT-Ophthalmoscope. Here pixel-to-pixel correspondence was seen in both channels, which ultimately may be a better set-up than the technique proposed here. However, this combination will need further optimization before it becomes clinically applicable. At this moment, our examples suggest that multimodal mapping is possible between the OCT-Ophthalmoscope and existing imaging modalities, without various imaging modalities being incorporated into the system. Combining several imaging techniques provides complementary information which helps in interpreting and understanding the findings of each modality. Additional modalities which could possibly be fused with transversal OCT include microperimetry and (multifocal) electroretinography. These would provide functional correlations to structural alterations in the retina. Conclusion ========== Combining conventional angiographic images and en-faceOCT C-scans assists in the interpretation of both imaging modalities. By combining the physiopathological information in the angiograms with the structural information in the OCT scan, zones of leakage can be correlated to structural changes in the retina or RPE. Fusion of different imaging modalities can contribute to our understanding of complex retinal or choroidal diseases and it can optimize our patient management. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= MvV carried out the acquisition of the OCT images, made the software-assisted overlay images and was responsible for writing this paper. KdV participated extensively in acquiring the overlay images and contributed to writing this paper, especially the methodology section. FV participated in the design of this study, contributed to the interpretation of the acquired overlay images and to writing the paper. CP participated in the design of this study and coordinated the acquisition of the overlay images. MdS participated in the design and coordination of this study, contributed in interpreting the images and in writing the paper. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2415/5/12/prepub>	PubMed Central	2024-06-05T03:55:59.494603	2005-6-13	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180453/", "journal": "BMC Ophthalmol. 2005 Jun 13; 5:12", "authors": [ { "first": "Mirjam EJ", "last": "van Velthoven" }, { "first": "Koos", "last": "de Vos" }, { "first": "Frank D", "last": "Verbraak" }, { "first": "Chris W", "last": "Pool" }, { "first": "Marc D", "last": "de Smet" } ] }
PMC1180454	Background ========== Ophthalmic Viscosurgical Devices (OVDs) have played a key role in the success of phacoemulsification. An ideal OVD is one which is able to maintain the anterior chamber during the procedure particularly capsulorhexis, phaco probe entry and initial phacoemulsification and during intraocular lens implantation, maintain mydriasis and media clarity, protect the endothelium from phaco energy and also prevent postoperative rise in intraocular pressure. Although, till date, no OVD can be considered ideal, corneal endothelial cell loss during cataract surgery has been minimized by the use of viscoelastic substances \[[@B1],[@B2]\]. However, the incidence of variable degree of endothelial decompensation in patients after cataract surgery remains around 17% \[[@B3],[@B4]\]. Hence, there is always the search of an ophthalmic viscosurgical device which is ideal and that can minimize damage to corneal endothelium and other intraoperative complications. Over the past few years, a partial success in this area has led to a decrease in the incidence of bullous keratopathy following cataract surgery requiring penetrating keratoplasty \[[@B5],[@B6]\]. In this study, we compared the efficacy and safety of three ophthalmic viscosurgical devices namely, Viscoat^®^(Sodium chondroitin sulfate 4.0% & sodium hyaluronate 3.0%), Healon GV^®^(Sodium hyaluronate 1.4%) and Healon 5^®^(Sodium hyaluronate 2.3%) in hard nucleus (grade 3 and 4). Methods ======= A randomized clinical trial was performed which comprised of 56 consecutive patients of immature senile cataract with hard nucleus, who attended ophthalmic out patient department at the Rajendra Prasad Centre for Ophthalmic Sciences, All India Institute of Medical Sciences, New Delhi, India. All patients were informed and a written consent was obtained from them to participate in the study. Only those patients were included in the study, who were more than 40 years of age, had senile cataract, nucleus hardness of grade 3/4, not having any evidence of subluxation or pseudoexfoliation and not having any other associated ocular pathology. Patients with preoperative diagnosed glaucoma and/or IOP greater then 20 mmHg preoperatively were excluded from the study. Other exclusion criteria were intraoperative events like manual dilatation of pupil, posterior capsular rent and placement of IOL in the sulcus. Pre-operative evaluation of the patients included the measurement of uncorrected and best corrected visual acuity (UCVA and BCVA respectively), an examination of the anterior segment of the eye under slit lamp biomicroscope, central corneal thickness and fundus evaluation through a dilated pupil. Intraocular pressure was recorded pre-operatively with a non-contact tonometer Topcon CT -- 80 (Topcon America Corporation, Paramus, NJ) in all the patients. Keratometry was performed using Bausch and Lomb keratometer, while the axial length was measured using an A scan biometer (Appasamy associates, Chennai). Anterior chamber depth was recorded by Sonomed 5500 Digital A/B Scan (Latham and Phillips Ophthalmic Products Inc., Grove City, Ohio). The power of the intraocular lens was calculated in all the patients using SRK (Sanders, Retzlaff, Kraff) formula. History of any systemic illness was excluded and blood pressure was measured on admission. Grading (0 to 4+) of the nuclear sclerosis (a combination of opalescence and yellowing) was performed on slit lamp biomicroscope. Corneal thickness was measured using Sonomed ultrasonic pachymeter (Micropach Model 200P; Latham and Phillips Ophthalmic Products Inc., Grove City, Ohio). Endothelial cell counts were performed with Topcon sp-2000P noncontact specular microscope (Topcon America Corporation, Paramus, NJ). Randomization was done pre-operatively by the statistical random table. Surgery was performed by a single surgeon (RBV) and the operating surgeon was told on the operation table to use single viscosurgical device allocated for the patient for the entire procedure as per the randomization. Those examiners (NS, JST) who performed postoperative follow up and examination were masked and unaware about the type of viscoelastic used in a particular case. Surgical technique ------------------ Surgeries were performed under topical anesthesia (proparacaine 0.5%). Pupillary dilatation was achieved by a combination of topical 0.5% tropicacyl and 5% phenylephrine. A 3.2 mm clear corneal tunnel was created either superiorly or temporally depending upon the steeper axis; side port entry was made with the help of microvitreoretinal (MVR) blade. OVD was injected into the anterior chamber according to the randomization, and capsulorhexis was performed. Hydrodissection and hydrodelineation were performed to achieve free rotation of nucleus. Phacoemulsification was done using crater and chop technique by Storz protégé machine (Storz Protégé, Bausch & Lomb, NY, USA) which was followed by a thorough irrigation and aspiration of the cortical matter. Capsular bag was inflated with viscoelastic and a single piece acrylic foldable IOL (ACRYSOF^®^SA60AT; Alcon laboratories, Fort Worth, TX) was implanted in all the patients. The OVD was completely aspirated out with the rock and roll technique. The irrigating solution and the phaco machine were similar in all the three groups. Postoperatively, patients were prescribed 1% Prednisolone acetate and ciprofloxacin 0.3% QID each for 4 weeks and tropicamide 1% TID for 1 week. The intraoperative parameters recorded were the type of OVD, phacoemulsification time, average ultrasonic energy used, time required to remove the viscoelastic material, maintenance of anterior chamber depth and total surgical time. Postoperatively, all the parameters were recorded on day 1, 1 week, 1 month and 3 months using the same method and instruments. Statistical analysis -------------------- Data were recorded on a predesigned proforma and managed on Excel spread sheet. For quantitative variables, approximate normal distribution was assessed and subsequently mean & standard deviation (SD) was computed as summary statistics. Preoperative values for all the parameters were statistically compared amongst the three groups. Group 1 (n = 19), comprised of eyes in which Viscoat^®^(Sodium chondroitin sulfate 4.0% & sodium hyaluronate 3.0%) was used; Group 2 (n = 19) comprised of those in which Healon GV^®^(Sodium hyaluronate 1.4%) was used and Group 3 (n = 18) comprised of those in which Healon 5^®^(Sodium hyaluronate 2.3%) was used. The base line values of all the parameters were statistically similar, therefore we used one-way analysis of variance followed by Scheffe\'s post hoc ANOVA, if required, to compare the mean values at every line point between the three groups. Analysis of co-variance was used to compute mean and SD values at various postoperative follow up. Repeated measures of analysis of variance were used to determine changes from the preoperative values. To compare categorical variables between the groups, Chi square or fisher\'s exact test, as appropriate, was used. STATA 8.0 and SPSS version 10.0 statistical softwares were used for data analysis. P value of \<0.05 was considered as statistically significant. Results ======= The mean age of the patient was 68.73 ± 8.96 years and 43 (76.71%) patients were male. The three groups were comparable preoperatively in terms of age, anterior chamber depth, central corneal thickness and central endothelial cell count (Table [1](#T1){ref-type="table"}). The mean preoperative uncorrected visual acuity in the three groups was 0.12 ± 0.14, 0.09 ± 0.13 and 0.11 ± 0.9 respectively, which improved to 0.51 ± 0.16, 0.61 ± 0.20 and 0.61 ± 0.20 respectively at the final follow up at 3 months. The mean preoperative best corrected visual acuity in the three groups was 0.18 ± 0.11, 0.17 ± 0.19 and 0.18 ± 0.13 respectively and this improved to 0.69 ± 0.24, 0.88 ± 0.18 and 0.79 ± 0.23 respectively at 3 months. The total number of eyes with grade 3 nuclear sclerosis was 8 in Group 1, 11 in Group 2 and 4 in group 3. The eyes with grade 4 nuclear sclerosis were 11, 8 and 14 in group 1, group 2 and group 3 respectively. The mean phaco energy (%) utilized was higher in Group 3 in comparison to Group 2 (p \< 0.03) and Group 1 (p \< 0.05) (Table [2](#T2){ref-type="table"}). The total operating time was longer in Group 3 in comparison to Group 1 (p \< 0.011). Time taken for removal of viscoelastic after IOL insertion in the capsular bag was higher in Group 1 when compared to Group 2 and Group 3 (Table [2](#T2){ref-type="table"}). The mean increase in central corneal thickness was 15.17% in group 1; 17.26% in group 2 and 16.21% in group 3 on the first postoperative day and this change was comparable in the three groups (Table [1](#T1){ref-type="table"}). Preoperative central endothelial cell density was 2311 cells/mm^2^in group 1; 2359 cells/mm^2^in group 2 and 2526 cells/mm^2^in group 3 (Table 3). Postoperatively, the density of endothelial cells on day 1 decreased by 290 cells/mm^2^(12.54%) in group 1 \[Not significant (NS)\], 324 cells/mm^2^(13.76%) in group 2 (NS) and 330 cells/mm^2^(13.06%) in group 3 (NS). On comparative evaluation, there was no significant difference in the change in the endothelial count between the three groups. The mean preoperative intraocular pressure in group 1, group 2 and group 3 were 13.3 ± 2.0, 14.0 ± 2.2 and 13.2 ± 3.2 respectively. On day 1 after surgery, the mean intraocular pressure in the three groups were 16.0 ± 4.7, 12.2 ± 4.7, and 12.3 ± 4.8 respectively. On comparative evaluation between the three groups, the rise in intraocular pressure in was found to be slightly higher in group 1 and significant statistically (1 vs 2 & 1 vs 3: p = 0.02; oneway ANOVA). At 1 week follow up, the mean intraocular pressure in the three groups were 13.8 ± 4.6, 12.0 ± 4.5 and 12.2 ± 4.7 respectively and a comparative evaluation between the groups was not significant. The mean intraocular pressure in the three groups at 3 months were 13.0 ± 3.4, 12.1 ± 3.7 and 12.3 ± 3.6 respectively (p = ns). There was no episode of peripheral extension of the capsulorhexis margin in any of the eyes. There was no evidence of any significant intraoperative or postoperative complication in any eye. Discussion ========== Optimal space maintenance during different stages of the surgical procedure is essential to minimize mechanical damage to the intraocular structures \[[@B7]\]. An increasing number of viscoelastics with different compositions and characteristics are now available \[[@B8]\]. These products differ in their space-maintaining capabilities and other properties. In the present study, we observed that the anterior chamber was well formed in all the patients in the three groups during the entire procedure of capsulorhexis and phaco probe entry and initial phacoemulsification. This property not only helps in maintaining the chamber for better performance of the procedure but also counteracts positive vitreous pressure during the procedure. In our study, greater number of grade 4 nucleus was seen in group 3. Hence the mean phaco energy and total surgical time was higher in this group in comparison to group 1 and group 2. However, there was no significant increase in central corneal thickness in the immediate postoperative period and also at the end of 3 months following surgery. However, pachymetry might be slightly inaccurate as it is impossible to measure exactly the same corneal area every time and measurements of several corneal parts (e.g. superior, nasal, inferior, temporal and central) should have been performed. There was no significant difference in the change in endothelial count in eyes among the three groups. It is reported in literature that endothelial cell loss rate after phacoemulsification with IOL implantation is greater in eyes with a hard nucleus than those with a soft nucleus \[[@B9]-[@B11]\]. However, statistically, all the three viscosurgical devices were comparable in their endothelial protective capabilities. Ravalico G et al performed a study comparing Healon, Healon GV, Viscoat and Hymecel (2% hydroxypropylmethylcellulose) and reported that Healon GV and Viscoat are comparable in their endothelial protective property \[[@B12]\]. However, another study reported a superiority of Viscoat over Healon GV and Healon in terms of protection to endothelium during phacoemulsification \[[@B13]\]. Holzer MP et al reported in their study that Healon 5 is superior to Healon GV and Viscoat in terms of endothelial protective property \[[@B14]\]. Several techniques have been reported in the literature for removal of the OVDs. These include: Rock and roll technique, two-compartment technique and bimanual irrigation/aspiration technique \[[@B15]\]. Healon-5^®^has been reported to be a viscoadaptive OVD as it has different functions at different flow rates. At lower flow rates it behaves as a very cohesive viscoelastic like a Healon-GV^®^. At higher flow rates, e.g. in chopping techniques, it begins to fracture and behaves similar to a dispersive viscoelastic, such as Viscoat^®^. In the present study, greater time was required to remove the viscoelastic after IOL implantation in group 1. This suggests that Viscoat has a more retentive nature in comparison to Healon GV and Healon 5 which come out in bolus and require lesser time. Again OVD removal time was significantly higher in group 3 in comparison to group 2. This suggests that Healon 5 has a greater dispersive quality than Healon GV. Postoperatively, there was no evidence of retention of residual OVD in the anterior chamber of any eye on slit lamp biomicroscope. There was a significant increase in the mean intraocular pressure in group 1 (although in the normal range) in comparison to group 2 (p = 0.02) and 3 (p = 0.02) on day 1 of the postoperative phase. However, at subsequent follow up visits, there was no significant difference in the mean intraocular pressure among the three groups. This transient rise in the mean intraocular pressure in group 1 suggests that although on slit lamp biomicroscopy, there was no direct evidence of retained viscosurgical device in the anterior chamber, there may be microscopic retention in the trabecular meshwork that can go undetected on routine examination. This microscopic retention can decrease the aqueous outflow transiently and result in variable increase in intraocular pressure. Subsequently, it gets drained off and the intraocular pressure returns close to the baseline value. A similar study comparing Healon 5 with Viscoat reported that the IOP in the early postoperative period was higher in the Viscoat group than in the Healon 5 group \[[@B16]\]. However, another study comparing Healon GV, Healon 5 and Iviz (Sodium hyaluronate 1%) concludes that although Healon 5 takes longer time for removal, there is no significant difference in the postoperative intraocular pressure \[[@B17]\]. Conclusion ========== In the present study, we conclude that the safety and efficacy of the three viscosurgical devices namely Viscoat^®^, Healon GV^®^and Healon 5^®^in performing phacoemulsification is comparable. However, Viscoat^®^can result in a mild transient rise in the intraocular pressure. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= RBV designed the study and performed surgeries. KV performed the data collection. RS wrote the manuscript. JST and NS followed up the patients. RMP performed the statistical evaluation. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2415/5/17/prepub> Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Mean Endothelial Cell Count. ::: ![](1471-2415-5-17-1) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Pre operative and Post operative characteristics of the patients ::: ----------------------------------------------------------------------------------------------------------------------------------------------- Parameter Group 1 n = 19 Group 2 n = 19 Group 3 n = 18 One-way ANOVA, F-Value P value ------------------------------------- -------------------- -------------------- -------------------- ---------------------------- ------------- Age 69.6 ± 9.2 65.8 ± 7.8 70.8 ± 9.9 1.58 NS ACD 2.6 ± 0.20 2.8 ± 0.37 2.59 ± 0.45 1.97 NS UCVA Pre-operative 0.12 ± 0.14 0.09 ± 0.13 0.11 ± 0.9 0.18 NS Day one 0.28 ± 19^a^ 0.39 ± 23^c^ 0.37 + 0.27^b^ 1.22 NS Three month 0.51 ± 0.16^c^ 0.61 ± 0.20^c^ 0.61 ± 0.20^c^ 1.51 NS Repeated measure within group ANOVA F = 15.0\ F = 28.29\ F = 23.5\ P = 0.0001 P = 0.0001 P = 0.0001 BCVA Pre-operative 0.18 ± 0.11 0.17 ± 0.19 0.18 ± 0.13 0.04 NS Day one 0.46 ± 0.23^b^ 0.69 ± 0.26^c^ 0.55 ± 0.31^a^ 1.70 NS Three month 0.69 ± 0.24^c^ 0.88 ± 0.18^c^ 0.79 ± 0.23^c^ 2.73 NS Repeated measure within group ANOVA F = 34.17\ F = 43.8\ F = 23.5\ P = 0.0001 P = 0.0001 P = 0.0001 Corneal thickness Pre-operative 560.7 ± 91.5 567.5 ± 44.07 576.0 ± 50.2 0.25 NS Day one 645.8 ± 94.1^c^ 665.5 ± 141.9^b^ 670.6 ± 117.1^b^ 0.32 NS Three month 573.4 ± 73.6 586.1 ± 29.0 586.7 ± 36.0 1.50 NS Repeated measure within group ANOVA F = 17.5\ F = 11.4\ F = 11.19\ P = 0.0001 P = 0.0001 P = 0.0001 Endothelial cell count. Pre-operative 2311.73 ± 288.0 2359.78 ± 383.2 2526.6 ± 305.4 2.36 NS Day one 2021.20 ± 201.0 2035.05 ± 377.0 2196.4 ± 378.2 0.11 NS Three month 1973.00 ± 537.0 1959.00 ± 503.0 2132.0 ± 482.0 0.82 NS Repeated measure within group ANOVA F = 4.2\ F = 2.20\ F = 17.5\ P = 0.004 P = NS P = 0.0001 ----------------------------------------------------------------------------------------------------------------------------------------------- a: p \< 0.05, b: \<0.01, c: \<0.001 as compared to preoperative parameter UCVA = Uncorrected visual acuity BCVA = Best corrected visual acuity ANOVA = Analysis of Variance NS = Not Significant ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Intra-operative oarameters ::: ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Parameter Group 1 n = 19 Group 2 n = 19 Group 3 n = 18 One-way ANOVA, F-Value P value Post hoc ANOVA, (Scheffe\'s test) -------------------------------- -------------------- -------------------- -------------------- ---------------------------- ------------- --------------------------------------- Mean phaco energy (%) 0.28 ± 0.06 0.31 ± 0.06 0.38 ± 0.09 6.75 0.0024 I Vs III: p = 0.03\ II Vs III: p = 0.05 Mean phaco time (min) 2.0 ± 1.0 1.8 ± 0.8 2.1 ± 0.7 0.026 \-- \-- Effective phaco time (sec) 33.8 ± 17.7 36.1 ± 20.4 49.5 ± 23.1 3.31 \-- \-- Mean total surgical time (min) 12.9 ± 3.3 15.2 ± 3.5 16.5 ± 3.6 5.06 1 Vs III: p = 0.011 OVD removal time (sec) 66.6 ± 11.2 45.1 ± 9.0 55.47 ± 6.6 26 0.0001 I Vs II: p = 0.0001\ I Vs III: p = 0.003\ II Vs III: p = 0.006 ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- OVD = Ophthalmic viscosurgical device min = minutes sec = seconds :::	PubMed Central	2024-06-05T03:55:59.496755	2005-7-15	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180454/", "journal": "BMC Ophthalmol. 2005 Jul 15; 5:17", "authors": [ { "first": "Rasik B", "last": "Vajpayee" }, { "first": "Kamna", "last": "Verma" }, { "first": "Rajesh", "last": "Sinha" }, { "first": "Jeewan S", "last": "Titiyal" }, { "first": "RM", "last": "Pandey" }, { "first": "Namrata", "last": "Sharma" } ] }
PMC1180455	Background ========== Interpersonal communication problems are present throughout the health care system. These problems take on greater significance when compounded by language and cultural barriers. The Deaf community, whose predominant language is American Sign Language (ASL), faces the same barriers to accessing health information and services as other communities whose members use English as a second language \[[@B1]-[@B22]\]. In a prior research study with the Deaf community, study participants reported that educational materials needed to be culturally and linguistically aligned if they are to optimally address the health disparities pervasive in the Deaf community \[[@B7]\]. The mode with which health information reaches the Deaf community is key to its value and impact. Given that this community relies heavily on the visual receipt of information, programs that are presented in ASL and enriched with open captioning and pictures are the best formats in which to distribute information \[[@B6],[@B16],[@B23]-[@B26]\]. This paper reports on the evaluation of a prostate and testicular cancer education video, filmed in ASL, to increase Deaf men\'s access to health information. Prostate cancer is the most common, life-threatening cancer in older American men (age 50 and older), and testicular cancer is a potentially curable disease that affects younger men (ages 15--40) \[[@B27]\]. Deaf men\'s empowerment and self-advocacy in the health care setting begins with greater access to health information. Methods ======= The 52-minute video used for this study, entitled Prostate and Testicular Cancer: Know your Options, was produced in 2002 through a partnership composed of Deaf Community Services of San Diego, Inc. (DCS), the Rebecca and John Moores UCSD Cancer Center, Bovee Productions, and Gallaudet University. The film was a New York Festivals Certificate of Distinction Finalist Winner in 2004. The educational video, filmed in ASL, shows two trainers providing answers to their audience\'s questions. Although the focus of this study was on the Deaf community, the addition of open text captioning and voice overlay without ambient music enabled the video to be accessed by both deaf and hard-of-hearing individuals, as well as their hearing loved ones. The video can be viewed at <http://cancer.ucsd.edu/deafinfo>. Participants in the authors\' prior studies had recommended that the ASL videos should include the same amount and complexity of information as that which would be available to the hearing community. Further, they recommended that men, like women, needed access to quality information about their gender specific cancers. In the light of the demonstrated need for increased access to health information, the authors opted to create a video that addressed prostate cancer and testicular cancer. This offered a way of reducing redundancy, encouraging men of diverse ages to interact in a supportive, culturally aligned, learning environment, and also covered more topics on the partnership\'s limited budget. As a result, the final video was complex, particularly in light of the community\'s limited prior exposure to cancer information and the many new medical terms and concepts that needed to be introduced. As such, the film was intended for multiple viewings to achieve mastery all of the information. The study looked at the impact of only a single viewing. Using a train-the-trainers model, the video was presented to small groups of Deaf men and women in community settings. Recruitment strategies were accomplished in collaboration with Deaf community service agencies throughout California: Deaf Community Services of San Diego, Inc.; Center on Deafness-Inland Empire; Deaf and Hard of Hearing Service Center, Inc.; Deaf Counseling, Advocacy and Referral Agency; NorCal Center on Deafness; Greater Los Angeles Agency on Deafness, Inc. These strategies included in-person and e-mailed word-of-mouth dissemination and IRB-approved flyers \[[@B28],[@B29]\]. Study prospects were told that participants would be helping to evaluate a prostate and testicular cancer education program that had been specifically designed to improve the Deaf community\'s access to health information. As a thank you for their participation, a meal was provided to all members of the audience before the presentation and a \$25 incentive was offered to those who completed all three survey documents. Participants were also given their own copy of the video and encouraged to view it again and share it with other members of the Deaf community. After passing a knowledge competency exam, six native signers were trained how to use the educational video. Sixteen educational sessions were held with a total of 102 men and 59 women attending. Attendance per session ranged from five to 32 persons. Only men were invited to take part in the evaluation of the video, and of those 102 men who attended the sessions, all consented to participate. All participants were offered the option of having the consent process, the consent documents, and the surveys translated into ASL, a service required to varying degrees. The consented participants completed the baseline survey before viewing the video. The video had been filmed in a question and answer format to make it easy for the trainers to stop and start the video to stimulate audience discussion and assure comprehension of content. Study participants and other attendees were to watch the video under the direction of the local grassroots trainers with the oversight of the project coordinator, both native signers. At the conclusion of the video, participants completed post-education intervention surveys. Participants and other audience members (both male and female) then engaged in a focus group discussion lead by the facilitators, lasting approximately 30 minutes to 60 minutes. Two months later, follow-up meetings were set up via e-mail, TTY, or in-person for completion of the final survey, with 95 men (93% of original participants) completing the follow-up survey. The paper and pen surveys collected socio-demographic data and used multiple-choice, true or false, and open-ended questions related to the presentation. All data collection instruments were approved by the University\'s Institutional Review Board (IRB). Statistical analysis -------------------- The surveys were analyzed using t-tests and McNemar chi-square tests to assess for differences between pre, post, and two-month responses. Comparisons made between the pre/post video surveys and the two-month follow-up surveys examined only the 95 men who responded at two months. Unanswered questions were omitted from the analysis. Sample description ------------------ The study participants all resided in Southern California and were between 18 and 86 years of age (mean age 44.35; SD 17.39). Participants completed an average of 13.68 (SD 3.32) years of school with a range from four years to 24 years. While 28.4% (29) of the group completed high school, 28.4% (29) completed some college, 20.6% (21) completed college, 14.7% (15) completed education beyond college, and 6.9% (7) did not answer the question. The group was composed of: 62.7% (64) Caucasians; 7.8% (8) African Americans; 5.9% (6) Asian/Pacific Islanders; 16.7% (17) Hispanics; 2.9% (3) Mixed; and 3.9% (4) Other. For health care expenses, 6.9% (7) paid for their health care out of pocket; the others had MediCal/Medicaid (27.5% (28)), Medicare (25.5% (26)), other health insurance (39.2% (40)), or other sources (1% (1)). Of the sample, 43.1% (44) reported having friends with prostate cancer, and 19.6% (20) reported having friends with testicular cancer. There were seven men (6.9%) who reported having had prostate cancer themselves, and there was one man (1%) who reported having had testicular cancer. Results ======= Table [1](#T1){ref-type="table"} shows the health care providers\' educational interventions recalled by the participants. When asked about screening for prostate cancer, 26.1% (18/69) of the men under age 50 and 75.8% (25/33) of the men over age 50 reported that they had been examined by their doctor at some time for prostate cancer. When asked about screening for testicular cancer, 17.6% (6/34) of the men under age 35 and 33.9% (21/62) of the men over age 35 reported having been examined. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Reported educational efforts of participants\' health care providers ::: Question Responded Yes -------------------------------------------------------------------------------- ------------------- Has your doctor ever talked with you about prostate cancer? 27 (26.5%) Has your doctor ever talked with you about testicular cancer? 13 (12.7%) Has another health care provider ever talked with you about prostate cancer? 13 (12.7%) Has another health care provider ever talked with you about testicular cancer? 6 (5.9%) Have you been trained by a health care provider to do a testicular self-exam? 9 (8.8%) Do you know how to do a testicular self-exam? 24 (23.5%) ::: When the men were asked whether they felt that there needed to be more programs on cancer and other health concerns specifically made for the Deaf community, 94.1% (96) responded affirmatively. Before and after the educational intervention, the participants were asked to rate their perception of the Deaf community\'s access to health information on a one to five scale with one being \"very little\" access and five being \"a lot\" of access. The results, shown in Table [2](#T2){ref-type="table"}, reveal that most men rated health information access as \"very little\" or \"little\" before the video was shown, but perceived access to be higher after viewing the video, a statistically significant shift. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Perception of access to health information on 1--5 scale (N = 102) (1-Very little, 2-Little, 3-Some, 4-Quite a bit, 5-A lot) ::: Pre-test (mean) ± SD Post-test (mean) ± SD ------------------- ---------------------- ----------------------- Prostate Cancer 2.06 ± 1.10 3.06\* ± 1.53 Testicular Cancer 1.88 ± 1.04 3.01\* ± 1.54 \T-test showed significant increase in mean score between pre and post-test surveys. ::: A total of 25 knowledge questions were asked on the surveys to assess whether there was any significant change in the participants\' knowledge about prostate and testicular cancer from pre- to post-intervention and whether this increase was maintained at the two-month time point. From baseline to immediately post-intervention, participants\' overall knowledge increased significantly (pre-video X~1~= 15.2, post-video X~2~= 18.4, t = -9.698, p \< 0.05). X denotes the mean number of correct responses; all the participants\' scores (number correct/25) were added, and the total was divided by the number of participants. From baseline to the two-month follow-up, participants maintained their overall statistically significant increase in knowledge (pre-video X~1~= 15.3, two-month X~3~= 17.1, t = -4.726, p \< 0.05). These questions are clustered by format, and the results are shown in Tables [3](#T3){ref-type="table"} and [4](#T4){ref-type="table"}. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Percentage of correct responses to true or false questions ::: Testicular true-false statements* Pre-video Post-video 2 month post --------------------------------------------------------------------------------------------------------------------------- --------------- ---------------- ------------------ Testicular cancer usually occurs in men 15--40 years old. (True) 48 (47.5%) 94 (93.1%)\* 80 (84.2%)\^ Older men are more likely to get testicular cancer than younger men. (False) 34 (33.7%) 78 (77.2%)\* 56 (58.9%)\^ Testicular cancer can be cured. (True) 59 (59%) 68 (68%) 70 (73.7%)\^ After treatment for testicular cancer, most men can still have children. (True) 49 (49.5%) 84 (84.8%)\* 65 (69.9%)\^ Your testicle needs to be removed if you have testicular cancer. (True) 73 (73%) 92 (92%)\* 76 (80%) When testicular cancer is suspected, a biopsy will be recommended. (False) 17 (16.8%) 43 (42.6%)\* 29 (30.9%) If testicular cancer is found in one testicle, the doctor will remove both testicles. (False) 70 (72.2%) 72 (72.4%) 75 (83.3%) Prostate true-false statements Benign prostatic hyperplasia is a type of prostate cancer. (False) 41 (41%) 38 (38%) 34 (37%) Older men are more likely to get prostate cancer than younger men. (True) 84 (82.4%) 89 (87.3%) 78 (82.1%) \"Watchful waiting\" is an option for some cases of prostate cancer. (True) 50 (50%) 85 (85%)\* 57 (60%) Early detection of prostate cancer increases your treatment options. (True) 85 (84.2%) 92 (91.1%) 86 (90.5%) When prostate cancer is suspected, a biopsy is recommended. (True) 80 (79.2%) 85 (84.2%) 75 (78.9%) Men who are at high risk of getting prostate cancer should be offered screening every year beginning at age 45. (True) 79 (78.2%) 66 (65.3%)\* 67 (70.5%) Men who are at average risk of getting prostate cancer should be offered screening every year beginning at age 50. (True) 66 (65.3%) 94 (93.1%)\* 74 (78.7%)\^ \Significant increase in knowledge between pre- and post-test using McNemar chi -square test, alpha\<0.05 \^Significant increase in knowledge maintained between pre- and two-month test using McNemar chi-square test, alpha\<0.05 ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Ability to select correct multiple choice answers about prostate and testicular cancers ::: Question* Pre-video Post-video 2 month post ---------------------------------------------------------------------------------------- --------------- ---------------- ------------------ What is cancer? A disease where your cells become mutated and divide uncontrollably. 73 (74.5%) 88 (89.8%)\* 78 (83%) What is a Prostate Specific Antigen Test? A test to measure the amount of PSA in a man\'s blood. 56 (56%) 70 (70%)\* 62 (66.7%) What increases your risk of getting prostate cancer? Getting older. 78 (78%) 76 (76%) 70 (74.5%) Being White or African American. 25 (25%) 71 (71%)\* 56 (59.6%)\^ Family history of prostate cancer. 60 (60%) 79 (79%)\* 68 (72.3%) What is a Digital Rectal Examination (DRE)? An examination of a man\'s prostate through his rectum. 66 (67.3%) 79 (80.6%)\* 66 (70.2%) Which of the following are possible side effects of treatment for prostate cancer? Losing one\'s ability to have an erection. 58 (57.4%) 79 (78.2%)\* 66 (70.2%)\^ Losing one\'s ability to hold urine (pee). 75 (74.3%) 87 (86.1%)\* 73 (77.7%) What increases your risk of getting testicular cancer? Family history of testicular cancer. 63 (63%) 81 (81%)\* 68 (73.1%) Undescended testicle in childhood. 18 (18%) 47 (47%)\* 35 (37.6%)\^ Which of the following is the treatment of testicular cancer? Surgical removal of the testicle that contains the cancer. 72 (72%) 82 (82%) 71 (77.2%) \Significant increase in knowledge between pre- and post-test using McNemar chi -square test, alpha\<0.05 \^Significant increase in knowledge maintained between pre- and two-month test using McNemar chi-square test, alpha\<0.05 ::: When the individual questions were analyzed, significant changes were also noted among the three time points. Of the seven true or false questions dealing with testicular cancer displayed in Table [3](#T3){ref-type="table"}, five showed a statistically significant improvement in the frequency of correct responses immediately post-intervention. At two months, participants\' responses showed that they maintained their increased knowledge from baseline for three of these five questions. Knowledge related to a sixth question (Testicular cancer can be cured) continued to increase at each time point and became significant at two months. As is evident from Table [3](#T3){ref-type="table"}, some the of the most striking improvements in participants\' understanding related to the age ranges during which men are at risk for testicular cancer and the consequences of treatment on fertility. Table [3](#T3){ref-type="table"} also shows the results of seven true or false questions for prostate cancer. In contrast to baseline testicular cancer knowledge, baseline knowledge of prostate cancer was generally much higher. As a result, there was less room for improvement. Scores significantly increased from the pre-video to the post-video in only three of the individual questions and remained significantly higher for only one. After viewing the video, 85% (85) of the men correctly identified \"watchful waiting\" as an option for managing prostate cancer, representing a significant gain in knowledge. However, this post-test gain dropped to 60% (57) at two months, a value that was not significantly different than that of the pre-test survey. In addition, while most men seem to have learned that prostate screening is recommended for men of average risk beginning at 50 years old, they were not able to recall that men of higher risk should begin screening at age 45. The other three questions did not show a significant increase in knowledge, and it was clear that both before and after the video, most men were confused about the definition of Benign Prostatic Hyperplasia (BPH). As shown in Table [4](#T4){ref-type="table"}, nine of the 11 multiple-choice questions showed a statistically significant increase in knowledge immediately following the video presentation. For example, 47% of participants correctly acknowledged, after viewing the video, that having an undescended testicle is a risk factor for testicular cancer, an increase from 18% at baseline. The remaining two questions showed a high rate of correct responses before and after the video, indicating that the participants were already informed about these concepts. While there was generally still greater knowledge at two months than at baseline, the increase remained statistically significant for only three questions. Only for the question related to aging as a risk factor for prostate cancer was there a decrease in the number of correct responses from baseline to two months. The participants were asked to rank order their best sources of health information from a list including the following: doctor, nurse, friends, family, newspaper, TV, magazine, health books/pamphlets, Internet, DCS, and special health education programs. The results shown in Table [5](#T5){ref-type="table"} reveal that while doctors remain the number one trusted source of information in this group, other sources like DCS, special health education programs, and the Internet are also important avenues of communication. Of note, written sources of information were not ranked as highly as person-to-person sources of information. Each box in Table [5](#T5){ref-type="table"} shows the percentage ranking of the top four responses for the first, second, third, and fourth response categories. ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Best sources of health information ::: ------------------------------------------------------------------------------------------------------------------------------------- Rank* Pre-video Post-video 2 month post ---------- ---------------------------------------- ---------------------------------------- ---------------------------------------- First Doctor (47%)\ Doctor (48%)\ Doctor (42.2%)\ DCS (15.7%)\ DCS (16%)\ DCS (18.6%)\ Internet (10.8%)\ Internet (13.7%)\ Internet (13.7%)\ Health education programs (9.8%) Health education programs (9.8%) Health education programs (7.8%) Second Internet (18.6%)\ DCS (21%)\ DCS (17.6%)\ DCS (15.7%)\ Internet (15%)\ Doctor (15.7%)\ Health education programs (14.7%)\ Health books/pamphlets (12.7%)\ Health education programs (12.7%)\ Health books/pamphlets (7.8%) Doctor (12.7%) Internet (9.8%) Third DCS (17.6 %)\ Health education programs (19.6%)\ DCS (16.7%)\ Health books/pamphlets (16.7%)\ DCS (15%)\ Health education programs (14.7%)\ Family (12.7%)\ Health books/pamphlets (12.7%)\ Health books/pamphlets (11.8%)\ Doctor (11.8%) Friends (12.7%) Internet (11.8%) Fourth Health education programs (13.7%)\ Health education programs (15.7%)\ Health education programs (20.6%)\ Doctor (13.7%)\ Family (15.7%)\ DCS (11.8%)\ Friends and family (11.8%)\ DCS (14.7%)\ Doctor (11.8%)\ Health books/ pamphlets (11.8%) Health books/pamphlets (11.8%) Health books/pamphlets (10.8%) ------------------------------------------------------------------------------------------------------------------------------------- ::: Participants also completed an open-ended question that asked them to list up to four sources of cancer information. Of the 102 participants, 76 listed at least one source, 48 listed at least two sources, 30 listed at least three sources, and 10 listed four sources. The results of these responses were coded according to whether English literacy was required and according to whether the learning was based on social interaction. For example, pamphlets, Internet, and books were coded as non-interactive, English language based methods of learning, while doctors, DCS, friends, and family were coded as interactive, socially based methods of learning. Analysis revealed an overwhelming preference for the latter. Seventy-one percent listed an interactive source of cancer information for response one, 69% for response two, 67% for response three, and 60% for response four. The results given above are taken from the baseline pre-video survey; the results from the same question in the post-video and the two-month surveys were not statistically different. In the immediate post-intervention survey, 98 (96.1%) felt that the video provided useful information, but 88 (86.3%) felt that too much information was provided. Most of the men (85 (83.3%)) agreed that the video needed to be watched more than once to understand all the information, and 92 (90.2%) of the men planned on watching the video again. Seventy-six (74.5%) felt that most in the Deaf community would be interested in the video, and 95 (93.1%) were comfortable showing the video to their friends. At the two-month survey, 67 (65.7%) of the participants reported they had watched the video again, and 28 (27.5%) had already shared the video with someone else. Discussion ========== The goals of this study were to gain knowledge of participants\' baseline screening behaviors and knowledge, evaluate the effectiveness of a prostate and testicular cancer video program in increasing knowledge about these cancers in the Deaf community, and gain feedback from the participants about their experience with the program and with other cancer information sources. An additional aim was to increase the Deaf community\'s awareness of available health resources. This study demonstrates that the provision of culturally and linguistically aligned educational strategies can increase the community\'s access to health information. There was near universal agreement that the video provided useful information and that there needs to be more special health programs designed for the Deaf community. Most men viewed the video again, and many shared it with others. After viewing the video, which provided websites and TTY numbers to access additional information, and learning that the video could be accessed on the Internet, participants perceived that the Deaf community had increased access to health information. Consensus reached in the focus group discussions was that the video was a culturally aligned method of education about these two cancer topics. Thus, the participants\' perception of an increase in access to information was the result of having information available in ASL and having knowledge of where to find additional resources. Participants also demonstrated a significant increase in knowledge after viewing the video just once. Overall, there was an increase in the number of correct responses after the video in sixteen of the 25 knowledge questions. While this gain was not fully maintained at two months, it still remained above baseline at a statistically significant level. Since the men all received copies of the video and were aware of the video\'s presence on the Internet, they also knew they could access the information again to review or to deal with a health related crisis. While including as much information in this video as possible was an admirable goal, given the dearth of information available in ASL, participants\' feedback suggests that a single health topic focus would be better for future videos. Most of the participants felt that the video offered too much information to absorb in a single viewing. This is not surprising, given that the video attempted to explain highly complex scientific concepts like chemotherapy, metastasis, benign prostatic hypertrophy, and malignancy to a non-medical audience. While the video\'s focus on two cancers may have reduced participants\' long-term retention of details, it allowed men of diverse ages to lend support and encouragement to each other during the training sessions and facilitated subject recruitment. Because the video was filmed as question and answer dyads, it is possible to view the two topics, prostate cancer and testicular cancer, separately. Another study is currently underway to explore the impact of just viewing the video dyads related to testicular cancer. This feedback highlights the need for programs that continually reinforce health concepts in the Deaf community rather than just one-time seminars. This study also reinforces the importance of long-term follow-up in the assessment of educational interventions. Studies that simply survey participants before and after the intervention fail to recognize that long-term retention of information does not always correlate with short-term learning. Moreover, this study underscores the challenge health care providers face even when trying to explain complex medical issues through a medical interpreter. Videos such as this one should help to bridge the communication gaps. While a strength of this study is its diverse sample, generalizations must still be drawn with caution since this group of men collectively had a relatively high level of education. It is unclear whether this video and educational format would be equally effective for deaf persons who have a low ASL literacy or more limited education. Additional evaluation of this educational format is therefore warranted with those audiences. Further, as with all studies, these men were self selected by their interest in learning about cancer, so their motivation to gain knowledge was probably higher than average. They may not be representative of all patients who will present for medical care. The most popular sources of information among the surveyed individuals included physicians, Internet, DCS, and health education programs. The available literature reports communication problems and mistrust between deaf and hard-of-hearing persons and their health care providers \[[@B13],[@B21],[@B30],[@B31]\]. In contrast, the findings in this study show that most of the men rely on their doctors for health information. They also report a high frequency of prostate exams and a relatively high knowledge of prostate cancer information at baseline. Thus while there may be a general distrust between the Deaf community and health care providers, these participants and their health care professionals are clearly succeeding at overcoming at least a proportion of the barriers to information and health care. Alternately, while the participants are accessing this information and care, they may still be reporting the difficulties they experience in achieving these outcomes. This educational intervention was created specifically to help health care providers meet the information needs of the Deaf community. Since the video can be viewed on the Internet at the Rebecca and John Moores UCSD Cancer Center website <http://cancer.ucsd.edu/deafinfo> and through the National Association of the Deaf\'s Caption Media Resource library, health care providers can encourage their patients to view this and other videos provided in ASL or with captioning. While such educational resources will offer a valuable tool to providers and their deaf patients, the focus group discussions showed that the provision of high level certified medical interpreters would be the single most important improvement health care providers could offer to their deaf patients to reduce the difficulties they experience when seeking medical care. With the growing number of certified medical interpreters and technology advances that have made video relay services available, it should be increasingly easier for physicians and their patients to access high level interpreting services more efficiently. When the participants\' responses regarding sources of cancer information were analyzed according to whether English literacy was required and according to whether the learning was based on social interaction, participants overwhelmingly preferred interactive forms of learning. This further reinforces the importance of providing information in the correct format to properly meet the needs of Deaf community. Ideally, an ASL video should be accompanied by person-to-person interaction between the viewer and another individual familiar with the health topic in order to facilitate better retention and understanding of the material. Television was rarely mentioned and radio was never mentioned as a source of information; this highlights the fact that these popular sources of information for the hearing public are not adequate means to address the needs of the Deaf community. An unforeseen benefit of this program for the Deaf community was that it exposed many individuals to the process of research studies and allowed them to become familiar with the consenting process, data collection procedures, and protocol adherence. Considerable time was devoted to explaining the reasons why one does a pre- and post-survey and how the results of these surveys could be used to design even better programs for the Deaf community. Study coordinators remarked on the audience\'s enthusiasm about understanding how the video was made and how the research study worked. They were uniformly appreciative of the intervention and willing to share the experience with loved ones and friends outside of the sessions. In the future, this positive exposure may encourage members of the Deaf community to consider invitations to participate in clinical trials and other research studies. Conclusion ========== ASL videos provide an effective tool for bringing cancer information to the Deaf community. Deaf men in the study showed significant increases in their short and long-term understanding of prostate and testicular cancers after viewing the film, Prostate and Testicular Cancer: Know Your Options. At the two-month follow-up survey, this increase in knowledge was still significant, although slightly diminished. This relative lack of sustainability is due in part to the complexity of the material, but it also highlights that multiple viewings are needed to gain the full benefit of the educational intervention. The Deaf community, like other communities, will benefit from being exposed to multiple repetitions of the same message. Future videos should be focused on one cancer and shorter in duration. The men in the study confirmed the need for more health education materials aimed at the Deaf community and rated doctors as their number one source of health information. Ideally, increasing distribution of health education materials to physician\'s offices would facilitate better access to and trust in educational interventions. Competing interests =================== The author declares that he has no competing interests. Authors\' contributions ======================= AF wrote the script and designed the graphics for the video, participated in the data analysis, and drafted the manuscript. GRS conceived the study, oversaw the day-to-day operation of the project, and drafted the manuscript. CK performed the statistical analysis and assisted with review of the manuscript. PB and SM were co-leaders in the study\'s training sessions, conducted the informed consenting process, collected and managed the database, and assisted with the cultural interpretation of the results. MB produced the video, and assisted with review of the manuscript. All authors have read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2458/5/63/prepub> Acknowledgements ================ The Alliance Healthcare Foundation, the California Endowment, and the National Cancer Institute Grant R25 CA65745 funded this project. Support and resources were provided by the American Cancer Society, the National Cancer Institute, and the Rebecca and John Moores UCSD Cancer Center. The video was produced through a partnership of Deaf Community Services of San Diego, Inc., the Rebecca and John Moores UCSD Cancer Center, Bovee Productions, and Gallaudet University. The Center on Deafness-Inland Empire; Deaf and Hard of Hearing Service Center, Inc.; Deaf Community Services of San Diego, Inc.; Deaf Counseling, Advocacy and Referral Agency; Gallaudet University; NorCal Center on Deafness; and Greater Los Angeles Agency on Deafness, Inc. aided in the recruitment of study subjects and facilitated the educational programs. Melanie Nakaji, MA, MS reviewed the manuscript and offered important suggestions on how to make the paper more culturally sensitive, relevant, and accurate for the Deaf community. Carol Salem MD, Joseph Schmidt MD, Michael Albo MD, Regina Gandour-Edwards MD, and Stephen Seagren MD participated as medical advisors in the drafting of the video manuscript. Thomas Duva Jr., Thomas Galey MS, Leslie Elion JD, and Raymond Trybus PhD of the Deaf Community Services of San Diego, Inc. served as vital contacts to the local Deaf community and as advisors on the project goals. Thomas Duva Jr. and Thomas Galey MS also served as hosts in the educational video used in this study, Prostate and Testicular Cancer: Know Your Options. Yoon Lee provided assistance in making the video appropriate in content and context for the Deaf community.	PubMed Central	2024-06-05T03:55:59.499118	2005-6-6	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180455/", "journal": "BMC Public Health. 2005 Jun 6; 5:63", "authors": [ { "first": "Ann", "last": "Folkins" }, { "first": "Georgia Robins", "last": "Sadler" }, { "first": "Celine", "last": "Ko" }, { "first": "Patricia", "last": "Branz" }, { "first": "Shane", "last": "Marsh" }, { "first": "Michael", "last": "Bovee" } ] }
PMC1180456	Background ========== At the turn of the 21st century, social inequalities in health continue to be a key public health problem in advanced societies, including European countries \[[@B1]\]. With regard to mortality, mean difference in life expectancy between those at the top and at the bottom of a society, as defined by education, income and employment status, are anywhere from 4 to 10 years \[[@B1]\]. In recognition of the importance of social inequalities in health, the World Health Organization (WHO) has set a special goal of reducing these inequalities in its global program \'Health for all in the Year 2000\' \[[@B2],[@B3]\]. Improved living conditions have contributed to better health and a decrease in death rates across all classes in developed countries \[[@B4]\]. Ecological studies suggest that living areas which are highly contaminated with air pollutants are over-proportionately inhabited by people of lower socio-economic status \[[@B5]\]. Studies done immediately after the German re-unification \[[@B5],[@B6]\] demonstrated that social inequalities in health-related living conditions already existed under the social system in East Germany. In both parts of Germany children of the lower class were more likely to live at busy streets and were therefore exposed to a greater degree to traffic-related air pollution and a higher noise level \[[@B5]\]. Dwellings of people of lower social class were more likely to be located near industrial areas, and there was less room per person in the houses \[[@B5]\]. Gas was used more frequently for cooking and warm water supply, and finally the dwellings were more frequently heated with individual coal-burning stoves than upper class dwellings \[[@B5]\]. After Germany re-unified in 1990, dramatic social, economic and environmental changes happened during the first decade in East Germany, the former socialist GDR. The aim of this study was to test the hypothesis that the changes in East Germany had influenced social inequality in health-related living conditions of 6-year-old children during the ten-year period starting a few months after the German re-unification in 1990. This investigation has not been done before. Methods ======= Study areas ----------- The data for our study are derived from an environmental epidemiological study, organized by the Medical Institute for Environmental Hygiene Duesseldorf and the District Hygiene Institute of Magdeburg. The purpose of this study was to investigate the health outcomes of school beginners influenced by air pollution from 1991 to 2000 \[[@B7]\]. The study was conducted in the cities of Leipzig, Halle, Magdeburg, and several small rural towns. Leipzig is one of the largest cities (with about 492,000 inhabitants) in Saxony. The specific study area (South-West Leipzig) is characterized by old dilapidated housing close to small-scale industry. Halle is an industrialized city in Saxony-Anhalt with 238,000 inhabitants. Magdeburg, with about 236,000 inhabitants, is the capital of Saxony-Anhalt. The rural areas used as reference in the environmental study are the small towns Salzwedel, Gardelegen, Osterburg and Kloetze in the Altmark. Study design and data collection -------------------------------- Since 1991 cross-sectional investigations have been repeated every year until 2000 in East Germany. All boys and girls entering the elementary school from 1991 to 2000 and living in the geographically defined areas of East Germany were eligible to participate. A letter was mailed to the parents asking for participation of the child and for completion of a questionnaire at home. Written consent of the parents of the examined children was requested. During the ten-year period, altogether 25,864 6-year-old children (response 83%) participated in the environmental study. 99.2% of the children in the study (51.2% boys and 48.8% girls) had the German nationality. Every third year the questionnaire investigation was extended to cover the whole city area of Halle, Magdeburg and Salzwedel. Measures -------- ### Parental educational level The main social indicator for this study is the parental educational level. Parental respondents reported their education and the education of their current partner if applicable. These responses served as the basis for calculating the educational level of the most highly educated parent of a child. Parental educational level was classified into three categories by the highest school grade (years) completed by either the mother or the father as follows: less than 10 years = \'low\'; 10 years = \'middle\'; more than 10 years = \'high\' \[[@B8]\]. ### Parental employment status Parental employment status was classified into two categories: mother and/or father \'in paid employment\' (including full-time and part-time jobs) and \'not in paid employment\' (including those parents who were not in the paid labour force, i.e. the income of the child\'s household was not based on regular paid labour). In the following, we will shortly refer to \'not in paid employment\' as \'not employed\'. The data on employment of both mother and father were only available from 1996 to 2000 for this study, because the information on employment status was not asked for before 1996. ### Living space We have also asked for the number of persons living in the children\'s homes and for the dimensions of the homes. The questions were \"How many people are living in the child\'s home?\" and \"How many square meters is the child\'s home?\" Per capita living space was defined by square meter/per person. We divided per capita living space into two groups: \< 20 m^2^(\'small living space\') and ≥ 20 m^2^. ### Health-related living conditions The questionnaire included living conditions which were considered as relevant for children\'s\' health. Parents were asked about the type of heating in the home, whether the house could be described as damp and how far the next street with heavy traffic was. Here, \'damp housing condition\', \'single oven heating\' (single room heating with coke, gas, oil) and \'living at busy road\' (distance to traffic street \< 50 m) \[[@B9]\] are considered as being unfavourable health-related living conditions. ### Region The distribution of the socio-economic status and, similar, of the educational level is different between big cities and rural areas. Therefore, we introduced in the analysis of this study a variable which distinguished urban from rural area as potential confounder. The urban area included the cities Leipzig, Halle and Magdeburg; while the rural area included the small rural towns Salzwedel, Gardlegen, Kloetze and Osterburg. Statistical analyses -------------------- The data were analysed with the SAS statistical software, Version 8.2 \[[@B10]\]. The time course of variables was described by the included graphics. Cross tables were used to show relationships between social variables, regions and educational levels; a chi-square test was used to test associations. Multiple logistic regression analyses were conducted to evaluate the influence of educational level, region and time as independent variables on the health-related living conditions as dependent variables. As to determine an increase or decrease of the difference between lower and higher social status with respect to the dependent variable, the interaction of time and educational level was modelled too: ![](1471-2458-5-64-i1.gif) where p is the probability for the health-related living condition present; M and L are equal 1 if the educational level is \'middle\' or \'low\', respectively, else 0; T (0,1,2,\...,9) represents the year of observation; R is equal 1 for urban area else 0. The estimated model parameters b~i~were presented as adjusted odds ratios together with their corresponding 95% confidence intervals to indicate the impact of the specified independent variable on the particular health-related living condition adjusted for all other independent variables of the regression model. The \'high\' parental educational level served as reference category for both the \'low\' and \'middle\' parental educational level. The influence of the urban area environment was compared to the rural environment. The odds ratio for the factor time refers to a one-year increment. The significance of the interaction term was indicated by the two significance levels p \< 0.05 and p \< 0.01, respectively. Results ======= Descriptive analyses -------------------- Table [1](#T1){ref-type="table"} shows the sample size and the distribution of parental educational level by study region from 1991 to 2000. Figures [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}, [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"}, [6](#F6){ref-type="fig"} present the time courses of the frequencies of parental social indicators and health-related living conditions of 6-year-old children in East Germany from 1991 to 2000. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Time courses of sample distribution of parental educational level classified by the highest school grade completed by either mother or father of 6-year-old children in East Germany from 1991 to 2000 (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10); ::: ![](1471-2458-5-64-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Time courses of prevalence of mother and father not in paid employment of 6-year-old children by parental educational level in East Germany from 1991 to 2000 (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10); ::: ![](1471-2458-5-64-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Time courses of prevalence of 6-year-old children\'s home with small living space (per-capita less than 20 m^2^) by parental educational level in East Germany from 1991 to 2000 (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10); ::: ![](1471-2458-5-64-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Time courses of prevalence of 6-year-old children\'s living under damp housing condition by parental educational level in East Germany from 1991 to 2000 (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10); ::: ![](1471-2458-5-64-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Time courses of prevalence of 6-year-old children\'s home with single oven heating system by parental educational level in East Germany from 1991 to 2000 (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10); ::: ![](1471-2458-5-64-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Time courses of prevalence of 6-year-old children living at busy road (distance closer than 50 m to a traffic street) by parental educational level in East Germany from 1991 to 2000 (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10). ::: ![](1471-2458-5-64-6) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Distribution of parental education (\'low\': less than grade 10, \'middle\': grade 10, \'high\': more than grade 10) of 6-year-old children in East Germany (1991--2000), by study region and year ::: Year Urban^a^ Rural^b^ Total(n) ------ ---------- ---------- ---------- ------ ----- ----- ------ ------ ------ 1991 3125 5.2 41.4 53.4 922 7.7 47.2 45.1 4047 1992 1225 7.2 42.9 50.0 924 5.0 50.5 44.5 2149 1993 983 6.0 43.3 50.7 878 4.6 50.1 45.3 1861 1994 3814 4.9 43.4 51.8 901 5.2 54.8 40.0 4715 1995 1223 5.6 42.0 52.3 746 4.2 51.6 44.2 1969 1996 1097 5.0 41.1 53.9 688 6.0 48.6 45.5 1785 1997 2390 6.2 46.9 46.9 570 5.8 51.6 42.6 2960 1998 891 6.7 50.1 43.2 325 8.6 50.5 40.9 1216 1999 842 5.6 50.4 44.1 359 6.7 52.4 41.0 1201 2000 1837 6.8 47.6 45.6 339 6.5 53.7 39.8 2176 ^a^Leipzig, Halle and Magdeburg; ^b^Salzwedel, Gardelegen, Osterburg and Kloetze. ::: More than 90% of children\'s parents in the study sample (Figure [1](#F1){ref-type="fig"}) received 10 or more years of school education. A complete reverse of proportions of high to middle educational level can be seen since 1997. We found a rising rate of parents not being in paid employment at all parental educational levels (Figure [2](#F2){ref-type="fig"}). Despite the very high rate of not employed parents among the low educated parents in 1996, we observe a further increase of 20% until 2000 for this group. In general, the percentage of children with homes of small living space was decreasing for all parental educational levels, but with the highest decreasing rate for the high educational level and the lowest for the low educational level (Figure [3](#F3){ref-type="fig"}). The proportion of damp housing conditions had partially increased (Figure [4](#F4){ref-type="fig"}) until 1996 in East Germany, but thereafter, reconstruction measures of homes around that time might be responsible for the slight decline of the frequency of damp housing conditions. The difference decreased by about 10% when comparing low educational level to high educational level during the study period. A very clear improvement of the situation regarding single oven heating systems at children\'s homes was found (Figure [5](#F5){ref-type="fig"}) and the related differences in parental educational levels extremely decreased by about 30% when comparing low educational level with high educational level over time. A slight increase of the proportion of children living closer than 50 m to traffic street could be observed (Figure [6](#F6){ref-type="fig"}) for low and middle parental educational levels, whereas for high parental educational level after a slight increase during the first five years the proportion reached in 2000 nearly the same level as in 1991. The difference increased by about 10% when comparing low educational level with high educational level during the study period. Table [2](#T2){ref-type="table"} shows a slight shift of parental education to the lower levels in rural regions compared to urban areas. Explanations for this difference between rural and urban areas could be, first, the fact that the educational level is lower in rural regions compared to larger cities or, second, the consequence of different migration after re-unification or, third, the result of a different response behaviour. Table [2](#T2){ref-type="table"} also demonstrates that the health-related living conditions were consistently better in rural areas compared to urban areas. The strongly significant differences between the three parental educational levels with respect to not employed and unfavourable health-related living conditions over the whole study period are clearly demonstrated in Table [3](#T3){ref-type="table"}. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Distribution of parental education, parental employment status, small living space and health-related living conditions of 6-year-old children in East Germany (1991--2000), by study region ::: Total Urban^a^ Rural^b^ p-value^c^ ---------------------------------------------------- --------- ------------ ------------ -------------- ------ --------- Parental educational level (highest school grade): Low (less than grade 10) 1380 997 5.7 383 5.8 Middle (grade 10) 11113 7730 44.4 3383 50.9 \<.0001 High (more than grade 10) 11586 8700 50.0 2886 43.4 Mother and father not in paid employment 738 599 14.6 139 10.2 \<.0001 Small living space (per-capita \< 20 m^2^) 13570 10133 56.9 3437 50.5 \<.0001 Damp housing condition 2617 2115 12.1 502 7.5 \<.0001 Single oven heating at child\'s home 8617 6725 40.4 1892 29.5 \<.0001 Living at busy road (distance \< 50 m to traffic) 14482 10812 62.0 3670 55.3 \<.0001 ^a^Leipzig, Halle and Magdeburg; ^b^Salzwedel, Gardelegen, Osterburg and Kloetze; ^c^p-value of chi-square test ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Distribution of parental employment status, small living space and health-related living conditions of 6-year-old children in East Germany (1991--2000), by parental educational level ::: Parental educational level -------------------------------------------- ---------------------------- ----- ------ ------ ------ ------ ------ --------- Mother and father not in paid employment 730 157 78.5 408 17.9 165 5.64 \<.0001 Small living space (per-capita \< 20 m^2^) 13163 982 71.2 6514 58.6 5667 48.9 \<.0001 Damp housing condition 2581 295 22.0 1306 11.9 980 8.5 \<.0001 Single oven heating at child\'s home 8472 682 56.0 3993 38.6 3797 34.2 \<.0001 Living at busy road (\< 50 m to traffic) 14238 921 71.3 6785 62.3 6532 56.9 \<.0001 ^a^Low: less than school grade 10, ^b^middle: grade 10, ^c^high: more than grade 10; ^d^p-value of chi-square test ::: Logistic regression analyses ---------------------------- In Table [4](#T4){ref-type="table"}, odds ratios, estimated by logistic regression and, therefore, mutually adjusted for the independent variables of the regression model, quantify the association between the health-related living conditions of the 6-year old children and their determinants parental educational level, time and region, respectively. Differences of change over time between the three educational levels are documented by strata-specific odds ratios, and their significance is indicated by the significance level of the respective interaction term of the regression model. The odds ratios were all significantly greater 1 for not employed, small living space, damp housing conditions, single oven heating and living at busy road when comparing low and middle educational levels to high educational level in accordance with the results in Table [3](#T3){ref-type="table"}. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Adjusted ^§^odds ratios (OR) with 95% confidence intervals (95% CI) of parental employment status and health-related living conditions of 6-year-old children in East Germany (1991--2000) for the determinants parental educational level, time and region, by means of logistic regression analysis ::: Independent variables Dependent variables -- OR (95%CI) --------------------------------------------- ----------------------------------- ----------------------- ----------------------- ----------------------- ----------------------- Parental educational level high (reference) 1.00 1.00 1.00 1.00 1.00 Parental educational level middle 2.89 (1.13--7.39) 1.34 (1.22--1.47) 1.82 (1.58--2.11) 1.55 (1.41--1.70) 1.15 (1.05--1.26) Parental educational level low 24.82 (3.90--158.10) 2.22 (1.78--2.76) 4.28 (3.38--5.41) 4.83 (3.83--6.10) 1.57 (1.27--1.94) Time (parental educational level high) 1.16 (1.04--1.29) 0.85 (0.84--0.86) 1.00 (0.98--1.03) 0.82 (0.80--0.83) 0.98 (0.97--1.00) Time (parental educational level middle) 1.20 (1.13--1.29) 0.89\\ (0.88--0.90) 0.95\\ (0.91--1.00) 0.78\\ (0.77--0.79) 1.01\\ (0.99--1.02) Time (parental educational level low) 1.34 (1.09--1.71) 0.90\\ (0.87--0.94) 0.92\\ (0.88--0.96) 0.71\\ (0.23--0.74) 1.03\* (0.90--1.18) Region rural area^a^ (reference) 1.00 1.00 1.00 1.00 1.00 Region urban area^b^ 1.55 (1.25--1.93) 1.43 (1.35--1.51) 1.78 (1.61--1.97) 1.90 (1.78--2.04) 1.35 (1.27--1.43) § Adjusted for all other factors included in the model and give in the Table. ^a^Region rural area: Salzwedel, Gardlegen, Osterburg and Kloetze; ^b^urban area: Leipzig, Halle and Magdeburg, \* p-value for interaction term of the regression model: p \< 0.05 \\ p-value for interaction term of the regression model: p \<0.01 ::: The odds for not employed increased by 34% for low educational level and by 20% for middle educational level compared to 16% for high educational level per year. Despite the clear difference in change over time, the interaction terms were not significant. The odds for small living space decreased by about 10% for low and middle educational levels compared to 15% for high educational level per year. The differences in change over time were significant in favour for the higher educational level. The odds for damp housing conditions slightly, but significantly decreased over time for low and middle educational levels, but not for high educational level. The prevalence of single oven heating strongly decreased for all three educational levels over time, but the low educational level showed the strongest decline of prevalence. Living at busy road slightly, but significantly increased for low and middle educational levels compared to high educational level over time. If comparing urban areas to rural areas, the adjusted odds for not employed, small living space, damp housing conditions, single oven heating at child\'s home and living at busy road, respectively, were significantly higher in urban areas. Discussion ========== The present study provides new important information on the details of changing social inequality with respect to health-related living conditions of 6-year-old children in the former socialist country, East Germany, on its transition to social market economy. The data reveals that the parental educational level as a main social indicator was consistently and significantly linked to parental employment status, living space and health-related living conditions during the whole study from 1991 to 2000 (Table [3](#T3){ref-type="table"}, Figure [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}, [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"}, [6](#F6){ref-type="fig"}). The time courses of the two social indicators, employment status (both parents not in paid employment) and living space (per capita living space less than 20 m^2^), respectively, developed unfavourably and less favourably, respectively, for the two lower educational levels. For the health-related living condition \'living at busy road\' (distance less than 50 m), we also observed an increasing, almost doubling gap between different social status to the disadvantage of children with low educated parents. The proportions of children living under damp housing conditions or being exposed at home to single oven heating were significantly different between the educational levels at the time of re-unification, but, fortunately, nearly identical after 10 years. The East German society of the former socialist GDR was regarded as characterized by a relatively uniform distribution of resources and living conditions. The standard of living was higher than that in other communist countries, and a comprehensive social insurance system covered medical, disability, unemployment, and other expenses \[[@B11]\]. Education was free and compulsory through 10th grade \[[@B11]\], the economic structure was similar to that in the Soviet Union \[[@B11]\], with state ownership and centralized control \[[@B11]\], income differences between social groups were relatively small, basic health care was equally accessible to all groups of the society \[[@B8]\]. Although living conditions and access to consumer goods in former socialist countries were much more uniform than in the West, disparities between social groups did exist, as shown by our study. The health-related variables of our study were damp housing conditions, single oven heating and living at busy road. Several large epidemiological studies have identified damp living conditions as a major risk factor for respiratory symptoms in children \[[@B12]\]. There are only few publications on health inequalities in GDR and it is difficult to assess whether health inequalities do have different reasons in East Germany \[[@B13]\]. Although equality was a major political goal in the GDR there have been health inequalities comparable to those in Western European countries \[[@B13]\]. Conditions after re-unification were not easy for the East German labour force as the difficulties of adjustment to the Western system caused many hardships, including unemployment \[[@B14]\]. By 1991, a massive decline in employment with a loss of about 3.5 million jobs (35% of the labour force) by the end of 1992; unemployment rose from almost zero at the beginning of 1990 to 15.4% of the labour force in 1992 \[[@B15]\]. In our study (Table [4](#T4){ref-type="table"}), odds for parents not in paid employment increased in the average per year by 34% for \'low\' educated parents, by 20% for \'middle\' and by 16% for \'high\'. Since 1989, about 1.7 million East Germans left for the West, the birth rate fell by about 60% in the period from 1989 to 1994, and the number of marriages and divorces declined sharply \[[@B16]\]. In 1991 (Figure [1](#F1){ref-type="fig"}), about 50% of the parents had an educational level of more than 10 years of school, but, after 1996, this proportion dropped to 40% and the proportion of parents with 10 years school education raised to 50%. 1997 is the first study year where children born after the German re-unification have been investigated. Although the size of the investigation areas were not changed during the time course of the study, the number of investigated children decreased from 4047 children in 1991 to 2960 in 1997. This massive decrease can only partly be explained by the 10 percent decrease in response rate (93% in 1991 and 83% in 1997), but mostly by the massive decline in birth rates and migration to other areas. 1997 is the same year when the percentage of middle educated parents outnumbered the percentage of highly educated parents for the first time. This means that the decrease in numbers due to declining birth rates and migration was steeper to children with highly educated parents than for children with middle educated parents. This observation has also been made by other groups \[personal communication\]. Around the year 1994, we observed an increase of \'small living space\' which was caused by the different composition of the surveys between 1993 and 1995. While the survey in 1993 had a proportion of 47% of school beginners in rural areas, the survey in 1994 showed just a fraction of 19%, whereas this fraction increased again to 38% in the survey 1995 (see Table [1](#T1){ref-type="table"}). The cross-sectional design of our study could only identify associations between social status and health-related living conditions, but could not provide the evidence of a causal relation by itself. For instance, the causal chain from a low parental educational level over low income to a cheap or unfavourable (with respect to children\'s health) home could not be explored. Two further limitations of this study need to be discussed. First, in the analyses presented in this paper we have considered social status differences by using parental educational level (of the most highly educated parent) for classifying children. The information about household income and occupational status of parents had not been asked in this environmental study. But an expert committee in Germany also recommended using educational level as a measure for social status in epidemiological and social medical studies \[[@B17]\]. Education is an important determinant of individuals\' work and economic circumstances, which are themselves linked to health through specific work conditions and levels of consumption \[[@B18]\], related to health outcomes through its influence on lifestyle behaviours (e.g. exercise, diet), problem-solving capacities and values (e.g., importance of preventive health behaviours) \[[@B19]\]. Education as \'year of education completed\', is a core variable in the MONICA project \[[@B13]\]. Nevertheless, the choice of a single indicator of parents\' socioeconomic status may be subject to debate \[[@B20]\]. The educational level paralleled factors which are considered to be characteristic of modern domestic comfort, such as central heating, and was inversely related to the degree of crowding in the home. The second limitation was that the environmental study from which the data for this study were derived focused on airway diseases and atopic manifestations and, therefore, the selection of living conditions of the 6-year old children was restricted to their relevance for these health outcomes. Other equally important health-relevant living conditions were missing in our study, such as nutrition or access to health services. Conclusion ========== The results of our repeated cross-sectional studies in East Germany are in agreement with the well known fact of a strong association between parental educational level and health-related living conditions of children. Furthermore, we observed that the domestic situations of small living space, damp housing condition and single oven heating system improved from 1991 to 2000; while parental employment status (1996--2000) and living at busy road (1991--2000) did not improve, but even deteriorated. Our basic hypothesis that the changes in East Germany had influenced social inequality was confirmed. We found decreased social inequality for single oven heating and damp housing condition; while increased social inequality was seen for parental employment status, small living space and living at busy road. These findings lead to the conclusion that the lower social class is often, but not always the looser of large socio-economic changes as happened in East Germany. The \"winning\" effect seems to be bound on strong general improvements, such as the replacement of single oven heating by central heating. A further level of investigation may be to ask how these social differences in health-related living conditions modify the association between exposure to environmental pollution and health. We know little about the effect modification of social inequality on environmental hazards or about the impact of social inequality on environment-related disease. In future, a closer co-operation between environmental-epidemiological and socio-epidemiological research would be needed. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= XDP wrote the paper and performed the statistical analysis. UK coordinated the epidemiological cross-sectional studies in East Germany. UR designed the study, assisted the statistical analysis and helped to draft the manuscript. All authors reviewed the final version of the manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2458/5/64/prepub>	PubMed Central	2024-06-05T03:55:59.502734	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180456/", "journal": "BMC Public Health. 2005 Jun 8; 5:64", "authors": [ { "first": "Xianming", "last": "du Prel" }, { "first": "Ursula", "last": "Krämer" }, { "first": "Ulrich", "last": "Ranft" } ] }
PMC1180457	Background ========== Tetrandrine (TET), a bisbenzylisoquinoline (Fig. [1a](#F1){ref-type="fig"}), isolated from the root of the plant Stephania tetrandrahas a number of potential medicinal properties. These include blockage of voltage-gated Ca^2+^channels \[[@B1]\], large-conductance Ca^2+^activated K^+^(BK) channels, and intracellular Ca^2+^pumps \[[@B1]-[@B6]\]. TET also has anti-inflammatory \[[@B2],[@B7]\] and anti-cancer activities \[[@B8],[@B9]\]. TET stimulates prostaglandin (PG)E~2~production by macrophages \[[@B10]\], probably after first releasing the substrate, arachidonic acid (AA) by altering phospholipase (Plase) activities. TET also induces apoptosis in many cell types including human leukemic (U937), human lung carcinoma (A549), human hepatoblastoma (HEPG2), neuro 2a mouse neuroblastoma and rat glioma cells (C-6) \[[@B11]-[@B14]\]. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### a): Tetrandrine (TET), isolated from the plant Stephania tetrandra(Structure reproduced with permission from G. Wang \[9\]) and b): Thapsigargin(THAP), isolated from the plant Thapsia garganica(Structure reproduced with permission from S. B. Christensen \[15\]). ::: ![](1471-2210-5-12-1) ::: Thapsigargin (THAP), a hexaoxygenated tetracycle sesquiterpine lactone, (Fig. [1b](#F1){ref-type="fig"}) isolated from the plant Thapsia garganica, also has a number of potential medicinal applications \[[@B15]\]. However, THAP is classified as a weak tumor promoter as measured in the two-stage model of mouse skin carcinogenesis \[[@B16]\]. Nevertheless, THAP \[[@B17]\] and its enzymatically modified analog \[[@B18]\] have been proposed as targeted therapy for prostate cancer. THAP, like TET, blocks intracellular calcium pumps resulting in increased cytoplasmic Ca^2+^, (\[Ca^2+^\]~i~) \[reviewed in 15\]. It also affects ion channels. THAP induces a Ca^2+^-dependent release of AA from \[^3^H\]-AA labelled macrophages and stimulates AA metabolism in the rat peritoneal macrophages \[[@B19]\]. THAP induces apoptosis in many cells including human neuroblastoma, colon cancer and prostate cancer cells and thymocytes \[[@B17],[@B20]-[@B22]\]. Based on the stimulation of AA release by known cancer chemopreventative agents, I have proposed that AA release by cells is associated with cancer chemoprevention \[[@B23]-[@B27]\], possibly, but not necessarily, by activating a secreted tumor suppressor phospholipase A~2~(PLA~2~) \[[@B28],[@B29]\]. In this report, evidence is presented that TET, a potential cancer chemopreventive compound, and THAP, a weak tumor promoter that also possesses potential cancer preventative properties for androgen-independent prostate cancer, both stimulate AA release from human colon carcinoma and rat liver cells. Both compounds also stimulate prostacyclin (PGI~2~) production in rat liver cells. The release of AA and AA metabolites appears to be initiated by different mechanisms. Results ======= TET and THAP release AA from human colon carcinoma (HT-29) cells and rat liver (C-9) cells in a concentration-dependent fashion (Fig. [2a -- 2d](#F2){ref-type="fig"} respectively). As little as 0.1 to 0.3 μM THAP stimulates AA release. With both HT-29 and C-9 cells, THAP is about 10 to 30 times more potent than TET. Characterizations of these effects are shown in Table [1](#T1){ref-type="table"}. Pre-incubation with actinomycin D partially inhibits stimulation by THAP but does not affect stimulation by TET. TET\'s stimulation of AA release does not require new mRNA synthesis, whereas THAP\'s stimulation does. As shown below, THAP\'s stimulation is mediated, in part, by the Ca^2+^-dependent PLA~2~an induced enzyme. The absence of extracellular Ca^2+^partially inhibits THAP\'s stimulations but does not affect the AA release stimulated by TET. Depolarization of the cells with 100 mM KCl does not affect TET\'s stimulations, but does partially inhibit release of AA stimulated by THAP. Pre-incubation with the L-TYPE Ca^2+^channel blocker, diltiazem, has no effect on the actions of either TET or THAP in HT-29 or C-9 cells. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Effect of a) TET and b) THAP on AA release from HT-29 cells. Effect of c) TET and d) THAP on AA release from C-9 cells. The data are representative of several experiments. The analyses were performed with triplicate or quadruplicate dishes. \* = Statistically significant vsMEM/BSA ::: ![](1471-2210-5-12-2) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Effects of Diltiazem (50 μg/ml), Actinomycin D (1 μg/ml), EGTA (1 mM), KCl (100 mM), and BABTA/AM (16 μg/ml) on AA Release from HT-29 or C-9 Cells stimulated by TET or THAP. ::: Some biological effects\* Agent tested Action of test agent AA Release ------------------------------- ------------------ ------------------------------- ---------------- ------ TET Diltiazem Blocks L-type Ca^2+^channels NI NI Ion Channels Actinomycin D Inhibits RNA synthesis NI NI Apoptosis EGTA Chelates extracellular Ca^2+^ NI NI Depolarization 100 mM KCl Depolarizes NI NI \[Ca^2+^\]~I~ BAPTA/AM Chelates \[Ca^2+^\]~i~ NI \\ THAP Diltiazem Blocks L-type Ca^2+^channels NI NI Ion Channels Actinomycin D Inhibits RNA synthesis ↓ ↓ Apoptosis EGTA Chelates extracellular Ca^2+^ ↓ ↓ Depolarization 100 mM KCl Depolarizes ↓ ↓ \[Ca^2+^\]~i~ BAPTA/AM Chelates \[Ca^2+^\]~I~ ↓ \\ \* = References in text NI = No Inhibition (or stimulation) ↓ = Inhibition: statistically significant \\ = BAPTA/AM (16 μg/ml) stimulates AA release (6.17 ± 0.088 (4)), MEM/BSA control vs(11.6 ± 0.322 (4)), BAPTA/AM (16 μg/ml). Thus, the effect of BAPTA/AM on C-9 cells is not recorded. ::: The role of \[Ca^2+^\]~i~in these stimulations is striking. While pre-incubation with the \[Ca^2+^\]~i~chelator, BAPTA/AM (16 μg/ml), does not affect TET\'s stimulation of AA release from HT-29 cells, such treatment does inhibit THAP\'s stimulation of AA release from the HT-29 cells (Fig. [3](#F3){ref-type="fig"}, Table [1](#T1){ref-type="table"}). Thus, a role for intracellular Ca^2+^pumps is strongly suggested only for the action of THAP. Ohuchi et al\[[@B19]\] had shown that \[Ca^2+^\]~i~increased within 2 min after treatment with THAP as measured by fluorescence changes in Quin 2 loaded peritoneal cells. A similar rise in \[Ca^2+^\]~i~has been measured after administration of THAP to thymocytes, rat liver microsomes \[[@B30]\] and in macrophages, astrocytoma cells, fibroblasts and human tumor lymphocytes \[[@B31]\]. The findings that the increased AA release is stimulated 5 minutes after incubation of HT-29 and C-9 cells with THAP is consistent with a role for increased \[Ca^2+^\]~i~in the activities of THAP (Fig [4a, 4b](#F4){ref-type="fig"}). ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Effect of BAPTA/AM, 16 μg/ml, on AA release by a) TET and b) THAP from HT-29 cells. The data are representative of two experiments, each with similar results. The analyses were performed on quadruplicate dishes. \* = Statistically significant vsMEM/BSA. ::: ![](1471-2210-5-12-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Time-course of AA release from a) HT-29 and b) C-9 cells by TET (50 μM) and THAP (2 μM). The analyses were performed on duplicate dishes. After the 60 minutes incubation, the AA release by TET is not statistically significant. TET\'s stimulation of AA release from HT-29 cells was not done. ::: ![](1471-2210-5-12-4) ::: While both cells express COX activity \[[@B32],[@B33]\], only the major product of COX activity (PGI~2~) can be quantitatively estimated at the low cell densities used in these studies. Both TET and THAP stimulate PGI~2~(measured as 6-keto PGF~1α~) production in these C-9 cells (Fig. [5a, 5b](#F5){ref-type="fig"}). As with AA release, THAP is at least 10-times more potent at stimulating PGI~2~production than TET. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Effects of a) TET and b) THAP on PGI~2~(6-keto-PGF~1α~) production in rat liver cells (C-9). The data are representative of several experiments. The analyses were performed with triplicate dishes. \* = Statistically significant vsMEM/BSA. ::: ![](1471-2210-5-12-5) ::: The likely role of \[Ca^2+^\]~i~in the stimulation of PGI~2~production by THAP is shown in Fig. [6](#F6){ref-type="fig"}. Chelation of the increased \[Ca^2+^\]~i~with BAPTA/AM completely inhibits the stimulation by THAP but has no effect on the stimulation with TET. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Effect of BAPTA/AM, 16 μg/ml on THAP\'s stimulated PGI~2~production by rat liver cells. The data are representative of two experiments, each with similar results. \* = Statistically significant vsTHAP or BAPTA/AM. \\ = Statistically significant vsTHAP. ::: ![](1471-2210-5-12-6) ::: Discussion ========== It has been proposed that the cancer chemopreventative properties attributed to TET reside in its ability to effect ion channels which leads to inhibition of cell proliferation and apoptosis \[[@B9]\]. Retinoic acids, tamoxifen, PPAR-agonists, (e.g GW-7845), some non-steroidal anti-inflammatory drugs, vitamin D~3~, anti-oxidants, (e.g resveratrol and caffeic acid phenylester) and statins all release AA from cells in culture \[[@B23]-[@B26]\]. All have been shown to have cancer preventative properties. Results of studies on correlation, if any, of AA release and cancer chemoprevention by TET or THAP have not been published. For the purpose of this study, however, I have postulated that the cancer chemopreventive properties of the test agent are causally related to the capacity of that ligand to release AA. I have also suggested that most, if not all, of these agents at μM concentrations, may be intercalating into the cell membrane and releasing AA as a result of activated PLAse activity \[[@B25],[@B26]\]. The mechanism of action of non-steroidal anti-inflammatory drugs that leads to apoptosis and cancer chemoprevention also involves stimulation of AA release and is associated with sphingomyelin to ceramide conversion \[[@B34]\]. The AA release probably results in ceramide-mediated apoptosis \[[@B35]\]. Both TET and THAP stimulate the release of AA from human colon carcinoma and rat liver cells. TET does prevent cancer \[reviewed in 8\] but THAP is considered to be a weak tumor promotor as measured in the two-stage model of mouse skin carcinogenesis \[[@B19]\]. THAP and some of its analogs, however, do induce apoptosis in androgen-independent prostatic cancer cells \[[@B17],[@B18],[@B36]\]. Apoptotic effects of THAP also have been reported in thymocytes and mouse lymphomas, including the WRH17-2 and WHB12 cell lines \[[@B17],[@B20]-[@B22]\]. Clear differences in the pathways of AA release stimulated by TET and THAP were found in this study. TET stimulates AA release (Fig. [2](#F2){ref-type="fig"}) in HT-29 and C-9 cells and PGI~2~production in C-9 cells (Fig. [5](#F5){ref-type="fig"}). It has been reported that TET blocks voltage-gated Ca^2+^channels \[[@B1]-[@B7]\] and depolarizes the cells \[[@B8]\], but blockage of these channels does not affect AA release (Table [1](#T1){ref-type="table"}). Nor is the AA release stimulated by TET blocked by pre-incubation of the cells with actinomycin D, in the presence of 100 mM KCl or in the absence of extracellular Ca^2+^. It is interesting to note that cycloheximide (178 μM), did not affect TET\'s induced apoptosis in human lymphoblasts (CEM-C7) \[[@B37]\]. TET\'s stimulation of AA release is not affected by the \[Ca^2+^\] chelator, BAPTA/AM (Fig. [3](#F3){ref-type="fig"}). In contrast, THAP\'s release of AA by HT-29 cells is blocked by BAPTA/AM (Fig. [3](#F3){ref-type="fig"}), is inhibited by pre-incubation with actinomycin D, is inhibited by EGTA and is inhibited when the cells are incubated in 100 m M KCl (Table [1](#T1){ref-type="table"}). That the rise in \[Ca^2+^\]~i~resulting from treatment with THAP appears to be associated with the stimulation of AA release is suggested by the relatively early increase, after 5 minutes, of AA release (Fig. [4](#F4){ref-type="fig"}). Both TET and THAP stimulate PGI~2~production in the C-9 cells (Fig. [5](#F5){ref-type="fig"}), but only the PGI~2~production stimulated by THAP is inhibited by BAPTA/AM (Fig [6](#F6){ref-type="fig"}). The relationship between AA release and cancer chemoprevention by TET and THAP could be explained, in part, by implicating the PLA~2~enzymes that catalyze the release of AA from phospholipids. The tumor suppressing sPLA~2~may be only one of sixteen structurally different PLA~2~enzymes \[[@B38],[@B39]\]. If the tumor suppressor genes are overexpressed or activated, at least two cancer preventative pathways may result, one from the activity of the group 11A tumor suppressor \[Reviewed in \[39\]\] and a second from the AA released. The proximity of the PLA~2~to the AA-esterified phospholipid after treatment of the cells is of importance, and may depend on the location of binding of the test agent. For example, celecoxib binds at the upper hydrocarbon core, close to the phospholipid head groups whereas rofeco xib binds at the polar head groups of the membrane \[[@B40]\]. Celecoxib and rofecoxib differ in their release of AA \[[@B41]\]. TET may activate a tumor suppressing sPLA~2~, while THAP induces the Ca^2+^-dependent cPLA~2~. Both would lead to AA release. Conclusion ========== Cells treated with tetrandrine and thapsigargin share several common features including pathways that lead to induction of apoptosis. These properties probably reflect their interaction with cell membranes and the altered expression of signaling processes. One reaction that does not appear to be shared is deesterification of a phospholipid by a PLA~2~. Based on the effects of inhibition of mRNA synthesis, 100 mM KCl, extracellular and especially intracellular Ca ^2+^, thapsigargin\'s release is mediated, in part, by a Ca ^2+^-dependent PLA~2~, whereas tetrandrine\'s stimulated release is mediated by a secretory PLA~2~. In addition to the altered signaling properties that accompany the membrane intercalation, both tetrandrine and thapsigargin release biologically active AA. Methods ======= The rat liver (C-9 cell line) were purchased from the American Type Culture Collection (Manassas, VA, USA) and the human colon carcinoma (the HT-29 cell line) was obtained from Dr. Basil Rigas, Chief, Division of Cancer Prevention, SUNY at Stony Brook, NY, USA. They were maintained in Eagle\'s minimum essential medium (MEM) supplemented with 10% fetal bovine serum. \[^3^H\] AA (91.8 Ci/mmol) was obtained from NEN Life Science Products, Inc. (Boston, MA, USA). BAPTA/AM, the \[Ca^2+^\]~i~chelator, diltiazem, the L-type Ca^2+^channel agonist, tetrandrine and thapsigargin were purchased from BIOMOL International, (Plymouth, PA, USA). All other reagents were from Sigma Chemical Co. (St. Louis, MO, USA) or Calbiochem, (San Diego, CA, USA). Two days prior to experiments, the HT-29 or C-9 cells were treated with 0.25% trypsin-EDTA and, after addition of minimal essential media (MEM) containing 10% fetal calf serum, the floating cells were seeded on to 35 mm culture dishes. The plating densities varied from 0.1 to 0.5 × 10^5^cells/35 mm dish. The freshly seeded cultures were incubated for 24-h to allow for cell attachment. After decantation of MEM containing the fetal bovine serum, 1.0 ml fresh MEM containing 10% fetal bovine serum and \[^3^H\] AA (0.2 mCi/ml) were added and the cells incubated for another 24-h. The cells were washed 4 times with MEM and incubated for various periods of time with 1.0 ml of MEM containing 1.0 mg BSA/ml (MEM/BSA) and different concentrations of each compound. The culture fluids were then decanted, centrifuged at 2000 × g for 10 min, and 200 μl of the supernate counted for radioactivity. The MEM/BSA values are the control values. Radioactivity recovered in the washes before the incubation was compared to input radioactivity to calculate the % radioactivity incorporated into the cells. For experiments on the effects of actinomycin D, the HT-29 or C-9 cells were washed with MEM 4 times, incubated with actinomycin D (1 μM) for 2-h, washed once and incubated again for 6-h in the presence or absence of actinomycin D (1 μM). For the effects of BAPTA/AM, the pre-incubation was 50 minutes. For PGI~2~production, 1.0 ml of MEM supplemented with 10% fetal bovine serum, void of \[^3^H\] AA, was added after the first 24-h incubation. The cells were incubated for another 24-h, washed three times with MEM, then incubated with the compounds in MEM/BSA for various periods of time. The culture fluids were decanted and analyzed for 6-keto-PGF~1α~, the stable hydrolytic product of PGI~2~, by radioimmunoassay \[[@B42]\]. The \[^3^H\] AA release is presented as a percentage of the radioactivity incorporated by the cells. Except for the time-course experiments, which used duplicate dishes, three to five culture dishes were used for each experimental point. The data are expressed as mean values ± SEM. The data were evaluated statistically by the unpaired Student\'s t-test. A Pvalue \< 0.05 was considered significant. Acknowledgements ================ My thanks to Hilda B. Gjika for preparation of the manuscript and to Dr. Armen H. Tashjian, Jr., Department of Genetic and Complex Diseases, Harvard School of Public Health, for his continuing interest in these studies.	PubMed Central	2024-06-05T03:55:59.505794	2005-6-24	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180457/", "journal": "BMC Pharmacol. 2005 Jun 24; 5:12", "authors": [ { "first": "Lawrence", "last": "Levine" } ] }
PMC1180458	Background ========== Vasectomy is recognized as a simple, safe, and effective contraception method. However, the occlusive and contraceptive effectiveness of the procedure varies widely according to the surgical technique used to occlude the vas deferens.\[[@B1]\] Ligation with suture material and excision of a small vas segment is believed to be the most common method used world-wide.\[[@B2]\] The risk of occlusive failure with this technique has been traditionally considered to be in the order of 1% to 5%.\[[@B3]\] However, recent studies have shown that the risk could be much higher, ranging from 8% to 13%, based on data from semen analyses.\[[@B4],[@B5]\] The risk of contraceptive failure may also be unacceptably high. A study involving 1052 men in Nepal showed that within 3 years after vasectomy 4.2 % had an unplanned pregnancy.\[[@B6]\] A similar failure rate (4.1%) was also found in Vietnam after more than 5 years of follow-up.\[[@B7]\] In a study conducted in China, among 1,555 couples using vasectomy as a contraceptive method, the risk an unplanned pregnancy was 9.5% after 5 years.\[[@B8]\] Alerted by this information Family Health International (FHI) and EngenderHealth, two non-governmental organizations promoting best practices in family planning worldwide, initiated an international research program to evaluate the effectiveness of alternative surgical techniques to LE. They recently completed two major studies. The first was a multicenter randomized controlled trial (RCT) comparing occlusion by suture LE with versus without fascial interposition (FI). The interim analysis showed a clear advantage of FI and recruitment was halted in May 2001.\[[@B9]\] At that time 841 men had been enrolled. Final results from this cohort were published recently.\[[@B10]\] Using a definition of failure as \> 5 million motile sperm / mL at 14 weeks or later or the presence more than 100,000 sperm with any motility at 26 weeks or later, they found 24 (5.9%) failures in the FI group versus 53 (12.7%) in the non-FI group. Most of the failures appeared to be due to early recanalization. \[[@B10]\] Given the results of the RCT showing relatively high failure rates even with FI, the second study, an exploratory observational study of cautery of the vas lumen,\[[@B11]\] was initiated based on the recommendations from the Expert Consultation on Vasectomy Effectiveness, a meeting organized by FHI and EngenderHealth.\[[@B12]\] This study was conducted at four centers that routinely used cautery for vas occlusion. Frequency of semen analyses and laboratory procedures were similar in both the cautery study and the RCT, but follow-up was only through 24 weeks and some of the sites could not provide data on sperm motility. Using a definition of early failure as \>10 million sperm / mL at 12 weeks regardless of motility, the risk of early failures was 4/389 (1.0%). Applying the same definition of failure to the RCT data set, early occlusive failure risks were 4.9% and 12.5% in the groups with and without FI, respectively.\[[@B13]\] Though the results are encouraging for the use of cautery in vasectomy, they must be interpreted with caution based on this non-randomized comparison. In addition, while FI was showed to be important in improving vasectomy occlusion success when LE are the primary occlusion method, this exploratory study of cautery cannot definitively confirm that FI is as useful when cautery is used as the primary occlusion method. However, these findings support the results from numerous large case series showing that the occlusive effectiveness of cautery, especially when combined with FI on the prostatic end, is high, with failures well below 1%. \[[@B1],[@B14]-[@B21]\] In December 2003, FHI and EngenderHealth organized a three-day expert consultation on vasectomy techniques and services (Summary available on FHI\'s web site at <http://www.fhi.org>). The vasectomy experts recommended that 1) training of vasectomy providers emphasize the potential increased effectiveness of vasectomy when FI is added to the standard technique of LE; 2) providers now using simple LE consider adopting FI, with appropriate training as needed; 3) where resources, training, and logistical support are available, cautery can be considered as an effective and safe method to block the vas. FI and/or cautery are already widely used in developed countries.\[[@B22]\] This might not be the case in developing countries\[[@B2]\] but there are no good data on the specific techniques used at the level of one country or a region in the developing world. It is just recently, in the 2003 edition of No-scalpel vasectomy: An illustrated guide for surgeons, that EngenderHealth has started to promote FI along with LE as the preferred occlusion technique, and cautery, with or without FI, as the alternative.\[[@B23]\] The rapid adoption of the most effective vasectomy occlusion techniques is essential considering that the lack of resources in most developing countries precludes most men from verifying the success of their vasectomy with semen analysis. EngenderHealth currently recommends that men use another form of contraception during the first 12 weeks after vasectomy. However, many barriers could prevent the adoption of these techniques in the low-resource settings. Firstly, while \"low tech\" hand-held battery-driven thermal cautery devices are available, the instruments and supplies needed to perform cautery and FI (cautery device, tips, batteries, and suture material) may be difficult to procure, to use properly -- including to sterilize adequately -- and to maintain for further use. Secondly, performing cautery and FI require additional surgical skills over LE. Specific hands-on training is essential to master these techniques although cautery is much easier to perform adequately than FI. Thirdly, personal, professional, social constraints, or country standards may limit the adoption of new surgical techniques, especially if the changes are imposed without adequate scientific justification and appropriate training. Accurate information on the current situation is essential to evaluate these barriers and to successfully introduce and adequately evaluate the adoption of new vasectomy occlusion techniques. This paper reports on the vasectomy techniques currently used in major vasectomy clinics or programs in Cambodia, Thailand, India, Nepal, and Bangladesh, and on the factors that could facilitate or obstruct the introduction of vasectomy occlusion techniques using cautery and FI in these countries. Methods ======= Countries were selected based on the involvement of national or international organizations supporting family planning programs in each of the countries, and geographical proximity to each other. These countries represent a wide range of cultural and religious backgrounds facilitating the generalization of the results. In all the selected countries, the use of vasectomy is currently much lower than the use of tubal ligation (Table [1](#T1){ref-type="table"}). This represents a window of opportunity in most of these countries for promoting best practices related to male sterilization and to enhance the popularity of vasectomy. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Proportion of women currently married or in union using sterilization as a family planning method in selected Asian countries. ::: Countries Using Male Sterilization % Using Female sterilization % Year(s) of Assessment ---------------- -------------------------------- ---------------------------------- --------------------------- Cambodia \- 1.5 2000 Thailand 2.0 22.0 1996--7 India 1.9 34.2 1998--9 Nepal 6.3 15.0 2001 Bangladesh 0.5 6.7 1999--2000 From: Family Planning Worldwide 2002 Data Sheet. Population Reference Bureau. Washington DC, June 2002. ::: Officers and program managers of national or international organizations involved with vasectomy in each country were contacted by e-mail beforehand in order to inform them about the purpose of the project and to seek their help in identifying (a) the most relevant individuals to meet, and (b) the most relevant vasectomy centers to visit. Based on this information, a field visit schedule was planned. The targeted international organizations were the United Nations Population Fund (UNFPA), Family Health International (FHI), EngenderHealth, International Planned Parenthood Federation (IPPF), Marie Stopes International (MSI), and Marie Stopes Clinic Society (MSCS). These organizations provide support to family planning programs in one or more of the selected Asian countries. Data were collected between December 2003 and February 2004. In each country, 5 to 11 key informants such as vasectomy providers, directors of family planning clinics, national level family planning program managers, and/or managers of international organizations involved in family planning programs were interviewed. Information about the current situation with vasectomy, locally and in general in the country, and the interest and open-mindedness regarding the use of cautery and/or FI was gathered. Visits to 3 to 6 major vasectomy centers in urban and rural areas were done. Vasectomy techniques performed in each center visited were precisely described based on direct observation, when possible, using a data collection grid previously used in a FHI/EngenderHealth study \[[@B11],[@B13]\]. In addition, with the consent of local authorities and the patients involved, vasectomy procedures performed by the providers were videotaped for further meticulous analyses. Audio-visual material on vasectomy techniques using cautery and FI was presented and hands-on demonstrations were performed by one author (ML) according to the local situation and interest. ML has performed over 9000 vasectomies over the last 20 years, most using NSV combined with various occlusion techniques including cautery and FI. \[[@B1],[@B15],[@B24]-[@B26]\] Cautery handles and tips manufactured in USA (Advance Meditech International) and Canada (Walsh Medical Devices Inc.) were brought along in order to assess the feasibility of carrying out procedures under local conditions. Results ======= Current state of vasectomy in visited countries ----------------------------------------------- ### Cambodia Vasectomy was almost non-existent in Cambodia until 2000. Since then RACHA (Reproductive and Child Health Alliance), a program initially managed by EngenderHealth which has grown into a Cambodian non governmental organisation (NGO), has promoted vasectomy in selected provinces in the countries. A total of 40 providers have been trained to perform both tubal ligation by minilaparotomy and no-scalpel vasectomy. In 2001, 2002, and 2003, 416, 199, and 281 vasectomies were performed, respectively \[[@B27]\]. These numbers, still very marginal, varied greatly in each center, and depended on the funding available to provide free access to vasectomy and to reimburse patients\' travel expenses to the vasectomy center. ### Thailand In Thailand vasectomy is available mainly through PDA (Population and Community Development Association), a NGO providing and promoting male sterilization since 1976. PPAT (Planned Parenthood Association of Thailand), government hospitals, and private clinics also provide vasectomy services. Vasectomy incidence has steadily decreased over the past 20 years. At PDA, from a peak of 7,836 vasectomies performed in 1983, it felt to a low of 767 in 2003, a ten-fold decrease. \[[@B28]\] Since 2003, sterilization (male and female) is not subsidized anymore as national government program. National statistics on vasectomy were not available. ### India India possesses a structured and comprehensive national program promoting the use of no scalpel vasectomies (NSV). This program is funded by the United Nation Population Fund (UNFPA) with the Government of India providing centers for training and making available the necessary infrastructure at the training sites. As of December 2002, 309 NSV courses had been organized all across the country involving 51 states trainers, 58 district trainers, and 1,080 trainees. A total of 153,687 procedures were performed during these training sessions. \[[@B29]\] A national NSV meeting is organized on an annual basis by the NSV Surgeons of India <http://www.nsvsi.com>. In 2003--04, 113,092 vasectomies were performed with an incidence of 0.043 new acceptors per 100 women of reproductive age. During the same period 4,873, 530 tubal ligations were performed. ### Nepal Providing vasectomy service to the population is embedded in the Nepal National Policy, Strategy, and Plans. Each year a projected number of cases is calculated for each district and region. Among the countries visited, Nepal has the highest incidence and prevalence of vasectomy. Nevertheless, the incidence has been decreasing over the recent years due to increased access to other contraceptive methods. In 2002--2003 surgeons performed 20,588 vasectomies with an incidence of 1.64 new acceptors per 100 women of reproductive age.\[[@B30]\] ### Bangladesh There is a national program supporting family planning including vasectomy services in Bangladesh. The use of vasectomy has been steadily decreasing from 151,125 vasectomies performed in 1985--86 to a low of 7,603 in 1996--97. \[[@B31],[@B32]\] Since then numbers have slowly increased in the remaining of the decade to reach 43,203 in 2002--03 exceeding the number of tubal ligations (32,761).\[[@B31]\] Trends forecast an even higher number of vasectomies in 2003--04. However, the incidence of 0.12 new vasectomy acceptors per 100 women of reproductive age in 2002--03 remains relatively low. Vasectomy surgical techniques currently used in visited Asian countries ----------------------------------------------------------------------- Overall 21 vasectomy centers were visited in five countries (Table [2](#T2){ref-type="table"}). Almost all facilities were training centers and most vasectomies observed were performed by certified trainers or master trainers. In many centers, the first author (ML) assisted the surgeon and/or performed parts of NSV and occlusion of the vas using thermal cautery and/or FI. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Vasectomy surgical techniques used in visited Asian countries ::: ------------------------------------------------------------------------------- Centers Visited Procedure\ NSV Cautery Excision (cm) Suture\ FI observed material ----------------- ------------ ----- --------- --------------- ---------- ----- Cambodia 1 No Yes No 1--1.5 V 3-0 Yes 2 No Yes No 1 V 3-0 Yes 3 No Yes No 1 V 3-0 Yes Thailand 1 Video Yes No 2--3 S Yes 2 Yes ± No 1.5 S ± 3 Yes Yes No 1 S 3-0 Yes India 1 Yes Yes No 1 C 10 No 2 Yes Yes No 1 S 2-0 Yes 3 No Yes No 1 S 2-0 No 4 Yes Yes No 1 S 2-0 Yes Nepal 1 Yes Yes No 1 S 2-0 Yes 2 Yes Yes Electro 0.5--1 S 1-0 No 3 Yes Yes No 1 S 2-0 No 4 No ± No 1--1.5 S 1-0 No 5 Video Yes Thermal 1 S 2-0 Yes 6 Yes Yes No 1 S 2-0 No Bangladesh 1 Yes Yes No 1 S 3-0 ± 2 Yes Yes No 1 S 1-0 No 3 Yes Yes No 1 S 3-0 ± 4 Yes Yes No 1 S 2-0 ± 5 Yes Yes No 0.5 S 2-0 ± ------------------------------------------------------------------------------- NSV = No Scalpel Vasectomy, FI = Fascial Interposition, V = Vicryl, S = Silk, C = Cotton. ::: ### Isolation of the vas To isolate and expose the vas, NSV procedure combined with vasal block was performed in all centers visited. However, the technique was not equally mastered by all providers and trainers, revealing the needs for some training updates. In general the quality of the instruments was acceptable but in some centers performing NSV technique properly was limited by the use of inadequate oval-designed ring clamps or blunted dissecting forceps. ### Occlusion of the vas Simple ligation with suture material and excision of a small vas segment (LE) was performed in nearly all centers visited. There were two exceptions, both in Nepal. In one hospital-based center in addition to LE, the tips of the stumps were electro cauterized. In another, thermal cautery combined with FI interposition was used. The frequency of combined use of FI with LE varied from on country to the other. However, at all sites visited, nearly all surgeons were using or were taught the FI technique evaluated \[[@B10]\] and promoted \[[@B23]\] by EngenderHealth and FHI (Figure [1](#F1){ref-type="fig"}). This technique involves ligating the vas sheath around the abdominal stump thus covering the testicular end with the fascia. \[[@B23]\] In Cambodia where the vasectomy program is rather young, much emphasis has been put on performing FI along with LE. Although no cases were observed, it was said to be performed routinely in the three centers visited. In Thailand, the majority of vasectomy services are currently provided by PDA surgeons who routinely combine FI with LE. In the South Asian countries the situation was different. Although all trainers and providers met during the visit were aware of FI and the majority had some training in performing the technique, it was estimated that more than 95%, 97%, and 99% of vasectomies were done with simple LE without FI in India, Nepal, and Bangladesh, respectively. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Vasectomy procedure using ligation and excision combined with fascial interposition over the testicular end. ::: ![](1471-2490-5-10-1) ::: There were many reasons reported for not performing FI. Firstly, this technique is difficult to master according to many trainers and providers. In a high number of procedures failed attempts to perform FI was observed. Training of providers may be insufficient as many trainers themselves do not routinely use the technique. One technical factor which may prevent surgeons from achieving proper FI was the size of the suture material. In Thailand where the technique was successfully performed routinely, Silk 3-0 was used to tie the vas. This fine thread would not interfere with the tied testicular stump of the cut vas sliding into its sheath when both stumps are returned back into the scrotum before pulling out the abdominal vas end to identify the vas sheath.\[[@B23]\] However, in South Asia many providers were using Silk 2-0, and even Silk 1-0 (Table [2](#T2){ref-type="table"}). Even if available, most of the surgeons said they would be reluctant to use Silk 3-0 because such fine thread is believed to cut through the vas and to decrease the effectiveness of vasectomy. However, using larger size suture material may preclude from doing adequate FI. Secondly, FI takes time. Even when well-mastered, performing FI may add 2 to 4 minutes to the LE procedure. Skilled NSV providers can complete a full LE vasectomy in about the same time. In high volume settings such as vasectomy mobile camps where several hundred men may be vasectomized each day, even trainers who teach FI do not do it because of the time constraints. Since a very high proportion of vasectomies are done in mobile camps in South Asia, the FI technique as now recommended may never be performed on a large scale for this reason. Thirdly, the national standards of practice in the South Asian countries selected did not include FI as a mandatory step of vasectomy. There was no mention of FI in the Nepal standards\[[@B33]\], FI was optional (\"preferable\") in the Indian standards,\[[@B34]\] and although mandatory in the Bangladesh standards,\[[@B35]\] it was not specifically mentioned in the most recent national training manual\[[@B36]\]. There is a lack of data on the effectiveness and complication risks associated with the techniques currently used. A common belief is that the failure rate of vasectomy as currently performed is about 1% but one center reported a pregnancy rate as high as 4%. Semen analysis (SA) was available in some training centers but compliance was said to be low. All reported compliance under 30% except one center in Nepal reporting 90%. In this center, the failure rate based on repeat vasectomy was estimated to be 2 to 3%. Two centers had collected data on their failure rate. In Nepal, in a cohort of 644 vasectomized men using LE, vasectomy was repeated in 4 (1.6%) of the 263 men who had a SA performed. (Dr Kiran Shrestha, personal communication) In India, 3 (1.2%) pregnancies were encountered in 258 vasectomies performed with simple LE. In the same center, adding FI to LE resulted in no need to repeat vasectomy in 130 vasectomized men who all had at least one SA (Dr Kaur Baljit, personal communication). No data on complications were available but were said to be rare. Feasibility of introducing and evaluating cautery and FI in visited Asian countries ----------------------------------------------------------------------------------- While FI was known by all vasectomy providers met, thermal cautery was new to most of them. Hand-held battery-driven thermal cautery devices specifically designed for vasectomy did not appear to be available in visited countries. About 20 vasectomies were performed involving local providers using the cautery devices brought from America. The technique used combined intraluminal thermal cautery and covering the prostatic end of the cut vas putting a free tie on the fascia. Beside the cautery device (handle and tips), all other material resources necessary to perform the technique including alkaline AA batteries were available in all countries visited, even in rural areas. The repeated use of a thermal cautery device and tips was tested in a suburban mobile camp in Nepal and proved to be feasible. Reusable cotton sheaths for inserting the cautery handle were designed and made locally. They were autoclaved along with other sterile drapes. Cautery tips were decontaminated, brushed, washed, and processed with high level disinfection. \[[@B37]\] In general infection prevention procedures were adequate for vasectomy except in some rural areas where, among other pitfalls, the surgical instruments were left in open air for many hours. Many providers are performing minilaparotomy for tubal ligation in addition to vasectomy, maintaining the same infection prevention standards for both procedures. With few exceptions, vasectomy surgical equipment was well maintained suggesting that cautery devices (handles and tips) could be kept functional with proper instructions and minimal training. Providers demonstrated much interest in learning the use of cautery and FI as described in Figure [2](#F2){ref-type="fig"}. At least 10 trainers or master trainers performed vasectomy using the device. After reviewing the technique on video and observing one live case, all were able to demonstrate adequate use of the cautery device and FI under supervision. All considered that the technique was easy to learn, to master, and to teach. In the hands of skilled NSV providers, time to perform thermal cautery with FI for the first time was similar to that observed with current LE and FI technique. However, it definitely took more time on average than performing simple LE. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Vasectomy procedure with thermal cautery combined with fascial interposition over the abdominal end. ::: ![](1471-2490-5-10-2) ::: Cautery was already recommended as an optional occlusion technique in the Nepal standards of practice \[[@B33]\] but was not in India\[[@B34]\] and Bangladesh\[[@B35]\]. There was much interest to participate in the evaluation of efficacy and possible implementation of cautery on the part of leaders of the reproductive health/vasectomy programs and the service providers in South Asian countries. Potential structures to conduct evaluation research appeared to be in place in some locations. Some centers in Nepal had collaborated in international studies. Data were already collected in a structured format in all centers although with regional variations. Centers in India were using extensive data collection forms. Some centers in Nepal and India were already producing their own statistics. However, in most centers reinforcement of structures would be needed to improve the process of data collection and the validity of data collected. Facilitating factors and barriers to the introduction and evaluation of cautery and FI in Asia are summarized in Table [3](#T3){ref-type="table"} and [4](#T4){ref-type="table"}. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Facilitating factors and barriers to the implementation of fascial interposition in Asia ::: Fascial Interposition --------------------------- -------------------------------------------------------------- -------------------------------------------------------------------------------------------------------------------------------------------------------------------- Facilitating factors Barriers Technical aspects FI already implemented although not generalized FI is difficult to learn and to master Use of FI difficult to implement in high volume settings because of time required to perform. Not a mandatory step in the national standard, and training protocol Human resources Interest in learning a new technique Belief that current techniques are effective Interest in improving efficacy and decreasing complications Changing current behavior Training Training infrastructures already in place in South Asia Training new providers may take more time than training with simple LE Need to retrain existing providers Need to retrain surgical assistants Supplies No new supply needed (except extra suture material) No supply of Silk 3-0 in national program Policy and program Program supporting sterilization (South Asia) No program supporting sterilization (Thailand) FI already mentioned in some national standards of practice FI not mandatory in most national standards of practice Evaluation Some infrastructure in place to conduct operational research Low rates of follow-up and compliance to SA ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Facilitating factors and barriers to the implementation of thermal cautery in Asia ::: Thermal Cautery ------------------------ ---------------------------------------------------------------------------------------------- ------------------------------------------------------------------ Facilitating factors Barriers Technical aspects Easiness to learn and to master thermal cautery Need to modify FI technique when using cautery Thermal cautery may be used alone with probably better efficacy than simple LE Cautery alone is faster to perform than any technique combined with FI Human resources Interest in learning a new technique Belief that current techniques are effective Interest in improving efficacy and decreasing complications Changing current behavior Training Training infrastructures already in place in South Asia Need to retrain existing vasectomy providers Need to train support staff (cautery device use and maintenance) Supplies \"Low tech\" supplies Cost of new supplies (including batteries) Most supplies already in place Thermal cautery devices not currently available Positive pilot field assessment of feasibility of processing and maintaining cautery devices Processing and maintaining new material AA alkaline batteries readily available Policy and program Program supporting sterilization (South Asia) No program supporting sterilization (Thailand) Cautery included in some national standards of practice Cautery not included in most national standards of practice Evaluation Some infrastructure in place to conduct operational research Low rates of follow-up and compliance to SA FI = Fascial Interposition, LE = Ligation and excision, SA = Semen Analysis. ::: Discussion ========== The objectives of this project were to determine the extent of the vasectomy surgical techniques currently used in some Asian countries and to evaluate the feasibility of introducing and assessing the use of cautery and FI to occlude the vas in these countries. It was not possible within the limits of the project to do an exhaustive survey of all vasectomy techniques performed in South and South East Asia. However, we included two strategies that we believe were sufficient to achieve our objectives and to provide a sound basis for planning future operational research addressing the issues. First, in each country key-informants from various levels of the health care system related with male sterilization program were interviewed. To the exception of Thailand, this included the national authorities who are responsible for the vasectomy program, and who could provide an overview of the global situation in each country. Second, a convenience sample of 21 urban and rural vasectomy centers from various Asian countries was visited, including participation in daily clinical activities in most centers. Although these centers may not be fully representative of all vasectomy centers in Asia, there were very strong national standards regarding how family planning services must be provided in the countries visited. We thus expect much less variations in techniques used in the countries visited than in North American or European countries. Most vasectomies in Asia were performed with NSV and simple LE. NSV is recognized as the best approach to expose the vas.\[[@B1]\] Based on current evidence if LE is used to occlude the vas, FI should also be performed to improve effectiveness of vasectomy.\[[@B1],[@B10]\] This latter technique was largely known and even taught in the Asian countries visited but was seldom performed in South Asia countries (India, Nepal, and Bangladesh). The main reasons reported for not adopting the technique were: 1) insufficient surgical skills, 2) time needed to perform the technique, and 3) technique not being mandatory according to country standards. However, all through Asia, vasectomy program leaders were conscious of the importance of implementing FI as a routine procedure. As per example, FI was the theme of the 2^nd^Indian National NSV Conference held in May 2004. The use of a hand-held battery device thermal cautery for vas occlusion may be feasible in Asia. Timing for introducing cautery/FI would be right as training or retraining of vasectomy providers on combined use of FI and LE is currently needed in most Asian countries. Providers showed great interest in the use of the technique, but taking into account the fact that most were experienced trainers this may not necessarily reflect the views of the majority of providers. Pilot assessment on a small scale showed that the technique can be safely and effectively performed by Asian providers with human and material resources currently available. On the other hand, it has to be kept in mind that the major benefit of introducing cautery with FI in Asia is related to the high occlusive and thus contraceptive effectiveness of the technique. Although, cautery combined with FI appears to be much more effective than LE combined with FI, firm and conclusive evidence of the superiority of one technique over the other are still lacking.\[[@B1]\] Moreover, introducing cautery with FI may be associated with the same implementation barriers encountered with introducing FI on a large scale. In addition, new direct costs (cautery devices and batteries) and indirect costs (training, processing, and maintenance of the devices) would have to be considered before implementing cautery on a large scale. PATH (Program for Appropriate Technology in Health) working in coordination with FHI and EngenderHealth has estimated that battery-driven cautery handles and tips could be manufactured at a very low price in Asia. Moreover, bench studies suggest that cautery handles and tips currently available in the United States and Canada are durable and can be safely reused (Dr D Sokal, personal communication). Conclusion ========== One of the characteristics of a successful vasectomy program in developing countries worldwide is the availability of skilled providers.\[[@B38]\] This means that providers must offer the most effective and the safest vasectomy method. Thermal cautery may prove to be this method. Further studies are needed however before thermal cautery is introduced in Asia on a large scale. These studies should assess effectiveness and surgical complications concomitantly with quantitative and qualitative outcomes related to the implementation of this new technique. Until thermal cautery is introduced in South and South East Asia, vasectomy providers should perform FI along with LE to maximize effectiveness of vasectomy procedure. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= ML, JP, and DS conceived of the study and participated to its design. ML, RK, SSB, MR, JB, GDB, TMV participated in the data collection. All authors provided input in the data analysis. ML drafted the manuscript and JP and DS revised the early versions. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2490/5/10/prepub> Acknowledgements ================ We wish to thank all vasectomy providers, program managers, and individuals from the five countries visited who generously help us during the conduct of this project and who contributed to its success. This project was supported in part by Family Health International and EngenderHealth.	PubMed Central	2024-06-05T03:55:59.507174	2005-5-25	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180458/", "journal": "BMC Urol. 2005 May 25; 5:10", "authors": [ { "first": "Michel", "last": "Labrecque" }, { "first": "John", "last": "Pile" }, { "first": "David", "last": "Sokal" }, { "first": "Ramachandra CM", "last": "Kaza" }, { "first": "Mizanur", "last": "Rahman" }, { "first": "SS", "last": "Bodh" }, { "first": "Jeewan", "last": "Bhattarai" }, { "first": "Ganesh D", "last": "Bhatt" }, { "first": "Tika Man", "last": "Vaidya" } ] }
PMC1180459	I. Background ============= Acupuncture is an essential component of Traditional Chinese Medicine and is finding increased application in therapeutic concepts for treating various illnesses. The most important effective structural elements are acupoints which are divided topographically into specific meridians. Many references regarding the qualitative characteristics of acupoints are available and histological tests seem to prove this for a number of acupoints \[[@B1]\]. Attempts to visualize the course of \"energetic paths\" in acupuncture using physical-technical methods, are described by different authors. However, the results from investigations available up until now are controversial and generally accepted proof for meridians cannot be considered as being given \[[@B2]-[@B5]\]. One goal of this study was to examine whether possible stimulation-induced meridian-like structures, as recently described by other authors \[[@B2]\], can be visualized and objectified using modern biomedical engineering methods of thermography \[[@B6]\], simultaneously at different infrared wavelength ranges. II. Methods =========== A. Thermography --------------- Two infrared cameras at different wavelength ranges were used for thermographic control of possible stimulation effects. The Agema Thermovision™ 470 PRO infrared camera system with optics of 20° (Flir Systems Inc., Portland, USA) operates at a wavelength range from 2 -- 5 μm. The temperature measurement range lies between -20°C and +500°C and can be increased to 2000°C using filters. Sensitivity is 0.1°C at 30°C. The system is ready for use in 15 -- 20 seconds. A second infrared camera was used simultaneously with the system above during all measurements. Here, a FLIR ThermaCAM™ S65 system with 24° optics was used. The spectral range of this camera lies between 7.5 and 13 μm and the temperature range lies between -40°C and 1500°C (2000°C optional). Temperature differences of less than 0.08°C (30°C, 50 Hz) can be registered with this camera. Using a special detector (FPA) with 320 × 240 pixel geometric resolution of 76.800 pixel per picture can be achieved. The determined data was transferred to a Notebook using ThermaCAM QuickView™ software and analysed with ThermaCAM Researcher Pro 2.8 software (Flir Systems Inc., Portland, USA). B. Laser-Doppler-flowmetry -------------------------- In addition to thermography, temperature and microcirculatory parameter flux ( = product of average speed and concentration of moving red blood cells in the tissue sample volume) at the selected point on the thigh (compare Fig. [4](#F4){ref-type="fig"}), were registered as comparative measurements using a Laser-Doppler device (DRT4) from Moor Instruments Ltd. (Devon, England). Laser wavelengths were 780 nm, whereby the raw signal was filtered with a digital filter from 20 Hz to 22.5 kHz. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Experimental set-up to demonstrate elevated false temperature values in the thermograms (left 2 -- 5 μm; right above 7.5 -- 13 μm) predominantly caused by reflection. In the thermogram, the actual surface skin temperature is noted as 31.2°C measured at a point next to the reflection using additional temperature sensor equipment (above center). ::: ![](1475-925X-4-38-4) ::: C. Stimulation methods ---------------------- ### a. Moxibustion-cigar and infrared warmth stimulation A burning moxibustion-cigar made of mugwort (length 20 cm, diameter 1.5 cm, Hunan, China) was used for stimulation. This radiates an average temperature of 550°C (measured with FLIR ThermaCAM S65) with an energy maximum ranging from 3 to 3.5 μm \[[@B7]\]. The second stimulation method (infrared heat radiation) was performed with an infrared lamp (EWT, 150 W, Infrared R95E, Philips, Eindhoven, Netherlands). ### b. Needle- and laserneedle stimulation In addition to the first two stimulation methods without actual skin contact, a further standardized acupuncture scheme was used, from which we know it induces proven, reproducible effects in the brain \[[@B8]-[@B10]\]. Following acupoints were stimulated with manual needle and laserneedle acupuncture: Neiguan (Pe.6), Qihai (Ren 6), Zusanli (St.36) and Sanyinjiao (Sp.6) \[[@B8]\]. Additional moxibustion was performed at acupoint Qihai \[[@B8]\]. The laserneedles used emitted continuous laser light at a wavelength of 685 nm and output of 30 -- 40 mW per laserneedle. Stimulation time was 10 minutes resulting in an energy density of 2.3 kJ/cm^2^at each laserneedle and an average total value of 9.2 kJ/cm^2^for the entire duration of laser stimulation \[[@B11]-[@B16]\]. Identical acupoints as by needle acupuncture were used. No moxibustion was done during this type of stimulation. D. Volunteers ------------- A total of 6 healthy volunteers were examined. Two volunteers were women and 4 were men, mean age ± SD was 28.7 ± 3.7 years (range: 25 to 35 years). Average body height was 176.2 ± 11.0 cm and the average body weight was 83.7 ± 23.8 kg. Volunteers declared that they did not take any medication. The study was approved by the Ethics Committee of the Medical University of Graz (13-048) and all participants gave their written consent. E. Procedure ------------ A standardized study design was used on all volunteers and included the following steps: The volunteers were laid relaxed on a bed for 10 minutes before the examination was started. During this time, the required measurement devices were applied (Fig. [1](#F1){ref-type="fig"}). The first stimulation method was performed with a moxa-cigar. Different areas of the body (left leg, right leg, upper body) were stimulated at a distance of about 10 cm for about 5 minutes. After a resting period, the procedure was repeated using the infrared lamp. In the third part of this study design a standardized acupuncture scheme was used and stimulation with laserneedles, manual needle acupuncture and additional moxibustion of the acupoint Qihai for a duration of 10 minutes was performed. The order of stimulation was selected at random \[[@B9]\]. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Infrared analytical measurement set-up (center), stimulation with moxa-cigar (right) and stimulation with laserneedle acupuncture (left) in the biomedical laboratory at the Medical University of Graz. ::: ![](1475-925X-4-38-1) ::: III. Results ============ We could not visualize structures which could be connected with possible meridians, as described in Traditional Chinese Medicine, in any of our volunteers. We were able to clearly objectify and quantify technical reflection artifacts; however, indications for a biological correlation could not be found (Tab. [1](#T1){ref-type="table"}). In the cases when only needle or laserneedle (685 nm) stimulation was used no thermal reflection phenomena were present and also no meridians could be detected (compare Tab. [1](#T1){ref-type="table"}). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Meridian structures, as described in Traditional Chinese Medicine, could not be proven in any of the volunteers. However, artifacts were evident in all volunteers. ::: Visualization ------------------- ---------------------------------- ----------------------------- ------------------------------- ------------ ----------------------------- -------------------- ------------------------- reflection phenomena (artifacts) meridian structures volunteer \# moxa-cigar infrared warmth stimulation arbitrary meridian-structures moxa-cigar infrared warmth stimulation needle acupuncture laserneedle acupuncture 1 \+ \+ \+ \- \- n. \- 2 \+ \+ \+ \- \- n. \- 3 \+ \+ \+ \- \- n. \- 4 \+ \+ \+ \- \- n. \- 5 \+ \+ \+ \- \- \- n. 6 \+ \+ \+ \- \- \- n. \+ \... detectable; -- \... not detectable; n. \... not performed ::: Figure [2](#F2){ref-type="fig"} shows the reflection (mirror-like) artifacts which result from reflection and scattering, in a 30-year-old volunteer. According to the position of the moxa-cigar (circular white area) in dependency with the camera position, reflections showing an optical linear path on the thermogram, can be applied to optional areas of the body. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Thermograms from a 30-year-old volunteer with reflection artifacts. The moxa-cigar is visible as a circular white area. Dependent on the angle of reflection, technical reflection effects visible as lines at optional parts of the body (a -- d) can occur. The thermographic scale was defined from blue (25°C) to red/white (41°C). ::: ![](1475-925X-4-38-2) ::: In this manner, typical artifacts from each volunteer can be determined in the \"thermographic pictures\". Figure [3](#F3){ref-type="fig"} shows a further example for this phenomenon. Here, simultaneous measurements in two different wavelength ranges were made using both infrared cameras. While the first camera (7.5 -- 13 μm; Fig. [3a](#F3){ref-type="fig"}) yielded less reflection artifacts with an all over better optical resolution, the second camera (2 -- 5 μm) revealed more distinct artifacts. The energy maximum of the moxa-cigar lies in the range of 3 -- 3.5 μm. Since the radiation curve markedly flattens at 7.5 -- 13 μm obviously greater reflection occurs within the range of 2 -- 5 μm. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Simultaneous monitoring of reflection artifacts with an infrared camera at a 7.5 -- 13 μm (a) and a camera at a 2 -- 5 μm (b) wavelength range. Note the stronger expression of the artifact (white-red area on the foot) at a wavelength of 2 -- 5 μm. The moxa-cigar is shown as the white circular area next to the leg. For further details see Fig. 2. ::: ![](1475-925X-4-38-3) ::: Figure [4](#F4){ref-type="fig"} demonstrates the occurrence of reflection phenomena in a 35-year-old volunteer at the thigh. Simultaneously, surface skin temperature and flux were determined with a combined temperature and microcirculatory sensor which was applied to the skin. The reflection artifacts at the area of the sensor led to false temperature representation of 38.5°C at the sensor surface (smooth plastic surface) in the thermograms from both cameras. The neighboring skin surface revealed a 4.2°C higher temperature (35.4°C instead of 31.2°C) in the thermograms. The taped on temperature sensor confirmed the correct value of 31.2°C for surface skin temperature at this point. Figure [5](#F5){ref-type="fig"} shows the structures which were registered on a black carton (Fig. [5](#F5){ref-type="fig"} left). Due to the optical characteristics in the infrared range, the marked (linear type) structure after external stimulation with an infrared heat source is visible in the thermogram. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Visualization of marked paths and points on a black carton (a) in the thermogram (b). Note that this visualization was created by a stimulation with a heat source at a distance of 10 cm. ::: ![](1475-925X-4-38-5) ::: IV. Discussion ============== Thermic and infrared energy lies within a wavelength range no longer detectable by the human eye. This energy lies within the region of the electro-magnetic spectrum which is perceived as being warm. Contrary to visible light, every object in this region with a temperature over absolute zero radiates heat. When the temperature of an object is higher, the intensity of emitted infrared radiation is also higher. Infrared cameras produce pictures of invisible infrared- or heat radiation, thus enable exact temperature measurements. This technique of temperature measurement has already been used in several studies dealing with acupuncture research \[[@B2]-[@B6]\]. Advantages of this method are the high local resolution as well as optical data registration without requiring skin contact. In addition, the method is passive, i.e. no energy is transmitted into the body and thus is completely harmless. This allows longer examination times and therefore can also be used on infants \[[@B17]\]. Dynamic changes in temperature distribution, as possible during acupuncture can be registered well and reliably \[[@B6]\]. Areas which produce less heat or are less perfused are shown as much cooler on the calculated pictures as those with better circulation. Differences in temperature which are less than 0.08°C can be objectified. An important factor is the emission coefficient of the skin, which lies at 0.97 in this study. This factor is dependent upon the direction of radiation. For this reason, border regions are visible as being much cooler. The misrepresentation of the measurement value resulting from geometrics of the radiated structures must be considered during interpretation \[[@B7]\]. Errors can also occur because of false emission coefficients at the skin because e.g. greasy skin creams can lead to an obvious reduction in the emission coefficient \[[@B6]\]. Our experience shows that material characteristics such as reflection or refraction, two important examples in this study, are dependent upon the wavelength of optical radiation, and can show strong differences in the visible spectral range. Thermal radiation spreads straight out in a homogenous medium and is influenced by border areas. For example, when air borders on a smooth metal surface, reflection occurs. An example for directed reflection is the surface mirror. Ideal diffuse reflection can be achieved with bright white finishes and coarse surfaces. On most border surfaces, mixed types of reflection occur. Further, part of the radiation is absorbed by the surface. Reflection is very important, particularly in thermography. In the majority reflection is unwanted in the radiation path of optical instruments and should be minimized. Often, background rays mask the measurement signal due to the reflection in irregular surfaces of the measurement object including the human body. This effect partially determines the limited applicability of infrared cameras. It is important to note, that the optical laws are valid for the entire spectral range; obvious deviations are evident at the borders of extreme ultraviolet or extreme infrared radiation \[[@B7]\]. The technical, but obviously not biological reflection phenomena described in this study can be applied to any desired area of the body. \"Thermic pictures\" with partially well known characteristics in the temperature range between 20°C and 40°C above the filmed body region can be made. The fact that we are here dealing with technical artifacts and not biological phenomena can be justified by the following: 1\. Reflections can be applied to optional parts of the body depending on the position of the moxa-cigar in connection with the angle of incidence of both cameras. In addition, the structure and form of phenomena (circular, linear) in dependence with the values described above can be varied. Thus, correspondence with the meridian structures according to Traditional Chinese Medicine is not given (compare Fig. [2](#F2){ref-type="fig"},[3](#F3){ref-type="fig"},[4](#F4){ref-type="fig"}). 2\. The moxa-cigar radiates more than 550°C with maximum energy ranging between 3 and 3.5 μm. The radiation curve flattens markedly between 7.5 and 13 μm, thus, stronger reflections in the range of 2 -- 5 μm occur than in the long-wave infrared range. This could be illustrated for the first time using two infrared cameras (compare Fig. [3](#F3){ref-type="fig"}). 3\. Reflection phenomena (artifacts) are also reproducible in both wavelength ranges in non-vital structures such as bed linens or paper (compare Fig. [5](#F5){ref-type="fig"}). 4\. Simultaneous measurement of surface temperature and microcirculation did not show any changes in regard to temperature and circulation. On the contrary, temperature differences up to 7°C could be registered between both methods at the examined regions of the body which underlines the probability of an artifact (compare Fig. [4](#F4){ref-type="fig"}). 5\. So-called moxa-induced meridian paths can be projected (reflected) horizontally or diagonally at any angle (compare Fig. [2](#F2){ref-type="fig"}). 6\. Stronger reflections at both wavelength ranges (2 -- 5 μm and 7.5 -- 13 μm) can be triggered by using an infrared stimulation device. 7\. No path equivalent meridians on the body surface could be visualized in any volunteer during acupuncture stimulation (manual needle acupuncture and laserneedle stimulation as well as moxibustion). V. Conclusion ============= In conclusion, we note that thermographic methods such as infrared cameras at wavelength ranges of 2 -- 5 μm and 7.5 -- 13 μm and other High-Tech methods \[[@B8]-[@B16],[@B18]-[@B25]\] are effective complementary methods in acupuncture research which support demystification of this treatment method. However, the validity of the method for proving meridian structures according to the view of Traditional Chinese Medicine, must be considered critically and analysed scientifically. In a publication which attempted to visualize meridians and their path using infrared thermography, the authors \[[@B2]\] described that the left stomach meridian could be visualized during stimulation with a moxa-cigar near the left foot. At the same time, the right spleen meridian should be visualized on the right foot. Based on the foundations of Traditional Chinese Medicine, it is improbable that exactly these two meridians can be activated simultaneously. According to current technical standings, the visualization of energetic paths in the sense of meridians as described in recent literature is not possible using thermography. On the contrary, the supposed thermographic reproduction of meridians is dependent upon physical-technical artifacts caused by thermic reflections. According to current technical standings and to the method proposed by other authors \[[@B2]\], the visualization of energetic paths in the sense of meridians seems to be not possible using thermography. Further scientific studies are required regarding the possible visualization of meridians. Acknowledgements ================ The author would like to thank Ing. Andreas Angerer (nbn Electronics, Graz, Austria) for his valuable support in data registration and data analysis. We thank nbn Electronics Graz for supplying both thermographic systems. In addition, our gratitude to Gerhard Schwarz (Prof MD), Head of the Dept. of Anesthesiology for Neurosurgical and Craniofacial Surgery and Intensive Care at the Medical University of Graz for his critical reviewing of this manuscript. We also want to thank Lu Wang MD and Ingrid Gaischek MSc (both Biomedical Engineering and Research in Anesthesia and Intensive Care Medicine, Medical University of Graz) for their competent support in acupuncture and preparing the manuscript and illustrations.	PubMed Central	2024-06-05T03:55:59.521023	2005-6-15	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180459/", "journal": "Biomed Eng Online. 2005 Jun 15; 4:38", "authors": [ { "first": "Gerhard", "last": "Litscher" } ] }
PMC1180460	Background ========== Radio-frequency (RF) ablation has become of considerable interest as a minimally invasive treatment for primary and metastatic liver tumors. Hepatocellular carcinoma is one of the most common malignancies, worldwide with an estimated annual number of 500,000 deaths \[[@B1]\]. Surgical resection offers the best chance of long-term survival, but is rarely possible. In many patients with cirrhosis or with multiple tumors, hepatic reserve is inadequate to tolerate resection and alternative means of treatment are necessary \[[@B2]\]. In RF ablation, RF current of 450 to 500 kHz is delivered to the tissue via electrodes inserted percutaneously or during surgery. Different modes of controlling the electromagnetic power delivered to tissue can be utilized. Power-controlled mode (P= constant), temperature-controlled mode (T= constant) and impedance-controlled mode (Z\< constant) are commonly used. The electromagnetic energy is converted to heat by resistive heating. Tissue damage can occur at temperatures above 43°C with heating times of several hours; at 50°C cell necrosis occurs after \~3 min \[[@B12]\]. A commonly used mode is temperature-controlled ablation, where the tip temperature of the electrodes is kept at a predetermined value, usually around 100°C. The hepatic electrodes (Fig. [1](#F1){ref-type="fig"}) used for temperature-controlled ablation have temperature sensing elements (thermistors or thermocouples) embedded in the prong tips. The sensors report temperature back to the generator, which then applies an appropriate amount of power to the electrode to keep the temperature constant. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Geometry of fully deployed Rita model 30 umbrella probe used in FEM model. The prongs and the distal 10 mm of the shaft conduct RF current. ::: ![](1475-925X-4-42-1) ::: Researchers have been using the finite element method (FEM) to simulate both cardiac and hepatic RF ablation \[[@B3]-[@B11]\]. When using the FEM to model temperature-controlled ablation, the applied voltage has to be adjusted to keep the tip temperature constant. Previously, most researchers used manual adjustment of applied voltage and trial-and-error methods to perform temperature-controlled ablation \[[@B3],[@B5]-[@B8]\]. Implementation of temperature-controlled feedback in models is of importance to obtain results comparable to clinical devices that use this type of control. We modeled a clinically used hepatic ablation electrode (15 gauge, RITA medical systems model 30) as described previously \[[@B6]\]. We implemented a control algorithm for a PI controller -- a commonly used controller type -- in a C++ program to change the applied voltage between the time steps. We used a closed loop system simulation to optimize control parameters for the PI controller. Methods ======= 1. Finite element method ------------------------ RF ablation destroys tissue by thermal energy, which is converted from electric energy. The current flows from the conductive electrode through the tissue to a surface dispersive electrode. Tissue in close vicinity of the electrode tip is heated by resistive heating. The heating of tissue during RF ablation is governed by the bioheat equation: ![](1475-925X-4-42-i1.gif) where ρis the density (kg/m^3^), cis the specific heat (J/(kg·K)), and kis the thermal conductivity (W/(m·K)). *Jis the current density (A/m^2^) and Eis the electric field intensity (V/m). T~bl~is the temperature of blood, ρ~bl~is the blood density (kg/m^3^), c~bl~is the specific heat of the blood (J/(kg·K)), and w~bl~is the blood perfusion (1/s). h~bl~is the convective heat transfer coefficient accounting for the blood perfusion. Q~m~(W/m^3^) is the energy generated by metabolic processes and was neglected since it is small compared to the other terms \[[@B13]\]. Equation (1) defines the solution in the spatial domain encompassing electrodes and tissue. Tissue properties were assumed temperature independent and are described in more detail in \[[@B6]\]. We used the commercial software ABAQUS/Standard 6.3 (Hibbitt, Karlsson & Sorensen, Inc., Pawtucket, RI) for solving the coupled thermo-electrical analysis. All analysis was performed on a SUN Blade-1000 workstation equipped with 2.5 GB of RAM and 80 GB of hard disk space. Fig. [1](#F1){ref-type="fig"} shows the geometry of the RITA model 30, 4-prong electrode we used in our models. The electrode was placed within a cylinder (80 mm diameter, 50 mm length) of liver tissue. The outer surfaces of the cylinder were set to 37°C (thermal boundary condition), and 0 V (electrical boundary condition). We performed quasi-static analysis. Due to the symmetry of the arrangement, we could reduce computing time by only modeling a quarter of the cylinder. We used the same model as in a previous study \[[@B3]\]; the model consisted of \~35,000 tetrahedral elements and \~7,000 nodes. The node spacing was small next to the electrode (0.2 mm) and larger at the model boundary (2 mm). Perfusion was included in the model according to the Pennes model \[[@B14]\]. The blood perfusion wbl used in this model was 6.4·10^-3^1/s \[[@B15]\]. An input file was submitted to ABAQUS, in which geometry, material properties, step time and boundary conditions were specified. The applied voltage was one of the boundary conditions and remained constant during each step. ABAQUS created a results file in which temperatures of all nodes of the FEM model at the end of the step were written. We created a C++ program that reads the temperature at the electrode tip from the results file and creates a new input file where the applied voltage is set according to a control algorithm; the C++ program then calls ABAQUS and waits until the next step is completed. We implemented a PI controller in our control algorithm. 2. Closed loop system simulation -------------------------------- Since the FEM model takes several hours to complete, using the FEM model to optimize control parameters was not feasible. Therefore we used a closed loop simulation to adjust parameters of the controller, and then used these parameters in the FEM model. Before implementing the controller, we analyzed the dynamic system (i.e. the FEM model) to be controlled. This system consisted of the ablation electrode, tissue, and dispersive electrode. The input variable of the system was the voltage applied to the electrode. The output variable was the temperature measured at the tip of the electrode. Initially, we determined the step response by applying a constant voltage of 25 V for 180 s. We then approximated the transfer function of this system by a time-discrete transfer function of the following form: ![](1475-925X-4-42-i2.gif) We used the control system simulation software ANA 2.52 (Freeware, Dept. of Control Engineering, Tech. Univ. Vienna/Austria) to analyze the control system. This software allowed us to approximate the system from its step response by a recursive least square algorithm and gave the parameters a~0~, a~1~, a~2~, b~1~and b~2~of (2). Fig. [2](#F2){ref-type="fig"} shows the step responses of the original dynamic system (i.e. the FEM model) and of the approximation according to (2). The parameters used for the approximation in (2) were: a~0~= 1.0, a~1~= -0.959, a~2~= 0.127, b~1~= 1.734, b~2~= -1.150. The sampling time of the approximated system was 10 s, since that was also the sampling time used later in the digital PI controller. This ensured a more accurate simulation model of the closed-loop control system. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Step response of the original dynamic system (FEM model) and of the approximation. ::: ![](1475-925X-4-42-2) ::: Once we had identified an approximation of the dynamic system (FEM model), we designed a feedback control system. There are different ways of controlling the tip temperature such as PID control, adaptive control, neural network prediction control and fuzzy logic control. We chose the relatively simple PI controller for our control system. Fig. [3](#F3){ref-type="fig"} shows the complete closed-loop control system. T~s~is the desired set tip temperature and T~t~is the current tip temperature. The input of the PI controller e= T~s~- T~t~. The output of the PI controller u(corresponds to the applied voltage) is fed into the dynamic system. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Closed loop system incorporating a PI controller and the dynamic system (FEM model). ::: ![](1475-925X-4-42-3) ::: The transfer function of the PI controller is: ![](1475-925X-4-42-i3.gif) Equation (3) can be described in the time-domain by: ![](1475-925X-4-42-i4.gif) To implement this controller in software, we have to use the discrete time-domain version of (4), ![](1475-925X-4-42-i5.gif) The second term of (5) represents the approximation of the integral term in (4) by trapezoidal numerical integration. The behavior of the PI controller is determined by the two parameters K~p~and K~i~. We simulated the behavior of the closed-loop system with the software ANA. From in-vivo experiments in pigs with the RITA 500 generator and the Model 30 electrode we determined that it takes between 1 and 2 min for the tip temperature to reach a target temperature of 100°C, which is the temperature used in clinical cases at our institution. The RITA 500 generator does not allow recording of the tip temperature signal, so we don\'t have exact data of the tip temperature over time. We empirically chose parameters for the PI controller to minimize overshoot and obtain similar temporal behavior as in the in-vivo experiments. The parameters we used for the PI controller were K~p~= 0.02, K*~i~= 0.0064. Fig. [4](#F4){ref-type="fig"} shows the results of the closed-loop control system simulation. The target tip temperature of 100°C was reached 100 s after start of the ablation. The maximum overshoot was 11%, which was reached after 148 s. The maximum voltage of 24.5 V was applied after 100 s. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Temporal behavior of tip temperature and applied voltage of the closed loop system. The upper curves show the tip temperature, the lower curves show the applied voltage. The light curves show the results of the control system simulation. The dark curves show the results of the FEM model controlled by the control software. ::: ![](1475-925X-4-42-4) ::: 3. Controlled FEM model ----------------------- We implemented the PI controller with the parameters resulting from the control system simulation in a control program written in C++. The software control program determined the applied voltage and the step time. An input file for ABAQUS was created and ABAQUS solved the FEM model. ABAQUS created a result file, which included the temperatures at all nodes. The tip temperature was read from the result file by the control program. The program then determined the applied voltage and step size for the next step, and ABAQUS was restarted with the modified input file. With decreasing step size (i.e. time that ABAQUS simulated the model with constant voltage) total computation time gets longer. As a compromise, we chose 10 s as the initial step size. Note that the FEM solver divided each step into smaller increments, starting at 0.05s. Also, the FEM solver performed convergence tests to ensure that the increment size was sufficiently small. This scheme was repeated until the ablation had been simulated for the desired time. Fig. [6](#F6){ref-type="fig"} is a flow chart of the algorithm implemented in the control program. The initial step size was 10 s. Subsequently, the step size increased once the tip temperature change between the steps decreased below a certain value. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Flowchart of the control software implementing a PI control algorithm. ::: ![](1475-925X-4-42-6) ::: We simulated ablation for 12 min using the FEM model and the control program. The simulation took 250 min to run and was divided into 35 steps. The individual steps took between 6 and 11 min to run in ABAQUS. Without using the control program, manual interaction would be necessary after each step to manually change the input file for ABAQUS and apply a new input voltage and set the step size. In this case the operator would have to check the results file after each step (i.e. in 6 to 11 min intervals), modify the input file and restart ABAQUS. Results and Discussion ====================== Fig. [4](#F4){ref-type="fig"} shows the resulting tip temperature and the applied voltage for the first 200 s of the closed loop system consisting of the FEM model and the control software. No results are shown after 200 s since there was little change once the set tip temperature was reached. Fig. [4](#F4){ref-type="fig"} also shows the results of the control system simulation. There is good correlation between the control simulation and the closed loop system consisting of the FEM model and the control software. We explain deviations between the two by inaccuracies of the approximation (see (2)) and differences in step sizes. In the control system simulation a constant sampling time (i.e. step time) of 10 s was used. However, the algorithm implemented in the control program did not use constant step times. Step time was increased if the change in tip temperature between the steps was below a certain value. Note that the parameters of the controller must be modified when boundary conditions (e.g. perfusion), electrode geometry etc. are changed. Otherwise there will be changes in the dynamic behavior of the closed-loop system. Fig. [5](#F5){ref-type="fig"} shows the temperature and voltage of the FEM model controlled by the same PI controller for two cases. The light graphs show the behavior without incorporating perfusion in the FEM model, the dark graphs show the behavior of the same FEM model with perfusion (as in Fig. [4](#F4){ref-type="fig"}). Both the overshoot and the settling time of tip temperature for the case with perfusion were higher. Also, the voltage required to keep the tip temperature at the set temperature value was higher because the perfusion carried heat away. To obtain the same performance during the initial period in both cases, different controllers had to be used, e.g. the parameters of the PI controller had to be modified. Our model only included a quarter of the actual electrode. In models where non-uniform heating due to vessels occurs, the electrodes will obtain different temperatures \[[@B6]\]. In this case, the temperature of the hottest electrode should be used for control to avoid tissue overheating. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Temporal behavior of tip temperature and applied voltage of the closed loop system. The upper curves show the tip temperature, the lower curves show the applied voltage. The light curves show the results of the controlled FEM model without perfusion. The dark curves show the results of the controlled FEM model incorporating perfusion. ::: ![](1475-925X-4-42-5) ::: In hepatic RF ablation, ablation times clinically used go up to 35 min. Since the heat-up period is comparably much smaller (1 to 2 min.), it is not of essential importance that the temporal behavior of the control algorithm reproduces the control algorithm used in clinical devices during the heat-up period. From our experience, the temperature distribution in the FEM model reaches close to steady state at the end of the simulation due to the long simulation times. As long as the tip temperature is kept within a small range around the target temperature after the initial heat-up period, the model results (i.e. final temperature distribution) should not differ significantly. However, with knowledge of the actual control parameters and algorithms of commercial devices (e.g. obtained from measurements), accurate simulations of these devices is possible using our methods. Conclusion ========== We implemented a closed loop control system to a FEM model, to automate simulation of temperature-controlled RF ablation. We further used a closed loop control system simulation to optimize control parameters. Previously, researchers often applied constant power, or used time-consuming trial-and-error methods to determine required voltage. Furthermore, if control parameters and algorithms of commercial devices are known or can be measured, an accurate simulation of commercial devices is possible. Authors\' contributions ======================= DH carried out computer simulations. JG participated in design of the study. All authors read and approved the final manuscript. Acknowledgements ================ This work was supported by NIH Grant HL56413.	PubMed Central	2024-06-05T03:55:59.531450	2005-7-14	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180460/", "journal": "Biomed Eng Online. 2005 Jul 14; 4:42", "authors": [ { "first": "Dieter", "last": "Haemmerich" }, { "first": "John G", "last": "Webster" } ] }
PMC1180461	Introduction ============ Lung cancer is the most common malignancy both in terms of incidence and mortality in the world. In 1999, there were 1.04 million new cases (12.8%) and 921,000 deaths (17.8%) due to lung cancer all over the world \[[@B1]\]. It is expected that in 2004, there will be 173,770 new cases of lung cancer diagnosed and 160,440 people will die of lung cancer in the United States \[[@B2]\]. Small cell lung carcinoma (SCLC), which comprises of 20% of lung cancer, is characterized by a rapid growth rate and early metastases \[[@B3],[@B4]\]. Despite recent advances in therapy, the 2-year survival in limited stage small cell lung cancer is only 20--30% \[[@B4]\], while the median survival in extensive stage small cell lung cancer (ESSCLC) is less than a year (10.1 months) \[[@B5]\]. Given these dismal survival rates, it is imperative that factors like biomarker profiles be evaluated in an attempt to define survival characteristics. We had previously described the overexpression of HER-2/neuin approximately 29% of patients with ESSCLC. In that same study, we also showed that this overexpression led to a poorer outcome \[[@B6]\]. Micke et al found a definite amplification of c-erbB-2 oncogene in 13% of patients with SCLC; predominantly in ESSCLC. They thereby suggested that HER-2/neuoverexpression was more important in advanced stages of the disease \[[@B7]\]. In contrast, however, Schneider and associates found minimal or no overexpression of HER-2/neuin four out of four cell lines derived from small cell lung cancer \[[@B8]\]. Similarly, Bunn et al evaluated eleven small cell lung cancer cell lines and found no evidence of HER-2/neuoverexpression \[[@B9]\]. In trying to explain these differences, we had hypothesized in our previous report that the difference in the methods used to detect HER-2/neu status (immunohistochemistry vs. fluorescent in situ hybridization) and the population studied (exclusively ESSCLC in our study), could have been responsible for the difference found in overexpression \[[@B6]\]. However, we were unable to definitively explain the possible mechanisms responsible for this high rate of HER-2/neuoverexpression seen in our study. We have also demonstrated earlier that CD117 (c-kit) overexpression occurred in about a third of patients with ESSCLC \[[@B10]\]. The proto-oncogene c-kit encodes a transmembrane tyrosine kinase receptor (c-KIT/CD117) related to the platelet-derived growth factor (PDGF)/colony-stimulating factor-1 (CSF-1) (c-fms) receptor subfamily \[[@B11]\]. CD117 is thought to play an important role in hematopoiesis, spermatogenesis, melanogenesis and, more recently, in carcinogenesis \[[@B12]-[@B14]\]. Overexpression of CD117 has previously been documented in myeloid leukemia, neuroblastoma, breast tumor, colon tumors, gynecological tumors, testicular germ cell tumors and SCLC \[[@B15]-[@B18]\]. Presented with two potential targets for therapy and/or possible predictors of survival, the objective for our current work was to undertake a retrospective comparison of the degree of overexpression of HER-2/neuand CD117 in ESSCLC samples, with a possible link to prior pesticide exposure in our largely rural study population. Materials and methods ===================== After approval from the Institutional Review Board and the Human Subjects Committee, a two-phase retrospective study was performed. The records of all patients with a diagnosis of SCLC from January 1991 through April 2001 were reviewed. The data collected included the following variables regarding the patient\'s cancer history: age, sex, socioeconomic status, family history of cancer (if positive, type of cancer), smoking, performance score (ECOG), stage of cancer, treatment modality/modalities for SCLC, metastatic sites (other organs) involved, recurrent disease: symptoms at first recurrence age of recurrence, months from initial diagnosis and stage of disease at recurrence, treatment received for the recurrence, cause of death, other co-morbidities (non-malignant illnesses) and associated malignancies (if any) in the same patient. In addition, we conducted a population-based epidemiologic study in an effort to establish a relationship, if any, between pesticide exposure, ESSCLC and HER-2/neuor CD117 overexpression. Pesticide risk assessment interviews were performed (by a single member of the team for consistency) via telephone on the basis of a pre-determined questionnaire. The questionnaire investigated occupations and hobbies with special emphasis on: \(a) nature of exposure: insecticides and herbicides, organic solvents (paints, varnishes, solvents and glues) and petroleum products (diesels, petrol, oils, greases, dyes, inks and colorings), \(b) type of exposure (preparation and/or spraying of pesticide solutions, direct handling of solvents containing materials or petroleum products), \(c) use of protective measures in the workplace (dissolving or spraying the pesticide with pressurized containers, using glues or varnishes with adequate ventilation, etc.). The subject was considered as \'non-exposed,\' in the presence of these effective protection measures. \(d) duration of exposure. An exposure index was calculated for each interviewed subject according to the following formula: hours/day × days/year × years \[[@B19]\]. Patients with an exposure index \>2400 h were considered as \'exposed\', since previous reports have indicated that this figure represents heavy exposure to genotoxic agents \[[@B20],[@B21]\]. HER-2/neuoncoprotein overexpression was assessed by immunohistochemistry (IHC). Adequate pathologic samples were available on 193 subjects. IHC staining was carried out on formalin fixed, paraffin-embedded material, using the Hercep test developed by DAKO^®^Immunostaining was classified as follows: 0 = no staining; 1+ = faint, incomplete membranous pattern; 2+ = moderate, complete membranous pattern; 3+ = strong membranous pattern \[[@B22]\]. A trained pathologist (M.K.), who was blinded from the clinical and exposure history of the patient, interpreted the IHC results. An IHC score of 2+ or greater was considered as positive for HER-2/neuoverexpression \[[@B23],[@B24]\]. Detection of CD117 overexpression was assayed using an IHC technique on archival paraffin-embedded tissue specimens. Results were recorded on a semi-quantitative scale to record the number of positive cells (10%, 10--50% and \>50%) and the intensity of reaction. CD117 status was reported as positive if it was \>10% and negative if it was \<10%. Immunohistochemical staining for c-kit (CD117) was performed using a 1:250 dilution of the rabbit polyclonal antibody A4502 (IMPATH, Los Angeles, CA, USA) with the EnVision detection system (IMPATH). An antigen retrieval method was not utilized. Appropriate positive and negative controls were used throughout the testing process. We used the A4502 antibody because it has shown consistent performance with a low background, because it seems to be the most widely used c-kit antibody, and because it is the antibody specified for CD117 testing in the large cooperative clinical trials of selective tyrosine kinase inhibitor STI-571. All interpretation of CD117 testing was performed by a single pathologist (M.K.) who was blinded from the clinical data of patients in our study population. Statistical analysis was carried out using the chi-square test to evaluate the association between pesticide exposure and HER-2/neuvs. CD117 overexpression, and the Mann Whitney U test to assess the impact of HER-2/neuand CD117 overexpression on survival. Overall survival was calculated from the date of diagnosis of lung carcinoma by the Kaplan-Meier product limit method \[[@B25]\]. Logistic regression analysis was performed to estimate the magnitude of the effect of pesticide exposure on HER-2/neuand CD117 status. A multivariate analysis was performed to analyze the prognostic impact of overexpression of either oncoproteins. Statistical analysis was performed using SPSS-10^®^ Results ======= Between 1991 and 2001, 223 patients with ESSCLC were identified of whom 209 patients had complete records of treatment. The most common chemotherapeutic regimen involved using either cisplatin/carboplatin and etoposide or carboplatin and paclitaxel. The most common second-line agent used was docetaxel. 174 patients (98 males, 76 females) with a mean age of 68.5 years (range: 42--90 years) had adequate pathological specimens available for definitive HER-2/neuand CD117 testing by IHC. The major symptoms at presentation and other relevant demographic data of the study population are described in Table-[1](#T1){ref-type="table"}. In our study, weight loss (52.8%) and cough (59.0%) were the two most common presenting complaints that showed a trend towards a decreased survival (p = 0.03) in patients with ESSCLC. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Demographic data and symptoms at initial presentation in relation to HER-2/neuand CD117 overexpression (n = 174)\. ::: HER-2/neunegative (n = 120)* *HER-2/neupositive (n = 53)* CD117 negative (n = 119) CD117 positive (n = 54) ------------------------------------ ----------------------------------- ---------------------------------- ------------------------------ ----------------------------- Mean age at diagnosis (years): 65 66 63 65.5 Mean age at death (years): 67 68 67.5 67 Sex: • Male (n = 98) 62 36 65 33 • Female (n = 76) 59 17 65 21 Performance score: • Scores 0 -- 2 (n = 102) 71 32 69 33 • Scores 3 -- 4 (n = 72) 49 21 50 21 *Most common symptoms\:** • Weight loss (n = 61) 47 14 44 17 • Cough (n = 53) 36 17 33 20 • Dyspnea (n = 33) 26 7 27 6 \* not all symptoms are included and some patients had more than one symptom at presentation. ::: Fifty-three (30.4%) of the 174 specimens revealed HER-2/neuoverexpression by IHC while the remaining 121 (69.6%) did not (Table [1](#T1){ref-type="table"}). After adjusting for age, smoking, ECOG score and treatment, HER-2/neuoverexpression was associated with a statistically significant diminished survival (p \< 0.001) \[[@B6]\]. Of the 53 patients with HER-2/neuoverexpression, we were able to ascertain pesticide exposure in 41 patients (77.4%), while of the 121 without HER-2/neuoverexpression, 47 (38.8%) had a history of significant pesticide exposure, a statistically significant difference (odds ratio: 5.38; p \< 0.01, 95% CI: 2.5 -- 11.2). After adjusting for age, gender and smoking history, the odds ratio for developing a HER-2/neu positive tumor following pesticide exposure was still significant (odds ratio: 15.7; 95% CI 5.3 -- 46.6; p \< 0.01). The pesticide exposure occurred any time between 2 and 21 years prior to the diagnosis of the lung cancer. Patients with an exposure index \>2400 h were considered as exposed although the length of exposure varied from 2400 to 7100 hrs. As for CD117, of the 174 available specimens for review, 54 of 174 (31.0%) demonstrated IHC-verifiable CD117 overexpression, whereas 120 (69.0%) did not. Twenty-nine of the 54 patients overexpressing CD117 (53.7%) had evidence of pesticide exposure, whereas 59 of the 120 patients devoid of CD117 overexpression (49.2%) also had pesticide exposure (odds ratio 1.18; p = 0.12) (Figure [1](#F1){ref-type="fig"}). Moreover, in contrast to HER-2/neu, CD117 overexpression carried with it no adverse prognostic significance (p = 0.025). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Characteristics of CD117 overexpression and pesticide exposure in extensive stage small cell lung carcinoma (ESSCLC). ::: ![](1477-3163-4-8-1) ::: Discussion ========== HER-2/neu(also known as c-erbB-2) oncogene is the second member of the epidermal growth factor receptor family. It is overexpressed in many different types of human malignancies, notably, breast, lung, ovarian, gastric, pancreatic, colorectal, and cancers of the female genital tract \[[@B26]-[@B33]\]. HER-2/neuoverexpression seems to induce chemoresistance in certain experimental conditions. Also, HER-2/neuoverexpression has been associated with a poor overall survival and the development of metastatic phenotypes \[[@B26]\]. These findings suggest that HER-2/neumay serve as an excellent target for developing targeted anticancer agents, and indeed, the benefit of the monoclonal antibody against HER-2/neu(trastuzumab) has been demonstrated in the treatment of metastatic breast carcinoma \[[@B34]\]. CD117 (c-kit) is a receptor-tyrosine kinase involved in regulation of hematopoiesis as well as cell migration of germ cells. Mutations in several key exons give rise to constitutive activation of the receptor that ultimately is responsible for the carcinogenesis of gastrointestinal stromal tumors. This finding has in part led to the creation of novel, tailored molecular therapy in the form of imatinib mesylate (STI571) that has been successful where many other anti-cancer therapies have failed \[[@B35]\]. Although animal studies demonstrate that many pesticides are carcinogenic, (e.g., organochlorines, creosote, and sulfallate) while others (notably, the organochlorines DDT, chlordane, and lindane) are tumor promoters, human data, however, are limited by the small number of studies that evaluate individual pesticides \[[@B36]\]. In a case control conducted to evaluate the hypothesis that past exposure to the pesticide lead arsenate led to an excess mortality from respiratory cancer, Wicklund et al found that the presence, the intensity, nor the duration of exposure to lead arsenate differed between cases and control subjects \[[@B37]\]. McDuffie et al reported an absence of correlation of lung cancer risk with occupational exposure to any specific pesticide or pesticides grouped by chemical composition \[[@B38]\]. In contrast, Blair et al studied the mortality experience of a cohort of 3,827 white men licensed to apply pesticides in Florida to investigate health effects associated with chronic exposure to pesticides. They found an increasing risk of lung cancer with number of years licensed suggesting that some pesticides may be carcinogenic in humans \[[@B39]\]. Similarly, De Stefani et al found that workers employed in the construction industry, as well as those exposed to DDT may have an excess risk of lung cancer \[[@B40]\]. Becher and associates also showed an increased overall cancer mortality and mortality of respiratory cancer after long-term exposure to phenoxy herbicides and dioxins \[[@B41]\]. Pesatori and co-workers showed that the risk of lung cancer was greater among those who worked as pest control operators than non-pest control workers \[[@B42]\]. In a predominantly agricultural state like North Dakota, where farming is the main occupation, exposure to various pesticides and herbicides is an everyday occurrence. Tessier and Matsumara have shown that in the human prostate cancer cell lines LNCaP and PC-3, erbB-2 kinase was activated by pesticides of different chemical classes: the organochlorine insecticides beta-hexa-chlorocyclohexane (beta-HCH), o,p\'-dichlorodiphenyltrichloroethane (o,p\'-DDT), heptachlor epoxide, the pyrethroid insecticide trans-permethrin, and the fungicide chlorothalonil \[[@B43]\]. Similarly, in another study, Enan and Matsumara examined the effect of o,p\'-DDT, the most estrogenic congener of the DDT family of chemicals and beta-HCH on protein phosphorylation activities in MCF-7, a line derived from human breast cancer cells. They found that o,p\'-DDT activated c-neu (HER-2/neuproduct protein) at extremely low concentrations \[[@B44]\]. In a previous report, we had shown that pesticide exposure was associated with HER-2/neu overexpression \[[@B45]\]. With regards to CD117 overexpression, our prior work has demonstrated that CD117 is overexpressed in about 30% of patients with ESSCLC \[[@B10]\]. Although the techniques to ascertain CD117 testing were different, since our reported prevalence of CD117 overexpression was different from that described in previous studies \[[@B46],[@B47]\], and the fact that our study population was predominantly rural and mostly consisted on of the farming community, we decided to explore the link between CD117 overexpression and pesticide exposure. In addition, despite the evidence on the role of pesticides in the pathogenesis of lung cancer via potential avenues such as HER-2/neu, the exact molecular role of pesticides on tumorigenesis in humans is unclear at this time. Although there are no studies in lung cancer, either in patients or on cell lines demonstrating the biologic effects of pesticides or their components, our results indicate that pesticide exposure may indeed be related to HER-2/neubut not CD117 overexpression in our select population with ESSCLC. We found that patients with HER-2/neuoverexpression were significantly more likely to have been exposed to pesticides (41/53; 77.4%), than those without HER-2/neuoverexpression (47/121; 38.8%) (Odds ratio: 5.38; p \< 0.01). Our findings seem to agree with the field cancerization theory of Gazdar which states that all or much of the aerodigestive tract epithelium has been mutagenized, perhaps as the result of exposure to carcinogens leading to multiple molecular changes in individual cells ultimately resulting in carcinogenesis \[[@B48]\]. Although there is a relative dearth of clinical studies, experimental studies using cancer cell lines revealed that HER-2/neuwas activated by various insecticides and fungicides of different chemical classes \[[@B43],[@B44]\]. However, whether this activation translates into clinical malignancy has not been studied thus far. Our data suggest that the findings of these experimental models may in fact be true in the carcinogenesis of small cell lung cancer. Interestingly, however, CD117 overexpression was not more prevalent in patients with pesticide exposure (29/54; 53.7%) than those lacking such exposure (59/120; 49.2%) (Odds ratio 1.18, p = 0.12). Thus, the difference in our reported prevalence of CD117 overexpression in ESSCLC and previous studies possibly was not due to pesticide exposure. Hibi et al. \[[@B47]\] reported that over 60% of cases with SCLC overexpress CD117. In another series, Sekido et al. \[[@B46]\] reported a higher incidence of CD117 overexpression (81%) in SCLC. The reason for the difference in our findings and the previous reports may be attributable to the fact that our large study included only patients with ESSCLC. Although the presence of CD117 does not carry definite prognostic value, it is possible that overexpression may be preferentially observed in patients with limited-stage SCLC. Earlier reports used northern or western blotting techniques for assessing CD117 expression; both of which are more sensitive but less specific than IHC techniques \[[@B49],[@B50]\]. This difference in methodologies may also contribute to the observed difference in CD117 overexpression. We realize that as with most questionnaire-based studies, a drawback of our study-design is the distinct possibility of a recall bias among the case population. In addition, in cases where the patient was deceased, the data was obtained from the next of kin, who may or may not have had detailed knowledge regarding pesticide exposure of the subject in question. Interviewing technique plays a central role in improving accuracy of the recall and is under the control of the investigator \[[@B51]\]. We tried to minimize the effect of interviewing technique on the results by having one single investigator conduct all the interviews. Also, we used prompted questions, rather than open-ended question during the collection of exposure-data as that has been shown to decrease recall bias \[[@B52]\]. Finally, another potential drawback to our study was the fact that although we gathered enough data regarding the extent of pesticide (insecticide/herbicide) exposure, while implementing our questionnaire, we failed to investigate the extent of exposure to specific pesticides/their components. However, it is possible that the effect of individual pesticides/pesticide components on HER-2/neuoverexpression may be better evaluated initially using SCLC cell lines \[[@B44]\]. In conclusion, HER-2/neuoverexpression was seen in slightly less than a third of patients with ESSCLC, and was associated with decreased survival. Conversely, CD117 overexpression was noted in approximately half of ESSCLC patients, but was not correlated to prognosis. Our data also suggest that pesticide exposure seems to be related to HER-2/neuoverexpression and may explain the increased incidence of overexpression detected in our study population. However CD117 expression appears to be uninfluenced by pesticides, at least at the level of exposure analyzed in our data. Future epidemiologic studies are needed to validate our findings, and also to investigate which pesticide(s)/pesticide components are actually related to HER-2/neuoverexpression, so that proper preventative measures may be adopted. Moreover, despite the lack of an evident relation between pesticides, prognosis, and CD117, this cell marker may nonetheless provide a valuable therapeutic target for select patients with ESSCLC as it does for certain patients with gastrointestinal stromal tumors. Further studies are warranted to investigate that possibility.	PubMed Central	2024-06-05T03:55:59.533267	2005-6-9	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180461/", "journal": "J Carcinog. 2005 Jun 9; 4:8", "authors": [ { "first": "Anil", "last": "Potti" }, { "first": "Apar Kishor", "last": "Ganti" }, { "first": "Sascha A", "last": "Tuchman" }, { "first": "Kaley", "last": "Sholes" }, { "first": "Eric", "last": "Langness" }, { "first": "Vijay", "last": "Koka" }, { "first": "Michael", "last": "Koch" } ] }
PMC1180462	Introduction ============ Recently there has been a great fervor in the health field around the inclusions of non-traditional disciplines \[[@B1]\]. For example, eastern medicines, naturopathy and healing have become popular and were recently introduced into medical school curricula. These changes have made Chinese medicine \[[@B2],[@B3]\] attainable in the west just as Western medicine is to the east. However, open-mindedness \[[@B4],[@B5]\] or legality \[[@B6],[@B7]\] does not necessarily imply true integration between the medicinal approaches. The aim of this article is to suggest a novel integrative view of the key concept in Chinese medicine, qi, meridians and acupuncture, with Western biochemistry. The crux of the hypothesis is that the transmission of qi along the meridians, involves informational molecules, which travel viaan intercellular communication system, and, acupuncture at specific points enhances this communication system. Nitric oxide (NO) is proposed here as a prime candidate for such a signaling molecule in the meridian system. One analogy that can be made is to consider the word HELLO as an example to the signaling molecule. The information that HELLO transfers is not only the meaning of the word, which is HI, but mainly the way by which it is expressed e.g., loud, smiling, sad, laughing. If we think of NO in the context of a word like HELLO, it is suggested that the meaning of qi is not only a word like HI but rather an abundance of bodily information. Discussion ========== Nitric Oxide ------------ NO has been the most widely studied signaling molecule for more than a decade. It regulates blood pressure, contributes to the immune responses, controls neurotransmission and participates in cell differentiation and in many more physiological functions \[[@B8]\]. Nitric oxide (NO) the diffusible, signaling gas, is synthesized by 3 nitric oxide synthase isoforms (NOS): a neuronal (nNOS), inducible (iNOS) \[[@B9]\], or endothelial NO synthetase (eNOS) \[[@B10],[@B11]\]. Using BH4 (tetrahydrobiopterin) \[[@B12]\] as a cofactor, these enzymes convert arginine to citruline and NO. The levels of NOS are altered by a variety of pathophysiological conditions such as hypertension, hypercholesterolemia, aging, cigarette smoking, diabetes, heart failure and under physical activity \[[@B13]\] and dementia \[[@B14]\]. NOS were also shown to be associated with a plethora of more than 20 proteins that affect the activity and spatial organization of NO synthesis within the cell. Some of these interactions may have functions beyond NO formation \[[@B15]\]. NO is such a crucial entity that the mammalian genome encodes three NOS genes with partially overlapping function that may compensate for each other if one of them is not functioning. For example, the essentiality and flexibility of the NO system allows nNOS enzyme to compensate for eNOS, in mutant mice lacking eNOS \[[@B16],[@B17]\]. Recently NOS was shown to be essential for development (in Drosophila) by taking advantage of the fact that, unlike those of mammals, the genome of the fruit fly has only one NOS gene and is thus more amenable to mutagenesis \[[@B18]\]. Qi, meridians and acupuncture ----------------------------- The idea of qi is fundamental to Chinese medical thinking. Qi is involved with all biological functions and circulates throughout meridians reaching the entire human body \[[@B19],[@B20]\]. The goal of acupuncture as well as other Eastern medicinal methods (herbs, massage, Qigong) is to ensure the unobstructed flow of qi. When qi is strengthened or balanced, it can improve health and ward off or slow the progression of disease. Chinese medicine considers sickness or pain, a result of the qi blockage or unbalanced qi energy in the body \[[@B2]\]. Like NO, qi is essential for life, as Guanzi said \[[@B21]\] in 5^th^century BCE, \"When there is qi, there is life\". Much information has accumulated regarding the physiological effects of acupuncture. In 1998, a NIH panel concluded that acupuncture is an effective technique for relieving nausea and vomiting and an effective agent for relieving pain \[[@B22]\]. Needling of acupuncture points has subjective (needle sensation) and objective (e.g., serum cortisol increase or Ca(2+) oscillations) effects \[[@B23],[@B24]\]. Acupuncture increases blood flow \[[@B25]\] and acupuncture points have high electrical conductance \[[@B26],[@B27]\]. A relationship between acupuncture points and meridians to connective tissue planes \[[@B28]\] and perivascular space \[[@B29]\] has also been suggested. Possible mechanisms of acupuncture have been reviewed \[[@B30]\] and data are available at the site of the National Center for Complementary and Alternative medicine \[NCCAM\] \[[@B31]\]. They include conduction of electromagnetic signals, activation of opioid systems or changes in brain chemistry, sensation and involuntary body functions. This article does not deal with the physiological effects of qi manipulation but rather with the nature of qi flow along the meridians. The hypothesis is that the transmission of qi is based on an intercellular communication system and that nitric oxide (NO) is a prime candidate to be a signaling molecule in the meridian system. Information and signals ----------------------- Information links increase the ability to handle complexity and thus, control mechanisms in the body can be regarded as information processing. This was proposed by H. R. Maturana & F. Varela in the concept of Autopoiesis, the process by which systems organize themselves out of disorder, forming a responsive, self-maintaining network characteristic of life \[[@B32]\]. Autopoiesis is parallel to the concept of qi, streaming along the meridians\' net, regulating our well being. It is expected that the qi type signal molecule in the body, will be an essential entity and probably a gas so that it can spread as well as dissipate very quickly to allow following information to be transferred. Furthermore, the dual nature of NO \[[@B33],[@B34]\], being either beneficial or detrimental, is parallel to the dual nature of qi \[[@B19]\] and fits the basic character of a signal. Signals can not always be beneficial, in that \"cost\" must be associated with them to ensure their \"integrity\", e.g., those who wish to transmit information about their richness, are likely to exhibit it by expensive jewels or costly cars, signals which are hazardous to poor people who can\'t afford them \[[@B35],[@B36]\]. With biochemical signals, such integrity can be attained by variety of costly characters, see for example cAMP and Ca2+ \[[@B37]\] or Glutamate and GABA \[[@B38]\]. One of such characters may be the toxicity of a molecule like NO which demands very cautious and strict handling. NO differs from other neurotransmitters in that its levels are regulated solely by synthesis, rather than by storage, release or targeted degradation. NO is impossible to live without, short lived, highly diffusible and toxic \[[@B39]\] and is thus an excellent candidate for a cellular communication signal which carries the qi information. Conclusion ========== What is known and what needs to be studied ------------------------------------------ The proposal presented here of NO as the carrier of qi along the meridians already has initial scientific support. Data are being accumulated indicating the tight relations between NO and acupoint manipulation. First, in rats, NO has been shown to play a role in mediating the cardiovascular responses to electroacupuncture at point 36ST \[[@B40],[@B41]\], Second, acupuncture was shown to modulate the expression of NOS under diabetic conditions \[[@B42],[@B43]\]. Additional experiments demonstrated that acupuncture increased the level of iNOS mRNA in macrophages \[[@B44]\] and while examining the distributions of NO in the skin points of rats it was shown that NO contents and nNOS expression were consistently higher in the skin acupoints/meridians associated with low electric resistance \[[@B45]\]. Together these data show that NO is associated with the functions of acupoint/meridian including their low electric resistance. An initial human clinical study was performed in China and showed that the NO contents of peripheral blood increased significantly after acupuncture/moxibustion at point 36ST \[[@B46]\]. It is important to note that because NO is involved in multiple body functions, its presence in the peripheral blood or in an acupuncture point is a supportive but not a sufficient proof for the hypothesis. In general a necessary approach in this case is to incorporate into the NO measures the interrelations of the meridian points based on the Chinese medicine principles \[[@B19],[@B47]\]. Manipulating a distant point on the meridian and measure NO in another part of the body, or blocking NO in one part of the body and measure inhibition of a distant meridian manipulation, will demonstrate the meridian net via the NO biochemistry and the essentiality of NO in the control mechanisms of Chinese medicine. Final words ----------- Traditional Chinese medicine, which was created over 2500 years ago, can play the key role in the formation of a novel integrative medicine. It can be characterized as holistic with an emphasis on the integrity of the human body and its close relationship with the environment. It is interesting to note that ancient mummies (stone aged) were discovered in Europe and in Peru, these mummies have tattoo lines partly overlapping meridians \[[@B48]\], so it seems that not only the Chinese have noticed this biological net. The elucidation of the molecular basis of qi will open a vast opportunity to integrate the Chinese medical philosophy with current biological research on cellular communication, as well as with information technology in general, a theme which has become one of the main issues of our time \[[@B49],[@B50]\]. Moreover, If NO is indeed a qi transmitter, the Chinese knowledge can be used as a clue to the sought after framework for NO \[[@B8]\] in Western medicine and further establish the link between Chinese and Western medicines. Are meridians the generic net of life? Are the major acupoints it\'s hubs? \[[@B51]\].	PubMed Central	2024-06-05T03:55:59.535349	2005-5-18	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180462/", "journal": "Cell Commun Signal. 2005 May 18; 3:8", "authors": [ { "first": "Dina", "last": "Ralt" } ] }
PMC1180463	The delivery of treatments for diseases has always been the central mission of medicine as the delivery of preventative or control measures has always been the mission of public health. This capacity to deliver interventions has been present over the history of both disciplines, independent of the degree of knowledge of causes of diseases or effectiveness of treatment or preventative measures. The development and reaffirmation of epidemiology as a scientific discipline has been closely associated with its ability to search for causes of health-related events. Causal reasoning was always part of human thinking and of philosophical concerns. However, the first to transform philosophical causal concerns in an organized logical system from which causal relationships could be inferred was the philosopher J. Stuart Mill in the mid-19^th^century. He formulated the so called Mill\'s Canons \[[@B1]\]. At the same time, bacteriologists excited with their newly-emerging discipline were looking for more specific systems to help them to check etiological associations between infectious agents and specific diseases, and to attend this need the Henlé-Kock postulates were formulated. After that, it would take nearly a century for new causal formulations related to health-related events to be developed. The rise of the chronic diseases as the main component of the morbidity and the mortality in rich countries and the consequent changes that this had upon the causal thinking in epidemiology (until then dominated by infectious disease) was the stimulus for the development of the widely known Hill criteria of causality \[[@B2]\]. In order to address problems raised by newly discovered infectious agents that did not behave according to the Henlé-Kock postulates and in an effort to unify the causal verification of infectious and non-infectious health problems, Evans \[[@B3]\] presented a new set of postulates. To make the lives of epidemiologists difficult, the causes of health-related events are multiple and very heterogeneous -- varying from a micro-infectious agent to a macro-social factor. Ideas of multicausality have long been part of epidemiological thinking, and multi-causal models have been built-up. However, the common way epidemiology search for causes continue to be throughout the test of one-by-one potential causes, even if it is part of a multi-causal complex where grouping all the others causes are grouped under the general label of confounders. It is not by chance that the classical study designs used in epidemiology are well suited for this task. During the 1990s several leading epidemiologists intensified alerts about the insufficiency and limits of such strategies used in epidemiology \[[@B4]-[@B8]\]. While the focus of their criticisms varied from theoretical or paradigmatic questions to more applied ones, together they raised serious concerns about the role of epidemiology to fulfill its mission of producing sound knowledge and informing effective actions compatible with present and future population health needs. As a consequence, this is a special moment in which the clarification of aspects related with causation could help epidemiology to overcome its epistemological difficulties and keep on track with its mission, including the study of the determinants of health-related states or events. It is worth noting that this is not a crisis restricted to epidemiology; it also affects different disciplines involved in the understanding of human and social events, and the roots of their causes. We are living in a moment when crucial philosophical and scientific questions are being debated. More than ever words like chaos, complexity, dynamic models, etc. have been present in the philosophical, scientific and lay literatures. In this context, the causality debate has been intensified, but now in connection with the complex ways that human and social events are now perceived. Causality is understood as a relational phenomenon, which has theoretical, but also practical implications. A cause can be the presence or the absence of an action depending of the position of the observer. A cause can increase or decrease the occurrence of a health-related event -- causation and prevention are different faces of the same coin. A cause is always the outcome of other causes. But, as far prevention is concerned, when suppression or activation of a specific cause is feasible and generates the desired prevention, there is no immediate need to understand its own causes. A cause is an analyzable factor but some times it is also consequence of a deliberate intervention -- the quality of environment affects health, but a clean environment can be a natural occurrence in forests or can be the consequence of an intervention directed to decrease pollution in urban places. Similarly, housing is an important factor related to health-events, but it also consequence of programs implemented to change the housing situation of a given population. Epidemiology has used two different approaches to study causes of the health-related events: the experimental and the observational. Experimental studies (frequently randomized community trials) evaluate a cause by comparing similar groups of individuals exposed and non-exposed to an intervention targeted to suppress or stimulate this cause. This characteristic means that randomized trials are considered by many to be the ultimate standard for definition of a causal association. For some radical minds, the observational approach cannot even be considered in causal discussion! However, for several reasons, including operational and ethical ones, a great part of the causal knowledge accumulated in epidemiology comes from observational studies where very often the comparison groups are not similar or even do not exist. Consider the situation where the vaccine Xis a \"cause\". While it is possible to test its effect using observational studies there is a great consensus that this is best done by a randomized trial. However, if the situation of a vaccination program using the same vaccine, a new causal problem is born, and use of a randomized trial is not as feasible as before. And for a great number of causal problems, experimental studies are totally unfeasible. In some special cases, the two approaches could be used complementarily: for instance, the effect of vitamin A deficiency on child morbidity (diarrhea and acute lower respiratory infections) was verified using observational studies and later, the effect of vitamin A supplementation of deficient children on morbidity was tested experimentally. The questions put forward by Greenland in the paper published here \[[@B9]\] is part of a great effort made by him and others \[[@B10]-[@B12]\] to understand the complex nature of causation. In this intricate world, advancements in the understanding of causality is a task that needs to conjugate very wise philosophical (but not metaphysical) and empirical (but not empiricist) perspectives. Epidemiologists have a difficult time as they must see causality as a dynamic and multivariate process, but without losing the opportunities for prevention and without getting lost in the web of causation. Competing interests =================== The author(s) declare that they have no competing interests.	PubMed Central	2024-06-05T03:55:59.536690	2005-5-24	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180463/", "journal": "Emerg Themes Epidemiol. 2005 May 24; 2:3", "authors": [ { "first": "Mauricio L", "last": "Barreto" } ] }
PMC1180464	Background ========== Skin disease has long been recognized as having an adverse psychosocial impact on patients \[[@B1]-[@B3]\]. During the past decade, the formal assessment of patient health-related quality of life (HRQL) has been included in studies to assess the management of chronic skin disease and evaluate new treatments. The Dermatology Life Quality Index (DLQI), in particular, has been used in a number of studies of dermatological diseases including eczema, psoriasis, and chronic idiopathic urticaria (CIU) to evaluate the impact of treatment in these patient populations \[[@B4]-[@B9]\]. The DLQI was originally developed as a brief questionnaire for routine clinical use to assess the limitations related to the impact of skin disease and has been shown to be responsive to clinical changes in a study of dermatology \[[@B9]\]. While the DLQI and other outcomes measures provide a useful benchmark by which to evaluate the effectiveness of treatment, there has been a growing interest among clinicians and regulatory agencies, such as the U.S. Food and Drug Administration, in identifying meaningful change in HRQL. The concept of the minimal important difference (MID) refers to the smallest difference in a score that is considered to be worthwhile or important \[[@B10]\]. Juniper and colleagues \[[@B11]-[@B13]\] define a minimal clinical important difference as \"the smallest difference in a score\...which patients perceive as beneficial and which would mandate, in the absence of troublesome side effects and excessive cost, a change in the patient\'s management\". For clinicians, it could be used as a threshold by which they recommend a therapy to their patients. The purpose of this study is to estimate a MID for the DLQI in patients with CIU, a chronic skin condition defined by recurrent pruritic welts or wheals. Methods ======= Data for the analyses are based on two Phase III randomized, double-blind, parallel group, placebo-controlled multi-center clinical trials that were conducted to assess the efficacy and safety of fexofenadine HCl in the treatment of CIU. The two studies are labeled Studies A and B, and were similar in design. Details of these studies are reported elsewhere \[[@B5],[@B6]\]. The primary objective of both studies was to assess the clinical efficacy and safety of a range of fexofenadine HCl doses (20, 60, 120, and 240 mg bid) compared to placebo for the relief of CIU symptoms. Both studies involved a 24-hour, single-blind placebo lead-in, followed by a four-week, double-blind treatment period. Each study included three site visits approximately two weeks apart. A secondary objective was to assess health-related quality of life (HRQL) among study subjects using the DLQI. The analyses described in this report do not evaluate treatment effects on HRQL. This report focuses on examining the MID of the DLQI among patients with CIU who participated in the two clinical trials. Subjects and inclusion criteria ------------------------------- Male and female subjects aged 12 to 65 years with a diagnosis of CIU, defined as the presence of urticarial wheals for at least 3 days per week for six consecutive weeks, were eligible to participate in these studies. In addition, subjects had to have a minimum of one to five wheals with moderate to severe itching during the previous 12 hours. A total of 476 and 468 subjects participated in Studies A and B respectively. Both studies recruited subjects from multiple clinical sites in the U.S. and Canada. Measures -------- ### Dermatology Life Quality Index (DLQI) The DLQI was used to assess health-related quality of life among study participants. Subjects were asked to complete the DLQI at three scheduled study visits: baseline, week 2, and week 4 (end of study). The DLQI is designed to assess the impact of a wide range of skin disease on patient health-related quality of life (HRQL) \[[@B9]\]. It consists of ten items and covers six domains including symptoms and feelings, daily activities, leisure, work and school, personal relationships, and treatment. Response categories include \"not at all,\" \"a little,\" \"a lot,\" and \"very much,\" with corresponding scores of 0, 1, 2, and 3 respectively; the response \"not relevant\" (and unanswered items) are scored as \"0\". A total score is calculated by summing the score of all items, resulting in a maximum score of 30 and a minimum score of 0. Scale scores are calculated for each domain. Higher scores indicate poorer HRQL (i.e., more impairment). ### Patient-assessed Pruritus severity The primary efficacy measure for the clinical trial was the change in mean pruritus score over the four-week treatment period. Patients assessed urticarial symptom severity reflectively over the last 12 hours and recorded scores in a daily diary twice a day, in the morning and evening, just before taking their medication. Pruritus severity was rating on a scale of 0 to 4, where 0 = none; 1 = mild, not annoying or troublesome; 2 = moderate, annoying and troublesome, may interfere with sleep/daily activities; 3 = severe, very annoying, substantially interfering with sleep/daily activities; and 4 = very severe, warrants a physician visit. A daily score for patient-assessed pruritus severity was calculated as the average of the morning and evening assessments made each day. Change in mean pruritis severity was calculated as the difference between scores for baseline and end of study. Data analysis ------------- We used two approaches to determine the MID for the total DLQL score including a distributional criterion approach and an anchor-based approach. Our primary approach was based upon distribution-based methods based on Wyrwich\'s work using the standard error of measurement (SEM) \[[@B14],[@B15]\] and on the standard deviation of the measure of interest \[[@B16]\]. The SEM describes the error associated with the measure and is estimated by the standard deviation of the measure multiplied by the square root of one minus its reliability coefficient. The advantage of using the SEM is that it is not sample dependent due to the inclusion of the sample\'s reliability and variability in the SEM computational formula. Therefore in repeated samples drawn from the same population, the SEM values should be equivalent, except in cases where particular samples have a large number of subjects on the extreme ends of the distribution. Based on evidence supported by Wywrich, a one-SEM criterion was used to reflect a minimal clinically important difference in individual patient scores. Baseline assessments of total DLQI scores were used to calculate the SEM. More recently, there has been discussion \[[@B16]\] that a number of studies have demonstrated that one-half a standard deviation of a measure represents a good approximation of the minimally important difference, so this distributional approach was used as well. As a means of confirming findings using the SEM-criterion approach, we also used an anchoring technique, whereby we examined changes in the DLQI total score by changes in disease severity using the patient-assessed mean pruritus score. Clinical meaningful change using this approach was defined as the difference in mean change of the DLQI total score for patients classified as \"responders\" and the mean change score for patients classified as \"non-responders.\" Responders were defined as those with mean change score in pruritus severity greater than or equal to one, while non-responders were defined as those with a change score of less than 1 and equal to or greater than 0. Those with a mean change score in pruritus severity of less than zero (i.e., condition worsened) were not included in the analysis. Given that Study A and Study B were two independently conducted studies, results were analyzed separated. Results ======= Distribution criteria --------------------- A total of 403 and 423 assessments were obtained for the DLQI total score at baseline for Studies A and B respectively. As can be seen in Table [1](#T1){ref-type="table"}, mean total score for the DLQI at baseline was 9.64 (SD = 6.19) for Study A and 9.32 (SD = 5.61) for Study B. The coefficient alphas for the DLQI total scores in Studies A and B were 0.87 and 0.84 respectively. Based on these findings, results from both Studies A and B demonstrate that the SEM for the DLQI total score is 2.24. One-half the standard deviation for Studies A and B were 3.10 and 2.81, respectively. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Distributional Characteristics of Total DLQI Score at Baseline for Studies A and B ::: Item N Floor (%) Ceiling (%) Mean SD Cronbach\'s Alpha SEM ------------------------ ------- --------------- ----------------- ---------- -------- ----------------------- --------- Study A Symptoms/Feelings 403 0.25 11.66 3.49 1.49 . . Daily Activities 403 24.07 3.72 1.92 1.69 . . Leisure 403 37.72 3.72 1.46 1.64 . . Work and School 403 27.05 15.14 1.21 1.01 . . Personal Relationships 403 53.85 2.73 1.04 1.53 . . Treatment 403 63.03 3.23 0.52 0.79 . . Total DLQI 403 2.98 0.25 9.64 6.19 0.87 2.24 Study B Symptoms/Feelings 423 0.24 9.69 3.48 1.39 . . Daily Activities 423 23.88 2.84 1.84 1.60 . . Leisure 423 37.35 4.02 1.38 1.56 . . Work and School 423 30.97 16.08 1.15 1.04 . . Personal Relationships 423 53.19 2.36 0.96 1.39 . . Treatment 423 62.65 2.84 0.51 0.76 . . Total DLQI at Baseline 423 0.24 0.47 9.32 5.61 0.84 2.24 Floor = percent who answered minimum value. Ceiling = percent who answered maximum value. ::: Change over time to disease improvement --------------------------------------- A total of 319 patients from Study A and 359 patients from Study B had both a baseline and end-of-study assessment for the DLQI total score. Of these, 9 patients from Study A and 7 patients from study B were excluded from the analysis, since their health worsened over the course of the study. Among the 310 patients from Study A included in the analysis, 150 (48%) were classified as \"responders,\" while 160 (52%) were classified as \"non-responders\". Among the 352 patients from Study B included in the analysis, 208 (59%) and 144 (41%) were classified as \"responders\" and \"non-responders,\" respectively. As can be seen in Table [2](#T2){ref-type="table"}, in Study A, the mean change in DLQI total scores from baseline to end of study was -7.06 (SD = 5.95) for \"responders\" and -3.85 (SD = 5.30) for \"non-responders\". While these findings suggest an improvement in quality of life for both \"responders\" and \"non-responders,\" the magnitude of change among \"responders\" is nearly twice that of \"non-responders\". Similarly, findings from Study B indicate that the mean change in DLQI total scores from baseline to end of study was -6.94 (SD = 5.32) for \"responders\" and -3.97 (SD = 5.79) for \"non-responders\". Thus, the difference in change scores of the DLQI total score between \"responders\" and \"non-responders\" was 3.21 and 2.97 for Studies A and B respectively. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Mean change in DLQI scores (baseline to end of study) for responders and non-responders as assessed by patient-reported pruritus severity score ::: Baseline End of Study Mean Change in DLQI\* ------------------- -------------- ------------------ --------------------------- ------------- -------------- -------------- Study A (N = 310) 9.39 (5.56) 9.33 (6.35) 2.33 (3.14) 5.48 (5.53) -7.06 (5.95) -3.85 (5.30) Study B (N = 352) 9.13 (5.38) 9.10 (5.72) 2.18 (2.89) 5.13 (4.99) -6.94 (5.32) -3.97 (5.79) \*End of study minus baseline ::: Discussion ========== This study determined that a change in the DLQI total score in the range of approximately 2.2 to 3.2 could be considered clinically relevant in patients with chronic idiopathic urticaria (CIU). Our results are based on two different approaches for determining the minimal important difference (MID), including a distributional approach and a patient-based anchoring technique. These findings represent conclusions drawn from two separate Phase III clinical trials to assess the efficacy and safety of fexofenadine HCl in the treatment of CIU. To the best of our knowledge, this is the first study to determine a MID for the DLQI that could be used to differentiate patients treated for CIU who experience clinically meaningful change from those who do not in both therapeutic trials and routine care settings. Using a one SEM distributional-based approach, we found that the MID for the DLQI total score in this patient population is 2.24. The SEM describes the error associated with the measure and Wyrwich has been able to show that this approach closely mirrors results using an approach based on patient global assessment of change \[[@B14],[@B15]\]. For our study, we used a one SEM threshold. However, threshold values of 1.96 SEM \[[@B17]\] and 2.77 SEM \[[@B15],[@B17]\], also have been suggested for defining clinically meaningful change. While 1.96 represents the value on a standard normal curve associated with a 95% confidence interval, the 2.77 value incorporates a multiplier of two to adjust for sampling error when using data from two samples (test and retest) versus one. Since our analysis is based on Cronbach\'s alpha coefficients obtained from two independent samples, as opposed to test-retest reliability coefficients from a single sample, we did not make this adjustment to the calculation. Using the one-half standard deviation approach, we estimate the MID to be 2.81 to 3.10 (for studies B and A, respectively). These estimates provide estimates that are slightly larger than the estimates obtained from the SEM approach. Finally, our finding of the MID threshold of 2.24 -3.10 is supported by similar results obtained when we used an anchoring technique based on change in patient assessment of their pruritus severity. Using this technique, the threshold for clinical change was 2.97 and 3.21 for the two clinical trials respectively. While anchor-based measures are typically based on perceived changes using a global question to address overall health (i.e., patients are asked to rate the degree to which their overall health improved or worsened over the course of follow-up), such a measure was not included in either clinical trial used for our study. Our results are based on the change in patient assessment of symptoms related to the degree of itching that patients reported at baseline and end-of-study and thus may not accurately reflect an overall assessment of health. The fact that the estimated MID in this study is less than the improvement shown by the \"non-responder\" group (improvement = 3.85 and 3.97 in the two studies, respectively -- see Table [2](#T2){ref-type="table"}) is anomalous. It may well reflect the fact that the two placebo groups in the two clinical trials upon which this study were based \[[@B5],[@B6]\] showed significant improvement, although significantly less improvement than the active treatment groups. It should be noted that direction of change may be important in defining clinically meaningful change. There are reports to suggest that a smaller amount of change may be required to be considered clinically important when a patient is improving compared to worsening \[[@B18]-[@B20]\]. Our analysis of the MID based on patient assessment of CIU symptom change did not include patients who worsened over the course o the study, since less that three percent of the study sample reported worsening symptoms. The magnitude of estimates of the MID on the DLQI for CIU are similar to the estimates that can be derived from data recently presented using the DLQI to assess change in patients with moderate to severe psoriasis. Using the tables provided in the article by Shikiar and colleagues \[[@B21]\], one can derive SEM estimates of the DLQI of 2.44 and 2.42, one-half standard deviation estimates of 3.4 and 3.36, respectively, for the two studies reported on in their manuscript. In addition, they demonstrated that the difference in change scores over time between non-responders and partial-responders was 2.48 in one study and 4.34 in a second study. With the exception of the last value cited, these results are generally in line with estimates obtained in the present study using CIU patients. In addition, another estimate of the MID of the DLQI for use in a general dermatologic population was found to be 5.0 \[[@B22]\]. Conclusion ========== In conclusion, we were able to identify the MID threshold that could be used to determine meaningful clinical change in patients treated for CIU when using the DLQI total score to assess their quality of life. This threshold, in the range of 2.24 to 3.10, could be used to interpret meaningful change in both clinical studies comparing alternative treatments and within the context of continuing care in a clinical practice setting. Further validation of these results are needed to explore the MID of the DLQI in other dermatology-related conditions and skin disease. Authors\' contributions ======================= RS and GH jointly developed the analytical approach and shared responsibility for interpreting the results and preparing the manuscript. ML reviewed results of the analyses, suggested additional analyses, and helped prepare the manuscript. RL reviewed the manuscript and provided helpful suggestions for its revision.	PubMed Central	2024-06-05T03:55:59.537985	2005-5-20	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180464/", "journal": "Health Qual Life Outcomes. 2005 May 20; 3:36", "authors": [ { "first": "Richard", "last": "Shikiar" }, { "first": "Gale", "last": "Harding" }, { "first": "Michael", "last": "Leahy" }, { "first": "Richard D", "last": "Lennox" } ] }
PMC1180465	Introduction ============ Asthma is a growing public health problem within the United States. It affects approximately 15 million people and results in at least 2 million emergency room visits and more than 5,000 deaths each year \[[@B1]\], with an estimated 6.7% prevalence of childhood asthma \[[@B2]\]. Moreover, there is mounting evidence showing that asthma prevalence in childhood has plateaued over the past two decades \[[@B1],[@B3]-[@B5]\]. Surveillance records from the Centers for Disease Control and Prevention (CDC) also show an upward trend within the neighborhood of 1.74 times for the period between 1980 and 1996 \[[@B1]\]. Asthma is a chronic inflammatory disease of the airways which is associated with reversible airway obstructive, hyperresponsiveness to triggers, clinical symptoms of wheezing, chest tightness, or cough and increased mucous production. It is a major respiratory illness among children and disproportionately affects minorities \[[@B6]\]. Most children diagnosed with asthma have mild to moderate symptoms, however, there are those whose symptoms result in numerous visits to the hospital emergency room and multiple hospitalizations. Asthma remains a major burden within the Buffalo community located near the U.S.-Canada border, and there is growing evidence from past studies \[[@B7]-[@B20]\] confirming this statement. For example, several published population-based and health care utilization studies have produced persuasive evidence suggesting a combination of contributing factors to asthma exacerbations and prevalence rates in neighborhoods located in close proximity to major traffic zones in western New York \[[@B7]-[@B14]\]. The evidence emerging from these reports lends credence to the hypothesis that traffic-related pollutants play a major role in the worsening of asthma and its development. At the present time, there is consistency among findings reported by previous studies, and at least four explanations are emerging: (1) Increases in hospitalizations due to asthma over a decade were associated with increased truck traffic in ZIP code areas downwind of the Peace Bridge Complex (PBC) and the major roadways supplying it \[[@B15]\]. Following the World Trade Center tragedy on September 11, 2001, there was a decrease in traffic at the border crossing. As a result, there was an associated decrease in hospitalization rates for respiratory illnesses. This effect was reversed when traffic recovered \[[@B13]\]. The Public Bridge Authority measured a six-fold increase in PM~2.5~(particulate matter that is 2.5 micrometers or smaller in size) levels during the period of September 11, 2001 due to increased delays caused by more detailed customs inspections \[[@B16]\]. (2) A study that focused on analyzing health effects of the U.S. Environmental Protection Agency (EPA) designated sites \[[@B17]\] provided strong evidence to support the hypothesis that asthma risk increases as distance to these focus sites decreases. (3) Two house-to-house surveys of home environmental factors \[[@B18],[@B19]\], conducted six years apart, suggested that household triggers such as smoking, humidifiers, and age of housing units were associated with increased asthma prevalence. (4) A separate analysis of socioeconomic factors \[[@B14],[@B18],[@B20]\] suggested that asthma prevalence varied by race, gender, and maternal history of asthma among children 4 to 13 and women 18 to 54 years of age. Additionally, a recent study of risk factors for asthma showed that location, gender, age, and race were significant factors even after adjusting for age of housing, pets, molds, animal triggers, and smoking \[[@B14]\]. While findings from previous studies have assisted in the characterization of the magnitude and geographic extent of asthma, none of the studies have focused on childhood asthma in the city of Buffalo and its environs. In this study, we report on hospital visits by children diagnosed with asthma. The rationale for studying childhood asthma stems from the need to understand health care utilization rates among children in the study area. The study takes advantage of readily available automated administrative datasets for childhood asthma patients since 2000 by Kaleida Health Systems. Results and discussions ======================= Kaleida asthma and gastroenteritis databases for children contained 6,425 and 5,132 hospitalization records, respectively. Of these records only 89.2% and 89.7% listed a residential address. The geocoding processing yielded over 90% address matching for both databases. The geocoding success was due, in part, to improved data management practices established after previous studies. Table [1](#T1){ref-type="table"} presents exposure in geographic locations identified at the ZIP code level. The highest percentages of diagnosed childhood asthma were seen in ZIP codes 14215 and 14211, located east of Buffalo, and ZIP codes 14201 and 14213, located west of Buffalo. However, the lowest percentages were seen in ZIP codes located farther away. The highest childhood asthma hospitalization rate per 10,000 population was recorded in ZIP code 14203, however, this was not statistically significant. We observed statistically significant positive associations at the 95% significance level between exposure sites and outcomes in ZIP codes 14212, 14204, 14211, and 14214, but a negative association was found in ZIP codes 14201, 14202, 14225, 14221, and 14224. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Exposure based on Geographic Locations Identified at the ZIP Code Level: Odds Ratios from a Case-Control Study, 2000--2002 ::: Case Patients (n = 5,731) Control patients (n = 4,604) ------- ------------------------------- ---------------------------------- ------- -------- ------ ---------------------- 14203 0.63 356.79 0.48 218.04 1.31 \[0.779, 2.218\] 14212 7.17 267.51 4.87 145.80 1.51 \[1.280, 1.778\]\\ 14204 4.34 248.01 3.52 161.35 1.24 \[1.019, 1.518\]\\ 14213 11.22 246.55 12.19 215.11 0.91 \[0.807, 1.027\] 14201 5.69 239.55 6.93 234.40 0.84 \[0.718, 0.988\]\* 14211 11.85 229.47 8.84 137.55 1.39 \[1.220, 1.575\]\\ 14208 4.26 184.75 3.82 133.26 1.12 \[0.920, 1.364\] 14215 12.84 165.45 12.05 124.76 1.08 \[0.956, 1.209\] 14209 1.97 132.21 2.09 112.32 0.94 \[0.715, 1.240\] 14207 4.96 123.55 5.41 108.33 0.91 \[0.766, 1.088\] 14214 3.35 87.83 2.13 44.83 1.59 \[1.252, 2.026\]\\ 14206 2.62 65.25 2.39 47.85 1.10 \[0.858, 1.408\] 14210 1.41 48.56 1.67 46.16 0.84 \[0.613, 1.157\] 14202 0.33 45.94 0.63 70.12 0.52 \[0.286, 0.952\]\* 14216 1.80 43.67 2.28 44.52 0.79 \[0.595, 1.037\] 14222 0.92 36.90 0.83 26.45 1.11 \[0.732, 1.682\] 14220 1.68 36.16 1.52 26.37 1.11 \[0.813, 1.507\] 14217 1.20 27.94 1.24 23.08 0.97 \[0.679, 1.378\] 14218 0.96 27.22 1.22 27.72 0.78 \[0.537, 1.146\] 14226 1.36 26.55 1.61 25.19 0.84 \[0.610, 1.164\] 14225 1.61 25.63 2.45 31.48 0.65 \[0.490, 0.866\]\* 14223 0.89 21.17 0.96 18.26 0.93 \[0.617, 1.391\] 14228 0.68 19.78 0.74 17.25 0.92 \[0.578, 1.460\] 14227 0.59 13.84 0.93 17.50 0.63 \[0.398, 1.005\] 14221 0.96 10.69 1.46 13.02 0.65 \[0.453, 0.944\]\* 14224 0.61 8.64 1.15 13.08 0.53 \[0.339, 0.822\]\* 14219 0.16 6.90 0.30 10.73 0.53 \[0.224, 1.264\] Note. Two denominators are reported, one is derived from the number of children registered in the Kaleida Health System and the other from population data from the 2000 US Census; and case patients and control patients derived from hospitalization and outpatient visits for asthma (ICD-9 code 493) and gastroenteritis (ICD-9 code 558) from Kaleida database, 2000--2002 \\ Positive association between exposure and outcome at the 5% significance level \* Negative association between exposure and outcome at the 5% significance level ::: Table [2](#T2){ref-type="table"} presents spatial analysis results of a case-control study showing odds ratios. This study found that proximity to the EPA-designated toxic emission sites \[Odds Ratios (OR) = 1.91, 95% confidence interval (CI) = 1.211--3.011\] was associated with statistically significant increased odds of having diagnosed asthma but not of non-respiratory disease. However, we also observed a negative association for multiple emission sites \[OR = 0.81, 95% CI = 0.703--0.92\], PBC \[OR = 0.69, 95% CI = 0.48--0.99\], and interstate highway \[OR = 0.81, 95% CI = 0.69--0.95\]. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Spatial analysis of Case-Control Study Showing Odds Ratios ::: Odds Ratios 95% Confidence Interval ------------------------------------------------ ------------- ------------------------- ----------- ----------- Peace Bridge Complex 0.69\* 0.87 0.48,0.99 0.69,1.09 Air release 0.77 0.87 0.58,1.03 0.70,1.09 Toxic release 1.91\\ 1.23 1.21,3.01 0.88,1.72 Multiple release 0.80\* 0.93 0.70,0.92 0.83,1.05 Interstate 190 0.90 0.81\* 0.74,1.08 0.69,0.95 Interstate198 and Route 33 0.96 0.95 0.85,1.09 0.85,1.07 Main St, Bailey Ave, Niagara St, and Seneca St 0.92 0.96 0.82,1.02 0.86,1.08 Delaware Ave 0.85 1.01 0.68,1.07 0.83,1.24 Note. Case patients and control patients derived from hospitalization and outpatient visits for asthma (ICD-9 code 493) and gastroenteritis (ICD-9 code 558) from Kaleida Database, 2000--2002 \\ Positive association between exposure and outcome at the 5% significance level \* Negative association between exposure and outcome at the 5% significance level ::: Table [3](#T3){ref-type="table"} summarizes multiple comparison results for seven EPA-designated focus sites using Diggle\'s model. The seven sites were statistically significant, hence the need to compare them further. The comparison results are ranked in order of adjusted significance level according to Bonferroni\'s and Holm\'s correction methods. The test showed that EPA-designated focus sites located in Buffalo\'s east side showed a closer association to childhood asthma than sites located elsewhere (p \< 0.008). EPA-designated focus sites located near the grain factory and major roadways also showed a strong correlation with childhood asthma. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Multiple-Comparison Test for Model Fittings according to Diggle\'s model ::: EPA-Designated Pollution Sites Original P-value Modified Holm\'s Remarks ------------------------------------ -------------------- ------------------ --------------------------------------------------- Nabisco Company 0.00000 0.0043 Located in the east near the grain factory Truck Parking Lot, Nabisco Company 0.00000 0.0047 Located in the east near the grain factory Marnap Industries 0.00000 0.0051 Located in the east near the grain factory Harrison Radiator 0.00063 0.0127 Located in the east near the major roadway Miken Company 0.00096 0.0169 Located in the west near the major roadway Peace Bridge Complex 0.00150 0.0253 Located in the west near the major roadway (I-90) Birge Company 0.03148 0.05 Located in the west near the major roadway ::: Figures [2](#F2){ref-type="fig"} and [3](#F3){ref-type="fig"} both illustrate spatial clusters of diagnosed asthma in Buffalo neighborhoods. In Figure [2](#F2){ref-type="fig"}, only one geographic region had over 234 adult asthma cases per 1,000 population \[[@B12],[@B17]\], however, in Figure [3](#F3){ref-type="fig"}, four geographic regions had over 244 childhood asthma cases per 1,000 population. Similar spatial patterns and distributions of asthma are apparent in both figures, but further analysis revealed that asthma is more prevalent in children than adults. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### shows spatial clusters of adult asthma. Spatial clusters were detected in Buffalo\'s west side, parts of the downtown areas, and only 1 geographic region had over 234 asthma cases per 1,000 population. ::: ![](1476-072X-4-14-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### shows spatial clusters of childhood asthma. Spatial clusters were detected in Buffalo\'s west side, east side, parts of the downtown areas, and 4 geographic regions had over 244 asthma cases per 1,000 population. ::: ![](1476-072X-4-14-3) ::: Establishing the effects of ambient air pollution on asthma ----------------------------------------------------------- The need exists for establishing the effects of particulate air pollution on adults and children from the study region. Results of the adult study are already reported in Oyana et al. \[[@B17]\]. The adult and children studies were conducted during 1996--2000 and 2000--2002, respectively. The two studies could assist us in elucidating the effects of ambient air pollution in asthma, especially among the children. Based on the outcome measures we should be able to tease out the health effects given that ambient pollution sources being considered have been in operation since the 1980s and transport mechanisms have not showed any significant changes in the past 10 years. Previously, Oyana et al. \[[@B17]\] conducted the study involving 3,717 and 4,129 adult asthma and gastroenteritis patients. In the current study, the numbers of child asthma and gastroenteritis patients were 5,731 and 4,604, respectively. A comparative analysis was conducted both at ZIP code level and at different sites to establish the role of different exposures on asthma in adults and children. At the ZIP code level, a positive association was found between exposure and outcome at a 5% confidence interval in adults in the ZIP codes 14201, 14213, 14207, and 14204. In a similar study for children, the ZIP codes 14212, 14204, 14211, and 14214 showed a positive association. Similarly, negative association between exposure and outcome at a 5% confidence interval in adults was observed in the ZIP codes 14221, 14214, 14217, 14150, and 14227, whereas in children, it was observed in ZIP codes 14201, 14202, 14225, 14221, and 14224. Only the ZIP code of 14204 had a positive association in both adults and children, whereas, the ZIP code of 14221 had a negative association. However, there were some differences in both studies. For example, ZIP code 14201 showed a positive association in the adult study, while a negative association was observed for the children\'s study. Likewise, ZIP code 14214 had a negative association for the adult study, but a negative association was found for the children\'s study. In adults, we observed the highest odds ratio of 15.77 at air release sites at 0.5 versus 2 km, however, in children, we found the highest odds ratio of 1.91 in the toxic release site at 0.5 versus 2 km. When comparing the 1 km and 2 km scenario, the adult study showed two positive associations, whereas the child study showed none. The adult study revealed a strong positive relationship for the Interstate 190 roadway for both scenarios; however, in the child study, none of the sites showed a positive relationship in either the 0.5 km versus 2 km or 1 km versus 2 km scenarios. In addition, although there are four sites with positive associations in the adult study, it does not include the toxic release site, which is the only site with positive association in the child study. Table [3](#T3){ref-type="table"} shows the results of the multiple comparison test using Diggle\'s method. The data shows that EPA-designated focus sites on the east side of Buffalo are more statistically linked to childhood asthma than those on the west side. Three sites in the east have a p-value = 0.0000 and have shown the strongest statistical significance in a multiple comparison study. The two studies reveal that EPA-designated focus sites including the PBC are significant contributors to increasing the number of asthma incidents in the study area. The spatial clusters of adult asthma seen in Figure [2](#F2){ref-type="fig"} are on Buffalo\'s west side and parts of downtown. However, spatial clusters of childhood asthma (Figure [3](#F3){ref-type="fig"}) were found not only on Buffalo\'s west side and parts of downtown, but also on its east side. This could be explained by African-American and Hispanic groups, which comprise a large portion of the population on the Eastside. The adult study reveals that only one geographical area has more than 234 asthma cases per 1,000 population, while the children\'s study shows that there are four geographical areas with more than 244 asthma cases per 1,000 population. Figures [2](#F2){ref-type="fig"} and [3](#F3){ref-type="fig"} suggest that children are more susceptible to asthma than adults in the study region, possibly due to incomplete development of the immune system in children. An ongoing study is collecting field measurements of particulate matter concentrations based on sampling sites identified within and outside of the spatial clusters in Figures [2](#F2){ref-type="fig"} and [3](#F3){ref-type="fig"}. Preliminary findings of large variations of PM~2.5~concentrations were observed in locations of high asthma prevalence and there are consistent with the results of the GIS disease model. The site observations reported in this ongoing study confirms the relevance of identified spatial clusters and lends further credence to their use in GIS modeling. There are two major observations that can be made regarding both studies: (1) Findings in the children\'s study are consistent with previous findings in the adult study. Current findings support the hypothesis that traffic levels at the U.S.-Canada border crossing point and specific EPA-designated focus sites are associated with high incidences of asthma and prevalence rates, especially in areas that are located in the west and east of Buffalo. The two studies have further established credible evidence of statistical associations between current traffic levels on major roadways and specific EPA-designated focus sites \[[@B18]-[@B20]\]. (2) A huge disease risk and burden exists especially among children. This burden, or risk, is confined not only to children residing on Buffalo\'s west side but also to those children on the east side. This second observation highlights the need to focus preventive and mitigation measures and efforts on residents living in close proximity to the border crossing point, especially on children. The policy implication of our study is targeting exposure reduction. This would be justified on the grounds of maximizing public health benefits. Differential distribution of adverse health effects also need to be considered alongside differential distribution of the benefits related to the emission sources. Conclusion ========== This childhood asthma study, as did the adult study, demonstrates the effects of ambient pollution sources on individuals with asthma and suggests these sources are the contributing factors both in the west and east of the study area. Identification of asthma clusters associated with different sources may provide insights into how mixtures of pollutants interact and lead to development of asthma in susceptible individuals. These findings provide a basis for a better understanding of outdoor environmental factors that might be related to the spatial distribution of asthma prevalence and morbidity in this Buffalo community. Materials and Methods ===================== Study area and Population ------------------------- The study area, as shown in Figure [1](#F1){ref-type="fig"}, is the second-largest city in New York State, with a population near 600,000. Most of these people reside in the city of Buffalo and its surrounding areas. The study area, which consists of approximately 27% children between age 1 and 18 years of age, serves as the main traffic corridor between the United States and Canada. It has a rich history of industrial activity and is also an emerging international trade corridor for the United States, Mexico, and Canada under the North American Free Trade Agreement (NAFTA), with a mandate to promote trade among the member states. It is one of the busiest trade corridors, with increased truck traffic traveling through densely populated residential areas and increasing the potential for the release of pollutants such as nitrogen dioxides and particulate matter that have an adverse effect on human health. In fact, a number of studies have associated these pollutants with the exacerbation of respiratory diseases \[[@B21]-[@B28]\]. The study area also includes both industrialized and non-industrialized neighborhoods in order to provide access to diverse populations. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### shows the study area, major roadways, EPA-designated point pollution sources, and the location of the Peace Bridge Complex (PBC). ::: ![](1476-072X-4-14-1) ::: Data sources ------------ Three data sources were analyzed in this study: (1) hospitalization and outpatient visits for asthma (ICD-9 code 493) and gastroenteritis (ICD-9 code 558); and (2) EPA-designated focus sites. 1\. Hospitalization and Outpatient Visits for Asthma and Gastroenteritis for children age 1 to 18 years, encompassing admissions from December 2000 to December 2002 available from Kaleida Health System, which includes Buffalo General Hospital, Millard Fillmore Hospitals (Gates and Suburban), DeGraff Memorial Hospital, and Children\'s Hospital of Buffalo. Each record in the database was assigned a unique identifier ensure patient confidentiality. Data on child age, residential addresses, and insurance status were derived from this source. Home locations of patients were the basis for measuring possible residential exposure to known pollution sites. Gastroenteritis patients were designated as the non-respiratory control group because it has no connection to ambient air quality and in some ways this \"adjusts\" for other aspects of a person that lead to access to care/hospitalization such as socioeconomic status (SES). Overall, processing and management of the datasets were of high quality as a result of previous knowledge of adult asthma \[[@B12]\] and also because of strong experience in geocoding residential addresses in the study area. Moreover, the Kaleida Health System is western New York\'s largest healthcare provider with a market share of 52% and serves approximately 80% of the total population in the study area. 2\. EPA-designated focus sites used for this study are similar to those reported in Oyana and Lwebuga-Mukasa \[[@B12]\] and Oyana et al. \[[@B17]\]. The childhood asthma study, like the adult study, searched for potential spatial relationships between these sites and home locations of children who were diagnosed with asthma. We calculated distances between residential locations for each patient to each EPA-designated focus site. Additional geographic information, such as the US Census data on population, major roadways, boundary files, was obtained from Cornell University Geospatial Information Repository (CUGIR) and the Federal Geographic Data Committee (FGDC) Clearinghouse Node for New York State. Data analysis ------------- Statistical and spatial techniques were used to study spatial location of case patients and control patients in relation to their proximity to different pollution sources. The details of these techniques have been discussed in earlier reports \[[@B17]\] and will only be presented briefly. In general, the data are explored in a number of ways. 1\. Description of ZIP code level differences in rates of asthma and gastroenteritis hospitalization and outpatient visits 2\. Case control analysis of proximity to important sources of ambient air pollutants 3\. Cluster identification (Diggle\'s method) Data processing, GIS mapping, and analysis were conducted in ESRI ArcGIS 9.0 (ESRI, Inc., Redlands, California) and Microsoft Excel (Microsoft, Inc., Seattle, Washington). ClusterSeer 2.03 (TerraSeer, Inc., Ann Arbor, Michigan) was used to implement cluster analysis and Diggle\'s model. Final map production was completed using Corel Draw 11 (Corel Corporation, Ltd., Ottawa, Ontario). Diggle\'s method is a focused cluster detection approach appropriate for handling spatial data at the individual level \[[@B29],[@B30]\]. The method compares the spatial pattern of case locations with the spatial pattern of control locations, for instance using a more common \"control\" disease. The control acts as a null model of no clustering and normally reflects the spatial pattern of the population-at-risk. The test is based upon maximizing the likelihood of the sample of cases and controls, which in turn is based on an exponential decline in risk as the squared distance from the source increases. It was assumed that those who lived within 1 km of the emission sites and busily traveled roadways were exposed to vehicle exhaust fumes and pollutants from suspected sources of pollution, and those living further away more than 2 km were assumed to be unexposed. Rijnders et al. \[[@B31]\] recommend that variables such as degree of urbanization, traffic density, and distance to a nearby highway or any potential pollution source can be used to estimate exposure to traffic-related air pollution. Milligan et al. \[[@B32]\] also used distance of more than 2 km in their study to estimate exposure due to traffic-related air pollution. Similarly, the use of data on location of home with respect to roads and of data on traffic density on those roads resulted in observations of significant relationships with specific respiratory hospital admission rates in Toronto \[[@B33]\], with childhood asthma hospitalization rates in Erie County, New York \[[@B11]\], and with childhood asthma medical care visits in San Diego County, California \[[@B34]\]. Epidemiological methods based on odds ratios (OR) and 95 percent confidence intervals were used to compute the spatial risk relationships between cases and controls (using a significance level of ρ ≤ 0.05). A 2 by 2 table analysis was conducted to demonstrate the relationship between two dichotomous or binary variables (exposed and unexposed groups). The variable we are measuring for the 2 × 2 table is the percentage of children who not only were clinically diagnosed with asthma, but also who reside near major roadways. The other variable represents those who reside farther away. We were interested in this type of analysis because previous studies \[[@B31]-[@B37]\] have shown that there is a significant risk among patients living near major traffic zones. The emerging evidence suggests that major traffic zones influence the patients\' susceptibility to respiratory illnesses, especially among persons with asthma. Moreover, by measuring the residential locations of patients thought to be exposed and comparing it with those living farther away, we were able to quantify the odds of the disease burden among the study population which could be attributed to the exposure or non-exposure sources. Limitations of Current Work --------------------------- Nevertheless, it is likely that we were not able to fully control for the effect of all confounders. Other potential confounders that we were not able to measure include duration of residence, comorbidity, smoking, and exposure to other pollutants in vehicle emissions. Previous studies suggest that control for duration of residence has little influence on effect estimates \[[@B33]-[@B35]\], possibly because of an acute effect of exposure. We considered the use of consumer purchasing data to control for area-level smoking, but available data were of questionable validity. Given the similar dispersion characteristics of PM~2.5~and other pollutants in vehicle emissions (e.g., NO~2~), some of the observed effect may be caused by exposure to other pollutants. Finally, this study does not take into account other possible factors for the prevalence of asthma, such as exposure to indoor pollutants and occupational exposures. Protection of Human Subjects ============================ All research reported in this article was approved by the Southern Illinois Unversity Carbondale and the University at Buffalo Human Investigation Review Board in accordance with national and institutional guidelines for the protection of human subjects. Authors\' contributions ======================= Dr. Tonny J. Oyana designed and developed the study\'s spatial and GIS approach. He also participated in data processing, geocoding, data analysis, and GIS modeling, and he wrote the manuscript. Dr. Patrick A. Rivers advised on healthcare issues and participated in the final editing and reviewing of the manuscript. Acknowledgements ================ Supported in part by SIUC Faculty Start-up Fund and SIUC ORDA Faculty Seed Grant. Dr. Jamson S. Lwebuga-Mukasa, Department of Internal Medicine, UB School of Medicine and Biomedical Sciences, Kaleida Health Systems Buffalo General Division, provided the data sets for this study. Special thanks to three anonymous reviewers of this research article.	PubMed Central	2024-06-05T03:55:59.539883	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180465/", "journal": "Int J Health Geogr. 2005 Jun 8; 4:14", "authors": [ { "first": "Tonny J", "last": "Oyana" }, { "first": "Patrick A", "last": "Rivers" } ] }
PMC1180466	Background ========== Lung cancer is the most common cause of cancer related death in both men and women in Western countries. More than 25% of all cancer deaths are attributable to lung cancer \[[@B1],[@B2]\]. Intraabdominal metastases are not exceptional, but their clinical manifestations are rare \[[@B3]\]. We report such a case presenting with small bowel perforation. Case Report =========== A 56-year-old man (varnisher by profession, heavy smoker with 160 pack years) presented in September 2002 with a five month history of cough, fever, night-sweat and a weight loss of 5 kg in 12 months. A bronchoscopy with biopsy revealed a squamous cell carcinoma of the lung. A FDG-PET/CT scanning showed a disseminated praetracheal, aortoal, supraclavicular and infracarinal lymphadenopathy and a spleen metastasis (stage IV) \[[@B4]\]. The patient received six cycles of Cisplatinum-Gemcitabine chemotherapy, followed by a second line Mitomycine-Vinorelbine polychemotherapy. A re-evaluation after one year revealed liver metastases and increasing mediastinal lymphadenopathy, followed by additionally Taxotere polychemotherapy. A brain magnetic resonance imaging after a pathologically Romberg Test in December 2003 showed a left hemisphere brain metastasis. At New Year\'s Day 2004, the patient presented again with a three days history of increasing abdominal pain, diffuse abdominal guarding and rebound tenderness. Laboratory evaluation was without pathological findings except a marked leukocytosis (WBC = 24,000/mm^3^). Abdominal CT-scan displayed free intraperitoneal air and a perforated jejunum (Figure [1](#F1){ref-type="fig"}). The patient was operated on with the diagnosis of perforated jejunum. At surgery, a 2 cm mass was found at the surface of a loop of the jejunum with perforation of 1 cm in diameter, furthermore left upper quadrant peritonitis with pus, stool and a fibrin coating. The affected segment of the jejunum was resected with an end-to-end anastomosis and an abdominal lavage was performed. Postoperative pathologic analysis confirmed the diagnosis of metastatic pulmonary carcinoma (Figure [2](#F2){ref-type="fig"}). The surgical margins were negative for tumor. Postoperatively, the patient was given a regular diet on postoperative day four and discharged home fourteen days after surgery. The patient died 8 weeks after surgery at home on tumor progression. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### CT scan with intravenouse contrast: Free intraperitoneal air and a enlarged jejunum. ::: ![](1477-7800-2-13-1) ::: Conclusion and review of the literature ======================================= Despite the early and widespread dissemination of lung carcinoma to many body sites, clinically significant metastases in the small bowel are rare and typically occur only in the advanced stage of the disease \[[@B5]\]. But metastases to the gastrointestinal tract from lung cancer may not be as rare as thought according to autopsy series. Antler et al. report in an autopsy series of 423 cases of primary tumors of the lung gastrointestinal tract involvement in 58 cases (14%) \[[@B6]\]. Out of these 58 cases, 14 were caused by direct extension of the tumor from the lung. 44 cases (10%) had distant metastasis without contact of the primary tumor to the gastrointestinal tract. The most common site of metastasis was the oesophagus (29 cases), followed by ileum (13 cases) and stomach (10 cases). Mc Neill et al. reported in their autopsy study of 431 cases small/large bowel metastases without oesophageal/stomach metastases in 46 cases (11%) \[[@B7]\]. All patients with small bowel metastasis had at least one other site of metastatic disease with an average of 4.8 sites. In an autopsy study of Burbige et al. 18 out of 147 (12%) patients with a diagnosis of primary carcinoma of the lung had metastatic lesions in the gastrointestinal tract \[[@B8]\]. Twelve of the 18 patients (67%) had signs or symptoms suggesting gastrointestinal involvement but only four were diagnosed premortem \[[@B8]\]. Mc Neill et al. and Antler et al. \[[@B6],[@B7]\] report in their autopsy studies, the most common histologic type of lung tumor causing gastrointestinal tract metastasis was squamous cell, followed by large cell carcinoma as shown in Table [1](#T1){ref-type="table"}. In all three autopsy studies large cell carcinoma is relative to the tumor incidence prone to small bowel metastasis \[[@B6]-[@B8]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Incidence of cell type causing gastrointestinal (Antler et al.) or small/large bowel (McNeill et al.) metastases ::: Histologic cell type Antler et al. 1982 \[6\]: 58 gastrointesinal metastasis out of 423 cases McNeill et al. 1987 \[7\]: 46 small/large bowel metastasis out of 431 cases ---------------------- -------------------------------------------------------------------------- ----------------------------------------------------------------------------- Squamous cell 19 out of 173 (11%) 15/5 out of 199 (10%) Adenocarcinoma 6 out of 72 (8%) 13/6 out of 108 (18%) Large cell 17 out of 87 (20%) 12/6 out of 31 (58%) Small Cell 11 out of 48 (23%) 6/5 out of 73 (15%) Others 5 out of 43 (11%) 0/2 out of 20 (10%) ::: Reviewing the literature between 1961 and 2003 we were able to document 58 clinically manifest metastasis to the small bowel \[[@B3],[@B7],[@B9]-[@B37]\]. One report even describes eleven cases in a thirteen year period in a single institution. This result strongly suggests that metastasis to the small bowel may not be as rare as thought. Over 80% of patients with metastases to the small bowel were male, with ages ranging from 36 to 78 years with a median of 60 years. Metastases to the small bowel may present as perforation (34/58; 59%), obstruction (17/58; 29%), six (10%) patients had haemorrhage, and in one patient (2%) the small bowel metastasis was discovered in a staging CT scan. Of those patients with known location of metastasis (n = 38), 21 (55%) had jejunal metastasis, 12 (32%) ileal, three (8%) duodenal and two (5%) patient had both jejunal and ileal metastasis. In 34 patients we were able to document the histological tumor type. Squamous cell carcinoma was the most common histological cell type (17/34; 50%) followed by large cell carcinoma (8/34; 24%) and adenocarcinoma (7/34; 21%). In 7 (12%) patients the initial diagnose of lung cancer was discovered because of the abdominal symptoms, before any pulmonary disease was considered. Five out of these seven patients presented with abdominal perforation. This fact could retract the statement that particularly chemotherapy induces necrosis of the tumor cells which has replaced a portion of the small bowel wall, thus causing perforation of abdominal metastasis \[[@B21],[@B23]\]. According to McNeill et al. lung cancer metastasis have a greater tendency to undergo necrosis compared with other malignant tumor metastasis \[[@B7]\]. Therefore, there may be a greater tendency for these metastases to cause perforation before attaining enough bulk to cause obstruction \[[@B7]\]. Phillips et al. \[[@B38]\] reviewed the literature for indications for laparotomy in 27 patients presenting with isolated small bowel metastases from extra-abdominal primary sites in a 22 year time period. The author found only six small bowel perforations and all six were in patients with carcinomas of the lung as primary site. Small bowel metastasis typically occurs in the final states of widespread disease \[[@B7]\]. Surgery in this group of patients will have a high morbidity and mortality due to cancer, age and associated medical problems \[[@B36]\]. Of those patients who were known to have died (n = 39), mean postoperative survival was 3 weeks. Nine (9/39; 23%) patients died within the first week, 25 (25/39; 64%) within the first month, four patients however were alive more than one year after surgery. Review of the literature indicates that metastases of primary pulmonary carcinoma to small bowel occur late in the course of the disease, are more common than thought and may be associated with serious clinical complications. When symptomatic, patients usually present with a small bowel perforation, obstruction or haemorrhage. Squamous cell carcinoma and large cell carcinoma of the lung are more likely to result in gastrointestinal metastases than the other cell types. Treatment of obstructed or perforated small bowel is similar to any acute surgical abdomen: prompt exploratory laparotomy. When perforation of small bowel lesions is found, the procedure of choice is resection of the involved small intestine with primary enteroenterostomy, obstructions could be treated with bypass procedures. List of abbreviations ===================== FDG-PET/CT:Fluorodeoxyglucose-Positron Emission Tomography/Computed Tomography CT-scan: Computed Tomography WBC: White blood cell count Competing interests =================== No reimbursements, fees, funding or salary from an organization were received. Furthermore I have no other financial or non-finacial competing interests. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Small bowel metastasis; Stain: Hematoxylin & Eosin (amplification 1:62). ::: ![](1477-7800-2-13-2) ::: Acknowledgements ================ Written consent was obtained from the patients relatives for publication of study.	PubMed Central	2024-06-05T03:55:59.543135	2005-6-6	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180466/", "journal": "Int Semin Surg Oncol. 2005 Jun 6; 2:13", "authors": [ { "first": "Andreas", "last": "Hillenbrand" }, { "first": "Joern", "last": "Sträter" }, { "first": "Doris", "last": "Henne-Bruns" } ] }
PMC1180467	Introduction ============ During the last several decades a great deal of work has been undertaken for developing devices to accelerate recovery from injuries, operations and other complications. Many successful devices and methods have come out of this work. This included a general division of the recovery process into several phases. In the early stages of therapy, passive rehabilitation is often a preferred method for reducing swelling, alleviating pain, and restoring range of motion. This consists of moving the limb with the muscles remaining passive and often involves devices such as Continuous Passive Motion (CPM) machines. The next stage of rehabilitation is often an active-assistive movement phase, which involves using external assistance to assist the muscles in moving the joint in order to reestablish neuromuscular control. Various different methods are presently used for this purpose, including various braces, orthoses, and large machines. The final stages aim at returning an individual to normal activities via resistance exercises that are usually focused at regaining muscle strength. Isokinetic machines are well known, ideally suited systems for achieving this final goal. In this paper we present a compilation of several new developments in the area of portable and smart rehabilitation devices, being developed by the Northeastern University Robotics and Mechatronics Laboratory. The devices that will be presented in this paper are: a\) a transportable continuous passive motion device ideally suited for nearly any aspect of the earlier rehabilitation stages of the elbow, b\) an electro-rheological fluid based device for resistance exercises and control of the knee; c\) an electrical stimulation and biofeedback device for active-assistive exercises of the knee. The presented devices span across all three of the mentioned phases in rehabilitation and exhibit many advantages over current technology. All three devices have been developed to increase the efficiency in rehabilitation exercises while remaining compact and portable. In each case, the capabilities of present technology have been taken into consideration and each device is designed to have similar characteristics. The most notable difference however, between this new breed of rehabilitation devices and currently used equipment is their highly adaptive, versatile and reprogrammable nature. Computer control is intrinsic to the design of each device presented in this paper and is a central theme behind their operation. This makes for highly effective tools for a wide range of applications. More specifically, the advantages of our advanced orthotics can be divided into four main categories: cost, portability, real-time abilities, and versatility. Cost ---- The designed advanced rehabilitation devices resolve several issues with cost with present-day technology. Every initial feasibility prototype fell just short of \$2,000 to build. With all the electrical and sensory components that need to be added to each device for a final functional and marketable product, it is estimated to cost approximately \$,3500. A state of the art, computer controlled Isokinetic Machine, such as the Biodex System 3 Pro, can be bought for over \$40,000. Clearly, direct cost comparisons warrant the use of advanced rehabilitation devices over the comparable rehabilitation machines. Indirectly, the smaller size of the advanced rehabilitation devices also brings down costs by eliminating concerns with storage, portability, and weight. Rehabilitation machines are inherently large and require a permanent or semi-permanent set-up. The facility able to house such a device along with the personnel required for operating them is at a large economical disadvantage to smaller facilities using these much more compact advanced rehabilitation devices. Numerous devices, at an overall lower cost than a single machine, could be stored in something as simple as a closet. The devices themselves could easily be transported by the patients for use at home as well, saving time and money in the costly trips to specialized facilities. Portability ----------- The most important feature of such a device is the fact that it is a portable and wearable form of rehabilitation. The compact and lightweight characteristics of these advanced rehabilitation devices allow them to be used in an average chair, while standing, or perhaps even during ambulatory motion. Their application is limited by only the user\'s abilities, meaning weaker patients can use it for resistive exercises while stronger patients can use it for both weight training as well as proper gait training. Equally noteworthy is the new capability for patients to take the device with them and exercise on their own time, from the comfort of their own home or office, or for use during their every day routines. All exercises being recorded, a physical therapist could simply download the data remotely and analyze the effectiveness and efficiency of the device, without ever needing the patient to revisit the medical facility. Real-Time Abilities ------------------- The ability of rehabilitation machines to function in real-time, is what separates them from their less efficient counterparts, the conventional orthotics. The inclusion of this feature is intrinsic to the utilization of compact advanced actuators and smart sensors in our portable and smart rehabilitation devices. They are easily computer controlled, and can react in the order of milliseconds. With such controllability, a rehabilitation regime can be perfectly tailored to each patient\'s individual needs very easily. Ideally, with closed loop control, feedback from the sensors would allow a computer to calculate the efficiency of each specific exercise and alter them in real-time accordingly to achieve optimal levels of rehabilitation. Versatility ----------- Probably the most unique advantage of these devices arises from their versatility. With comparable abilities to modern day rehabilitation machines and similar functionality to several different types of these machines, the all-encompassing nature of these advanced orthotics alone makes them equally as versatile. However, due to all their additional strengths and advantages, including size, portability, and real-time computer control, the applications of these devices goes above and beyond those of the competing technologies. In the area of rehabilitation, these advanced orthotics could be a valuable tool in the development of new rehabilitation exercises and regimes. With complete control and tunability of the device, any type of complex algorithm defining the motion or resistance of the patient\'s knee could be easily implemented. Whole new concepts in rehabilitation or weight training could potentially be developed using this device as a research instrument, providing all the force and feedback necessary for any type of investigation. For more complicated medical disabilities, for instance in the case of gait correction in stroke patients, both analysis and implementation of newly developed methods could also be easily performed. Other potential applications, showing the extreme versatility of this device, include virtual reality simulations and athletic training, such as in rowing and weight-lifting. Portable continuous passive motion elbow device =============================================== Overview -------- A transportable elbow rehabilitation device for use throughout the entire process of rehabilitating patient\'s with severe elbow trauma was designed, built, tested and optimized. The apparatus has three settings -- passive, active and bracing. The device consists of a D.C. motor, gearbox, encoder, clutch and brake located in a portable unit, attached through a flexible shaft to an absolute encoder located on an elbow brace. In the passive setting, the device moves the forearm about the elbow joint to regain the range of motion. It acts as a \"smart\" continuous passive motion machine because constant sensor feedback enables the device to push to the patient\'s maximum range of motion during each cycle. Torque and speed of the passive movement is controlled through the current and voltage, respectively, drawn by the motor. In the active setting, variable resistance is applied using the brake. Both settings are controlled, monitored and recorded using a LabVIEW program on a personal computer, with specific protocol defined by a physician, physical therapist or athletic trainer. Currently available CPM machines are not transportable, do not sense the patient\'s range of motion and do not allow for an active setting. By combining three different functions (active mode, passive mode and bracing) of the device into one transportable unit, the next generation of elbow rehabilitation devices was created. Significance and Background --------------------------- Following surgery, stroke or other injury to the elbow, a patient\'s range of motion is reduced due to trauma experienced at that location. Increasing the user\'s range of motion is the first step in a full recovery. This is accomplished through passive motion, where the patient\'s forearm is actively forced to flex and extend, followed by strength training. At this point, most doctors or physical therapists begin to use a continuous passive motion (CPM) machine. A CPM machine moves the forearm about the elbow joint to regain the patient\'s range of motion. Unfortunately, current CPM machines often involve a complicated set up, are non-portable, and are most importantly inefficient. Their inefficiency arises from their inability to recognize when the user\'s range of motion has increased. The machine must be continuously monitored and manually reset to further increase the range of motion. A related concern is the possibility of forcing the patient\'s arm past his or her range of motion resulting in further damage to the joint. The range of motion can only be increased in very small increments and movement about the elbow is nonproductive once the preset range is achieved. There are several patents covering the range of elbow rehabilitation devices \[[@B1]-[@B6]\]. Several companies such as Breg, Dyna Splint, Ultra Flex, Biodex CPM and the Bledsoe Extender Arm Brace have products out on the market that immobilize the injury and prepare the elbow for rehabilitation \[[@B7]-[@B12]\]. However, the only portable devices that are available provide either spring tension against an elbow contracture to achieve increased motion or locking mechanisms to restrict motion and prevent further injury. There are currently no commercially available devices that are portable and provide the passive motion required in the beginning stages of elbow rehabilitation. Design and Prototype -------------------- A wearable and portable CPM device that senses increases in the patient\'s range of motion and simultaneously increases its range of motion has been developed in our laboratory. The patient\'s torque and motion limits are inputted into a computer interface. The program then monitors and controls all of the components of the device, progressively increasing the user\'s range of motion about the joint within the torque and motion range. Through sensory input, the computer senses when the user\'s muscular resistance has reached its limit and signals a reversal in direction of motion, allowing for maximum range of motion to be reached quickly and efficiently, without harm to the patient. This new transportable elbow rehabilitation device also safely and efficiently assists throughout the rest of the entire rehabilitation process, including bracing the joint and building muscle mass. The device has adjustable settings for each stage of rehabilitation. The passive motion setting, as mentioned, uses constant sensor feedback that enables the device to progressively increase the user\'s range of motion. The device is also capable of applying variable resistance about the elbow joint to build muscle mass once the patient\'s ideal range of motion has been achieved. This mode is very similar to Isokinetic machines. Finally, the device is also capable of acting as a simple brace: either locking in place to prevent the user from moving his or her arm, or disengaging entirely to provide mediolateral support. The combination of all three modes with adjustable settings within each mode, allows this device to be utilized through the entire rehabilitation process for a variety of elbow injuries. The elbow device is lightweight, easily programmable and transportable. A CAD rendering of the device and its components can be seen in Figure [1](#F1){ref-type="fig"}. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### CAD rendering of portable elbow device. ::: ![](1743-0003-2-18-1) ::: The device can be split into two subsystems. The first is the brace worn by the patient. It is designed around an Orthomerica Prime Elbow System brace and includes an optical encoder, for measurements of position and velocity and an attachment point for a flexible shaft. This flexible shaft connects the brace to the second subsystem, a tabletop drive assembly unit that provides the functionality of the device. It houses a DC motor, an electrically controlled clutch and magneto-resistive fluid brake and is designed to fit in a backpack. The flexible shaft allows the user to move freely while the device is in use and easily detaches from the brace, providing the patient with a protective elbow brace to continue daily routines when not in use. The motor-gearbox combination provides the passive exercise motion for the patient to increase his or her range of motion. A current limiter set in the motor control box ensures that the patient does not exceed his or her range of motion. The current measurement is converted to torque resistance in the computer and once the preprogrammed limit is exceeded; the motor direction is reversed Between the motor and flexible shaft is the electrically controlled clutch. It serves mainly as a safety feature for the patient. It disengages if the user hits the stop switch, if the current exceeds the motors limited levels, or when the active feature is in use. This active feature functions with the use of a magneto-resistive fluid (MRF) brake. The brake is manufactured by Lord Corporation and features a simple yet rugged design, high torque, and quiet operation. It provides smooth, controllable resistance to the patient for building muscle and tissue strength in the elbow joint. The MRF brake and motor in-line assembly can also be used in combination. This provides the user with an extra impulse of motion after they have used the resistive feature to their maximum range of motion or active-assistance. The device was constructed for feasibility analysis. Figure [2](#F2){ref-type="fig"} shows the full assembly. The device has a mobility range of 155 degrees. The motor, gearbox (1:134 gear ratio), and clutch combination was found to be capable of producing 10 N·m of torque. The MR brake was found to have a maximum resistive torque capability of 5.6 N· m. All these performance characteristics can be found listed in Table [1](#T1){ref-type="table"}. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Portable elbow device: full assembly. ::: ![](1743-0003-2-18-2) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Elbow Device Design Summary ::: System Characteristics: ------------------------------------------- --------- Range of motion (0° being full extension) ± 77.5° Continuous Passive Motion capabilities 10 N·m Isokinetic capabilities 5.6 N·m ::: The motor, encoder, brake, and clutch are controlled through a LabVIEW 7.0 program on the PC. The user interface is simple, and utilizes tab controls that allow the user to select either the active or passive setting. In the active setting, the user inputs the resistive torque required for exercise and can see a real-time plot of the joints position and resistance level. Inputs in the passive setting include the number of repetitions, speed, and minimum and maximum angles. The user can view real time plots of position and torque being applied to their joint during the exercise routine. The graphic user interfaces for the passive motion can be seen in Figure [3](#F3){ref-type="fig"}. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Passive motion graphical interface. ::: ![](1743-0003-2-18-3) ::: Electro-rheological fluid based knee resistance device ====================================================== Overview -------- This device aims to demonstrate the feasibility of using Electro-Rheological Fluid (ERF) actuators in orthotics, creating a new breed of rehabilitation devices. ERFs are fluids that experience dramatic changes in rheological properties, such as viscosity or yield stress, in the presence of an electric field. Using the electrically controlled rheological properties of ERFs, compact actuators with an ability to supply high resistive torques in a controllable and tunable fashion, have been developed. This study involves the design, fabrication and testing of an ERF based knee orthotic device and the innovative ERF actuators it uses. The knee orthotic is achieved through a standard brace design with a polycentric hinge and gear system. Coupled to this are two Flat-Plate ERF actuators, given that name for their characteristic set of parallel flat plates allowing for actuation of the fluid. The overall knee orthotic system is designed to resist up to 25.4% of an average human knee\'s torque abilities and be controlled in real-time. The goal of this work is to provide a much more efficient means of rehabilitation over the average orthotic, while matching the proficiency of rehabilitation machines, all in a smaller, simpler, and more cost efficient design. Significance and Background --------------------------- An orthotic device by strict definition is a specialized mechanical device that supports or supplements weakened or abnormal joints or limbs. The majority of these devices can be categorized as passive, meaning the resistance or support they provide is not changed in real time. The Sports Medicine Committee of the American Academy of Orthopedic Surgeons has further classified these types of braces, specifically used for the knee, into four categories: prophylactic, rehabilitative, functional and patellofemoral. All provide stability, apply precise pressure, and/or help maintain alignment of the knee joint at set constants. Some of the more innovative designs allow torsion to be applied at the knee joint and new technology has further improved their efficiency by allowing the torque to be adjusted. However, the lack of real-time abilities is a significant downside for these devices that limits their overall effectiveness in rehabilitation. The inclusion of active components has been a widely accepted method of improving upon this deficiency. This seemingly small addition has considerable drawbacks though. The application of traditional active elements increases the overall size, cost, weight, and other related characteristics. Equally important are the concerns with control and sensory feedback, which would also be considered necessary with the addition of active components. All these combined, along with the obvious goals of making the systems as efficient and beneficial to an individual during rehabilitation as possible, force their designs to go beyond that of a portable orthosis, and more so a machine. In terms of rehabilitation, the most effective methods known today are these rehabilitation machines. They are commonly used for rehabilitating and strengthening patients, subjects, and athletes while providing quantitative measurements of their performance. They provide high resistive and sometimes assistive forces, while providing a unique tailoring of the rehabilitation regime to nearly any individual. This ability dramatically increases their proficiency as a rehabilitation tool. Their services have been limited to primarily only physical therapy offices though, as a direct result of their shear size, weight, and cost. Electro-rheological fluids (ERFs) are fluids that experience dramatic changes in rheological properties, such as viscosity, in the presence of an electric field. Willis M. Winslow first explained the effect in the 1940\'s using oil dispersions of fine powders \[[@B13]\]. The fluids are made from suspensions of an insulating base fluid and particles on the order of one tenth to one hundred microns (in size). The volume fraction of the particles is between 20% and 60%. The electro-rheological effect, sometimes called the Winslow effect, is thought to arise from the difference in the dielectric constants of the fluid and particles. In the presence of an electric field, the particles, due to an induced dipole moment, rearrange into a more organized manner, or form chains along the field lines. These chains alter the ERF\'s viscosity, yield stress, and other properties, allowing the ERF to change consistency from that of a liquid to something that is viscoelastic, such as a gel. ERF\'s generally respond to changes in electric fields in a matter of only a millisecond or two. Good reviews of the ERF phenomenon can be found in \[[@B14],[@B15]\]. Control over a fluid\'s rheological properties offers the promise of many possibilities in engineering, especially actuation and control of mechanical motion. Devices that rely on hydraulics can benefit from ERF\'s quick response time and reduction in device complexity. Their solid-like property in the presence of a field can be used to transmit forces over a large range and have found a number of applications. A list of many engineering and practical applications of ERFs can be found in \[[@B16]\]. Our team has developed several prototypes of ERF-based linear and rotary actuation elements \[[@B17],[@B18]\], which can apply controllable resistive forces and torques such as the Flat Plate (FP) rotary actuator concept which is the primary component of the ERF actuated knee orthosis described below. Design and Prototype -------------------- The ERF knee device possesses the ability to accurately provide large resistive forces with full real-time control while remaining completely portable and wearable. These characteristics make it an ideal apparatus for several applications. For active rehabilitation exercises it replaces the need for overly cumbersome and reasonably outdated machines, by remaining a lightweight portable system that is capable of all the same forces, control, and more. Similarly, it replaces the need for large weight-lifting machines. For gait-training purposes, such as in stroke patients with hyperextension difficulties, it is a viable clinical device. Through the sensors embedded in the device, computer closed-loop control, and clinical training these disabilities are overcome by providing real-time resistance that limits motion and supports the weight of the user, to retrain a proper gait. Additionally, the portability of the device adds a whole new dimension to rehabilitation and exercising in general, where the patient is now able to take a powerful isokinetic machine home, to work, on vacation, or wherever else they may travel. The design of this innovative device consists of three major subsystems -- an ERF based resistive actuator, a gear system, and the structural brace frame. The ERF based resistive actuators, which provide a bias force to the knee joint, simulating whatever forces desired, consist of multiple parallel rotating electrode plates and they are called Flat Plate resistive actuators. They are attached via a gear system to a standard brace as seen in the CAD rendering of Figure [4](#F4){ref-type="fig"}. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### CAD rendering of electro-rheological fluid based knee orthosis. ::: ![](1743-0003-2-18-4) ::: Several circular copper plates (shown in Figs. [5a](#F5){ref-type="fig"} and [5b](#F5){ref-type="fig"}) are located parallel to each other, on a fixed axis. On a parallel, concentric axis, are another set of copper plates, which lie parallel and alternate with the fixed plates. The latter set of plates can rotate relative to the fixed plates, and the small gap between the plates contains ERF. Applying an electric field across the gap causes the fluid properties to change (in a matter of milliseconds), resulting in an increase in yield stress. The change physically alters the fluid from the consistency of thin oil to that of a thick gel. This property is used to control the resistive forces of the ERF FP actuator. The copper electrode plates with an inner and outer radius, r~i~and r~o~, respectively and a gap of d between plates can be seen in Figure [5a](#F5){ref-type="fig"}. Based upon the dimensions of the variables r~i~, r~o~, and d the design of the FP resistive actuator can be adjusted to produce a device capable of the resistive torques needed for any application. Figure [5b](#F5){ref-type="fig"} shows the assembly of a multiple Flat-Plate ERF element in CAD. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### \(a) Electrode plates (b) Internal assembly of the FP ERF resistive actuator. ::: ![](1743-0003-2-18-5) ::: The entire ERF assembly is housed in a casing that seals in the fluid and attaches to a gearbox. The gearbox transmits and multiplies the torque output of the FP resistive actuators while supporting them on the frame. The brace frame is an off-the-shelf knee brace with all the features necessary for the device. It boasts a polycentric hinge, comfortable strapping method, and a lightweight, rigid frame. Included in the design were optical encoders for measuring angle, speed, and acceleration of the knee. An initial prototype of the design was built for feasibility analysis. The final actuator was rapid prototyped using a 3D Systems Viper 2000i^2^machine. It was bench tested and was found to produce a maximum resistive torque of 9.16 N·m. A Don Joy 4TITUDE™ knee brace was donated by the company Don Joy Orthopedics, slightly disassembled and machined to allow for attachment of the gearbox and actuator. A gear ratio of 1:1.67 was used resulting in an overall device resistance of approximately 30.16 N·m. The final system successfully demonstrated an accurate and easy controllable system for resisting knee motion. In Table [2](#T2){ref-type="table"} a summary of the device characteristics and the actuator parameters can be found. Below are several images of the prototype and close-ups of some of the individual parts (Figs. [6](#F6){ref-type="fig"}, [7](#F7){ref-type="fig"}, [8](#F8){ref-type="fig"}). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### ERF Based Knee Device Design Summary ::: Actuator Parameters: ----------------------------- ----------- Gap size (d) 1.0 mm Inner Radius (r~i~) 20.0 mm Outer Radius (r~o~) 45 mm Number of Plates 17 Actuation Voltage 4.25 kV Maximum Actuator Torque 9.17 N·m System Characteristics: Gear Ratio 1:1.67 Torque Produced by Device 30.16 N·m ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Components of the brace\'s ERF FP actuator.Fabricated case with o-ring seal (top left); CNC machined electrodes with rapid prototyped mounts (top right); fabricated rotating shaft with steel output shaft and commuter installed (bottom left); actuator shaft with rotating plates attached (bottom right). ::: ![](1743-0003-2-18-6) ::: ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Close up views of the ERF actuated brace.Fabricated gearbox (left); Inner hinge (center left); Attached Actuator casing made with slots so inside plates are visible (center right); Fabricated actuator attached, filled with fluid, and encoder mounted (right). ::: ![](1743-0003-2-18-7) ::: ::: {#F8 .fig} Figure 8 ::: {.caption} ###### First version prototype of the ERF driven knee rehabilitation orthosis.Left actuator casing is made with slots so inside plates are visible, right actuator is filled with fluid. ::: ![](1743-0003-2-18-8) ::: So far tests were performed to verify the capabilities of the actuators. Since two identical actuators were used, the verification of one of these actuators would be theoretically as accurate as creating a duplicate of a human knee joint for the purpose of testing the whole device. The average torque output of the actuator at each voltage was plotted and compared to the predicted theoretical equation\'s results. Figure [9](#F9){ref-type="fig"} was the result and the two plots show a very close resemblance. The accurate results therefore suggest that the proposed system is capable of the forces desired. ::: {#F9 .fig} Figure 9 ::: {.caption} ###### Theoretical vs. experimental torque of final designed/fabricated actuator for ERF driven knee rehabilitation orthosis. ::: ![](1743-0003-2-18-9) ::: Electrical stimulation and biofeedback knee device ================================================== Overview -------- A knee brace that can be used in multiple stages of rehabilitation by using various therapy techniques was developed by our team. Following knee surgery most patients experience muscle atrophy and in some cases nerve damage. To overcome these problems physical therapists have turned to the use of electrical stimulation (E-Stim) and biofeedback (EMG) as the preferred methods of treatment. These forms of therapy help to increase the range of motion of the knee and improve neuromuscular re-education. By incorporating these units, along with a rotary encoder, into a post-operative brace it is possible to monitor the progress of the patient in a unified computer controlled setting. It also allows the patient to perform the rehabilitation while walking in a stable brace which promotes proper gait. A graphical user interface was built in LabView to monitor the various sensing units of the biofeedback knee brace. Significance and Background --------------------------- During physical therapy, an articulated controlled motion and exercise is crucial in a successful physical therapy process. Controlled motion benefits the ligaments, bones, and soft tissue and prevents them from becoming degenerative. Resistance exercises help build muscle mass and restore functionality to the limbs. Devices such as the two previously mentioned are ideal for these two cases. Alternative systems exist however, that use very different methods for overcoming the same aspects in rehabilitation. These include electronic stimulation and biofeedback. Electrical Stimulation (E-Stim) is a rehabilitative treatment that stimulates nerves by sending an electrical current through the skin. In knee surgery rehabilitation the E-Stim is typically used to activate the muscles around the knee for the purposes of neuromuscular re-education. In the early post-operative stages the E-Stim is used for active rehabilitation, where it stimulates the motor nerves of muscles without the patient\'s effort. In the secondary stages of therapy the E-Stim is used in active-assisted motion, where patient uses their muscles along with the external stimulation to move the joint. E-Stim can also be used while exercising muscles as well. Biofeedback implemented by EMG is a device that monitors muscle activity. The feedback provides valuable information regarding progress and muscle performance. The data from this device allows therapists to gain a better understanding of how the patient is responding to the treatment. This means that the therapy can be better tailored to the individual. Following knee surgery many patients experience muscle atrophy and in some cases nerve damage. To overcome these problems, physical therapists have turned to the use of electrical stimulation and biofeedback as the preferred methods of treatment. These forms of therapy help to increase the knee range of motion as well as reduce pain, swelling, and total recovery time. Design and Prototype -------------------- A knee brace combining electrical stimulation with the sensory information from biofeedback and other sensors has been developed as shown schematically in Figure [10](#F10){ref-type="fig"}. The information from the biofeedback and a rotary encoder are fed to a computer. The computer compares the EMG information to the data it receives from the linear encoder. If it detects a bio-signal being sent from the brain to the leg, but there is no motion in the brace, the E-Stim is triggered to assist the patient. All information gathered by the sensors is presented to the operator within a graphical interface, where it is also possible to adjust the settings for proper customizing of the exercises (see Figure [11](#F11){ref-type="fig"} for a screen capture of this interface). ::: {#F10 .fig} Figure 10 ::: {.caption} ###### Schematic of the smart knee brace. ::: ![](1743-0003-2-18-10) ::: ::: {#F11 .fig} Figure 11 ::: {.caption} ###### Graphical controls interface for the electrical stimulation and biofeedback knee device. ::: ![](1743-0003-2-18-11) ::: The flexibility and computer control of this device results in a valuable autonomous tool for a variety of rehabilitation exercises. The system, as described, can be used in place of or in conjunction with any exercise involving passive rehabilitation or active-assisted (the two earlier post-operative rehabilitation stages). With the addition of the foot switch, which is placed under the sole of the patient\'s shoe, the system can be used to aid in walking and regaining proper mobility. When pressure is exerted on the switch, it triggers the electrical stimulation on the quadriceps as well as the tibialis anterior, to assist in such situations as with those who suffer from foot drop due to nerve damage. Furthermore, the system boasts the added benefits of allowing the patient to perform the walking exercises while wearing a stable brace (promoting a more proper gait) and establishes an easy way to monitor their progress. A prototype was developed to demonstrate the proposed concept as shown in Figure [12](#F12){ref-type="fig"}. This prototype uses a Don Joy TROM knee brace as a frame for the device. The padding for this brace was adapted to allow the e-stim and EMG electrodes to be adhered to the proper locations on the leg. Polyethylene brace supports were also added to the upper and lower sections of the brace. This adds stability and makes the brace significantly easier to put on. In addition, the hinge on the brace was modified so that a 1/4\" diameter shaft rotates with the lower section of the brace. A Renco rotary encoder was attached to this shaft so that the angle and velocity of the knee brace could be determined. The wiring from the e-stim, EMG and the rotary encoder is covered by a custom wiring conduits located along the rails of the brace. Once the wiring leaves the brace it is hooked up to a control box, which houses all of the electronics. This includes the ProComp Infiniti biofeedback unit, the Respond II e-stim unit, as well as two solid state relays. This control box is also connected to a personal computer. Figure [13](#F13){ref-type="fig"} is a picture of the control box with the components labeled. ::: {#F12 .fig} Figure 12 ::: {.caption} ###### Electrical stimulation and biofeedback knee device. ::: ![](1743-0003-2-18-12) ::: ::: {#F13 .fig} Figure 13 ::: {.caption} ###### Control box for the electrical stimulation and biofeedbackdevice. ::: ![](1743-0003-2-18-13) ::: A graphical user interface in Labview is responsible for controlling the response of the system (see Figure [11](#F11){ref-type="fig"}). The information from the EMG biofeedback and the rotary encoder are fed to a computer via Labview. The computer will then compare the EMG information to the data it receives from the rotary encoder. If it detects a signal being sent from the brain to the leg, but there is no motion in the brace, the e-stim will be triggered to assist the patient. This will help re-educate the neuro-muscular system. This information will be presented to the operator with a graphic interface. By controlling the operation of the e-stim through adjustable set points in the lab view program the physical therapist will be able to apply the device throughout the entire rehabilitation process. The device was tested to ensure all the components were coordinated correctly. The test subject was instructed to perform a steady extension of the knee as shown in Figure [14](#F14){ref-type="fig"}. In the testing, the subject attempts to do seated leg extensions with the help of the brace. In this exercise the subject sits on the edge of a table and attempts to extend their lower leg horizontally. When the subject reaches a point where they can\'t extend their leg any further under their own power, the EMG biofeedback senses this data. Then the logic in the controls will compare the readings with the information from the rotary encoder. If the EMG signals are above the threshold this is set and the rotary encoder is stationary then the e-stim is triggered. The e-stim causes the quadriceps to contract helping the subject to achieve full extension. Since we were testing on a healthy subject, the threshold was set to a low value so that the e-stim was triggered in the middle of the exercise. Figs. [15a--d](#F15){ref-type="fig"} shows the results of one of these tests. Once reaching his maximum point, the EMG unit is seen detecting the muscle trying to move with no movement registered by the encoder. This triggered the E-Stim device and current was sent to the muscle, forcing the muscle to complete the extension movement. All processes worked correctly setting the ground work for a very versatile and efficient system. ::: {#F14 .fig} Figure 14 ::: {.caption} ###### Human subject wearing the instrumented biofeedback knee device. ::: ![](1743-0003-2-18-14) ::: ::: {#F15 .fig} Figure 15 ::: {.caption} ###### Test on quadriceps muscles (extension).(a) Position; (b) Velocity; (c) EMG Biofeedback; (d) Electrical stimulation. ::: ![](1743-0003-2-18-15) ::: Conclusion ========== The designs of the three systems described involve the meshing of standard, mechanical solutions and novel, new age ones. Besides being innovative the devices were all designed with a few key factors in mind. Compactness and portability were desired to compete with present day equipment that is currently quite large, permanent, and tedious to set-up and use. This was obviously met in each case, where all three devices are wearable, can be transported easily and involve computer control that simplifies their use significantly. Finally, the prototypes of portable rehabilitation devices presented here did demonstrate that these concepts are capable of the performance their commercially available but non-portable counterparts exhibit. After proving this feasibility, analyzing the efficiency of these devices is the next obvious phase. All three devices will be undergoing redesigning and construction to optimize their operation in order to proceed to human testing. Collaborations with large companies have been established for the development of specialized braces for these projects. Additionally, collaborations and arrangements are already in place for human testing at Spaulding Rehabilitation Hospital located in Boston, Massachusetts. If these tests are successful, they will open the door to a new era in rehabilitation where the recovery process could be dramatically improved through the use of a whole new breed of rehabilitation devices. Acknowledgements ================ Special thanks to Paul Canavan and Sue Lowe, of the Northeastern University Physical Therapy Department; Doctors Joel Stein and Paolo Bonato, of the Spaulding Rehabilitation Hospital in Boston MA; and Dr. Peter Gerbino of Children\'s Hospital in Boston MA, for their advice and input on the projects.	PubMed Central	2024-06-05T03:55:59.544365	2005-7-12	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180467/", "journal": "J Neuroengineering Rehabil. 2005 Jul 12; 2:18", "authors": [ { "first": "Constantinos", "last": "Mavroidis" }, { "first": "Jason", "last": "Nikitczuk" }, { "first": "Brian", "last": "Weinberg" }, { "first": "Gil", "last": "Danaher" }, { "first": "Katherine", "last": "Jensen" }, { "first": "Philip", "last": "Pelletier" }, { "first": "Jennifer", "last": "Prugnarola" }, { "first": "Ryan", "last": "Stuart" }, { "first": "Roberto", "last": "Arango" }, { "first": "Matt", "last": "Leahey" }, { "first": "Robert", "last": "Pavone" }, { "first": "Andrew", "last": "Provo" }, { "first": "Dan", "last": "Yasevac" } ] }
PMC1180468	Background ========== Recent developments in cell-free systems offer new and promising possibilities for producing recombinant proteins \[[@B1]\]. The RTS, commercialized by Roche, is an exchange cell-free system with improved productivity \[[@B2]\]. The continuous supply of consumable substrates and removal of reaction products provide a yield of several milligrams of protein. This system uses bacteriophage T7 RNA polymerase to perform transcription, while an enriched E. coliS30 extract provides the translational machinery. Thus, protein production in RTS requires a preliminary cloning step of the target gene into a vector, downstream of the T7 promoter. For this purpose, the pIVEX family of expression plasmids has been optimized for in vitrouse. They include a T7 promoter comprising the T7 gene 10 translation enhancer \[[@B3]\], an efficient prokaryotic Shine-Dalgarno sequence with an optimum distance to the start codon, a multiple cloning site, and a T7 terminator which prevents 3\'-terminal exonucleolytic degradation of the mRNA. These vectors are very convenient since multiple cloning sites were designed to allow gene fusions with several tags either at the N- or C-terminal of target proteins by conserving the same restriction strategy and reading frame compatibility. Some investigations revealed that less than optimal expression in this cell-free system could be explained by the presence of stable secondary structures in mRNAs \[[@B4]\]. Indeed, transcription levels are kept very high due to the highly active T7 RNA polymerase. Consequently, no real coupling of transcription and translation can take place in vitro, and unprotected mRNAs may form stable secondary structures, notably in the translation initiation region that inhibit ribosome binding and limit expression levels. There are mRNA folding algorithms that can predict such unfavourable intramolecular secondary structures, but they do not give information about expression levels. Although similar regulatory mechanisms exist in E. coli\[[@B5]\], it is simple and useful to assess in vivoexpression levels from a pIVEX plasmid which gave poor protein yields in vitro. The influence of mRNA secondary structures on translation is not identical in both contexts. However, transformation of the most widely used T7 RNA polymerase-producing BL21(λDE3) strain \[[@B6]\] by pIVEX is impaired because the absence of the lacIgene, coding the lactose repressor, in these high copy number plasmids. Here, we report how to solve this problem by designing a simple protocol with a compatible plasmid carrying the lactose repressor gene. This method allows the direct comparison of in vitroand bacterial expression from pIVEX vectors. Results and Discussion ====================== Unlike other T7 promoter-based vectors, the pIVEX do not contain a lacoperator sequence downstream of the T7 promoter. Since expression from pIVEX is not repressed by LacI, they do not contain the corresponding gene. In fact, first attempts to transform, by a standard chemical procedure, the BL21(λDE3) strain with different pIVEX failed since no transformant was obtained on LB ampicillin-agar plates. On the assumption that basal expression from pIVEX may have adverse effects on bacterial growth, we decided to test various plasmid/host combinations in order to control more tightly transcription both at the lacUV5promoter of the T7 RNA polymerase gene in the host chromosome, and consequently at the T7 promoter in the pIVEX plasmid. Furthermore, other investigations indicated that the transformation efficiency of this E. coliB strain could be critical \[[@B7]\]. Therefore, we electroporated freshly BL21(λDE3) competent cells containing either the pLysS plasmid encoding T7 lysozyme \[[@B8]\], a natural inhibitor of T7 RNA polymerase, or the pDIA17 plasmid \[[@B9]\] harboring the lacIgene. Both resident plasmids are chloramphenicol resistant and compatible with pIVEX since they carry the origin of replication from plasmid p15A. To facilitate the analysis of expression among the recombinants, a pIVEX-GFP plasmid encoding the green fluorescence protein, GFP, was used for fluorescence screening of single colonies on agar plates. As a control for toxic expression, the same strains were also transformed with an empty pIVEX plasmid (pIVEX2.4d). The results presented in Table [1](#T1){ref-type="table"} show that 1) the presence of the cloned gene coding GFP into pIVEX has no influence on transformation efficiency, 2) pLysS did not increase the tolerance of BL21(λDE3) for either of the pIVEX plasmids employed, and 3) in contrast, pDIA17 increases the transformation efficiency by 4 orders of magnitude. These results suggest that the basal production of T7 RNA polymerase in BL21(λDE3) was incompatible with a high-copy number plasmid which does not encode the lacIgene, even without a cloned gene downstream the T7 promoter. Yet, a substantial increase in lactose repressor from pDIA17 in BL21(λDE3) cells is sufficient to permit the presence of pIVEX. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Transformation efficiencies of pIVEX vectors into BL21(DE3) strain ::: ------------------------------------------------------------------------------------------------------------------------------------------------------------ Resident plasmid (p15A derivative) Incoming plasmid (pIVEX vector) Transformation efficiency (transformants per μg of DNA)^a^ Colony fluorescence^b^ ------------------------------------ --------------------------------- ------------------------------------------------------------ ------------------------ none pIVEX2.4d 3.1 × 10^4^± 0.8 \- none pIVEX-GFP 3.5 × 10^4^± 0.6 85 % F\ 15 % NF pLysS pIVEX2.4d 4.5 × 10^4^± 0.8 \- pLysS pIVEX-GFP 4.6 × 10^4^± 0.7 90 % F\ 10 % NF pDIA17 pIVEX2.4d 7.5 × 10^8^± 1.5 \- pDIA17 pIVEX-GFP 8.2 × 10^8^± 2.5 100 % F ------------------------------------------------------------------------------------------------------------------------------------------------------------ a Results are mean of three experiments ± SD on LB plates containing antibiotics. b Transformation plates containing 100 to 200 colonies were observed under UV illumination, NF; non-fluorescent, F; fluorescent ::: Furthermore, the production of GFP could be directly assessed from the transformation plates. At 37°C in the absence of IPTG, GFP production from pIVEX-GFP, after overnight cell growth on solid LB medium, was sufficient to give fluorescent colonies under these conditions. These high levels of leaky expression is puzzling, but it has been shown that when uninduced BL21(λDE3) cells are grown to stationary phase in LB medium lacking glucose, the lacUV5promoter is derepressed \[[@B10]\]. Apparently, LacI levels from pDIA17 are sufficient to increase pIVEX tolerance in the competent state, but not to ensure stringent repression in the uninduced stationary state. Examination of transformation plates revealed that 100 % of colonies from BL21(λDE3)/pDIA17 cells were fluorescent, but only 90 % from those bearing pLysS (Table [1](#T1){ref-type="table"}). This result suggesting that some transformants lacked GFP production, is in good agreement with a recent report in which the highest production level of a recombinant hemoglobin, from a comparison of both strains, was found in the BL21(λDE3)/pDIA17 strain \[[@B11]\]. Although all transformants obtained with pIVEX-GFP in this strain displayed bright fluorescence on transformation plates, a similar lack of GFP production was observed by subculturing repeatedly these cells in liquid LB medium, containing both ampicillin and chloramphenicol. Indeed, many studies have already reported that poor protein synthesis or complete lack of protein production is a common and frequent problem when using the T7 promoter expression system. Production of toxic proteins causing plasmid instability is the conventional explanation for this observation \[[@B8]\]. However, a recent investigation has suggested that the decrease in protein production levels in the BL21(λDE3) strain is more attributable to chromosomal mutations reducing the level of functional T7 RNA polymerase than to significant plasmid loss or to mutations arising on the plasmid \[[@B12]\]. Consistent with this study, isolation of pIVEX-GFP DNA from the subcultured non-expressing cells and transformation into new competent BL21(λDE3)/pDIA17 cells restored GFP production. To compare both in vitroand in vivoexpression, we used the maltose-binding protein (MalE), the soluble receptor for the high-affinity transport of maltose of E. coli, as a model protein \[[@B13]\]. Previously, we showed that some additional sequences at the N-terminus of this protein could influence its production level in the RTS cell-free system \[[@B14]\]. The four used pIVEX plasmids allow protein fusion with two different peptide-tags, His- and Strep-tag, either at the N- or C-terminus of MalE \[[@B15]\]. Among these, one construct (pIV2.2ME) yielded only very low levels of protein in the cell-free expression system (Figure [1](#F1){ref-type="fig"}). Indeed, when the Strep-tag was fused to the N-terminus of MalE, the corresponding protein was undetectable on SDS-polyacrylamide gels stained with Coomassie blue. Since His-tagged MalE at the N-terminus was correctly produced, we hypothesized that the Strep-tag sequence rather than the N-terminal position had a negative influence, probably by forming an unfavorable secondary structure in the translation initiation region of mRNA, which hampers protein synthesis. Because this phenomenon could be minimized inside E. colicells, we examined malEexpression in BL21(λDE3) from the same four pIVEX plasmids. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Comparison of MalE overproduction between in vitroand in vivoexpression systems. RTS500 E. coliHY extracts (A) and BL21(DE3)/pDIA17 whole cell lysates were separated on 12% SDS-polyacrylamide gel electrophoresis followed by staining with Coomassie blue. Lanes 1, molecular weight marker; lanes 2, pIVEX2.4d; lanes 3, pIV2.1ME; lanes 4, pIV2.2ME; lanes 5, pIV2.3ME; lanes 6, pIV2.4ME. ::: ![](1475-2859-4-18-1) ::: To ensure correct cellular protein production, we freshly transformed BL21(λDE3) carrying pDIA17 with the various plasmids, and instead of selecting transformants on LB agar plates supplemented with ampicillin and chloramphenicol, the selection was carried out in liquid LB medium containing both antibiotics. Next day, the saturated cultures were diluted (1:100) into fresh liquid LB medium, supplemented only with ampicillin, and incubated at 30°C. We found that processing the whole population of transformants was less tedious than screening individual colonies for correct expression, and by lowering the growth temperature to 30°C, the copy number of the pUC plasmid was decreased \[[@B16]\]. Subsequently, cultures were induced during the log phase, and cells were harvested 1.5 h after adding IPTG. Following lysis, the steady-state levels of MalE were analyzed by SDS-polyacrylamide gel electrophoresis. As shown in Figure [1](#F1){ref-type="fig"}, all pIVEX yielded very large amounts of MalE under these conditions. Although expression from pIV2.2ME was lower than from other plasmids, the corresponding protein was easily detectable on the gel stained by Coomassie blue. Thus, in contrast to RTS expression, the N-terminus Strep-tagged MalE is correctly produced in E. colicells. In practice, a successfull cellular expression may help to validate a pIVEX plasmid in which the target gene is correctly introduced between the regulatory elements, but gave no detectable expression in vitro. Conclusion ========== Using a correct plasmid/host combination and a simple transformation-expression protocol, we could directly compare the production of the same pIVEX-encoded proteins in two related expression systems, but performed in vivoand in vitro. With the development of cell-free systems in structural genomic and functional proteomic programs, it will be interesting to systematically investigate, on a large set of proteins, whether expression behaves similarly in both environments. Materials and methods ===================== Bacterial strain and transformation ----------------------------------- E. coliB strain BL21(λDE3) was obtained from Novagen. Electroporation was performed using an Eppendorf Electroporator 2510. Bacteria were grown in 100 ml LB to A~600~of 0.7, chilled on ice, and harvested by centrifugation (10 min, 1000 g at 4°C). The pellet was washed twice with 100 ml ice-cold distilled water, and once with 5 ml ice-cold 10 % glycerol. The last bacterial pellet was suspended in a final volume of 500 μl in 10 % glycerol. Aliquots (50 μl) were mixed with 10 ng plasmid DNA in chilled cuvettes (0.2 cm electrode gap). A simple pulse of 12.5 kV/cm was applied and 1 ml of SOC was immediately added. Then, electroporated bacteria were transferred to polypropylene culture tubes and shaken for 1 h at 37°C before either plating on LB plates containing ampicillin (100 μg/ml) and chloramphenicol (25 μg/ml), or diluting into liquid LB medium supplemented with both antibiotics. Plasmids -------- Plasmid pLysS was obtained from Novagen, and pIVEX-GFP, encoding the GFP cycle 3 variant \[[@B17]\], was a generous gift from Cordula Nemetz. Plasmids pIV2.1ME, pIV2.2ME, pIV2.3ME, and pIV2.4ME, carrying the wild-type malEgene without signal sequence, under the control of T7 promoter, were previously constructed \[[@B14]\] by subcloning the same PCR product into the corresponding pIVEX2.1MCS, pIVEX2.2bNde, pIVEX2.3MCS, and pIVEX2.4bNde vectors (Roche). Plasmid pDIA17 is a pACYC184 derivative carrying the lacIgene under the control of the tetracycline promoter \[[@B9]\]. Protein synthesis reaction -------------------------- In vitroMalE synthesis was performed in RTS500 E. coliHY with the RTS ProteoMaster instrument, essentially as described in the Instruction Manual from Roche. The coupled transcription/translation reactions, initiated by adding 15 μg of pIVEX plasmid DNA, were carried out in 1 ml total volume for 20 hours at 30°C. In vivoexpression was performed in BL21(λDE3) cells bearing pDIA17 plasmid at 30°C in LB medium supplemented with ampicillin (100 μg/ml). Cultures were induced at an A~600~of 0.8 by addition of IPTG to a final concentration of 500 μM, and growth was allowed for 1.5 hours after addition of IPTG. Proteins from both in vitroand in vivoextracts were separated by 12% SDS-polyacrylamide gel and stained by Coomassie blue. Abbreviations used ================== GFP -- green fluorescent protein IPTG -- isopropyl -- β-D-galactoside LB -- Luria Bertani MalE -- maltose binding protein PCR -- polymerase chain reaction RTS -- Rapid Translation System SDS -- sodium dodecyl sulfate Authors\' contributions ======================= JR performed bacterial transformations. JMB performed protein synthesis and prepared the manuscript. Both authors read and approved the final manuscript. Acknowledgements ================ We thank Cordula Nemetz and Hélène Munnier-Lehmann for the gift of plasmids, and Emmett Johnson for carefully reading the manuscript. This work was supported by grants from Institut Pasteur and CNRS.	PubMed Central	2024-06-05T03:55:59.548478	2005-6-2	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180468/", "journal": "Microb Cell Fact. 2005 Jun 2; 4:18", "authors": [ { "first": "Julie", "last": "Rogé" }, { "first": "Jean-Michel", "last": "Betton" } ] }
PMC1180469	Background ========== Clinical studies have established that individuals with diabetes have a two- to six-fold increased risk for various manifestations of cardiovascular disease (CVD), compared to age matched nondiabetic subjects \[[@B1]-[@B3]\]. Although individuals with diabetes have a higher prevalence of traditional CVD risk factors (e.g., hypertension and dyslipidemia) compared to nondiabetics, these risk factors do not fully account for the excess mortality associated with diabetes \[[@B4]\]. Even before the clinical diagnosis of type 2 diabetes, a significantly elevated risk of CVD was found in a large cohort of women in the U.S. who subsequently developed diabetes compared to women who remained nondiabetic \[[@B5]\]. Thus, having diabetes is an established independent risk factor for the development of CVD. In an early meta-analysis assessing the relationship between plasma total homocysteine (tHcy) and CVD \[[@B6]\], it was reported that elevated tHcy was strongly and independently related to several manifestations of CVD in the general population. Subsequent meta-analyses have confirmed the significant relationship between tHcy and CVD risk in healthy populations \[[@B7],[@B8]\], although the strength of association in these studies was weaker than reported previously. These meta-analyses were based on studies in healthy populations, thus, the relationship between tHcy and CVD risk in subjects with a disease highly related to CVD risk itself, such as diabetes, is largely unknown. In a large cohort of 50--75y men from the Netherlands \[[@B9]\], the risk for any CVD was increased with a parallel elevation in tHcy across subjects stratified by glucose tolerance -- from normal glucose tolerance to impaired glucose tolerance to type 2 diabetes -- even after adjustment for traditional CVD risk factors and serum creatinine. In a prospective analysis of this cohort, elevated tHcy was related to 5-year mortality \[[@B10]\] and coronary events \[[@B11]\] in subjects with type 2 diabetes, independent of other CVD risk factors. In these studies, elevated tHcy was a stronger risk factor for CVD in diabetic than in nondiabetic individuals. It is unknown why tHcy is elevated in persons with diabetes. Elevated tHcy was more common in patients with complications of type 2 diabetes, compared to patients with uncomplicated disease and controls \[[@B12]\]. Increased tHcy was also reported in patients with poorly controlled type 2 diabetes, compared to well controlled patients and matched controls in a clinic based study in Poland \[[@B13]\]. Results of a recent prospective study in an older cohort of Italian adults \[[@B14]\] demonstrated that tHcy was decreased in patients with type 2 diabetes who had a modest improvement in metabolic control (assessed by glycohemoglobin, HbA~1c~), whereas tHcy increased in subjects who had an increase in HbA~1c~and was unchanged in patients who had no change in HbA~1c~. Poor metabolic control was also associated with elevated tHcy levels in patients with type 1 diabetes \[[@B15]\], consistent with studies in which lower tHcy levels were reported for well controlled, insulin-treated type 1 diabetics than in controls \[[@B16]\]. Together, these studies suggest that tHcy levels in persons with diabetes may be at least partially related to metabolic control. Similarly, in-vitroevidence demonstrates that important enzymes of intracellular homocysteine metabolism were directly regulated by insulin and glucose concentrations \[[@B17]\]. This study tested the hypothesis that fasting plasma glucose status independently predicts fasting tHcy in a large, representative sample of non-diabetic U.S. adults. This study also examined correlates of fasting tHcy among a large sample of U.S. adults with a self-report history of diabetes. Methods ======= This report is based on data from the combined 1999--2000 and 2001--2002 National Health and Nutrition Examination Survey (NHANES). The study design is a stratified, multistage probability sample of the civilian non-institutionalized U.S. population. Approximately 9,965 persons aged 2 months to 85 years were studied in NHANES 1999--2000, and 11,039 in NHANES 2001--2002. A sub sample of over 3,000 individuals from each survey was invited to attend a morning examination after having fasted overnight. The data collection procedure involved an initial home interview component comprised of a screener, sample person, and family interview questionnaire. All interviewed persons were then invited to complete the health examination component of the survey in a mobile examination center, which included a series of health questionnaires, a physical examination, and a laboratory component. Fully informed consent was obtained from all participants as approved by the National Center for Health Statistic\'s Institutional Review Board. Details of the NHANES protocol are available elsewhere. Briefly, height was measured in an upright position with a stadiometer, and weight at a standing position on a self-zeroing scale. Blood pressure measurements were performed by trained technicians using a standardized protocol. Three and sometimes four measurements were made on all subjects with a mercury sphygmomanometer, and the first and fifth Korotkoff sounds were recorded to represent the systolic and diastolic pressures. We used the average of three recorded measurements in all data analyses. Blood analytes were stored under appropriate frozen conditions until they were shipped to a central laboratory for analysis. Plasma tHcy was measured using a flourescence polarization immunoassay (Abbott Diagnostics). Plasma glucose was measured using an enzymatic technique employing the hexokinase/glucose-6-phospate dehyrogenase reaction. Glycohemoglobin was measured in whole blood using a fully automated glycohemoglobin analyzer (Primus Instruments) and boromate affinity high-performance liquid chromatography. Total cholesterol was measured in serum or plasma using an enzymatic technique. Serum levels of folate and vitamin B~12~were measured using a radioassay kit (Bio-Rad Laboratories), and serum creatinine was measured using a photometric method employing the Jaffe reaction. Creatinine clearance was estimated (C~Cr~) using the Cockcroft-Gault formula \[[@B18]\]. Data from each survey were linked using the unique survey participant identifier (SEQN). The primary analyses consisted of 3,485 adults (20--85y) with a fasting (at least 8 hours) plasma glucose value recorded and who responded \"No\" to a lead-in question on diabetes history (see below). Subjects with a fasting plasma glucose \< 100 mg/dL were categorized as having normal fasting glucose (NFG), those with a value ≥ 100 \< 126 mg/dL as having impaired fasting glucose (IFG), and those with a level ≥ 126 mg/dL as having a provisional diagnosis of diabetes (DFG) \[[@B19]\]. The second set of analyses consisted of 275 adults (20--85y) based on a personal interview on diabetes, including use of medications and symptoms associated with diabetes. All subjects were asked a lead-in question pertaining to history of diabetes (\"Have you ever been told by a doctor or health professional that you have diabetes or sugar diabetes?\"). Subjects with a missing response, those who refused to respond, and those who responded \"No\", \"Don\'t know\", or \"Borderline\" were excluded from these analyses. Pregnant females were also excluded from both sets of analyses. Statistical Analysis -------------------- Data were analyzed using SAS (Version 9.1) survey procedures. Both analyses used the four-year fasting weights (WTSFA4YR) to estimate means and 95% confidence intervals, and the masked variance units (pseudo-primary sampling units \[SDMVPSU\] and pseudo-stratum \[SDMVSTRA\]) to estimate standard errors of those means. Continuous variables were analyzed using PROC SURVEYMEANS and are presented as a mean (± standard error). Categorical variables were analyzed using PROC SURVEYFREQ and are presented as a frequency (n), weighted frequency (based on sampling weights), and proportion \[with 95% confidence interval\]. Proportions are based on a percentage of the weighted frequencies. Differences in continuous variables were tested univariately using the t-test for independent samples (using PROC SURVEYREG), and prevalence values for categorical variables were compared using the χ^2^test for proportions (in PROC SURVEYFREQ). Statistical significance was established at α = 0.05 a priori, and multiple comparisons were adjusted using the Bonferroni method. Results ======= The primary analyses included a sample of 3,485 adults. By definition (see Methodssection), none of these subjects had a self-report history of diabetes, and none reported taking insulin or diabetic pills to lower blood sugar. The sex distribution was roughly 50% male and female, and race distribution 24% Mexican-American, 5% other Hispanics, 51% non-Hispanic white, 18% non-Hispanic black, and 3% other (including multicultural). Using population-based sample weights, this was equivalent to a population of 179,557,944 adults 45.5 ± 0.6y with sex distribution 49% male and 51% female, and race distribution 7% Mexican-American, 6% other Hispanics, 73% non-Hispanic white, 10% non-Hispanic black, and 4% other. There were 114 subjects in this sample with a self-report of coronary heart disease (CHD) and 91 with a self-report of stroke, equivalent to less than 3% of this population with CHD or stroke. Differences in physiological variables by fasting glucose status are presented in Table [1](#T1){ref-type="table"}. Age, tHcy, glucose, HbA~1c~, insulin, BMI, and systolic blood pressure were significantly higher in IFG and DFG groups than the NFG group (Ps\< 0.017). In addition, total cholesterol and serum creatinine were higher (Ps\< 0.017) in IFG than NFG. Furthermore, age, glucose, HbA~1c~, insulin, BMI, and systolic blood pressure were higher in DFG than IFG (Ps\< 0.017). There were no differences (P\> 0.05) among groups with respect to vitamin status (serum folate and B~12~) and C~Cr~, although the comparison between DFG and IFG with respect to C~Cr~just missed significance (P= 0.02, adjusted α = 0.17). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Select physiologic variables by fasting plasma glucose status among 3,485 U.S. adults (≥ 20y) with no self-reported diabetes. ::: Variable *NFG (n= 2,244)* *IFG (n= 1,108)* *DFG (n= 133)* -------------------------- ---------------------- ---------------------- -------------------- tHcy (μmol/L) 7.4 \[7.3--7.5\] 8.3 \[8.1--8.5\]\* 8.6 \[8.0--9.2\]† Age (years) 42.3 ± 0.6 52.1 ± 0.7\* 58.2 ± 2.3†‡ Glucose (mg/dL) 90.8 ± 0.2 106.9 ± 0.2\* 170.8 ± 8.5†‡ HbA~1c~(%) 5.2 ± 0.0 5.5 ± 0.0\* 7.2 ± 0.3†‡ Insulin (uU/mL) 9.8 ± 0.1 14.1 ± 0.4\* 26.4 ± 1.9†‡ BMI (kg/m^2^) 26.7 ± 0.1 29.3 ± 0.2\* 33.5 ± 1.3†‡ SBP (mmHg) 120.5 ± 0.6 127.5 ± 0.6\* 135.7 ± 2.8†‡ TC (mg/dL) 198.9 ± 1.4 209.8 ± 1.6\* 211.2 ± 5.9 Folate (ng/mL) 14.1 ± 0.3 14.5 ± 0.3 14.4 ± 0.9 Vitamin B~12~(pg/mL) 492.2 ± 4.2 486.0 ± 8.5 486.5 ± 20.9 Creatinine (μmol/L) 68.4 ± 0.8 75.9 ± 2.4\* 72.2 ± 2.9 C~Cr~(mL/min) 131.6 ± 1.9 126.5 ± 2.5 142.5 ± 10.4 Data are from the combined 1999--2000 and 2001--2002 National Health and Nutrition Examination Surveys. Non-standard abbreviations are NFG, normal fasting glucose, IFG, impaired fasting glucose, DFG, provisional diagnosis of diabetes, tHcy, total plasma homocysteine, SBP, systolic blood pressure, TC, total lipoprotein cholesterol, and C~Cr~, estimated creatinine clearance. Values are mean ± standard error, except for tHcy, which was log transformed because of its highly skewed distribution and is presented as median with 95% confidence interval (antilog values). \*P\< 0.017 for NFG vs. IFG † P\< 0.017 for NFG vs. DFG ‡ P\< 0.017 for IFG vs. DFG ::: Correlations between tHcy and physiologic variables, by fasting glucose status, are shown in Table [2](#T2){ref-type="table"}. In the total sample, age (r= 0.41) and C~Cr~(r= -0.38) were the strongest correlates of tHcy (Ps\< 0.0001). This was consistent when correlates of tHcy were examined by individual groups, with correlations for age ranging from r= 0.38 to 0.44, and for C~Cr~ranging from r= -0.35 to -0.46 (Ps\< 0.001). Glucose and HbA~1c~were weaker, but still significant, correlates of tHcy in all subjects (r= 0.12 and 0.10, Ps\< 0.0001), and in NFG (r= 0.11 and 0.12, Ps\< 0.0001) and IFG groups (r= 0.15 and 0.07, Ps\< 0.05). In contrast to the other groups, there was a negative relationship between tHcy and glucose (r= -0.14), and tHcy and HbA~1c~(r= -0.19), with only the latter being significant (P\< 0.05). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Relationships between total plasma homocysteine and select physiologic variables, by fasting glucose status, among 3,485 U.S. adults (≥ 20y) with no self-reported diabetes. ::: Group Age Glucose HbA~1c~ Insulin BMI SBP TC Folate B~12~ S~Cr~ C~Cr~ ----------- --------- ------------- ------------- ------------- --------- --------- -------- ------------ ----------- ----------- ----------- Total 0.41\* 0.12\* 0.10\* -0.00 -0.01 0.28\* 0.11\* -0.10\* -0.25\* 0.35\* -0.38\* NFG 0.38\* 0.11\* 0.12\* -0.05‡ -0.02 0.30\* 0.14\* -0.15\* -0.24\* 0.34\* -0.35\* IFG 0.38\* 0.15\* 0.07‡ -0.06 -0.07‡ 0.19\* -0.01 -0.05 -0.26\* 0.32\* -0.42\* DFG 0.44\* -0.14 -0.19‡ -0.21‡ -0.19‡ 0.13 0.05 0.15 -0.28† 0.55\* -0.46\* Data are from the combined 1999--2000 and 2001--2002 National Health and Nutrition Examination Surveys. Non-standard abbreviations are NFG, normal fasting glucose, IFG, impaired fasting glucose, DFG, provisional diagnosis of diabetes, SBP, systolic blood pressure, TC, total lipoprotein cholesterol, S~Cr~, serum creatinine, and C~Cr~, estimated creatinine clearance. Values are Pearson correlation coefficients (r) using log transformed total homocysteine (log tHcy) as the dependent variable. \*P\< 0.0001 †P\< 0.01 ‡P\< 0.05 ::: Results from univariate and multivariate regression models are presented in Table [3](#T3){ref-type="table"}. Fasting glucose status was a significant predictor of tHcy (Model 1, Ps\< 0.001). However, after adjusting for demographic and physiologic variables that influence homocysteine metabolism and/or fasting tHcy concentrations, fasting glucose status was no longer a significant predictor (Models 2 and 3, Ps\> 0.05). Next, fasting glucose status was examined after adjusting for age, sex, and a measure of kidney function only. Similar to previous findings, fasting glucose status was not a significant predictor when serum creatinine was included in the model (Model 4, Ps\> 0.05). However, fasting glucose status was a significant predictor when C~Cr~was included in the model (Model 5, Ps\< 0.05). HbA~1c~was a significant predictor of tHcy in the model using C~Cr~as the measure of kidney function (Model 3, P\< 0.05), but not in the model using serum creatinine (Model 2, P\> 0.05). ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Univariate and multivariate linear models predicting total plasma homocysteine among 3,485 U.S. adults (≥ 20y) with no self-reported diabetes. ::: IFG DFG Sex Age HbA~1c~ Insulin BMI SBP TC Folate B~12~ S~Cr~ C~Cr~ ------- ---------- --------- ---------- ---------- ------------- ------------- --------- --------- -------- ------------ ----------- ----------- ----------- 1 0.1181\* 0.1462† 2 0.02 0.08 0.1143\* 0.0077\* -0.03 -0.00 -0.00 0.0012‡ 0.00 -0.0010\* -0.0003\* 0.0020‡ 3 0.02 0.08 0.1598\* 0.0047\* -0.0307§ -0.00 0.0070‡ 0.0014† 0.00 -0.0010\* -0.0003\* -0.0021\* 4 0.01 0.01 0.1391\* 0.0068\* 0.0019‡ 5 0.0326‡ 0.0551§ 0.1738\* 0.0051\* -0.0012\* Data are from the combined 1999--2000 and 2001--2002 National Health and Nutrition Examination Surveys. Non-standard abbreviations are NFG, normal fasting glucose, IFG, impaired fasting glucose, DFG, provisional diagnosis of diabetes, SBP, systolic blood pressure, TC, total lipoprotein cholesterol, S~Cr~, serum creatinine, and C~Cr~, estimated creatinine clearance. Values are regression coefficients (β) from univariate and multivariate linear models using log transformed total homocysteine (log tHcy) as the dependent variable. For simplicity, the intercept term was omitted from the table but was significant (P ≤ 0.0001) in all models. Normal fasting glucose and female sex were used as reference groups. \*P\< 0.0001 † P\< 0.001 ‡ P\< 0.01 §P\< 0.05 ::: The analyses completed on subjects by self-report diabetes history included a sample of 275 adults, with an equal sex distribution, 34% Mexican-American, 7% other Hispanics, 36% non-Hispanic white, 19% non-Hispanic black, and 4% other (including multicultural). Using population-based sample weights, this was equivalent to a population of 12,322,266 adults 56.5 ± 1.1y with sex distribution 45% female and 55% male, and race distribution roughly 8% Mexican-American, 9% other Hispanics, 62% non-Hispanic white, 13% non-Hispanic black, and 8% other. Thirteen subjects (5% of this population) reported taking insulin, and 205 subjects (67%) reported use of diabetic pills to lower blood sugar. There were 28 subjects in this sample with a self-report of CHD and 13 with a self-report of stroke (about 8% of this population had CHD and 3.6% had stroke). The median tHcy level was 8.3 \[7.9--8.6\] μmol/L, and the average physiologic values were the following: fasting plasma glucose 154.6 ± 3.6 mg/dL, HbA~1c~7.2 ± 0.1%, fasting insulin 20.7 ± 1.7 uU/mL, BMI 31.4 ± 0.8 kg/m^2^, systolic blood pressure 131.7 ± 1.2 mmHg, total cholesterol 199.2 ± 2.6 mg/dL, serum folate 15.2 ± 0.8 ng/mL and vitamin B~12~548.3 ± 22.4 pg/mL, serum creatinine 73.6 ± 3.2 μmol/L, and C~Cr~138.7 ± 7.3 mL/min. Similar to findings among non-diabetic subjects, age (r= 0.41) and C~Cr~(r= -0.46) were the strongest correlates of tHcy (Ps\< 0.0001), whereas glucose and HbA~1c~were weaker, but still significant, correlates of tHcy (r= -0.15 and -0.17, Ps\< 0.05). Finally, results from univariate and multivariate regression models among subjects with a self-report history of diabetes are presented in Table [4](#T4){ref-type="table"}. In the full models (1 and 2), measures of kidney function (either serum creatinine or C~Cr~) were again the strongest predictors of tHcy, whereas neither HbA~1c~nor insulin were significant predictors of tHcy in these models. In contrast, in the abbreviated models controlling for sex, age, and a measure of kidney function (models 3 and 4), serum creatinine was (P\< 0.05), but C~Cr~was not, a significant predictor of tHcy. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Univariate and multivariate linear models predicting total plasma homocysteine among 275 U.S. adults (≥ 20y) with a self-report history of diabetes. ::: Sex Age HbA~1c~ Insulin BMI SBP TC Folate B~12~ S~Cr~ C~Cr~ ------- ---------- ---------- ------------- ------------- --------- --------- -------- ------------ ----------- ----------- ----------- 1 -0.07 0.0114\* -0.01 -0.00 0.00 0.00 -0.00 -0.00 -0.0004‡ 0.0025† 2 -0.1139† 0.0058‡ -0.01 -0.00 0.0154† 0.00 -0.00 -0.00 -0.00 -0.0022† 3 -0.07 0.0094\* 0.0021‡ 4 -0.0979‡ 0.0074† -0.00 Data are from the combined 1999--2000 and 2001--2002 National Health and Nutrition Examination Surveys. Non-standard abbreviations are SBP, systolic blood pressure, TC, total lipoprotein cholesterol, S~Cr~, serum creatinine, and C~Cr~, estimated creatinine clearance. Values are regression coefficients (β) from univariate and multivariate linear models using log transformed total homocysteine (log tHcy) as the dependent variable. For simplicity, the intercept term was omitted from the table but was significant (P ≤ 0.0001) in all models. Female sex was used as the reference group. \*P\< 0.0001 † P\< 0.01 ‡ P\< 0.05 ::: Discussion ========== This study demonstrates that fasting tHcy levels are elevated in non-diabetic subjects as a function of fasting glucose status; tHcy concentrations increased progressively with worsening fasting glucose. Subjects with self-reported diabetes had levels similar to levels found in non-diabetic subjects who had abnormally high fasting gluocse levels. The difference in fasting tHcy was 1.2 μmol/L between groups representing the extremes of fasting glucose, and 0.9 μmol/L between subjects with a self-report history of diabetes and those without a self-report history of diabetes and normal fasting glucose. The clinical significance of such small differences is unknown, particularly in light of the fact that tHcy levels were within the normal range (less than 15 μmol/L) in all groups of subjects. Although fasting tHcy levels differed as a function of fasting glucose status, results from correlation and multivariate regression analyses were in agreement that age and kidney function (measured as serum creatinine or estimated C~Cr~) were the major correlates of fasting tHcy concentrations among non-diabetic and diabetic individuals. In contrast, these analyses provided much less support for the concept that glucose metabolism has a major influence on fasting tHcy concentrations. These findings were somewhat surprising, based on previous reports demonstrating that fasting tHcy levels were related to metabolic control in patients with type 1 and type 2 diabetes \[[@B13]-[@B16]\]. In a clinic-based study \[[@B13]\], fasting tHcy levels were nearly doubled in poorly-controlled than well-controlled patients with type 2 diabetes. In a prospective study \[[@B14]\], fasting tHcy levels decreased in patients with type 2 diabetes who experienced modest improvements in metabolic control, whereas fasting tHcy increased in patients who had worsened metabolic control over follow-up. Because measures of kidney function (blood urea nitrogen and serum creatinine) and vitamin status (serum folate and B~12~) were within normal clinical limits at baseline and remained unchanged during follow-up, these findings imply a direct effect of metabolic control on fasting tHcy. However, in the present study HbA~1c~was only weakly correlated with fasting tHcy in non-diabetic subjects, irrespective of fasting glucose status. The relationship between HbA~1c~and tHcy was highest in non-diabetic subjects with normal fasting glucose, and was negative in those with a provisional diagnosis of diabetes, which was opposite to the expected direction. Similarly, HbA~1c~was weakly and negatively correlated with tHcy in diabetic subjects. In multivariate regression analyses, age and measures of kidney function (serum creatinine or C~Cr~) were the major predictors of fasting tHcy in non-diabetic and diabetic subjects, whereas HbA~1c~was not a significant predictor in either group of subjects. One factor that might explain why the present results differ from others with respect to the role of metabolic control on fasting tHcy is that our study includes current data obtained from a broad cross-section of the U.S. population, whereas other studies were conducted in The Netherlands \[[@B9]-[@B11]\], Poland \[[@B13]\], and Italy \[[@B14]\]. Differences in race and/or lifestyle of individuals living in the various countries represented in these studies may have contributed to the differential results. In addition, the U.S. instituted a program of cereal grain fortification in 1998, whereas the other countries listed do not have a fortification practice in place, which could have attenuated fasting tHcy levels in subjects from the U.S. The present findings were also somewhat surprising in that fasting insulin had little influence on fasting tHcy levels. Although relationships between fasting insulin and fasting tHcy were negative in non-diabetic subjects, which is in the expected direction, the strength of these correlations was weak, and were essentially non-existent in diabetic subjects (data not shown). Furthermore, fasting insulin was not a significant predictor in any model of fasting tHcy. This is somewhat in contrast to findings in which acute hyperinsulinemia led to a decrease in tHcy in normal subjects, but not in insulin resistant type 2 diabetes patients, suggesting that insulin had a lowering effect on tHcy, except in the case of patients who were resistant to this effect of insulin \[[@B20]\]. On the other hand, in support of the present findings a recent report \[[@B21]\] failed to show any influence of insulin resistance or degree of metabolic control on fasting tHcy levels in obese patients with type 2 diabetes patients treated with daily insulin over a one-month period. Although the findings with respect to metabolic control and insulin were somewhat surprising and in contrast to other studies, what was not surprising was that kidney function played a major role in modulating fasting tHcy concentrations. Because many of the studies mentioned previously only included a single measure of serum creatinine, used as a surrogate marker of renal function, estimated C~Cr~was included in the present analyses to provide a better assessment of the impact of kidney function on fasting tHcy levels. The rationale for this was that an estimate of glomerular filtration rate is the best measure of overall kidney function in health and disease \[[@B22]\]. Furthermore, the kidney has substantial tHcy handling and metabolizing capabilities \[[@B23]\] and thus it is possible that previous studies did not adequately account for differences in kidney function by using serum creatinine alone. The results of the present study support this contention. For example, estimated C~Cr~was a slightly stronger correlate of fasting tHcy than serum creatinine alone in both diabetic and non-diabetic subjects. In addition, estimated C~Cr~was a slightly stronger predictor of fasting tHcy in multivariate models. This is in contrast to a recent report \[[@B24]\] that fasting tHcy concentrations were closely and independently associated with estimated glomerular filtration rate, but not with serum creatinine, in groups of patients with type 1 and type 2 diabetes undergoing intensive insulin treatment. On the other hand, the present results are supported by previous studies, including Stabler and colleagues \[[@B25]\], and Davies and colleagues \[[@B26]\], who demonstrated the importance of renal function on modulating tHcy levels, at least in patients with type 2 diabetes. The present results also suggest that there is a differential impact of C~Cr~, compared to serum creatinine, on fasting tHcy levels. In non-diabetic subjects, fasting glucose status was a significant predictor of tHcy after adjusting for age, sex, and C~Cr~, but not when serum creatinine was substituted for C~Cr~. Likewise, sex and age, but not C~Cr~, were significant predictors of tHcy in the abbreviated models (see Models 3 and 4 in Table [4](#T4){ref-type="table"}), whereas age and serum creatinine, but not sex, where significant predictors of tHcy when serum creatinine was substituted for C~Cr~in the models. Some factors inherent to the study design may have contributed to the differential findings in this study compared to others, including differences in groups with respect to demographic and health status variables. For example, among non-diabetic subjects, the normal fasting glucose group had more females than males (\~ 56% vs. 44%), whereas both the impaired fasting glucose and provisional diagnosis of diabetes groups had more males than females (\~ 60% vs. 40% in both groups). This could have contributed to the higher tHcy levels in the latter two groups because fasting tHcy is typically higher in men than women \[[@B27],[@B28]\]. Furthermore, sex was a significant predictor of fasting tHcy in all regression models in non-diabetic subjects. In light of the differences in sex distribution, separate regression models were run for women and men to determine if there were differences in major correlates of fasting tHcy according to sex. Among females, in the full regression model using serum creatinine, age, insulin, total cholesterol, folate, B~12~, and creatinine were significant predictors of fasting tHcy. Among males, age, HbA~1c~, systolic blood pressure, folate, and B~12~, but not creatinine, were significant predictors of fasting tHcy. Among females, in the full regression model using C~Cr~, age, BMI, folate, B~12~, and C~Cr~were significant predictors of tHcy. Among males, age, HbA~1c~, systolic blood pressure, folate, B12, and C~Cr~were significant predictors of tHcy. Differences between the sexes with respect to major correlates of tHcy were also apparent in the abbreviated regression models. Among females, age and creatinine were both significant predictors of tHcy. Among males, age, but not creatinine, was a significant predictor of tHcy. Among females, the overall effect of fasting glucose status, age, and estimated C~Cr~were significant predictors of tHcy. Among males, fasting glucose status was not a significant predictor of tHcy, after adjusting for age and estimated C~Cr~(both terms were significant). Second, there were some differences with respect to self-reported health status among non-diabetic subjects. Roughly 2% of non-diabetic subjects with normal fasting glucose had a self-report of CHD, whereas 4% of non-diabetic subjects with impaired fasting glucose and 9% with a provisional diagnosis of diabetes reported CHD. The same pattern emerged for stroke (about 2%, 3%, and 7% for normal, impaired, and provisional diagnosis of diabetes groups). Because macroangiopathy \[[@B12]\] has been shown to impact fasting tHcy, at least in patients with type 2 diabetes, differences in the prevalence of CHD and stroke may have contributed to the group differences in tHcy. However, these differences were not so large that self-reported health status alone would likely account for differences in fasting tHcy levels among the various groups of non-diabetic subjects. The cross-sectional nature of NHANES does not allow for the determination of the underlying cause(s) of the elevated fasting tHcy levels in non-diabetic subjects with impaired fasting glucose/a provisional diagnosis of diabetes, or in diabetic subjects. However, results from the various analyses were consistent in demonstrating the importance of age and kidney function in modulating fasting tHcy levels, while providing much less support for an important role of glucose metabolism in this regard. The results of this study were also fairly consistent with those of Russo and colleagues \[[@B29]\], who found that age, creatinine, vitamin status, sex, smoking, methylene tetrahydrofolate reductase (MTHFR) genotype, and systolic blood pressure were significantly associated with tHcy in a clinic-based study of Italian adults with type 2 diabetes. Similarly, that study found no significant associations between tHcy and diabetes-related variables, such as diabetes duration, fasting glucose, HbA~1c~, current treatment regimen, and diabetes complications. Although data on MTHFR genotype was not available in NHANES, it is unlikely that large differences existed with respect to genotype distribution among the various groups because the data were drawn from a large, stratified random sample. However, this possibility cannot be entirely ruled out. Conclusion ========== This study supports the concept that age and kidney function, and not measures of glucose metabolism (fasting glucose, level of metabolic control, and insulin), are the primary correlates of fasting tHcy levels among non-diabetic and diabetic adults in the U.S. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= GED conceived the study, participated in its design and analysis, and drafted the manuscript. SML and XHZ performed the statistical analyses and edited the manuscript. All authors read and approved the final manuscript. Note ==== http://www.cdc.gov/nchs/about/major/nhanes/datalink.htm\#1999%20Current%20NHANES Acknowledgements ================ This work was supported by K01 DK61999 to GED.	PubMed Central	2024-06-05T03:55:59.550303	2005-5-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180469/", "journal": "Nutr Metab (Lond). 2005 May 26; 2:13", "authors": [ { "first": "Glen E", "last": "Duncan" }, { "first": "Sierra M", "last": "Li" }, { "first": "Xiao-Hua", "last": "Zhou" } ] }
PMC1180470	Introduction ============ Epidemiological studies have demonstrated an increased morbidity and mortality by particulate air pollution \[[@B1],[@B2]\]. The highest relative risk for mortality and hospital admissions were observed in subjects with existing pulmonary disease including asthma and COPD \[[@B1],[@B2]\]. The exact mechanism by which PM can adversely affect humans remains unknown, but several hypotheses have been forwarded. These include that PM causes pulmonary inflammation causing release of factors that influence blood coagulation \[[@B3]\], reduced lung function \[[@B4]\], increased blood plasma viscosity \[[@B5]\], reduced heart rate variability \[[@B6],[@B7]\] and destabilisation of atheromatous plagues \[[@B8]\]. Some of these effects are attributed to translocated nanoparticles based on their potential effects on vascular function \[[@B9],[@B10]\], blood coagulation \[[@B11]\], mitochondrial function \[[@B12]\] and Ca-flow \[[@B13],[@B14]\]. Nanoparticles have been shown to translocate from lung to the circulation \[[@B15]-[@B18]\], but most of the inhaled dose remains in the lung interstitium \[[@B19]\] even up to several years \[[@B20]\]. Therefore it seems that not the epithelial but the endothelial barrier is more important in prevention of translocation to the blood. Enhanced lung permeability has been measured by increased Clara-cell protein in blood \[[@B21]\] or enhanced DTPA clearance in the lung \[[@B22]\] after ozone and hyperoxia. Recent work in rabbit isolated perfused lungs shows that nanoparticles themselves can influence microvascular permeability measured by weight gain after occlusion \[[@B23]\]. However, since the mechanisms of nanoparticle transport on a sub-cellular level are unknown it remains to be determined whether the above indices of lung permeability are related to translocation of nanoparticles. Particles may also cause the release of vasoactive mediators such as histamine, which was shown to be increased in plasma of hamster after instillation of diesel exhaust particles \[[@B24]\]. Histamine is well known to induce vascular permeability through its action on endothelial H~1~-receptor \[[@B25]\]. Finally, by oxidative stress mechanisms ambient and nanoparticles can cause activation of lung alveolar macrophages and epithelial cells that result in the production of pro-inflammatory cytokines such as TNF and Il-1 in humans \[[@B26]\] and rat models \[[@B27]\] that are typically associated with increased lung permeability \[[@B28],[@B29]\]. The objective of this study was to assess nanoparticle translocation in relation to permeability changes for small molecules and integrity of epithelial and endothelial monolayers. In order to manipulate permeability in the absence of neutrophil recruitment and activation we used an isolated perfused rat lung. Several treatments to modify lung permeability in-vitro were applied including oxidative stress by instillation of hydrogen peroxide and endothelial permeability by histamine in the perfusate. These treatments were selected for their relevance to conditions of patients with pulmonary or systemic complications. Particle translocation was assessed by the inherent radioactivity of 18 nm size iridium nanoparticles (^192^Ir-UFP). Materials & methods =================== Animals and surgical procedure ------------------------------ Adult, healthy, male Wistar-Kyoto rats (WKY/Kyo@Rj rats, Janvier, France) (200--250 g) were housed in pairs in a humidity (55% relative humidity) and temperature (22°C) controlled room. They were maintained on a 12-h day/night cycle. Rats were allowed to acclimate to the facility for a minimum of 10 days prior to use. When the experiments were performed rats were more than 17 weeks of age. The studies were conducted under federal guidelines for the use and care of laboratory animals and were approved by the Oberbayern Government and by the GSF Institutional Animal Care and Use Committee. Surgical procedure for lung isolations was done according the method of Uhlig and Wollin \[[@B30]\]. Briefly, rats were anaesthetized intraperitoneally with 80 mg/kg ketamin. Deep anaesthesia was characterized by a lack of response to toe pinching. Heparin (500 IU) was injected via the tail vein. A midline incision was made from the pelvic region to the neck of the rat. With the ventilator operating, the trachea was cannulated using a rigid catheter and the catheter was attached to the ventilator. Therefore lung were ventilated from the start of the whole procedure. The animals were exsanguinated opening the aorta abdominalis after deep intraperitoneal anesthesia with ketamine (100 mg/100 g body weight) and xylazine (0.5 mg/100 g body weight). After anesthesia a longitudinal ventral incision was made to open the thoracic and abdominal cavity and it was held open using clamps. The thymus was removed and the apex of the heart was cut off to introduce a cannula into the pulmonary artery. A slight perfusion flow of around 1 ml/min was maintained before inserting cannula. Care was taken not to introduce any air bubbles into the pulmonary artery. The left atrium was cannulated by advancing the venous cannula through the mitral valve. A ligature was placed around the heart to keep both cannula\'s in place. The aortic cannula was then attached to the lining fed through the Perspex lid of the 500 mL negative-pressure-chamber. After the Perspex lid was fully mounted on the chamber negative pressure ventilation started. The respiratory settings during the negative pressure ventilation were 65 breaths per min. Regular sighs were introduced (hyperventilation) to improve function of lungs. Isolated lung perfusion ----------------------- The IPL-4401 Isolated lung ventilation perfusion system (FMI GmbH Oberbach) was used for our study. Additional negative pressure chamber has been constructed by GSF -National Research Centre for Environment and Health. The system in brief consists of a small animal ventilator, jacketed upper media reservoir, negative-pressure-chamber holding the heart-lung-bloc, perfusion lines, peristaltic pump and pulmonary artery pressure transducer. The computer software (FMI GmbH Oberbach) was operated by a 386 personal computer and allowed constant monitoring of pulmonary blood pressure. The upper media reservoir, ventilation chamber and perfusion lines were held at 37°C by a re-circulating water-bath. The perfusion medium was selected based on its extensive use in isolated organ perfusion and consisted of a modified Krebs-Ringer -bicarbonate buffer. Krebs-Ringer composition was as follows (mM): NaCl 118 ; KCl 5.9 ; CaCl~2~2.5 ; MgSO~4~1.2 ; NaH~2~PO~4~1.2 ; NaHCO~3~24.9 ; glucose 11.1, pH was adjusted at 7.4. It was then mixed at a ratio 1:1 with Haemacell solution (Hoechst Marion Roussel). The buffer was pre-warmed and gassed with 95% O~2~and 5% CO~2~at a rate low enough to prevent excessive frothing of the medium. The medium flowing through the system passed through a bubble trap prior to reaching the lungs and the buffer pH was continuously monitored throughout the experiment. Respiratory rate was set at 65 breaths per min. Lungs were inflated at a maximum negative pressure of -1.5 kPa in the chamber. Stroke volume was set at 10 ml to achieve a tidal volume of usually 3--4 ml (because of the altered compliance of the lungs). The lungs were expanded (sighed) every 4 minutes applying a negative pressure of -2.5 kPa to the chamber. Optimal perfusion settings included perfusion rate of 5 ml/min and a medium pH of 7.4. Perfusion pressure was not constant but was kept between 10 and 14 kPa. Particle generation and exposure -------------------------------- Aerosols of ultrafine iridium particles (Ir-UFP) radiolabeled with ^192^Ir were produced with a spark generator as described previously \[[@B15]\]. Size distribution and number concentration were monitored continuously by a differential mobility particle sizer (DMPS 3070, TSA instruments) and a condensation particle counter (CPC 3022A, TSI Instruments). The size distribution of the ^192^Ir-UFP was aimed to a count median diameter of 17--20 nm (geometric standard deviation 1.6) at a particle concentration of 10^7^cm^-3^aiming for a tidal volume of 3--4 cm^3^at a frequency of 65 /min. The estimated dose under these conditions is 180 μg/hour. A schematical description of the experimental system is shown in Figure [1](#F1){ref-type="fig"} ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Diagram of the experimental perfusion system used for this study. The ultrafine iridium particles (Ir-UFP) radiolabelled with ^192^Ir were produced in the spark generator. At the exit of spark generator the aerosol was quasi-neutralized by a radioactive ^85^Kr source. The aerosol was diluted with nitrogen and with oxygen and adjusted to obtain 20% oxygen and was air conditioned at 50--60 % relative humidity. The particle size distribution and number concentration were monitored by a differential mobility particle sizer (DMPS) and a condensation particle counter (CPC). ^192^Ir-UFP radioactivity of the aerosol was determined by continuous aerosol sampling of a measured volume and integral radioactivity counting. The lungs were perfused at a perfusion rate of 5 ml/min and a stroke volume of 10 ml. Respiration rate set at 65 breaths per minute. Negative ventilation pressure in chamber was regulated with animal ventilator. Lungs were manually expanded (sighed) every 4 minutes by applying a negative pressure of -2.5 kPa to the chamber. ::: ![](1743-8977-2-3-1) ::: Experimental design ------------------- Following a 15 minute period of equilibration, during which the lungs were already ventilated and perfused, the experiment started by the inhalation of freshly produced ^192^Ir-UFP from the aerosol line. Intratracheal instillation of ^99m^Tc-DTPA or other instillations were performed at this starting time point. The perfusate was collected continuously and sampled at 15-min time intervals (Figure. [2](#F2){ref-type="fig"}). The following treatments were investigated, ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Experimental protocol for isolated perfused rat lungs. All perfusions were done under negative pressure ventilation. Following a 15 minute period of equilibration, during which the lungs were already ventilated and perfused, the experiment started by the inhalation of freshly produced ^192^Ir-UFP from the aerosol line. Intratracheal instillation of ^99m^Tc-DTPA or other instillations were performed at this starting time point. The perfusate was collected continuously and sampled at 15-min time intervals. The following treatments were investigated, group 1: control group, only ^192^Ir-UFP inhalation for 120 min; group 2: instillation of 50--100 μL ^99m^Tc-DTPA, 500 μl H~2~O~2~bolus (0.5 mM), ^192^Ir-UFP aerosol inhalation for 120 min; group 3:instillation of 50--100 μL ^99m^Tc-DTPA, histamine continuously pefrused during the next 2 hours at concentration 10 μM, ^192^Ir-UFP aerosol inhalation for 120 min; group 4: instillation of 50--100 μL ^99m^Tc-DTPA and 500 μl histamine bolus instillation at a concentration of 10 mM, ^192^Ir-UFP aerosol inhalation for 120 min; group 5: instillation of 50--100 μL ^99m^Tc-DTPA. For each group 3--4 animals were used. ::: ![](1743-8977-2-3-2) ::: group I: control group, only ^192^Ir-UFP inhalation for 120 min, ; group II: instillation of 50--100 μL ^99m^Tc-DTPA, 500 μl H~2~O~2~bolus (10 mM), ^192^Ir-UFP aerosol inhalation for 120 min; group III: instillation of 50--100 μL ^99m^Tc-DTPA, histamine continuously perused during the next 2 hours at concentration 10 μM, ^192^Ir-UFP aerosol inhalation for 120 min; group IV: instillation of 50--100 μL ^99m^Tc-DTPA and 500 μl histamine bolus instillation at a concentration of 10 mM, ^192^Ir-UFP aerosol inhalation for 120 min; group V: instillation of 50--100 μL ^99m^Tc-DTPA. Evaluation of ^192^Ir-UFP translocation --------------------------------------- The perfusate samples as well as the heart-lung-blocs were analysed for ^192^Ir-UFP activity in a shielded 1-L-well-type gamma-spectrometer. Analysis of ^192^Ir activity was performed in those samples studied for ^99m^Tc-DTPA permeability when the ^99m^Tc activity had decayed -- see below. Activity measurements of both isotopes were decay and background corrected. ^192^Ir activity in the perfusate samples were given as a fraction of the total activity found in the perfusate and the heart-lung-bloc. Evaluation of lung permeability ------------------------------- Technetium-99^m^labelled DTPA (^99m^Tc-DTPA; DRN 4362 TechneScan-DTPA, Malinckrodt Medical BV, The Netherlands) was used to evaluate lung permeability. The lyophilised DTPA powder was dissolved in 10 ml sterile ^99m^Tc activity containing saline, which was eluted from the ^99m^Tc generator. The solutions were then allowed to equilibrate for 15 minutes at room temperature. The volume instilled in the trachea was 50--100 μL at a DTPA concentration of 120--250 μg. and a ^99m^Tc activity of 5--10 MBq. The ^99m^Tc-DTPA permeability was studied measuring the activity in the heart-lung-bloc and the perfusate samples. The ^99m^Tc radioactivity was also analysed in the shielded 1-L-well-type gamma-spectrometer at the appropriate photo peak of ^99m^Tc. Since the ^99m^Tc activity was chosen to be at least an order of magnitude higher than the ^192^Ir deposition in the lungs, interference of Compton rays in the ^99m^Tc window originating from ^192^Ir was negligible. Permeated ^99m^Tc activity in the perfusate samples was given as a cumulative fraction of the total instilled activity recovered in the perfusate and the heart-lung-bloc. Tissue preparation and microscopy --------------------------------- Immediately after the termination of the lung perfusion the radioactive particles treated lungs were air dried with room air at a pressure 3.5 kPa and subsequent imaging for particle distribution. For histopathology only lungs treated with non-radioactive iridium particles were used. After the experiment, the trachea and pulmonary vein of the IPL were perfused with 2.5 % glutaraldehyde in 0.1 M phosphate buffer (pH 7.2, 340 mOsm) at 25 cm fixative pressure. Post-fixation of the lungs was done by immersion of the whole lung in the same fixation solution for 2 hours at room temperature. t 25 cm fixative pressure. Two slices from left- and right caudal lobes of each animal were embedded in paraffin and 5 μm thick sections were stained with hematoxylin and eosin. Small portions of the left lobe of a a sub-group of 7 animals were embedded in Epon ^®^, and semithin sections (1 μm) were stained by toluidine blue. Biochemical analysis of the perfusate ------------------------------------- Perfusate samples were analysed for histamine with an ELISA kit from IBL-Hamburg (reference no. RE 59221). The detection limit of the kit was 0,3 ng/ml when using plasma. Angiotensin converting enzyme (ACE) was measured according the kinetic method of Maguire and Price \[[@B31]\] using standards from Bühlmann Laboratories AG, Switzerland (Reference KK-ACK), Protein determination was done according the Bicinchonic acid (BCA) protein assay \[[@B32]\]. Clara-cell protein was measured in perfusate using a sensitive latex immunoassay \[[@B21]\] with a detection limit 1 μg/l perfusate. Alkaline Phosphatase (ALP) determination was done with KIT manufactured by Diasys Diagnostics GmbH Germany Cat no: 104019990314. Samples measured (Beckman DU 640 spectrophotometer) at 25°C at wavelength 405 nm. The increase of the extinction was measured each minute for 3 min and enzyme activity was measured as the difference in extinction divided by the minutes multiply a constant factor 2754 \[[@B33]\]. Statistical Analysis -------------------- Results are expressed as means ± SD, and/or as individual experiments (Fig [5](#F5){ref-type="fig"}). Differences between treatments were tested for statistical significance by Mann Whitney-test. A value of P \< 0.05 was considered significant. All statistics were run with SPSS for Windows XP. Results ======= Particle distribution and deposition ------------------------------------ The first set of experiments measured the distribution of ^192^Ir -UFP related radioactivity in the lungs. Deterioration of lung performance was noted based on increasing frequency of inflation to maintain tidal volume, but could not be quantified during negative pressure perfusion in the experimental set-up due to maintain a closed system for radiation protection safety reasons. In a positive pressure using the same equipment, tidal volume, respiration pressure and weight did not change over a 2-hour perfusion period. The particle size distribution in the inhaled aerosol was well reproducible and the count median diameter (CMD) ranged between 16 and 18 nm of particle diameter (Fig [3](#F3){ref-type="fig"}). Geometric standard deviation (GSD) always was 1.6. Deposition of particles in isolated perfused lungs was compared to animals exposed parallel to the same aerosol and showed similar homogenous distribution, with somewhat lower deposition (data not shown). Translocation of ultrafine particles after modified permeability ---------------------------------------------------------------- In a large set of perfusions in control lungs no translocation of ^192^Ir-UFP particles was noted and the variance between different perfusions is small (\< 5 %) as shown in Fig. [4A](#F4){ref-type="fig"}. Then several treatments were applied to investigate the role of epithelial and endothelial permeability on particle translocation. First, hyperinflation to double tidal volume every minute was applied but did not lead to increased translocation of nanoparticles (data not shown). An initial bolus injection of H~2~O~2~into the trachea of the IPL to reach a final concentration of 0.5 mM, caused particle translocation to start at 60 min after onset of the inhalation of radioactive aerosol. A significant difference (P \< 0.05, Mann-Whitney U-test) in particle-related radioactivity in the perfusate was observed between control and H~2~O~2~group at 90, 105 and 120 minutes after onset of inhalation (Fig. [4A](#F4){ref-type="fig"}). At other time-points beyond 60 minutes the differences to untreated lungs were of borderline significance (P \< 0,1; Mann Whitney-test). The variance between perfusions upon this treatment in Fig [4A](#F4){ref-type="fig"} was much higher than in control perfusions. However, individual presentation of the experiments of H~2~O~2~pretreated lungs (Figure [5A](#F5){ref-type="fig"}) show a similar trend in all perfusions. Increased radioactivity in perfusate was only detected beyond 60 minutes of perfusion. A similar translocation versus time profile was observed in lungs upon presence of 1 μM histamine in the vascular perfusion fluid (Fig [4A](#F4){ref-type="fig"}). However, here statistical significance in this condition versus control lungs was only attained after 120 minutes of perfusion (Fig [4A](#F4){ref-type="fig"}), which is best explained by the individual experiments shown in Fig [5B](#F5){ref-type="fig"}. On the other hand, in the lungs treated with a histamine bolus injection no ^192^Ir-UFP radioactivity was detected in the perfusate. Interestingly, the kinetics of translocation of DTPA (Fig. [4B](#F4){ref-type="fig"}) and Ir-UFP are very different. Whereas Ir-UFP only starts to increase in perfusate after 60 min of inhalation, DTPA is measured in perfusate within a few minutes after intratracheal instillation. On the other hand the effects of H~2~O~2~and vascular histamine on particle translocation are also reflected in the DTPA -clearance (Fig. [4B](#F4){ref-type="fig"}). Although not significant, both treatments caused trends of a higher rate of translocation of DTPA one hour after administration, which is also observed for translocation of ^192^Ir-UFP. The histamine bolus injection, with a final target concentration in the lumen of 0.5 mM caused a considerable slowing-down of DTPA permeability (Fig [4B](#F4){ref-type="fig"}) and no observed effects on ^192^Ir-UFP translocation. Biomarkers of epithelial and endothelial damage ----------------------------------------------- Alkaline phosphatase (ALP) was measured in control and pre-treated lungs as a marker of type II cell damage. No significant differences in ALP activity (15--135 minutes) were observed between perfusate of the control IPL and H~2~O~2~pre-treated lungs after exposure to ^192^Ir-UFP (Fig [6A](#F6){ref-type="fig"}). In the IPL perfused with vascular histamine a significantly lower activity of ALP was seen at 15 and 30 minutes in comparison to lung perfusions that only received ^192^Ir-UFP by inhalation. At all later time points ALP showed no difference to control lungs (Fig [6A](#F6){ref-type="fig"}). The histamine bolus group did also not differ from control group. To evaluate endothelial damage, angiotensin converting enzyme (ACE) was measured in the lung perfusate (Figure [6B](#F6){ref-type="fig"}). No significant differences were observed in ACE activity between the control group and isolated lungs treated with H~2~O~2~, histamine in perfusate or histamine delivered as a bolus in the trachea (ANOVA, post-hoc Tukey and Mann Whitney-test). To check whether the H~2~O~2~effect was mediated by histamine release we measured histamine in the perfusate but did not detect significant differences to histamine levels in control perfusions (data not shown). Also a measure of total protein for lung permeability did not detect differences between the different treatments used in these experiments. Morphology of lungs ------------------- The outcomes of the histo-pathological analyses are summarised in Table [1](#T1){ref-type="table"} and illustrated in Fig. [7](#F7){ref-type="fig"}. Overall there is quite extensive damage at the end of the perfusion experiments, but there are not many differences between the different treatments. Sub-epithelial round cell infiltration, interstitial dilation along with moderate to severe oedema was present in all the groups. Occasionally alveolar dilation, alveolar inflammation and fluid in the alveolar lumen were detected. The in vitro perfusion procedure might be responsible for the perivascular and peribronchial dilation in all the lungs. H~2~O~2~might be directly responsible for the epithelial damage of the proximal bronchi in the H~2~O~2~instilled group. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Morphological evaluation of the rat lungs after 2 hours perfusion and inhalation exposure to Ir-UFP, using lung sections and HE- or toluidine blue staining The regions investigated included the bronchial segment, the blood vessels and the alveolar region ::: Treatment Bronchial region Blood vessels Alveolar region ------------------------- ------------------------------------------------------------------------------------- ------------------------------------------ ------------------------------------------------------------- Control Sub epithelial round cell filtration; interstitial dilation Interstitial dilation Small alveolar dilation in periphery Hydrogen peroxide Desquamation of epithelial layer, partially destruction of sub epithelial structure Moderate to severe dilation Histamine Perfusate Sub epithelial round cell filtration; interstitial dilation Moderate to severe interstitial dilation Alveolar dilation in periphery Histamine (Bolus) Sub epithelial round cell filtration; interstitial dilation Moderate to severe interstitial dilation Focal alveolar dilation, oedematous fluid in alveolar lumen ::: Discussion ========== The purpose of our study was to evaluate the role of epithelial and endothelial barrier in the translocation of ultrafine particles across the lung into the systemic circulation using the isolated perfused lung model. Translocation of Iridium (Ir)particles was monitored by radioactivity of the particles themselves, and not by any attached radioactive label. No translocation of ^192^Ir-UFP (17--20 nm) was detected in isolated perfused rat lungs. However lungs pre-treated in-situwith histamine on the endothelial side (1 μM) or H~2~O~2~(0.5 mM) in the alveolar lumen showed small amounts of radioactivity in the single pass perfusate after a lag-time of 60 min. Although kinetics of DTPA and particle translocation were different the in-situtreatments histamine and H~2~O~2~caused unidirectional in both processes, in the absence of biochemical evidence for epithelial and endothelial damage. In this study we applied several ex-vivotreatments of the isolated lungs with either H~2~O~2~or histamine. Using H~2~O~2~we anticipated inducing an oxidative stress, which is also caused by PM inhalation in the lung both by direct radical formation by PM constituents and indirectly by recruited inflammatory cells (review: \[[@B34]\]). Oxidative stress has been forwarded as a central hypothetical mechanism in the adverse effects of PM, including ultrafine particles \[[@B35]\]. Actually, the oxidative capacity of PM was shown by us to be a predictor of bronchial inflammatory response to PM after installation in normal human volunteers \[[@B26]\]. Earlier on Rhaman et al \[[@B36]\] forwarded that oxidative stress and depletion of GSH can affect lung permeability allowing for greater particle passage via lung epithelium into the interstitium. This concept is supported by our data using a high concentration of H~2~O~2~(5 mM) by bolus injection into the lung, to reach a final concentration of 0.5 mM. In fact, lung-lining fluid of COPD patients has been shown to contain levels of H~2~O~2~up to 5 μM \[[@B37]\]. Although this model does certainly not meet all conditions of an inflammatory response, similar models have been applied in other ex-vivo permeability studies \[[@B38],[@B39]\]. In isolated perfused rat lungs, a low concentration of H~2~O~2~(0.25 mM) in the perfusate was shown to increase capillary permeability in the absence of lipid peroxidation \[[@B38]\]. A short-term treatment with H~2~O~2~(100 mM) on the epithelium of human airway tubes caused a six-fold increase in translocation of ^111^In-DTPA, which was explained by the opening of paracellular pathways \[[@B39]\]. In our study, we assume that a final luminal concentration of 0.5 mM H~2~O~2~is reached and we found an increased translocation of both ^192^Ir-UFP as well as a trend of increased translocation of ^99m^Tc-DTPA after a lag-time of about 60 min. However, no temporal relationship between both markers of translocation was seen, which suggest that they operate through different routes. More data on the translocation route of ^192^Ir-UFP in the lungs are given by the results obtained with histamine administered both on the luminal side and through the microvasculature. These data show that histamine at very low levels in the perfusate (1 μM) caused an increased translocation of particles as well as ^99m^Tc-DTPA-permeation after a lag-time of about 60 min (Fig [4](#F4){ref-type="fig"}). On the other hand luminal administration of histamine (0.5 mM) did not increase ^192^Ir-UFP translocation and actually showed a slightly reduced ^99m^Tc-DTPA permeation. Histamine is a very potent vasoactive and bronchial mediator that has been shown to be involved in both local and systemic effects of diesel particles \[[@B40]\]. First, upon mast cell degranulation histamine is the major mediator of bronchial constriction as observed in allergic airway response \[[@B41]\]. This constriction probably also explains our reduction in ^99m^Tc-DTPA clearance after a histamine bolus injection into the trachea. The approach using vascular histamine is relevant because histamine has been shown to increase after instillation of particles in isolated tracheally perfused rabbit lung ex-vivo (Nemmar et al, 1999)\[[@B42]\], hamsters in vivo (Nemmar et al, 2003)\[[@B11]\] and healthy human volunteers (Salvi et al, 1999)\[[@B43]\]. We assume that histamine at low concentrations (10 ^-6^M) in our system increases endothelial permeability \[[@B44]\] and allows an increase in intercellular transport of ^192^Ir-UFP located in the interstitium through the endothelium into the perfusate. A similar effect of 10^-4^M vascular histamine was found recently in isolated perfused rabbit lungs (Nemmar et al, 2005)\[[@B45]\]. In the latter study latex particles (24--190 nm) translocated from the vascular compartment into the lumen, as observed by subsequent bronchoalveolar lavage. The amount of reverse translocation was about 2.5 % of administered dose within 2 hrs of perfusion. No translocation of latex particles with different size (24--190 nm) and surface chemistry (carboxylate versus amine) and charge were seen under normal physiological conditions in the rabbit lungs. In our studies 18 nm iridium particles also did not translocate through rat lung barriers, although we used negative pressure ventilation which caused considerable damage to the lung tissue. Taken together these findings suggest that the translocation of ultrafine particles in the lung occur through different routes. Among different uptake routes we can discriminate transcytosis and para(inter)-cellular transport. Hermans et al \[[@B46]\] stated that radiolabelled tracers such as ^99m^Tc-labelled DTPA permeate the epithelial barrier by passing through intercellular junctions. Since kinetics of DTPA and ^192^Ir-UFP translocation in untreated lungs is so different, we suggest that ^192^Ir-UFP are translocated along different pathways. In fact, ultrafine particles may use different transcytotic pathways such as clathrin-coated pits, pinocytosis and non-coated pits, called caveolae \[[@B47]\]. Caveolae are the most likely route of uptake for the ^192^Ir-UFP used in our study (18 nm) as derived from studies by Gumbleton \[[@B47]\]. The alveolar epithelial cells, comprising 95 % of the lung surface have a cell thickness of 400 nm and about 600.000 to 900.000 caveolae that are 50--60 nm wide. Such a high number of possible transport units lead to assumption that NP smaller than 60 nm can be rapidly taken up and transported through the epithelial cell layer. Recently Kato et al (2003)\[[@B48]\] showed that lecithin-coated polystyrene latex beads (240 nm) got incorporated into the Type I and II alveolar epithelial cells as well as in the capillary lumen. It was suggested that these latex beads move from the alveolar epithelial cells to the capillary lumen via transcytosis. Kapp et al (2004)\[[@B49]\] found Ti0~2~(29 nm) particles as intracellular clusters forming needle shape particles or rounded shape particles. This may also explain the relatively low and variable translocation as observed in other, previous in vivostudies \[[@B15]-[@B18]\]. We therefore used histamine to modify or facilitate trans-endothelial passage and indeed found that this enhanced translocation of ^192^Ir-UFP to the perfusate. However, we cannot discriminate between transcytosis and para-cellular transport in endothelium. We assume that the long lag-time (60 min) detected which is needed before passage is caused by the fact that an interstitial load has to be built by epithelial passage of ^192^Ir-UFP in the lung. The methods used in our study and the study by Nemmar et al (2005)\[[@B45]\] do not allow to evaluate whether translocation has occurred through primary particles or by aggregates. The inhalation in our study assured single UFP deposition in the alveolar region and virtually no agglomeration on the epithelium because of the alveolar surface and the number of deposited particles. However, upon vascular injection aggregates are formed, unless surface modifications are used that impede this process. It may therefore be that translocation observed in Nemmar\'s study occurs as aggregates by above mechanisms, or facilitation by phagocytic cells but not in our study. With this respect the studies by Heckel et al (2004)\[[@B50]\] have demonstrated by TEM that 4 nm gold-particles really pass membranes and reach the lumen as single particles. The difference between 4 and 20 nm particles may however be huge since 4 nm AU-particles are not recognized by the reticulo-endothelial system. Our findings may be criticized due to a number of factors that are associated to our experimental design and performance. First, it must be taken into account that isolated and perfused lungs are not under physiological conditions since, for example, lymph flow is altered, bronchial perfusion is suppressed, autonomic innervation is disconnected and no blood cells (including inflammatory cells) are present in the perfusate. However the artificial negative pressure perfusion of isolated lung resembles respiratory conditions and the time span of experiments was limited up to 2 hours maximum to avoid excessive decrease of function. Knowing this, the lack of concomitant physiological measurement in the negative pressure ventilation is a major shortcoming in our data. The obvious reason for this is that given the use of radioactive ^192^Ir-UFP inclusion of measurement devices was allowed for radiation protection safety reasons, since they could lead to an open system and radioactive particle emissions. We have tried to compensate for this lack of know-how by measurement of biochemical indices of damage in perfusate and performing histology in lung after the experiment. No evidence for extensive lung damage in control conditions or after in-situ treatment to the essential barriers of the lung was noted. The release of ALP and ACE as biomarkers of epithelial and endothelial integrity are not elevated during the perfusion and do not correspond to the small translocation of ^192^Ir after a lag-time of about 60 min -UFP. Although microscopical analysis in lung sections showed desquamation of the epithelial layer in the lungs treated in situ with hydrogen peroxide, ALP levels in perfusate were not different from that in the control group. The experimental conditions also did not affect the integrity of the endothelial layer since perfusate levels of ACE did not change during perfusion. However, in most lungs oedema was noted in microscopy as interstitial dilation (Table [1](#T1){ref-type="table"}) at the end of the experiment. The formation of oedema is due to the imbalance between fluid transvascular filtration and clearance. Also the excess fluid causes an overhydration of the interstitium associated with accumulation of oedema fluid in the loose connective tissue \[[@B51]\]. It could be argued that oedema might affect the translocation of ^192^Ir-UFP from the lumen via the interstitium across the endothelium. In fact, a recent study using iv injection of colloidal gold particles (4 nm) in rabbits, showed that a small but significant percentage (7 %) of the UFP was taken up in endothelial and epithelial cells of the lung \[[@B50]\]. After LPS infusion, causing mild pulmonary oedema, transendothelial transport was boosted five-fold, while a significant amount of gold particles accumulated in the interstitium (14 %) and even reached the alveoli (11 %). Although this suggests a potential interference for oedema in our study, one should realize that in the above study \[[@B50]\] oedema was present at the very beginning and translocation is followed in a different direction. Nevertheless an effect of oedema on particle translocation is considered unlikely in our experimental setup since no ^192^Ir-UFP translocation was observed in the control groups and after a histamine bolus for up to 2 hours after onset of inhalation. The minute translocation of ^192^Ir-UFP in the isolated lung perfusion system conforms to our previous in vivo findings \[[@B15]\] using the same particles by inhalation at similar dose in rats. Other studies however reported very different amounts and kinetics of translocation. Nemmar et al \[[@B16]\] studied particle translocation after intratracheal instillation of uF particles in hamsters in vivoand observed a rapid (3 % within 5 min) translocation of 80 nm albumin particles coated with ^99m^Tc but no translocation was observed after the 15 minutes. On the other hand, the same group could not find latex (24--190 nm) particle translocation isolated perfused rabbit lungs at positive pressure \[[@B45]\]. Surface chemistry did not affect this process. In a similar approach using latex fluorescent beads and positive pressure ventilation, we also did not find particle translocation (data not shown). In contrast Brooking et al \[[@B48]\] showed a continuous increase in translocation with time up to 180 minutes during nasal inhalation of latex particles between 50 and 250 nm by rats. The latter study highlighted the importance of particle size as the smallest ultrafine particles showed higher uptake rates than the larger particles. In addition they demonstrated that particle surface chemistry was an important characteristic \[[@B52]\]. Oberdorster et al \[[@B18]\] reported quite extensive (\> 20 %) translocation of uF carbon particles (18 nm) after short-term inhalation. Whatever the mechanism or particle properties involved in passing the lung barriers, the question remains what particle translocation means in terms of systemic effects. The mainstream hypothesis is that lung inflammation causes and facilitates the release of mediators that adversely affect cardiovascular parameters \[[@B3]\]. Alternatively, translocation of particles to the brain (Oberdorster et al, 2004)\[[@B53]\] or the systemic circulation may also explain effects of PM exposure on heart and vascular tissue. The blood that leaves the lung first enters the heart before it is pumped to the other organs. In our previous work we showed that suspensions or filtrates of PM~10~could have direct effects on vessels \[[@B9]\]. However, the effects were rather due to the soluble components (i.e. transition metals) than to particles themselves \[[@B9]\] and are in contrast to in vivofindings on endothelial function with PM \[[@B54],[@B10]\] 2004). Although these data do not allow quantitative conclusions on the exact mechanism and the importance of systemic translocation of UFP as a mechanism in adverse effects of PM, we do confirm that ultrafine particles can translocate from the lung into the circulation using the isolated perfused rat lung upon pharmacological mediation. Permeability of the lung barrier to ultrafine particles seems to be controlled both at the epithelial and endothelial level and conditions that affect this barrier function such as inflammation may affect translocation of UFP. The conditions under which this does occur mimic conditions that are met in diseased, susceptible subjects including asthmatics and COPD-patients. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Average particle size distribution in a typical perfusion experiment over entire exposure time (120 min), revealing an average size distribution of count median diameter of16.9 nm, GSD of1.6, aerosol concentration 4.45 10^6^cm^-3^, SD 0.13 10^6^cm^-3^The example shown is from inhalation during histamine perfusion, as shown in Fig 3 (lower panel). ::: ![](1743-8977-2-3-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Translocation of ^192^Ir-UFP particles (upper panel) and ^99m^Tc-DTPA (lower panel) into perfusate of isolated perfused rat lung as a fraction of the deposited dose for inhaled particles and as fraction of the instilled dose for DTPA. Both control lungs (●) were used as well as lungs treated in-situ with H~2~O~2~bolus 0.5 mM (O), histamine (1 μM) in perfusate (□) or 0,5 mM instilled into the lungs (■). A stabilisation period of 15 minutes is done before treatment and collection of samples are taken place every 15 minutes for translocation and lung markers detection. Values in the upper panel depict the mean and SD\'s of 3 or 4 experiments; values in the lower panel depict only the mean of 3 or 4 experiments indicating the trend. ::: ![](1743-8977-2-3-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Translocation of Iridium particles in individual perfusions after treatment with (A) H~2~O~2~(bolus) and (B) histamine (10 μM) in the perfusate. Translocation represented as relative radioactivity of ^192^Ir in the perfusate. Data shown for individual heart-lung-blocs. ::: ![](1743-8977-2-3-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Release of Alkaline phosphatase (A) and Angiotensin converting enzyme (B) measured in lung perfusate during and after particle inhalation and different pre-treatments. Both control lungs (●) were used as well as lungs treated with H~2~O~2~(O), histamine (10 μM) in perfusate (□) or injected into the lungs (■). Values depict the mean of 3 or 4 experiments. ::: ![](1743-8977-2-3-6) ::: ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Examples of histopathological lesions encountered in lungs after 2 hour perfusion and inhalation of non-radioactive ^192^Ir-UFP in control lungs (A), lungs pretreated with a bolus of H2O2 in the lumen (B, D), showing septum and bronchus reduction, and (C) lung perfused with histamine, showing oedema and infiltration. ::: ![](1743-8977-2-3-7) ::: Acknowledgements ================ This study is supported by the German Ministry of Environment (BMU) and the Baden-Württemberg Environmental research program BW Plus, project number 20018. We thank dr Doris Hoehr and Erich Jermann for technical assistance during the start-up of the project, and dr Roel Schins for his valuable suggestions during the completion of the study.	PubMed Central	2024-06-05T03:55:59.552738	2005-6-27	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180470/", "journal": "Part Fibre Toxicol. 2005 Jun 27; 2:3", "authors": [ { "first": "James J", "last": "Meiring" }, { "first": "Paul JA", "last": "Borm" }, { "first": "Karim", "last": "Bagate" }, { "first": "Manuela", "last": "Semmler" }, { "first": "Jürgen", "last": "Seitz" }, { "first": "Shinji", "last": "Takenaka" }, { "first": "Wolfgang G", "last": "Kreyling" } ] }
PMC1180471	Background ========== The cytoplasmic expression of unspliced and incompletely spliced HIV-1 mRNAs encoding the HIV-1 structural proteins and enzymes is dependent upon the Rev protein \[[@B1]\]. Rev-dependent mRNAs are characterized by two types of cis-acting sequences, a single Rev response element (RRE) \[[@B2],[@B3]\] and several cis-acting repressive sequences (CRS) \[[@B4]-[@B6]\]. These sequences are removed in the completely spliced HIV-mRNAs, which therefore do not require Rev for cytoplasmic appearance and translation. The Rev protein, encoded by the completely spliced HIV-1 mRNA, is a nucleocytoplasmic shuttle protein that following nuclear import binds to and exports the RRE-containing RNAs to the cytoplasm \[[@B7],[@B8]\]. Genetic studies of the 116 residue Rev protein have defined several functional domains; including a basic domain (aa 35--50) that specifies nuclear and nucleolar localization of Rev (NLS/NOS) in addition to specific binding of Rev to RRE \[[@B3],[@B9]-[@B11]\]. An other essential domain (aa 75--84) signals active nuclear export of Rev (NES) \[[@B8],[@B12]-[@B14]\]. The Rev basic domain binds with high affinity to a site within the stem-loop IIB of the RRE and also to other sites after or upon oligomerization \[[@B15]\]. This binding of oligomeric Rev to target RNA is important for Rev function \[[@B16]\]. It is, however, not clear if Rev binds as a pre-formed complex or if oligomerization occurs after binding of the first monomer to the IIB sequence. The binding of monomeric Rev to IIB may induce conformational changes in the RRE secondary structure allowing binding of additional Rev molecules stabilized by protein-protein interactions \[[@B17]-[@B19]\]. However, Rev oligomerization has been shown to occur independently of RRE RNA both in vitro\[[@B3],[@B20]-[@B22]\] and in vivo\[[@B23]-[@B27]\]. The fact that Rev forms RNA-independent complexes indicates that complex formation may occur before binding to RNA. Although following binding of the first oligomeric Rev complex, additional complexes may bind to other low affinity sites within RRE. Interactions between the preformed complexes could then be mediated by residues different from those involved in the primary complex formation. This model could explain the apparently conflicting reports identifying different regions in oligomer formation. However, it is now generally agreed that sequences flanking the basic domain are involved in oligomer formation \[[@B3],[@B20],[@B21],[@B23],[@B25]-[@B27]\]. Of the regions reported to be essential for oligomerization, only the region N-terminal to the basic domain was found to be necessary for oligomer formation in the cytoplasm \[[@B26],[@B28]\]. One of these mutants (M4) is mutated at residues 23, 25 and 26 \[[@B29]\]. It is not clear whether the M4 mutations directly affect the residues that are involved in the oligomer formation or if the mutations cause perturbation of the structure and thus affect the ability to form oligomers \[[@B30]\]. In the current study, the M4 mutant was studied to clarifying why oligomer formation is essential for Rev activity by assessing the requirements for restoration of the activity of the mutant. Results ======= The intracellular localization of Rev and mutants ------------------------------------------------- The intracellular distribution of the M4 and the M4 derived Rev mutants (schematically outlined in figure [1](#F1){ref-type="fig"}) were tested by immunofluorescence in the absence or presence of 5 nM Leptomycin B (LMB) for 6 hours before fixation \[[@B31]\]. Wild type Rev localization was predominantly nuclear and nucleolar while the M4 mutant localized mainly to the cytoplasm with a weak nucleolar and nucleoplasmic staining (Figure [2](#F2){ref-type="fig"}, panels a and b). The addition of the three NLS from the large T-antigen enhanced nuclear import of the M4 mutant (Figure [2](#F2){ref-type="fig"}, panel c), whereas the M4-M4 dimer and the NOS-M4, which both contain two nuclear import signals, mostly localized to the cytoplasm. The nuclear staining was somewhat stronger than that of M4 (Figure [2](#F2){ref-type="fig"}, panels d and e). Treatment with LMB did not dramatically change the distribution of the wild type Rev protein (Figure [2](#F2){ref-type="fig"}, panel f). Unexpectedly, the LMB treated cells expressing the M4 mutants showed accumulation in the nucleus similarly to Rev, suggesting that the nuclear import of all the mutants occurred and that the nuclear export of the M4 mutants was mediated by an LMB-dependent pathway (Figure [2](#F2){ref-type="fig"}, panels g-j). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### A, Schematic diagram of wild type and Rev mutants. The location of the M4 mutations are indicated by arrows. The Rev basic domain is indicated as Rev-NOS, the three copies of the large T-antigen NLS are indicated as 3xNLS, the Rex overlapping NLS/NOS signal is shown as Rex-NOS. B, Schematic diagram of the reporter systems. The CAT gene and the 5\' and 3\' splice sites are indicated. The Rev and Rex responsive elements are indicated as RRE and RxRE respectively. The 6 copies of the Rev high affinity binding site IIB is indicated as 6xIIB. The Rev dependent gag mRNA with the CRS element fused to the RRE expressed in the A3.9 cells is shown below. The drawings are not to scale. ::: ![](1742-4690-2-39-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### The intracellular steady state localization of the wild type Rev and mutants is shown in the panels to the left (panels a-e). The panels to the rights show the nuclear accumulation of the wild type and mutant proteins after treatment with 5 nM LMB for 6 hours (panels f-j). The anti-Rev Mab 8E7 combined with FITC labeled anti-mouse IgG2a (Southern Biotech) was used for detection of Rev and the M4 mutants. ::: ![](1742-4690-2-39-2) ::: Testing Rev activity using reporters encoding the CAT gene within an intron --------------------------------------------------------------------------- The functional activity of the mutants was tested using the two reporter plasmids pDM138-RRE and pDM138-6xIIB (Figure [1B](#F1){ref-type="fig"}). Figure [3](#F3){ref-type="fig"} displays the results of one experiment showing the CAT expression in COS-7 cells co-tranfected with the reporter plasmids together with the M4 mutant plasmids and pcrev. The plasmid pctatwas included as a negative control. Table [1](#T1){ref-type="table"} shows the results of three or more additional and independent experiments related to the activity of wild type Rev, the negative control and to the relative amount of cell lysates in the samples. As expected, M4 displayed very low activity compared to the wild type protein (Figure [3A](#F3){ref-type="fig"} and [3B](#F3){ref-type="fig"}, Table [1](#T1){ref-type="table"}). The activity of the M4-3xNLS mutant was also low using the pDM138-RRE reporter plasmid. Some of the activity was rescued using pDM138-6xIIB but addition of 3xNLS to Rev also enhanced the relative Rev activity (Table [1](#T1){ref-type="table"}). In contrast to the M4 and M4-3xNLS mutants, M4-M4 and NOS-M4 were both active in co-transfection experiments using the Rev dependent pDM138 reporter plasmids containing RRE or six IIB high affinity binding sites (Figure [3A](#F3){ref-type="fig"} and [3B](#F3){ref-type="fig"}, Table [1](#T1){ref-type="table"}). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### The relative activity of Rev and M4 mutants ::: pDM138-RRE Rev activity as CAT produced (%) pDM138-6xIIB Rev activity as CAT produced (%) ------------------------ --------------------------------------------- ----------------------------------------------- Rev (positive control) 100 100 Rev-3xNLS 81 113 M4 7 0 M4-3xNLS 10 24 NOS-M4 67 72 M4-M4 73 61 Tat (negative control) 0 0 The numbers represent the mean value of three or more independent experiments related to the activity of wild type Rev, the negative control and the amount of cellular protein. ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Functional analysis of wild type Rev and M4 mutants by western blot analysis of COS-7 cells co-transfected with the Rev dependent reporter plasmids pDM138-RRE (A, C), pDM138-6xIIB (B, C) or the Rex dependent reporter plasmid pDM138-RxRE (C) together with the plasmids indicated above the lanes. The CAT protein was detected by polyclonal anti-CAT antibodies (Sigma) combined with HRP-labeled anti-rabbit Ig (Amersham) and developed using ECL. The lane numbers are indicated below. ::: ![](1742-4690-2-39-3) ::: There are conflicting reports of Rex\'s ability to rescue RRE RNA \[[@B32],[@B33]\]. Therefore, cells were co-transfected with the Rev dependent reporters pDM138-RRE and pDM138-6xIIB together with a vector encoding Rex. The Rex dependent reporter pDM138-RxRE was included as a positive control for Rex activity. Rex dependent CAT expression was only detected when using pDM138-RxRE containing the specific Rex responsive element (Figure [3C](#F3){ref-type="fig"}). This indicated that the Rex-NOS sequence in NOS-M4 did not bind to IIB in the co-transfection experiments using pDM138-RRE and pDM138-6xIIB. Testing Rev activity using a cell line expressing HIV-1 gagmRNA including a CRS element ----------------------------------------------------------------------------------------- Wild type Rev and the mutants were also tested by transfecting pcrevand the mutant plasmids into the stable cell line A3.9 expressing a gagmRNA fused to RRE \[[@B34]\]. No Gag p55 was detected in cells transfected with pcrevM4 and pcrevM4-3xNLS whilst Rev dependent Gag expression was observed in cells expressing Rev and the two mutants M4-M4 and NOS-M4 (Figure [4](#F4){ref-type="fig"}, right panels). However, compared to the cells transfected with pcrev, the number of Gag positive cells and the amount of Gag protein expressed in single cells was clearly less in cells transfected with pcrevM4-M4 and pcrevNOS-M4 (Figure [4](#F4){ref-type="fig"}). This trend was confirmed by western blot analysis of transfected A3.9 cells (not shown). In the A3.9 cells the cytoplasmic localization of the M4 mutants was even clearer than in the COS-7 cells (Figure [4](#F4){ref-type="fig"} left panels). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Functional analysis of wild type Rev and M4 mutants by single cell analysis of transfected methanol fixed A3.9 cells expressing Rev-dependent gagmRNA. The panels to the left show expression of Rev and the mutants detected by the anti-Rev Mab 8E7. The panels to the right show Gag expression in the same cells. Gag p55 was detected using the anti-p24 Mab. ::: ![](1742-4690-2-39-4) ::: Discussion ========== The intracellular distribution of M4 was previously found to be mainly cytoplasmic \[[@B14],[@B26]\]. Since this mutant has been shown to bind to RRE \[[@B3]\], an alternative explanation for the loss of function could be that cytoplasmic retention of M4 resulted in lack of M4 in the nucleus. The present study was conducted to test this hypothesis by using a combination of extra nuclear import signals for the M4 mutant and employing a reporter allowing binding of six Rev molecules to the RNA. The experiments using M4-3xNLS showed that neither efficient nuclear import nor binding of six monomers to intron RNA was sufficient for restoration of activity. Some of the activity of M4-3xNLS was rescued when the RRE sequence was replaced by 6xIIB allowing binding of six molecules to RNA. Control experiments with pcrev-3xNLS demonstrated that the addition of 3xNLS also enhanced the relative activity of Rev using the reporter pDM138-6xIIB (Table [1](#T1){ref-type="table"}). Thus, the extra lysine rich NLS signals may have improved the nuclear import or increased non-specific binding to the viral RNA. The M4-M4 mutant comprises two NES signals whilst NOS-M4 contains only one. Both mutants were, however, highly active in co-transfection experiments using the Rev dependent pDM138 reporter plasmids suggesting that the extra NES signal in M4-M4 is not responsible for the rescue of Rev activity (Figure [3](#F3){ref-type="fig"}, Table [1](#T1){ref-type="table"}). There was no significant difference in activity of wild type Rev or the NOS-M4 and M4-M4 mutants whether the reporter contained RRE or six IIB high affinity binding sites. This is in agreement with previous findings suggesting that the formation of oligomeric complexes on RRE is mainly dependent on protein-protein interactions and not so dependent on the RNA sequences specificity outside the IIB site \[[@B15]\]. The two mutants M4-M4 and NOS-M4 also activated Gag-expression in A3.9 cells. Thus, these mutants were also active in this essentially different reporter system. The pDM138 plasmids encode mRNAs with the CAT gene flanked by the HIV-1 splice sites. These splice sites are not present within the gag mRNA expressed in the A3.9 cell line. Although the presence of a cis-acting repressor element (CRS) requires Rev in transfor expression of the p55 Gag protein \[[@B34]\]. The common feature of NOS-M4 and M4-M4 is the presence of two functionally similar basic domains. The dimer included two copies of the Rev basic domain while NOS-M4 contained one domain from Rev and one from Rex. The function of the HTLV-I Rex protein is similar to that of Rev and the basic domains from the two viral proteins bind to Importin β during nuclear import \[[@B35],[@B36]\]. It is therefore likely that the two protein domains interact in a similar manner with other putative cellular cofactors. Previous reports have shown that Rex is able to rescue RRE containing RNA \[[@B32],[@B33]\]. However, the binding site for Rex was shown to be different from the Rev binding site IIB \[[@B37]\]. Other studies did not confirm that Rex replaces Rev in trans-activity assays \[[@B38]\]. The co-transfection experiments in this study showed that Rex was not able to induce CAT expression from the Rev-dependent pDM138 reporters indicating that the Rex NOS domain did not bind the IIB or RRE sequences in this context (Figure [3C](#F3){ref-type="fig"}). The rescue of activity of the mutants M4-M4 and NOS-M4 could be explained by at least two models. Both mutants comprise two basic domains with comparable functions. We found that Rex did not activate CAT expression from the Rev dependent pDM138 reporters. This indicated that the NOS-M4 mutant binds to the IIB RNA sequences by the Rev basic domain leaving the Rex domain free for other interactions. The previous observation that a peptide corresponding to the basic domain of Rev inhibited the in vitrosplicing of RRE containing mRNAs underscores the functional importance of this region \[[@B39],[@B40]\]. These results therefore support the suggestion that the basic domain participates in events other than binding to nuclear import factors and target RNA. This implies that at least one of the functional benefits of oligomer formation of Rev is that free basic domains can be exposed while the complex is tethered to RNA via other basic domains. Alternatively, the addition of extra sequences may have stabilized the structure of the otherwise unstable monomeric mutant, suggesting that dimer formation may be essential for obtaining a stable Rev structure. Conclusion ========== The present study showed that the activity of the negative monomeric M4 mutant was rescued by addition of an extra basic domain implying that two or more basic domains must be present within the complexes that bind to target RNA. This can be important for structural reasons or for leaving free basic domains for interaction with cellular co-factors when the Rev complex remains bound to viral RNA. Methods ======= Plasmids -------- The plasmids pcrev-3xNLS and pcrevM4-3xNLS encoding Rev and the M4 mutant with three C-terminal copies of the SV40 large T antigen NLS were constructed by amplifying the revcoding region from pcrevand pcrevM4 using the primer pair catgccatggcaggaaga agcggag / ccgctcgagttctttagttcctgactccaa \[[@B1],[@B29]\]. The PCR products were cloned into the NcoI-and XhoI-digested pCMV/myc/nuc vector (Invitrogen). The plasmid pcrevM4-M4 encodes an artificial M4 dimer with the monomers connected by a glycine rich linker. The revcoding region from pcrevM4 was amplified using the primer pairs tcgaagctagtcgacatctcctatg / cggggtaccgcctccttctttagctcc (PCR A) and cggggtaccggaatggcaggaagaagc / ctccagttggtagagagagcag (PCR B). The reverse primer in PCR B binds 70 nucleotides downstream from an XhoI site present in pcrevM4. The PCR product A was digested with SalI and KpnI while PCR B was digested with KpnI and XhoI. The two PCR products were ligated together into the SalI and XhoI digested pcrevvector. The NOS-M4 mutant contained the overlapping NLS/NOS from the HTLV-1 Rex protein fused to the N-terminal. The plasmid was constructed by inserting the NcoI-and XhoI digested fragment from pcrevM4-3xNLS into the NcoI-and XhoI digested pCMV/myc/cyto vector (Invitrogen), and inserting an NcoI flanked oligo encoding the overlapping NLS/NOS signal from the HTLV-1 Rex protein into the NcoI digested pcrevM4-myc vector. An overview of Rev and the Rev mutants is shown in figure [1A](#F1){ref-type="fig"}. The rexsequence was amplified from pcD-Srα Rex provided by Dr. Shida and cloned into the pcrevvector after excising the revgene \[[@B41],[@B42]\]. The Rev dependent pDM138-RRE and the Rex dependent pDM138-RxRE reporter plasmids contain the chloramphenicol acetyltransferase (CAT) gene and RRE or RxRE elements respectively within HIV intron sequences flanked by the 5\' and 3\' splice sites from the envgene \[[@B43],[@B44]\]. In the plasmid pDM138-6xIIB the RRE sequence was replaced by six repeats of the IIB high affinity site for Rev binding allowing binding of six monomers \[[@B40]\]. The Rev dependent reporter plasmids are schematically shown in figure [1B](#F1){ref-type="fig"}. Transfections and cell lines ---------------------------- The A3.9 cell line stably expressing the gag mRNA fused to the RRE sequence was provided by M.L. Hammarskjöld and D. Rekosh. Transfection of A3.9 cells and COS-7 cells in 35 mm wells were carried out using Lipofectamine Reagent 2000 (GIBCO BRL Life Technologies) according to the manufacturer\'s recommendations. One coverslip was added to each 35 mm well and CAT and Rev expression by western blot and immunofluorescence respectively were analysed in parallel. Each 35 mm well was transfected with 1 μg of the pDM138 reporter plasmids. The amount of revplasmid DNA varied from 0.2 -- 2 μg in order to obtain the same amount of Rev or mutant protein expressed in the cells. The Rev protein levels were estimated by immunofluorescence. Of the control plasmids pcrexand pctat, 4 μg and 1 μg respectively were added and the expression was confirmed by immunofluorescence. Immunofluorescence and western blot analysis -------------------------------------------- The cells were fixed or harvested 24 or 48 h post transfection for analysis by immunofluorescence and western blot respectively as previously described \[[@B28]\]. The immunofluorescence labeled cells were examined and the images were captured using the Leica DM RXA confocal scanning microscope with Leica PowerScan software attached. The figures were created using the program Adobe Photoshop version 3.0. The bands of the western blot analysis were scanned using an Agfa Snapscan 600 flatbed scanner and quantified using FUJIFILM LAS-1000 ProVer. 2.02 and Image Gauge V.3.45. The CAT bands were related to the activity of Rev set to 100 %, the negative control set to 0 % and the intensity of cellular bands representing the amount of cell lysate in the sample. Calculations were not performed when cellular background bands were not visible as in the experiments shown in figure [3](#F3){ref-type="fig"}. ### Antibodies The Rev proteins were detected by the anti-Rev Mab 8E7 \[[@B45]\]. Tat and Rex were detected using the Mabs 1D9 and 1F8 respectively (not shown) \[[@B42],[@B46]\]. The anti-Gag p24 Mab for detection of full length Gag p55 expressed in A3.9 cells was supplied by H.C. Holmes, Medical Research Council, London, UK \[[@B47]\]. The polyclonal rabbit antibody for detection of the CAT protein was from Sigma. List of abbreviations used ========================== RRE: Rev Responsive Element HIV: Human Immunodeficiency Virus CRS: cis-acting repressor sequences Mab: Monoclonal antibody CAT: Chloramphenicolacetyltransferase NOS: Nucleolar localization signal NLS: Nuclear localization signal NES: Nuclear export signal LMB: Leptomycin B Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= CF: Vector design and construction. Transfections and immuoassays. TA: Vector design and construction. Transfections and immuoassays. PA: Vector design and construction. JK: Experimental design, manuscript preparation AMSz: Experimental design. Transfections and immuoassays. Manuscript preparation. Acknowledgements ================ We thank B. Cullen, M. Malim, T. Hope, H. Shida, H. Holmes, M.L. Hammarskjöld, David Rekosh and B. Wolff for reagents, Rebecca Cox Brokstad for reading of the manuscript and Siri Brønlund for technical assistance. The study was supported by the University of Bergen and grants from The Faculty for Mathematics and Natural Sciences, University of Bergen.	PubMed Central	2024-06-05T03:55:59.556515	2005-6-10	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180471/", "journal": "Retrovirology. 2005 Jun 10; 2:39", "authors": [ { "first": "Clemens", "last": "Furnes" }, { "first": "Thomas", "last": "Arnesen" }, { "first": "Peter", "last": "Askjaer" }, { "first": "Jørgen", "last": "Kjems" }, { "first": "Anne Marie", "last": "Szilvay" } ] }
PMC1180472	Background ========== Bronchial asthma is a chronic disorder characterized as reversible airway obstruction, eosinophilic airway inflammation, mucus hypersecretion, and airway hyperresponsiveness (AHR) \[[@B1]\]. In the process of airway inflammation, various types of cells, such as eosinophils, mast cells, T lymphocytes, and dendritic cells are involved \[[@B2],[@B3]\]. AHR to nonspecific stimuli is a hallmark of asthma. However, the precise mechanism to induce AHR has not been fully elucidated. Persistence of eosinophilic airway inflammation is closely linked to induction of AHR \[[@B1],[@B4],[@B5]\]. However, the dissociation of AHR and eosinophilic airway inflammation often occurs \[[@B6]-[@B12]\]. For example, Leckie et al. showed that administration of neutralizing antibody (Ab) to interleukin (IL) -5 does not suppress AHR despite this treatment abrogating eosinophilia in blood and sputum \[[@B6]\]. In addition, even a protective role of eosinophils on the AHR induction has been recently proposed in a mouse model \[[@B9]\]. These findings suggest that other mechanism(s) than eosinophilic inflammation would be involved in inducing AHR. On the other hand, in patients with asthma, activated T cells, especially CD4^+^T helper (Th) 2 cells, also infiltrate into the airway, which is associated with disease severity \[[@B13]-[@B15]\]. In a mouse model, administration of blocking Ab to CD4 suppresses AHR and airway inflammation \[[@B11],[@B16]\]. Moreover, transfer of CD4^+^Th2 cells into naive mice and subsequent antigen-inhalation induce AHR and airway inflammation \[[@B17],[@B18]\]. These findings suggest that T cells, especially CD4^+^Th2 cells, are also important for the AHR induction. However, in these studies, AHR induced by CD4^+^Th2 cells accompanies eosinophilic airway inflammation. Therefore, it remains unclear whether T cells alone could directly induce AHR. We previously reported that systemic antigen sensitization alone directly induces AHR before development of eosinophilic inflammation in mice \[[@B19]\]. This raised a possibility that systemic immune response to antigen itself could directly induce AHR. The purpose of the present study was to investigate which component in the immunocompetent cells could directly induce AHR. In this study, we found that passive cell transfer of spleen cells obtained after antigen-sensitization reconstituted AHR in naive mice. Then, using this system, we studied the cell source essential for the AHR induction, and confirmed that antigen-sensitized CD4^+^CD62L^low^memory/effector Th2 cells would play an essential role for induction of basal AHR. Methods ======= Immunization of mice and transfer of spleen cells ------------------------------------------------- Mice were immunized as reported previously \[[@B19],[@B20]\]. Seven-week-old male BALB/cAnNCrj mice (Charles River Japan, Kanagawa, Japan) or IL-4 gene-deleted mice (BALB/c-IL4^tm2Nnt^; Jackson Laboratory, Bar Harbor, ME) were sensitized with an i.p. injection of 2 μg ovalbumin (OVA; Sigma, St. Louis, MO), bovine serum albumin (BSA; Wako, Osaka, Japan), or keyhole limpet hemocyanin (KLH; Calbiochem, LaJolla, CA) plus 2 mg aluminum hydroxide (alum) on days 0 and 11. Control mice received an injection of physiologic saline (SA) without alum on days 0 and 11. Some control mice received an injection of SA plus alum. In some experiments, we used complete Freund\'s adjuvant (CFA; Difco Laboratories, Detroit, MI) instead of alum as an adjuvant. On day 18, cell suspensions of spleens were obtained by pressing the tissues through a 70-μm nylon filters. Spleen cells prepared from the OVA-sensitized mice (1 × 10^6^, 5 × 10^6^, 2 × 10^7^, or 5 × 10^7^in 0.5 ml HBSS, respectively) or the SA-treated mice (5 × 10^7^) were transferred into syngenic recipients by intravenous injection. In some experiments, mice were sensitized with OVA on days 0 and 11, and then challenged with 3% OVA for 10 minutes from day 18 to day 20. On day 21, lungs were excised to observe eosinophilic airway inflammation. All animal experiments were approved by the Animal Research Ethics Board of the Department of Allergy and Rheumatology, University of Tokyo. Measurement of airway responsiveness (AR) ----------------------------------------- On day 22 (4 days after the transfer), AR to methacholine (Mch) was measured with the enhanced pause (Penh) system (Buxco, Troy, NY) as described previously \[[@B19],[@B20]\]. In some experiments, AR was assessed by measurement of airway resistance (Raw) \[[@B21],[@B22]\]. Briefly, anesthetized mice were tracheostomized and connected to a MiniVent ventilator (Hugo Sachs Elektronik, March, Germany), then ventilated with a tidal volume of 250 μl and a respiratory frequency of 120 breaths/minute. The mice were placed inside whole-body plethysmographs (Buxco) to measure Raw. Increasing doses of Mch were administered by ultraneblization for 3 minutes. The concentration of Mch that induced a 100% increase in Penh or Raw was expressed as PC~200~Mch (μg/ml) or PC~200~Mch Raw (μg/ml) as an indicator of AHR. Bronchoalveolar lavage (BAL) and histological examination --------------------------------------------------------- Bronchoalveolar lavage fluid (BALF) analyses were performed as described previously \[[@B19],[@B20]\]. The lungs were lavaged four times with SA (0.5 ml each), and approximately 1.6 ml was consistently recovered with gentle handling. The cell suspension was centrifuged at 1,500 rpm for 10 minutes at 4°C. The cells were resuspended into 1 ml of saline with 1% BSA, and the total cell numbers were counted with a hemocytometer. Cytospin samples were prepared by centrifuging the suspensions (200 μl) at 300 rpm for 10 minutes. To clearly distinguish eosinophils from the neutrophils, three different stains were applied: Diff-Quick stain, May-Grünwald-Giemsa stain, and Eosino (Hansel) stain \[[@B19]\]. On the basis of the findings with these stainings, cell differentials were counted with at least 300 leukocytes in each sample. Lung histological examinations were performed as described previously \[[@B19],[@B20]\]. Serum immunoglobulin (Ig) E and BALF cytokine concentrations were measured using ELISA kits (Pharmingen, SanDiego, CA) according to the manufacturer\'s instructions. The lower limits of sensitivity for the ELISA were 78 ng/ml (IgE), 7.8 pg/ml (IL-4), 15.6 pg/ml (IL-5), and 31.2 pg/ml (interferon (IFN)-γ), respectively. Fluorescence study and immunohistochemistry ------------------------------------------- On day 18, spleen cells (5 × 10^7^) from OVA-sensitized mice were labeled with fluorescent dye (PKH67; Sigma), and then transferred into syngenic recipients. In another experiment, CD4^+^CD62L^low^cells (4 × 10^6^) from OVA-sensitized mice were positively selected as described in \"depletion and positive selection study\", and then labeled and transferred. On day 19, after perfusion with saline, lungs were excised. Five-micrometer sections were cut and observed by fluorescence microscopy (BX51; Olympus, Melville, NY). Immunohistochemistry was performed using Vectastain ABC kits (Vector Laboratories, Burlingame, CA) as described previously \[[@B23]\]. T cells were detected by staining for CD3 (cytoplasm, blue). Proliferation was assessed by staining for proliferating cell nuclear antigen (PCNA; nucleus, brown). Double-staining analysis of a single section was performed. Briefly, the tissue was deparaffinized and rehydrated with decreasing concentrations of ethyl alcohol. The slides were boiled in 0.05 M citric acid for 7 minutes. After cooling down to room temperature, endogenous peroxidase activity was blocked by incubating the slides in 3% H~2~O~2~in methanol for 60 minutes. Next the slides were treated with blocking solution containing 5% normal rabbit serum, 2% casein, and 3% BSA for 45 minutes. Anti-CD3 Ab (5 μg/ml; Santa Cruz Biotechnology, Santa Cruz, CA) was applied to the tissue and incubated at 37°C for 30 minutes. After washing with PBS, biotinylated rabbit anti-goat IgG Ab was applied and incubated at 37°C for 30 minutes. After washing, avidin-biotin alkaline phosphatase complex was applied and incubated at 37°C for 30 minutes, followed by the addition of substrate solution. Color development was stopped by rinsing the slides in distilled water. Then, the slides were treated with blocking solution containing 5% normal goat serum, 2% casein, and 3% BSA for 45 minutes. Anti-PCNA Ab (2 μg/ml; Santa Cruz Biotechnology) was applied to the tissue and incubated at 37°C for 30 minutes. After washing with PBS, biotinylated goat anti-mouse IgG Ab was applied and incubated at 37°C for 30 minutes. After washing, avidin-biotin peroxidase complex was applied and incubated at 37°C for 30 minutes, followed by the addition of diaminobenzidine solution. Color development was stopped by rinsing the slides in distilled water. The slides were counterstained with neutral red. Positively immunostained cells were enumerated directly in 20 random high power fields (hpf; 40× objective). Depletion and positive selection study -------------------------------------- For depletion, on day 18, spleen cells from OVA-sensitized mice were incubated with biotinylated anti-CD4 monoclonal antibody (mAb; RM4-5; Pharmingen), anti-CD8 mAb (53-6.7; Pharmingen), anti-CD11b mAb (M1/70; Pharmingen), anti-CD11c mAb (HL3; Pharmingen), or anti-CD19 mAb (1D3; Pharmingen), and then incubated with streptavidin-microbeads (Miltenyi Biotech, Bergisch Gladbach, Germany). For depletion of invariant Vα14 (iVα14) natural killer T (NKT) cells, spleen cells from OVA-sensitized mice were incubated with α-galactosylceramide (α-GalCer; kindly provided by the Pharmaceutical Research Laboratory of Kirin Brewery Company, Gunma, Japan)-loaded mouse CD1d: Ig dimeric protein (Pharmingen) and then incubated with anti-mouse IgG1-microbeads (Miltenyi Biotech). Bead-bound cells were depleted using magnetic separation columns. Flow cytometry confirmed that greater than 98% of CD4^+^, CD8^+^, CD11b^+^, CD11c^+^, or CD19^+^cells were removed from splenocytes, and 88% of cells that bound α-GalCer-loaded mouse CD1d dimer were removed (data not shown). Syngenic recipients received CD4^+^, CD8^+^, CD11b^+^, CD11c^+^, CD19^+^, or iVα14 NKT cell-depleted splenocytes (5 × 10^7^each). For positive selection, on day 18, spleen cells from OVA-sensitized mice were incubated with anti-CD4 mAb-coated or anti-CD11c mAb-coated microbeads (Miltenyi Biotech). Bead-bound cells were then isolated using magnetic separation columns. The purities of the enriched CD4^+^and CD11c^+^cells were 95% and 85%, respectively (data not shown). Over 95% of the CD4^+^cells were CD3^+^T cells (data not shown). Syngenic recipients received CD4^+^(1.25 × 10^7^) or CD11c^+^(1 × 10^6^) cells. We prepared a CD4^+^CD62L^low^subset or a CD4^+^CD62L^high^subset using an anti-FITC multisort kit (Miltenyi Biotech), FITC anti-CD4 mAb (RM4-5; Pharmingen), and anti-CD62L mAb-coated microbeads (Miltenyi Biotech). The purities of each subset were 80% (data not shown). Syngenic recipients received CD4^+^CD62L^low^(4 × 10^6^) or CD4^+^CD62L^high^(8.5 × 10^6^) cells. AR was measured on day 22 (4 days after the transfer). In vitroOVA stimulation and polarization to Th1 or Th2 phenotype ------------------------------------------------------------------ On day 18, spleen cells (5 × 10^6^cells/ml) from OVA-sensitized mice were incubated with OVA (200 μg/ml) for 4 days in vitro. On day 22, syngenic recipients received these stimulated cells (1 × 10^6^). In some experiments, dead cells were removed from cultured splenocytes using Percoll (Pharmacia Biotech, Uppsala, Sweden) gradient centrifugation. For polarization to Th1 cells, recombinant IL-12 (10 ng/ml; Genzyme Techne, Minneapolis, MN) and anti-IL-4 Ab (0.1 μg/ml; Genzyme Techne) were added to the culture medium. For polarization to Th2 cells, recombinant IL-4 (100 ng/ml; Genzyme Techne) and anti-IL-12 Ab (0.25 μg/ml; Genzyme Techne) were added. Polarization was confirmed by measuring IL-4 and IFN-γ in supernatant using ELISA. On day 22, CD4^+^T cells were positively selected, and syngenic recipients received CD4^+^Th1 cells or CD4^+^Th2 cells (5 × 10^5^each). AR was measured on day 26 (4 days after the transfer). In vitroproliferation and cytokine assays ------------------------------------------- Positively selected CD4^+^T cells, CD4^+^CD62L^high^T cells, and CD4^+^CD62L^low^T cells (2.5 × 10^5^cells/well, respectively) from OVA-sensitized mice were cultured with freshly isolated mitomycin C (Sigma)-treated splenocytes (2.5 × 10^5^cells/well) in the presence or absence of OVA. After 48 hours, the proliferation was assessed by a cell proliferation ELISA bromodeoxyuridine (BrdU) kit (Roche Applied Science, Mannheim, Germany). After 72 hours, cytokine concentrations in the supernatants were measured using ELISA. Statistics ---------- Values are expressed as means ± SEM. Statistical analysis was performed by one-way ANOVA followed by Fisher\'s least significant difference test or Student\'s ttest. A p value \< 0.05 was considered significant. Results ======= Passive cell transfer of spleen cells from antigen-sensitized mice induces AHR ------------------------------------------------------------------------------ As reported previously \[[@B19]\], immunization with OVA alone induced a significant increase in AR (OVA ip; PC~200~Mch; 3,870 ± 518 μg/ml) as compared with saline injection (SAip; 5,725 ± 1,009 μg/ml; Figure [1A](#F1){ref-type="fig"}). Injection with alum alone provoked a slight non-specific increase in AR, but it was not significant (data not shown). When OVA-sensitized mice received OVA inhalation challenges, then prominent infiltration of eosinophils was provoked and AR further increased (OVA/OVA; 2,564 ± 343 μg/ml; Figure [1A](#F1){ref-type="fig"}). In the group of mice that received spleen cells from OVA-sensitized mice (defined as \"TROVA-mice\"; 5 × 10^7^), PC~200~Mch was 4,191 ± 203 μg/ml, which was significantly lower than that of the group of mice that received the same number of spleen cells from SA-treated mice (defined as \"TRSA-mice\"; 5 × 10^7^; 6,357 ± 835 μg/ml; Figure [1A](#F1){ref-type="fig"}). Transfer of spleen cells from mice that were injected with SA plus alum did not induce AHR (TRAlum; 6,596 ± 697 μg/ml). Therefore, transfer of OVA-sensitized spleen cells reconstituted moderate AHR in a naive mouse to a similar degree of AHR induced by systemic sensitization with OVA antigen. In the TROVA-mice, AR increased in a cell-number dependent-manner (Figure [1B](#F1){ref-type="fig"}; 7,415 ± 2,176 μg/ml (1 × 10^6^), 6,343 ± 1,392 μg/ml (5 × 10^6^), 4,803 ± 572 μg/ml (2 × 10^7^), respectively). To confirm the reliability of the data obtained from the Penh system, we examined AR by measuring Raw with ventilated mice treated with the same immunization protocol. Similar results on AHR were obtained by measuring Raw (Figure [1C](#F1){ref-type="fig"}; PC~200~Mch Raw; SAip, 47,205 ± 4,767 μg/ml, OVA ip, 20,668 ± 1,562 μg/ml, TRSA, 46,702 ± 6,653 μg/ml, TROVA, 22,450 ± 9,535 μg/ml). So we used Penh system for the following experiments. In a time course study, the TROVA-mice revealed a significant increase in AR from 4 to 10 days after the transfer (Figure [2](#F2){ref-type="fig"}). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Passive cell transfer of spleen cells from OVA-sensitized mice induces airway hyperresponsiveness (AHR). (A) Transfer of spleen cells from OVA-sensitized mice induces a moderate AHR. Mice were sensitized with OVA or SA on days 0 and 11. On day 18, recipients received the spleen cells (5 × 10^7^) from SA (without alum)-treated mice (TRSA), OVA-sensitized mice (TROVA), or SA (with alum)-treated mice (TRAlum). Airway responsiveness (AR) to methacholine (Mch) was measured with Penh methods on day 22 (4 days after the transfer) as described in Methods. Some OVA-sensitized mice were inhaled with OVA from day 18 to day 20. AR was measured on day 18 in mice received that i.p SA injection (SAip) or OVA injection only (OVAip), or on day 21 in mice that received OVA-sensitization and -inhalation (OVA/OVA). Values are presented as means ± SEM for 5 to 14 mice per group. \* p \< 0.05 compared with PC~200~Mch of TRSA (5 × 10^7^). \# p \< 0.05 compared with PC~200~Mch of SAip. (B) AR to Mch increases in a transferred-cell-number-dependent manner (n = 5--14 per group). Recipients received spleen cells from SA-treated mice (TRSA; 5 × 10^7^) or OVA-sensitized mice (TROVA; 1 × 10^6^, 5 × 10^6^, 2 × 10^7^, or 5 × 10^7^). AR was measured 4 days after the transfer. \* p \< 0.05 compared with PC~200~Mch of TRSA (5 × 10^7^). (C) AR was assessed by measurement of airway resistance (Raw) (n = 6--12 per group). \* p \< 0.05 compared with PC~200~Mch Raw of TRSA (5 × 10^7^). \# p \< 0.05 compared with PC~200~Mch Raw of SAip. ::: ![](1465-9921-6-46-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Change in AHR following cell transfer. Recipients received spleen cells (5 × 10^7^) from SA-treated mice (TRSA) or OVA-sensitized mice (TROVA). AR was measured at the indicated time after the transfer (n = 8 per group). \#\# p \< 0.01 compared with the baseline value (before the transfer) (ANOVA). \* p \< 0.05 compared with PC~200~Mch at the same time point of TRSA. ::: ![](1465-9921-6-46-2) ::: Antigens that elicit Th2-type, but not Th1-type, immune response can induce AHR by cell transfer ------------------------------------------------------------------------------------------------ Next we confirmed other antigens than OVA could also induce AHR. Transfer of spleen cells from BSA-sensitized mice or KLH-sensitized mice also induced a significant AHR (Figure [3A](#F3){ref-type="fig"}; TRSA, 5,814 ± 638 μg/ml, TROVA, 4,112 ± 147 μg/ml, TRAlum, 6,224 ± 680 μg/ml, TRBSA, 4,633 ± 279 μg/ml, TRKLH, 4,123 ± 280 μg/ml). In another experiment, we confirmed systemic sensitization alone with BSA or KLH induced AHR without eosinophilic inflammation (data not shown). We also confirmed that systemic sensitization with BSA or KLH increased serum IgE concentration (data not shown). When BSA or KLH sensitized mice received airway antigen challenge, eosinophilic airway inflammation was provoked (data not shown). On the other hand, use of CFA instead of alum as an adjuvant, which is known to elicit Th1-type immunity \[[@B24]\], did not induce transfer-mediated AHR (Figure [3B](#F3){ref-type="fig"}; TRSA, 5,814 ± 638 μg/ml, TROVA/Alum, 4,112 ± 147 μg/ml, TROVA/CFA, 5,224 ± 74 μg/ml, TRCFA, 5,708 ± 945 μg/ml). Therefore, it could be generally considered that antigens that elicit Th2-type immune response could induce a significant AHR by cell transfer. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Antigens that elicit Th2 type response, but not Th1 type response, induce transfer-mediated AHR. (A) Antigens that elicit Th2 type response induce transfer-mediated AHR. Recipients received spleen cells (5 × 10^7^) from SA (without alum)-treated mice (TRSA), from OVA-sensitized mice (TROVA), from SA (with alum)-treated mice (TRAlum), from BSA-sensitized mice (TRBSA), or from KLH-sensitized mice (TRKLH). AR was measured 4 days after the transfer (n = 4--9 per group). \* p \< 0.05 compared with PC~200~Mch of TRSA. (B) CFA, an adjuvant that elicits Th1 type response, does not induce transfer-mediated AHR. Recipients received spleen cells (5 × 10^7^) from SA-treated mice (TRSA), from OVA (with alum)-sensitized mice (TROVA/Alum), from OVA (with CFA)-sensitized mice (TROVA/CFA), or from CFA-treated mice (TRCFA). AR was measured 4 days after the transfer (n = 4--9 per group). \* p \< 0.05 compared with PC~200~Mch of TRSA. ::: ![](1465-9921-6-46-3) ::: The transfer-mediated AHR is provoked in an eosinophil-independent manner ------------------------------------------------------------------------- In the TROVA-mice, the number of total cell, macrophage, and lymphocyte in BALF slightly increased, whereas eosinophil number did not (Table [1](#T1){ref-type="table"}). In the TROVA-mice, a slight infiltration of inflammatory cells into the peribronchial area was detected in some specimens. However, prominent infiltration of eosinophils was not detected (Figure [4A](#F4){ref-type="fig"}, left). In contrast, in the OVA-sensitized and challenged mice (OVA/OVA), prominent infiltration of eosinophils into the peribronchial interstitial area or bronchial wall was observed (Figure [4A](#F4){ref-type="fig"}, right). These results indicated that cell transfer induced AHR without prominent infiltration of eosinophils in the lung. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### BALF findings ::: Total cells(×10^2^) Macrophage(×10^2^) Lymphocyte(×10^2^) Neutrophil(×10^2^) Eosinophil(×10^2^) ----------------- --------------------- -------------------- -------------------- -------------------- -------------------- TRSA 5 × 10^6^ 919 ± 115 893 ± 111 25 ± 6 1 ± 1 0 ± 0 TROVA 1 × 10^6^ 980 ± 149 945 ± 140 33 ± 7 3 ± 3 0 ± 0 TROVA 5 × 10^6^ 1070 ± 132 1030 ± 132 34 ± 4 6 ± 4 0 ± 0 TROVA 2 × 10^7^ 950 ± 78 898 ± 77 44 ± 7 6 ± 2 3 ± 1 TROVA 5 × 10^7^ 1004 ± 105 942 ± 98 55 ± 11 6 ± 2 1 ± 1 Values are presented as means ± SEM for 5 to 14 mice per group. ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Histologic findings. (A) H&E stain. Lung sections from mice that received spleen cells from OVA-sensitized mice (TROVA) and from mice that received OVA-sensitization and aerosol OVA-challenge (OVA/OVA) are shown. Scale bar, 200 μm. (B) Fluorescence study. Spleen cells from OVA-sensitized mice were labeled with fluorescent dye (PKH67), and then transferred into recipients. Five-micrometer sections of the lungs were observed by fluorescence microscopy 24 hours after the transfer (TROVA). A lung section from naive mice without transfer is shown (naive). Scale bar, 100 μm. (C) Double staining analysis of a single section by immunohistochemistry. T cells were detected by staining for CD3 (cytoplasm, blue). Proliferation was assessed by staining for PCNA (nucleus, brown). Lung sections from mice that received spleen cells from OVA-sensitized mice (TROVA) and from mice that received spleen cells from SA-treated mice (TRSA) are shown. Proliferating T cells were clearly detected (black arrow). Scale bar, 40 μm. (D) Histologic findings of colon (H&E stain). Colon sections from mice that received spleen cells from OVA-sensitized mice (TROVA) and from SA-treated mice (TRSA) are shown. Scale bar, 200 μm. (E)Transferred cells recruit into the colon. Spleen cells from OVA-sensitized mice were labeled, and transferred. Five-micrometer sections of the colons were observed by fluorescence microscopy 24 hours after the transfer (TROVA). A colon section from naive mice without transfer is shown (naive). Scale bar, 100 μm. ::: ![](1465-9921-6-46-4) ::: Some transferred cells recruit into the lung, and some T cells proliferate without further airway antigen challenge ------------------------------------------------------------------------------------------------------------------- We next analyzed lung sections from mice that received fluorescently labeled spleen cells from OVA-sensitized mice. The transferred cells were clearly detected in the lung 24 hours after the transfer, particularly in lung interstitial areas (Figure [4B](#F4){ref-type="fig"}, left). In contrast, the mice that had not received cell transfer did not show this finding (Figure [4B](#F4){ref-type="fig"}, right). Similar results were observed 3 days after the transfer (data not shown). Immunohistochemistry revealed that some T cells in the lung proliferated without further airway antigen challenge in the TROVA-mice (1.6 ± 0.2/hpf; Figure [4C](#F4){ref-type="fig"}, left). In contrast, in the TRSA-mice, proliferation of T cells was less detected (0.5 ± 0.1/hpf; Figure [4C](#F4){ref-type="fig"}, right). Transferred cells also distribute in other tissues and induce mild inflammation ------------------------------------------------------------------------------- In the TROVA-mice, a slight infiltration of inflammatory cells was also detected in the mucosa of colon in some specimens (Figure [4D](#F4){ref-type="fig"}, left). In contrast, it was less detected in the TRSA-mice (Figure [4D](#F4){ref-type="fig"}, right). Similar results were obtained in tissues other than colon such as stomach and small intestine (data not shown). Fluorescence study revealed that transferred cells also distributed among other tissues such as colon (Figure [4E](#F4){ref-type="fig"}), stomach, liver, and spleen (data not shown). Therefore, transferred cells did not recruit specifically into the lung, but distributed throughout the body. Th2 cell-type cytokines, but not IgE, mediate transfer-induced AHR ------------------------------------------------------------------ In the TROVA-mice (5 × 10^7^), the concentrations of IL-4 and IL-5 were significantly higher than those of the TRSA-mice (5 × 10^7^) (Figure [5A](#F5){ref-type="fig"}). The concentrations of IL-13 (data not shown) and IFN-γ levels (Figure [5A](#F5){ref-type="fig"}) also slightly increased in the TROVA-mice (5 × 10^7^), but these values were not significantly different from those of the TRSA-mice (5 × 10^7^). Thus, Th2 cell-type cytokines increased in BALF, and their increases might play a pivotal role in the transfer-mediated AHR. We also measured serum IgE concentration. No significant increase in IgE was detected (Figure [5B](#F5){ref-type="fig"}), suggesting that IgE did not mediate transfer-induced AHR. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Concentrations of BALF cytokines and serum IgE. (A) BALF cytokine concentrations. Four days after the transfer, BAL was performed and then the centrifuged supernatant was assayed for IL-4, IL-5, and IFN-γ concentrations by ELISA, respectively (n = 5--14 per group). \* p \< 0.05 compared with the values of TRSA (5 × 10^7^). (B) Serum IgE concentrations (n = 5--14 per group). \\\* p \< 0.001 compared with the values of SAip. ::: ![](1465-9921-6-46-5) ::: IL-4 plays an important role in transfer-mediated AHR ----------------------------------------------------- As reported previously \[[@B19]\], IL-4 played a pivotal role in AHR that induced by antigen sensitization alone. So, we next examined the role of IL-4 in this transfer-mediated AHR. Transfer of spleen cells from OVA-sensitized, IL-4-deficient mice failed to induce the development of AHR (Figure [6](#F6){ref-type="fig"}; TRSA, 6,554 ± 758 μg/ml, TROVA, 4,209 ± 287 μg/ml, TRSA/IL-4^-/-^, 6,723 ± 765 μg/ml, TROVA/IL-4^-/-^, 6,593 ± 698 μg/ml) and an increase in BALF IL-4 concentration was not detected (data not shown). These results suggested that IL-4 production by OVA-sensitized spleen cells played an important role in the induction of transfer-mediated AHR. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### IL-4 plays an important role in transfer-mediated AHR. Recipients received spleen cells (5 × 10^7^) from SA-treated wild-type mice (TRSA), from OVA-sensitized wild-type mice (TROVA), from SA-treated IL-4-deficient mice (TRSA/IL-4^-/-^), or from OVA-sensitized IL-4-deficient mice (TROVA/IL-4^-/-^). AR was measured 4 days after the transfer (n = 6--8 per group). \* p \< 0.05 compared with PC~200~Mch of TRSA. \# p \< 0.05 compared with PC~200~Mch of TROVA. ::: ![](1465-9921-6-46-6) ::: In vitroOVA stimulation potentiates the intensity of transfer-mediated AHR ---------------------------------------------------------------------------- In another experiment, OVA-sensitized spleen cells were stimulated with OVA in vitroand then transferred. This treatment increased the intensity of transfer-mediated AHR (Figure [7A](#F7){ref-type="fig"}; TRSA, 6,556 ± 703 μg/ml, TROVA (1 × 10^6^), 6,848 ± 997 μg/ml, TROVA (5 × 10^7^), 4,607 ± 205 μg/ml, stimulated cell-transferred mice (TRSTIM) (1 × 10^6^), 3,654 ± 459 μg/ml). BALF cytokine concentrations of the recipients were also increased by this treatment (Figure [7B](#F7){ref-type="fig"}). This result indicated that stronger antigen stimulus induced stronger immune response, which resulted in the induction of higher AHR. ::: {#F7 .fig} Figure 7 ::: {.caption} ###### **In vitroOVA stimulation increases the intensity of transfer-mediated AHR*. (A) Effect of in vitroOVA stimulation on transfer-mediated AHR. On day 18, spleen cells (5 × 10^6^cells/ml) from OVA-sensitized mice were incubated with OVA (200 μg/ml) for 4 days in vitro. On day 22, recipients received these stimulated cells by intravenous injection (TRSTIM: 1 × 10^6^). AR was measured 4 days after the transfer (n = 5--10 per group). \ p \< 0.05 compared with PC~200~Mch of TRSA (5 × 10^7^). \# p \< 0.05 compared with PC~200~Mch of TROVA (1 × 10^6^). (B) BALF cytokine concentrations. Four days after the transfer, BAL was performed and then the centrifuged supernatant was assayed for IL-4, IL-5, and IFN-γ concentrations using ELISA, respectively (n = 5--10 per group). \* p \< 0.05 compared with the values of TRSA (5 × 10^7^). \# p \< 0.05 compared with the values of TROVA (1 × 10^6^). ::: ![](1465-9921-6-46-7) ::: CD4^+^Th2 cells induce AHR -------------------------- To determine which cells are most important for this AHR induction, we carried out a cell depletion study. Transfer of CD4^+^cell-depleted splenocytes into naive mice failed to induce the development of AHR (Figure [8A](#F8){ref-type="fig"}; TRSA, 6,240 ± 577 μg/ml, TROVA, 3,858 ± 325 μg/ml, CD4 (-), 5,695 ± 543 μg/ml, CD8 (-), 3,738 ± 368 μg/ml, CD11b (-), 3,528 ± 327 μg/ml, CD11c (-), 4,077 ± 206 μg/ml, CD19 (-), 3,694 ± 434 μg/ml, iVα14 NKT (-), 3,497 ± 345 μg/ml) or failed to increase BALF cytokine concentrations (data not shown). We next carried out positive selection. We determined the numbers of CD4^+^and CD11c^+^spleen cells to be transferred based on the physiologic ratio of these cells (CD4^+^spleen cells were 25% and CD11c^+^spleen cells were 2% of total spleen cells, respectively). Transfer of positively selected CD4^+^T cells into naive mice induced AHR (Figure [8B](#F8){ref-type="fig"}; TRSA, 7,061 ± 831 μg/ml, TROVA, 4,381 ± 102 μg/ml, CD4, 4,526 ± 560 μg/ml, CD11c, 5,637 ± 1,040 μg/ml) and an elevation of BALF cytokine levels (data not shown), which were consistent with the results of depletion study. Next, we elucidated which of the two CD4-mediated response play a major role for AHR induction. Spleen cells were polarized to either Th1 or Th2 phenotype (Figure [8C](#F8){ref-type="fig"}) and each CD4^+^population was transferred. Transfer of CD4^+^Th2, but not Th1, cells induced AHR (Figure [8D](#F8){ref-type="fig"}; Th1, 5,732 ± 508 μg/ml, Th2, 4,384 ± 151 μg/ml). ::: {#F8 .fig} Figure 8 ::: {.caption} ###### CD4^+^Th2 cells directly induce AHR. (A) Depletion study. Recipients received spleen cells (5 × 10^7^) from SA-treated mice (TRSA) or OVA-sensitized mice (TROVA). Other recipients received CD4^+^cell-depleted (CD4(-)), CD8^+^cell-depleted (CD8(-)), CD11b^+^cell-depleted (CD11b(-)), CD11c^+^cell-depleted (CD11c(-)), CD19^+^cell-depleted (CD19(-)), or iVα14 NKT cell-depleted (iVα14 NKT(-)) spleen cells from OVA-sensitized mice (5 × 10^7^each), respectively. AR was measured 4 days after the transfer (n = 6--10 per group). \* p \< 0.05 compared with PC~200~Mch of TROVA. (B) Effect of CD4^+^or CD11c^+^cells on transfer-mediated AHR. Recipients received unfractionated spleen cells from SA-treated mice (TRSA; 5 × 10^7^), or unfractionated (TROVA; 5 × 10^7^), CD4^+^(CD4; 1.25 × 10^7^), or CD11c^+^(CD11c; 1 × 10^6^) spleen cells from OVA-sensitized mice. AR was measured 4 days after the transfer (n = 5--8 per group). \* p \< 0.05 compared with PC~200~Mch of TRSA. (C) Polarization to Th1 or Th2 phenotype. On day 18, spleen cells (5 × 10^6^cells/ml) from OVA-sensitized mice were incubated with OVA (200 μg/ml) for 4 days in vitro. For polarization toward Th1 cells, IL-12 and anti-IL-4 Ab were added to the culture medium. For polarization toward Th2 cells, IL-4 and anti-IL-12 Ab were added. On day 22, positively selected CD4^+^T cells were cultured with freshly isolated mitomycin C-treated splenocytes and OVA (200 μg/ml). After 96 hours, IL-4 and IFN-γ concentrations in the supernatants were assayed. \\\* p \< 0.001 compared with the value of the CD4^+^Th1 cells. \#\#\# p \< 0.001 compared with the value of the CD4^+^Th2 cells. (D) Effect of Th1 or Th2 phenotype on transfer-mediated AHR. Recipients received CD4^+^Th1 or Th2 cells (5 × 10^5^). AR was measured 4 days after the transfer (n = 5--6 per group). \* p \< 0.05 compared with PC~200~Mch of mice that received CD4^+^Th1 cells. ::: ![](1465-9921-6-46-8) ::: CD4^+^CD62L^low^memory/effector T cells play an essential role in the transfer-induced AHR ------------------------------------------------------------------------------------------ The results obtained so far indicated that antigen-stimulated CD4^+^Th2 cells reached the lung and proliferated there, then produced Th2-type cytokine, which resulted in the direct induction of AHR. They also indicated that antigen-pulsed memory/effector T cell phenotype might play an important role in the transfer-mediated AHR induction. So, finally we examined their role in our system. CD4^+^CD62L^high^T cells and CD4^+^CD62L^low^T cells were prepared by Ab-coated microbeads and column separation. In an in vitrostudy, the CD62L^low^memory/effector subset proliferated and produced Th2-type cytokines in response to OVA, whereas the CD62L^high^subset did not (Figure [9A](#F9){ref-type="fig"} and [9B](#F9){ref-type="fig"}). Moreover, the CD62L^low^memory/effector subset from OVA-sensitized mice produced Th2-type cytokines even without further antigen stimulation, although the values were low (data not shown). Then, we performed transfer study. We determined the numbers of CD62L^high^and CD62L^low^cells to be transferred based on the physiologic ratio of these phenotypes (CD62L^high^cells were 68% and CD62L^low^cells were 32% of splenic CD4^+^T cells, respectively). Transfer of the CD62L^low^memory/effector subset, but not the CD62L^high^subset, induced AHR (Figure [9C](#F9){ref-type="fig"}; CD4, 3,968 ± 258 μg/ml, CD62Lhi, 6,549 ± 645 μg/ml, CD62Llo, 3,824 ± 420 μg/ml). When we evaluated AHR by measuring Raw, similar results were obtained (Figure [9D](#F9){ref-type="fig"}; PC~200~Mch Raw; CD4, 23,840 ± 3,350 μg/ml, CD62Lhi, 41,146 ± 6,451 μg/ml, CD62Llo, 22,146 ± 6,150 μg/ml). We also confirmed that some transferred CD4^+^CD62L^low^cells actually recruited into the lung (Figure [9E](#F9){ref-type="fig"}). Moreover, in the mice that received the CD62L^low^subset, some T cells in the lung proliferated there without further antigen stimulation (Figure [9F](#F9){ref-type="fig"}). These results strongly indicated that CD4^+^CD62L^low^memory/effector T cells were essential for this transfer-mediated, antigen-induced AHR. ::: {#F9 .fig} Figure 9 ::: {.caption} ###### CD4^+^CD62L^low^memory/effector subset produces Th2-type cytokines and directly induces transfer-mediated AHR. (A and B)CD4^+^CD62L^low^memory/effector subset proliferates and produces Th2-type cytokines with antigen stimulation. On day 18, CD4^+^T cells (CD4), CD4^+^CD62L^high^T cells (CD62Lhi), or CD4^+^CD62L^low^T cells (CD62Llo) from OVA-sensitized mice were positively selected by magnetic cell sorting as described in Methods. Then, these cells were cultured with freshly isolated mitomycin C-treated splenocytes in the presence of OVA. After 48 hours, the proliferation was assessed by BrdU incorporation using ELISA (A). The maximum proliferation observed in response to OVA for CD4^+^CD62L^low^T cells from OVA-sensitized mice was set as control (100%). After 72 hours, Th2-type cytokine levels in the supernatants were assayed using ELISA (B). \\\* p \< 0.001 compared with the value of CD4^+^CD62L^high^T cells. (C and D) CD4^+^CD62L^low^memory/effector T cells induce AHR. Recipients received CD4^+^(CD4; 1.25 × 10^7^), CD4^+^CD62L^high^(CD62Lhi; 8.5 × 10^6^), or CD4^+^CD62L^low^(CD62Llo; 4 × 10^6^) cells from OVA-sensitized mice. AR was measured 4 days after the transfer. (C) AR was measured by Penh methods (n = 5--6 per group). \* p \< 0.05 compared with PC~200~Mch of mice that received CD4^+^CD62L^high^T cells. (D) AR was assessed by measurement of Raw (n = 6--8 per group). \* p \< 0.05 compared with PC~200~Mch Raw of mice that received CD4^+^CD62L^high^T cells. (E)Fluorescence study. CD4^+^CD62L^low^cells from OVA-sensitized mice (4 × 10^6^) were labeled with fluorescent dye, and then transferred into recipients. Five-micrometer sections of the lungs were observed by fluorescence microscopy 24 hours after the transfer (CD62Llo). Scale bar, 100 μm. (E) Double staining analysis by immunohistochemistry. T cells were detected by staining for CD3 (cytoplasm, blue). Proliferation was assessed by staining for PCNA (nucleus, brown). Lung sections from mice that received OVA-sensitized CD4^+^CD62L^low^cells (CD62Llo; 4 × 10^6^) are shown. Proliferating T cells were clearly detected (black arrow). Scale bar, 40 μm. ::: ![](1465-9921-6-46-9) ::: Discussion ========== In the current study, we demonstrated that transfer of antigen-induced cellular immune response into naive mice reconstituted AHR in an antigen free setting. We found that CD4^+^CD62L^low^Th2 cells play an essential role in this process. Our results strongly suggest that in sensitized individuals, memory/effector T cells could reach the lung tissue and locally act on the airways, thus would directly induce and maintain basal AHR independently of eosinophils, although the intensity could be moderate. AHR is one of the most characteristic features of asthma \[[@B1]\]. However, the precise mechanism for its induction has not been fully clarified. It is considered that eosinophilic airway inflammation is closely linked to the AHR induction \[[@B1],[@B4],[@B5]\]. However, a causal link between eosinophilic airway inflammation and AHR has not been established. On the other hand, CD4^+^T cells, especially CD4^+^Th2 cells, are also involved in the induction of AHR \[[@B11],[@B13]-[@B18]\]. However, the significance of eosinophils or CD4^+^T cells on the AHR induction has been clarified only in the effector phase, under the condition of airway antigen challenge. Therefore, the role of these cells in the AHR induction has not been evaluated in an antigen free setting. We previously reported that systemic antigen sensitization alone directly induces AHR before development of eosinophilic airway inflammation \[[@B19]\]. In addition, in the current study, transfer of spleen cells obtained from antigen-sensitized mice induced a significant AHR in naive mice without airway eosinophilia (Figure [1](#F1){ref-type="fig"}, Figure [3](#F3){ref-type="fig"}, Figure [4A](#F4){ref-type="fig"}, and Table [1](#T1){ref-type="table"}). These results indicate that antigen-sensitized spleen cells could directly induce AHR. In humans, we previously reported that some patients with atopic dermatitis who are highly sensitized to mite antigen have a moderate AHR regardless of the lack of any history of asthma \[[@B25]\]. This would support the speculation that sensitization to an antigen could directly induce AHR also in humans. We measured AR mainly by Penh system throughout the current study. Penh system has been widely used for measurement of AR to Mch in BALB/c mice \[[@B19],[@B20],[@B26]\]. Measuring Penh is superior to measuring Raw of ventilated mice in terms of its conciseness and non-invasiveness. In addition, sampling bias caused by maneuver in Penh system seemed to be lower than that in invasive ventilator system. Based on these advantages of Penh system, we could measure AR of large numbers of mice simultaneously with a good reproducibility. However, the accuracy of Penh as an indicator of AR has been recently criticized because Penh does not correlate with Raw especially in C57BL/6 mice \[[@B27]-[@B29]\]. Measuring Penh is more frequently affected by the heat and humidification than measuring Raw \[[@B30],[@B31]\]. Considering these experimental and theoretical problems, we also examined AR by measuring Raw with ventilated mice in the most important experiments (Figures [1C](#F1){ref-type="fig"} and [9D](#F9){ref-type="fig"}). We confirmed that similar results were obtained by the two systems. Therefore, we considered that the data obtained from the Penh system in the current study were, to a certain extent, reliable. Next we studied the mechanism of the cell transfer-induced AHR. The fluorescence study and double staining demonstrated that these transferred cells reached the lung and some T cells actually proliferated in the lung without further airway antigen challenge (Figure [4B](#F4){ref-type="fig"} and [4C](#F4){ref-type="fig"}). Byersdorfer et al. reported that some transferred Th1 cells migrate in the airway before antigen challenge \[[@B32]\]. Julia et al. reported that antigen sensitization alone distributes antigen-specific T cells in the BALF and in the lung \[[@B33]\]. These reports support the present findings. Therefore, some antigen specific T cells could have reached the lung in an antigen free setting. On the other hand, transferred cells reached the tissues other than lung such as colon, which induced a slight inflammation there (Figure [4D](#F4){ref-type="fig"} and [4E](#F4){ref-type="fig"}). These results suggested that antigen sensitization or transfer of antigen-sensitized spleen cells would distribute antigen-specific T cells among tissues, and thus could induce some immure response there even without local antigen challenge. The result of BALF cytokine concentrations (Figure [5A](#F5){ref-type="fig"}) and transfer of CD4^+^Th2 cells (Figure [8D](#F8){ref-type="fig"}) demonstrated that CD4^+^Th2 cells played an essential role in the induction of the AHR. In addition, results from positive and negative selection studies clarified that CD4^+^CD62L^low^T cells were essential for transfer-induced AHR induction, while other phenotypes such as CD4^+^CD62L^high^T cells or iVα14 NKT cells were not (Figures [8](#F8){ref-type="fig"} and [9](#F9){ref-type="fig"}). There are two distinct populations of memory CD4^+^T cells, which are distinguished by the expression of CCR7 \[[@B34]\]. One is CCR7^high^ central memory T cells which home preferentially to lymph nodes, another is CCR7^low^effector memory cells which traffic more efficiently to non-lymphoid tissues and acquire effector function more rapidly \[[@B34],[@B35]\]. Central memory T cells express high levels of CD62L \[[@B34]\]. In contrast, effector memory T cells express CD62L to a lower and variable extent \[[@B34]\]. Therefore, CD62L^low^memory T cells are thought to be a subset of effector memory T cells \[[@B34],[@B36]\]. We demonstrated that the CD62L^low^subset proliferated and produced Th2-type cytokines in response to OVA, while the CD62L^high^subset did not (Figure [9A](#F9){ref-type="fig"} and [9B](#F9){ref-type="fig"}). Moreover, transferred CD62L^low^subset recruited into the lung, and proliferated there without local antigen challenge (Figure [9E](#F9){ref-type="fig"} and [9F](#F9){ref-type="fig"}). Therefore, in our system, some antigen specific T cells could have acquired a memory/effector phenotype (effector memory T cells), reached the lung, proliferated and produced cytokines, then finally induced AHR in an antigen free setting. Furthermore, local antigen stimuli would further accelerate immune response and thus airway eosinophilia would be induced \[[@B37]\], which could further increase AHR (Figures [1A](#F1){ref-type="fig"} and [4A](#F4){ref-type="fig"}). So far, whether immunocompetent cells would modify intrinsic AHR has remained unclear. De Sanctis et al. reported that T lymphocytes mediate intrinsic AHR \[[@B38]\]. However, Hadeiba et al. later reported that transfer of CD4^+^T cells that are not stimulated with antigen does not alter the intrinsic AHR \[[@B39]\]. The latter finding is strongly supported by the present findings that antigen-stimulated cells mediated AHR induction, while cells that had not encountered antigen did not. In vitroOVA stimulation of OVA sensitized spleen cells enhanced transfer activity to induce AHR (Figure [7A](#F7){ref-type="fig"}) and cytokine concentration in BALF (Figure [7B](#F7){ref-type="fig"}). Wise et al. reported that in vitroOVA stimulation enhances transfer activity to provoke eosinophilic airway inflammation \[[@B37]\]. We needed larger numbers of OVA-sensitized spleen cells to induce AHR than other studies \[[@B17],[@B18]\], because we did not stimulate cells with OVA in vitrobefore transfer in most part of this study. Spleen cells from OVA-sensitized mice produced greater amount of cytokines upon further stimulation with OVA in vitro(data not shown). Therefore, the more Th2-type cytokine was produced by transferred cells, the greater the intensity of AHR increased (Figure [7](#F7){ref-type="fig"}). Among cytokines produced by CD4^+^Th2 cells, which is most important for developing of AHR has not been fully clarified. IL-13 is a candidate for developing AHR independently of eosinophilic inflammation \[[@B12]\]. On the other hand, some previous studies \[[@B10],[@B40]\] including ours \[[@B19]\] showed that IL-4 also plays an important role. In the current study, IL-4 and IL-5 concentrations in BALF increased by cell transfer (Figure [5](#F5){ref-type="fig"}), while IL-13 did not increase significantly (data not shown). Moreover, transfer of spleen cells from antigen-sensitized IL-4-deficient mice did not induce AHR (Figure [6](#F6){ref-type="fig"}). So in the present experimental system, IL-4 would play a major role. Airway smooth muscle cells express IL-4Rα \[[@B41],[@B42]\]. We speculate that one of the mechanisms of AHR induction by cell transfer could be a direct influence on smooth muscle cells \[[@B43]\]. The relationship between Th2 cytokine and smooth muscle cell should be further elucidated. NKT cells are another cell source of IL-4 \[[@B44]-[@B47]\] and IL-13 productions \[[@B44]\], and would influence the subsequent adaptive immune response and T cell polarization \[[@B45],[@B47]\]. The development of AHR is abrogated in NKT-cell deficient mice \[[@B44],[@B46]\]. However, in the current study, iVα14 NKT cells did not influence this transfer-mediated AHR (Figure [8A](#F8){ref-type="fig"}). This indicates that although NKT cells might play some important role in a certain phase of sensitization, once conventional T cells are activated and gain effector function, iVα14 NKT cells would have little contribution to the induction of AHR. Conclusion ========== In conclusion, the current study clarified the significance of antigen-specific memory/effector CD4^+^T cell-mediated Th2-type immune response as an essential factor to induce basal AHR in an antigen independent setting. It would also propose that suppression of antigen-specific immune response itself should be a critical target for controlling allergic asthma. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= KN and MD carried out animal experiments and preparation of the manuscript. KO, YT, AS and RT carried out animal experiments. YK and KN assisted with positive/negative cell selections. KY participated in the direction of the study. All authors read and approved the final manuscript. Acknowledgements ================ We thank I. Makino, E. Kasukawa, A. Tobimatsu and M. Katakawa for technical assistance.	PubMed Central	2024-06-05T03:55:59.558423	2005-5-28	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180472/", "journal": "Respir Res. 2005 May 28; 6(1):46", "authors": [ { "first": "Kazuyuki", "last": "Nakagome" }, { "first": "Makoto", "last": "Dohi" }, { "first": "Katsuhide", "last": "Okunishi" }, { "first": "Yasuo", "last": "To" }, { "first": "Atsushi", "last": "Sato" }, { "first": "Yoshinori", "last": "Komagata" }, { "first": "Katsuya", "last": "Nagatani" }, { "first": "Ryoichi", "last": "Tanaka" }, { "first": "Kazuhiko", "last": "Yamamoto" } ] }
PMC1180473	Background ========== The lung represents a site for the invasion of various bacteria or bacterial products. Along with alveolar macrophages, pulmonary epithelial cells are the first cells to be challenged by pathogenic microorganisms. The gram-negative bacterium Chlamydophila (Chlamydia) pneumoniae (C. pneumoniae) is an obligate intracellular pathogen causing acute and chronic pneumonia \[[@B1],[@B2]\]. The Toll-like receptor (TLR)-family is an integral part of the innate immune system and recognizes conserved pathogen-associated molecular patterns (PAMPs) on microorganism. The interaction of TLRs with pathogen components initiates a signaling cascade that activates the adaptive immune response mechanisms which subsequently lead to inflammatory response and to the elimination of the pathogen \[[@B3]\]. TLRs are mainly expressed in professional immune cells in the alveolus. However, TLRs have also been found on type II pneumocytes \[[@B4]-[@B6]\] and could thus play an important role in the innate immune response at the alveolar surface area. It is assumed that different TLRs recognize different classes of PAMPs \[[@B7]\]. TLR2 recognizes lipoproteins, peptidglycans and lipoteichoic acid. TLR4 is the receptor for lipopolysaccharide (LPS) and mediates the LPS signal transduction together with other molecules such as CD14, MD-2, myeloid differentiation factor 88 (MyD88), and so on \[[@B8]\]. Heat-shock protein (HSP) is one of the most phylogenetically conserved proteins in prokaryotes and eukaryotes \[[@B9]\]. Recent studies suggest, that chlamydial HSP stimulates innate immune and inflammatory responses by a TLR-mediated pathway, that is independent from LPS \[[@B10],[@B11]\]. Recognition of PAMPs by TLR results in early host defense as well as in the activation of an inflammatory response pathway that involves mitogen-activated protein kinase (MAPK) and nuclear factor-kappaB (NF-κB). Furthermore, recognition of PAMPs induces the production of cytokines and stimulates the maturation of antigen-presenting cells \[[@B8],[@B3]\]. Type II pneumocytes are responsible for the metabolism of alveolar surfactant and have recently been suggested to play an important role in the inflammatory response of the lung. Little is known about events that are induced by an interaction of bacteria with type II cells. We have recently shown that the contact of C. pneumoniae with microvilli of type II cells induces changes in cytoskeleton and leads to activation of NF-κB pathway \[[@B12]\]. Here, we tested whether C. pneumoniae stimulates expression of TLR2 and/orTLR4 in type II cells to induce the production of the pro-inflammatory cytokine TNF-α and chemokine MIP-2 via Rho-GTPase-and NF-κB-dependent pathways. The identification of the involved pathways is essential for the understanding of the innate immune defense mechanisms and pulmonary inflammatory response of the alveolus. Methods ======= Cell culturing -------------- C. pneumoniae strain TW183 was cultured and purified as described previously \[[@B13]\]. Briefly TW183 was grown to high titers in cycloheximide-treated HEp-2 cells. Infected monolayers were harvested from culture flasks and sonicated for 30 sec. Cellular debris was removed by centrifugation at 500 × g for 10 min at 4°C. Titer determination in cycloheximide-treated HEp-2 cells was performed with a thawed aliquot (in triplicate). Aliquots diluted with an equal volume of sucrose-phosphate-glutamate buffer supplemented with 10% foetal calf serum (FCS) were stored at -80°C until use. C. pneumoniae concentrations used were expressed as inclusion forming units (IFU)/ml. Type II pneumocytes were isolated from the lungs of adult male Wistar rats (body weight 120--140 g) according to Dobbs et al. \[[@B14]\]. Rat alveolar macrophages were separated from bronchial lavage fluid. The lungs were lavaged six times with phosphate-buffered saline (PBS) containing 0.2 mM EGTA (ethylene glycol tetraacetic acid). Cells were then collected by centrifugation at 230 × g for 10 min. The cell pellet was re-suspended in modified Eagle medium (MEM) (PAA laboratories GmbH, Linz, Austria). Isolated type II cells were pre-incubated with the following substances: mevastatin (10 μM), also known as compactin, a specific inhibitor of the Rho-GTPase family; the calcium chelator BAPTA/AM (5 μM); SN50 (50 μg/ml), a cell-permeable peptide that inhibits the nuclear translocation of NF-κB/Rel complexes; SN50M (50 μg/ml), a cell-permeable control peptide; parthenolide (25 μM), an inhibitor of NF-κB through prevention of IκB phosphorylation (all chemicals were obtained from Calbiochem, Darmstadt, Germany) or MEM with corresponding concentrations of dimethyl sulfoxide (DMSO). Bacteria were treated with polymyxin obtained from Sigma, Deisenhofen, Germany. Thereafter, type II cells and macrophages in MEM in absence of FCS were incubated with or without C. pneumoniae with a multiplicity of infection (MOI) of 2 (2 × 10^6^IFU per 1 × 10^6^target cells) as used in previous experiments \[[@B12]\]. The viability of isolated alveolar macrophages or type II cells was assessed by trypan blue staining and was more than 95% or 98% respectively and remained above 94% after cell incubation with or without C. pneumoniae and different inhibitors. Cell purity, estimated by hematoxylin staining, was more than 95% or 92% respectively. RNA extraction and Semi-Quantitative Polymerase Chain Reaction (PCR) -------------------------------------------------------------------- Cellular RNA was extracted from 5 × 10^5^type II cells using an RNeasy™ extraction kit (Quiagen, Hilden, Germany) according to the manufacturers instructions. RNA concentration was quantified by determining optical density at 260 and 280 nm (Ultraspec 2000 Spectrophotometer, Pharmacia Biotech, Freiburg, Germany) and adjusted to 100 pg/μl. Semi-quantitative polymerase chain reaction was used to assess TLR2 and TLR4 mRNA expression under different conditions in rat alveolar type II cells. The mRNA expression levels of TLR2 and TLR4 genes were determined in total RNA from pooled samples of three independent experiments. Reverse transcription (RT) of 1 μg RNA into cDNA was performed by using random hexamers, AMV reverse transcriptase and 1× PCR buffer (20 mM Tris-HCL, pH 8.3, 50 mM KCl, 2 mM MgCl, 100 μg/ml bovine serum albumin). RT was performed at 42°C for 30 min followed by 95°C for 5 min to inactivate the reverse transcriptase enzyme. The cDNA was amplified by PCR in a total volume of 50 μl using 2.5 U Ampli-Taq DNA polymerase, 100 μM dATP, dCTP, dGTP, 50 μM dTTP and 0.5 μM of each primer in 1× PCR buffer. Incorporation of digoxigenin was performed by addition of 10 μM digoxigenin-11-dUTP to the PCR reaction mixture. For GAPDH housekeeping gene, PCR was performed with 25 cycles of 1 minute each at 95°C, 54°C and 72°C in a microprocessor-driven thermal cycler (Fa. Landgraf, Hannover, Germany). For TLR2 and TLR4 amplification, PCR was performed with 35 or 40 cycles of 1 minute each at 95°C, 65°C and 72°C. Primer sequences for GAPDH were (GenBank BC059110; product size 303 bp): 5\'-CAG TGC CAG CCT CTG CTC AT and ATA CTC AGC ACC AGC ATC AT-3\'. Primer sequences for rat TLR2 were (GenBank AY151255; product size 199 bp): 5\'-GTC CAT GTC CTG GTT GAC TGG and GAT ACC ACA GCC CAT GGA AAT-3\'. Primer sequences for rat TLR4 were (GenBank NM\_019178; product size 150 bp): 5\'-GAG CCG GAA AGT TAT TGT GG and AGC AAG GAC TTC TCC ACT TTC T-3\'. Products were stained with ethidium bromide and electrophoresed through a 1.5% agarose gel. The PCR products were transferred to nylon membranes by capillary blotting using 20× SSC as blotting buffer. The nylon membranes were fixed by UV light and the digoxigenin-UTP that had been incorporated into the PCR products was visualized by staining with anti-digoxigenin antibody conjugated to alkaline phosphatase \[[@B15]\]. Luminescence of the substrate (Lumigen™ PPD) was documented by short exposure to X-ray film. Densitometric calculations of digital film images were performed with the analysis program Scion Image (Scion Corporation, Frederick, MD). Western blot analysis --------------------- Isolated type II cells were incubated with C. pneumoniae in suspension. Nuclear and cytoplasmatic extracts and membrane fractions were prepared from cells as reported previously \[[@B16],[@B17]\]. The protein content of nuclear, cytoplasmatic and membrane fraction extracts was determined by Bradford assay using bovine serum albumin (BSA) as a standard method (Bio-Rad Laboratories protein assay kit, Richmond, USA). Protein fractions were stored at -80°C. TLR4 protein was assayed in membrane and cytoplasmatic extracts, IκBα protein in cytoplasmatic extracts, NF-κB p65 protein in nuclear extracts and Rac1 and RhoA protein in membrane extracts of type II cells by Western blotting. Proteins (40 μg or 60 μg for TLR4, 40 μg for IκBα, 10 μg for p65 and 40 μg for Rac1 and RhoA) from each sample were mixed with 2 × sodium dodecyl sulfate (SDS) sample buffer, heated at 95°C for 5 minutes, and separated by 10% (TLR4, p65), 12.5% (IκBα) and 15% (Rac1 and RhoA) SDS-polyacrylamide gels. The separated proteins were blotted onto nitrocellulose membranes. Non-specific binding sites were blocked with 10% nonfat dry milk in Tris-buffer saline (TBS)-0.05% Tween. Goat IgG anti-TLR4 (L-14) sc-16240, mouse monoclonal IgG anti-IκBα (H-4) sc-1643, rabbit IgG anti-NF-κB p65 (C-20) sc-372, goat IgG anti-Rac1 (T-17) sc-6084 and rabbit IgG anti-RhoA (119) sc-179 antibodies (from Santa Cruz Biotechnology, Heidelberg, Germany) were used for primary detection. Peroxidase-conjugated anti-goat IgG, anti-mouse IgG and anti-rabbit IgG antibodies (Dianova, Hamburg, Germany) were employed for secondary detection. The specific protein bands were visualized on film by enhanced chemiluminescence (ECL) (Amersham Biosciences, Inc, Piscataway, NJ) according to the manufacturers. The bands were quantified by scanning densitometry using a GS-710 Imaging Densitometer (Bio-Rad, Hercules, CA). All other standard reagents and chemicals were of analytical grade and obtained from different suppliers. Immunocytochemistry and CLSM ---------------------------- The isolated type II cells were plated on glass cover slips for 3 hours at 37°C. For TLR4-binding experiments, type II cells were incubated with and without C. pneumoniae for 10 min at 37°C, washed and then without permeabilization incubated with goat anti-TLR4 antibody for 60 min at 22°C. Cells were pre-incubated with or without drugs or with corresponding concentrations of DMSO and incubated with C. pneumoniae as described by Wissel et al. \[[@B12],[@B18]\]. Cells were fixed in 1% paraformaldehyde/250 mM Hepes (pH 7.4) and permeabilized with 0.04% saponin. For staining of TLR4 cells were incubated with goat anti-TLR4 antibody for 20 hours at 4°C. In experiments designed to block NF-κB activity, type II cells were pre-incubated with 20 μg/ml anti-TLR4 monoclonal antibody (HTA125, preservative free) or isotype control antibody (QBIOGENE-ALEXIS GmbH, Grünberg, Germany) for 30 min at 37°C before incubation with C. pneumoniae for 60 min at 37°C. To stain the NF-κB subunits within the cells, the following antibodies were used: mAb anti-IκBα and mAb anti-NF-κB p65 subunit, recognizing predominantly p65 only when IκBα is not bound to p65 (Chemicon, Hofheim, Germany) and incubated for 20 hours at 4°C. For staining of Rac1 and Rho protein cells were incubated with goat anti-Rac1 and rabbit anti-RhoA antibody for 2 hours at 22°C. After washing steps in PBS, cells were incubated with one of the following secondary antibodies: Alexa^®^488 or Alexa^®^594 anti-goat IgG, anti-mouse IgG labeled with Alexa^®^488 or Alexa^®^594, or with anti-rabbit IgG labeled with Alexa^®^594 (Molecular Probes Europe BV, Leiden, Netherlands) for 2 hours at 22°C. To study the F-actin cytoskeleton, Alexa^®^488-conjugated phalloidin (Molecular Probes) was used (2 hours at 22°C). Nuclear DNA was stained with 4\',6-diamidino-2-phenylindole (DAPI) (Molecular Probes) for 20 minutes. For TLR4-binding experiments samples were analyzed with a Leica CLSM (Leica Lasertechnik GmbH, Heidelberg, Germany), equipped with a krypton-argon laser as described previously \[[@B18]\]. Images were captured with a NPL Fluotar 40×/1.3 oil immersion objective. To visualize the cell boundaries, images were recorded simultaneously with bright-field contrast in relation with the channel system of CLSM in the equatorial plane of cells. The triple staining was analyzed with a Zeiss laser scanning microscope LSM 510 META with Axiovert 200 M (Carl Zeiss Jena GmbH, Jena, Germany) as described previously \[[@B12]\]. Images were taken using a Plan-Apochromat 63×/1.4 oil immersion objective and fluorescence excitation at 488 nm (30 mW Ar-Laser), 543 nm (1 mW HeNe-Laser) and 405 nm (Diode Laser). To minimize signal crosstalk, a sequential scan with fast change of excitation lines was performed. The semi-quantitative estimation of NF-κB subunits was made according to the methods described previously by Wissel et al. \[[@B12]\] and analyzed with Zeiss LSM 510 system with Axiovert microscope as described above. Thirty randomly chosen cells from six different regions were analyzed. Confocal settings were optimized and maintained for all images. Images were recorded with a scanning speed at 8s/frame with a resolution of 512 × 512 pixels. Serial sections of cells (z stack) at a depth of 0.5 μm were performed. Fluorescence intensity was determined on confocal images using a computer-based image analysis system equipped with MetaView™ software (obtained from Visitron Systems GmbH, Puchheim, Germany) by measuring the average intensity per pixel within a fixed-area circle placed over the relevant area of the cell. TNF-α ELISA and MIP-2 ELISA --------------------------- Three-hour adherent type II cells and macrophages (2 × 10^6^cells/ml) on plastic dishes were incubated with or without C. pneumoniae for different periods of time. Adherent type II cells were pre-incubated with BAPTA/AM, SN50, SN50M or parthenolide followed by incubated with C. pneumoniae for 3 to 6 hours. Cell culture supernatants were collected and analyzed for TNF-α and MIP-2 secretion by enzyme-linked immunosorbent assay (ELISA). The levels of TNF-α production were determined using rat TNF-α Quanticine ELISA (R&D Systems Inc, Minneapolis, USA). MIP-2 concentration in the samples was determined using rat MIP-2 ELISA kit (Biosource Europe, Nivelles, Belgium) following the manufacturer\' s instructions. Statistics ---------- Statistical comparisons were performed by analysis of variance (ANOVA) with subsequent Fisher\'s protected least significant difference (PLSD) test. The level of significance was set at p \< 0.05. Results ======= C. pneumoniae increase the expression of TLR4 mRNA but not of TLR2 mRNA ----------------------------------------------------------------------- Incubation of rat type II pneumocytes with C. pneumoniae was associated with an increased expression of TLR4 mRNA (Figure [1](#F1){ref-type="fig"}). The mRNA concentration for TLR4 roughly doubled after 2 hours of incubation with C. pneumoniae compared to non-infected controls. The ratio between TLR4 mRNA signal and GAPDH signal was 0.8 in control type II cells and 1.4 after 2 hours of treatment with C. pneumoniae. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### TLR4 mRNA expression in type II cells. Detection of TLR4 (upper lane) and TLR2 mRNA expression (middle lane) in type II cells after exposure to C. pneumoniae for 2 hours (1) and in non-infected control cells (2) by semiquantitative polymerase chain reaction. GAPDH (lower lane) served as internal control. After 2 hours of incubation with C. pneumoniae TLR4 mRNA expression in type II cells approximately doubled whereas TLR2 mRNA remained unchanged. ::: ![](1465-9921-6-51-1) ::: In contrast, 2 hours of incubation did not affect TLR2 mRNA expression, the ratio of TLR2 mRNA and GAPDH signal was 1.3 in control and 1.1 in C. pneumoniae treated cells (Figure [1](#F1){ref-type="fig"}). The density of the shown GAPDH bands was homogenous for both infected and non-infected type II cells. The second GAPDH band showed a 0.89-fold expression compared to the first band. Effect of C. pneumoniae on TLR4 protein expression -------------------------------------------------- To test whether C. pneumoniae contact rapidly increases the TLR4 protein expression and whether newly synthesized TLR4 protein is required, we first characterized C. pneumoniae-mediated TLR4 protein expression time-dependent in isolated rat type II cells. The presence of TLR4-protein was confirmed by Western blotting in non-infected and in type II cells after 30, 60, 120 and 180 minutes incubation with C. pneumoniae (Figure [2A](#F2){ref-type="fig"} and [2B](#F2){ref-type="fig"}). As quantified by scanning the chemiluminescence\'s bands from four different experiments. The TLR4 protein expression was highest at 30 and 180 min (1.6-fold increase versus control cells). After 60 min and 120 min incubation with C. pneumoniae the TLR4 expression did not differ from non-infected cells. These two peeks suggest that after 30 min contact with C. pneumoniae TLR4 protein is expressed constitutively at a post-transcriptional level whereas 3 hours incubation with bacteria mediated newly synthesized TLR4 protein expression. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### TLR4 protein expression in type II cells has two peeks. (A) The cytoplasm protein fraction of type II cells was subjected to SDS-PAGE and immunoblotting, non-infected cells (lane 1), 30 min (lane 2), 60 min (lane 3), 120 min (lane 4) and 180 min (lane 5) exposure with C. pneumoniae. (B) The TLR4 protein expression was estimated by densitometry. Values of n = 4 experiments are given as means ± SD in arbitrary units \[AU\]). Asterisk indicates a significant difference to controls (1), 60 min (3) and 120 min (4) (P \<0.05). ::: ![](1465-9921-6-51-2) ::: TLR4 protein was only detectable with a goat polyclonal anti-TLR4 antibody (L-14), mapping an epitope within the extra cellular domain. Distribution of TLR4 in type II cells ------------------------------------- Confocal imaging revealed rapid recycling of TLR4 from cytoplasm to the cell surface in response to C. pneumoniae incubation. Binding-assays with C. pneumoniae incubation of type II cells demonstrated a TLR4 expression on the external membrane of the cells but not on non-infected cells (Figure [3A](#F3){ref-type="fig"}). ::: {#F3 .fig} Figure 3 ::: {.caption} ###### TLR4 is localized extra-and intracellularly in presence of C. pneumoniae. (A) Representative CLSM images after binding of C. pneumoniae at type II cells. Cells were without permeabilization stained with anti-TLR4 antibody that was detected by a secondary antibody coupled to Alexa488 (green). TLR4 is expressed at the external cell membrane. In un-infected cells no TLR4 expression was seen. Transmission images with pseudo-color red was used to highlight the outline of the cells. (B) Representative CLSM images of type II cells after incubation with C. pneumoniae at 37°C. Cells were after permeabilization stained with anti-TLR4 antibody that was detected by a secondary antibody coupled to Alexa594 (red) and labeled for F-actin with Alexa488-conjugated phalloidin (green). Nuclear DNA was stained with DAPI (blue). TLR4 mobilization to the external cell membrane reached a peak value after 10--30 min after exposure to C. pneumoniae. Areas of overlap between red and green are yellow, the bar equals 10 μm. Shown are representative examples, three more experiments gave similar results. ::: ![](1465-9921-6-51-3) ::: Time-dependent incubation with and without C. pneumoniae at 37°C, that was followed by fixation and permeabilization of the cells, revealed a TLR4 mobilization from the cytoplasm to the external cell surface that reached a peak value between 10--30 min after challenge with C. pneumoniae. The surface TLR4 expression started to decrease after 60 min in C. pneumoniae incubated cells. In non-infected cells TLR4 was partially co-localized with the intact F-actin cytoskeleton close to the cell membrane. A significant change in the F-actin cytoskeleton was found in cells after 30 and 60 minutes of C. pneumonia incubation (Figure [3B](#F3){ref-type="fig"}). Heat-inactivated C. pneumoniae have no effect on TLR4 expression ---------------------------------------------------------------- Western blot analysis was used to determine the influence of protein components and/or LPS of C. pneumoniae on the expression of TLR4 in type II pneumocytes (Figure [4A](#F4){ref-type="fig"} and [4B](#F4){ref-type="fig"}). Incubation of type II cells with C. pneumoniae for 30 min stimulated TLR4 expression 2.1-fold when compared with non-infected controls. The TLR4 expression was significantly lower after incubation of cells with heat-inactivated bacteria (56°C for 30 min) than after 30 min of treatment with viable C. pneumoniae. Pre-treatment of C. pneumoniae with polymyxin B (2 μg/ml) did not alter the TLR4 expression. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### TLR4 expression is stimulated by a heat-sensitive component of C. pneumoniae. (A) TLR4 was visulalized by Western blot technique and (B) the expression determined by densitometry in non-infected cells (lane 1), non-treated C. pneumoniae (lane 2), heat-inactivated C. pneumoniae (lane 3) and polymyxin pre-treated bacteria (lane 4). Values of n = 4 experiments are given as means ± SD in arbitrary units. Asterisk indicates a significant difference to non-treated C. pneumoniae (2) and polymyxin pre-treated bacteria (4) (P \< 0.0001). ::: ![](1465-9921-6-51-4) ::: Anti-TLR4 antibodies inhibit C. pneumoniae-mediated NF-κB- activation --------------------------------------------------------------------- In experiments designed to block NF-κB activity, type II cells were pre-incubated with 20 μg/ml anti-TLR4 monoclonal antibodies (HTA125 or isotype control antibody) for 30 min and thereafter incubated with C. pneumoniae for 30 min. The effect on IκBα and NF-κB p65 was studied qualitatively (Figure [5A](#F5){ref-type="fig"}) and semi-quantitatively by CLSM. In non-infected type II cells IκBα was found in the cytoplasm and co-localized with the F-actin cytoskeleton. 30 min of bacterial contact caused a significant (p \< 0.0001) decrease of the amount of IκBα in the cytoplasm when compared with control cells (15.9 ± 6 vs. 90.3 ± 16). Staining with anti-p65 antibody (recognizes p65 only when IκBα was not bound to p65) showed almost no activation of p65 in cytoplasm and nucleus in non-infected cells. Activated p65 was significantly (p \< 0.0001) increased in the nucleus after C. pneumoniae contact (114 ± 10 vs. 6.3 ± 1.4 in control cells). Pre-treatment of cells with anti-TLR4 monoclonal antibodies, followed by C. pneumoniae incubation prevented IκBα decrease and p65 activation. Isotype control antibody had no effect on C. pneumoniae-mediated NF-κB activation (results not shown). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Inhibition of C. pneumoniae-mediated NF-κB activation with anti-TLR4 antibodies. (A) Representative CLSM images after incubation of type II cells pre-treated or not with anti-TLR4 antibodies followed by C. pneumoniae incubation for 30 min. Cells were stained with anti-IκBα antibody and anti-p65 antibody both detected by a secondary antibody coupled to Alexa594 (red), F-actin was stained with Alexa488-conjugated phalloidin (green) and nuclear DNA with DAPI (blue). In C. pneumoniae infected cells IκBα is decreased and p65 is localized in the nucleus. In cells that were treated with anti-TLR4 antibodies IκBα is not decreased and p65 is not localized in the nucleus. The effect on IκBα and NF-κB p65 was studied semi-quantitatively by CLSM as described in Materials and Methods. The C. pneumoniae-mediated IκBα degradation resulting in a 5.6-fold decrease versus control cells and NF-κB p65 translocation in the nucleus resulting in a 18-fold increase versus control cells. Areas of overlap between blue and red are pink, and between red and green are yellow, the bar equals 10 μm. (n = 5 experiments). (B and C) The amount of cytoplasmatic IκBα and of nuclear p65 of type II cells in controls (lane 1), incubated with C. pneumoniae for 30 min (lane 2) and pre-incubated with anti-TLR4 (HTA) antibodies prior to C. pneumoniae exposure for 30 min (lane 3) was quantified by Western blot analysis. Values of n = 3 experiments are given as means ± SD in arbitrary units. Open bars, IκBα; solid bars, p65. Asterisk indicates a significant difference to controls (1) and cells pre-incubated with anti-TLR4 antibodies (3) (P \< 0.0001). ::: ![](1465-9921-6-51-5) ::: The amount of cytoplasmatic IκBα and of nuclear p65 was measured by Western blot analysis (Figure [5B](#F5){ref-type="fig"} and [5C](#F5){ref-type="fig"}). In comparison to control cells a 4-fold decrease in the amount of IκBα protein in the cytoplasm and a 10-fold increase of p65 in the nucleus was found 30 min after C. pneumoniae contact. In cells pre-treated with 20 μg/ml anti-TLR4 (HTA) antibodies, followed by C. pneumoniae incubation for 30 min neither C. pneumoniae-mediated IκBα degradation in the cytoplasm nor NF-κB activation of p65 in the nucleus was found. BAPTA-AM, parthenolide and SN50 suppress TLR4 mRNA and protein -------------------------------------------------------------- As shown in Figure [6A](#F6){ref-type="fig"}, mRNA expression for TLR4 approximately doubled in type II cells exposed to C. pneumoniae for 3 hours compared to controls. The TLR4 mRNA concentration remained at baseline values in BAPTA-AM and SN50 pre-incubated type II cells followed by 3 hours incubation with C. pneumoniae. The ratio between TLR4 mRNA signal and GAPDH signal was 1.7 in control cells, 1.0 in C. pneumoniae treated cells, 1.6 in BAPTA-AM pre-treated and infected cells and 1.8 in parthenolide pre-treated and infected cells. The density of the shown GAPDH bands was homogenous with GAPDH bands showing a 0.91, 0.98 and 0.84 fold expression compared to the first band, respectively. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### BAPTA-AM, parthenolide and SN50 inhibited C. pneumoniae-mediated TLR4 mRNA and protein expression. (A) Detection of TLR4 (upper lane) and GAPDH mRNA expression (lower lane) in type II cells by semi-quantitative polymerase chain reaction. The C. pneumoniae-induced TLR4 mRNA increase (2) returned to control values (1) after combined exposure to BAPTA-AM (3). Parthenolide (4) exerted an even stronger inhibitory activity and reduced TLR4 mRNA expression below baseline values. (B) Type II cells in suspension controls (lane 1) incubated with C. pneumoniae for 3 hours (lane 2), pre-incubated with BAPTA-AM (lane 3) and SN50 (lane 4) prior to C. pneumoniae exposure were prepared for Western blot analysis of TLR4 protein. (C) Their protein expression was determined by densitometry. Values of n = 3 experiments are given as means ± SD in arbitrary units. Asterisk indicates a significant difference to control (1), BAPTA-AM (3) and SN50 (4) (P \< 0.05). ::: ![](1465-9921-6-51-6) ::: The amount of TLR4 protein was analyzed by Western blot (Figure [6B](#F6){ref-type="fig"}) and quantified by scanning the chemiluminescence\'s bands from three different experiments (Figure [6C](#F6){ref-type="fig"}). TLR4 protein expression exposed to C. pneumoniae for 3 hours was 2.5-fold higher than in control cells. In contrast, the TLR4 protein expression in BAPTA-AM and SN50 pre-incubated type II cells followed by 3 hours of incubation with C. pneumoniae remained at control levels. C. pneumoniae-induced secretion of TNF-α and MIP-2 is NF-κB dependent --------------------------------------------------------------------- Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"} show that C. pneumoniae stimulate the release of TNF-α and MIP-2 from type II cells in a time-dependent manner. TNF-α concentration in the supernatant increased after 3--6 hours incubation with bacteria, and remained constant over the next 12 hours. The MIP-2 concentration in the supernatant continuously increased after bacterial incubation. To rule out a stimulatory effect due to contamination by macrophages (isolated type II cells contain less than 10% macrophages), the release of TNF-α and MIP-2 from isolated macrophages obtained from rat bronchial lavage fluid was measured. As tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"} show, an estimated contamination of 10% macrophages would not influence the TNF-α or MIP-2 concentration in the supernatant. Pre-incubation of cells with BAPTA-AM, parthenolide and SN50 suppressed TNF-α and MIP-2 production (Table [3](#T3){ref-type="table"}). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### TNF-α release in C. pneumoniae infected type II cells ::: Type II cells + C. pneumoniae Macrophages + C. pneumoniae Type II cells -- 10% macrophages ---------- ------------------------------- ----------------------------- ---------------------------------- 0 min 7 ± 1 13.7 ± 2.7 5.6 30 min 21 ± 7.5 69.5 ± 11.3 14 1 hour 109 ± 22 174 ± 9 92 2 hours 183 ± 29 303 ± 69 153 3 hours 656\* ± 103 661§ ± 41 590 6 hours 1006\\ ± 74 1070§S ± 58 899 12 hours 996 ± 294 1038 ± 207 893 Mean values are given in pg/ml ± SD of three independent experiments. \* P \< 0.0001 vs. 0--2 and 6--12 hours + C. pneumoniae \\ P \< 0.0001 vs. 0--3 hours + C. pneumoniae § P \< 0.0001 vs. 0--2 and 6--12 hours + C. pneumoniae §§ P \< 0.0001 vs. 0--3 hours + C. pneumoniae ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### MIP-2 release in C. pneumoniae infected type II cells ::: Type II cells + C. pneumoniae Macrophages + C. pneumoniae Type II cells -- 10% macrophages ---------- ------------------------------- ----------------------------- ---------------------------------- 0 min 160 ± 21 231 ± 22 137 30 min 210 ± 93 439 ± 49 166 1 hour 509 ± 76 1390 ± 329 370 2 hours 2405 ± 486 2715 ± 587 2134 3 hours 3367 ± 262 3600 ± 106 3007 6 hours 5274 ± 1095 11255 ± 2498 4149 12 hours 16179\* ± 4577 24511§ ± 5161 13728 Mean values are given in pg/ml ± SD of three independent experiments. \* P \< 0.0001 vs. 0--6 hours + C. pneumoniae § P \< 0.0001 vs. 0--6 hours + C. pneumoniae ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### BAPTA-AM, SN50 and parthenolide inhibited TNF-α and MIP-2 release in C. pneumoniae infected type II cells ::: BAPTA-AM SN50 SN50M parthenolide ------------- -------------- ------------ ---------------- -------------- ------------ -------------- TNF-α (3 h) 22.2\* ± 7.5 703.4 ± 91 155.5\* ± 30.7 122.6\* ± 41 705.8 ± 65 125\* ± 15.2 MIP-2 (6 h) 211\* ± 28 4516 ± 182 2725\* ± 413 2436\* ± 253 3648 ± 49 1602\* ± 24 Mean values are given in pg/ml ± SD of four independent experiments. Asterisk indicates a significant difference (P \< 0.0001) compared to C. pneumoniae infected type II cells (column 2) or C. pneumoniae infected type II in the presence of SN50M (column 5). ::: Rho-GTPases are involved in C. pneumoniae-mediated TLR4 protein expression -------------------------------------------------------------------------- To investigate whether C. pneumoniae-mediated TLR4 protein expression interferes with the F-actin-mediated Rho-GTPase activation, we examined the Rac1 and Rho protein expression after C. pneumoniae incubation for 30 min by CLSM. C. pneumoniae resulted in an increased expression of membrane-bound Rac1 and RhoA (Figure [7A](#F7){ref-type="fig"}). Furthermore, we tested the ability of mevastatin to down-regulate Rac1 and RhoA protein expression. As shown in Figure [7A](#F7){ref-type="fig"} mevastatin treatment reduced C. pneumoniae-mediated expression of both proteins. In non-infected type II cells, pre-treatment with mevastatin caused a destruction of the F-actin cytoskeleton (results not shown). We also observed that C. pneumoniae-induced stimulation of TLR4 can be decreased by mevastatin pre-treatment (Figure [7A](#F7){ref-type="fig"}). ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Mevastatin inhibited C. pneumoniae-mediated Rac1, RhoA and TLR4, expression. (A) Representative CLSM images of type II cells pre-treated and non-treated with mevastatin followed by C. pneumoniae incubation for 30 min. Detection with antibody for Rac1, RhoA and TLR4 followed by Alexa594-labeled secondary antibody (red). Labelled for F-actin with Alexa488-conjugated phalloidin (green) and nuclear DNA with DAPI (blue). The red labeled Rac1, RhoA and TLR4 appear yellow when co-localized with green labeled F-actin. The bar equals 10 μm. (B) Rac1, RhoA and TLR4 were visualized by Western blot technique and (C) their expression in membrane fractions was determined by densitometry in control cells (lane 1), C. pneumoniae incubated cells for 30 min in absence (lane 2) and in presence of mevastatin (lane 3). The values are means ± SD in arbitrary units of n = 3 experiments. Open bars, Rac1; solid bars, RhoA; hatched bars, TLR4. Asterisk indicates a significant difference compared to controls (1) and mevastatin (3) (P \< 0.05). ::: ![](1465-9921-6-51-7) ::: Next, the Rac1, RhoA and TLR4 expression was examined by Western blotting of the membrane fractions of cells incubated with C. pneumoniae for 30 min in absence and presence of mevastatin (Figure [7B](#F7){ref-type="fig"} and [7C](#F7){ref-type="fig"}). As shown, C. pneumoniae resulted in a greater expression of Rac1 (1.5-fold), RhoA (3-fold) and TLR4 (2.5-fold) versus control cells. Mevastatin reduced C. pneumoniae-mediated Rac1, RhoA and TLR4 expression to control values. Discussion ========== Toll like receptors are crucial in mediating cellular responses to microbial components. The extra cellular portion of the TLR is directly involved in the recognition of various pathogens. C. pneumoniae are known to enter and persist in humane type II pneumocytes in chronic but not acute pneumonia \[[@B2]\]. The role of TLR in the pulmonary host defense system is unknown. The present study on the interaction of type II pneumocytes with C. pneumoniae helps to understand the role of TLR4 within the innate immunity and inflammatory response mechanism of the lung. The study demonstrates that TLR2 and TLR4 mRNA is expressed constitutively in isolated rat type II cells. Incubation of alveolar type II cells with C. pneumoniae for 2 hours increased TLR4 but not TLR2 mRNA expression. Thus, TLR4 expression of rat type II cells is highly responsive to C. pneumoniae. TLR4 is localized in the cell membrane and cytoplasm of type II cells. Incubation of type II cells with C. pneumoniae increased the expression of TLR4 at the external cell membrane surface within 10--30 min. Whereas the expression of TLR4 reached a peak value 30 min after C. pneumoniae contact, the TLR4 expression was less prominent after 60--120 min of incubation. An incubation time of 180 min was again associated with a high TLR4 expression. The data suggests, that constitutively expressed TLR4 in type II pneumocytes is rapidly responding to C. pneumoniae contact and mediates cellular responses to C. pneumoniae components at a post-transcriptional level. Thus, it seems as if C. pneumoniae is captured by constitutively expressed receptors very rapidly. To enable further binding of C. pneumoniae components, newly synthesized TLR4 is required. The decreased TLR4 expression due to C. pneumoniae binding at the cellular surface and the subsequent up-regulation of TLR4 expression could represent a barrier to prevent uncontrolled stimulation. Various studies show that TLR4 is constitutively expressed in distinct human alveolar and bronchial epithelial cells. Lipopolysaccharide (LPS) stimulates the activation of TLR4 signaling in human alveolar type II-like (A549) and tracheo-bronchial (BEAS-2B, 16-HBE, and NT-1) epithelial cell lines \[[@B6]\]. Furthermore, LPS increased the expression of TLR2 and TLR4 mRNA and protein in human primary type II pneumocytes \[[@B5]\]. In normal human lung TLR2 mRNA and protein is mainly expressed in type II cells \[[@B4]\]. Whereas TLR4 is generally believed to be the receptor that mediates LPS responses \[[@B19]\], TLR2 mediates the inflammatory response to components of gram-positive bacteria and mycobacteria. However, TLR2 also mediates LPS-induced cellular signaling \[[@B20]\]. Different cells are activated by C. pneumoniae or cellular chlamydial components in a TLR2 and/or TLR4-dependent pathway \[[@B10],[@B11],[@B21]-[@B23]\]. Our data demonstrate that the TLR4 protein expression in rat type II cells is independent of chlamydial LPS. In addition, our results with heat-inactivated bacteria implicate that TLR4 expression is most likely mediated by a heat-sensitive chlamydial protein. C. pneumoniae and LPS free components of chlamydial heat shock protein 60 (cHSP60) were described to trigger inflammatory and cytokine responses in a TLR2-or TLR4-dependent way \[[@B10],[@B21],[@B23],[@B24]\]. It can be speculated that the TLR ligand is a protein component of the outer membrane of C. pneumoniae, with cHSP60 being a possible candidate. One regulator of TLR signaling is the NF-κB transcription factor, which controls the expression of many genes involved in the inflammatory response \[[@B3]\]. In vivo studies showed, that LPS induces NF-κB activation of type II cells \[[@B25],[@B26]\]. The interaction of TLR2 and TLR4 with HSP60 triggers a signal cascade of protein kinase p38 and JNK1/2, mitogen-activated protein kinases ERK1/2 and IκB kinase (IKK) \[[@B11],[@B27],[@B28]\], that subsequently leads to NF-κB activation. We have recently shown that C. pneumoniae contact with the type II cell membrane activates the NF-κB pathway \[[@B12]\]. In the present study blockade of the TLR4 receptor at the cell membrane level reduced the C. pneumoniae-induced NF-κB activation in type II cells. Thus, results are in accordance with studies on mouse macrophages and human endothelial cells showing, that activation of NF-κB by chlamydial heat shock protein 60 (cHSP60) requires TLR4 \[[@B10]\]. We demonstrate that C. pneumoniae-mediated TLR4 mRNA and protein expression is attenuated by the calcium chelator BAPTA-AM and by inhibition of nuclear factor NF-κB with parthenolide or SN50. LPS-mediated TLR2 mRNA up-regulation in murine alveolar macrophages was attenuated by inhibition of NF-κB with sulfasalazine or SN50 \[[@B29]\]. BAPTA-AM, SN50 and parthenolide inhibited C. pneumoniae-mediated TNF-α and MIP-2 release in rat type II pneumocytes. It can be concluded that the calcium-mediated NF-κB activation is required for the transcriptional induction of TNF-α and MIP-2. Our findings support the idea that contact of C. pneumoniae with TLR4 triggers signal transduction and cytokine release. Contact of C. pneumoniae with TLR2 and TLR4 bone marrow-derived murine dendritic cells (BMDDC) leads to the translocation of NF-κB as well as secretion of TNF-α \[[@B22],[@B24]\]. HSP60 from C. pneumoniae causes a TLR2-and TLR4-dependent increase in serum levels of MIP-2 in mice \[[@B11]\]. Intrapulmonary administration of Haemophilus influenzae in the mouse leads to TLR4-mediated TNF-α and MIP-2 expression by an activation of the NF-κB pathway in epithelial cells of the conducting airways. This suggests that the airway epithelia contribute to recognition of H. influenzae infection and induction of the innate immune response \[[@B30]\]. The role of alveolar type II cells in the regulation of lung inflammation is less clear. In vitro studies suggest that LPS-treatment of primary cultured type II cells from adult rat lungs leads to a TNF-α release, which initiates the synthesis and release of chemokines like MIP-2. Subsequently, the recruitment of inflammatory cells into the lung is promoted. These data indicate that a direct bacterial stimulation of alveolar epithelial cells can initiate the local host defense \[[@B31]-[@B33]\]. Recently, we demonstrated that NF-κB activation is involved in the calcium-mediated reorganization of the F-actin cytoskeleton after C. pneumoniae contact \[[@B12]\]. LPS induces the depolymerization of the cytoskeleton by mediating the TNF-α production in alveolar type II cells and may thus facilitate the migration of inflammatory cells \[[@B32]\]. Small GTPases of the Rho family (Rac and Rho) play an important role in several signal networks including actin cytoskeleton dynamics, transcriptional regulation, and membrane trafficking \[[@B34]\]. RhoA controls the assembly of actin and myosin stress fibers. Rac1 regulates the formation of membrane ruffles \[[@B35]\]. Data of the present study demonstrate that C. pneumoniae-induced TLR4 activation is Rac1 and RhoA protein mediated. Mevastatin, a specific inhibitor of the Rho-GTPase family, reduced the C. pneumoniae-mediated activation of TLR4, Rac1 and RhoA protein. In dendritic cells, TLR activation stimulates the antigen capture via F-actin remodeling \[[@B36]\]. Mevastatin (compactin) blocked the invasion of C. pneumoniae in Hep2 cells \[[@B37]\]. C. pneumoniae infection of vascular smooth muscle cells increased Rac1 and RhoA proteins and NF-κB expression that was suppressed by statin \[[@B38]\]. These results suggest a signaling cascade by which C. pneumoniae contact with the cell membrane leads to the activation of Rho family GTPases. The subsequent changes in actin cytoskeletal structure are essential for TLR4 expression and the activation of the NF-κB signaling pathway. The latter controls the expression of many genes involved in the innate immune system and adaptive inflammatory immune response. We have recently shown, that C. pneumoniae changes the cytoskeleton of type II pneumocytes by NF-κB activation \[[@B12]\]. The current work demonstrates that NF-κB activation induces an inflammatory cascade that could support the local host defense. In healthy lungs, the pulmonary surfactant represents a barrier that can prevent the invasion of bacteria. However, C. pneumoniae inhibits the surfactant metabolism of isolated type II cells \[[@B18]\]. Thus, C. pneumoniae induce an immune response that prevents the invasion of C. pneumoniae into type II cells. Simultaneously, the impaired surfactant metabolism will enable bacteria to enter type II cells. The resulting net effect remains to be studied in vivo. Clinical data show, that no intracellular C. pneumoniae are found in patients with acute pneumonia, but in patients with chronic pneumonia \[[@B2]\]. It could be assumed that C. pneumoniae-induced alterations of type II cell cytoskeleton and surfactant metabolism will impair the host defense. Subsequently, more type II cells are invaded by C. pneumoniae, causing the clinical picture of chronic pneumonia. Conclusion ========== The TLR4 protein expression in rat type II cells is most likely stimulated by a heat-sensitive C. pneumoniae protein. The induction of a rapid Ca^2+^-mediated NF-κB activity is required for TLR4 expression and TNF-α and MIP-2release. A Rac1 and RhoA protein-dependent mechanism seems to be involved. These results implicate that TLR4 plays an important role in the innate pulmonary immune defense mechanism and in the inflammatory response of the alveolus. Authors\' contributions ======================= HW performed immunocytochemistry and CLSM, conceived and co-ordinated the study, acquired the data and wrote the manuscript. CS performed the ELISA analysis. PK performed the mRNA analysis. ER was involved in CLSM. MM participated in the C. pneumoniae strain TW183 culturing. MR was involved in the revision of the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ We gratefully thank Ruth Herrmann, Helga Kemmer, Silke Wilitzky (Clinic of Neonatology, Campus Charité Mitte, University Children\'s Hospital, Humboldt-University Berlin), Petra Klein (Institute of Biology, Department of Membrane Physiology, Humboldt University Berlin, Germany) and Anke Hellberg (Institute of Medical Microbiology and Hygiene, University of Lübeck, Germany) for technical assistance. This study was supported by Grants Wi 2074/1-1 from the Deutsche Forschungsgemeinschaft.	PubMed Central	2024-06-05T03:55:59.561653	2005-6-3	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180473/", "journal": "Respir Res. 2005 Jun 3; 6(1):51", "authors": [ { "first": "Heide", "last": "Wissel" }, { "first": "Christian", "last": "Schulz" }, { "first": "Petra", "last": "Koehne" }, { "first": "Ekkehard", "last": "Richter" }, { "first": "Matthias", "last": "Maass" }, { "first": "Mario", "last": "Rüdiger" } ] }
PMC1180474	Background ========== It is recognized that genetic factors influence lung function \[[@B1],[@B2]\]. The identification of genetic variants that determine either lung function development or decline is particularly important for diseases in which low lung function is a feature, such as chronic obstructive pulmonary disease and asthma, since this provides insight into the pathophysiology of these disorders. This may also be relevant to non-pulmonary disorders that have been associated with low lung function, such as cardiovascular disorders \[[@B3],[@B4]\] and diabetes \[[@B5]\], in which these genes may control systemic mechanisms (e.g. inflammation) that contribute to both low lung function and disease development. Signal transducer and activator of transcription 3 (STAT3) is a member of a protein family of transcription factors, which was discovered in the course of studies of interferon-induced intracellular signal transduction \[[@B6]\]. These proteins are latent in the cytoplasm and become activated through tyrosine phosphorylation which typically occurs through cytokine receptor associated kinases (the Janus kinase-signal transducer or JAKs). The JAK-STAT pathway transmits information received from extracellular polypeptide signals, through membrane receptors, directly to target gene promoters in the nucleus, providing a mechanism for transcriptional regulation without second messengers. The gene, STAT3, is induced by a wide-array of cytokines, including interleukin (IL)-6, IL-10, and IL-13, and has been implicated in the regulation of cell growth, inflammation, immune tolerance and early embryonic development. Recently, STAT3was implicated in asthma pathogenesis in a study that showed that STAT3-dependent pathways induced by IL-13 in lung myofibroblasts were inhibited by the administration of the inhaled corticosteroid, fluticasone \[[@B7]\]. This suggests a role in airway inflammation and remodeling in asthmatics, which may affect lung function level. In a previous study of the pharmacogenetics of asthma treatment, STAT3was one of the candidate genes that we screened for association with response to cortocosteroid treatment \[[@B8]\]. Single nucleotide polymorphisms (SNPs) in STAT3were genotyped and tested in a screening dataset from an adult asthma clinical trial. No effect of STAT3SNPs on asthma drug response was seen in that study. However, the polymorphisms affected baseline lung function. In this report, we present our analysis of the association of STAT3SNPs with lung function in adults with asthma, and replicate our findings in a cohort of children with asthma. Methods ======= Populations and Study Samples ----------------------------- We used information from two asthma clinical trials, as previously reported \[[@B8]\]. All patients or their legal guardians consented to the study protocol and ancillary genetic testing. The Adult Study was a multicenter 8-week randomized clinical trial comparing the effect of once-daily high-dose inhaled flunisolide therapy with that of standard inhaled corticosteroid therapy (i.e. high vs standard dose inhaled corticosteroid therapy) among moderate to severe adult asthmatics \[[@B8]\]. Inclusion criteria were a history of asthma, ≥ 12% improvement in FEV~1~with albuterol, and use of inhaled steroids at randomization. Exclusion criteria were non-asthma pulmonary disease, smoking (≥ 10 pack-years), and recent asthma exacerbations requiring systemic steroids. Subjects were phoned weekly and had spirometry at 4 and 8 weeks. For this analysis, we included only the 401 Caucasian participants. The Childhood Asthma Management Program (CAMP) is a multicenter, randomized, double-blinded clinical trial testing the safety and efficacy of inhaled budesonide vs. nedocromil vs. placebo over a mean of 4.3 years. Trial design and methodology for CAMP have been published \[[@B9],[@B10]\]. CAMP enrolled 1,041 children ages 5 to 12 years with mild to moderate asthma. Entry criteria included asthma symptoms and / or medication use for ≥ 6-months in the previous year and airway responsiveness with a provocative concentration of methacholine causing a 20% reduction in FEV~1~(PC~20~) ≤ 12.5 mg/ml. Data for 652 Caucasian children were included in this analysis. Phenotypes ---------- The primary phenotype of interest in both cohorts was baseline, pre-bronchodilator forced expiratory volume in one second, as a percentage of predicted (PPFEV~1~). In the Adult Study, baseline PPFEV~1~was measured after an open-label 4-week period that demonstrated stability on the study drug or on standard inhaled therapy. Spirometry was then performed at monthly intervals, for a total of three spirometric measurements. In CAMP, baseline spirometry was performed at randomization, after a 28-day period during which only as-needed albuterol was allowed. Follow-up spirometry was perfomed at 2, 4, 12, 16, 24, 28, 36, 40, and 48 months. In addition to the analysis of the baseline PPFEV~1~, a repeated measures analysis was performed in both cohorts, making use of the longitudinal follow-up for each subject. In CAMP, we included information on parental smoking obtained from the baseline questionnaire. SNP Selection and Genotyping ---------------------------- SNPs were selected from two sources, public databases and genomic DNA sequencing performed at the Whitehead Institute. Three SNPs were discovered as a result of the sequencing effort: G3363a3, G3363a4, and G3363a16. These three SNPs have been submitted to the public database and correspond precisely to rs8075442, rs2293152, and rs2306581, respectively (dbSNP: <http://www.ncbi.nlm.nih.gov/SNP>). Five additional SNPs were chosen from public databases for genotyping, with the overall goal of having, on average, at least one SNP every 10 kilobases. Two of these SNPs -- rs1803125 (exonic) and rs744284 (promoter) -- were found to be monomorphic in the Adult Study subjects, and were not subsequently genotyped in the CAMP cohort. Three additional SNPs were successfully genotyped in both cohorts: rs1026916, rs1905340, and rs957971. These are all intronic SNPs (Figure [1](#F1){ref-type="fig"}\[[@B11],[@B12]\]), and flanking sequences are given in Table [1](#T1){ref-type="table"}. Linkage disequilibrium (LD) between each pair of SNPs was calculated and plotted using the LDPlotter tool <http://innateimmunity.net/IIPGA2/Bioinformatics/>, and expressed as the r^2^LD statistic \[[@B13]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Flanking sequences for STAT3SNPs ::: ------- ----------- -------------------------------------------------------------------- STAT3 G3363a16 GGGAAAATGAGATCAGGAGATAAAG \[G/T\]GGCACCCTTTGGTCTTGTAAAGCCTTTTTTA STAT3 G3363a3 ACAGACATCATTTGAACTAGAGACTCT \[G/A\]TCTTTATTCAGAGATCTTCATTTTGTGGAC STAT3 G3363a4 TCCCCTTCACAAAGGGCCTCTGGCTGC \[C/G\]GGAGAGGGCTAGGGAGAGCCTCACAG STAT3 rs1026916 AGGAAAAAGTTTAACCCAAAGACTGT \[A/G\]TGGATCTTCTCTACCCTACATCTCCAATCT STAT3 rs1905340 TATTTGAGAATCTAAGAAAGTAGATCA \[A/C\]ACTAAATATTGATATGCAGACACTAAAATC STAT3 rs957971 TGTTATATGAAGTGAATTAACCTCCTAT \[C/G\]GTACTTCAGTTTTCTCTATGCTAAAAGTGT ------- ----------- -------------------------------------------------------------------- ::: ::: {#F1 .fig} Figure 1 ::: {.caption} ###### *Positions of the STAT3SNPs genotyped in both the Adult Study and the CAMP cohort.From UCSC Genome Browser <http://genome.ucsc.edu>, May 2004 Human Reference Sequence Assembly\[11,12\]. UCSC webpage was accessed on March 18, 2005. ::: ![](1465-9921-6-52-1) ::: SNPs were genotyped via a SEQUENOM MassARRAY MALDI-TOF mass spectrometer (Sequenom, San Diego, CA) for analysis of unlabeled single-base extension minisequencing reactions with a semiautomated primer design program (SpectroDESIGNER, Sequenom). Our protocol implemented the very short extension method \[[@B14]\], whereby sequencing products are extended by only one base for three of the four nucleotides and by several additional bases for the fourth nucleotide (representing one of the alleles for a given SNP), permitting clearly delineated mass separation of the two allelic variants at a given locus. Statistical Analysis -------------------- Single SNP association analyses were performed with SAS statistical software (SAS Institute, Inc., Cary, NC). Univariate associations between SNPs and the phenotype of interest (PPFEV~1~) were tested by univariate linear regression, as implemented in Proc Reg in SAS. Multivariable linear regression models, were used to control for potential confounders. In these models, the genotype for each SNP was coded as three-level categorical variables (additive genetic model) or as dummy-coded variables. Hardy-Weinberg equilibrium for each SNP was tested using the chi-square goodness-of-fit test as implemented in the ALLELE Procedure in SAS. The repeated measures analyses were carried out using a mixed linear model as implemented in the MIXED Procedure in SAS. A mixed linear model is a generalization of the standard linear model where the data are permitted to exhibit correlation and nonconstant variability, thereby providing the flexibility of modeling the variances and covariances of the data, in addition to the means. The covariance structure for the lung function data was specified using an unstructured model, which provided the best fit for the data after testing other covariance matrices (compound symmetry, spatial exponential, autoregressive, and autoregressive-heterogeneous). All models adjusted for time (in weeks) and contained a SNP × time interaction term. However, since the time interaction terms were not significant in any of the models, they were dropped from the final models presented in the results. Multiple testing for the single SNP association analyses was addressed by controlling the false discovery rate (FDR) using the method of Benjamini and Liu\[[@B15],[@B16]\] (FDR tool available at <https://innateimmunity.net/IIPGA2/Bioinformatics/>. Control of the FDR was set at the 0.05 threshold. Haplotype associations were explored with score tests that account for linkage phase ambiguity \[[@B17]\]. The score tests, derived from generalized linear models, are used for global tests of association, as well as haplotype-specific tests. The haplo.stats program implements the methods of Schaid et al, and was used for these analyses. Haplotypes were imputed and frequencies estimated using the modified EM algorithm estimation facility in haplo.stats. Analyses were run with and without adjustment for nongenetic factors. We modified the method to include data from individuals with partially missing marker information. The minimum haplotype frequency was set at 2.5%. As previously reported \[[@B8]\] in an analysis of 59 SNPs across the genome, we found no evidence for population stratification in either population. Results ======= Baseline characteristics and genotype frequencies of the six SNPs in both cohorts are shown in Table [2](#T2){ref-type="table"}. G3363a3 in the Adult Study was the only SNP out of Hardy-Weinberg equilibrium, due to one rare individual who was homozygous for the minor allele. Genotype frequencies were also similar for both cohorts. Figure [2](#F2){ref-type="fig"} plots the LD patterns among the six SNPs, with the corresponding r^2^values in Table [3](#T3){ref-type="table"}. The LD patterns are similar in the two cohorts. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Baseline characteristics and genotype frequencies for the 6 SNPs in both cohorts. ::: Adult Study N = 401* CAMP N = 652 ---------------------------------------------------------- --------- ------------------------- ------------- ------------------ ------------- Age, mean(sd) 39.2 (13.7) 8.9 (2.1) Gender, male, n(%) 171 (42.8) 393 (60.3) FEV~1~(%predicted), mean (sd) 71.7 (12.8) 94.3 (14.2) Number of positive skin tests, median (minimum, maximum) \-\-- 4 (0, 18) Eosinophil count, geometric mean (sd) \-\-- 5.86 (2.06) IgE, geometric mean (sd) \-\-- 404.0 (4.9) Paternal asthma, n(%) \-\-- 125 (19.2) Maternal asthma, n(%) \-\-- 164 (25.2) Maternal cigarette smoking during pregnancy, n(%) \-\-- 88 (13.5) SNP genotypes, n(%) HWE p-value HWE p-value G3363a16 GG 145 (36.2) 245 (37.6) GT 153 (38.0) 0.6 291 (44.6) 0.7 TT 46 (11.5) 80 (12.3) missing 57 (14.2) 36 (5.5) G3363a3 GG 342 (85.2) 582 (89.3) GA 1 (0.2) 0.004 2 (0.3) 1.0 AA 1 (0.2) 0 missing 57 (14.2) 68 (10.4) G3363a4 GG 137 (34.0) 218 (33.4) GC 175 (43.8) 0.3 291 (44.6) 0.5 CC 70 (17.5) 85 (13.0) missing 19 (4.8) 58 (8.9) rs1026916 GG 164 (41.0) 227 (34.8) GA 181 (45.0) 0.8 233 (35.7) 0.2 AA 52 (13.0) 74 (11.3) missing 4 (1.0) 118 (18.1) rs1905340 CC 203 (50.8) 254 (39.0) CA 159 (39.5) 1.0 191 (29.3) 0.6 AA 30 (7.5) 41 (6.3) missing 166 (25.5) rs957971 CC 156 (39.0) 242 (37.1) CG 172 (42.8) 0.6 260 (39.9) 1.0 GG 54 (13.5) 70 (10.7) missing 80 (12.3) ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Linkage disequilibrium (r^2^) among the six STAT3SNPs\. ::: Adult Study* G3363a16 G3363a3 G3363a4 rs1026916 rs1905340 rs957971 ----------------- -------------- ------------- ------------- --------------- --------------- -------------- G3363a16 0.0089 0.1662 0.9804 0.6985 0.9936 G3363a3 0.0016 0.0003 0.0076 0.0018 0.0017 G3363a4 0.1668 0.0012 0.1418 0.1051 0.1666 rs1026916 0.9754 0.0037 0.1627 0.6693 0.9718 rs1905340 0.6563 0.0002 0.0888 0.6644 0.6910 rs957971 0.9734 0.0036 0.1615 0.9854 0.6536 \r2 between SNPs for the Adult Study are above the diagonal and for the CAMP Study are below the diagonal ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Linkage disequilibrium (LD) plots among the six STAT3SNPs for both cohorts.LD is expressed as the r^2^statistic. ::: ![](1465-9921-6-52-2) ::: Table [4](#T4){ref-type="table"} and Figure [3](#F3){ref-type="fig"} show the results of linear regression models for baseline PPFEV~1~. In the Adult Study, three SNPs showed an effect on lung function. For G3363a16, subjects who were homozygous TT had PPFEV~1~levels that were 6.85% lower than levels for GG homozygotes. Likewise, for rs1026916 and rs957971, subjects who were homozygous for the minor allele had lower PPFEV~1~values than did subjects who were homozygous for the major allele. These results remained significant after controlling for the FDR (i.e. the p-values associated with each SNP were smaller than the FDR threshold p-value, thus we reject the null hypothesis of no significant association). In the CAMP trial, only rs1026916 was significantly associated with baseline PPFEV~1~, however, this did not remain statistically significant after controlling for the FDR (i.e. the p-value was greater than the FDR threshold p-value, thus we are unable to reject the null hypothesis of no significant association). Although not statistically significant, the direction of the changes in lung function associated with variation in G3363a16 and rs957971 paralleled those of the Adult Study. Similar results were obtained when we used raw pre-BD FEV~1~measures, adjusted for age, sex, and height. Additional control for exposure to maternal smoking in utero or post-natal maternal or paternal smoking did not change the results. Interactions between individual SNPs and parental smoking variables were not significant. There was no significant association of any of the SNPs with forced vital capacity. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Results of linear regression models for baseline PPFEV~1~\ ::: [Adult Study]{.underline} [CAMP]{.underline} ----------------- ------------------------------- ------------------------------------- ------------------------------- -------------- ------------------------------------- ------------------------------- Stat3 SNPs β (se)† Additive genetic model p-value‡ FDR threshold (decision§) β (se)† Additive genetic model p-value‡ FDR threshold (decision§) G3363a16 Intercept (GG) 74.03 (1.06) 95.67 (0.92) GT -3.35 (1.48) 0.0007 0.007 -1.93 (1.25) 0.1 0.02 TT -6.85 (2.15) (reject) -2.08 (1.84) (accept) G3363a4 Intercept (GG) 70.76 (1.10) 94.19 (0.98) GC 1.18 (1.47) 0.3 0.05 1.06 (1.29) 0.6 0.03 CC 2.04 (1.89) (accept) -2.04 (1.86) (accept) rs1026916 Intercept (GG) 73.59 (0.93) 96.21 (0.95) GA -2.90 (1.38) 0.004 0.01 -2.21 (1.33) 0.02 0.007 AA -5.22 (2.03) (reject) -4.11 (1.90) (accept) rs1905340 Intercept (CC) 73.04 (0.89) 95.52 (0.89) CA -3.00 (1.34) 0.04 0.03 -1.18 (1.36) 0.8 0.05 AA -2.77 (2.49) (accept) 0.70 (2.39) (accept) rs957971 Intercept (CC) 73.48 (1.02) 95.56 (0.01) CG -2.10 (1.41) 0.007 0.02 -2.07 (1.26) 0.08 0.01 GG -5.34 (2.00) (reject) -2.64 (1.91) (accept) \* Results are from individual linear regression models. † β \'s and p-values are from linear regression models with dummy-coded genotype categories. ‡ p-values refer to the categorical genotype variable (0, 1, 2), for the number of minor alleles present. §Refers to the decision of whether to accept or reject the null hypothesis of no genotype effect. Please see text for further details. ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### *The association between STAT3SNPs and FEV~1~. Mean (± sd) percent predicted FEV~1~in the Adult Study plotted against genotype for each SNP. Additive genetic models were statistically significant for each SNP: p = 0.0007, 0.0043, and 0.007, respectively for G3363a16, rs1026916, and rs957971. ::: ![](1465-9921-6-52-3) ::: We then took the SNPs that were significant in the baseline analysis and performed a repeated measures analysis in each cohort, to take advantage of the multiple measures of PPFEV~1~. In the Adult Study, all 401 participants had complete data on PPFEV~1~at all three time points. In the repeated measures analysis, results were similar to those of the analyses on baseline PPFEV~1~(Table [5](#T5){ref-type="table"}), although only the results for G3363a16 and rs957971 remained significant after control for FDR. Since there were no differences in PPFEV~1~levels at baseline or at any time point between the two treatment groups, we did not additionally control for treatment group in the multivariable longitudinal models. For CAMP, 6358 observations were available for analysis: 557 (85.3%) children had complete information for the 10 time points, 56 (8.6%) had information for 9 time points, 18 (2.8%) had information for 8 time points, and 22 (3.4%) had information for 3 to 7 time points. In CAMP, rs1026916 was associated with lung function over the 4 years of observation, and stratified analyses showed similar effects of the SNP across treatment groups. Although treatment group had a significant effect on PPFEV~1~over time, a formal test for interaction between SNP and treatment group was not significant. Thus, treatment group was left as a covariate in the final models. In both cohorts, the SNP × time interaction terms were not significant, meaning that the effects of the individual SNPs did not vary over the time of each study. ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Effects of Stat3 SNPs on repeated measures of PPFEV~1~over time\ ::: [Adult Study]{.underline} [CAMP]{.underline} ---------------- ------------------------------- ------------------------------------- ------------------------------- -------------- ------------------------------------- Stat3 SNPs β (se) Additive genetic model p-value‡ FDR threshold (decision§) β (se) Additive genetic model p-value‡ G3363a16 Intercept (GG) 73.85 (1.21) GT -2.72 (1.69) 0.008 0.009 \-\-- \-\-- TT -6.33 (2.46) (reject) rs1026916 Intercept (GG) 73.46 (1.13) 97.01 (0.93) GA -2.25 (1.56) 0.028 0.017 -0.19 (0.99) 0.02 AA -4.74 (2.30) (accept) -4.24 (1.49) rs957971 Intercept (CC) 73.36 (1.16) CG -1.51 (1.60) 0.036 0.038 \-\-- \-\-- GG -5.04 (2.29) (reject) \* β\'s obtained from mixed models of repeated measurements of percent predicted FEV~1~over time. All models are adjusted for time in weeks, with a maximum of 3 observations for the Adult Study, and 10 observations for the CAMP study. The model for CAMP is adjusted additionally for treatment group. §Refers to the decision of whether to accept or reject the null hypothesis of no genotype effect. Please see text for further details. ::: Haplotype analyses on baseline PPFEV~1~were performed for each cohort. Table [6](#T6){ref-type="table"} presents the results for the Adult Study. There were five haplotypes that had frequencies above 2.5%. The global statistic was significant at p = 0.02. Haplotype 5, which is comprised of the major alleles for G3363a16, rs957971, and rs1026916, was positively associated with PPFEV~1~, meaning that this haplotype was associated with higher PPFEV~1~values. On the other hand, haplotype 1, which contained the minor alleles of these three SNPs, was negatively associated with PPFEV~1~. These results were consistent with those of the single SNP analysis. Haplotype analysis in CAMP was also consistent with the single SNP analyses, but did not reach statistical significance (data not shown). ::: {#T6 .table-wrap} Table 6 ::: {.caption} ###### Association of haplotypes in the STAT3gene and baseline PPFEV~1~in the Adult Study. ::: Haplotypes Frequency Haplotype-Specific Score Statistic p-value ---------------- ----------------------- ---------------------------------------- ---------------------------- ---------------------------- ------------------------- ----------------------------- --------- ------- ------ [G3363a4]{.underline} [G3363a3]{.underline} [G3363a16]{.underline} [rs957971]{.underline} [rs1905340]{.underline} [rs1026916]{.underline} (1) C G T G A A 0.04501 -1.91 0.05 (2) G G T G A A 0.23441 -1.73 0.08 (3) G G T G C A 0.06895 -1.60 0.1 (4) G G G C C G 0.27914 0.96 0.3 (5) C G G C C G 0.35557 2.58 0.01 Global Score Statistics: Global Statistic: 13.3, df = 5 Global p-value = 0.02 ::: Discussion ========== This is the first report of an association between SNPs in the STAT3gene and lung function in human populations. In this study, we show an association between SNPs in the STAT3gene and FEV~1~among asthmatics. These results are reasonably robust and are consistent in a cohort of adult asthmatics and a cohort of childhood asthmatics. Although only one SNP was significant in both cohorts, the direction of the effect of each individual SNPs was similar in the two cohorts. These effects were seen when we analyzed baseline FEV~1~and the repeated measures of FEV~1~over 8 weeks in the adult cohort and over 4 years in the childhood asthma cohort. In the cohort of childhood asthmatics, these effects were independent of parental smoking and asthma treatment group. STAT proteins comprise a family of transcription factors latent in the cytoplasm, that are activated by a series of extracellular signaling proteins such as cytokine, growth factors, and hormones that bind to specific cell-surface receptors. The resulting signal transduction pathways permit them to play different roles in normal physiological cell processes, such as differentiation, proliferation, apoptosis, and angiogenesis \[[@B18],[@B19]\]. Whereas other members of this gene family have generally demonstrated specificity in individual signaling pathways, STAT3is deployed in various, sometimes disparate, physiological processes \[[@B6]\], including cell growth and differentiation \[[@B20]\], apoptosis \[[@B21]\], and anti-inflammatory processes mediated by IL-10 \[[@B22]\], to name a few. Additionally, while the different functions of the members of this gene family have been elucidated via targeted gene ablation, ablation of STAT3leads to embryonic lethality in transgenic mice \[[@B23]\], underscoring its importance in embryogenesis. In the lung, the function of STAT3has not been fully elucidated. However, STAT3appears to play a role in the regulation of surfactant \[[@B24],[@B25]\], and in the inflammatory response in acute lung injury \[[@B26],[@B27]\]. Additionally, STAT3is an important mediator in the pro-inflammatory effects of the Th2 cytokine IL-13 on lung myofibroblasts \[[@B7],[@B28]\]. It is plausible that our results are due to an effect of the STAT3gene during the embryonic stage of lung development. However, the results in the CAMP cohort show that this effect, if present, is likely a small one. On the other hand, STAT3had stronger effects on lung function in the adult asthmatics, and a potential explanation for our findings is that STAT3interacts with pro-inflammatory environmental stimuli, such as tobacco smoke, to affect FEV~1~level. It is known that at least three factors determine lung function at a particular point in adult life: (1) the maximally attained level of lung function; (2) the onset of decline of lung function (or alternatively, the duration of the plateau phase); and (3) the rate of decline of lung function \[[@B29]\]. Whether STAT3affects only a particular phase of lung growth or decline, or affects all phases remains to be seen. Furthermore, it needs to be determined whether this effect of STAT3on FEV~1~level is unique to asthmatics or also applies to non-asthmatics. Because the effects of STAT3were stronger in the Adult Study, we additionally hypothesized that STAT3may interact with environmental exposures to cause a decrement in lung function. Since cigarette smoking is established as the major environmental risk factor for low lung function\[[@B30]\], we hypothesized that exposure to cigarette smoke (either personal smoking or environmental tobacco smoke) could potentially interact with STAT3. We were unable to test the interaction between STAT3and smoking in the Adult Study, because participants were non-smokers. In CAMP, we did not see a significant interaction effect between the individual SNPs and parental smoking variables (in utero smoke exposure, maternal smoking, paternal smoking), and too few children smoked to permit any meaningful interaction analyses. It is also possible that other environmental exposures that we did not measure could be interacting with STAT3. A limitation of our study is the lack of complete sequence information on the gene. The sequencing efforts focused only on the exons, thus limiting our knowledge of the full linkage disequilibrium pattern of the gene. Furthermore, since all the SNPs we successfully genotyped were in intronic regions of the gene, it is likely that these are not responsible for the association but are rather in linkage disequilibrium with the functional variant of this gene, or with variants in another gene. Studying other populations, such as general population samples or occupational cohorts with exposure information, may help elucidate the effects of this gene on FEV~1~. Additionally, more functional studies of SNPs in STAT3are needed to elucidate its role in lung function development. Airway hyperresponsiveness was not assessed in the Adult Study. However, the subjects had a physician diagnosis of moderate to severe asthma (as evidenced by the levels of lung function), were on inhaled steroids at baseline, and had a significant bronchodilator response to albuterol. The combination of a physician diagnosis and bronchodilator response is a reasonable definition of asthma in genetic studies\[[@B31]\]. Furthermore, significant smoking and non-asthma respiratory disorders were excluded. Additionally, we did not have information on allergy outcomes in the Adult Study, either. In CAMP, we performed additional analyses, however, there were no associations between any of the STAT3SNPs or either serum IgE level or skin test reactivity. We controlled for multiple testing by controlling the false discovery rate. An additional strategy we took to minimize the effect of multiple testing is by performing a screening analysis in the Adult Study, then performing a replicate analysis in CAMP. Population stratification is another potential concern \[[@B32]\]; thus, we included only Caucasian subjects in this analysis. Furthermore, in previous testing utilizing a panel of 59 random markers, we found no evidence for stratification in the Caucasian subjects in either of these two cohorts \[[@B8]\]. Conclusion ========== We have shown that polymorphisms in STAT3are associated with FEV~1~in asthmatics. We show these effects both in a cohort of adult asthmatics and in a cohort of childhood asthmatics. In both cohorts, we excluded gross population stratification by testing with a panel of random markers. The precise mechanism for the effects of this gene on FEV~1~remains unknown. However, while we see an association between SNPs in this gene and FEV~1~in young asthmatics, the effects were stronger in the adult asthmatics, suggesting a role of STAT3in chronic inflammatory pathways that may have an impact on lung growth and decline. Authors\' contributions ======================= AAL participated in the conceptualization of the analysis, designed the analysis, performed the analyses and interpretation of results, and drafted the manuscript. KGT participated in the selection and genotyping of SNPs, conception and design of the study, acquisition of the data, and drafting and critically revising the manuscript. SL participated in the statistical analysis of the data and in critically revising the manuscript. RL participated in the conception and design of the analysis, and in critically revising the manuscript. BGR participated in the preparation of the data for analysis and in critically revising the manuscript. SG participated in the selection and genotyping of SNPs and in critically revising the manuscript. ESS participated in the design of the analysis and in critically revising the manuscript. STW conceived of the study, participated in its design and its coordination, participated in acquisition of the data, and participated in critically revising the manuscript. Grant Support ============= This work was supported by: U01 HL065899 -- The Pharmacogenetics of Asthma Treatment; the Childhood Asthma Management Program (CAMP)by contracts N01-HR-16044, 16045, 16046, 16047, 16048, 16049, 16050, 16051, and 16052; and the CAMP Genetics Ancillary Studyby P01 HL067664, all awarded by the NHLBI. Acknowledgements ================ The authors wish to acknowledge Dr. David Altshuler and Dr. Eric Lander for their assistance with the generation of sequence data utilized in the genotyping assays. We thank all the CAMP families for their enthusiastic participation in the CAMP Genetics Ancillary Study. We also acknowledge the Childhood Asthma Management Program (CAMP) investigators and research team, supported by the National Heart, Lung, and Blood Institute (NHLBI) at the National Institutes of Health (NIH), Bethesda, Maryland USA, for collection of CAMP Genetics Ancillary Studydata. All work on data from the CAMP Genetics Ancillary Studywas conducted at the Channing Laboratory, Brigham and Women\'s Hospital under appropriate CAMP policies and human subjects protections.	PubMed Central	2024-06-05T03:55:59.565462	2005-6-3	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180474/", "journal": "Respir Res. 2005 Jun 3; 6(1):52", "authors": [ { "first": "Augusto A", "last": "Litonjua" }, { "first": "Kelan G", "last": "Tantisira" }, { "first": "Stephen", "last": "Lake" }, { "first": "Ross", "last": "Lazarus" }, { "first": "Brent G", "last": "Richter" }, { "first": "Stacey", "last": "Gabriel" }, { "first": "Eric S", "last": "Silverman" }, { "first": "Scott T", "last": "Weiss" } ] }
PMC1180475	Background ========== We have recently demonstrated that the ability of deep inspirations to dilate constricted airways is impaired in subjects with COPD \[[@B1]\]. We have suggested that the lack of deep inspiration-induced bronchodilation may be one of the major factors that contribute to persistent airway narrowing in chronic obstructive pulmonary diseases. However, the mechanism accounting for the reduction in the bronchodilatory effect of deep inspiration in COPD has not yet been elucidated. When a deep inspiration takes place, radial traction is applied to the outer airway walls by virtue of the forces of interdependence between the airways and the surrounding parenchyma \[[@B2]\], which are sustained by the connective tissue of the lungs. As a consequence, lung inflation produces transient airway distension. If the airways are constricted, the stretch imposed by airway distension may produce bronchodilation \[[@B3],[@B4]\]. The loss of the bronchodilatory effect of lung inflation in COPD may result from factors that unlink the parenchyma from the airways. COPD is accompanied by destructive changes of alveolar walls and consequent reduction in the number of alveolar attachments on the airways \[[@B5]-[@B7]\]. We postulated that, because of the alveolar wall destruction, mechanical decoupling between airways and parenchyma results in diminished airway wall stretch, thus impairing the postulated primary step in the mechanism of bronchodilation by deep inspiration. The current study showed that the reduction in alveolar attachments on the airways correlates with the reduction in the bronchodilatory ability of deep inspiration. Methods ======= In order to obtain lung tissue for morphometric and correlative analyses, we enrolled subjects undergoing lobar resection for peripheral pulmonary nodule, with the assumption that a significant proportion would have been smokers and would presumably show evidence of reduced integrity of alveolar attachments \[[@B6]\]. Subjects were recruited from the Unit of Thoracic Surgery, \"V. Cervello\" Palermo Hospital, Italy. After giving their informed consent, subjects were referred to the Istituto di Medicina Generale e Pneumologia, Palermo University, for potential participation in this protocol. Exclusion criteria for participation in the study were a large involvement of the lung and/or mediastinal organs by the tumor, a history of myocardial infarction, congestive heart failure, arrhythmia, and any unstable clinical condition, such as bronchial exacerbations. The diagnosis of COPD was made according to the diagnostic criteria of the GOLD (Global Initiative for Chronic Obstructive Lung Disease) guidelines \[[@B8]\]. Several subjects with COPD were using bronchodilators and three subjects were using inhaled corticosteroids. The study was approved by the local Ethic Committee. Study design ------------ ### Clinical and functional assessment Prior to surgical lung resection, each subject underwent a clinical evaluation and a complete respiratory functional assessment that included spirometry, and determinations of lung volume and of the single-breath CO diffusing capacity. The clinical evaluation included a questionnaire that derives from the IUALTD (International Union Against Lung and Tuberculosis Disease) bronchial symptom questionnaire \[[@B9]\] and a physical examination. Total lung capacity (TLC) was determined by bodyplethysmography (Sensor Medics Corporation V6200 AutoBox, Yorba Linda, CA). TLC was expressed as percent predicted based on the prediction equation of Goldman and Becklake \[[@B10]\]. Single-breath diffusing capacity for CO was determined using a fully-computerized water-sealed Stead-Wells spirometer (Baires System; Biomedin, Padua, Italy) and the transfer factor of the lung for CO (TLCO) was measured. At least two determinations of TLCO that were within 5% of each other were obtained, and the highest value was retained for analysis. In a series of subsequent visits, the bronchodilatory effect of deep inspirations was determined, as previously described \[[@B1],[@B11]\]. Inhaled short-acting β-agonists and/or anticholinergic agents were withheld for at least 8 h and long-acting β-agonists for at least 24 h before each visit that involved either lung function or methacholine bronchoprovocation. A series of single dose methacholine bronchoprovocations (a single dose per challenge) were performed, in the absence of deep inspirations, using stepwise increasing doses of methacholine. All subjects started with 0.025 mg/ml and increased the dose by a factor of 2 at each visit until an at least 15% reduction in inspiratory vital capacity (IVC) from baseline, under deep breath prohibition (Figure [1](#F1){ref-type="fig"}). Methacholine was delivered through an ampul-dosimeter (Mefar Elettromedicali; Bovezzo, Italy), which was activated by an inspiratory effort for 0.5 seconds at a time. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Schematic of the single dose methacholine bronchoprovocation protocol designed to induce at least a 15% reduction in inspiratory vital capacity (IVC), in the absence of deep breaths, and to calculate the bronchodilatory effect of deep inspirations. The combination spirometric maneuvers (partial forced expiration followed by full forced expiration) used to determine IVC are also depicted. ::: ![](1465-9921-6-55-1) ::: To measure IVC, the subject forcefully expired from end-tidal volume to residual volume (partial expiratory maneuver) and immediately inhaled to total lung capacity (IVC = TLC -- RV). The maneuver continued with a forced expiration to RV. This forced expiration allowed us to also calculate FEV~1~and FVC. At baseline, 3 acceptable combined partial/maximal forced expiratory maneuvers were performed, and the best was retained for analysis. Subjects were then instructed to abstain from taking deep breaths for a period of 20 minutes. Thereafter, the single dose of methacholine was administered as five tidal breaths followed, 3 minutes later, by a single partial/maximal combined spirometric measurement, as described above. If the targeted reduction in IVC (at least 15%) was not attained, another single dose challenge was performed. This was done during the same session (at one hour) if the reduction in IVC was within 5% of baseline, or postponed to the next day. Single dose provocations with increasing doses were conducted in this manner until the expected level of reduction in IVC was reached. The provocation in which the targeted reduction in IVC from baseline was attained was extended with 4 deep inspirations taken immediately after the post-methacholine IVC spirometry. Following these deep inspirations, another IVC maneuver was performed. The difference between the IVC obtained after the 4 deep inspirations and the post-methacholine IVC that preceded the 4 deep inspirations was used to calculate the bronchodilatory effect of deep inspirations. This was expressed as percent change from the post-methacholine IVC (% bronchodilation). We have used IVC in studying the effects of deep inspiration because, assuming that TLC does not change \[[@B12]\], the primary determinant of a change in IVC is the change in residual volume. Although IVC is sensitive to the effects of deep inspirations, its actual measurement, in contrast to that of FVC or FEV~1~, is not influenced by a lung inflation maneuver because RV is reached through a partial forced expiration. FEV~1~and FVC data were utilized in secondary analyses. ### Morphometric assessment For tissue morphometry, we applied the methodology previously described by Saetta and colleagues \[[@B6]\]. Two to seven randomly selected tissue blocks were taken from the subpleural parenchyma of the resected lobe that was tumor-free. Specimens were fixed in 10% neutral buffered formalin (pH 7.2) for at least 24 h and embedded in paraffin wax. Four-μm sections were attached to microscope slides pretreated with polylysine solution (Sigma Aldrich). After dewaxing and rehydratation, all slides were stained with haematoxylin and eosin. Tissue samples were coded and evaluated blindly by two independent investigators using a light microscope (Leica, Wetzlar, Germany). The images were analyzed by a computerized system (Quantimet 500 MC software, Leica, Wetzlar, Germany). All airways with internal diameter ≤ 2 mm were retained for analysis. Non-respiratory bronchioles with incomplete walls at the edges of the sections or with a short/long diameter ratio \< 1/3 were excluded. After this selection, each patient had at least four non-respiratory bronchioles suitable for morphometry. In each airway, the external perimeter (P~e~), the internal perimeter along the subepithelial basement membrane (P~bm~), the lumenal diameter (Dl), the external area (Ae), the internal area (Ai), and the muscle area (WAm) were evaluated. The thickness of the nonrespiratory bronchioles (wall area, WAtot) was obtained by the difference between the external area and the internal areas (WA = Ae -- Ai). Values of Dl, WAm and WAtot were normalized by dividing for P~bm~\[[@B13]\]. According to the method of Saetta and colleagues \[[@B6]\], alveolar attachments (AAi) were identified as the alveolar septa that extend from the outer wall of the nonrespiratory bronchioles. Those attachments showing rupture or discontinuity were defined as destroyed alveolar attachments (AAd). The number of destroyed alveolar attachments, expressed as a percentage over the total number of alveolar attachments, represented the primary outcome of the study. The data used for analysis were averages of those obtained independently from each the two study pathologists. ### Data analysis Linear regression analysis was performed to correlate the bronchodilatory effect of deep inspirations with the morphometric variables obtained from the lung tissue. Secondary analysis using the same approach was employed to assess the relationship between the magnitude of bronchoconstriction that was induced in the absence of deep breaths (in terms of IVC and FEV1) and the morphometric variables. Unpaired t-tests were used to assess differences between groups. In all analyses, two-tailed values of p = 0.05 were considered statistically significant. Results ======= Descriptive findings -------------------- A total of fifteen subjects took part in the study (age: 67 ± 2.2 yrs, mean ± SEM). Seven of them had a diagnosis of COPD, confirmed by our clinical and functional evaluation. No subject received a diagnosis of asthma. Eleven out of the 15 subjects were smokers (69 ± 27 pack-years, mean ± SD). None of the non-smokers received the diagnosis of COPD. Baseline lung function and lung tissue morphometric characteristics for each individual are presented in Table [1](#T1){ref-type="table"}. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Functional and morphometric characteristics of study participants. ::: Mean ± SEM Range ------------------- ------------- ------------ FEV1, % predicted 82 ± 6.2 43--118 FEV1/FVC 0.65 ± 0.03 0.44--0.77 TLC, % predicted 117 ± 14.3 82--164 TLCO, % predicted 74 ± 3.9 55--92 WAtot/Pbm, μm 80 ± 7.9 52--154 WAm/Pbm, μm 11 ± 1.6 6--22 AAd, % 33 ± 4.0 12--61 ::: The median single methacholine dose required to induce the targeted reduction in IVC, in the absence of deep inspiration, was 25 mg/ml (range: 0.025--75 mg/ml). The % reduction in IVC in the protocol devoid of deep inspiration was 20 ± 1.8% (mean ± SEM). The % bronchodilation by deep inspiration was 4.3 ± 2.1% with a range of -13% to 18%. Correlative findings -------------------- We found a significant inverse correlation between the bronchodilatory effect of deep inspiration and the percentage of destroyed alveolar attachments (r = -0.51, p = 0.05, Figure [2](#F2){ref-type="fig"}). In addition, the bronchodilatory effect of deep inspiration correlated directly with the airway smooth muscle area (r = 0.59, p = 0.03). In contrast, no correlation with the magnitude of the total wall area (r = 0.39, p = 0.15) was observed. The multiple regression analysis, in which the bronchodilatory effect of deep inspiration is the dependent variable, and the percent of destroyed alveolar attachments and the airway smooth muscle area serve as independent variables, yielded a p value of 0.02; however, neither the alveolar attachment (p = 0.09) nor the airway smooth muscle (p = 0.06) entered the model. The bronchodilatory effect of deep inspiration did not differ between subjects with COPD and the 4 subjects without COPD, but with a history of smoking (2.6 ± 4.2% vs. 5.2 ± 3.6%, respectively; p = 0.68). Similarly, no differences were found between COPD subjects and the non-COPD smokers with respect to the percentage of destroyed alveolar attachments (40 ± 7.2% vs. 35 ± 3.9%, respectively; p = 0.64). When the entire group was considered, the % destroyed attachments showed a strong inverse correlation with TLCO% predicted (r= -0.75, p = 0.003) (Figure [3](#F3){ref-type="fig"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Relationship between the bronchodilatory effect of deep inspiration and the percentage of destroyed alveolar attachments. ::: ![](1465-9921-6-55-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Relationship between TLCO and the percentage of destroyed alveolar attachments. ::: ![](1465-9921-6-55-3) ::: Discussion ========== We have recently documented the lack of deep inspiration-induced bronchodilation in COPD \[[@B1]\]. The results of this study confirm and extend our previous report. Herein, we provide a possible explanation for this phenomenon, by showing that the impairment of the bronchodilatory effect of deep inspiration is associated with reduction in the alveolar attachments to the airway walls. A significant correlation is not a proof for a causative relationship, but it is a pre-requisite for it. Moreover, there is good theoretical reason to propose that the reduction in the number of alveolar attachments is the most important factor responsible for the impairment in deep inspiration-induced bronchodilation, in smokers and individuals with COPD. A body of evidence has suggested that the effect of deep inspiration on the airways is a function of the interdependence between the airways and the parenchyma \[[@B14],[@B15]\], provided by the alveolar attachments that act by distending the airways when lung volume increases (Figure [4](#F4){ref-type="fig"}), and by the relative magnitudes of airway and parenchymal hystereses \[[@B16]\]. According to this theory, equal degrees of hysteresis result in no effect of a deep inspiration on airway caliber. If parenchymal hysteresis prevails, such as in COPD \[[@B17]\], a deep inspiratory maneuver fails to dilate airways, and may even result in bronchoconstriction. Therefore, the impairment in the bronchodilatory effect of deep inspiration in subjects with COPD could be explained by the increased ratio of parenchymal over airway hysteresis. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Pathology picture showing the intact (a) and the destroyed (b) alveolar attachments. ::: ![](1465-9921-6-55-4) ::: We reasoned that structural alterations of the lung parenchyma, specifically the destruction of alveolar attachments, would reduce the effectiveness of the distending forces in a manner that a deep inspiration would not be capable of stretching narrowed airways and/or reopen closed airways. However, other explanations need to be considered: first, increased airway smooth muscle mass could render the muscle too stiff to stretch, or generate higher forces that could counteract bronchodilation. However, we found that larger smooth muscle area was related to stronger bronchodilation by deep inspiration. Second, COPD could be associated with enhancement of a bronchoconstriction reflex that is activated by lung inflation, or with the failure to release bronchodilatory agents. Third, under a condition of reduced stretch, the airway smooth muscle could develop peculiar rearrangement of the contractile elements that would induce a state of increased resistance to the effect of deep inspiration. Finally, in a condition of lung hyperinflation, which is often recorded in subjects with emphysema, the amplitude of a deep inspiration could be severely reduced. Corsico et al. \[[@B18]\] recently reported findings that have similarities to ours. The authors showed that the loss of alveolar attachments is associated with a bronchoconstrictor effect of deep inspiration. In our previous study, which was conducted on subjects with COPD \[[@B1]\], we observed that, in those subjects with the lowest TLCO, deep inspirations led to bronchoconstriction, instead of bronchodilation. We have the same observation in this study (Figure [2](#F2){ref-type="fig"}), in individuals who are among those with the highest percentage of destroyed alveolar attachments (\>40%). In the study of Corsico and coworkers, the percentage of destroyed alveolar attachments is higher than that of the current study (46% vs. 33%). It is plausible that mild parenchymal alterations, such as those observed in smokers \[[@B6],[@B19]\], would attenuate bronchodilation by deep inspirations, whereas more advanced abnormalities of the lung would convert the beneficial effect of deep inspiration into a detrimental one. Whereas the morphometric approach was identical in the two studies, the functional assessment was different, in that, Corsico and colleagues employed the baseline ratio of maximal over partial expiratory flows (M/P), which may be a measure of deep inspiration-induced distensibility, rather than deep inspiration-induced bronchodilation. In other words, our protocol assesses the consequences of the deep inspiratory maneuver after the maneuver is completed, whereas the M/P ratio describes the difference in flow between a partial and a maximal expiration without necessarily predicting what the state of airway will be at the end of the maneuver. The correlation between the loss of the bronchodilatory ability of deep inspiration and the loss of alveolar attachments becomes even stronger if it is viewed in the context of the fact that the range of deep inspiration-induced bronchodilation that we observed in our subjects was quite narrow (18 to -13%). Overall, bronchodilation was substantially reduced in this group (4.3 ± 2.1%), compared to an average value of around 20% that we would have expected in healthy individuals of the same age, based on our previously published data \[[@B11]\]. It is also important to note that the average bronchodilation by deep inspiration we report in this group of subjects is the same as in a group of individuals, all diagnosed with COPD, that we have reported earlier \[[@B1]\]. Although the number of subjects is too small for meaningful conclusions to be drawn, it is interesting that the deep inspiration effect appeared to be reduced even in smokers without the diagnosis of COPD. In our previous study on subjects with COPD, the bronchodilatory effect of deep inspiration correlated with TLCO, but not with spirometric outcomes such as FEV~1~or FEV~1~/FVC \[[@B1]\]. The significant inverse correlation between TLCO and the percentage of destroyed attachments we found in this study offers an explanation for the above-cited relationship. The lack of correlation between the thickness of airway wall and the attenuation of the bronchodilatory effect of deep inspiration is also in agreement with the report by Corsico and colleagues \[[@B18]\], and indicates that, when parenchymal destruction is present, airway wall factors play a secondary role in determining the magnitude of the beneficial effects of deep inspiration. This may be different in asthma, where parenchymal involvement appears to be minimal \[[@B20],[@B21]\]. The presence of a direct correlation between the bronchodilatory ability of deep inspiration and the airway smooth muscle area is difficult to explain. One possibility is that increased smooth muscle mass leads to more bronchoconstriction and this may, up to a point, increase the bronchodilatory effects of deep inspiration by increasing radial traction \[[@B15],[@B22]\]. Indeed, we have previously shown that the bronchodilatory effect of deep inspiration cannot be measured when the induced bronchoconstriction is relatively small \[[@B3]\]. Conclusion ========== The results of our study support the hypothesis that the attenuation of airway stretch due to loss of alveolar attachments represents an important cause for the impairment in the bronchodilatory effect of lung inflation in COPD. Whether the progressive impairment of the beneficial effect by deep inspiration has clinical and prognostic implications in these subjects needs to be addressed in future studies. Competing interests =================== Dr. Togias\' participation in this work was supported by NIH grant RO1 HL61277. The authors declare that they have no competing interest. Authors\' contributions ======================= NS conceived and designed the study, performed the clinical and functional assessments, analyzed and interpreted the data, drafted the manuscript; AB carried the morphometric assessment and participated to the interpretation of the findings; RM carried the clinical and functional assessments and participated to the interpretation of the data; AMV participated to the design and the coordination of the study and the interpretation of the results; AT conceived and participated to the design of the study, the analysis of the data and the interpretation of the results, and contributed significantly to draft the manuscript; VB helped in the design and the organization of the study, as well as the interpretation of the results. Acknowledgements ================ The authors wish to thank the colleagues of the Unit of Thoracic Surgery, \"V. Cervello\" Hospital, Palermo (Dr. Balistreri, Dr. Regio, Dr. Agneta, Dr. Caronia, Dr. Mazzotta) who provided the lung tissue, and the personnel of the Unit of Pathology, \"V. Cervello\" Hospital, Palermo, who provided the tissue blocks. The authors are also indebted to the national Council of Research of Palermo for providing the equipment for morphometric analysis. All authors read and approved the final version of the manuscript.	PubMed Central	2024-06-05T03:55:59.569873	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180475/", "journal": "Respir Res. 2005 Jun 8; 6(1):55", "authors": [ { "first": "Nicola", "last": "Scichilone" }, { "first": "Andreina", "last": "Bruno" }, { "first": "Roberto", "last": "Marchese" }, { "first": "Antonio Maurizio", "last": "Vignola" }, { "first": "Alkis", "last": "Togias" }, { "first": "Vincenzo", "last": "Bellia" } ] }
PMC1180476	Background ========== Signaling networks and metabolic pathways in living cells are regulated through a complex web of enzyme cascades. The regulatory architecture of these covalent modification cascades in combination with allosteric interactions determines the control of cellular processes \[[@B1],[@B2]\]. A prototypical example of such an enzyme cascade system is the regulation of glycogen phosphorylase (GP) and glycogen synthase (GS), enzymes involved in glycogen degradation (glycogenolysis) and synthesis (glycogenesis) respectively \[[@B3]-[@B6]\]. To circumvent a futile cycle, simultaneous activation of glycogenolysis and glycogen synthesis is prevented through reciprocal regulation of glycogen phosphorylase and synthase activities by a unique regulatory network \[[@B5],[@B6]\]. Although this reciprocal regulation is identical in all tissues, there are subtle differences indicating distinctive adaptation strategies in different cell types. For example, in skeletal muscle, phosphoprotein phosphatase-1 (PP1) is allosterically inactivated by inhibitor-1, whereas in the liver no such specific inhibitor has been observed \[[@B3],[@B7]\]. Instead, it has been demonstrated that active GP itself plays a similar inhibitory role, regulating the GS cascade by allosterically inactivating the corresponding phosphatase \[[@B8]\] (Fig. [1](#F1){ref-type="fig"}). In liver, the phosphorylation states of GP and GS are regulated by glucose and glucose-6-phosphate, whereas in muscle, GP and GS are regulated mainly by cyclic AMP (cAMP) and calcium concentration \[[@B9]\]. In the absence of glycogen in the liver, i.e. under starved condition, both GP and GS appear to co-exist in an active form constituting a futile cycle, thus overcoming the reciprocal regulation existing in a normally-fed condition \[[@B10]\]. In the present work, we have quantified the glycogen cascade system at steady state to examine the effect of the network architecture on its performance in liver and muscle. We have also gained insights into the operation of the system in liver under fed and starved conditions. The steady state model incorporates the cascade structure, multi-step and zero-order effects and inhibitor sensitivity in response to cAMP and glucose. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Enzyme cascades involved in the regulation of glycogen synthesis and degradation in (A) Skeletal Muscle (B) Liver. Nomenclature: Active enzyme form is indicated by an affix \'a\'and the corresponding inactive form is indicated by an affix \'b\'. R~2~C~2~, cyclic AMP dependent protein kinase (CAPK); C, catalytic subunit of CAPK; PP1, phosphatase-1; PrP2, phosphatases-2A; PK, Phosphorylase kinase; GP, glycogen phosphorylase; GS, Glycogen synthase; Glu6P, glucose-6-phosphate; PP1 Inhibitor-1, Inhibitor of PP1; Km~1~to Km~8~are Michaelis-Menten constants, k~1~to k~8~are rate constants, K~11~, K~22~, Kd are dissociation constants as shown in the figure. Positive and negative signs indicate the activation and inhibition of a reaction respectively. In the muscle (Fig. 1A), cAMP activated CAPK catalyzes the phosphorylation of GS, PK and inhibitor-1. Phosphorylated PK activates GP-b. Active phosphatase-2A is assumed to inactivate inhibitor-1, whereas PP1 catalyzes the dephosphorylation of GS, GP and PK. In liver (Fig. 1B), GP-acatalyzes the allosteric inactivation of GS phosphatase and inhibitor-1 does not appear to be involved in the regulation of PP1. ::: ![](1742-4682-2-19-1) ::: The regulatory system for glycogen synthesis and breakdown mainly consists of phosphorylation and dephosphorylation of phosphorylase kinase (PK), which further regulates the activities of GP and GS \[reviewed in \[[@B3]-[@B6]\], \[[@B9]-[@B12]\]\] (Fig. [1](#F1){ref-type="fig"}). The activities of these enzymes depend on extracellular signals as hormones and on cellular-metabolic signals such as glucose and cAMP levels \[[@B5],[@B11]\]. Phosphorylation of GP and GS converts them to catalytically more active (a-form) and inactive (b-form) species than their respective dephosphorylated forms. GP is activated by PK, which in-turn is activated by cAMP-dependent protein kinase (CAPK). GS is inactivated by multiple protein kinases including CAPK and PK \[[@B9]\]. PP1 is one of the main phosphatases catalyzing the dephosphorylation of PK, GP and GS. The regulation of PP1 activity is quite different in muscle and liver, which are the major sites of glycogenolysis and glycogenesis (Fig. [1](#F1){ref-type="fig"}). In liver, GS phosphatase is allosterically inactivated by active GP, whereas in muscle, PP1 is allosterically inactivated by CAPK-activated inhibitor-1 \[[@B3],[@B5],[@B9],[@B12]\]. Thus, an increased cAMP level in the muscle cytosol not only increases the phosphorylation of PK, GP and GS, but also decreases their dephosphorylation by regulating the corresponding phosphatases. In addition to covalent modification, GP and GS are also regulated by allosteric interactions. AMP is an allosteric activator, whereas ATP and glucose-6-phosphate are allosteric inhibitors of phosphorylase-b\[[@B3]\]. Synthase-bis allosterically inhibited by physiological concentrations of ATP, ADP and inorganic phosphate, and is also allosterically activated by glucose-6-phosphate \[[@B9]\]. Experimental and theoretical quantifications \[[@B13]-[@B23]\] have revealed that there are significant advantages in having an interconvertible enzyme cascade structure in place of a simple allosteric interaction. These may include signal amplification, flexibility, robustness, ultrasensitivity and signal integration \[[@B22]\]. Ultrasensitivity has been defined as the response of a system that is more sensitive to changes in the concentration of a ligand than the normal hyperbolic response represented by the Michaelis-Menten equation \[[@B20]\]. The Hill coefficient has been used as a sensitivity parameter to quantify the steepness of sigmoidal dose-response curves \[[@B22]\]. A Hill coefficient greater than one indicates an ultrasensitive response, and a value less than one indicates a subsensitive response. The existence of ultrasensitivity in covalent modification cycles is due to the operation of enzymes in a region of saturation with respect to their substrates (termed zero order sensitivity) \[[@B14],[@B15]\], involvement of the same effector in multiple steps of a pathway \[[@B15]\], and the presence of stoichiometric inhibitors \[[@B20]\]. All these requirements for ultrasensitivity appear to be fulfilled by the enzyme cascades involved in glycogen synthesis and degradation. Edstrom and coworkers \[[@B24],[@B25]\] have provided experimental proof of zero order ultrasensitivity in the muscle glycogen phosphorylase cascade. Theoretical analysis of the glucose-induced switch between phosphorylase and glycogen synthase in the liver showed the possibility of a sharp threshold in the response \[[@B26]\]. Furthermore, the multistep effects of cAMP in the glycogen cascade system are brought about by activation of the forward step and indirect inhibition of the reverse step (inhibition of phosphatases), thus satisfying the requirement for ultrasensitivity \[[@B27]\]. Although it is known that the second messenger cAMP affects five different steps in the glycogen cascades, its effective role in multistep ultrasensitivity has not been quantified. The output performance of the phosphorylase and glycogen synthase cascade in the presence of an inhibitor has also not been characterized. The main objective of the current work was to compare the regulatory structure of the glycogen cascade system prevalent in the liver and the muscle through steady state analysis. The quantification incorporates the influences of all the effectors that regulate the output response of the glycogen cascade system. The simulation results revealed that the cascade system exhibits highly sensitive switch-like responses to changes in cAMP concentration and the output responses are surprisingly different in muscle and liver. In muscle, glycogen phosphorylase is more sensitive than glycogen synthase to cAMP, while the opposite is observed in liver. The steady state analysis indicates that, when liver undergoes a transition from starved to fed state, different proportions of active GP and GS can coexist. The transition from such a futile cycle to reciprocal regulation depends on the varying inhibition of GS phosphatase by GP and this regulation may be necessary to meet the challenges that exist under starved conditions. Materials and methods ===================== The enzyme cascades involved in the regulation of glycogen synthesis and degradation in muscle and liver are schematically shown in Fig. [1A](#F1){ref-type="fig"} and [1B](#F1){ref-type="fig"} respectively. The concentrations of the metabolites ATP, AMP and PPi are assumed to be constant throughout the analysis. Allosteric regulations of GP and GS by these metabolites and effectors are also neglected. Detailed information on the set of equations and list of parameters used for the simulation are given in the Appendix.Most of the parameters and enzyme concentrations are taken from literature sources and the same set has been used for simulating the glycogen cascade system of skeletal muscle and liver. In the present work, the cAMP concentration is considered to be the primary input to the glycogen cascade. The fractional activations of GS (dephosphorylated form) and GP (phosphorylated form) are taken as the output responses of the glycogen system. The effects of cAMP on the enzyme cascade are mediated through activation of the allosteric enzyme CAPK. In the absence of cAMP, CAPK exists as an inactive holoenzyme, R~2~C~2~, with tightly bound subunits of the regulatory dimer R~2~and the catalytic subunit C. However, in the presence of cAMP, R~2~C~2~becomes activated through the binding of cAMP to the regulatory subunit and subsequent dissociation of the holoenzyme into cAMP-bound regulatory subunits and the free catalytic subunit \[[@B17]\]. The overall reaction scheme of CAPK activation is, R~2~C~2~+ 4(cAMP) ↔ 2C+ R~2~(cAMP)~4~ \[1\] In the present work, CAPK activation by cAMP is assumed to be a stepwise dissociation of the catalytic subunits. The analytic expression for quantifying the CAPK activation is taken from Shacter et al.\[[@B17]\] and it is assumed that the complex between the catalytic subunit of CAPK and its target enzyme is negligible compared to the total concentration of CAPK. The activation of CAPK in terms of catalytic subunit formation is quantified using the following cubic equation (see Appendixfor details): ![](1742-4682-2-19-i1.gif) where (R~2~C~2~)~t~denotes the total CAPK, C is the catalytic subunit, (cAMP) is the total cAMP concentration, and K~11~and K~22~are the dissociation constants of the first and second catalytic subunits respectively. A valid root was obtained as total CAPK catalytic subunit concentration using Eq. 2 and is taken as the input for modification of downstream target enzymes. Figure [1A](#F1){ref-type="fig"} shows the schematic of the enzyme cascades involved in regulation of glycogen synthesis and breakdown in the skeletal muscle. Although dual phosphorylation of PK and multiple phosphorylation of GS have been observed in vitro\[[@B5],[@B9]\], for simplicity we have considered a single phosphorylation site for these enzymes. To incorporate the PK and CAPK catalyzed phosphorylation of GS, it is assumed that both the enzymes form a pool before catalyzing the GS phosphorylation. Ca^+2^, which acts as another input stimulus to the system, is assumed to be present at concentrations corresponding to full activation of PK. Phosphorylated Inhibitor-1 inactivates PP1 by an allosteric reaction but it fails to inhibit phosphatase-2A. Here, we consider phosphatase-2A as a dephosphorylating enzyme of active inhibitor-1, as inhibitor-1 does not inhibit its own dephosphorylation even at saturating concentration \[[@B3]\]. Figure [1B](#F1){ref-type="fig"} shows the schematic of the glycogen cascade structure in liver. In vitrostudies have shown that glucose-6-phosphate can stimulate dephosphorylation of GS and inhibit phosphorylation of GP-band GS-a, whereas glucose acts as an allosteric activator of GP phosphatase \[[@B28]-[@B33]\]. In the present work, we have incorporated these effects along with the allosteric inhibition of PP1 by GP-a. It is assumed that glucose and glucose-6-phosphate influence the phosphorylation and dephosphorylation reactions by decreasing the respective Michaelis-Menten constants (see Appendixfor equations). Glucose concentration was varied between 0.1 mM to 100 mM and the corresponding level of glucose-6-phosphate was calculated to be in the physiological range of 0.1--0.5 mM. The intracellular cAMP level is regulated by glucose concentration through hormonal signals such as glucagon. The inverse relationship between glucose and cAMP levels was incorporated to estimate the cAMP levels from the glucose concentration (details in Appendix) The performance of the enzyme cascades in response to different cAMP input stimuli was analyzed by the steady state operating equation from the classic work of Goldbeter and Koshland \[[@B14]\]. For illustrative purposes, we present the following cubic equation, which quantifies the fractional activation inhibitor-1 (Fig. [1A](#F1){ref-type="fig"}) by taking all (Michaelis-Menten) complexes of a cascade into account: ![](1742-4682-2-19-i2.gif) where f~1~= I/I~t~, I~t~is the total inhibitor concentration, (PP2)~t~is the total phosphatase-2A and other terms are as given in Fig. [1A](#F1){ref-type="fig"}. From the constraint 0 \<f~1~\< 1, a valid root was obtained as a fractional unmodified inhibitor using Eq. 3. The fractional phosphorylated inhibitor (i.e. I~p~/I~t~) can then be obtained from the following relationship, ![](1742-4682-2-19-i3.gif) The operating equation for the allosteric interaction of PP1 with inhibitor-1 and phosphorylase is taken from our earlier work \[[@B34]\]. The following quadratic equation was used to simulate the allosteric inhibition of muscle PP1 by phosphorylated inhibitor-1, given by ![](1742-4682-2-19-i4.gif) where PP1.I~p~is inactive PP1 and K~d~is the dissociation constant: ![](1742-4682-2-19-i5.gif) where (PP1)~t~is the total PP1 and f~3~is the fractional inactivated PP1 (i.e., (PP1.I~p~)/(PP1)~t~). The fractional free (active) species of PP1 (i.e., f~4~= (PP1)/(PP1)~t~) can be estimated by f~4~= 1-f~3~. In the present work, the cascade-connecting complexes are neglected. For example, complexes of PK with GP-band PK with GS-aare neglected in the total PK balance (details in Appendix). The steady state operating equation for individual covalent modification cycles and allosteric interaction were sequentially connected to evaluate the output response of the cascade structure i.e. fractional modification of GP and GS to the primary input stimulus, cAMP in muscle and glucose in liver (details in Appendix). These equations were solved simultaneously using Matlab (The Mathworks Inc. USA) to obtain dose-response curves for fractional steady state activation of all the component enzymes at various input stimulus levels. Since most of the parameters are taken from different experimental reports, we performed the sensitivity analysis on the complete data set. To assess the sensitivity to variations in individual parameters, each parameter was varied over a 10-fold while holding all the other parameters constant. Results ======= The steady state model was used to obtain dose-response curves for the fractional activations of the component enzymes in glycogen synthesis and degradation. Figure [2A](#F2){ref-type="fig"} shows the fractional modification of GP, GS, PK, CAPK and inhibitor-1 at various concentration of cAMP in skeletal muscle. The dose-response curves show an increase in signal amplification and sensitivity as the signal propagates down the cascade. The fractional activation of CAPK at various concentrations of cAMP (curve \'e\' Fig. [2A](#F2){ref-type="fig"}) shows a response curve with an apparent Hill coefficient (![](1742-4682-2-19-i6.gif)) of 1.12 and the simulated results are in agreement with in vitroexperimental studies reported by Beavo et al.\[[@B35]\]. The fractional modifications of GP and GS demonstrate ultrasensitivity with apparent Hill coefficients of 34 and 7.3 respectively (Fig. [2A](#F2){ref-type="fig"}). Previous experimental and theoretical studies by Edstrom and coworkers on the glycogen phosphorylase cascade reported a Hill coefficient of 2.3 in the absence of inhibitor-1 action in muscle \[[@B24]\]. In subsequent work, they observed that the phosphorylase cascade exhibits greater sensitivity in the presence of phosphatase inhibitor \[[@B25]\]. To assess the contribution of individual parameters on the output response of the system, we carried out the sensitivity analysis on the parameter set. The results indicate that the sensitivities of GP and GS display switch-like outputs in response to variation over a wide range of parameters (Table [1](#T1){ref-type="table"}). Further, it can be noted that the sensitivity of the GP response is always greater than that of GS in skeletal muscle irrespective of the range considered for the parameter set. Our simulated results show that, in the absence of PP1 inhibition by inhibitor-1, the steepness of the dose-response curves and signal amplification decreased (see Fig. [2B](#F2){ref-type="fig"}). The fractional activations of GP and GS show apparent Hill coefficients of 3.8 and 1.9 respectively, as compared to a highly sensitive response in the presence of inhibitor action. This demonstrates that inhibitor ultrasensitivity plays a major role in imparting sensitivity to the GP and GS responses in muscle. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Parametric sensitivity analysis for the glycogen cascade system. The term \'standard\' indicates the parameter set used for simulation in this work and the value is indicated in parenthesis. These parameters were varied over a wide range to assess the sensitivity of the response. The star symbol indicates that the output response of a particular enzyme did not reach full activation. ::: Sensitivity analysis for glycogen cascade system of skeletal muscle ------------------------------------------------------------------------- ------------------------------ ------------------ ------------------------------------------------------------ --------------- -------------- ------------------------ Apparent Hill coefficient (Standard) to cAMP levels S. No. Parameter (standard set) Varied Range GP (33) GS (6.4) PK (7) Inhibitor -1 (1.4) Rate constants (sec^-1^) 1 k1 (1.4) 0.14 -- 14 12.2 -- 48 2.4 -- 17.8 13 -- 3.6 1.3 -- 1.3 2 k2 (0.01) 0.001 --0.1 48 -- 12.2 17.8 -- 2.4 3.6 -- 13.9 1.3 -- 1.34 3 k3 (20) 2 -- 200 48 -- 12.3 16.2 -- 2.5 3.6 -- 13.9 1.34 -- 1.34 4 k4 (5) 0.5 -- 50 12.3 -- 48 2.5 -- 16.2 13.9 -- 3.6 1.34 -- 1.34 5 k5 (20) 2 -- 200 48.7 -- 12.2 6.4 -- 6.4 7 -- 7 1.34 -- 1.34 6 k6 (5) 0.5 -- 50 12.2 -- 48.7 6.4 -- 6.4 7 -- 7 1.34 -- 1.34 7 k7 (20) 2 -- 200 33.8 -- 33 17.7 -- 2.4 7 -- 7 1.34 -- 1.34 8 k8 (0.05) 0.005 -- 0.5 33.8 -- 33 2.4 -- 17.7 7 -- 7 1.34 -- 1.34 *Michaelis-Menten Constants (μM)* 9 Km1 (5) 0.5 -- 50 49 -- 13.6 11.6 -- 2.7 5.9 -- 12.9 1.85 -- 1.2 10 Km2 (0.7) 0.07 -- 70 40.5 -- 27 3.9 -- 19.4 18 -- 2.6 1.85 -- 1.1 11 Km3 (0.4) 0.04 -- 4 32 -- 42.5 6 -- 9 11.9 -- 3.1 1.34 -- 1.34 12 Km4 (1.1) 0.11 -- 11 48.9 -- 12 16.3 -- 2.5 11.4 -- 8.8 1.34 -- 1.34 13 Km5 (10) 1 -- 100 57.9 -- 25 6.4 -- 6.4 7 -- 7 1.34 -- 1.34 14 Km6 (5) 0.5 -- 50 55 -- 11.5 6.4 -- 6.4 7 -- 7 1.34 -- 1.34 15 Km7 (15) 1.5 -- 150 33.8 -- 33.8 3.8 -- 16 7 -- 7 1.34 -- 1.34 16 Km8 (0.12) 0.012 -- 1.2 33.8 -- 33.8 3 -- 7.8 7 -- 7 1.34 -- 1.34 Sensitivity analysis for glycogen cascade system of Liver S. No. Parameter (standard set) VariedRange Apparent Hill coefficient (Standard) to glucose levels GP (6.3) GS (13.6) PK (1.6) \-\-\-- Rate constants (sec^-1^) 1 k3 (20) 2 -- 200 6 -- 6 13.7 -- 14 \* -- 2.9 \-\-\-- 2 k4 (5) 0.5 -- 50 6 -- 6 14 -- 13.7 \* -- 2.9 \-\-\-- 3 k5 (20) 2 -- 200 5.3 -- 5.4 21 -- 14.1 1.6 -- 1.6 \-\-\-- 4 k6 (5) 0.5 -- 50 5.4 -- 5.4 14.1 -- 21 1.6 -- 1.6 \-\-\-- 5 k7 (20) 2 -- 200 6.3 -- 6.3 13 -- 20.1 1.6 -- 1.6 \-\-\-- 6 k8 (4) 0.4 -- 40 6.3 -- 6.3 20.1 -- 13 1.6 -- 1.6 \-\-\-- *Michaelis-Menten Constants (μM)* 7 Km3 (0.4) 0.04 -- 4 6.3 -- 6.2 13.6 -- 13.4 4 -- \* \-\-\-- 8 Km4 (1.1) 0.11 -- 11 10.6 -- 5.4 14.5 -- 13.8 \* -- 2.9 \-\-\-- 9 Km5 (10) 1 -- 100 11.2 -- 5 12.2 -- 18.7 1.6 -- 1.6 \-\-\-- 10 Km6 (5) 0.5 -- 50 8 -- 3.8 27 -- 11.3 1.6 -- 1.6 \-\-\-- 11 Km7 (15) 1.5 -- 150 6.3 -- 6.3 20.5 -- 13 1.6 -- 1.6 \-\-\-- 12 Km8 (0.12) 0.012 -- 1.2 6.3 -- 6.3 13.9 -- 12.3 1.6 -- 1.6 ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Predicted dose-response curves in case of skeletal muscle. The star symbol (\) represents the experimental data from Beavo et al.\[35\]. (A) Dose-response curves in the presence of inhibition of PP1 by inhibitor-1. The sensitivity of the fractional dose-response curve of glycogen synthase (curve a, Apparent Hill coefficient ![](1742-4682-2-19-i6.gif)\~6.4), glycogen phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~33.8), phosphorylase kinase (curve c, ![](1742-4682-2-19-i6.gif)\~7), inhibitor-1 (curve d, ![](1742-4682-2-19-i6.gif)\~1.3), CAPK activation (curve e, ![](1742-4682-2-19-i6.gif)\~1.12). (B) Dose-response curves in absence of inhibition of PP1 by inhibitor-1. The sensitivity fractional dose-response curve of Glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~1.2); Glycogen phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~3.8); Phosphorylase kinase (curve c, ![](1742-4682-2-19-i6.gif)\~0.8); Inhibitor-1 (curve d: ![](1742-4682-2-19-i6.gif)\~1.3); CAPK activation (curve e, ![](1742-4682-2-19-i6.gif)\~1.12). ::: ![](1742-4682-2-19-2) ::: The analysis was extended to the glycogen cascade system in liver. The coordinated changes in the phosphorylation of PK, GP and GS are under the influence of cAMP, glucose and glucose-6-phosphate concentrations (Fig. [1B](#F1){ref-type="fig"}). Figure [3](#F3){ref-type="fig"} shows the predicted performance of the glycogen cascade system in liver at different concentration of glucose, glucose-6-phosphate and cAMP. The results are surprisingly different from those obtained in muscle. Figure [3A](#F3){ref-type="fig"} shows that the fractional activation of GS exhibits a steeper response with an apparent Hill coefficient of 13.6, while GP demonstrates a response with an apparent Hill coefficient of 6.3 with respect to glucose. The response sensitivity of GS was found to be highly dependent on the GP-aconcentration. This result is seems to be in agreement with a recent study showing that hepatic glycogen synthesis and glycogen synthase activity is highly sensitive to phosphorylase activity \[[@B36]\]. Because of the stronger binding between GP-aand GS phosphatase, GS becomes activated only when the GP-alevels drop below 1%. This inverse switching between the inactivation of GP and activation of GS occurs at a glucose concentration of \~10 mM. This result is in agreement with the experimental observation that GS becomes activated once GP-ainhibition of GS phosphatase becomes negligible, and this shift in activity occurs after meals when the glucose concentration rises above 10 mM \[[@B10],[@B37]\]. Sensitivity analysis of the parameter set indicated that the fractional modifications of GS and GP to glucose levels display switch-like outputs (Table [1](#T1){ref-type="table"}). It was noted that the sensitivity of the GS response is always greater than that of GP in liver irrespective of the range considered for the parameter set. The simulated dose-response curves for fractional activation of GP-aand GS-aat various concentrations of cAMP also show an ultrasensitive response. The threshold concentration of cAMP required to activate GP and inactivate GS is higher in liver (\~1 nM) than in muscle (\~0.01 nM). The dose-response curve for fractional modification of the enzymes with respect to glucose-6-phosphate demonstrates that the switching between GP and GS occurs at 20 μM with an ultrasensitive response (Fig. [3C](#F3){ref-type="fig"}). Our result is consistent with earlier observations showing an inverse correlation between the activity of GP-aand the concentration of glucose-6-phosphate \[[@B33]\]. Similarly, a direct correlation exists between GS-alevels and glucose-6-phosphate concentration. The threshold activation of phosphorylase and glycogen synthase is shown explicitly in Fig. [3D](#F3){ref-type="fig"} by plotting the active fraction of synthase against the active fraction of phosphorylase. GS is activated only when GP is mostly inactive, demonstrating the inverse relationship between the activities of the two enzymes. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Simulated results of glycogen cascade system in liver, incorporating glycogen synthase phosphatase inhibition by phosphorylase-a**. (A) Fractional modification of enzymes at various concentration of glucose. The sensitivity of the fractional dose-response curve of glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~13.6), phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~6.3), phosphorylase kinase (curve c, ![](1742-4682-2-19-i6.gif)\~1.6), CAPK (curve d, ![](1742-4682-2-19-i6.gif)\~1.12). (B) Fractional modification of enzymes at various concentrations of cAMP. The sensitivity of fractional dose-response curve of glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~6.8), phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~3.2), phosphorylase kinase (curve c, ![](1742-4682-2-19-i6.gif)\~1.6), CAPK (curve d, ![](1742-4682-2-19-i6.gif)\~1.12). (C) Fractional modification of enzymes at various concentrations of glucose-6-phosphate. The sensitivity of the fractional dose-response curve of glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~14.2) and phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~6.4). (D) Fractional modification of phosphorylase as a function of glycogen synthase demonstrating reciprocal regulation. The dissociation constant (Kd) of phosphorylase-a binding to glycogen synthase phosphatase is taken as 0.002 μM. ::: ![](1742-4682-2-19-3) ::: The inhibition of GS phosphatase by GP-adepends on glycogen concentration in liver and it has been shown that a minimal concentration of glycogen is essential for this inhibition \[[@B38],[@B39]\]. To simulate the fasted or glycogen depleted state in liver, the steady state analysis was repeated with the inhibition constant of GP-areduced. The simulated results (Fig [4](#F4){ref-type="fig"}) show that, at a 1000 fold decrease (Kd value of 2 μM) in the inhibition of GS phosphatase by GP-a, the liver may have appreciable amounts (about 50%) of both GP-aand GS-aat 4 to 9 mM glucose. This result is in agreement with the experimental observation reported by Massillon et al.\[[@B38]\]. We observe that this decrease in the steepness of the GS response curve is due to reduction in the phosphatase inhibition by GP-a. A decrease of similar extent in the ultrasensitivity of the GS response was observed with respect to cAMP and glucose-6-phosphate (see Fig. [4B](#F4){ref-type="fig"} and [4C](#F4){ref-type="fig"}). Furthermore, plotting the active fraction of GP as a function of the active fraction of GS demonstrates the absence of reciprocal regulation in the fed state (Fig. [4D](#F4){ref-type="fig"}). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Simulated results of glycogen cascade system in liver under starved conditions. (A) Fractional modification of enzymes at various concentrations of glucose. The sensitivity of the fractional dose-response curve of glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~10.4), phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~6.2), phosphorylase kinase (curve c, ![](1742-4682-2-19-i6.gif)\~1.6), CAPK (curve d, ![](1742-4682-2-19-i6.gif)\~1.12) (B) Fractional modification of enzymes at various concentrations of cAMP. The sensitivity of the fractional dose-response curve of glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~5.2), phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~3.1), phosphorylase kinase (curve c, ![](1742-4682-2-19-i6.gif)\~1.6), CAPK (curve d, ![](1742-4682-2-19-i6.gif)\~1.12) (C) Fractional modification of enzymes at various concentrations of glucose-6-phosphate. The sensitivity of the fractional dose-response curve of glycogen synthase (curve a, ![](1742-4682-2-19-i6.gif)\~10.5) and phosphorylase (curve b, ![](1742-4682-2-19-i6.gif)\~6.4). (D) Fractional modification of phosphorylase as function of glycogen synthase. The dissociation constant (Kd) of phosphorylase-abinding to glycogen synthase phosphatase is taken as 2 μM (\~1000 fold higher Kd than used to simulate results shown in Fig 3). Appreciable amounts of both glycogen synthase and phosphorylase exist in such a fasted state. ::: ![](1742-4682-2-19-4) ::: The exact percentage reduction in the inhibition of GS phosphatase by GP-ais unknown. When liver undergoes a metabolic shift from completely starved to fed state, the inhibition of GS phosphatase can vary over a wide range. This was simulated by changing the inhibition constant (Kd) of GS phosphatase from 0.002 μM to a very high Kd value to represent no inhibition. These results are shown in Fig. [5](#F5){ref-type="fig"} as a plot of the active fraction of GP against the active fraction of GS at different inhibitor constants. In the complete absence of inhibition, both GS and GP exist in 100% active states indicating a futile cycle (curve \'g\' Fig. [5](#F5){ref-type="fig"}). In such a state, the cells would not accumulate glycogen due to continuous glycogenolysis by GP-a. In the fed state, i.e. in the presence of appreciable amounts of glycogen in the liver, the inhibition of GS phosphatase by GP-ais high and a reciprocal regulation of GP and GS activity is observed (curve a, Fig. [5](#F5){ref-type="fig"}). Different proportions of active fractions of GP-aand GS-acan coexist when conditions change from starved to fed state, owing to varying net glycogen concentrations in the liver (curves b-f, Fig. [5](#F5){ref-type="fig"}). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Variable fractional levels of active phosphorylase-a and synthase-a in the liver under fasted (glycogen depletion) state. The dissociation constant of phosphorylase-a binding to glycogen synthase phosphatase was varied from 0.002 μM to no-inhibition (very high Kd), to simulate the metabolic transition from fasted to fed state. The values of dissociation constants (Kd) used are, curve a: 0.002 μM; curve b, 0.2 μM; curve c, 2 μM; curve d, 5 μM; curve e, 10 μM; curve f, 20 μM; curve g, very high dissociation constant (\~10^6^). The active fraction of glycogen synthase and phosphorylase coexist in liver in the no-inhibition state (starved condition), while simultaneous activation of phosphorylase and inactivation of synthase is seen in liver in the fed state. The fractional active form of glycogen synthase and phosphorylase varies over a wide range between these operations. ::: ![](1742-4682-2-19-5) ::: Discussion ========== The coordinated regulation of glycogenolysis and glycogenesis in the liver and the skeletal muscle is dependent on a network of interacting enzymes and effectors that determine the fractional activation of GP and GS \[[@B3]-[@B6],[@B9]-[@B12]\]. In the present work, the cascades involved in the regulation of glycogen synthesis and breakdown were analyzed at steady state to gain an insight into the inherent design principle of the regulatory cascades existing in muscle and liver. Using experimental data from the literature for rate and Michaelis-Menten constants, the simulation results revealed that, in muscle, the response of GP to cAMP input is more highly sensitive (![](1742-4682-2-19-i6.gif)\~34) than that of GS (![](1742-4682-2-19-i6.gif)\~6.5), whereas in the liver, the GS sensitivities to glucose (![](1742-4682-2-19-i6.gif)\~13.6) and cAMP (![](1742-4682-2-19-i6.gif)\~6.8) are high compared to that of GP (![](1742-4682-2-19-i6.gif)\~6.3 for glucose and ![](1742-4682-2-19-i6.gif)\~3.2 for cAMP). The sensitivity analysis indicated that this differential performance of GS and GP in liver and muscle is due to the presence of a distinctive regulatory design and not to selection of a particular parameter set. CAPK-activated inhibitor-1 inhibits PP1, which is a major dephosphorylating enzyme in muscle, whereas GP-ainhibits GS phosphatase in liver, representing this distinctive design. The simulation results indicate that the response sensitivity of GS with respect to glucose and cAMP is highly dependent on the GP concentration in liver. Similarly, the sensitivities of the PK, GP and GS responses are dependent on inhibitor-1 concentration in muscle. The ultrasensitive response of these enzymes may be attributed to the known system-level mechanisms, namely, multistep ultrasensitivity due to cAMP, inhibitor ultrasensitivity due to phosphatase inhibitor and zero order effects due to the pyramidal relationship in enzyme component concentrations. However, the significance of this switch-like response of GP in muscle and GS in liver is unclear. It can be argued that glycogen breakdown in muscle has to be sensitive to the second messenger cAMP in order to meet the urgent requirement for glucose during exercise or the fight and flee response. Similarly, glycogen synthesis in liver has to be sensitive to blood glucose concentration, so that GS can start synthesizing glycogen whenever the blood glucose concentration increases beyond a toxic level. In muscle, the ultrasensitive response of GP can be directly attributed to the presence of zero order effects (GP concentration about \~70 μM) and compounded by the inhibitor ultrasensitivity imparted by inhibitor-1. Such a direct effect is not observed in GS owing to its minimal zero order effects (GS concentration about \~3 μM). The primary stimulus, cAMP, not only increases the phosphorylation of PK, GP and GS, but also indirectly decreases their dephosphorylation through inhibitor-1. In liver, the ultrasensitive response of GS can be attributed mainly to the inhibitor ultrasensitivity caused by GP on the GS modification cycle. In this case, the zero order effect actually resides in the GP cascade, which transmits it to the GS cycle by inhibiting the dephosphorylation reaction. Furthermore, the stimulatory effect of glucose on dephosphorylation of GP-a, the inhibitory effect of glucose-6-phosphate on phosphorylation of GP-band GS-a, and the stimulation of GS dephosphorylation by glucose-6-phosphate, enhance the sensitivity of GS. Thus, the ultrasensitivity of GS in liver is brought about by the combined action of the multistep effects of cAMP, the inhibition of GS phosphatase by active GP and the influence of glucose and glucose-6-phosphate concentration. It is noteworthy that the simultaneous activation and deactivation of GP and GS respectively in muscle and liver results in reciprocal regulation of these enzymes by the primary stimulus. This reciprocal regulation, although identical in all tissues, still imparts a distinctive adaptive strategy in different cell types owing to subtle differences in the network. For example, the inhibition of GS phosphatase by GP in liver can compromise the reciprocal regulation in the absence of liver glycogen (i.e. starved state), while in muscle the reciprocal regulation cannot be compromised owing to an independent inhibitor-1. Our simulation of the glycogen cascade system under starved condition demonstrates that the sensitivity of GS reduces because of the reduction in inhibitor ultrasensitivity caused by GP. The percentage reduction in the inhibition of GS phosphatase is unknown. It is possible that when the liver undergoes a transition from starved to fed state, GS phosphatase can experience varying degrees of inhibition by GP. This results in a shift from a highly futile cycle with no inhibition to reciprocal regulation in the fed state. This causes GS to be always active, while GP is active only in the starved state in the presence of high glucose (see Fig. [5](#F5){ref-type="fig"}). Hallenbeck and Walsh \[[@B40]\] observed that, if the quantity of phosphorylase sequestered in the glycogen particle compartment of rabbit muscle is taken into account, then the local concentration of GP can be very high (up to 2--5 mM). Furthermore, GP interaction with the glycogen particle is known to lower the Michaelis-Menten constants of PK and PP1, thus enhancing the zero order effects further \[[@B29],[@B25]\]. Considering these observations, Meinke and Edstrom \[[@B25]\] estimated an apparent Hill coefficient of 51 for the activation of 3.5 mM phosphorylase. Our simulation results show that at 3.5 mM phosphorylase the system can actually show a highly ultrasensitive response with an apparent Hill coefficient as high as 200 (results not shown). This apparent Hill coefficient value is far higher than any known ultrasensitive system or any of the cooperative enzymes. Though the utility of such a highly sensitive response in vivois unclear at present, various observations indicate that the multi-enzyme cascade system has the potential to exhibit higher sensitivity. Signaling by hormones such as glucagon and epinephrine is known to elicit responses within a fraction of a second, incorporating amplification of the input signal and enhanced sensitivity to allosteric effectors \[[@B2],[@B3],[@B27],[@B41]\]. It has also been shown, in the contraction of resting muscle, that GP-bis converted to GP-awithin a second followed by immediate initiation of glycogenolysis \[[@B3]\]. Such rapid and sensitive responses are known to be the characteristic behavior of enzyme cascades with progressive increase in enzyme concentration down the cascade \[[@B2]\]. This effect can also be brought about by the opposing action of the same effector on modifying and demodifying enzymes \[[@B18]\] and the presence of a stoichiometric inhibitor \[[@B20]\]. It appears that living systems use these ultrasensitive regulatory mechanisms to coordinate multiple input signals, show varied responses to different signals, exhibit rapid responses at an invariably low stimulus concentration \[[@B2],[@B3],[@B27]\] and, most importantly, use negligible amount of cellular energy \[[@B42],[@B43]\]. Theoretical quantification of a regulatory system as presented here reveals insights into system level properties. Ultrasensitivity, signal amplification, flexibility in operation and signal integration are all system level properties, and are not apparent in isolated components. These properties can be studied by connecting different functional units and defining the quantitative relationship between different components of a system. Our simulation results revealed that the switch-like responses of GP and GS in liver and muscle are comparable with that of the MAPK cascade in Xenopusoocytes \[[@B21]\]. At the metabolic level, GP and GS are also regulated by calcium levels and feedback loops constituted by effectors such as ATP, AMP, cAMP, glucose and glycogen \[[@B3]-[@B6],[@B9]-[@B12]\]. Furthermore, GS and PK are known to have multiple phosphorylation sites \[[@B5],[@B9]\]. Regulatory networks made up of multiple feedback loops and multiple phosphorylation cycles, as seen in the activation of maturation-promoting factor and the MAP kinase cascade during oocyte maturation \[[@B44],[@B45]\], can yield multiple steady state responses. Although we have not incorporated the overall regulatory network, our analysis suggests that the enhanced sensitivity observed in the glycogen cascade system may act as a selective pressure in evolution favoring tissue-specific adaptive strategies and compartmental regulatory modules. The abbreviations used are ========================== GP: Glycogen Phosphorylase; GS: Glycogen Synthase; cAMP: cyclic AMP; PP1: Phosphoprotein Phosphatase-1; PK: Phosphorylase Kinase; CAPK: cAMP dependent Protein Kinase; Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= VKM and KVV conceived and designed the experiments. VKM performed the experiments. VKM and KVV analyzed the data. VKM and KVV conceptualized the manuscript. All authors have read and approved the final manuscript. Appendix ======== The following equations were solved in Matlab (The Mathworks Inc. USA) to obtain dose-response curves for fractional steady state activation of component enzymes at various cAMP levels. The steepness of these stimulus dose-response curves can be approximated using the Hill equation. The global output response (fractional modification of phosphorylase and glycogen synthase) can then be quantified in terms of apparent Hill coefficients and half saturation constants, with respect to the input stimulus concentration. Here, the half saturation constant is the amount of input stimulus required for 50% fractional modification of the corresponding protein kinase. Thus, the half saturation constant indicates a mid-point on the unmodified to modified kinase transition curve. The apparent Hill coefficient can also be calculated by estimating the primary input concentration required for 10% to 90% modification of the target enzyme by using the following equation: ![](1742-4682-2-19-i7.gif) where I~0.1~and I~0.9~are the input concentrations required for 10% to 90% modification of target enzyme and ![](1742-4682-2-19-i6.gif) is the apparent Hill coefficient. In the following section we detail the solution strategy employed in simulations. The following equations are derived for the glycogen cascade system schematically shown in Fig [1](#F1){ref-type="fig"}. (I) Activation of cAMP dependent protein kinase (CAPK) by cAMP -------------------------------------------------------------- Two cAMP molecules bind to each R subunit of CAPK (R~2~C~2~) through an infinitely cooperative mechanism and this results in stepwise dissociation of the catalytic subunit \[[@B17]\] ![](1742-4682-2-19-i8.gif) ![](1742-4682-2-19-i9.gif) where K~11~and K~22~are the dissociation constants and Cis a catalytic subunit. ![](1742-4682-2-19-i10.gif) ![](1742-4682-2-19-i11.gif) Mass balance on catalytic subunit yields \[C~t~\] = 2\[R~2~C~2~\] + \[R~2~C(cAMP)~2~\] + \[C\] \[A6\] ![](1742-4682-2-19-i12.gif) Using equations \[A6\] and \[A7\], we obtain a cubic equation for the active catalytic subunit C. ![](1742-4682-2-19-i13.gif) where \[R~2~C~2~\]~t~denotes total CAPK, C is the catalytic subunit, \[cAMP\] is total cAMP concentration. A valid root was obtained as total catalytic subunit concentration of CAPK using Eq. A8, and is taken to be same in both liver and muscle. In the current work, the following cascade-connecting complexes were neglected in the total interconvertable balance: complexes between CAPK catalytic subunits and inhibitor-1, phosphorylase kinase and glycogen synthase in the CAPK balance; inhibitor-1 complex with PP1 in the inhibitor-1 balance; PP1 complexes with phosphorylase kinase, phosphorylase and synthase in the PP1 balance; phosphorylase kinase complexes with phosphorylase and synthase in the phosphorylase kinase balance; liver glycogen phosphorylase complex with PP1 in the phosphorylase balance. We have verified the extent of formation of these complexes and they were found to be negligible compared to the corresponding total interconvertable enzymes. This assumption is valid when the dose-response curve of each target enzyme exceeds 90% phosphorylation \[[@B23]\]. (II) Operating equations for covalent modification cycles \[[@B14]\] involved in regulation of glycogen synthesis and breakdown in the muscle --------------------------------------------------------------------------------------------------------------------------------------------- ### Cubic equation for phosphorylation-dephosphorylation cycle of Inhibitor-1 ![](1742-4682-2-19-i14.gif) where f~1~= I/I~t~, I~t~and Iare total and unphosphorylated inhibitor concentration, (PP2)~t~is total phosphatase 2A, k~1~and k~2~are rate constants for phosphorylation and dephosphorylation of inhibitor-1 respectively. K~m1~and K~m2~are Michaelis-Menten constants for phosphorylation and dephosphorylation of inhibitor-1 respectively. From the constraint 0 \<f~1~\< 1, a valid root was obtained as a fractional unmodified inhibitor using Eq. A9. The fractional phosphorylated inhibitor (i.e. I~p~/I~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i15.gif) where f~2~= I~p~/I~t~* ### Quadratic equation for allosteric interaction of phosphorylated phosphorylase with PP1 \[[@B34]\] ![](1742-4682-2-19-i16.gif) where P~p~is phosphorylated phosphorylase, PP1.P~p~is inactive PP1 and Kdis the dissociation constant. ![](1742-4682-2-19-i17.gif) where (PP1)~t~is total PP1 and f~3~is fractional inactivated PP1 (i.e., (PP1.I~p~)/(PP1)~t~). The fractional free (active) species of PP1 (i.e. f~4~= (PP1)/(PP1)~t~) can be estimated by f~4~= 1-f~3~. ### Cubic equation for phosphorylation-dephosphorylation cycle of phosphorylase kinase ![](1742-4682-2-19-i18.gif) where f~5~= K/K~t~, K~t~and Kare total and unphosphorylated phosphorylase kinase concentration, k~3~and k~4~are rate constants for phosphorylation and dephosphorylation of phosphorylase kinase respectively. K~m3~and K~m4~are Michaelis-Menten constants for phosphorylation and dephosphorylation of phosphorylase kinase respectively. From the constraint 0 \<f~5~\< 1, a valid root was obtained as a fractional unmodified phosphorylase kinase using Eq. A13. The fractional phosphorylated phosphorylase kinase (i.e. K~p~/K~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i19.gif) where f~6~= K~p~/K~t~ ### Cubic equation for phosphorylation-dephosphorylation cycle of phosphorylase ![](1742-4682-2-19-i20.gif) where f~7~= P/P~t~, P~t~and Pare total and unphosphorylated phosphorylase concentrations, k~5~and k~6~are rate constants for phosphorylation and dephosphorylation of phosphorylase respectively. K~m5~and K~m6~are Michaelis-Menten constants for phosphorylation and dephosphorylation of phosphorylase respectively. From the constraint 0 \<f~7~\< 1, a valid root was obtained as a fractional unmodified phosphorylase using Eq. A15. The fractional phosphorylated phosphorylase (i.e. P~p~/P~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i21.gif) where f~8~= P~p~/P~t~ ### Cubic equation for phosphorylation-dephosphorylation cycle of glycogen synthase ![](1742-4682-2-19-i22.gif) where f~9~= S/S~t~, S~t~and Sare total and unphosphorylated glycogen synthase concentrations, k~7~and k~8~are rate constants for phosphorylation and dephosphorylation of glycogen synthase respectively. K~m7~and K~m8~are Michaelis-Menten constants for phosphorylation and dephosphorylation of glycogen synthase respectively. From the constraint 0 \<f~9~\< 1, a valid root was obtained as a fractional unmodified glycogen synthase using Eq. A17. The fractional phosphorylated glycogen synthase (i.e. S~p~/S~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i23.gif) where f~10~= S~p~/S~t~ A plot of fractional activation of catalytic subunit, inhibitor-1 (f~2~), phosphorylase kinase (f~6~), phosphorylase (f~8~) and glycogen synthase (f~10~) at different cAMP input concentrations in muscle is shown in Fig [2](#F2){ref-type="fig"} of the main text. (III) Operating equations for covalent modification cycles involved in regulation of glycogen synthesis and breakdown in liver ------------------------------------------------------------------------------------------------------------------------------ In this case, glucose is considered to be the primary input to the enzyme cascades. Glucose-6-phosphate levels were estimated from various concentration of glucose using the following relationship: ![](1742-4682-2-19-i24.gif) where g6ptrepresents physiological (maximum) concentration of glucose-6-phosphate, g6pis the concentration of glucose-6-phosphate in relation to the concentration of glucose and Kgis an activation constant. Glucose concentration regulates intracellular cAMP levels through hormonal signals such as glucagon. The inverse relationship between glucose and cAMP levels is incorporated by the following equation: ![](1742-4682-2-19-i25.gif) where cAMPtrepresents the physiological (maximum) concentration of cyclic AMP, cAMPis the concentration of cyclic AMP in relation to the concentration of glucose and Kirepresents the inhibitor constant. The superscript 2 and the parameters including Ki, kgand kg2are suitably chosen so that glucose-6-phosphate and cAMP are relatively in the physiological range. cAMP is further taken as an input to CAPK activation. The analytic expression for this interaction is the same as given in Eq. A8. ### Cubic equation for phosphorylation-dephosphorylation cycle of phosphorylase kinase ![](1742-4682-2-19-i26.gif) where f~11~= K/K~t~, K~t~and Kare total and unphosphorylated phosphorylase kinase concentrations, k~3~and k~4~are rate constants for phosphorylation and dephosphorylation of phosphorylase kinase respectively. K~m3~and K~m4~are Michaelis-Menten constants for phosphorylation and dephosphorylation of phosphorylase kinase respectively. From the constraint 0 \<f~11~\< 1, a valid root was obtained as a fractional unmodified phosphorylase kinase using Eq. A21. The fractional phosphorylated phosphorylase kinase (i.e. K~p~/K~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i27.gif) where f~12~= K~p~/K~t~ ### Equations for glucose and glucose-6-phosphate influence on enzyme cascades in liver #### Glucose-6-phosphate inhibition of phosphorylase b phosphorylation ![](1742-4682-2-19-i28.gif) where K~m5~is the Michaelis-Menten constant for phosphorylation of phosphorylase-b, K~m51~represents K~m5~modified by glucose-6-phosphate effects. #### Activation of dephosphorylation of phosphorylase a by glucose ![](1742-4682-2-19-i29.gif) where K~m6~is the Michaelis-Menten constant for dephosphorylation of phosphorylase a, K~m61~represents K~m6~modified by glucose effects. S2 is a multiplicative factor and kgi represents the activation constant. ### Cubic equation for phosphorylation-dephosphorylation cycle of phosphorylase ![](1742-4682-2-19-i30.gif) where f~13~= P/P~t~, P~t~and Pare total and unphosphorylated phosphorylase concentrations, k~5~and k~6~are rate constants for phosphorylation and dephosphorylation of phosphorylase respectively. K~m5~and K~m6~are Michaelis-Menten constants for phosphorylation and dephosphorylation of phosphorylase respectively. From the constraint 0 \<f~13~\< 1, a valid root was obtained as a fractional unmodified phosphorylase using Eq. A25. The fractional phosphorylated phosphorylase (i.e. P~p~/P~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i31.gif) where f~14~= P~p~/P~t~ ### Quadratic equation for allosteric interaction of phosphorylated phosphorylase with PP1 \[[@B34]\] ![](1742-4682-2-19-i32.gif) where P~p~is phosphorylated phosphorylase, PP1.I~p~is inactive PP1 and Kdis the dissociation constant ![](1742-4682-2-19-i33.gif) where (PP1)~t~is total PP1 and f~15~is fractional inactivated PP1 (i.e. (PP1.P~p~)/(PP1)~t~). The fractional free (active) species PP1 (i.e. f~16~= (PP1)/(PP1)~t~) can be estimated by f~16~= 1-f~15~. ### Equations for glucose and glucose-6-phosphate influence on enzyme cascades in liver #### Activation of glycogen synthase dephosphorylation by glucose-6-phosphate ![](1742-4682-2-19-i34.gif) where K~m8~is the Michaelis-Menten constant for dephosphorylation of synthase, K~m81~represents K~m8~modified by glucose-6-phosphate effects. S1 is a multiplicative factor and kg2 represents the activation constant. #### Inhibition of glycogen synthase phosphorylation by glucose-6-phosphatase ![](1742-4682-2-19-i35.gif) where K~m7~is the Michaelis-Menten constant for phosphorylation of synthase, K~m71~represents K~m7~modified by glucose-6-phosphate effects. ### Cubic equation for phosphorylation-dephosphorylation cycle of glycogen synthase ![](1742-4682-2-19-i36.gif) where f~17~= S/S~t~, S~t~and Sare total and unphosphorylated glycogen synthase concentrations, k~7~and k~8~are rate constants for phosphorylation and dephosphorylation of glycogen synthase respectively. K~m7~and K~m8~are Michaelis-Menten constants for phosphorylation and dephosphorylation of glycogen synthase respectively. From the constraint 0 \<f~17~\< 1, a valid root was obtained as a fractional unmodified glycogen synthase using Eq. A31. The fractional phosphorylated glycogen synthase (i.e. S~p~/S~t~) can then be obtained using the following relationship: ![](1742-4682-2-19-i37.gif) where f~18~= S~p~/S~t~ A plot of fractional activation of catalytic subunit, phosphorylase kinase (f~12~), phosphorylase (f~14~) and glycogen synthase (f~18~) at different glucose, glucose-6-phosphate and cAMP concentrations in liver is shown in Fig [3](#F3){ref-type="fig"} and [4](#F4){ref-type="fig"} of the main text. ### Parameters from the literature used for the simulations #### Rate Constants k1 = 1.4 sec^-1^rate constant for phosphorylation of inhibitor \[[@B48]\] k2 = 0.01 sec^-1^rate constant for dephosphorylation of inhibitor \[assumed\] k3 = 20 sec^-1^rate constant for phosphorylation of phosphorylase kinase \[assumed\] k4 = 5 sec^-1^rate constant for dephosphorylation of phosphorylase kinase \[assumed\] k5 = 20 sec^-1^rate constant for phosphorylation of Phosphorylase \[[@B42]\] k6 = 5 sec^-1^rate constant for dephosphorylation of Phosphorylase \[[@B49]\] k7 = 20 sec^-1^rate constant for phosphorylation of glycogen synthase \[assumed\] k8 = 0.05 sec^-1^rate constant for dephosphorylation of glycogen synthase \[assumed\] #### Michaelis-Menten constants Km1 = 5 μM for inhibitor phosphorylation \[[@B48]\] Km2 = 0.7 μM for dephosphorylation of Inhibitor \[[@B52]\] Km3 = 0.4 μM for Phosphorylation of phosphorylase kinase \[assumed\] Km4 = 1.1 μM for dephosphorylation of phosphorylase kinase \[[@B52]\] Km5 = 10 μM for phosphorylation of phosphorylase \[[@B25]\] Km6 = 5 μM for dephosphorylation of phosphorylase \[[@B47]\] Km7 = 15 μM for phosphorylation of glycogen synthase \[assumed\] Km8 = 0.12 μM for dephosphorylation of glycogen synthase \[[@B50]\] Kd = 0.002 dissociation of PP1 and phosphorylated PP1 Inhibitor, and also phosphorylase a with synthase PP1 \[[@B47]\] #### Total Concentrations capkt = 0.25 μM total R2C2 ie. cAMP dependent protein kinase, CAPK \[[@B3]\] It = 1.8 μM total Inhibitor concentration \[[@B3]\] kt = 2.5 μM total Phosphorylase kinase \[[@B35]\] pt = 70 μM total Glycogen Phosphorylase \[[@B3]\] st = 3 μM total Glycogen synthase \[[@B3]\] PP1 = 0.25 μM PTPase 1 \[[@B33]\] PP2A = 0.025 μM PTPase 2 \[[@B3]\] #### Other parameters: (chosen as per various qualitative observations are in physiological ranges as given in \[[@B3],[@B9],[@B12],[@B17],[@B25],[@B27],[@B29],[@B33],[@B35],[@B42],[@B46]-[@B54]\]) k11 = 0.043 μM Dissociation constant of cAMP \[[@B35]\] k22 = 0.7 μM Dissociation constant of cAMP \[[@B35]\] ki = 100 μM cAMP inhibition constant campt = 10 μM maximum cAMP \[[@B3]\] kg = 349500 μM activation constant of glucose-6-phosphate for synthase PP1 g6pt = 700 μM maximum glucose-6-Phosphate \[[@B33]\] kgi = 10000 μM activation constant of glucose for phosphorylase phosphatase s1 = 100 a multiplicative factor for glucose-6-phosphate effect on glycogen synthase dephosphorylation kg2 = 500 μM inhibition due to glucose-6-phosphate = 0.05 mM s2 = 0.0010 a multiplicative factor for glucose effect on phosphorylase phosphatase Sensitivity analysis -------------------- The above parameter set was used for simulating the dose-response curves of the glycogen cascade system. To assess the sensitivity to variation in individual parameters, each parameter was varied over a 10-fold change while holding all other parameters constant. The response sensitivity is quantified using a Hill coefficient and is given in Table [1](#T1){ref-type="table"}. The results indicate that at different parameter sets, the output responses of GP and GS are switch-like and display different degrees of signal amplification. Acknowledgements ================ KVV acknowledges financial support from the Swarnajayanti fellowship, Department of Science and Technology, India.	PubMed Central	2024-06-05T03:55:59.572233	2005-5-20	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180476/", "journal": "Theor Biol Med Model. 2005 May 20; 2:19", "authors": [ { "first": "Vivek K", "last": "Mutalik" }, { "first": "KV", "last": "Venkatesh" } ] }
PMC1180477	Background ========== Mammalian reoviruses are the prototypic members of the Reoviridaefamily, which also includes the pathogenic rotaviruses, coltiviruses, seadornaviruses and orbiviruses. These viruses share elements of their replication cycle as well as structural features, including a non-enveloped multi-layered capsid that surrounds a segmented dsRNA genome. In humans, mammalian reoviruses are typically associated with mild and self-limiting enteric and respiratory infections. However, studies in neonatal mice reveal that reoviruses can spread to distant tissue sites in immunocompromised hosts (reviewed in\[[@B1]\]). The factors that determine reovirus cellular host range are poorly understood. Because reovirus attaches to cells through interactions with broadly expressed receptors, one or more subsequent steps in the viral life cycle must help to regulate host range and pathogenesis. Our recent studies suggest that one such step is proteolysis of the capsid protein σ3 \[[@B2],[@B3]\]. In cell culture, the first step in infection is attachment to cellular receptors through interactions with the viral protein σ1 \[[@B4],[@B5]\]. σ1 interacts with two known receptors: sialic acid and junctional adhesion molecule 1 \[[@B6]-[@B8]\]. Following binding, virions are internalized by receptor-mediated endocytosis \[[@B9]\]. Endocytosis is an essential step in the viral life cycle under standard infection conditions \[[@B10]\]. Within the endosomal and/or lysosomal compartment, proteases convert virions into particles that resemble in vitro-generated intermediate subvirion particles (ISVPs) \[[@B10]-[@B14]\]. These uncoating intermediates, typically prepared using chymotrypsin or trypsin, lack σ3 and have a cleaved form of μ1. Studies using ISVPs and ISVPs recoated with recombinant outer capsid proteins reveal that σ3 plays a key role in regulating reovirus cell entry by interacting with, protecting, and controlling the conformational status of the underlying penetration protein μ1 \[[@B15]-[@B18]\]. In cells that cannot efficiently mediate σ3 degradation during uncoating, reovirus infection is slow or blocked; these cells can be productively infected by particles that lack σ3 \[[@B2]\]. In vitro, ISVP-like particles can be generated by a variety of proteases in addition to chymotrypsin and trypsin, including proteinase K, thermolysin, endoproteinase lys-C, Cat L, Cat B and Cat S\[[@B3],[@B19]-[@B21]\]. Recent work has provided insight into the cellular determinants of reovirus uncoating. In murine fibroblasts, where reovirus entry has been best studied, the cysteine proteases Cat L, and to a lesser extent Cat B, are required for σ3 removal, whereas the aspartyl protease Cat D is not \[[@B14],[@B21]-[@B25]\]. Virion disassembly in murine fibroblasts also requires acidic pH\[[@B10],[@B26],[@B27]\]. Recently, we demonstrated that reovirus uncoating in the macrophage-like cell line P388D is mediated by the acid-independent lysosomal cysteine protease Cat S\[[@B3]\]. This finding revealed that in different cell types, distinct proteases can facilitate reovirus uncoating. Our results suggested a model in which infection in some cells is acid-dependent because the proteases that mediate σ3 removal in those cells require acidic pH for maximal activity. Thus, in fibroblasts or other cells in which the acid-dependent proteases Cat L and Cat B mediate σ3 removal, infection is acid-dependent \[[@B21],[@B23],[@B28]\], whereas in Cat S-expressing cells it is not \[[@B3]\], because Cat S maintains its activity at neutral pH \[[@B29]\]. Insight from the analysis of reovirus cell entry facilitated the recent discovery that activation of the Ebola virus glycoprotein also depends on the activity of the acid-dependent endosomal proteases Cat B and Cat L \[[@B30]\]. The role that specific intracellular and extracellular proteases play in regulating reovirus tropism, spread, and disease in animals is largely unknown, except in the murine intestinal tract where pancreatic serine proteases have been shown to mediate σ3 removal \[[@B31],[@B32]\]. Reovirus also naturally infects hosts via the respiratory tract \[[@B33]-[@B35]\]. One protease with well-described effects in the respiratory tract is elastase 2 (GenBank NM\_001972), an inflammatory serine protease of the chymotrypsin family, which is predominantly expressed by neutrophils \[[@B36]\]. NE plays a prominent role in wound repair \[[@B37]-[@B39]\] and in controlling microbial infections \[[@B38]-[@B40]\]. NE expression can also promote pathogenesis; it has been implicated in smoke-induced emphysema \[[@B41]\], respiratory syncytial viral bronchiolitis \[[@B42]\] and in the respiratory syndrome ARDS \[[@B4]\]. The fact that reovirus replication in the rodent lung causes an influx of neutrophils \[[@B35],[@B43]\] and that reovirus infection can recapitulate ARDS \[[@B44]\], led us to ask whether NE could mediate productive reovirus uncoating. We investigated reovirus infection in the monocyte-like cell line U937, because it is known to express NE \[[@B45]\]. Experiments described in this report demonstrate that reovirus infection in U937 cells does not require cysteine protease activity and is not blocked in the presence of agents that raise vesicular pH. Studies using protease inhibitors suggest that, in the absence of cysteine protease activity, NE is largely responsible for productive infection of U937 cells. NE can directly mediate σ3 removal from reovirus virions; the resultant particles are infectious and do not require additional intracellular proteolysis. Our data raise the possibility that NE is involved in reovirus replication in the respiratory tract. Results ======= Reovirus infection of U937 cells does not require cysteine protease activity ---------------------------------------------------------------------------- The promonocytic cell line U937 expresses large amounts of elastase \[[@B45]\] and provided a suitable system to analyze the role of this protease in reovirus infection. To determine if NE can facilitate reovirus infection of U973 cells, we first established conditions under which lysosomal cysteine protease activity was inhibited. Cells were treated with 300 μM E64, a broad-spectrum cysteine protease inhibitor \[[@B46]\], and protease activity was assessed using the Cat L and Cat B-specific fluorogenic substrate Z-Phe-Arg-MCA. We analyzed enzyme activity at two time points: first after 3 h of treatment, because we typically pre-treat cells with inhibitors for 3 h prior to infection, and second at 3 d, the time point at which viral yield would be quantified. As shown in Fig. [1A](#F1){ref-type="fig"}, treatment with 300 μM E64 completely abolished cysteine protease activity in U937 cells. Consistent with our previous findings \[[@B3]\], E64 also completely blocked cysteine protease activity in L929 cells. Raw values are provided, to illustrate the relative difference in Cat L/B enzyme activity levels between U937 cells and L929 fibroblasts. In the absence of inhibitor, Cat L and B activity was significantly lower in U937 cells than in L929 cells. This may be a consequence of high expression in U937 cells of cystatin F, an intracellular cysteine protease inhibitor with specificity for Cat L and papain \[[@B47]\]. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Analysis of viral replication in L929 and U937 cells treated with E64.A. 3 × 10^6^L929 and U937 cells were untreated (-; black) or treated (+; grey) with 300 μM E64 for 3 h or 3 d. Cysteine protease activity was assessed using the fluorogenic substrate Z-Phe-Arg-MCA (Sigma) and plotted in arbitrary units. Activity levels in treated cells were so low (in L939 cells, 254 units at 3 h and 231 units at 3 d; in U937 cells, 200 units at 3 h and 115 units at 3 days) that they cannot be visualized on this graph. B. L929 (L; black bars) and U937 (U; grey bars) cells were treated with 300 μM E64 for 3 h prior to infection. Cells were then infected with reovirus strain Lang virions or ISVPs at an MOI of 3. Infectious virus present at 3 d p.i. was determined by plaque assay on L929 cell monolayers. Each time point represents the mean (+/- SD) derived from three independent samples. ::: ![](1743-422X-2-48-1) ::: Next, we compared reovirus replication in E64-treated U937 and L929 cells. Cells were pre-treated for 3 h and infected with Lang virions or ISVPs at a multiplicity of infection (MOI) of 3. The results of a representative experiment are shown in Fig. [1B](#F1){ref-type="fig"}. In the absence of E64, both L929 and U937 cells supported reovirus replication, consistent with the fact that these cells express Cat L. As expected, E64 blocked virion infection of L929 cells; however, viral yields in E64-treated U937 cells were only slightly reduced relative to untreated cells. ISVPs, which lack capsid protein σ3, replicated efficiently in treated cells, indicating that 300 μM E64 was not toxic to either cell type. These results demonstrate that productive infection of U937 cells by Lang virions does not require the activity of E64-sensitive, papain-like cysteine proteases. Infection of U937 cells is acid-independent ------------------------------------------- Acidic pH is required for productive reovirus infection of murine L929 fibroblasts \[[@B10],[@B27]\], in which the acid-dependent proteases Cat L and Cat B mediate uncoating \[[@B21],[@B23]\]. Serine proteases, including NE, and metalloproteases function over a broader pH range. Therefore, to gain insight into the nature of the protease(s) that can promote reovirus uncoating in U937 cells, we investigated the requirements for acidic pH. L929 and U937 cells were left untreated or pre-treated with E64 in the presence or absence of bafilomycin A1 (Baf) or NH~4~Cl. These latter agents raise vesicular pH by blocking the vacuolar H^+^-ATPase pump or by acting as a weak base, respectively \[[@B48]-[@B50]\]. After pre-treatment, cells were infected with Lang virions at an MOI of 3 and viral yields were determined at 3 days post infection (d p.i.). A representative experiment is shown in Fig. [2](#F2){ref-type="fig"}. Treatment with either Baf or NH~4~Cl did not inhibit viral replication in U937 cells; yields reached 2.9 and 2.7 logs, respectively. Furthermore, these agents had little effect on viral replication in U937 cells even when the cells were also treated with E64 to inhibit cysteine protease activity. In contrast, Baf or NH~4~Cl alone completely blocked reovirus replication in L929 cells, consistent with the requirement for Cat L/B-mediated σ3 removal in these cells. Given that reovirus uncoating is an essential step in the viral life cycle \[[@B10]\], these findings revealed that a non-cysteine protease that functions at neutral pH can facilitate this step in U937 cells. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Effects of agents that raise vesicular pH on reovirus replication in U937 and L929 cells.A. U937 cells were pre-treated without (-; black bars) or with 300 μM E64 (E64; grey bars) in the presence or absence of 25 nM Baf or 20 mM NH~4~Cl. Following pre-treatment, cells were infected with reovirus strain Lang at an MOI of 3 and viral yield was measured at 3 d.p.i as described in the legend to Fig. 1B. B. L929 cells were pre-treated without (-) or with E64, 25 nM Baf or 20 mM NH~4~Cl. Pre-treated cells were infected with reovirus strain Lang at an MOI of 3 and viral yield was measured at 3 d p.i. as described in the legend to Fig. 1B. ::: ![](1743-422X-2-48-2) ::: Analysis of reovirus replication in U937 cells differentiated by PMA -------------------------------------------------------------------- Treatment of the promonocytic U937 cells with phorbol ester derivatives results in their differentiation into macrophage-like cells \[[@B51],[@B52]\]. This differentiation is characterized by several major phenotypic changes, including increases in expression of urokinase plasminogen activator receptors, upregulation of collagenase activity and a significant decrease in the expression of NE and Cat G \[[@B51],[@B52]\]. We predicted, therefore, that PMA treatment might decrease the capacity of reovirus virions to replicate in U937 cells when cysteine proteases were inhibited. To confirm that there was a significant decrease in NE expression in U937 cells differentiated with PMA, U937 cells were treated with 150 nM PMA for 72 h and expression of NE was analyzed by immunoblotting. As shown in Fig. [3A](#F3){ref-type="fig"}, NE was expressed in untreated U937 cells, but its expression was dramatically reduced following PMA-induced differentiation. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Analysis of reovirus replication in U937 cells differentiated with PMA.A. Lysates generated from 10^5^U937 cells that were untreated (-) or treated with 150 nM PMA for 72 h were resolved on SDS-12% polyacrylamide gels and electrophoretically transferred to a nitrocellulose filter. The filter was subsequently incubated with a polyclonal goat antibody against human NE (1:400) (Santa Cruz Biotechnology). The filter was washed and incubated with a secondary anti-goat antibody conjugated to horseradish peroxidase (1:5000) (Santa Cruz Biotechnology). Protein bands were detected using reagents that generate a chemiluminescent signal (Amersham). B. U937 cells that were undifferentiated (-; black bars) or differentiated (PMA; grey bars) with 150 nM of PMA for 72 h were left untreated (-) or were treated with 300 μM E64. Following pre-treatment with the protease inhibitor, cells were infected with Lang virions or ISVPs at an MOI of 3. Viral yield was quantified at 3 d p.i. as described in the legend to Fig. 1B. ::: ![](1743-422X-2-48-3) ::: To examine the effect of U937 cell differentiation on reovirus infection, PMA-treated and untreated U937 cells were left untreated or were treated with E64 for 3 h and infected with Lang virions or ISVPs at an MOI of 3. Yields were measured at 3 d p.i. and the results of a typical experiment are shown in Fig. [3B](#F3){ref-type="fig"}. In the absence of E64, PMA-treated U937 cells were permissive to infection by virions. PMA treatment only decreased yields by \~0.5 log relative to untreated cells. In contrast, when PMA-differentiated U937 cells were treated with E64 to inhibit cysteine protease activity, they no longer supported productive infection by Lang virions. Because these results could be explained if E64 was toxic to PMA-treated U937 cells, we examined the replication of ISVPs. In the presence of E64, ISVPs replicated to high yields in both undifferentiated and differentiated U937 cells. Since PMA-induced differentiation of U937 cells caused a substantial decrease in NE expression, these results are consistent with the hypothesis that NE or another similarly regulated neutral protease facilitates productive reovirus infection in promonocytic (pre-differentiated) U937 cells. NE can facilitate reovirus infection in U937 cells -------------------------------------------------- We directly examined the capacity of NE to facilitate reovirus infection by using the irreversible elastase inhibitor, N-(methoxysuccinyl)-Ala-Ala-Pro-Val-chloromethyl ketone \[[@B53]\]. This inhibitor is highly specific for NE and does not inhibit the activity of the related serine protease, Cat G \[[@B53]\]. First, we established the efficacy and specificity of inhibitor treatment under our experimental conditions. U937 cells were treated with the NE inhibitor, E64, Baf or NH~4~Cl for either 3 h or 2 d and the activity of NE in cell lysates was examined using a colorimetric substrate. As shown in Table [1](#T1){ref-type="table"}, the NE inhibitor was active at both time points. In cells treated with the specific inhibitor, NE activity was less than 9% of that in untreated U937 cells. In contrast, in U937 cells treated with E64, Baf or NH~4~Cl, NE activity was only modestly reduced, remaining above 80% even after 2 d. These results are consistent with the capacity of NE to function at neutral pH. To verify the specificity of the NE inhibitor, we also examined its effect on Cat L/B activity using the fluorogenic substrate Z-Phe-Arg-MCA. As expected, Cat L/B activity was completely inhibited by E64 but largely unaffected by the NE inhibitor. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Inhibition of NE activity in U937 cells.^a^ ::: \% NE Activity^b^ \% Cat Activity^c^ ----- ------------------- -------------------- ------------- ------------ ------------ ------------ 3 h 7.5 ± 5.0 84.3 ± 5.8 87.0 ± 9.0 88.3 ± 9.4 ND ND 2 d 8.8 ± 3.5 89.5 ± 9.1 88.8 ± 10.0 86.3 ± 3.6 -4.7 ± 1.6 98.8 ± 3.9 ^a^U937 cells were treated with the indicated inhibitors for 3 h or 2 d. ^b^NE activity was assessed using the colorimetric substrate MeOSuc-Ala-Ala-Pro-Val-ρNA and percent activity relative to untreated cells was calculated. ^c^Cathepsin L and B activity were assessed using the fluorogenic substrate Z-Phe-Arg-MCA and percent activity relative to untreated cells was calculated. ::: To examine the effect of the NE inhibitor on reovirus replication in U937 cells, we pre-treated them for 3 h with E64 in the presence or absence of the NE inhibitor, infected them with Lang virions or ISVPs at an MOI of 3, and quantified viral yields at 2 d p.i. A representative experiment is shown in Fig. [4](#F4){ref-type="fig"}. Consistent with the results shown in Fig [1](#F1){ref-type="fig"}, virion replication was not blocked in E64-treated U937 cells. However, in the presence of both E64 and the NE inhibitor, yields were significantly reduced. ISVPs replicated to high yields in treated cells, indicating that the combination of inhibitors was not toxic to U937 cells. These results demonstrate that NE plays a critical role in reovirus infection of U937 cells when cysteine proteases are inhibited. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Analysis of viral replication in U937 cells treated with E64 and NE inhibitor.U937 cells were pre-treated with 300 μM E64 in the absence (-; black bars) or presence (NEI; grey bars) of 200 μM N-Ala-Ala-Pro-Val chloromethylketone, an inhibitor of NE. Following inhibitor pre-treatment, cells were infected with Lang virions or ISVPs at an MOI of 3. Viral yield was analyzed at 2 d p.i. as described in the legend to Fig. 1B. ::: ![](1743-422X-2-48-4) ::: NE-generated subviral particles are infectious and do not require additional proteolytic processing --------------------------------------------------------------------------------------------------- NE, like many cellular proteases, is expressed as a proenzyme that becomes activated only after its pro-region is removed \[[@B54]\]. We envisioned two models by which NE could facilitate reovirus infection of U937 cells. In the first, NE could directly mediate σ3 degradation, leading to the generation of an ISVP-like particle. In the second, NE could act indirectly by activating another protease. To try to distinguish between these models, we examined the capacity of purified NE to directly mediate σ3 removal from Lang virions in vitro. Purified Lang virions were treated with NE for 1 and 4 h and the treated virus particles were analyzed by SDS-PAGE. As shown in Fig. [5A](#F5){ref-type="fig"}, NE efficiently removed σ3 from Lang virions; after 1 h very little intact σ3 remained on viral particles. After 4 h of NE treatment, σ3 was completely removed and the underlying μ1C was cleaved to the δ and φ fragments (φ was not retained on the gel). When we assayed the infectivity of the resultant particles by plaque assay we found that NE treatment did not negatively affect the titer of Lang particles (data not shown). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Analysis of NE treatment of reovirus virions.A. Lang virions (7.5 × 10^10^) were incubated with purified 25 μg/ml NE (Calbiochem) at 37°C for the indicated times (in min) and analyzed on SDS-12% polyacrylamide gels and proteins were visualized by Coomasssie staining. The mock sample (M) consisted of virions held in reaction buffer in the absence of protease for 4 h. The positions of reovirus capsid proteins are labeled. B. L929 cells were left untreated (-) or were pre-treated with 300 μM E64 and infected with Lang virions (black bars), ISVPs (white bars) or NE-treated particles (grey bars) at an MOI of 3. Viral yield was determined at 1 d p.i. C. L929 cells were left untreated (-) or were pre-treated with 20 mM NH~4~Cl, 25 nM Baf or 25 μM monensin and infected with Lang virions or protease-treated particles. Cell extracts prepared at 1 d p.i. were analyzed for the expression of the viral non-structural protein μNS by immunoblotting using antiserum against μNS (diluted 1:12500). ::: ![](1743-422X-2-48-5) ::: To determine if NE-generated SVPs required further proteolytic processing of σ3, L929 cells were pre-treated with E64 to block cysteine protease activity and infected at an MOI of 3 with Lang virions, ISVPs or NE-generated subviral particles (NE-SVPs). Viral yields were determined at 1 d p.i. As expected, E64 blocked infection of L929 cells by virions. In contrast, both ISVPs and NE-SVPs replicated efficiently in the presence of the cysteine protease inhibitor (Fig. [5B](#F5){ref-type="fig"}). Because virion disassembly in L929 cells requires acidic pH \[[@B10]\], we also examined the capacity of NE-SVPs to infect L929 cells treated with Baf, NH~4~Cl or monensin, three agents that raise vesicular pH by distinct mechanisms. Cells were treated with these agents and then infected with virions, ISVPs or NE-SVPs at an MOI of 10. At 18 hours post infection (h p.i.), cell lysates were harvested and expression of the reovirus non-structural protein μNS was analyzed by immunoblotting (Fig. [5C](#F5){ref-type="fig"}). As expected, when treated cells were infected with virions, viral protein expression was blocked. In contrast, μNS expression was evident even in the presence of agents that raise pH when infections were initiated with ISVPs or NE-SVPs (Fig. [5C](#F5){ref-type="fig"}). Together, these results demonstrate that NE can directly mediate σ3 removal from virions to generate infectious particles that do not require further proteolytic processing by acid-dependent cysteine proteases in L929 cells. Discussion ========== NE, an acid-independent serine protease, can promote productive reovirus infection in U937 promonocytes ------------------------------------------------------------------------------------------------------- Serine proteases are involved in reovirus infection in the mammalian intestinal tract \[[@B31]\] and in this report we provide evidence that they can mediate uncoating and promote infection in U937 cells. This expands the range of proteases that promote reovirus infection in cell culture to include NE as well as the cysteine proteases Cat L, Cat B, and Cat S. Several lines of evidence now support the notion that protease expression is a cell-specific host factor that can impact reovirus infection. For example, some reovirus strains are inefficiently uncoated by Cat S and thus do not replicate to high yield in P388D macrophages \[[@B3]\]. In this report we demonstrate that PMA-induced differentiation influences the type of protease that mediates reovirus uncoating in U937 cells. In these cells, PMA treatment is reported to increase Cat L expression \[[@B55]\] and decrease expression of the serine proteases NE and Cat G \[[@B56],[@B57]\]. Accordingly, when we used PMA to induce U937 cell cultures to differentiate, reovirus infection became sensitive to the cysteine protease inhibitor E64. We suspect that Cat L is largely responsible for uncoating in these PMA-differentiated cells, but the acid-independent protease Cat S may also play a role. We are currently addressing this question by analyzing infection in PMA-differentiated cells treated with either Baf or NH~4~Cl. Does the serine protease Cat G also play a role in reovirus infection of U937 cells? ------------------------------------------------------------------------------------ Our data do not completely resolve this question. Cat G is expressed by U937 cells and, like NE, it is down-regulated by PMA treatment. Furthermore, we found that in vitrotreatment of reovirus virions with purified Cat G generates SVPs that behave like NE-SVPs in that they are infectious in the absence of further proteolytic processing (data not shown). Results of our experiment with the NE-specific inhibitor suggest that NE is largely responsible for the E64-resistant infection in U937 cells. While this inhibitor is reported not to inhibit Cat G \[[@B53]\], we have not independently confirmed this. Another approach to assess the role of Cat G in reovirus infection of U937 cells would be to examine the effect of Cat G-specific inhibitors on infection. We tried one such inhibitor, Cathepsin G Inhibitor I (Calbiochem)\[[@B58]\], but found that it was cytotoxic to U937 cell cultures. Given that both NE and Cat G can generate infectious reovirus SVPs, more work needs to be done in order to understand the role that these two proteases play in infection in these cells. The acid-dependence of reovirus infection is a reflection of the requirements for protease activation ----------------------------------------------------------------------------------------------------- Previously, we reported that virion uncoating mediated by Cat S does not require acidic pH \[[@B3]\]. These results were consistent with the acid-independence of Cat S activity \[[@B37]\]. Together, the results in Fig. [2](#F2){ref-type="fig"} and Fig. [4](#F4){ref-type="fig"} reveal that, like Cat S, NE-mediates infection in an acid-independent manner. This finding thus provides further support for a model in which the requirement for acidic pH during reovirus infection of some cell types reflects the requirement for acid-dependent protease activity in those cells rather than some other requisite acid-dependent aspect of cell entry. The small effect of Baf and NH~4~Cl on E64-resistant reovirus growth (Fig. [2](#F2){ref-type="fig"}) may reflect the participation of one or more acid-dependent proteases (such as Cat D) in the activation of NE. Does NE mediate uncoating intracellularly or extracellularly in U937 cell cultures? ----------------------------------------------------------------------------------- Elastase is stored in azurophilic granules that are the major source of acid-dependent hydrolases in neutrophils \[[@B59]\]. Although these granules do not contain LAMP-1 or LAMP-2 \[[@B60]\] they contain the lysosomal markers LAMP-3 \[[@B61]\] and CD68 \[[@B62]\] and are accessible to endocytosed fluid-phase markers under conditions of cellular stimulation \[[@B63]\]. NE can be released from neutrophils during degranulation \[[@B64]\] and its cell surface expression can be induced upon PMA treatment \[[@B65]\]. However, studies in U937 cells have shown that NE is predominantly retained intracellularly and that little if any activity is present in the extracellular medium \[[@B45]\]. Consistent with this, we have been unable to generate ISVP-like particles by treatment of virions with U937 culture supernatants (data not shown). This observation, together with our finding that PMA treatment decreases the capacity of E64-treated U937 cells to support reovirus infection, leads us to favor a model in which NE-mediated virion uncoating in U937 cell cultures occurs intracellularly. Implications for infection in the host -------------------------------------- In vivo, a number of viruses, including dengue and respiratory syncytial virus, induce the release of IL-8, a cytokine that serves as a chemoattractant for neutrophils and promotes their degranulation \[[@B66],[@B67]\]. Reovirus replication in the rat lung results in neutrophilic invasion \[[@B35],[@B43]\] and studies in cell culture indicate that reovirus infection can induce IL-8 expression \[[@B68]\]. Thus, the capacity of reovirus to induce IL-8 secretion in vivomight facilitate the release of neutrophilic lysosomal hydrolases, including NE, into the extracellular milieu. In this report, we have shown that mammalian reovirus can utilize this acid-independent serine protease for uncoating. Our data suggest that, in vivo, one consequence of reovirus-induced IL-8 expression would be the generation of infectious NE-SVPs. Like ISVPs, these particles would be predicted to have an expanded cellular host range because they can infect cells that restrict intracellular uncoating \[[@B2]\]. Thus, inflammation might be predicted to exacerbate reovirus infection by promoting viral spread. Future studies using mice with deletions in the NE gene will be required to elucidate the role this protease plays during reovirus infection in the respiratory tract and other tissues. Finally, given the recent finding that endosomal proteolysis of the Ebola virus glycoprotein is necessary for infection \[[@B30]\], our results raise the interesting possibility that NE or other neutrophil proteases may play a role in cell entry of other viruses. Methods ======= Cells and viruses ----------------- Murine L929 cells were maintained as suspension cultures as described previously \[Kedl, 1995 \#94\]. U937 cells were maintained in RPMI medium (GIBCO-BRL, Gaithersburg, MD) supplemented to contain 10% fetal calf serum (Gibco-BRL), 50-units/ml penicillin (GIBCO-BRL), 50 μg/ml streptomycin (GIBCO-BRL) and 2 mM glutamine (GIBCO-BRL). Where indicated, U937 cells were differentiated by treatment with 150 nM of PMA (Sigma) for 48 h prior to infection. Third-passage lysate stocks of reovirus were prepared in L929 cell cultures. Purified virions were prepared by CsCl density gradient centrifugation of extracts from cells infected with third-passage lysate stocks \[Furlong, 1988 \#81\]. ISVPs were prepared by treating purified virions with chymotrypsin as described elsewhere \[Nibert, 1992 \#95\]. Measurement of cysteine protease activity ----------------------------------------- Cysteine protease activity was measured as described previously \[[@B23]\] with some minor modifications. Briefly, P388D U937 and L929 cells (2 × 10^6^each) were incubated in the presence or absence of 300 μM E-64, 5 nM LHVS, or 5 μM CA074 for the times indicated. After incubation, cells were trypsinized, collected by centrifugation at 179 × gfor 10 min at 4°C and washed once in PBS. Cell pellets were resuspended in 100 μl of lysis buffer (100 mM sodium acetate \[pH 5.5\], 1 mM EDTA, and 0.5% Triton X-100), incubated on ice for 30 min and cell debris was pelleted by centrifugation at 89 × gfor 10 min at 4°C. For each sample, 20 μl of clarified cell lysate was added to 80 μl of reaction buffer (100 mM sodium acetate \[pH 5.5\], 1 mM EDTA, 4 mM dithiothreitol) in a well of a black 96-well plate (Corning). To measure Cat B activity, 100 μM Z-Arg-Arg-7-amido-4-methylcoumarin (Z-Arg-Arg MCA) (Calbiochem) was included in the reaction buffer. To measure Cat L and Cat B activity, 100 μM Z-Phe-Arg-7-amido-4-methylcoumarin (Z-Phe-Arg-MCA) (Calbiochem) was added to the reaction buffer. Reactions were incubated for 30 min at room temperature with gentle tapping every 10 min. Fluorescence was measured using an FL600 microplate reader (Bio-Tek Instruments, Inc., Winooski, VT) with an excitation of 390 nm and emission at 460 nm. Measurement of serine protease activity --------------------------------------- NE activity was determined by incubating 3 × 10^6^U937 cells in the presence or absence of 200 μM NE inhibitor (N-(methoxysuccinyl)-Ala-Ala-Pro-Val-chloromethyl ketone) (Sigma) for the indicated times. After treatment, cells were collected by centrifugation at 179 × g for 10 min at 4°C, washed twice in PBS and lysed in TLB (10 mM Tris \[pH 7.5\], 2.5 mM MgCλ~2~, 100 NaCl, 0.5% Triton X-100, 5 μg/μl of leupeptin \[Sigma\], 1 mM PMSF) for 30 m on ice. The lysate was clarified by centrifugation at 89 × gfor 10 min at 4°C. For each sample, 20 μl of cell lysate was added to 80 μl of virion dialysis buffer (VDB) (150 mM NaCl, 10 mM MgCl~2~, 10 mM Tris \[pH 7.5\]) containing 500 μM of NE substrate (MeOSuc-Ala-Ala-Pro-Val-pNA) (Calbiochem) and incubated for 30 min at room temperature with gentle tapping every 10 min. Absorbance was measured at 405 nm using an EL340 BioTek microplate reader (Bio-Tek Instruments). Analysis of viral growth ------------------------ Cells were infected (in triplicate) at the indicated MOI and adsorption was allowed to proceed for 1 h on ice at 4°C. After adsorption, cells were pelleted by low speed centrifugation and resuspended in fresh media. Virus and cells were then added to dram vials (2 × 10^5^cells/vial) containing 1 ml of chilled medium. Prior to infection, some cells were pre-treated for 3 h with 300 μM E64 and/or 25 nM Baf (Sigma), 20 mM NH~4~Cl (Sigma) and 200 μM NE inhibitor. Inhibitors were included in the medium throughout the time course for treated samples. Time zero samples were immediately frozen at -20°C and remaining samples were incubated at 37°C until the desired time point was reached. Samples were frozen and thawed three times and titrated by plaque assay on L929 cells as described elsewhere \[[@B69]\]. Viral yields were calculated according to the following formula: log~10~(PFU/ml)~t\ =\ x\ hr~- log~10~(PFU/ml)~t\ =\ 0~+/- standard deviation (SD). In vitroanalysis of NE-mediated uncoating ------------------------------------------- NE digestions were performed as follows. Purified virions (7.5 × 10^10^) were incubated with 25 μg/ml of purified NE (Calbiochem) in 20 μL VDB at 37°C for the times indicated. Mock-treated samples were incubated in VDB for the longest time point. 1 mM PMSF and 200 μM of NE inhibitor were added to the samples to terminate the reactions. Protein sample buffer (0.125 M Tris \[pH 8.0\], 1% SDS, 0.01% bromphenol blue, 10% sucrose, and 5% β-mercaptoethanol) was added to each reaction mixture and samples were resolved on SDS-12% polyacrylamide gels. The protein gels were stained with Coomassie Brilliant Blue. Immunoblot analysis of NE expression ------------------------------------ To analyze NE expression, cell lysates were generated from U937 cells, either treated or untreated for 48 h with 150 nM PMA as described for the analysis of viral protein expression. Lysate from the equivalent of 1 × 10^6^cells was run on SDS-12% polyacrylamide gels and transferred to nitrocellulose. Membranes were blocked overnight in TBST containing 10% nonfat dry milk. NE expression was analyzed using a polyclonal antibody against NE (1:400 in TBST) (Santa Cruz Biotechnology Inc, Santa Cruz, CA). Membranes were washed with TBST and incubated with a horseradish peroxidase-conjugated anti-goat IgG (1:5000 in TBST). Bound antibody was detected by treating the nitrocellulose filters with enhanced chemiluminescence (ECL) detection reagents (Amersham) and exposing them to Full Speed Blue X-ray film (Henry Schein, Melville, NY). Analysis of viral protein expression in infected cells ------------------------------------------------------ Cells were plated at 10^6^/well in a 6-well plate 18--24 h prior to infection. Virus was allowed to adsorb to cells for 1.5 h at 4°C. At this temperature, virus binds to cells but is not internalized \[[@B70]\]. After adsorption, the cultures were incubated at 37°C in fresh medium. Prior to some infections, cells were pre-treated for 3 h with 300 μM E64, 100 nM Baf, 25 μM monensin (Sigma), or 20 mM NH~4~Cl. In those instances inhibitors were also included in the post-adsorption culture medium. At the indicated times p.i., cells were collected by centrifugation at 179 × g, washed twice in chilled PBS and lysed in TLB. After centrifugation at 179 × g to remove cellular debris, samples were resuspended in sample buffer. Protein samples (representing 1 × 10^5^cells) were analyzed by electrophoresis on SDS-12% polyacrylamide gels and transferred to nitrocellulose membranes for 2 h at 100 V in 25 mM Tris-192 mM glycine-20% methanol. Nitrocellulose membranes (Bio-Rad Laboratories, Hercules, Calif.) were blocked overnight at 4°C in TBST (10 mM Tris \[pH 8.0\], 150 mM NaCl and 0.05% Tween) containing 5% nonfat dry milk, rinsed with TBST, and incubated with a rabbit anti-μNS polyclonal antiserum \[[@B71]\] (1:12500 in TBST) for 1 h. Membranes were subsequently washed with TBST and incubated for 1 h with horseradish peroxidase-conjugated anti-rabbit immunogloblin G (IgG) (1:7500 in TBST) (Amersham, Arlington Heights, Ill.). Bound antibody was detected by treating the nitrocellulose filters with enhanced chemilumescence (ECL) detection reagents (Amersham) and exposing the filters to Full Speed Blue X-ray film (Eastman Kodak, Rochester, N.Y.). Generation of NE subviral particles for infection ------------------------------------------------- Purified virions (1.4 × 10^11^) were incubated with 25 μg/ml of purified neutrophil elastase (Calbiochem) in 40 μL of VDB at 37°C for 3 h. Reactions were terminated by adding 1 mM PMSF and 200 μM NE inhibitor to the reaction mixture. 5.0 × 10^10^particles were run on SDS-12% polyacrylamide gels stained with Coomassie Brilliant Blue to confirm the removal of σ3. Viral infectivity was determined by plaque assay on L929 cell monlayers. Analysis of virus titer after NE treatment of virions ----------------------------------------------------- Purified Lang virions (1.4 × 10^11^) were treated with 25 μg/ml of NE in 40 μL of VDB at 37°C for the times indicated. Reactions were terminated as described above. To verify σ3 removal, the proteins from 5.0 × 10^10^particles were separated on SDS-12% polyacrylamide gels and visualized with Coomassie Brilliant Blue staining. Viral infectivity for each time point was determined by plaque assay on L929 cell monolayers. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= J.W.G. performed all experiments and was responsible for the experimental design and data analysis. J.W.G. also wrote the initial draft of the manuscript. L.A.S. is corresponding author, participated in experimental design, data analysis and critically edited the manuscript. Both authors read and approved the final manuscript. Acknowledgements ================ We express thanks to Stephen Rice and Max Nibert for critically reviewing this manuscript. We also thank members of the Schiff, Rice and Bresnahan laboratories for their constructive input throughout these studies. This work was supported by NIH grant AI45990 and University of Minnesota Graduate School Grant-In-Aid \#20017 to L. A. S. J. W. G. received support from NIH training grant 2T32 AI0742.	PubMed Central	2024-06-05T03:55:59.577290	2005-5-31	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180477/", "journal": "Virol J. 2005 May 31; 2:48", "authors": [ { "first": "Joseph W", "last": "Golden" }, { "first": "Leslie A", "last": "Schiff" } ] }
PMC1180478	Findings ======== Human papillomavirus (HPV) is a ubiquitous sexually transmitted DNA virus. A subset of mucosal HPVs are termed \"high-risk\" (for example, types 16, 18 and 31) because of an increased association with cervical cancer (review\[[@B1]\]). Among this group, HPV16 is the most common type, being found in about fifty percent of invasive cancers worldwide \[[@B2]\]. Two HPV16 oncoproteins, E6 and E7, are actively expressed in cervical cancer cells and are responsible for host cell transformation and cancer progression \[[@B3],[@B4]\]. As a polycistronic gene, the transcription of the E6E7 cassette yields E6E7 full-length mRNA as well as two spliced products, E6\IE7 and E6\IIE7. These splicing events are only found in the high-risk HPVs, but not in the low risk group \[[@B5]\]. From previous studies, the E6\I transcript has been found to be the most abundant one detected in HPV16 transformed cells, transgenic animals, cervical cancer cell lines and clinical samples \[[@B3],[@B5],[@B6]\]. While E7 is predicted to translate from spliced products as well as full-length transcripts, E6 protein can only be encoded from full-length transcripts \[[@B5],[@B7]\]. The splicing has been proposed to promote E7 translation by providing space for ribosome initiation to occur \[[@B5],[@B8]\]. However, it has been revealed that the translation of E7 from the full-length transcript is as efficient as those from spliced transcripts and that splicing is not required for E7 synthesis \[[@B9]\]. In earlier studies, difficulty in detection of the truncated E6 proteins raised the possibility that truncated forms of E6 might be present in such diminishingly small quantities that they may not have significant function \[[@B10]\]. However, detection of E6\I and E6\II proteins in HPV18-containing cervical cancer cells established in nude mice \[[@B7]\] and HPV16 transfected cells \[[@B11]\] strengthens the argument that these products have a role in tumorigenesis and in the viral lifecycle. Furthermore, studies of in vitrosynthesis of E6\I protein from wheat germ extract, but not rabbit reticulocyte lysate, suggest rapid protein turnover in the absence of E6AP \[[@B9],[@B12]\]. To date, various properties of E6\I protein have been studied. For example, HPV16 E6\I protein has the ability to transactivate the adenovirus E2 promoter as well as HPV P97 promoter \[[@B13]\]. The ability of HPV18 E6\I protein to bind to full-length E6 and E6AP (an E3 ubiquitin ligase) has been previously reported \[[@B14],[@B15]\]. The interaction of E6 protein with E6AP was found to be important in E6 functions (review in \[[@B16]\]). However, the information regarding E6 and E6\I protein localization with respect to E6AP protein in the cells is still unclear. Thus, we were interested in analyzing the co-localization of E6 and E6\I protein with E6AP to gain further insight into functional relationships that might correlate with localization. To overcome the difficulty in E6 protein detection by use of antibodies, GFP-linked E6 constructs were created (Figure [1](#F1){ref-type="fig"}). HPV16 E6 (nucleotide 101 to 559) and E6\I (nucleotide 101 to 417, without nucleotide 226--409) were amplified from SiHa cDNA and cloned in frame to the C-terminus of pEGFP-C1 (Clontech) using XhoI and KpnI sites. From previous studies \[[@B14],[@B17]\], similar GFP-E6 constructs have been shown to generate both full-length E6 as well as E6\I proteins. The E6 splice donor mutant (E6MT) was generated by site-directed mutagenesis at the spliced donor site with a G to C mutation at nucleotide 226 resulting in a V to L amino acid substitution. Plasmids were transfected into human embryonic kidney cells, 293T, and immortalized human keratinocyte cells, HaCaT, using FuGene 6 transfection reagent (Roche) using methods prescribed by the manufacturer. At 48 hours after transfection, cells were fixed with 4% paraformaldehyde in PBS, permeabolized with 0.1--0.5% NP-40, blocked with 5% fetal bovine serum in PBS and incubated with monoclonal antibody against E6AP (Sigma). After three washes, cells were incubated with Alexa 568-conjugated to goat anti-mouse antibody (Molecular Probes). Cells were washed with PBS and mounted on the slides using Gel Mount (Sigma). To localize the nuclei, cells were incubated with 4\',6-Diamidino-2-phenylindole (DAPI; Roche) diluted in PBS before a final PBS washing step. Slides were observed using an Olympus FV500 Confocal microscope with 3 lasers giving excitation lines at 405 nm, 488 nm and 515 nm. The images were taken under 60 × or 100 × objective oil immersion lens and collected at 1,024 × 1,024 pixels or 512 × 512 pixels resolution. The images were captured at various scanning conditions suited for the individual samples in order to eliminate overlapping signals between channels. We observed the localization of the different forms of E6 protein in 293T cells and HaCaT cells. The results are shown in figure [2](#F2){ref-type="fig"}. The wildtype E6 construct, which yields both full-length E6 and E6\I proteins (the truncated form of E6 that contains 43 amino acids identical to the N-terminus of E6), was expressed throughout the cells, but preferentially in the nucleus. A similar result was obtained with E6\I transfected cells. Interestingly, when E6MT was expressed, the signal was predominantly detected in the nucleus. All three protein products were excluded the nucleoli. Similar results were obtained from both 293T and HaCaT cells. These results correspond to previous reports of HPV18 E6 localization \[[@B17],[@B18]\]. However, the localizations of E6 and E6\I proteins are different than one previous study of HPV16 E6 \[[@B19]\] which found that HPV16 E6 and E6\I proteins were primarily nuclear. Since HPV16 E6 used in our study was amplified from SiHa cells, which contains two mutations (position 83, L to V and position 103, Q to D), it is possible that these contribute to the observed differences. These conservative amino acid changes are not predicted to alter localization since neither are in regions responsible for nuclear import \[[@B19]\]. However, alteration of interactions with unknown cellular targets by these mutations cannot be ruled out. In our experiments, the level of E6\I and/or E6 proteins differed slightly from the other studies. Though the signals shown in our experiments from E6 and E6\I proteins were dispersed throughout the cells, these signals were consistently excluded from nucleoli. Hence, these signals are not likely to be artifactual. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Cellular localization of HPV16 E6s in 293T (A) and HaCaT (B). Cells were transiently transfected with plasmids expressing GFP, GFP-E6, GFP-E6\I and GFP-E6MT. Cells were grown and fixed with 4% paraformaldehyde in PBS on the coverslips at 48 h after transfection. The fluorescent images, phase contrast images and merge fluorescent-phase contrast images are shown. V indicates the pEGFP-C1 vector transfected cells. The scale bar represents 20 μm. ::: ![](1743-422X-2-50-2) ::: As the E6 protein has been shown to associate with E6AP in HPV-induced cervical cancers, the location of these functional complexes is of importance. From results in figure [3](#F3){ref-type="fig"}, co-localization of either E6 or E6\I protein and E6AP mainly occured in the cytoplasm whereas co-localization of E6MT and E6AP was restricted to areas of the nucleus and the perinuclear region. In essence, E6MT protein was rarely co-localized with E6AP. The localization of E6AP agreed with the previous studies showing that E6AP is predominately localized to the cytoplasm, and to a lesser extent, the nucleus\[[@B20],[@B21]\]. E6 protein expression has been shown to induce self-ubiquitination and degradation of E6AP with a dramatic reduction in E6AP half-life overall \[[@B22]\]. This likely explains the infrequent detection of E6MT and E6AP co-localization in the cytoplasm. The binding of E6AP with full-length E6 protein leads to the rapid degradation of E6AP. The detection of E6\I protein co-localized with E6AP is in agreement with previous observations which showed interaction between HPV18 E6\I protein and E6AP in vitro\[[@B14],[@B15]\] and that we would expect such complexes to be stable since E6\I inhibits E6AP function thereby allowing the complexes to be visible by immunofluoresence. The co-localization of E6 protein with E6AP resembles that of E6\I. The presence of translation products of E6\I protein in E6-transfected cells, may explain this similarity. There are two mechanistic possibilities; first, the E6\I protein may compete with E6 protein in binding to E6AP and this binding may protect E6AP from degradation induced by E6 protein. Second, the E6\I protein might bind to E6 full-length protein and the binding might modulate the function of E6 in induction of E6AP degradation. As shown in the microscopy results, the co-localization of E6AP is not exclusively to E6 protein. This raises the possibility that E6AP-independent functions of E6 \[[@B23]\], E6AP-independent degradation of E6 \[[@B24]\] or, in fact, E6-independent functions of E6AP may play a significant role. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Co-localization of E6s and E6AP in 293T (A) and HaCaT (B). Transiently transfected cells were analyzed for E6 or E6 variants fused to GFP, E6AP (Alexa 568 dye) and nuclear DNA (DAPI) by confocal microscopy. Slides were analyzed by microscopy with 3 lasers excitation lines. The images from the individual channels (DAPI, GFP, Alexa 568) as well as the merged image are shown. P and V represent non-transfected cells and pEGFP-C1 vector transfected cells, respectively. The scale bar represents 20 μm. ::: ![](1743-422X-2-50-3) ::: In addition, we observed stronger fluorescent signals with E6-transfected cells than from E6\I or E6MT-transfected cells (data not shown). More intense signals might suggest interaction with and stabilization by cellular proteins. This could involve the ability of E6 protein to oligomerize with E6 or E6\I protein \[[@B14]\]. In other studies, binding of E6\I and E6 protein has been shown to induce the degradation of both proteins \[[@B14]\], however, the increase in signal observed with E6 suggests that homo-oligomeric interactions could stabilize the protein complexes. The differing capacity of E6\I protein of different HPV types to interfere with E6 might result in differing downstream effects on E6-dependent targets. Since the similarity of E6\I protein between HPV16 and HPV18 is about 50% and the amino acids 28--31, which have been shown to be important in E6 and E6AP binding, differ slightly (amino acid \"IEIT\" from HPV18 versus \"IILE\" from HPV16), this may account for the differing capacity to bind E6 or E6AP. Furthermore, HPV16 E6 and HPV18 E6 have been shown to have different preferences in binding to p53 structural conformations \[[@B20]\]. Therefore, it might be informative to test E6\I protein from various HPV types for functional differences in their abilities to affect stability of cellular targets such as p53. In conclusion, our results suggest a functional role for expression of E6\I protein in high-risk HPV-infected cancer cells. We propose a model as shown in figure [4](#F4){ref-type="fig"}. E6\I protein may bind to either E6 or E6AP and binding may modulate their functions or interfere directly with degradation of these two proteins. Further studies in the regulation of E6 protein with E6\I protein provide useful insights into HPV-disease and a potential means to control development and progression of HPV-related cancers. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### A model for regulation of E6 protein function by E6\I*. E6\I protein binds to E6 or E6AP in the nucleus and cytoplasm. The binding may lead to the inhibition of E6-E6AP binding as well as inhibition of binding of E6 or E6AP to other cellular targets. In this manner, E6\I may inhibit either E6-induced or E6AP-dependent proteosome degradation function or, in fact, inhibit other functions of either E6 or E6AP. The question mark (?), represents other E3 ubiquitin ligases enzymes participating in E6-mediated degradation or E6AP-independent, E6-induced protein degradation. ::: ![](1743-422X-2-50-4) ::: List of abbreviations ===================== Human papillomavirus (HPV), E6-associated protein (E6AP), 4\',6-Diamidino-2-phenylindole (DAPI), green fluorescent protein (GFP) Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= K. V. designed and created the GFP fusion constructs, performed all of the described experiments and interpreted the data. S. C. created the E6MT mutant. P. M. and P. C. A. provided logistical, financial and material support and helped in data interpretation and manuscript preparation. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### E6 protein sequences. The secondary structure of E6 protein is shown. The E6 gene codes for full-length E6 as well as a truncated protein (E6\I, indicated in pink). The E6 splice donor mutant (E6MT) was generated by site-directed mutagenesis at the splice donor site resulting in a V to L substitution at amino acid 42 (arrow head). This construct expresses only E6 full-length protein. Whereas, the E6\*I construct expresses only the truncated form of E6 and not the full-length form. GFP was fused to the N-terminus of each of these constructs. ::: ![](1743-422X-2-50-1) ::: Acknowledgements ================ We thank Paul F. Lambert (UW-Madison) for critically reading and evaluating this manuscript. We thank Joe Zhou for providing his expertise in confocal microscopy. This research was supported in part by a COBRE grant to the Nebraska Center for Virology from the NCRR (P20 RR15635). K.V. was supported by a fellowship from Mahidol University, Bangkok Thailand through the Medical Scholars Program. We thank the National Science Technology and Development Agency (NSTDA).	PubMed Central	2024-06-05T03:55:59.580162	2005-6-16	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180478/", "journal": "Virol J. 2005 Jun 16; 2:50", "authors": [ { "first": "Kulthida", "last": "Vaeteewoottacharn" }, { "first": "Siriphatr", "last": "Chamutpong" }, { "first": "Mathurose", "last": "Ponglikitmongkol" }, { "first": "Peter C", "last": "Angeletti" } ] }
PMC1180479	Background ========== Bladder cancer is the second most common tumor of the urogenital tract with transitional cell carcinoma comprising of about 90% of all primary bladder malignancies \[[@B1]\]. Most of these tumors are superficial at initial presentation, limited to the mucosa, sub-mucosa or lamina propria. In superficial cancers, recurrence rates after initial treatment are 50--80% with progression to muscle invasive disease in 10--25% cases, whereas in muscle invasive disease there is 50% risk of loco-regional and distant metastasis. The transitional cell carcinomas (TCC) originates in the bladder mucosa, progressively invading the lamina propria and more sequentially involves muscularis propria, perivesical fat and contiguous pelvic structures with increasing incidence of lymph node involvement during progression \[[@B2],[@B3]\]. Radical cystectomy is the corner stone of treatment for invasive bladder cancers in patients whose medical condition allows major surgical procedure \[[@B1],[@B4]\]. Five-year life expectancy is 75% for T~2~lesions, 40% for T~3~and 25% for T~4~lesions \[[@B5]\]. The outcome for bladder cancer patients have improved primarily because of advances in surgical technique and better peri-operative care. However 70% of the patients with node positive disease develop tumor recurrence if treated by cystectomy alone. Proper staging of disease and timely institution of adjuvant treatment could potentially make a difference. In addition to the evaluation of the primary tumor in a cystectomy specimen, assessment of the regional lymph nodes is also important; not only to accurately stage the disease but also to identify the need for adjuvant treatment. Preoperative imaging studies CT scan or an MRI will miss microscopic nodal metastasis in up to 70% of the patients \[[@B3],[@B7],[@B8]\] and this leaves histological evaluation as the only reliable tool in accurately staging the disease. The reported incidence of regional lymph node involvement following radical cystectomy is between 14% and 28% \[\[[@B5],[@B6]\], and \[[@B8]\]\]. These rates correlate with the stage of the tumor; the higher the stage, the higher is the incidence of the involved lymph nodes \[[@B5],[@B10],[@B11]\]. Lymph node involvement is associated with increased risk of local recurrence and disease progression with survival rates varying from 20% to 40% in patients with and without lymph nodes metastasis respectively \[[@B5],[@B10]\]. Pelvic lymphadenectomy is routinely performed as part of radical cystectomy for bladder cancers; however, there is lack of consensus on the intent (therapeutic or diagnostic) and limits of lymph node dissection. It is most often performed as a staging procedure. Some of the leading authorities believe that it has a potential therapeutic adjunct to radical cystectomy \[[@B12],[@B13]\]. The absolute limits or extent of lymphadenectomy for patients undergoing surgery with a curative intent has not been precisely defined. The recent guidelines for the treatment of muscle invasive bladder cancer by The European Association of Urology recommend limited pelvic node dissection. This consists of removing the tissue in the obturator fossa in patients undergoing surgery with a curative intent \[[@B14]\]. Whereas several authors have noted an improved 5-year survival with extensive pelvic lymph node dissection in patients with node positive bladder cancer \[[@B2],[@B3],[@B15]\]. In the present review, we have attempted to analyze the current literature to see if there is convincing data to support the observation of some authors that extensive pelvic lymphadenectomy is associated with low pelvic and distant recurrence and survival. Extent of lymphadenectomy ========================= The minimum number of lymph nodes to be excised in order to obtain a therapeutic advantage remains an issue of controversy. Anatomical studies have defined the external and internal iliac lymph nodes and the obturator nodes as the site of primary lymphatic drainage of the bladder and the common iliac nodes as the site of secondary drainage, with the trigone and the posterior wall draining directly into the pre-sacral nodes \[[@B16]\]. The role and extent of lymphadenectomy in bladder cancer is not well-defined, in contrast, patients with gastric \[[@B17]\], breast \[[@B18]\], and colorectal \[[@B19]\] tumors the impact of lymphadenectomy has been well studied and there has been a consensus that complete lymph node dissection with removal of a minimum number of lymph nodes correlates with improved survival. Guidelines for the management of colon and rectal cancer indicate that at least 12--14 lymph nodes should be retrieved by the pathologist in patients with colorectal cancers \[[@B18]\] and at least 10 lymph nodes in patients with endometrial cancer \[[@B20]\], but this knowledge is lacking in cases of bladder cancers. Several authors have postulated that the number of lymph nodes examined in cystectomy specimen can have an impact on the outcome of patients treated for primary bladder cancer \[[@B9],[@B13]\], whereas others have reported an improved survival with extensive nodal dissection \[[@B15],[@B22]\]. Extensive lymphadenectomy increases the operating time and theoretically may increase the risk of hemorrhage and lymphocele formation, but studies have not shown a significant increase in morbidity with this procedure \[[@B22]\]. Number of lymph nodes and impact on survival ============================================ Many factors can influence the outcome of patients with bladder cancer including the T-stage, the N-stage and the total number of lymph nodes retrieved. Skinner et al,\[[@B13]\] noted that important prognostic factors included age, gender and lymph nodes status but the number of involved nodes was the single most important prognostic variable. Lerner et Al, \[[@B2]\] observed that the 5-year survival probability in cases with 5 or fewer involved nodes was almost double than those with 6 or more involved nodes. Mills et Al, \[[@B23]\] studied a group of 83 patients with nodes positive disease and reported similar results. The median number of retrieved nodes was 20/case. A significant survival difference was observed when patients with less than 5 involved nodes were compared with more than 5 involved lymph nodes (p = 0.0027), however this significance was lost on multivariate analysis. Stein et al,\[[@B22]\] verified the impact of positive nodes on univariate as well as multivariate analysis with a cut off of 8 involved nodes \[[@B18]\]. Vieweg et al, \[[@B24]\] reported that in cases with organ confined disease, there was no survival difference between N~0~and N~1~groups but survival benefit consistently decreased from N~1~to N~3~disease. Konety et al,\[[@B25]\] studied 20,799 patients treated for primary bladder cancer and reported similar results. They concluded that regardless of the tumor stage, at least 14 lymph nodes should be excised in patients undergoing radical cystectomy. Herr et al,\[[@B26]\] studied the impact of lymph nodes in a group of 322 patients who underwent pelvic node dissection. They observed that in patients with pN~0~status improved survival and local control was seen when the cut-off was 8 examined lymph nodes. However, in patients with pathologically involved nodes at least 9 lymph nodes should be removed. Anatomical extent of dissection =============================== Poulsen et al, \[[@B15]\] studied the influence of extent of nodal dissection on survival following radical cystectomy for bladder cancer. They compared 126 patients who underwent extensive pelvic dissection with proximal limits up to the bifurcation of aorta and 68 patients who underwent limited nodal dissection with proximal limits up to the common iliac vessels. The overall prevalence of nodal metastasis was slightly higher in patients with extended dissection as compared to ones with limited dissection. Extended dissection was associated with an improved 5-year recurrence free survival for patients with tumor confined to the bladder (pT~3a~or less), but this was not true for tumors penetrating the bladder wall. This was a retrospective, non-randomized study and the anatomical mapping and the number of retrieved lymph nodes was not mentioned. Herr et al, \[[@B21]\] introduced the term of lymph node density indicating the ratio between the numbers of nodes removed to the number of involved node. They found ratio based lymph node staging, which reflects the quality of lymph node dissection, was a significant prognostic variable for survival and local control in patients with lymph node positive bladder cancer after radical cystectomy. Later Stein et al, \[[@B22]\] reported their experience of 244 patients with involved lymph nodes treated for primary transitional cell carcinoma of bladder. They reported that the overall and recurrence free survival was significantly related to pathological subgroup of the primary bladder tumor, patients with organ-confined disease having a better survival as compared to those with extravesical tumors. Patients with a lymph node density of 20% or less had a better recurrence free survival when compared to those with more than 20% (p \< 0.001), however this study confined to only the lymph node positive cases, the lymphadenectomy was done in en blocfashion, and the exact anatomical location of the excised lymph nodes was not known. In an attempt to standardize the extent of dissection, Leissner et al\[[@B9]\] studied a group of 447 patients who underwent extended lymphadenectomy along with radical cystectomy for bladder cancer. The guidelines for lymphadenectomy were to remove obturator, internal, external and common iliac, pre-sacral and the lymph nodes on both sides of aortic bifurcation. The mean number of retrieved nodes was 16.7/patient. They observed that overall the tumor specific survival (P-value \< 0.013) and disease free survival was significant (P-value \<0.016) when patients with ≥ 16 excised lymph nodes were compared with those with \< 15 nodes removed, except for those with T-4 lesions. In presence of nodal metastasis patients only benefited by removal of at least five metastatic lymph nodes, thus supporting the therapeutic effect of extended lymphadenectomy. The two possible reasons for better survival in patients with extended lymphadenectomy were explained by lymphadenectomy resulting in a more accurate staging so adjuvant therapy may be started earlier. The other possible explanation could be the increased likelihood of removing nodal tumor deposits. Conclusive evidence could not be derived from a retrospective studies and literature review. Mills et al\[[@B23]\] reported their experience of 83 patients with node positive tumors who underwent pelvic lymphadenectomy. Anatomical mapping was done in the group of retrieved nodes. Survival correlated with the number of lymph nodes, adjuvant chemotherapy, capsular penetration and the site of metastatic nodes. On univariate analysis, survival was better when more than 5 positive lymph nodes was removed however this was not significant and only lymph node capsular penetration demonstrated a significant effect on survival. This was again a retrospective study but only the pelvic lymph nodes were dissected, therefore the nodal status of the proximal sites was unknown. Vazina et al\[[@B27]\] reviewed their results of 176 patients who were treated with lymph node dissection up to the aortic bifurcation. The median number of lymph nodes excised was 25/case. It was observed that the rate of nodal involvement gradually decreased from pelvic to aortic sites. They reported that these tumors did not exhibit skip metastasis as all. The patients except one who had involvement of lymph nodes at or above the aortic bifurcation or common iliac region also had positive lymph nodes in distal (pelvic and peri-vesical) areas. One patient who had involvement of the common iliac lymph nodes without involving the distal lymph nodes, had a primary lesion in the trigone explaining this to be presumably due to direct lymphatic drainage of the trigone area to the common iliac region. Bochner et al,\[[@B28]\] reported the survival status of 144 patients who were operated for primary bladder cancer, out of whom 56 underwent standard pelvic node dissection whereas 88 patients had extended dissection. Though, the yield of nodes was greater in extended dissection than in standard dissection, but both types had a similar percentage of patients with positive lymph nodes, however extended lymphadenectomy provided better prognostic information through a more accurate evaluation of the total number of lymph nodes excised or the number of involved lymph nodes. Extended lymphadenectomy identified 33% of patients with microscopic nodal metastasis involving the common iliac lymph nodes, explaining the therapeutic value of pelvic dissection. This is also supported by a European study of a group of 290 patients, who underwent extended radical lymphadenectomy along with radical cystectomy \[[@B29]\]. In this series, it was observed that 57% of patients with lymph node involvement at level I had nodal metastasis at level II and 31% had level III nodes involved. Similarly, 35% patients with level II nodes involved were found to have level III nodal involvement, hence they concluded with a strong recommendation for extended lymphadenectomy in all patients undergoing radical cystectomy for bladder cancer with a curative intent. The three lymphadenectomy levels were defined by the authors as; level I, distal to the bifurcation of common iliac artery, level II distal aortic bifurcation up to level I and level III proximal to aortic bifurcation. Conclusion ========== More than half the patients with locally advanced (pT~3a,~pT~3b,~pT~4a~and/ or pN+) treated by radical surgery will have disease progression in five years. Adequate staging affects the outcome of patients with bladder cancer. Such information is important not only for therapy and prognosis, but also in identifying individuals for adjuvant treatment. Evidence from contemporary studies indicates that more extensive lymphadenectomy significantly improved the prognosis of patients with bladder cancer. A growing body of evidence suggests that an extended lymph node dissection may provide not only improved prognostic information, but also a clinically significant therapeutic benefit for both lymph node-positive and -negative patients undergoing radical cystectomy. However, in the absence of randomized controlled trial, the scientific value of current evidence is at best level 2. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= MHA. Conceived of the idea, helped in literature search and drafted part of the manuscript. SFdid the literature search, drafted most of the manuscript. OSWrote the initial draft and helped in literature search. All authors have read and approved of the final draft.	PubMed Central	2024-06-05T03:55:59.581566	2005-7-15	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180479/", "journal": "World J Surg Oncol. 2005 Jul 15; 3:43", "authors": [ { "first": "M Hammad", "last": "Ather" }, { "first": "Sadaf", "last": "Fatima" }, { "first": "Orhun", "last": "Sinanoglu" } ] }
PMC1180512	Introduction {#s1} ============ Genetic linkage analysis has traditionally been applied to one or very few traits at a time. It is now possible to simultaneously measure thousands of related "traits" from high-throughput technologies such as DNA \[[@pbio-0030267-b01],[@pbio-0030267-b02]\] and protein microarrays \[[@pbio-0030267-b03]\]. It is therefore necessary to extend linkage analysis techniques so that thousands of traits can be simultaneously analyzed, particularly when the traits are complex and it is desirable to identify multiple loci contributing to phenotypic variation. Performing linkage analysis on many traits compounds difficulties that are already present in a conventional analysis \[[@pbio-0030267-b04]\], although it also provides the opportunity to borrow information across the traits in such a way that more informative conclusions can be drawn. Here, we focus on linkage analysis of genomewide expression data obtained from DNA microarrays. Recent studies in a variety of organisms \[[@pbio-0030267-b05]--[@pbio-0030267-b13]\] have unambiguously shown that heritable variation in gene expression levels is pervasive. Therefore, there is considerable interest in delineating the genetic architecture of transcriptional variation at the genomewide level. Existing linkage analysis techniques have already been applied to genomewide expression in yeast, mice, maize, and humans \[[@pbio-0030267-b05]--[@pbio-0030267-b07]\]. In a cross between two strains of the budding yeast Saccharomyces cerevisiae, linkage scans were performed separately on the expression levels of approximately 6,000 genes. It was shown that hundreds of these "gene expression traits" show linkage to at least one locus, and many traits appear to be influenced by multiple quantitative trait loci (QTL) \[[@pbio-0030267-b05],[@pbio-0030267-b07],[@pbio-0030267-b14]\]. However, multiple locus linkage analysis has not been applied to this dataset. A well accepted approach for mapping two loci, and for identifying epistatic interactions, is to perform an exhaustive two-dimensional (2D) linkage analysis in which all pair-wise positions in the genome are tested for linkage. However, 2D scans are extremely computationally demanding when applied to thousands of phenotypes and may suffer from low statistical power due to the large number of hypothesis tests performed. Although several other approaches exist for mapping multiple loci that are linked to a quantitative trait, none of these methods allows the individual and joint significance of the loci to be unequivocally assessed. Here, the joint significance is the case where all of the multiple loci are truly linked, not just a subset of them. Thresholding based on the joint significance becomes particularly important when examining many quantitative traits at once. One common approach for identifying multiple QTL is to use a model selection algorithm \[[@pbio-0030267-b15]--[@pbio-0030267-b18]\], where the goal is to identify the subset of loci comprising the best model according to some optimality criterion. Using this approach, it is difficult to search over the enormous number of potential models and to provide a criterion for the best model that is biologically meaningful. Furthermore, for models that include multiple QTL, it is difficult to determine which of these are true QTL and which have been included in the model by chance. Another approach is to a priori form a model that includes a certain set of pre-chosen loci spanning the genome, and then test whether each remaining locus significantly improves the model in this larger multiple locus model \[[@pbio-0030267-b19]--[@pbio-0030267-b23]\]. The motivation for this mapping approach and its derivatives such as multiple interval mapping is that the pre-chosen loci explain residual variation of other unknown QTL. It is difficult to formulate a pre-chosen model for each expression trait, and again it is difficult to interpret the joint significance among multiple loci. To address these issues, we have developed a new method for mapping multiple loci and identifying epistatic interactions when analyzing thousands of phenotypes, such as gene expression levels. Information shared across expression traits is employed in a way that allows us to make minimal assumptions about the statistical properties of the data. Our method permits easy-to-interpret statistical significance analysis of individual loci, as well as the overall joint significance of multiple loci identified for any given expression trait. Strengths of both the model selection and composite interval mapping methods have been incorporated, which turns out to be more straightforward when analyzing many traits simultaneously. Rather than trying to estimate the true model underlying the expression trait by seeking the "best model," or by assuming a certain model of genetic background and testing for the inclusion of additional loci, we propose to measure the probability that a locus is in the true model given the data, without ever specifically estimating the entire true model. This overcomes some of the difficulties incurred by applying the two existing approaches to this problem. However, if one were considering only a single trait, it would be very difficult to calculate this probability without making strong assumptions. Here, we use a nonparametric approach that allows us to calculate conservative estimates of these posterior probabilities of linkage. The nonparametric approach is possible because many related traits are considered simultaneously. We applied the method to the S. cerevisiae experiment and show that at least 37% of all gene expression traits show joint linkage to two loci. Pairs of jointly linking QTL could be identified with high confidence for 170 gene expression traits. Bioinformatics analysis of these 170 significant expression traits and their corresponding QTL begins to provide intriguing insights into the genetic architecture of transcriptional variation. In addition, we are able to show that epistatic interactions contribute to gene expression variation in at least 14% of all traits. Our proposed approach overcomes the inherent computational and statistical difficulties that arise when performing an exhaustive search on thousands of traits at once. Moreover, the availability of thousands of traits for a single set of meioses allows us to show that, for this experiment, a full 2D scan is not as powerful as the sequential search method we employ, even though the locus pairs selected by the two methods overlap substantially. We also show that 2D scans do not allow one to test for joint linkage of a pair of loci, only whether at least one of the pair is linked. This is shown to be particularly problematic when analyzing thousands of phenotypes. Results {#s2} ======= The data used in this study were derived from a cross between two haploid strains of the budding yeast S. cerevisiae: a standard laboratory strain and a wild isolate from a California vineyard. Gene expression measurements were obtained for 6,216 open reading frames in 112 haploid segregants, and dichotomous genotypes were identified at 3,312 markers covering 99% of the genome \[[@pbio-0030267-b14]\]. Using these experimental data, we developed and applied a new computationally efficient method for simultaneously mapping multiple gene expression QTL and identifying epistatic interactions. The models and methods used here are appropriate for haploid organisms, although the ideas may be extended to diploid and higher ploidy organisms in the usual way \[[@pbio-0030267-b24]\]. 2D Linkage Scan {#s2a} --------------- Initially, we applied an exhaustive 2D linkage scan in order to identify expression traits that are significantly linked to pairs of loci or that are significant for epistasis. In performing these significance tests, we considered a linear model that fully parameterizes the quantitative trait in terms of all four possible genotypes. This model can be written as Traditionally, genetic linkage is said to exist between the trait and a pair of loci if any of the locus effects are significantly different than zero, but not necessarily all of them \[[@pbio-0030267-b24]\]. Epistasis exists only if the locus1 × locus2 interaction is significantly different from zero. This linear model approach to identifying QTLs is well justified \[[@pbio-0030267-b24]\] and has been shown to be especially useful when the markers are densely sampled \[[@pbio-0030267-b18]\]. The test for pair-wise linkage was performed as follows. For each expression trait, a linear model was fit by least squares to each pair of loci. The locus pair with the largest F-statistic comparing the full model to the baseline model was selected for that trait. For the test of epistasis, a similar procedure was performed, except an F-statistic was computed that compared the full model to the purely additive model, which directly assesses the contribution of the interaction term. The significance of each locus pair selected for linkage was computed using a standard permutation technique against the null hypothesis of no linkage to either locus \[[@pbio-0030267-b25]\]. For the test of epistasis, we used a similar permutation technique to assess the significance of the locus1 × locus2 interaction \[[@pbio-0030267-b26]\]. The end-product of these tests is a p-value and a pair of loci for each expression trait. Using the false discovery rate (FDR) quantity to correct for multiple comparisons \[[@pbio-0030267-b27],[@pbio-0030267-b28]\], there were 3,540 traits significantly linked to a locus pair at the 5% FDR threshold. Note that in this case, a "false discovery" must be defined as a trait where both selected loci are false positives, and thus many of the "true discovery" traits could have one false-positive locus. In the test for epistasis, no significant results were obtained; the number of significant tests at each cut-off mirrored the number found under the null permutations. The exhaustive 2D search proved to be unsatisfactory for a number of reasons. Most obviously, the number of multiple-locus models that have to be considered is computationally and statistically challenging for pairs of loci, and prohibitive for three or more loci. With 3,312 markers and 6,216 expression traits, one has to consider more than 18 million single-locus models to simply test for linkage between every expression trait and locus. More than 27 billion tests have to be performed to consider all two-locus models for every expression trait, and more than 27 trillion tests to consider all three-locus models for every expression trait. In addition, it is likely that by searching over so many models, the statistical power to detect linkage is severely attenuated because of the multiple comparison problem. Secondly, when employing an exhaustive 2D scan, there is no statistically rigorous method to test for joint linkage, which exists only if both loci have nonzero terms in the full model. In other words, the significance of an individual locus selected for an expression trait is confounded with the overall significance of the pair of loci. Since p-values are calculated against the null hypothesis of no linkage, a highly significant result may be due to only one locus being truly linked while the other locus is included by chance. This confounding is especially problematic when considering thousands of traits simultaneously. Since pairs of loci that show large marginal effects are preferentially selected when testing thousands of traits for linkage, one cannot examine marginal effects of individual loci among the most significant linkages in an unbiased fashion. Therefore, by chance it may appear that both loci explain a large proportion of variance of the trait. Ideally, a measure of significance for each locus would be available, and then a joint measure of significance for all loci would be calculated. In our case, we did not want to call a trait significantly linked if either of the loci were a false positive. Finally, a decision must be made as to which traits to call significantly linked. If the goal is to avoid any false-positive loci when calling a trait significant, then there is no simple p-value that can be formed for this purpose. This follows because the null hypothesis consists of multiple scenarios, and there is no readily available null distribution to describe this. Therefore, a more sophisticated method must be used to assess the significance of thousands of multi-locus models. Stepwise Search More Powerful than 2D Search {#s2b} -------------------------------------------- One potential way to improve the exhaustive 2D scan is to use another method for selecting pairs of loci. In particular, one can select loci in a sequential manner, cutting down the number of models considered to 2 × 3,312, instead of more than 5 million. One readily available method for selecting loci in a sequential manner is forward stepwise regression. Here, one selects a primary locus that shows the most significant one-dimensional (1D) linkage, i.e., the one that has the largest LOD score. This is equivalent to identifying the locus that yields the smallest residual sum of squares when regressing the expression trait on the inheritance pattern at that locus \[[@pbio-0030267-b24]\]. Next, a secondary locus is chosen that yields the largest LOD score conditional on the primary locus being linked. Again, this is equivalent to choosing a locus that minimizes the residual sum of squares when regressing the expression trait on both the secondary locus and the primary locus contained in the same model. A tertiary locus may be selected by including the previous two loci in the regression, and so on. One can use this forward stepwise regression technique simply as a way to select pairs of loci for each trait, and then the significance analysis can be repeated as before. It has been hypothesized that failing to consider all possible two-locus models through an exhaustive 2D search (e.g., selecting loci in a sequential fashion) may lead to a loss in power or to missing important interactions between loci \[[@pbio-0030267-b29]\]. The yeast expression dataset presents a rare opportunity to test this idea; in a typical study, only one quantitative trait is measured and only one scan is performed, which makes it difficult to compare different multi-locus selection methods. Simulation-based comparisons make a number of assumptions that may not always be true. Here, we are able to make a direct comparison based on thousands of related scans, all conditional on the same set of meioses. Thus, we compared the exhaustive 2D search to a simple sequential search method by repeating every step exactly as above, except that the sequential search method was used to select pairs of loci (see [Materials and Methods](#s4){ref-type="sec"}). At FDR cut-offs of 1% and 5%, there are 2,780 and 4,271 significant pair-wise linkages, respectively. At these same cut-offs, the 2D search yielded substantially fewer: 1,715 and 3,540 significant linkages. [Figure 1](#pbio-0030267-g001){ref-type="fig"}A shows the number of significant pair-wise linkages over a range of p-value thresholds, where it can be seen that the sequential scan consistently finds a greater number. Since any given p-value threshold results in the same number of expected false positives for each type of search, this is empirical evidence that the sequential search is more powerful. A 2D search could still produce more biologically meaningful results. In order to assess this, we measured the overlap in selected locus pairs among the traits corresponding to the 3,000 most significant linkages identified by the 2D method. The locus pairs selected by the two methods for each given trait were considered to be equivalent if the same two chromosomes were identified and the respective chromosomal locations of the loci were within 50 kilobases (kb) of each other. Under this definition, 90% of the locus pairs were found to be equivalent between the two methods. ::: {#pbio-0030267-g001 .fig} Figure 1 ::: {.caption} ###### A Power Comparison of the 2D Locus Pair Search and the Sequential Search The number of significant traits over a range of p-value thresholds are shown. Since any given p-value threshold results in the same number of expected false positives, these plots give empirical evidence that the sequential search is more powerful than the 2D search. \(A) Plot of the number of traits significant for linkage versus the p-value threshold. \(B) Plot of the number of traits significant for epistasis versus the p-value threshold. The gray line, which shows the number of expected false positives for each p-value cut-off, is similar to the number called significant under the exhaustive 2D search. ::: ![](pbio.0030267.g001) ::: The sequential search was also more powerful for identifying epistasis relative to the exhaustive 2D scan. [Figure 1](#pbio-0030267-g001){ref-type="fig"}B shows the number of traits called significant for epistasis over a range of p-value thresholds, where it can be seen that the sequential search is again more powerful. Neither search method yielded a trait with high significance for epistasis. However, from the sequential method we are able to estimate that at least 14% of the traits are operating under epistasis, whereas due to a lack of power the 2D search estimate is 0%. This estimate is obtained by the following reasoning, which has been rigorously developed elsewhere \[[@pbio-0030267-b27],[@pbio-0030267-b28]\]. If the locus pair identified for each transcript were a false positive, the distribution of p-values across all transcripts would be flat and uniformly distributed between zero and one; thus, the shape of the observed distribution of p-values can be used to estimate the total proportion of false positives. The more powerful a set of statistical tests are, the more this flatness can be distinguished from the signal. We did not see much overlap in locus pairs selected among the two search methods, but this is not surprising given that the 2D search apparently produces only noise. Therefore, for this particular experiment, the sequential search is more powerful than the exhaustive 2D search in identifying pair-wise linkage and detecting epistasis. The sequential search also appears to extract a biological signal that is similar to that from the 2D search. However, it is not possible to conclude whether these properties would hold in other experiments or for different sample sizes. Also, the comparison was made based on significance assessed against the null hypothesis of no linkage, which is not a solution to the problem of detecting joint linkage. The sequential approach as implemented above still suffers from the problem that significance can be driven by a single locus while the other locus is a false positive. However, sequentially selecting loci allows their individual significance to be assessed, which we show is crucial in detecting true joint linkage. We discuss how to assess individual and joint significance for the sequential approach below; we note that the same methods would not work without a number of potentially unjustifiable assumptions for the exhaustive 2D search. Proposed Approach {#s2c} ----------------- We developed a method to overcome the following problems associated with existing approaches: a prohibitively large number of multi-locus models are considered, a clear measure of significance among individual loci is not available, and the desired alternative hypothesis that all selected loci are linked for each trait is not tested. The method can be summarized in four steps. Step 1. For each expression trait,L loci are identified through a sequential locus selection procedure, as above. Step 2. At each stage of the sequential search, a Bayesian technique is employed to calculate the probability that the locus is linked to the expression trait, conditional on the assumption that the previously chosen loci are also linked. Step 3. The locus-specific probabilities are combined to form the probability that all loci are simultaneously linked to the expression trait. The overall probabilities of linkage from Step 3 provide a ranking of the traits from most significant to least significant. It is then necessary to select a set of traits, each of which has a high probability of being linked to all loci simultaneously. In order to guide this choice, we propose a method to assess the statistical significance of a given set of traits. Step 4. A significant expression trait is called a false discovery if any of the loci selected for that trait is a false positive. That is, a true discovery is an expression trait where all selected loci are truly linked. A new approach for estimating the FDR among a set of significant traits is employed that directly utilizes the probabilities calculated in Step 3. The starting point for the method is to define a multi-locus model that may include varying numbers of loci, where it is clear how one modifies the model to include an additional locus. Here, we continue to use the fully parameterized model. For zero, one, and two loci, the model may be written, respectively, as where, for example, "locus1" is the main effect for the primary locus, and "locus1 × locus2" is the epistatic interaction between the primary and secondary loci ([Materials and Methods](#s4){ref-type="sec"}). A sequential search can then be performed to identify the top linked locus for each expression trait, which involves finding the locus that offers the greatest improvement in goodness of fit when comparing model M1 to model M0. This primary locus is then included in models M1 and M2, and a secondary locus is identified that provides the greatest improvement in goodness of fit when comparing model M2 to model M1. Continuing this process, an ordered set of L loci for each expression trait can be identified. Here we consider only L = 2 loci, but the method can be applied to larger numbers of loci. The Bayesian posterior probability that the primary locus for each trait shows linkage can be written as Pr(locus 1 linked \\|Data). Since the secondary locus is identified conditional on the presence of the primary locus, the probability that it is linked is calculated conditionally on the primary locus being linked: Pr(locus 2 linked \\| locus 1 linked, Data). Note that probabilities may be formed analogously for L loci, with the final probability beingPr(locus L linked \\|loci 1,\...,L-1 linked, Data). These conditional probabilities are conceptually consistent with the procedure used to select loci. For example, the secondary locus is not called significant unless the primary locus is also called significant, since it was used in identifying the secondary locus. The above probabilities give a measure of significance to each locus. However, one would also like to know the joint significance of the loci. The probability that all loci are linked to the expression trait is simply the product of the locus-specific probabilities. For example, These joint-linkage probabilities can be used to select traits that are significant for having all loci jointly linked by calling all traits significant that have a joint-linkage probability exceeding some threshold. For example, all traits with Pr(loci 1 and 2 are linked \\| Data) ≥ 0.90 may be called significant. This threshold is equivalent to ranking the traits for significance by the size of the joint-linkage probability. (Variations on this ranking procedure are possible, depending on the goals of the study; e.g., one may want to consider only traits that have all locus-specific linkages probabilities at 0.95 or greater---see [Materials and Methods](#s4){ref-type="sec"}.) In order to decide on a reasonable threshold, an error rate associated with the thresholding rule is assessed. For example, how reliable is the list of traits that have a joint-linkage probability of 0.90 or greater? The FDR concept is attractive in this case since thousands of traits are simultaneously being assessed, and we would like to select several without incurring too many mistakes. The FDR is typically defined and estimated in terms of multiple hypothesis tests \[[@pbio-0030267-b27],[@pbio-0030267-b28]\]. However, the "null hypothesis" here is complicated because it includes any scenario where one or both of the loci are not truly linked. However, when assuming a Bayesian model, the FDR has been shown to be equal to a Bayesian posterior probability \[[@pbio-0030267-b30]\], and the FDR concept and estimation methodology can be extended to accommodate our situation. A trait is defined to be a "false discovery" for joint linkage if any of its selected loci is a false positive. In standard multiple hypothesis testing situations, the false discovery has been estimated as the ratio of the estimated number of false positive divided by the observed number of tests called significant. For a given threshold we place on the traits, it is straightforward to count how many are called significant, but it is not as easy to estimate the expected number of false positives because the null distribution of the joint-linkage probabilities is not available. However, when identifying pairs of loci for each trait, the probability a trait is a false discovery is 1 −Pr(loci 1 and 2 are linked\\| Data). Therefore, the overall expected number of false discoveries is the sum of the 1 −Pr(loci 1 and 2 are linked\\| Data) over all traits called significant for two-locus linkage. An estimate of the proportion of false discoveries among significant linkages is then where again the summation is taken over all traits called significant for a two-locus linkage. This estimate can be justified in the context of Bayesian representations of the FDR, but it also has connections to p-value based estimates (\[[@pbio-0030267-b30]\]; [Materials and Methods](#s4){ref-type="sec"}). For example, in our study there are 72 traits that have two-locus joint-linkage probabilities of 0.90 or greater. Summing all 72 corresponding quantities 1−Pr (loci 1 and 2 are linked \\| Data), there are 4.8 expected false discovery two-locus linkages among these. Therefore, the FDR estimate of this particular threshold is 4.8/72 = 6.7%. In practice, the locus-specific and joint-linkage probabilities must be estimated. Due to the massive amount of available data, we form nonparametric estimates of the probabilities rather than making assumptions about their distributions. At each stage of the locus selection, the strength of linkage is quantified by a standard F-statistic used to compare two models (M1 versus M0 or M2 versus M1). The statistics associated with the primary and secondary loci for each trait are the maximal F-statistics among all loci. Since these maximal statistics do not have a known null distribution, the null distributions are simulated. The quantitative trait values are permuted and the maximal statistics are recomputed \[[@pbio-0030267-b25]\] to give permutation null statistics. Note that when the null statistics are simulated for the secondary loci, the fact that the primary loci are assumed to be truly linked is taken into account \[[@pbio-0030267-b26]\]. That is, the null statistics corresponding to the secondary loci are calculated conditionally on the genotypes of the primary loci. We performed five permutations to yield sets of 6,216 × 5 simulated null statistics corresponding to the primary and secondary locus selections. The observed statistics and null statistics corresponding to the primary loci are used to estimate the linkage probabilities Pr(locus 1 linked \\| Data) for each trait. Similarly, the observed and null statistics corresponding to the secondary loci are used to estimate Pr(locus 2 linked \\| locus 1 linked, Data) for each trait. A key aspect of our proposed approach is that loci are selected one at a time for each given expression trait. In the traditional approach, pairs of loci are selected together so that among these locus pairs, zero, one, or two loci may be truly linked. Therefore, it is not possible to model all three cases without making a number of assumptions. However, since we select only one locus at a time, there are only two possible outcomes at each selection step: the locus is either linked or not. The statistics calculated at each locus selection stage are a mixture of the two distributions corresponding to these linked and unlinked loci. The permutation null statistics represent one component of this mixture and can be used in conjunction with the observed statistics to conservatively estimate the locus-specific linkage probabilities. [Figure 2](#pbio-0030267-g002){ref-type="fig"} shows a plot of the strategy used to form these estimates. The solid black line is an empirical probability density (i.e., smoothed histogram) of the 6,216 observed statistics calculated from the secondary loci. This density is a mixture of a null density corresponding to statistics of unlinked loci, and an alternative density corresponding to statistics of linked loci. The solid grey line is an empirical probability density of the permutation null statistics, which has been drawn to reflect its relative contribution to the black mixture density of observed statistics. The observed statistics and permutation null statistics can be used to conservatively estimate the proportion of true linkages among all secondary loci (\[[@pbio-0030267-b27]\]; [Materials and Methods](#s4){ref-type="sec"}), which is the mixing proportion of the null density and the prior probability of linkage. In order to calculate the locus-specific linkage probability, the ratio of the null density to the mixture density must also be estimated. This ratio is estimated by adaptively considering the relative frequency of null statistics to observed statistics in small intervals around each possible value of the observed statistics ([Materials and Methods](#s4){ref-type="sec"}). Once the locus-specific linkage probabilities are estimated, these quantities are plugged into the above proposed procedure to obtain a set of significant two-locus linkages. ::: {#pbio-0030267-g002 .fig} Figure 2 ::: {.caption} ###### An Example of the Locus-Specific Linkage Probability Estimation Applied to the Secondary Loci The estimated density of the observed statistics is plotted (solid black). This density is modeled as a weighted mixture of probability densities corresponding to the "null" unlinked secondary loci (solid grey) and the "alternative" linked secondary loci (dashed grey). The estimated posterior probability of linkage is also shown (dashed black). ::: ![](pbio.0030267.g002) ::: Two-Locus Joint Linkage Applied to Gene expression Traits in Yeast {#s2d} ------------------------------------------------------------------ We applied the proposed method for two-locus linkage analysis to the S. cerevisiae experiment. Based on the joint-linkage probabilities, we estimate that 2,300 traits (approximately 37%) are jointly linked to two loci, although we cannot identify all of these with high confidence. Of these 2,300 traits, 170 can be identified at a FDR of 10%. Among these 170 traits, the primary locus FDR is less than 0.2%. Therefore, we expect that at most 17 of these joint linkages include a single false-positive locus, and about zero include two false-positive loci. Recall that when a more liberal definition of two-locus linkage was used, where only one locus was required to be linked, about 4,000 linkages were called significant at a FDR of 5%. However, in that situation it was not clear whether both loci or just a single locus were truly linked. Because we identify only 170 significant joint linkages at a FDR of 10%, it appears that many of the 4,000 significant linkages from the other approach were due to only a single locus being truly linked. When comparing our method to a traditional 1D linkage scan where the top two linkage peaks are taken as significant, we find 3.3 to 8.7 times more linkages at FDR cut-offs ranging from 1% to 10% ([Materials and Methods](#s4){ref-type="sec"}). These observations indicate that our proposed approach provides a new and statistically rigorous framework for distinguishing between genetic models. To better understand the molecular mechanisms underlying the observed linkages for these traits, we searched for cis-acting effects. Here a cis linkage is said to occur if one of the two linkage peaks coincides with the position of the encoding gene corresponding to the expression trait. In total, 58 traits demonstrate a cis linkage ([Table S1](#st001){ref-type="supplementary-material"}), which has two important implications. First, the observation of a cis linkage immediately suggests a candidate QTL that can be experimentally tested. Second, these results demonstrate that variation in the expression level for a given trait cannot simply be dichotomized into either cis or trans effects, as both can simultaneously contribute to variation in gene expression levels. Several previous linkage analyses of gene expression levels in yeast and other organisms have shown that linkages are nonrandomly distributed throughout the genome and tend to cluster into specific locations \[[@pbio-0030267-b05]-[@pbio-0030267-b07],31\]. In order to get a broad view of the distribution of joint linkages throughout the genome, we first divided the genome into 550-kb bins and counted the number of jointly significant traits at a FDR ≤ 10% in each pair-wise bin ([Figure 3](#pbio-0030267-g003){ref-type="fig"}A), where simulations demonstrate that the number of bins expected to have three or more two-locus linkages by chance is less than one. [Figure 3](#pbio-0030267-g003){ref-type="fig"}A indicates that the genomic distribution of joint linkages does not solely follow simple patterns, which is evidence that the "joint" linkage here is meaningful. To test whether similar observations would extend to pairs of linkages on a finer scale, we further divided the genome into 50-kb bins and counted the number of significant joint linkages occurring in each bin. We observed 10 pair-wise bins with three or more traits ([Figure 3](#pbio-0030267-g003){ref-type="fig"}B), where the number expected by chance is much less than one. This suggests that the same pair of QTL or closely linked QTL contribute to variation in the gene expression levels among all traits falling into any given 50-kb pair-wise bin. Not surprisingly, groups of traits defined by linkage bins possess similar biological functions ([Table S2](#st002){ref-type="supplementary-material"}). For example, the 12 traits that jointly link to nearly identical positions on Chromosomes 3 and 8 are predominantly involved in the mating response. The linkage peak on Chromosome 3 maps to the precise location of the yeast mating type locus MAT. The parental strains are of opposite mating type, and mating type segregates in the cross. We show elsewhere that variation in the expression of genes in this group is indeed explained by inheritance at MAT on Chromosome 3 and at the pheromone response gene GPA1 on Chromosome 8 \[[@pbio-0030267-b32]\]. Common biological themes can be assigned to the majority of the remaining clusters including amino acid and mitochondria metabolism (13 traits defined by linkage to regions on Chromosomes 3 and 13), mitochondrial tricarboxylic acid cycle (ten traits defined by linkage to regions on opposite ends of Chromosome 15), and response to stress (five traits defined by linkage to Chromosomes 6 and 10). [Table S2](#st002){ref-type="supplementary-material"} provides a complete list of genes and putative biological functions for the ten pair-wise bins with three or more traits. ::: {#pbio-0030267-g003 .fig} Figure 3 ::: {.caption} ###### A Plot of the Locus Pair Positions Corresponding to the 170 Traits Significant for Joint Linkage \(A) A plot of the significant locus pair positions when each chromosome has been partitioned into equally sized bins less than or equal to 550 kb. The number of significant traits showing linkage to locus pairs in each pair-wise bin is denoted. The number on each axis indicates the chromosome number; a dash denotes a bin division. \(B) A plot constructed analogously to (A), except bins less than or equal to 50 kb are used, and only bins with three or more traits significant for joint linkage are numbered. ::: ![](pbio.0030267.g003) ::: Another interesting observation that emerges from the spatial distribution of joint linkages is that distinct groups are connected by a common linkage peak. For example, of the 10 pair-wise bins with three or more linked traits, there are three that share a common linkage to the exact same position on Chromosome 15 ([Table S2](#st002){ref-type="supplementary-material"}). Many of the genes in these three groups are localized to the mitochondria, suggesting an important QTL on Chromosome 15 that mediates expression levels for numerous mitochondria related genes. An attractive candidate QTL for this region is IRA2, which is a regulator of the RAS-cAMP pathway \[[@pbio-0030267-b33]\] that is located in both the cytoplasm and mitochondria. More generally, these results intimate that multiple locus mapping of gene expression levels may be useful in reconstructing regulatory networks by identifying shared linkages across traits. Discussion {#s3} ========== We developed a new, computationally efficient statistical method for simultaneously mapping multiple QTL. Whereas conventional linkage analysis has been widely and successfully applied to study one or very few traits at a time, our method is appropriate for analyzing thousands of phenotypes. Pairs of loci were identified sequentially rather than considering all possible combinations, which was shown to be empirically more powerful. The model used to select pairs of loci included an interaction term allowing for possible epistasis. This sequential approach will of course miss locus pairs with primarily epistatic effects (i.e., little or no main effect for either locus), and these may be biologically interesting or important. Also, we have not included any special modifications to handle the case where two QTL are closely linked, although such modifications are likely possible. Even though it is not likely that two locus models give a complete picture of gene regulation \[[@pbio-0030267-b14]\], such analyses may still provide valuable information as we have shown here and elsewhere \[[@pbio-0030267-b32]\]. Since including only two loci may have an adverse effect on power when many QTL affect a trait, it may be helpful to adapt composite interval mapping methods to our approach. However, we were able to observe a number of significant linkages using two QTL models. A major challenge that our method overcomes is to assign joint significance to the pairs of loci. When identifying linked loci in a sequential manner, it is tempting to apply a readily available significance threshold at each stage. For example, existing p-value based FDR methods could have been applied at the first stage to identify a set of significant primary loci. The procedure could then have been repeated on this significant subset to obtain a set of significant secondary loci. Although this may initially appear to be valid, biases are incurred because of the high-dimensional nature of the problem. Specifically, the set of primary loci called significant at the first stage explain a large proportion of variance of their corresponding expression traits, even if the primary loci are false positives. This must be taken into consideration when assessing the significance of secondary loci, which is not the case in simplistic sequential applications of existing p-value based methodology. In our approach, we explicitly took into account the sequential selection procedure in order to obtain an overall significance measure of joint linkage. As technological advances in gene, protein, and metabolite profiling continue to be made, we anticipate that statistical methods such as the one proposed here will provide important insights into the genetic architecture of complex and quantitative traits. Materials and Methods {#s4} ===================== {#s4a} ### Expression measurements {#s4a1} These expression data have recently been reported elsewhere \[[@pbio-0030267-b14]\]. Briefly, 112 F1 segregants (one from each tetrad) were grown from a cross involving parental strains BY4716, isogenic to the lab strain S288C, and the wild isolate RM11-1a \[[@pbio-0030267-b05],[@pbio-0030267-b07],[@pbio-0030267-b14]\]. RNA was isolated and cDNA was hybridized to microarrays \[[@pbio-0030267-b05],[@pbio-0030267-b07],[@pbio-0030267-b14]\]. Each hybridization was done in the presence of the same BY reference material, and all reported expression values are log~2~(sample/BY reference), averaged over two dye-swapped arrays. Each array \[[@pbio-0030267-b34]\] assayed 6,216 yeast ORFs, 13 of which were spotted twice, and we did not incorporate special corrections for potential cross-hybridization \[[@pbio-0030267-b35]\]. ### Genotyping procedure {#s4a2} As previously reported \[[@pbio-0030267-b14]\], GeneChip Yeast Genome S98 microarrays were purchased from Affymetrix (Santa Clara, California, United States). Genomic DNA was isolated and genotype-calling algorithms were performed as before on all 112 F1 segregants \[[@pbio-0030267-b05]\]. The resulting genetic map of 3,312 markers covered more than 99% of the genome \[[@pbio-0030267-b05]\]. When genetic markers are sparse, interval mapping methods \[[@pbio-0030267-b36]\] can be used to impute mixtures of pseudo-markers in between known observed markers in an attempt to increase power to detect linkage. However, more than 90% of adjacent markers have five or less differences among the 112 progeny, and only 1,226 unique sets of alleles exist among the 3,312 typed. In this case, it sufficed to test for linkage only at these unique sets of alleles. The method proposed here is easily extended to the interval mapping paradigm. ### Exhaustive 2D test for linkage and epistasis {#s4a3} All pairs of loci were tested for linkage to a given trait based on an F-statistic comparing the least fitted two-locus full model to the null model of no linkage. In order to ease the computational burden, we considered only 613 equally spaced loci and we did not consider any pairs of loci located on the same chromosome. For each trait, the pair of loci with the largest F-statistic was selected. A p-value was calculated for each selected pair by using a standard permutation technique \[[@pbio-0030267-b25]\]: the ordering of the arrays was randomly permuted, and a new maximal F-statistic was recorded for each trait. Five permutations were carried out, and the p-value was calculated as the frequency of simulated null F-statistics that exceed the observed statistic. Note that in doing this, the null F-statistics were pooled across traits (giving 5 × 6,216 null statistics). This can be justified by noting that the F-statistic is a pivotal statistic and the number of observations (112) is reasonably large. For the test of epistasis, an F-statistic was formed for each pair of loci that compared the full model fit to a purely additive model fit, which directly tests for an interaction between the two loci. For each trait, the pair of loci with the largest F-statistic was selected. In calculating p-values, a similar permutation technique was performed that also takes into account the fact that the null model is the additive model \[[@pbio-0030267-b26]\]. The p-values were corrected for multiple testing by employing the FDR \[[@pbio-0030267-b27],[@pbio-0030267-b28]\]. ### Comparison between 2D and sequential selection procedures {#s4a4} In order to compare the power of the 2D and sequential selection procedures, the sequential locus selection procedure (described below) was also performed exactly as above, on the same loci, the same null permutations, etc. Therefore, the only aspect compared is the exact procedure used to choose a pair of loci. The p-values were obtained from a test against the null hypothesis of no linkage and a test against the null hypothesis of no epistasis. These were compared between the two procedures, and the sequential procedure showed more power in both scenarios (see [Results](#s2){ref-type="sec"}). ### Sequential selection of locus pairs {#s4a5} For each fixed trait i, the primary locus was chosen as the one showing the most single linkage to the trait. Specifically, an F-statistic was calculated for each locus that compares the goodness of fit of the least squares model under the case of no linkage to the least squares model under the case that a single locus is linked. The secondary locus is similarly chosen by fitting a least squares model of trait i on its primary locus and each additional locus under the full two-locus model (which includes their additive terms and an interaction term). The locus showing the best improvement in fit, again quantified by a standard F-statistic, is chosen as the secondary locus. Although pairs of loci residing on the same chromosome were not considered in the comparison to the exhaustive 2D search approach, we place no restriction on loci in the main proposed method, i.e., all available loci are considered at each stage of the sequential selection. ### Calculation of observed and null statistics used to estimate locus-specific linkage probabilities {#s4a6} Let F~i~ ~1~ be the maximal F-statistic corresponding to the primary locus for each trait i = 1, ... , 6,216. Let F~i~ ~2~ be the maximal F-statistic corresponding to the secondary locus for each trait i = 1, ... , 6,216. In general, L loci may be sequentially selected and F~ij~ analogously calculated, j = 1, ... , L. Statistics from the null distributions were simulated by randomly permuting the ordering of the arrays and calculating a new maximal F-statistic for each trait \[[@pbio-0030267-b25]\]. For the secondary locus null distribution, the permutations take place within each segregant group corresponding to the primary locus, which takes into account the fact that the null distribution on the secondary locus statistics is calculated under the assumption that the primary locus is truly linked \[[@pbio-0030267-b26]\]. Five permutations were carried out for each selection stage to yield sets of null statistics F~i~ ~1~ ^0b^ and F~i~ ~2~ ^0b^ for i = 1, ... , 6,216 and b = 1, ... , 5. The null F-statistics corresponding to a given locus selection stage were pooled across traits, yielding 5 × 6,216 null statistics. This can be justified again by noting that the F-statistic is an asymptotically pivotal statistic and the number of observations (112) is reasonably large. ### Nonparametric estimation of locus-specific and joint-linkage probabilities {#s4a7} The observed F~ij~ and null F ~i1~ ^0b^ are directly used to estimate the locus-specific linkage probabilities. Define ℓ~ij~ = 1 if the jth marker chosen for trait i is linked, and ℓ~ij~ = 0 if no linkage exists; i = 1, ... , 6,216 and j = 1, ... , L. A standard Bayesian analysis would parameterize a model for the data and also assign prior probabilities to the ℓ~i~ ~1~, ℓ~i~ ~2~, ... , ℓ~i~ ~L~. Since there is an enormous amount of data available, we can avoid making some of these assumptions. Let F~ij~ be the statistic corresponding to the jth locus chosen for trait i. First, we replace Pr(ℓ~ij~ = 1\\|ℓ~i~ ~1~ = 1, ... , ℓ~i,j~ ~−1~ = 1, Data) with Pr(ℓ~ij~ = 1\\|ℓ~i~ ~1~ = 1, ... , ℓ~i,j~ ~−1~ = 1, F~ij~) which may lead to a loss of information at the cost of making less assumptions. The F~ij~ and F~ij~ ^0b^ are calculated under the assumption that ℓ~i~ ~1~ = 1, ... , ℓ~i,j~ ~−1~ = 1 so this is a coherent formulation. All of the information shown in [Figure 2](#pbio-0030267-g002){ref-type="fig"} is not needed in order to simply estimate the locus-specific linkage probabilities. There are essentially only two components that need to be estimated. Take, for example, the calculation of the primary locus probability of linkage, and suppose that the null and alternative distributions of F~i~ ~1~ have probability density functions g ~0~ and g ~1~, respectively. Then if π~0~ of the primary loci are not linked and π~1~ are linked, a randomly selected F~i~ ~1~ follows the mixture density g = π~0~ g ~0~ + π~1~ g ~1~. (The density functions g ~0~ and g ~1~and prior probabilities π~0~ and π~1~ are not assumed to be the same at each locus selection stage. Also, if differ between the traits, one can view g ~0~ and g ~1~ as the average of these.) According to Bayes theorem, the posterior probability of linkage for the primary locus is Since F~i~ ~1~ are observations from g = π~0~ g ~0~ + π~1~ g ~1~ function, and the simulated null are observations from g ~0~, these two sets of statistics can be used to estimate the likelihood ratio g ~0~/g, where we define R(F) = g ~0~(F)/g(F). Anderson and Blair \[[@pbio-0030267-b37]\] have shown that with a fixed number of observations from two probability densities, it is valid to estimate their likelihood ratio by performing a logistic regression where, say, the F~i~ ~1~ are called "successes" and the are called "failures." Methods to perform this logistic regression with a nonparametric link function have been previously developed \[[@pbio-0030267-b37],[@pbio-0030267-b38]\]. Using this technique, we form an estimate of the likelihood ratio function denoted by . Specifically, the link function is parameterized by a natural cubic spline as previously described \[[@pbio-0030267-b38]\], with 6,216 knots evenly distributed among all observed and null statistics. A similar procedure has been applied in several applications, for example, in identifying differentially expressed genes \[[@pbio-0030267-b06]\]. The quantity π~0~ is estimated by This estimate was originally formulated for use in estimating p-value based FDRs \[[@pbio-0030267-b27],[@pbio-0030267-b28]\]. It is straightforward to show under our assumptions that the expected value of is greater than or equal to π~0~, thus providing a conservative estimate. Adjusting the tuning parameter c allows one to balance bias and variance in the estimate. In order to automatically deal with the choice of c, we smoothed over a range of c using a technique previously described \[[@pbio-0030267-b28]\]. In this context, another estimate of π~0~ has been suggested as \[[@pbio-0030267-b39]\]; however, we found this to be much too unstable. For the primary loci we estimate , and for the secondary loci . The overall primary locus linkage probability estimate is then for i = 1, . . . , 6,216. The secondary locus linkage probability estimate (and any subsequently selected locus) is formed analogously based on its observed statistics F~i~ ~2~ and simulated null statistics . Finally, the two-locus joint-linkage probability is estimated by ### FDR estimation {#s4a8} We ranked the traits for significance by the magnitude of the and chose significance cut-offs by calling all trait-locus pair combinations significant that have for some λ. Let S ~λ~ be the set of traits called significant with this threshold and be the number of traits called significant. Defining a trait to be a false discovery if either locus is a false positive, we estimated the FDR by Setting λ = 0.84, we estimate the FDR to be 10%. The estimate can be generalized to L loci by simply replacing with . This procedure can also be made more general by noting that any thresholding rule may be used. For example, if one wants to consider traits where both locus-specific linkage probabilities reach a certain level, one may define S ~λ~ by those traits where . Suppose that one wants to guarantee that , but it is unknown how many loci L to choose for each trait. Define ; if this is not true for any value set . Calling all traits i and top loci jointly linked (among traits with ), then it follows that over this set. Rather than motivating this estimate of the FDR from a model-based Bayesian framework \[[@pbio-0030267-b40],[@pbio-0030267-b41]\], we can justify it from the more common frequentist viewpoint. The FDR is usually estimated among multiple hypothesis tests where the null hypothesis is simple (i.e., contains only one parameter value). However, here the null hypothesis consists of the three scenarios where , , and . Using previously developed theory \[[@pbio-0030267-b30]\], it follows that In situations where the FDR can be written in this way, it has been shown in a variety of scenarios that estimates of the form control the FDR as long as the estimated number of expected false discoveries is conservative as the number of traits gets large \[[@pbio-0030267-b42]\]. Specifically, we can consider the to be random variables and view the numerator and denominator of the estimate to be based on the empirical distribution function of these random variables. It has been shown that as long as the empirical distribution function converges properly, then the above FDR estimate is conservative not only at a fixed threshold, but also at all adaptively chosen thresholds \[[@pbio-0030267-b42]\]. The main hurdle to overcome beyond this existing theory \[[@pbio-0030267-b42]\] is to show that is a consistent estimate of g ~0~(F)/g(F) in such a way that the convergence of the empirical distribution function is not adversely affected. Nonparametric logistic regression is quite flexible and provides consistent estimates in a fairly general sense as long as the smoothness of the link function decreases at a proper rate \[[@pbio-0030267-b38]\]. Thus, a reasonably general frequentist justification of our approach based on existing theory appears to be within reach. A battery of simulations (data not shown) indicates that our proposed approach provides reliable significance estimates in a variety of scenarios. ### Comparison between 1D linkage scan and proposed method {#s4a9} In a traditional 1D linkage scan, a statistic is calculated at each marker and a significance threshold is applied to these in order to find markers showing significant linkage. It is possible that more than one linkage statistic will exceed the threshold. Therefore, one could view this procedure as a multiple locus linkage analysis. We compared this approach to our proposed method by thresholding the two top linkage peaks for each trait. (Peaks were considered to be distinct if they lay on different chromosomes.) A p-value was calculated for each trait under the alternative hypothesis that both peaks are true linkages, making the hypothesis test equivalent to ours. Specifically, the p-value was defined to be the probability that a statistic exceeded the minimum of the two peaks under the case of no linkage. The p-values were then used to estimate the FDR at various significance cut-offs \[[@pbio-0030267-b28]\]. Since this 1D approach does not take into account any interaction between the two loci, we compared it to our proposed method when using a purely additive model in order to select loci. We found that our method yields 3.3 to 8.7 times more linkages at FDR cut-offs ranging from 1% to 10%. Supporting Information {#s5} ====================== Table S1 ::: {.caption} ###### Information about the 170 Traits That Possess Joint Linkage at FDR Less Than or Equal to 10% The "q-value-Joint" column gives the q-value for joint linkage, where a q-value is the FDR analog of the p-value. In the "Cis" column, zero denotes no cis linkage, one denotes cis linkage to the primary locus (locus 1), and two denotes cis linkage to the secondary locus (locus 2). (44 KB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Table S2 ::: {.caption} ###### Linkages That Cluster according to the Pair-Wise Position of the Two Loci The genome was split into 50-kb pair-wise bins, and the number of significant linkages at FDR less than or equal to 10% falling into each bin was recorded. For any bin containing more than three linkages, the exact marker positions and expression traits are listed below. (49 KB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Accession Numbers {#s5a} ----------------- The expression data reported in this paper have been deposited in the Gene Expression Omnibus (GEO) (<http://www.ncbi.nlm.nih.gov/geo>) database (GSE1990). This work was supported in part by NIH grants R01 HG002913--01 (JDS) and R37 MH59520--06 (LK), National Science Foundation Postdoctoral Fellowship 0305916 (JMA), and by the Howard Hughes Medical Institute (LK). LK is a James S. McDonnell Centennial Fellow. We thank J. Whittle for generating microarray data, R. Brem for several useful discussions, and three anonymous referees for helpful comments on the manuscript. Competing interests. The authors have declared that no competing interests exist. Citation: Storey JD, Akey JM, Kruglyak L (2005) Multiple locus linkage analysis of genomewide expression in yeast. PLoS Biol 3(8): e267. Author contributions. JDS, JMA, and LK conceived and designed the experiments, analyzed the data, and wrote the paper. 1D : one-dimensional 2D : two-dimensional FDR : false discovery rate kb : kilobase QTL : quantitative trait loci	PubMed Central	2024-06-05T03:55:59.582928	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180512/", "journal": "PLoS Biol. 2005 Aug 26; 3(8):e267", "authors": [ { "first": "John D", "last": "Storey" }, { "first": "Joshua M", "last": "Akey" }, { "first": "Leonid", "last": "Kruglyak" } ] }
PMC1180513	Introduction {#s1} ============ The first known left--right (L-R) symmetry-breaking event in mouse development occurs at the ventral surface of the node at embryonic day 8.0 (E8.0) \[[@pbio-0030268-b01]\]. Previous studies showed that hundreds of monocilia on the node rotate in a clockwise direction and generate leftward flow of extraembryonic fluid, referred to as nodal flow \[[@pbio-0030268-b02]\]. The direction of this flow determines which side will be the left, because the L-R decision is reversed when the flow direction is reversed by imposing artificial flow \[[@pbio-0030268-b03]\]. Although nodal flow is clearly important, at least in the mouse, several questions remain to be answered. First, it is unknown how the nodal flow works to control the L-R decision. Two models have been proposed. The nodal flow may transport a signaling molecule toward the left side \[[@pbio-0030268-b02]\]. Alternatively, mechanical stress generated by the flow may be sensed by a mechanosensor on node cells \[[@pbio-0030268-b04],[@pbio-0030268-b05]\]. Currently, there is no direct evidence for one model or the other. Second, it is not easy to envisage how unidirectional flow is generated by rotating cilia, because a rotational movement should cause a vortex, and the sum of the vortices should result in a circular flow around the edge of the node. In practice, however, we never observe such a vortical flow but only a linear leftward flow. Although theoretical models have been proposed \[[@pbio-0030268-b06]\] in which rotational movement can result in unidirectional flow, their validity has not been tested experimentally. Finally, it remains to be clarified whether the flow plays a conserved role among vertebrates, although fluid flow has been detected in the Kupffer\'s vesicle of zebrafish embryo, which may be equivalent to the mode node \[[@pbio-0030268-b07]\]. A recent paper by Cartright et al. \[[@pbio-0030268-b06]\], describing theoretical fluid dynamics simulations of nodal cilia rotation, showed that a linear flow could result if the rotation axis of the cilia has a posterior tilt. Their simulation predicted a three-layered flow pattern: Rotational movement of cilia directly makes a diphasic flow that is leftward at the level of the cilia tips and rightward closer to the node floor. The third layer is a boundary-induced return flow to the right, predicted on the basis of conservation of mass if Reichert\'s membrane seals the cavity surrounding the node. This theoretical simulation can explain how rotational motion is converted to linear flow. An interesting feature of the Cartwright model is the fact that the leftward flow that is usually observed in the node is balanced by two layers of rightward flow. Boundary-induced rightward flow across the top of the node is necessary because in utero development occurs within a closed volume inside of Reichert\'s membrane. Such a rightward flow, however, appears to be unnecessary for generating L-R asymmetry because wild-type embryos with Reichert\'s membrane removed develop normal situs, and an immotile cilia mutant (inversus viscerum \[iv\]) \[[@pbio-0030268-b08]\] can establish L-R asymmetry in artificial flow experiments \[[@pbio-0030268-b03]\] in which the flow pattern is completely linear and unidirectional. More problematic is the prediction that a second layer of active rightward-directed flow should occur near the node floor. This rightward flow near the surface of the node has never been observed in the mouse embryo \[[@pbio-0030268-b02],[@pbio-0030268-b03],[@pbio-0030268-b09]\]. Brokaw \[[@pbio-0030268-b10]\] has pointed out that surface interactions could produce a leftward flow from rotational ciliary movement without generating any significant rightward flow near the surface. Usually fluid contacting a solid wall will not move (the so-called no-slip boundary condition), and this static fluid layer should impede overlying fluid from being dragged by the cilia due to viscous force \[[@pbio-0030268-b11]\]. This effect should be greatest near the surface. Cartwright et al. \[[@pbio-0030268-b06]\] mention that the flow close to the floor is slower due to this effect, but in fact if this effect is great enough, it could effectively produce unidirectional propulsion toward the left ([Figure 1](#pbio-0030268-g001){ref-type="fig"}). As proposed by Brokaw \[[@pbio-0030268-b10]\], if a clockwise rotating cilium has posterior tilt, it passes closest to the surface during the rightward phase and farthest from the surface during the leftward, and thus should drag surrounding water the least efficiently during rightward motion and the most efficiently during leftward motion. In total, while at a very small scale, the local flow will have a rotational component; at large scales the leftward flow will predominate over the rightward, and the rightward flow cannot make a continuous flow layer. Brokaw\'s model can account for the fact that cilia-induced rightward flow was never observed in our previous bead-tracking experiments \[[@pbio-0030268-b02],[@pbio-0030268-b09]\], thus reconciling the essential aspects of Cartwright\'s tilted cilia model with the experimental data that rightward flow near the surface is not observed in vivo. Exploitation of surface effects, by dragging motile limbs close to the surface during a return stroke, is a common strategy for generating directed motile forces in small organisms such as copepods \[[@pbio-0030268-b12]\]. Brokaw\'s model proposes a fundamentally similar strategy for generating nodal flow. Indeed, this model is based on previous discussions of propulsion by ciliary movement in which return strokes occur near a surface \[[@pbio-0030268-b13]\], and thus the model does not invoke any unusual fluid dynamics beyond what has already been described in other systems. ::: {#pbio-0030268-g001 .fig} Figure 1 ::: {.caption} ###### Flow Generation Mechanism Circular clockwise motion of a cilium can generate directional leftward flow if its axis is not perpendicular to the cell surface but tilted posteriorly. Due to distance from the cell surface, a cilium in the leftward phase (red arrow) drags surrounding water more efficiently than the rightward phase (blue arrow), resulting in leftward force in total (purple arrow). See text for details. The rotating cilium is drawn slightly bent due to viscous resistance, as seen in [Video S1](#sv001){ref-type="supplementary-material"}. ::: ![](pbio.0030268.g001) ::: In this paper, we have studied how unidirectional flow is generated by the rotational movement of node cilia, by testing the validity of Brokaw\'s model \[[@pbio-0030268-b10]\]. Our data suggest that node cilia do not protrude perpendicular to the node surface but are tilted posteriorly, and that this posterior tilt is responsible for generating the leftward flow. Therefore, L-R symmetry breaking is achieved by integration of the pre-existing positional information, anterio-posterior and dorsoventral information. Results/Discussion {#s2} ================== Node Cilia Are Tilted Posteriorly {#s2a} --------------------------------- To test the validity of Brokaw\'s theoretical model experimentally, we first examined whether the node cilia are really tilted posteriorly. Trajectories of the cilia were tracked by a high-speed camera attached to a differential-interference contrast (DIC) microscope, followed by image processing to enhance subtle contrast ([Figure 2](#pbio-0030268-g002){ref-type="fig"}A, [Video S1](#sv001){ref-type="supplementary-material"}). If a cilium is rotating around an axis that is directed straight up (that is, toward the ventral axis), the movements of the tip would describe a circle with the root of the cilium located in the center ([Figure 2](#pbio-0030268-g002){ref-type="fig"}B, top). In contrast, if a cilium is tiltedposteriorly, the tip should appear as an ellipse and the position of the root should be displaced from the center anteriorly ([Figure 2](#pbio-0030268-g002){ref-type="fig"}B, middle). The majority of observed cilia had exactly such a pattern. Some cilia traced out the shape of a "D", suggesting that the trajectory of the tip was not strictly circular. This shape resembled that of Paramecium cilia \[[@pbio-0030268-b14],[@pbio-0030268-b15]\]. Paramecium cilia have an effective stroke and a recovery stroke that take place in different planes, resulting in a D-shaped pattern when viewed from the top. The change in bending plane between the effective and recovery strokes is thought to be mediated by rotation of the central pair microtubules in Paramecium cilia \[[@pbio-0030268-b16]\]. However, the nodal cilia are the 9+0 type, which lacks the central pair \[[@pbio-0030268-b17]\], making it unlikely that the node cilia could move with a Paramecium-like two-phase beat pattern. Furthermore, Paramecium cilia usually beat in a D-shape pattern \[[@pbio-0030268-b14]\], not in the elliptical paths that were observed for the majority of node cilia. ::: {#pbio-0030268-g002 .fig} Figure 2 ::: {.caption} ###### Trajectory of Node Cilia Movement \(A) Trace of node cilia in enhanced DIC images after background subtraction. Positions of root are indicated in black, and tip in blue, green, and orange. Most cilia have a pattern consistent with the projection of a tilted cone (blue and green, see text) whereas some cilia move in a D-shape (orange). A, P, L, and R refer to anterior, posterior, left, and right sides of the node, respectively. The direction of cilia rotation was clockwise (arrows). \(B) Relationship between essentially rotatory movement of cilia and their projected images at various tilt angles. ::: ![](pbio.0030268.g002) ::: An alternative interpretation of the observed D-shape pattern is to consider the D-shape pattern as simply a rotating cilium whose axis is tilted at a sufficiently acute angle relative to the surface that the tip cannot trace a circle without hitting the surface ([Figure 2](#pbio-0030268-g002){ref-type="fig"}B, bottom). In such a case, the curved portion of the "D" results when the cilium scrapes along the surface of the node, whereas the straight portion results when the cilium rotates up away from the surface. Given the orientation of the "D" shapes observed, it is notable that in all cases it appears that the movement away from the surface occurs during the leftward phase, whereas the scraping movement along the surface is during the rightward phase. As discussed by Brokaw \[[@pbio-0030268-b10]\], the no-slip boundary condition would then predict that during this rightward phase, little fluid would be moved by the cilium due to its proximity with the immobilized fluid layer at the surface. Therefore, the D-shaped trajectories of moving cilia emphasize the potential for surface interactions to have a major effect on flow generation, arguing strongly that such interactions should never be neglected. Second, we measured tilt of the cilia from scanning electron microscopy images ([Figure 3](#pbio-0030268-g003){ref-type="fig"}). Posterior tilt is immediately apparent from side-viewed images ([Figure 3](#pbio-0030268-g003){ref-type="fig"}A); however we performed three-dimensional (3D) stereographic analysis to eliminate any possibility that the tilt might be an illusion caused by the viewpoint. We used iv/iv mutants for stereographic analysis because, in the absence of dynein-driven motility \[[@pbio-0030268-b08]\], we expect that the direction of the rigid cilia \[[@pbio-0030268-b09]\] would indicate the orientation of the basal body and therefore represent the axis of the conic pathway around which wild-type cilia would rotate. We measured 225 cilia from six different embryos, and by comparing the orientation of each cilium with the orientation of a plane best-fit to the node surface, we found that the cilia showed an average 26.6° tilt towards the posterior. The individual average tilt angles within each embryo ranged from 15° to 35° but were always directed towards the posterior ([Figure 3](#pbio-0030268-g003){ref-type="fig"}D), as required by theoretical models for flow generation by tilted rotation \[[@pbio-0030268-b06],[@pbio-0030268-b10]\]. The fact that the same orientation is observed in multiple embryos fixed independently argues that the tilt is not due to artifacts generated during sample preparation. In the scanning electron microscopy images, we also noted that most of the node cilia protrude from the posterior region of the node pit cells. This was the case, not only for iv/iv cilia ([Figure 3](#pbio-0030268-g003){ref-type="fig"}C), but also for wild-type cilia ([Figure 3](#pbio-0030268-g003){ref-type="fig"}A), and was confirmed by immunofluorescent analysis of basal bodies ([Figure 3](#pbio-0030268-g003){ref-type="fig"}E and [3](#pbio-0030268-g003){ref-type="fig"}F). Out of 83 basal bodies examined, 73 were located in the posterior half of the cells ([Figure 3](#pbio-0030268-g003){ref-type="fig"}F). This posteriorly oriented location of the cilia within cells whose surface is rounded may be responsible for the posterior tilt of the cilia ([Figure 3](#pbio-0030268-g003){ref-type="fig"}G). Additionally, another paper recently published also reports posterior tilt of the node cilia \[[@pbio-0030268-b18]\]. ::: {#pbio-0030268-g003 .fig} Figure 3 ::: {.caption} ###### Node Cilia Are Posteriorly Tilted and Positioned \(A) Scanning electron micrograph of the wild-type node. Note that cilia emanate from the posterior part of the cells. The view angle is about 30° with respect to the horizontal line. (B, C) Scanning electron micrograph of the iv/iv node. (C) is a high-magnification picture of the region in (B) indicated by an arrow. \(D) Deduced tilt of iv/iv node cilia after stereography from multiple-tilt scanning electron micrograph images. Yellow lines indicate cilia, red dots their root positions, and a blue square a plane best-fit to the node surface. When we calculated the tilt of the cilia, we separated the tilt into A-P (anterior--posterior) and L-R components. The average tilt was 26.6° in A-P axis (toward the posterior) and 0.06° in L-R axis (towards the right). \(E) Immunofluorecence image of node cells shown as projection of 3D confocal data stack. Basal bodies and cell boundaries are shown by immunofluorescence against γ-tubulin (red) and ZO-1 (green), respectively. \(F) 3D reconstruction of (E) viewed from ventral side (top) and right side (bottom), showing posterior bias of basal body positions. White lines divide the cells into the anterior and the posterior halves. Basal bodies located in the anterior and the posterior are shown in yellow and red, respectively. \(G) Speculative interpretation of posterior bias of basal bodies in orientation and position of the node cilia. Because the node cells are somewhat rounded, if basal bodies were located at the posterior part of these cells, it would result in posteriorly tilted cilia even though the basal bodies remain perpendicular to the plasma membrane. ::: ![](pbio.0030268.g003) ::: Tilted Rotation Can Generate Unidirectional Flow {#s2b} ------------------------------------------------ Finally we confirmed experimentally that tilted rotation can generate leftward flow without any detectable rightward flow near the surface, as predicted by considerations of surface effects on flow \[[@pbio-0030268-b10],[@pbio-0030268-b13]\]. For this purpose we constructed a mechanical model of 1000 × in scale: Five wires (0.2-mm diameter, 6-mm long) rotate with conical pathways and stir viscous silicone fluid of 30,000 centipoise ([Figure 4](#pbio-0030268-g004){ref-type="fig"}A and [4](#pbio-0030268-g004){ref-type="fig"}B). As shown in [Figure 4](#pbio-0030268-g004){ref-type="fig"}C ([Video S2](#sv002){ref-type="supplementary-material"}) and 4D, this model can indeed generate leftward flow from ciliary rotation. The velocity of the leftward flow depends on the tilt angle: A faster flow was generated when the tilt angle was larger ([Figure 4](#pbio-0030268-g004){ref-type="fig"}D). Closer to the surface, the flow was still leftward ([Figure 4](#pbio-0030268-g004){ref-type="fig"}E; [Video S3](#sv003){ref-type="supplementary-material"}), indicating that any rightward flow that may occur is insignificant in magnitude relative to the leftward flow, consistent with our predictions based on the no-slip boundary condition. Unidirectional flow was not formed if the rotation axis of the wires was not tilted ([Video S4](#sv004){ref-type="supplementary-material"}). ::: {#pbio-0030268-g004 .fig} Figure 4 ::: {.caption} ###### Experimental Fluid Dynamics Model of the Posterior Tilt Mechanism \(A) Schematic representation of the model. Wires driven by stepper motors stir silicone fluid of 30,000 centipoise. Note that the fluid surface is in contact with the acrylic board. Resulting flow is visualized as movement of glitter particles inserted through holes on the board. Wires are connected to motor axles by elastomer connectors in order to smoothen the stepwise motion of the motors. The path of the wires is completely specified by the two angles θ and ψ, as shown in the diagram. \(B) Photograph of the model. Corresponding directions are indicated. \(C) Leftward flow. Flow generated by five rotating wires (visible as white lines in photo) drives glitter towards the left at the depth of 3 mm from the surface. Trajectory of the glitter becomes curved far from the wires (t = 3′) because the glitter has reached the chamber\'s wall. Scale = 1 cm. \(D) Flow velocity (leftward component) as a function of tilt and bend angles. The flow velocity shown here represents the flow speed at the depth of 3 mm. The flow is most efficient when the path of the wire is tangential to the board surface during part of its rotation (i.e., ψ + θ = 90°). \(E) Leftward flow near the surface. Glitter was injected on the right side at the surface of the fluid (no more than 1-mm depth). Only leftward flow was observed. Scale = 1 cm. ::: ![](pbio.0030268.g004) ::: To simulate fluid dynamics at different size scales, it is important to consider the Reynolds number (Re). Re is the ratio between inertial force to viscous force and is very low in microscopic phenomena, i.e., viscous force is dominant and inertial force negligible. The Re for ciliary rotation (Re~c~) has been calculated as Re~c~ ≍ 5 × 10^−4^ in node cilia \[[@pbio-0030268-b06]\]. From [Figure 2](#pbio-0030268-g002){ref-type="fig"}A we estimate the length of cilia is about 6 μm, then Re~c~ ≍ 1 × 10^−3^. In our mechanical model, Re~c~ = 2 × 10^−2^. Although the two Reynolds numbers are different, they both fall in the "creeping flow" regime of low Re in which inertial forces are negligible, thus the same laws of motion will govern fluid behaviour in both cases. To confirm this, another model was constructed that had a single cilia with Re~c~ = 4.6 × 10^−4^. This model also produced leftward flow in the layer close to the surface ([Video S5](#sv005){ref-type="supplementary-material"}). Concluding Remarks {#s2c} ------------------ Here we demonstrated that the node cilia are tilted posteriorly and that rotation of posteriorly tilted cilia can produce leftward flow instead of vortices. Therefore, L-R asymmetry of the body without preceding L-R asymmetry can be explained by relying on information obtained from the antero-posterior and dorso-ventral axes and the chirality of ciliary movement. In this model, dorso-ventral axis information is used to orient the basal bodies along the apical--basal axis of ventral node cells, and antero-posterior axis information is used to tilt this axis towards the posterior. Finally, the inherent chirality of the cilium is used to bias the rotation direction clockwise, thus leading to directed flow. Brown and Wolpert \[[@pbio-0030268-b19]\] hypothesized an "F-molecule" for establishing L-R asymmetry in vertebrate development. A chiral structure would create the first difference along L-R axis by orienting itself with the other axes, then this nanometer scale asymmetry would be amplified to cellular and organism scale by directional transport of another molecule. We propose that the node cilia fulfill the criteria for the F-molecule in the mammalian case. Materials and Methods {#s3} ===================== {#s3a} ### Motion analysis of live node cilia {#s3a1} E8.0 mouse embryos were dissected as previously described \[[@pbio-0030268-b02]\]. Motion of the cilia was recorded with a high-speed camera at 200 frames/s (HAS-200EX, Ditect, Tokyo, Japan) attached to a DIC microscope (Axiophot-2, Zeiss, Oberkochen, Germany) with 63× water-immersed objective and 2.5× intermediate lens. The background of the images was subtracted and the contrast enhanced using a custom-made Ruby script. All the traceable cilia were digitized at roots and tips using free Object-Image software developed by Norbert Vischer (<http://simon.bio.uva.nl/object-image.html>) ### Scanning electron microscopy and stereography {#s3a2} E8.0 mouse embryos were processed using standard procedures \[[@pbio-0030268-b02]\] and observed by a scanning electron microscope (S-800, Hitachi, Tokyo, Japan). Sets of photographs were taken with 0° and plus or minus 10° tilts at 6,000× magnification for cilia and one at 200× for the whole embryo to measure the direction of the antero-posterior axis. One embryo was excluded from further analysis because it had severely curved cilia, presumably due to a fixation artefact. Photographs were digitized and used to calculate 3D coordinates using custom software written within Matlab (MathWorks, Natick, Massachusetts, United States.). Because the high-magnification photographs cover only a small part of the node, the surface of the node can be approximated as a plane within each dataset. Tilt angles were calculated as the angle between each cilium and the normal vector to the planar surface. Results were visualized with free Rotater software written by Craig Kloeden (<http://casr.adelaide.edu.au/rotater/>). ### Immunofluorescence analysis {#s3a3} As in our previous paper \[[@pbio-0030268-b20]\], node tissue fixed with ice-cold acetone was stained by a rabbit polyclonal antibody to γ-tubulin (Sigma, St. Louis, Missouri, United States) and a mouse monoclonal antibody to ZO-1 kindly provided by S Tsukita \[[@pbio-0030268-b21]\]. Confocal images were obtained with a LSM 510 confocal microscope (Zeiss) and analyzed using Object-Image software. Lines dividing the cells into the anterior and the posterior halves were determined as the line that passes through the center of mass of the cell borders and is perpendicular to antero-posterior axis of the embryo. ### Fluid dynamic experiments using a mechanical model {#s3a4} Copper wire of 0.2 mm diameter was bent at 45° or 60°, then cut so that the length of wire protruding into the silicone fluid would be 6 mm. In the five-wire model, the wires were connected to small stepper motors (CAM-60, Canon, Tokyo, Japan) and rotated at 2.7 Hz in silicone fluid of 30,000 centipoise (Clearco, Bensalem, Pennsylvania, United States) in a rectangular Petri dish (AW2000, Eiken-Kizai, Tokyo, Japan) of inner dimensions 138 mm × 98 mm × 13 mm. To visualize fluid movement, glitter was directly applied to the surface up to a depth of 1 mm or 3 mm. In the single-wire model, a wire bent at 45° was connected to a DC motor with a reduction gearbox and rotated at 5.6 × 10^−2^ Hz. Glitter was applied up to a depth of 1 mm. Supporting Information {#s4} ====================== Video S1 ::: {.caption} ###### Motion of Node Cilia Described in [Figure 2](#pbio-0030268-g002){ref-type="fig"}A The video is 20 times slower than real time. See [Figure 2](#pbio-0030268-g002){ref-type="fig"}A for details. (4.8 MB MOV). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S2 ::: {.caption} ###### Flow Generated by the Mechanical Model Described in [Figure 4](#pbio-0030268-g004){ref-type="fig"}C The video speed is 30 times real time. (64 KB MOV). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S3 ::: {.caption} ###### Leftward Flow Generated near the Surface Shown in [Figure 4](#pbio-0030268-g004){ref-type="fig"}E The video speed is 30 times real time. (370 KB MOV). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S4 ::: {.caption} ###### Flow Resulting in a Big Vortex around Rotating Wires, Rather than Linear Flow, When Their Rotation Axis Is Not Tilted (θ = 0°, ψ = 45°) The video speed is 30 times real time. (398 KB MOV). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S5 ::: {.caption} ###### Leftward Flow Generated by a Single-Wire Model with a Lower Re~c~ The Re~c~ = 4.6 × 10^−4^. The video speed is 480 times real time. (266 KB MOV). ::: ::: {.caption} ###### Click here for additional data file. ::: We thank Masahide Kikkawa, Bob Shadel, Michael O\'Grady, Shoichiro Tsukita, Sachiko Tsukita, Tatsuhiko Kodama, Itsushi Minoura, Ken-ichi Wakabayashi, Shinji Kamimura, and Atsuko Namiki for technical assistance, insightful discussion, and reagents. This work was supported by CREST (Core Research for Evolutional Science and Technology) of the Japan Science and Technology Corporation and a University of California, San Francisco Sandler Program Opportunity Award. SN was supported by a fellowship from the Japan Society for the Promotion of Science for Japanese Junior Scientists. Competing interests. The authors have declared that no competing interests exist. Citation: Nonaka S, Yoshiba S, Watanabe D, Ikeuchi S, Goto T, et al. (2005) De novo formation of left--right asymmetry by posterior tilt of nodal cilia. PLoS Biol 3(8): e268. Author contributions. SN, TG, WFM, and HH conceived and designed the experiments. SN, SY, DW, SI, and TG performed the experiments. SN analyzed the data. TG contributed reagents/materials/analysis tools. SN, WFM, and HH wrote the paper. DIC : differential-interference contrast iv : inversus viscerum L-R : left--right Re : Reynolds number 3D : three-dimensional	PubMed Central	2024-06-05T03:55:59.588190	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180513/", "journal": "PLoS Biol. 2005 Aug 26; 3(8):e268", "authors": [ { "first": "Shigenori", "last": "Nonaka" }, { "first": "Satoko", "last": "Yoshiba" }, { "first": "Daisuke", "last": "Watanabe" }, { "first": "Shingo", "last": "Ikeuchi" }, { "first": "Tomonobu", "last": "Goto" }, { "first": "Wallace F", "last": "Marshall" }, { "first": "Hiroshi", "last": "Hamada" } ] }
PMC1180514	Introduction {#s1} ============ In the developing mammalian brain, axons elaborate to form complex arbors that innervate specific target cells. In the peripheral and central nervous systems, axons have an early phase of rapid growth, followed by a period of activity-dependent pruning \[[@pbio-0030272-b01]--[@pbio-0030272-b08]\]. But the notion that axons develop through sequential "growth then pruning" remains controversial \[[@pbio-0030272-b09]\], as axons may undergo simultaneous growth and pruning \[[@pbio-0030272-b10]--[@pbio-0030272-b12]\] or monotonic growth \[[@pbio-0030272-b13]\]. Time-lapse in vivo imaging has been critical for understanding axonal growth and refinement in lower vertebrates \[[@pbio-0030272-b14],[@pbio-0030272-b15]\]. In contrast, prior studies of the developing mammalian brain have relied on measurements in fixed brain preparations. Therefore, relatively little is known about how axonal projections are refined within their target regions, at the level of a single cortical layer or column, or about the exact time course of growth and pruning. Even less is known about the elaboration of interneuron axonal arbors. Axonal growth, guidance, and branching are key elements of axonal development. These developmental processes have been studied most thoroughly in cultured neurons. Axon elaboration involves the extension of distal tips and the formation of new processes by branching. The tips of growing axons classically exhibit elaborate growth cones with multiple filopodia \[[@pbio-0030272-b16],[@pbio-0030272-b17]\]. Growth cones are thought to be critical for guidance by contact-mediated or chemical cues \[[@pbio-0030272-b18]\]. For example, an individual axon\'s growth cone morphology can vary according to whether the axon is growing rapidly along a fiber tract or pausing at a decision-making point \[[@pbio-0030272-b19]--[@pbio-0030272-b23]\]. It is also possible that different growth programs for different axon types are associated with distinct growth cone geometries. The exact role of the terminal growth cone in branching is also a point of contention \[[@pbio-0030272-b22],[@pbio-0030272-b24]\]. Although direct (growth cone splitting \[[@pbio-0030272-b25]\]) and indirect (delayed branching \[[@pbio-0030272-b26]\]) roles of growth cones have been proposed for branching of some axons, interstitial branching independent of growth cones also plays a prominent role in axon elaboration in some cases \[[@pbio-0030272-b27]\]. The relative importance of these processes in the intact cortex is unknown. Branch elimination is another prominent feature of axon development. Two general processes have been described in developing neurons in various species. In Drosophila, axons of mushroom body neurons are eliminated during metamorphosis by axonal degeneration \[[@pbio-0030272-b28]\], a process reminiscent of Wallerian degeneration \[[@pbio-0030272-b29],[@pbio-0030272-b30]\]. In the mouse hippocampus, pruning is presumably manifested as axon tip retraction \[[@pbio-0030272-b31]\] (but see \[[@pbio-0030272-b32]\]). It is conceivable that different types of axons prune branches in different ways, or that these different modes of pruning may operate over different length scales \[[@pbio-0030272-b33]\]. To study axonal growth and pruning in the intact brain we used two-photon microscopy in transgenic mice expressing membrane-bound green fluorescent protein (GFP) under the thy-1 promoter \[[@pbio-0030272-b34]\]. We imaged the elaboration of cortical axons within layer 1 during early postnatal development. We compared long-range projection axons of thalamocortical (TC) neurons with local axons of Cajal-Retzius (CR) interneurons. TC axons begin to invade the cortex around the time of birth and elaborate their arbors over the following 2--3 wk \[[@pbio-0030272-b13],[@pbio-0030272-b35]--[@pbio-0030272-b39]\]. We imaged axons originating from thalamic neurons that project to barrel cortex. We focused on secondary (diffuse) thalamic projections \[[@pbio-0030272-b40]--[@pbio-0030272-b43]\] originating from the Pom (the medial part of the posterior nucleus of the thalamus), which target layer 5A and supragranular layers, including layer 1 \[[@pbio-0030272-b44],[@pbio-0030272-b45]\]. In contrast, primary thalamic projections from the ventral posteromedial (VPm) subdivision of the ventrobasal nucleus largely target layers 4 and 5B of sensory cortex, and do not project to layer 1 \[[@pbio-0030272-b13]\]. Individual Pom axons project more widely than VPm axons and target complementary territories. Pom axons span multiple barrel columns in primary somatosensory cortex and often send branches to higher-order sensory areas and motor areas \[[@pbio-0030272-b44],[@pbio-0030272-b45]\]. It has been suggested that diffuse thalamic projections mediate sensory-motor integration and thalamocortical synchrony \[[@pbio-0030272-b43]\]. CR neurons are born relatively early during embryonic life (embryonic day 10--14 \[E10--E14\]) \[[@pbio-0030272-b46]\], but their axons are still growing during the first 2 wk of postnatal life, as suggested by the presence of growth cones at their tips \[[@pbio-0030272-b47],[@pbio-0030272-b48]\]. The cell bodies, dendrites and axons of CR neurons are entirely contained within layer 1. Thus, CR cells are the earliest cortical interneurons \[[@pbio-0030272-b49]\]. Because of the strategic location in layer 1, it has been suggested that CR neurons are important for neuronal migration and the generation of the inside-out cortical layering \[[@pbio-0030272-b46]\]. Indeed, CR neurons secrete the protein Reelin, which is critical for the development of cortical lamination \[[@pbio-0030272-b50],[@pbio-0030272-b51]\]. CR axons synapse principally on the growing apical dendritic arbors of pyramidal cells \[[@pbio-0030272-b46]\]. Just like cortical interneurons in the adult brain, CR neurons function as active elements in an early cortical network by integrating synaptic activity of developing pyramidal neurons \[[@pbio-0030272-b46],[@pbio-0030272-b48]\]. We found that TC and CR axons concurrently exhibit pronounced growth and pruning, with the balance tipping slightly in favor of growth. TC and CR axons exhibit different modes of elaboration. Structural dynamics were much more pronounced for TC axons than for CR axons. TC axon tips lacked distinct growth cones and grew rapidly in straight paths. In contrast, CR axons had large growth cones and grew slowly along tortuous paths. Branching increased dramatically for TC axons as a function of development. After an early period of branching, CR axons elaborated by growth without branching. Results {#s2} ======= Here we present our studies on the development of TC and CR axonal arbors in layer 1. We begin by describing long-term in vivo imaging experiments focusing on TC and CR axons. We then analyze the structures and dynamic processes underlying axonal arbor development, including growth cones, axonal trajectories, axonal growth, and pruning. Finally, we investigate how branches are added in TC and CR axons over the first 2 wk of life. Together these observations reveal distinct modes of axonal development for TC and CR axons. Imaging Thalamocortical and Cajal-Retzius Axons In Vivo {#s2a} ------------------------------------------------------- We identified a transgenic line (line L21) of mice expressing membrane-targeted GFP in two types of cortical axons \[[@pbio-0030272-b34]\]. These mice express GFP in a small subset of TC projection neurons from VPm and Pom, both of which project to barrel cortex, as well as in some Cajal-Retzius (CR) cells, the local intracortical interneurons of layer 1 ([Figure S1](#sg001){ref-type="supplementary-material"}). Pom projections have collaterals in layer 1 \[[@pbio-0030272-b44],[@pbio-0030272-b45]\]. A few other neurons expressed GFP during the first 2 wk of postnatal development (e.g., hippocampal pyramidal neurons), but they did not project to somatosensory cortex. Thus, cortical axons within layer 1 presumably belong to either CR cells or thalamocortical projection neurons. GFP expression levels were sufficient for high-contrast in vivo two-photon imaging of individual axons in superficial cortical layers as early as postnatal day 3 (P3), and increased further thereafter ([Figure 1](#pbio-0030272-g001){ref-type="fig"}). ::: {#pbio-0030272-g001 .fig} Figure 1 ::: {.caption} ###### Long-Term In Vivo Two-Photon Imaging of TC and CR Axons in Layer 1 (A and B) Time-lapse images of two different TC axons. (A) The left photomicrographs (corresponding to the orange box in B) are projections of 19 sections; the right (corresponding to the blue box in B) are projections of of 11 sections. Scale bar = 50 μm. (B) Orange and blue boxes contain axons, in horizontal view, reconstructed with BREBIS (see [Figure S2](#sg002){ref-type="supplementary-material"}E; [Materials and Methods](#s4){ref-type="sec"}) after the last imaging session (P14). The black boxes in (B) correspond to the fields of view of the in vivo images in (A). \(C) Axons reconstructed with BREBIS (see [Figure S2](#sg002){ref-type="supplementary-material"}E; [Materials and Methods](#s4){ref-type="sec"}) in coronal view. The position of the pia is indicated in green. \(D) Overlay of the coronal view in (C) on a cartoon of a generic adult mouse brain. \(E) Time lapse images of a CR axon (left, 12 sections; right, eight sections). Scale bar = 100 μm. (F and G) Axonal arbor of a CR neuron (gray) imaged in vivo in (E). Views are horizontal (F) and coronal (G). The CR soma and dendrites are drawn in red. \(H) Image of the CR dendritic arbor shown in a projection of 14 sections. Note that there are two CR neurons in this image, including the horizontally oriented neuron we traced in red in (F) and (G). Scale bar = 50 μm. \(I) High-magnification view of CR dendritic spines (projection of five sections). Scale bar = 25 μm. ::: ![](pbio.0030272.g001) ::: Axons were identified as CR axons if they could be traced back to the cell body of a CR cell in layer 1 ([Figure 1](#pbio-0030272-g001){ref-type="fig"}E and [1](#pbio-0030272-g001){ref-type="fig"}F). These cells can be recognized on the basis of morphological criteria \[[@pbio-0030272-b48]\]. Their dendritic trees have spiny branches oriented vertically toward the pia ([Figure 1](#pbio-0030272-g001){ref-type="fig"}G--[1](#pbio-0030272-g001){ref-type="fig"}I). Furthermore, CR axons have peculiar branching patterns with sequential bifurcations, offset slightly (5--20 μm) along the dorsoventral axis, with daughter segments running in opposite directions, parallel to the pial surface and always within layer 1 ([Figure 1](#pbio-0030272-g001){ref-type="fig"}G). In contrast, TC somata were too deep to be imaged in vivo. Instead, the putative TC arbors in layer 1 converged to a single stem that originated deep in the cortex (more than 300 μm below the pia). A subset (n = 5) of presumed TC axons were further reconstructed in perfusion-fixed brains using brain reconstruction with en-bloc imaging and slicing (BREBIS) ([Figure S2](#sg002){ref-type="supplementary-material"}; see [Materials and Methods](#s4){ref-type="sec"}). Such reconstructions revealed that these TC axons coursed through the white matter, consistent with thalamic origin. In some examples (n = 3), it was possible to trace these axons all the way to the thalamus ([Figure 1](#pbio-0030272-g001){ref-type="fig"}A--[1](#pbio-0030272-g001){ref-type="fig"}D). The morphology of the reconstructed axons, with extensive branching in layers 1 and 5A, matched that of reconstructions of Pom thalamic axons \[[@pbio-0030272-b44]\]. Axonal Growth Cones {#s2b} ------------------- The complexity of axon growth cones can vary depending on the location of the growing axon \[[@pbio-0030272-b22],[@pbio-0030272-b23]\]. Given the identical terrain in which we examined the growth of CR and TC axons, we were surprised to observe that TC and CR axons differed in the structure of their growth cones. The growth cones of TC axons consisted of small swellings that often lacked distinct filopodia ([Figure 2](#pbio-0030272-g002){ref-type="fig"}A; [Video S1](#sv001){ref-type="supplementary-material"}). In contrast, during the first week of postnatal development, CR axon tips had large, complex growth cones studded with numerous long filopodia ([Figure 2](#pbio-0030272-g002){ref-type="fig"}B; [Video S2](#sv002){ref-type="supplementary-material"}). CR growth cones were four times larger than TC growth cones at P5--P6 (123.0 ± 14.3 μm^2^ versus 32.9 ± 3.2 μm^2^, p \< 0.01; [Figure 2](#pbio-0030272-g002){ref-type="fig"}C). Similarly, the number of filopodia per growth cone was higher in CR axons than in TC axons (4.45 ± 0.49 versus 0.97 ± 0.31, p \< 0.0001; [Figure 2](#pbio-0030272-g002){ref-type="fig"}D). Growth cones belonging to TC and CR axons were frequently observed in the same region, just a few micrometers apart from one another, suggesting that their structural differences were not imposed by their environment. These differences in growth cone morphology between TC and CR axons persisted throughout the first week of postnatal development. When a branch belonging to either a TC or a CR axon stopped growing (or when it retracted; see below) growth cones could no longer be found at its tip. Thus, the frequency of large growth cones on CR axons decreased with developmental age, and after P9 they were rarely seen. Large growth cones similar to those of CR axons were never encountered in TC axon arbors within superficial cortical layers ([Figure 2](#pbio-0030272-g002){ref-type="fig"}E). ::: {#pbio-0030272-g002 .fig} Figure 2 ::: {.caption} ###### Growth Cone Structure of TC and CR Axons \(A) Growth cones at three ages (P5, P6, and P7). Photomicrographs are projections of three or four sections. The numbers in the images correspond to the numbered branch tips in the reconstructions drawn below the images. Red dots mark tips that could not be imaged because their signal was absorbed by overlying blood vessels or bone. Scale bar = 100 μm. \(B) Growth cones of CR axons. Images are projections of five to nine sections. The cell body and dendritic tree of the CR neuron are drawn in red. Image numbering, red dots, and scale bars are as in (A). \(C) Growth cone surface area for TC and CR axons. For panels C and D we analyzed 38 TC axon tips from five mice and 62 CR axon tips from four mice at ages P5 and P6. Error bars represent SEM. \* p \< 0.01. \(D) Number of filopodia per growth cone for TC and CR axons. \* p \< 0.0001. \(E) Scatter plot of CR and TC growth cone according to their surface area and number of filopodia. ::: ![](pbio.0030272.g002) ::: Axonal Trajectories {#s2c} ------------------- We analyzed the tortuosity of TC and CR axons. TC axons grew in relatively straight paths ([Figure 3](#pbio-0030272-g003){ref-type="fig"}A). In contrast, CR growth cones followed tortuous paths ([Figure 3](#pbio-0030272-g003){ref-type="fig"}B). The large CR growth cones often transiently split into two or more distinct growth cones ([Figure S3](#sg003){ref-type="supplementary-material"}A). However, these bifurcations were not maintained and instead resulted in sharp turns of the axon trajectory. The differences in tortuosity between TC and CR axons were significant at all ages ([Figure 3](#pbio-0030272-g003){ref-type="fig"}C and [3](#pbio-0030272-g003){ref-type="fig"}D). The tortuosity index was greater for CR cells than TC axons at P8 (1.67 ± 0.2 versus 1.11 ± 0.02, p \< 0.001), and at P13 (1.48 ± 0.1 versus 1.2 ± 0.02, p \< 0.0001). The tortuosity of TC axons increased somewhat with age due to the late addition of short twisted branches. Note that even those late branches lacked large growth cones, implying that the difference between CR and TC axon growth cones was not due to the developmental (or chronological) age of the arbors. The disparity between TC and CR axonal trajectories suggests differences in the mechanisms of guidance for these types of neuron subtypes. In particular, the tortuous paths of CR axons suggest intimate interactions with local guideposts and specificity at the level of axon trajectory \[[@pbio-0030272-b52]\]. ::: {#pbio-0030272-g003 .fig} Figure 3 ::: {.caption} ###### CR and TC Axonal Trajectories \(A) Reconstruction of a TC axon segment over time (P5--P8). \(B) Reconstruction of a CR axon segment over time (P5--P8). \(C) The tortuosity was higher for CR axons than TC axons (P8, \* p \< 0.001, 85 segments; P13--P14, \* p \< 0.0001, 89 segments). Error bars represent SEM. \(D) The difference in tortuosity was independent of the length of the analyzed segments. ::: ![](pbio.0030272.g003) ::: Axonal Growth and Retraction {#s2d} ---------------------------- Does pruning of cortical axons follow a period of overgrowth \[[@pbio-0030272-b12],[@pbio-0030272-b53]\]? We investigated whether local pruning occurs in vivo within the axon\'s target region, and if so, whether it occurs concurrently with or subsequent to a period of growth ([Figure 4](#pbio-0030272-g004){ref-type="fig"}). We also asked whether rates of growth are different in CR versus TC axons, as might have been predicted by their dissimilar growth cone morphologies \[[@pbio-0030272-b21]--[@pbio-0030272-b23]\]. TC and CR axon tips were imaged 10--30 h over a range of developmental ages (P4--P19). Some tips grew over considerable distances between imaging sessions, while others retracted ([Figure 4](#pbio-0030272-g004){ref-type="fig"}A; see also [Figure S4](#sg004){ref-type="supplementary-material"}A and [S4](#sg004){ref-type="supplementary-material"}B). Individual axonal arbors simultaneously exhibited growth and retraction at different branch tips. There were no discernable morphological clues that helped predict whether a tip would grow or retract. TC axon tips grew or retracted at high rates, up to 35 μm/h. Length changes were less rapid for CR axon tips ([Figure 4](#pbio-0030272-g004){ref-type="fig"}C). The fraction of axon tips growing (65%) and retracting (35%) was similar for CR and TC, and across different ages (unpublished data). In the first week of postnatal development, rapid growth or retraction occurred at most branch tips, but after P12, axon tips grew or pulled back only over short distances ([Figure 4](#pbio-0030272-g004){ref-type="fig"}D; see also [Figure S4](#sg004){ref-type="supplementary-material"}C). On average, growth was almost balanced by retraction, although the net balance tipped in favor of growth until the third week of postnatal development ([Figure 4](#pbio-0030272-g004){ref-type="fig"}E). The overall net growth of axon tips was higher for TC axons than CR axons at every time point examined. ::: {#pbio-0030272-g004 .fig} Figure 4 ::: {.caption} ###### Axon Growth and Pruning (A and A′) Representative axon tips imaged at two time points 19 h apart (images are projections of 12--20 sections). Some TC axon tips grew (green arrows), while others retracted (red arrows), sometimes only to grow again in a slightly different direction. Pruning of some TC axons involved fragmentation of the local arbor (red arrowheads in A′). CR axons could also be seen in the same regions of interest (blue inset). (B and B′) Higher-magnification view of TC axonal degeneration (projections of 11 and 18 sections). For earlier stages of degeneration see [Figure S4](#sg004){ref-type="supplementary-material"}B. (C and C′) Higher-magnification view of the CR axon growth cones. \(D) Plot of growth and retraction at individual axon tips throughout postnatal development. Each line represents a single axon tip (1,036 TC axon tips and 292 CR axon tips). Imaging sessions were ∼10--30 h apart and the length changes are given as rates. The rates therefore represent a lower estimate of the actual movement dynamics. \(E) Average length change at individual axon tips, including growth and retraction. The average growth at TC axon tips was always greater than at CR axon tips. See also [Figure S4](#sg004){ref-type="supplementary-material"}C. Error bars represent SEM. \* p \< 0.05. ::: ![](pbio.0030272.g004) ::: Two Types of Pruning: Axonal Degeneration as Part of Normal Development {#s2e} ----------------------------------------------------------------------- Pruning is one of the cardinal features of axon arbor development. We observed two types of axonal pruning. In addition to branch tip retractions, pruning also occurred by degeneration of large portions of axonal arbors ([Figures 4](#pbio-0030272-g004){ref-type="fig"} and [5](#pbio-0030272-g005){ref-type="fig"}). Degeneration began with beading and swelling of the axon shaft, followed by detachment of branches from their parent shafts ([Figure S4](#sg004){ref-type="supplementary-material"}B). The degenerated axon shafts then disintegrated into small fragments ([Figure 5](#pbio-0030272-g005){ref-type="fig"}C), the entire process taking less than 12 h. The debris then disappeared within 24--48 h, presumably after being cleared by glia. This differed from retraction in that retracting axons did not leave fluorescent debris behind ([Figure S4](#sg004){ref-type="supplementary-material"}D). It was not possible to predict degeneration based on axonal morphology. Axonal degeneration often spanned all the imaged axon arbor of one neuron ([Figure 5](#pbio-0030272-g005){ref-type="fig"}). However, degeneration was typically restricted to a single axon in the field of view, as opposed to retraction, which affected branch tips of multiple neighboring axons. This suggests that axonal degeneration is intrinsic to a given axon and not a response of a population of axons to an extrinsic signal. To rule out the possibility that axonal degeneration was a pathologic phenomenon related to the surgery or in vivo imaging, we also imaged brains of perfusion-fixed, naive mice. In these control brains, axonal degeneration was also detected at the same young ages at a similar frequency ([Figure 5](#pbio-0030272-g005){ref-type="fig"}E), implying that degeneration is a normal developmental process. ::: {#pbio-0030272-g005 .fig} Figure 5 ::: {.caption} ###### Large Portions of TC Axon Arbors Are Pruned by Axonal Degeneration \(A) Horizontal views (left) of tracings of a TC axon imaged in vivo from P5 to P7. The coronal view of the P7 trace (right) shows that branches with layers 1 and 2/3 converged upon a single parent axon stem arising from deep in cortex. The pia is drawn in green. \(B) In vivo image of the same axon as in (A) at P7 (projections of 10--22 sections). A neighboring axon (yellow arrows) can also be seen. \(C) Same axons as in (A) 27 h later. The large axon arbor has degenerated, and fragmented debris (red arrow heads) is found in its place. \(D) Quantification of length lost through branch tip retraction (blue) and axon degeneration (red) for n = 92 pruning events. When adjusted for a 24-h interval, the average length lost per pruning event at P5--P9 was 782.3 ± 134.7 μm for degenerating arbors versus 80.6 ± 10.1 μm for retracting segments (p \< 0.0001). \(E) Axon degeneration in a control, in vivo imaging-naïve brain at P4 (projection of 38 sections). The animal was perfused transcardially with cold 4% paraformaldehyde and 10% sucrose in PBS, in order to avoid rosary-type beading that is commonly seen in perfusion-fixed preparations of rodent brains at these young ages. Two intact axons can be seen coursing through the neuropil, one of which has a small growth cone (yellow arrow) typical of TC axons. A third axon is seen undergoing degeneration (red arrowheads), the fragmented debris appearing indistinguishable from the fragmented axons encountered in vivo (compare to [Figures 4](#pbio-0030272-g004){ref-type="fig"}A′, 5B, and S4B). ::: ![](pbio.0030272.g005) ::: Axon degeneration was relatively infrequent (affecting less than 5% of GFP-positive axons). Degeneration was observed only at younger ages (P4--P9), and in TC axons but not in CR axons. Axon degeneration typically involved long stretches of TC axon with multiple branches. Sometimes an entire TC arbor that had grown over several days in layer 1 suddenly degenerated all the way down to deep cortical layers as far as we could image ([Figure 5](#pbio-0030272-g005){ref-type="fig"}A). Axon segments that underwent degeneration were always much larger than those undergoing retraction. The average size was 782.3 ± 134.7 μm for degenerating arbors versus 80.6 ± 10.1 μm for retracting segments at P5--P9 ([Figure 5](#pbio-0030272-g005){ref-type="fig"}D; p \< 0.0001). Since in many cases the entire axon in the field of view was fragmented, our estimates of the average length lost through axonal degeneration represent a lower bound. Therefore, degeneration constitutes a mechanism of axon elimination that is distinct from retraction and operates over longer lengths of axon. Axonal Branching {#s2f} ---------------- Branch formation has been proposed to occur via one of three distinct mechanisms: Splitting of growth cones, delayed growth cone branching, and interstitial (growth cone-independent) branching \[[@pbio-0030272-b18],[@pbio-0030272-b22],[@pbio-0030272-b24]\]. How do neocortical axons branch in vivo? We observed interstitial branch formation in TC and CR axons ([Figure 6](#pbio-0030272-g006){ref-type="fig"}). In TC axons, small growth cones appeared de novo from the main shaft (see [Figure S3](#sg003){ref-type="supplementary-material"}B), and then elongated orthogonal to the parent axon. In the case of CR axons, the new growth cones also emerged from the parent shaft (unpublished data), and then quickly enlarged to assume their typical large configuration. CR axon growth cones (but not TC axons) often split into two or more segments, but never produced two permanent daughter branches (see [Figure S3](#sg003){ref-type="supplementary-material"}A). These observations imply that, during axon elaboration within the target region, interstitial branching is the dominant form of axonal branching in the neocortex. ::: {#pbio-0030272-g006 .fig} Figure 6 ::: {.caption} ###### Dynamics of Branch Additions and Retractions \(A) Reconstruction of a representative TC axon imaged from P8 to P17. Insets show photomicrographs corresponding to regions of interest indicated by red boxes on drawings to the right. See also [Figure S3](#sg003){ref-type="supplementary-material"}B. Images are projections of 8--19 sections. Red dots mark tips that could not be imaged. Note that the tracing corresponding to the P9 image is not shown. \(B) Reconstruction of a representative CR axon imaged between P5 and P13. Image insets are projections of 13--27 sections. The CR soma and dendritic tree are drawn in green. ::: ![](pbio.0030272.g006) ::: However, the placement of new branches within the developing arbor differed between TC and CR axons. Interstitial branching in TC axons occurred essentially uniformly along the principal axon segments ([Figures 6](#pbio-0030272-g006){ref-type="fig"}A and [7](#pbio-0030272-g007){ref-type="fig"}A). In contrast, new branches in CR axons appeared closely behind the leading growth cone ([Figures 6](#pbio-0030272-g006){ref-type="fig"}B and [7](#pbio-0030272-g007){ref-type="fig"}B). Out of 69 new branches added to TC axons, only 11 (16%) appeared within 100 μm of the tip, compared to 19 (56%) out of 34 new branches added to CR axons in that location (p \< 0.005; [Figure 7](#pbio-0030272-g007){ref-type="fig"}C). ::: {#pbio-0030272-g007 .fig} Figure 7 ::: {.caption} ###### Development of Axonal Complexity \(A) Tracings of a representative TC axon from P8 to P14. \(B) Tracings of a representative CR axon from P5 to P10. \(C) Location of branch additions with respect to tip of parent axon for 69 new TC branches and 34 new CR branches. The difference between TC and CR axons was significant (p \< 0.05). \(D) The rate of branch additions per millimeter of axon per day was greater in TC axons (n = 30) than in CR axons (n = 12). Error bars represent SEM. \* The difference was significant at P8--P11 (p \< 0.05). \(E) Density of branch points versus postnatal age. The branch point density gradually increases for TC axons (n = 219) between P5--P6 and P14--P19 (p \< 0.0001), whereas it decreases slightly for CR axons (n = 76) during the same time period (p = 0.4). The overall difference in these trends between TC and CR axons was significant (p \< 0.05). Error bars represent SEM. \* The density of branch points for TC axons, which was lower than that of CR axons at P5--P6 (p = 0.064) and at P7--P8 (p = 0.05), was later significantly higher at P11--P13 (p \< 0.01) and at P14--P19 (p \< 0.001). ::: ![](pbio.0030272.g007) ::: Our ability to image the same axons over periods of up to 2 wk allowed us to determine the developmental progression by which axonal arbors elaborate. These long-term imaging studies revealed profound differences between TC and CR axons. During the first week of postnatal development, many TC axons were initially composed of very few (1--3) long branches that extended in straight paths for hundreds of microns in layer 1, without secondary branches ([Figure 7](#pbio-0030272-g007){ref-type="fig"}A; see also [Figures 2](#pbio-0030272-g002){ref-type="fig"}A and [6](#pbio-0030272-g006){ref-type="fig"}A). Over the next few days, short branches appeared all along these parent axons at increasing rates (from 2.2 ± 0.5 branches per millimeter of axon per day at P5--P7 to 2.8 ± 0.4 at P8--P11; p = 0.09; [Figure 7](#pbio-0030272-g007){ref-type="fig"}D). Many of these early branches were unstable with short lifetimes, but during the second week of postnatal development, new branches became stabilized. Thus, between P5--P6 and P14--P19 the branch density (number of branch points per millimeter of axon) increased by a factor of three ([Figure 7](#pbio-0030272-g007){ref-type="fig"}E; from 2.2 ± 0.4 to 6.0 ± 0.4 branch points per millimeter of axon; p \< 0.0001). At the time of the first imaging session, around P5, CR axons already had numerous (more than ten) branches close to the soma, and most branch tips were still growing ([Figure 7](#pbio-0030272-g007){ref-type="fig"}B; see also [Figures 2](#pbio-0030272-g002){ref-type="fig"}B and [6](#pbio-0030272-g006){ref-type="fig"}B). During the first postnatal week, branches were added at a low rate (1.3 branches per millimeter of axon per day), which continued to decrease with development ([Figure 7](#pbio-0030272-g007){ref-type="fig"}D). The rate of branch addition for CR axons across all ages was significantly lower than for TC axons (p \< 0.05). Because the main form of growth for CR axons was extension at axon tips without branching, the branch density decreased between P5--P6 and P14--P19 from 3.6 ± 0.6 branch points per millimeter at P5--P6 to 2.7 ± 0.6 at P14--P19 (p = 0.4). The difference in branch density trends during postnatal development between TC and CR axons was significant (p \< 0.05). During the third week of postnatal life, TC axons changed little, whereas CR branch tips retracted slightly. CR neurons and their axons became dimmer, presumably because of decreased GFP expression, and eventually disappeared. Discussion {#s3} ========== The development of axonal arbors is a critical step in the establishment of precise neural circuits. To discover the modes of axonal elaboration, we imaged axons of TC projection neurons and CR interneurons from early postnatal days until their arbors were mature, in the third week of postnatal development. These axons were intermingled within layer 1 and therefore experienced identical extracellular influences. We found that the elaboration of both CR and TC axons relies on a fine balance of growth and pruning. TC and CR axons have distinct structures and elaborate with different dynamics, suggesting that different neuronal subtypes use different intrinsic developmental programs to innervate their target cells. In addition, our data are relevant to controversies regarding the modes of axon pruning and branching. Axonal Trajectory, Growth, and Pruning {#s3a} -------------------------------------- Within cortical layer 1, TC axons have small growth cones and follow essentially straight paths, while CR axons have large growth cones that advance in tortuous paths. Elaborate growth cones are thought to be necessary for contact-mediated guidance \[[@pbio-0030272-b22]\]. Since tortuous axonal paths may imply interactions between axon growth cones and local guidance cues \[[@pbio-0030272-b52]\], our findings imply that TC axons might explore the local environment with little guidance from local cues, whereas CR axons are strongly shaped by such signals. Axon growth and pruning are cardinal features of nervous system development. Based on observations from fixed mammalian axons \[[@pbio-0030272-b04]--[@pbio-0030272-b07]\], it is often assumed that early overgrowth is followed by pruning \[[@pbio-0030272-b54]\]. But concurrent growth and pruning have also been described, especially in lower vertebrates such as Xenopus tadpoles \[[@pbio-0030272-b10],[@pbio-0030272-b14],[@pbio-0030272-b15]\]. We studied growth and pruning within the target area of mammalian cortical axons in quantitative detail and find that branch tip growth and retraction events occurred simultaneously. Growth and pruning were almost balanced, with a small preference toward growth (see [Figure 4](#pbio-0030272-g004){ref-type="fig"}E). As a result, the total size of the axonal arbor increased monotonically with age until the third postnatal week. We observed two distinct forms of axonal pruning: branch tip retraction and degeneration. Branch tip retraction occurred in both TC and CR axons. Relatively small branch segments (tens of micrometers) were retracted, leaving behind no detectable debris (see [Figure 4](#pbio-0030272-g004){ref-type="fig"}). Axonal degeneration was seen only in TC axonal arbors and resulted in the elimination of large portions (hundreds of micrometers) of the axonal arbor (see [Figure 5](#pbio-0030272-g005){ref-type="fig"}). Degeneration did not proceed sequentially in one direction (e.g., distal-to-proximal) but instead affected the entire axon synchronously. This is similar to what happens during Wallerian degeneration in transected nerves, where large-scale fragmentation of the entire disconnected axon (several hundred microns in length) occurs suddenly, within a few minutes \[[@pbio-0030272-b30]\]. The process of degeneration in TC axons occurred through beading and then fragmentation, which is also reminiscent of Wallerian degeneration. Similar forms of pruning have been observed in the Drosophila mushroom body \[[@pbio-0030272-b28]\] and in cultured hippocampal axons \[[@pbio-0030272-b32]\]. In our experiments, axon degeneration was not likely a pathologic phenomenon related to imaging. First, degenerating axons were seen only during a specific time period (P4--P9), suggesting that it is a developmentally regulated process. Second, axonal degeneration occurred at various times after the window surgery (up to 3 d later). Third, only a minority (less than 5%) of axons were affected by this type of pruning. Finally, fragmented axons were also identified in control, unoperated, perfusion-fixed mice at equivalent ages (see [Figure 5](#pbio-0030272-g005){ref-type="fig"}E). Degeneration of TC axons is also unlikely to be due to apoptosis of neurons in thalamus. The wave of naturally occurring cell death in the thalamus occurs mostly before birth in the mouse \[[@pbio-0030272-b55]--[@pbio-0030272-b57]\]. Furthermore, inhibition of caspase activity does not prevent Wallerian degeneration or pruning of Drosophila mushroom body axons \[[@pbio-0030272-b28],[@pbio-0030272-b58]\]. Both axon degeneration and branch tip retraction in cortical axons differed from axosome shedding at the neuromuscular junction \[[@pbio-0030272-b59]\]. Axosome shedding involves the loss of membranous organelles as axons slowly retract. In contrast, axonal degeneration occurred essentially synchronously along long segments of axons. Axon retraction did not involve retraction bulbs (seen in axosome shedding), and fragmented debris was not observed, even when we imaged at frequent intervals ([Figure S4](#sg004){ref-type="supplementary-material"}D). Axosome shedding may, therefore, represent a distinct third mechanism of pruning in the peripheral nervous system. During development, axonal retraction and degeneration probably play different roles in wiring the neocortex. Retraction is likely involved in the local refinement within a target area, while degeneration operates at the level of projections. It will be interesting to determine if axonal degeneration in developing mouse TC axons and Drosophila mushroom body neurons are mechanistically related. Furthermore, it will also be important to explore the mechanisms that govern maintenance versus retraction of branches in TC and CR axons, including the roles of neuronal activity and synaptogenesis. Axonal Branching {#s3b} ---------------- Experiments in the spinal cord and neocortex have demonstrated that many axons can branch at interstitial sites, anywhere along the parent axon shaft \[[@pbio-0030272-b27],[@pbio-0030272-b60]\]. However, some controversy remains regarding the role of the growth cones in branching \[[@pbio-0030272-b18]\], as some axons seem to exploit the splitting of growth cones for branching \[[@pbio-0030272-b25]\]. Growth cones of CR axons often split transiently into two branches, but one of these branches was subsequently retracted. Splitting growth cones therefore never gave rise to two bona fide daughter branches. Rather, time lapse imaging of individual CR growth cones suggests that growth cone splitting may be involved in guidance/turning behaviors (see [Figure S3](#sg003){ref-type="supplementary-material"}A). Our data show that the vast majority of branches are formed at interstitial sites (see [Figure 6](#pbio-0030272-g006){ref-type="fig"}), inconsistent with a direct role for growth cone splitting in branch formation. Axon growth cones have also been implicated in interstitial branching \[[@pbio-0030272-b24]\]. Two types of interstitial branching have been described. Experiments in dissociated cultures of cortical neurons have shown that growth cones can leave behind sites of unstable cytoskeleton, which later become branch points through a process known as delayed interstitial branching \[[@pbio-0030272-b26]\]. In other cases, interstitial branches appear days after the main axon stopped growing \[[@pbio-0030272-b24]\]. We found that TC axons add new branches all along their arbor, inconsistent with delayed interstitial branching. In contrast, CR axons preferentially add branches close to the growing tips, suggesting that a process similar to delayed interstitial branching may occur in CR neurons in vivo. However, we did not observe filopodia or lamellar remnants being left behind by the advancing growth cone, as described previously for delayed interstitial branching \[[@pbio-0030272-b26]\]. Distinct Modes of Elaboration for TC and CR Axons {#s3c} ------------------------------------------------- The function of cortical circuits in the mature brain relies on a precise circuit diagram. The physical organization of axonal and dendritic arbors is the structural basis of functional circuits. The elaboration of axons involves both intrinsic programs and extrinsic factors, including guidance cues and neural activity \[[@pbio-0030272-b10],[@pbio-0030272-b11],[@pbio-0030272-b61]\]. Because axons of different cell types are molecularly distinct, they are likely to respond differently to such external cues and grow in different ways. Time-lapse imaging revealed that TC and CR axons develop with distinct strategies over the first 2 wk of development ([Figure S5](#sg005){ref-type="supplementary-material"}; [Videos S3](#sv003){ref-type="supplementary-material"} and [S4](#sv004){ref-type="supplementary-material"}). TC axons grow and retract rapidly, following straight paths. With developmental age they become more complex, adding numerous branches to their arbors. CR axons grow and retract more slowly. Their elaboration in the postnatal brain is primarily through elongation of tortuous branch tips, with negligible interstitial branching. Some of these differences in behavior between CR and TC axons were due to differences in their developmental time-course. For example, much of CR branching had occurred before our first imaging session, while most of the TC branching within layer 1 was yet to occur. Yet, CR and TC neurons are born at approximately the same embryonic age \[[@pbio-0030272-b38],[@pbio-0030272-b46]\], and, at the ages we studied, CR axons still had many large growth cones and grew long distances over the imaging period. For these reasons, the differences in growth cone structure, axonal trajectory, and growth rates between these two cell types cannot be attributed to differences in developmental maturity between CR and TC axons. Axonal structure is cell-type dependent. For example, cortical pyramidal cell axons have straight paths, while cortical inhibitory interneuron axons have tortuous paths \[[@pbio-0030272-b62],[@pbio-0030272-b63]\]. Our data show that this distinction holds for TC projection neuron and CR interneuron axons. Computational analysis has further shown that projection neuron axons make as many "contacts" with synaptically unconnected dendrites as with connected dendrites \[[@pbio-0030272-b52],[@pbio-0030272-b64]\]. In contrast, interneurons make more contacts with their postsynaptic partners compared to other neurons in the vicinity. The different modes of elaboration we observed for TC and CR axons are consistent with this view. TC axons elaborate with large-scale growth and retraction, in straight paths, suggesting that these axons may be stabilized by activity-dependent selection \[[@pbio-0030272-b10]\]. In contrast, CR axons elaborate more gradually in tortuous paths, suggesting that these axons are strongly influenced by extracellular guidance cues. Axon Elaboration and Critical Periods {#s3d} ------------------------------------- Both the addition of new branches ([Figure 7](#pbio-0030272-g007){ref-type="fig"}D) and the degree of growth and retraction at individual tips of TC axons (see [Figure 4](#pbio-0030272-g005){ref-type="fig"}D) diminished after the second postnatal week, suggesting that arbors reach a mature state around that time. Interestingly, the end of the second postnatal week coincides with the end of the critical period for barrel cortex. During this critical period, layer 2/3 sensory maps \[[@pbio-0030272-b65],[@pbio-0030272-b66]\] and circuits \[[@pbio-0030272-b67]\] display rapid developmental and experience-based plasticity. Moreover, the maturation of dendritic branching \[[@pbio-0030272-b68]\] and spine motility \[[@pbio-0030272-b65]\] of layer 2/3 pyramidal neurons are also affected by sensory deprivation. It is likely that local axonal growth and pruning contribute to experience-dependent development of cortical circuits. Indeed, activity-dependent rearrangements of thalamocortical axon collaterals have been demonstrated in vitro \[[@pbio-0030272-b11],[@pbio-0030272-b69]\]. It remains to be seen whether manipulations of sensory experience, for example by whisker trimming, similarly affect axonal development. Materials and Methods {#s4} ===================== Animals and surgery {#s4a} ------------------- All experimental protocols were conducted according to the National Institutes of Health guidelines for animal research and were approved by the Institutional Animal Care and Use Committee at Cold Spring Harbor Laboratory. We used male and female c57/Bl6 transgenic mice in which a thy1 promoter drove the expression of GFP that was targeted to the cell membrane via a palmytoylated site (mGFP, line L21 \[[@pbio-0030272-b34]\]). Throughout the first 2 wk of postnatal development, these mice express GFP in Cajal-Retzius neurons, thalamic neurons in the Pom and VPm nuclei, and in CA1 pyramidal neurons in hippocampus. Importantly, cortical pyramidal neurons begin to express GFP only toward the end of the second postnatal week (with the exception of a small number of pyramidal neurons in the cingulate cortex). At P3--P12, mice were deeply anesthetized with an isoflurane-oxygen mixture delivered by an anesthesia regulator (SurgiVet, Waukesha, Wisconsin, United States) or with an intraperitoneal injection of a mixture of ketamine (0.13 mg/g body weight) and xylazine (0.01 mg/g). Following a midline scalp incision, the skull overlying the right somatosensory and visual cortices was carefully removed, leaving the dura intact. The dura was covered by a thin layer of 1.2% low-melting-point agarose (Type IIIA, Sigma, St. Louis, Missouri, United States) dissolved in Hepes-buffered artificial cerebrospinal fluid, and then a custom-made round cover glass (No. 1) was quickly laid over the agarose. This "window" (see [Figure S2](#sg002){ref-type="supplementary-material"}A) was sealed in place with dental cement. A titanium bar was also attached to the skull to later secure the mouse to the stage of the microscope. Between imaging sessions mice were housed with littermates and their mothers. The surgery itself seemed innocuous to the mice, as they recovered nicely from the surgeries and anesthesia, gained weight, and behaved indistinguishably from their unoperated littermates. Imaging {#s4b} ------- Imaging began after a recovery period of at least 1 h. Animals were anesthetized with an isoflurane-oxygen mixture. The animals woke up within a few minutes after stopping the flow of anesthetic. In vivo images of GFP-expressing neurons were acquired with a custom-built two-photon microscope \[[@pbio-0030272-b65]\], using a Ti:sapphire laser (Tsunami, Spectra Physics, Mountain View, California, United States), running at approximately 910 nm, pumped by a 10 W solid state laser (Millenia X, Spectra Physics). The objective (40×, 0.8 NA) and scan lens were from Zeiss (Oberkochen, Germany), the trinoc from Olympus (Tokyo, Japan), and the photomultiplier tube from Hamamatsu (Hamamatsu City, Japan). Detection optics with large apertures provided optimal fluorescence detection \[[@pbio-0030272-b70]\]. Image acquisition was achieved with ScanImage software custom-written in MatLab (MathWorks, Natick, Massachusetts, United States) \[[@pbio-0030272-b71]\]. Slightly overlapping image stacks (up to 40 per animal) were taken at low zoom (approximately 210 × 230 μm field of view; see [Figure S2](#sg002){ref-type="supplementary-material"}D). For high-magnification imaging of axon growth cone dynamics smaller ROIs (approximately 50 × 50 μm) were selected at random or centered on axon tips. All image stacks consisted of optical sections with 512 × 512 pixels separated by 1- to 3-μm steps. By following axons over long distances, we could distinguish between axons belonging to CR neurons (identified by their location within layer 1 and their characteristic dendritic trees; see [Figure 1](#pbio-0030272-g001){ref-type="fig"}G and [1](#pbio-0030272-g001){ref-type="fig"}H) and TC axons that originated from below layer 2/3 (\> 300 μm). Care was taken to achieve close to identical fluorescence levels across imaged regions and imaging sessions by adjusting the laser intensity at deeper focal planes or by compensating for increased expression levels of GFP or darkening of a particular ROI due to growing blood vessels or skull bone. Imaged axons were within 450 μm from the surface of the brain. All images in the Figures are projections of three-dimensional stacks. Labeling in the line 21 mGFP mice was very sparse (see [Figure S2](#sg002){ref-type="supplementary-material"}D), especially in mice less than 1 wk of age; this facilitated the task of tracing individual axons. In some time-lapse experiments, dendrites and axons expressing GFP accumulated at older ages. Therefore, in some Figures, particularly those consisting of projections of more than ten sections, distracting fluorescent processes were digitally removed for presentation. BREBIS {#s4c} ------ Retrospective reconstructions of selected axonal arbors were used to confirm their identity as TC axons. After the last imaging sessions, mice were perfused transcardially with 4% paraformaldehyde, and their brains were carefully removed and blocked to remove the cerebellum and left hemisphere. Brains were post-fixed overnight in the same fixative and then immersed in low-melting-point agarose and mounted on the well of a vibratome slicer (Leica) that was also fitted to be mounted on the stage of the two-photon microscope. Previously imaged axons were once again identified using blood vessel imprints on the dura (see [Figure S2](#sg002){ref-type="supplementary-material"}C), or by using injections of fluorescent beads as fiducial points. These axons were imaged in the same orientation as during in vivo imaging and as deep as possible (often up to 800 μm in a single set of image stacks). Next, the most superficial 300--600 μm of brain were sliced off and the brain was re-imaged, always keeping track of the location of the imaged axon. Following several cycles of serial imaging and slicing (see [Figure S2](#sg002){ref-type="supplementary-material"}E), the axon could be traced from layer 1 all the way down to the thalamus (see [Figure 1](#pbio-0030272-g001){ref-type="fig"}). Naïve animals were also processed for BREBIS and analyzed as controls (n = 3). These axons were imaged under identical conditions to in vivo imaging (optics, magnification, orientation) in whole mounts of the fixed brains. We randomly selected axon segments of similar TC axons imaged in vivo and confirmed with BREBIS that they indeed originated from the white matter or from thalamus. Since it was not possible to reconstruct every axon that was imaged, it is possible that some putative TC axons originated outside the thalamus, for example in the brainstem. However, such brainstem projections to neocortex, including monoaminergic axons, innervate cortex in a diffuse pattern that spans many layers, and are inconsistent with those we observed for putative TC axons \[[@pbio-0030272-b72]\]. Tracing of axons and analysis {#s4d} ----------------------------- The data in this study come from 22 mice imaged in vivo. Nine mice were imaged over multiple days, and 13 mice were imaged over two time-points at least one day apart. In addition, six control mice were imaged only after fixation. The data analyzed for this study were collected over a total of more than 100 separate imaging sessions. We analyzed a total of 253 mm of TC axons and 99 mm of CR axons. Tracing and analysis were performed independently by two investigators (CPC and RW). Axons were traced in three dimensions within tiled stacks using the confocal module of Neurolucida (Microbrightfield, Williston, Vermont, United States; see [Figure 2](#pbio-0030272-g002){ref-type="fig"}D). From these tracings we could easily derive data regarding axon segment lengths (growth and retraction over time), branch densities, and axon tortuosity. Axon extensions were considered to be true branches (as opposed to filopodia or elongated terminal boutons) if they measured at least 7.5 μm in length. We also imaged fixed brains that had never undergone surgery or in vivo imaging (n = 3 mice, nine axons, measuring a total of 22 mm) at P7--P8 and at P14--P15 to study the effects of surgery and/or chronic imaging on axonal development. The density of branch points was not different from that of experimental mice (at P7--P8, 2.1 ± 0.5 branch points per millimeter of axon in controls versus 1.9 ± 0.3 in experimental mice, p = 0.88; and at P14--P15, 5.4 ± 0.6 branch points per millimeter of axon in controls versus 5.9 ± 0.6 in experimental mice, p = 0.63; [Figure S6](#sg006){ref-type="supplementary-material"}), suggesting that there were no deleterious effects due to the surgery or chronic imaging. For the growth cone comparison (see [Figure 2](#pbio-0030272-g002){ref-type="fig"}), we analyzed 62 CR growth cones from four mice and 38 TC growth cones from five mice at ages P5 and P6. Growth cone size was determined by tracing a contour encompassing the base of the growth cone and the tips of all filopodia. For bulbous growth cones lacking filopodia, a contour of the bulbous swelling was measured. For tortuosity data (see [Figure 3](#pbio-0030272-g003){ref-type="fig"}), we analyzed 93 TC axon segments from six mice and 81 CR axon segments from six mice, and the average length of those segments was 328 μm and 332 μm (p \> 0.34), respectively. Only axons from mice that had been imaged both at P8 and at P13/P14 were included in the tortuosity analysis. Axon arbors were first broken down into the individual branches and then the longest segments were further broken down into smaller segments of approximately similar sizes such that all tortuosity data was were obtained from axon segments measuring more than 100 μm and less than 850 μm. The tortuosity index is the ratio of the true contour length of the axon segment divided by the end-to-end distance. The number of mice/axons/tips analyzed for short-term growth and retraction (see [Figure 4](#pbio-0030272-g004){ref-type="fig"}) was as follows: for TC axons: at P4--P7, 9/46/128; at P7--P9, 8/33/145; at P9--P11, 5/32/240; at P11--P14, 5/32/363; and at P14--P19, 3/21/160. For CR axons, values were: at P4--P7, 8/43/137; at P7--P9, 7/29/92; at P9--P11, 3/11/28; and at P11--P14, 4/10/35. For those data, in animals older than P10 we analyzed only axons that had been already followed for several days since P8 or younger. This was a precaution to avoid collecting data from axons belonging to cortical pyramidal neurons, which begin to express GFP toward the end of the second postnatal week. For the comparison between axon degeneration and retraction (see [Figure 5](#pbio-0030272-g005){ref-type="fig"}), 13 degenerating arbors and 79 retracting segments were analyzed. Branch density data were obtained only for axon segments larger than 100 μm. The number of mice/axons analyzed for branch density measurements were as follows: for TC axons: at P5--P7, 7/59; at P7--P8, 9/52; at P9--P10, 4/33; at P11--P13, 6/48; at P14--P19, 5/27. For CR axons, values were: at P5--P7, 4/29; at P7--P8, 4/29; at P9--P10, 2/7; at P11--P14, 3/11. Finally, for measurements of branch additions we analyzed 30 TC axons from eight mice and 12 CR axons from six mice. Significance was set at p \< 0.05 and evaluated using Student\'s t-test. Error bars represent the standard error of the mean (SEM). Supporting Information {#s5} ====================== Figure S1 ::: {.caption} ###### GFP Expression Pattern in L21 Transgenic Mice Immunocytochemistry was performed to detect GFP on coronal sections from line L21 brains at P7. \(A) Primary somatosensory cortex. Note that cortical pyramidal neurons do not express GFP at this early stage. Branching of TC axons in layer 4 delineates barrels (asterisks). Scale bar = 100 μm. \(B) High magnification of inset in (A) showing axons in layer 1 (arrows). Scale bar = 10 μm. \(C) Coronal view of a CR cell (arrowhead) and CR axon (arrow) in layer 1. Scale bar = 60 μm. \(D) Thalamus (Th). Insets 1 and 2 are shown at higher magnification in E and F, respectively. Scale bar = 220 μm. (E and F) Higher-magnification views of the thalamic neurons shown in (D). Cell bodies of bright relay neurons are indicated by arrow heads. Fi, fimbria of the hippocampus. Scale bars = 35 μm. \(G) Primary visual cortex and dorsal hippocampus. Insets are shown at higher magnification in H and I, respectively. Scale bar = 200 μm. (H and I) Higher-magnification views of TC axons (arrows) in layers 1 (H) and 2/3 (I) of visual cortex shown in (G). Scale bars = 60 μm (H) and 30 μm (I). (6.3 MB TIF). ::: ::: {.caption} ###### Click here for additional data file. ::: Figure S2 ::: {.caption} ###### Methods To image developing axons in vivo, we performed window surgeries on transgenic mice expressing membrane-bound GFP under the thy1 promoter. \(A) Photograph of a P3 mouse shortly after surgery. A titanium bar was used to immobilize the animal onto the microscope stage. \(B) Axons were imaged in superficial layers of cortex over several days and up to 2 wk. \(C) The unique pattern of blood vessels in each mouse was used to find the regions of interest from one imaging session to the next. \(D) Axons were then traced from overlapping stacks of images. \(E) In order to confirm the identity of thalamocortical axons, some animals were fixed by perfusion and the previously imaged axons were reconstructed using BREBIS (see [Figure 1](#pbio-0030272-g001){ref-type="fig"}; [Materials and Methods](#s4){ref-type="sec"}) (3.3 MB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Figure S3 ::: {.caption} ###### Tortuous Path of CR Axon Growth Cone and Interstitial Branching in TC Axon \(A) Time lapse sequence showing the tortuous path of a CR axon growth cone. Images are projections of 13--19 sections. \(B) Time lapse sequence of the birth of a TC axon branch, starting with the de novo appearance of an interstitial growth cone within the shaft of the TC axon. Time stamps in hours. (3.1 MB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Figure S4 ::: {.caption} ###### Examples of Growth, Branch Retraction, and Axonal Degeneration (A and B) Additional examples of growth (green arrows) and retraction (red arrows) of TC axon tips (see also [Figure 4](#pbio-0030272-g004){ref-type="fig"}). (A) Initial imaging session; (B) Same field of view imaged 10 h later. The fragmented axon shown here (red arrow heads) is at an earlier stage of degeneration than the one shown in [Figures 3](#pbio-0030272-g003){ref-type="fig"} and [4](#pbio-0030272-g004){ref-type="fig"}. \(C) Average length gained (growth) or lost (retraction only) for TC (black) and CR (gray) axon tips at various ages throughout postnatal development. The differences in growth between TC and CR axons were statistically significant at P4--P7 (p \< 0.0001), P9--P11 (p \< 0.05), and P11--P13 (p \< 0.05). The differences in retraction between TC and CR axons were statistically significant at P4--P7 (p \< 0.0001), P7--P9 (p \< 0.001), and P9--P11 (p \< 0.01). \(D) Time lapse images obtained at 1-h intervals of a cortical axon from a P4 mouse. Red arrows indicate branches that are retracted. Note that branch tip retraction is distinct from axosomal shedding in that it does not involve retraction bulbs and does not leave a trail of debris. Scale bar = 5 μm. Error bars represent the SEM. (2.0 MB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Figure S5 ::: {.caption} ###### Summary Cartoon of the Different Modes of Axon Elaboration Used by Local Intracortical Axons (CR) and Long-Range Extracortical Axons (TC) \(A) TC axons, which lack exuberant growth cones, grow fast during the first postnatal week. But this remarkable growth is accompanied by a surprisingly high amount of local pruning. At later stages, many branches are added to the original "bare" skeleton, and although growth slows down, so does the amount of branch retraction. \(B) CR axons branch extensively during the first week of postnatal development. Their large growth cones guide the axon tips in a slow, twisting path, with little pruning. Later, growth cones disappear and growth and branching slow down. Some branches are eventually lost. Thus, two modes of axon arbor elaboration exist in early postnatal cortex. On the one hand, local interneurons grow slowly in a meandering course, making few mistakes. On the other hand, long-range projection axons grow fast, leading to frequent errors and backtracking. Given the identical terrain upon which these axons grow, the differences must reflect intrinsic differences in the axons themselves, such as in their ability to sense gradients of signaling molecules or in their method of synapse formation. (265 KB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Figure S6 ::: {.caption} ###### Density of Branch Points in Imaged Mice Matches That in Control Mice To control for possible deleterious effects of chronic imaging, we examined perfusion-fixed, in vivo imaging-naïve mice (n = 3). The density of branch points in these control mice was not different than that of imaged mice at P8 (p = 0.88) and P14--P15 (p = 0.63). Time stamps in hours. Error bars represent the SEM. (194 KB PDF). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S1 ::: {.caption} ###### TC Growth Cone The time stamp is given in minutes. (1.1 MB AVI). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S2 ::: {.caption} ###### CR Growth Cone The time stamp is given in minutes. (1.9 MB AVI). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S3 ::: {.caption} ###### TC Axon Cartoon (2.3 MB AVI). ::: ::: {.caption} ###### Click here for additional data file. ::: Video S4 ::: {.caption} ###### CR Axon Cartoon (2.2 MB AVI). ::: ::: {.caption} ###### Click here for additional data file. ::: We are grateful to Rafael Yuste for support, Barry Burbach for expert technical assistance, and Anthony Holtmaat for help with experiments. We also thank Ed Ruthazer, Carol Mason, and members of the Svoboda and Yuste labs for helpful discussions and/or reading the manuscript. Funded by the John Merck Fund (to Rafael Yuste), the Howard Hughes Medical Institute, and the National Institutes of Health. Competing interests. The authors have declared that no competing interests exist. Author contributions. CPC and KS conceived and designed the experiments and wrote the manuscript. CPC and RMW performed the experiments and analyzed the data. VDP, PC, and KS contributed reagents/materials/analysis tools. ¤ Current address: Departments of Neurology and Neurobiology, Reed Neurological Research Center University of California, Los Angeles, California, United States of America Citation: Portera-Cailliau C, Weimer RM, De Paola V, Caroni P, Svoboda K (2005) Diverse modes of axon elaboration in the developing neocortex. PLoS Biol 3(8): e272. BREBIS : brain reconstruction with en-bloc imaging and slicing CR : Cajal-Retzius E\[number\] : embryonic day \[number\] GFP : green fluorescent protein P\[number\] : postnatal day \[number\] Pom : medial part of the posterior nucleus of the thalamus SEM : standard error of the mean TC : thalamocortical VPm : ventral posteromedial nucleus of the thalamus	PubMed Central	2024-06-05T03:55:59.590727	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180514/", "journal": "PLoS Biol. 2005 Aug 26; 3(8):e272", "authors": [ { "first": "Carlos", "last": "Portera-Cailliau" }, { "first": "Robby M", "last": "Weimer" }, { "first": "Vincenzo", "last": "De Paola" }, { "first": "Pico", "last": "Caroni" }, { "first": "Karel", "last": "Svoboda" } ] }
PMC1180515	On the outside, humans and other vertebrates seem to be bilaterally symmetrical. Draw a line down your body from the top of your head to your feet, and what is to the left of the line is pretty much the mirror image of what is to the right. But think about your internal organs. Your heart is on the left of your body, while your liver is over to the right. This left-right asymmetry, like the other asymmetries of vertebrate bodies, is established early in embryonic life. The symmetrical ball of cells formed after fertilization of the egg quickly develops a head and tail end (an anteroposterior axis) and a front and back (a dorsoventral axis). Once these identities have been established, the embryo then specifies its left and right sides---an event called left-right symmetry breaking. Work in early mouse embryos suggests that left-right symmetry breaking arises from the leftward flow of extra-embryonic fluid. This flow is somehow generated by beating cilia, rod-like unicellular organisms through water. During evolution, however, multicellular organisms retained cilia on cells that move extracellular fluid; for example, respiratory tract cells use cilia to flush away bacteria and other debris. And in the embryo, cilia on cells in a region called the node produce the symmetry-breaking leftward flow, or "nodal flow." As a result, in mouse embryos with mutant cilia that fail to beat, nodal flow is not established, and some mice develop with their internal organs on the wrong side. [](#pbio-0030291-g001){ref-type="fig"} ::: {#pbio-0030291-g001 .fig} ::: {.caption} ###### A scanning electron micrograph of mouse embryo node cilia helped researchers determine the cilia\'s tilt direction, a factor that contributes to embryo asymmetry ::: ![](pbio.0030291.g001) ::: But it\'s not clear how beating nodal cilia actually produce a leftward fluid flow. Unlike most cilia, which beat in a whip-like back-and-forth motion, nodal cilia beat by twirling in a circle; cilia twirling in fluid should act like a propeller and create a vortex (a circular flow of water). But nodal cilia don\'t produce a vortex; they produce a leftward flow. Models of fluid dynamics suggest that the nodal cilia might be able to do this if they are tilted. And as Hiroshi Hamada and colleagues now report, this is indeed the case; nodal cell cilia are tilted toward the embryo\'s tail end, rather than sticking straight up out of the plane of the node. To investigate whether the cilia in mouse embryonic nodes are tilted, the researchers first traced their trajectories using a high-speed camera. Tracing the path of the tip of a straight-up cilium as it beats should yield a circular trace, but the researchers actually recorded elliptical and D-shaped traces. The authors explain that elliptical traces are representative of beating cilia tips viewed at an angle, while the D-shaped traces result when the cilia are tilted so much that they slam into the "floor"---the embryonic cell surface---during their circuit. Scanning electron microscopy then revealed that all the cilia in the node tilted toward the posterior of the embryo and that each cilium was located toward the back of its node cell. Finally, the researchers built a 1,000× scale model of an embryonic node using wires to represent cilia and thick silicone fluid to represent the extra-embryonic fluid. As predicted by theoretical fluid dynamics, the liquid in this model always flowed leftward when the tilted wires were rotated clockwise. Hamada and colleagues propose that the position of the cilia and their tilt is determined by the pre-existing axis asymmetries (anteroposterior and dorsoventral asymmetries) in the embryo. The tendency of the cilia to rotate clockwise then produces the leftward fluid flow across the node that breaks embryonic left-right symmetry. Whether a similar mechanism occurs in vertebrates other than mice, and how exactly leftward flow contributes to left-right axis establishment remains to be determined, but at least the mystery of how rotating cilia can produce a linear flow has now been solved.	PubMed Central	2024-06-05T03:55:59.595730	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180515/", "journal": "PLoS Biol. 2005 Aug 26; 3(8):e291", "authors": [] }
PMC1180516	Using traditional statistical tools to analyze the modern wealth of biological data is a bit like trying to move a muscle car with a buggy whip---you\'re not likely to get anywhere very fast. The problem is perhaps most acute in the quest to understand how genes interact to regulate one another\'s expression. The amount of RNA made by any one gene is likely influenced by DNA at dozens of loci, or locations around the genome. Such loci are often situated within genes that participate in the same pathway as the gene being influenced, and a central goal is to understand this network of mutually influential genes and loci. Consider piecing together this puzzle for each of the many thousands of genes and many thousands of potentially influential loci, and the old analytical tools simply can\'t keep up. In this issue of PLoS Biology, John Storey and colleagues tackle the challenge with a new approach. The authors began by mating two strains of yeast that have minor differences in their DNA at more than 3,000 loci---creating over 3,000 markers---and then tracking the inheritance of these markers in the yeast offspring. Because the two genomes randomly reshuffle upon mating, any single offspring will contain some random combination of marker outcomes from each parent. The authors also examined the amount of RNA produced by over 6,000 individual genes in each offspring. The next step was to determine how these two large sets of data---variations at specific loci and variations in expression of specific genes---were correlated. Straightforward statistical tests performed on each gene\'s expression revealed the single most influential location in the DNA. But such tests don\'t reveal the more complicated reality that any single gene is likely to be influenced by more than one locus. Linking expression of a single gene (or "expression trait," in genetic parlance) to more than one locus has been stymied by the inability of conventional statistical approaches to cope with the mountains of data involved. Not only is an exhaustive pair-by-pair testing of all possible interactions computationally demanding, but it can also be very difficult to distinguish whether elevated expression is due to one or both of the loci being tested. The problem becomes exponentially harder as more potentially linked loci are tested. [](#pbio-0030294-g001){ref-type="fig"} ::: {#pbio-0030294-g001 .fig} ::: {.caption} ###### To unravel the tangled web of interactions between yeast genes, researchers bred two distinct yeast strains and statistically analyzed how gene expression varied between their offspring (Photo: David Byres) ::: ![](pbio.0030294.g001) ::: To overcome the limitations of standard approaches, Storey et al. used a novel statistical approach that exploited what they had the most of---data. They began by determining the single most significant locus for each expression trait. They then moved on to the next most significant locus for that trait, but tested its linkage (that is, its influence on expression) with the assumption that the first locus was also linked. The ability to assign a "probability of linkage" to the first locus greatly simplified the calculations for the subsequent locus, reducing by almost a thousand-fold the number of possibilities that needed to be tested. As opposed to standard methods, the authors show that their approach is able to assess true joint linkage of two loci to an expression trait, while requiring substantially less computation. In addition, they found that about one in seven expression traits is controlled by "epistatic," or hierarchical, relationships among the two loci, while the standard method revealed none. This method can be adapted to search for even larger numbers of linked loci, to provide insights into the many interlocking pathways that make up the gene regulatory network, and ultimately result in the organism itself.	PubMed Central	2024-06-05T03:55:59.596439	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180516/", "journal": "PLoS Biol. 2005 Aug 26; 3(8):e294", "authors": [] }
PMC1180517	Introduction {#s1} ============ Influenza A viruses are negative-strand RNA viruses of the Family Orthomyxoviridae that infect a wide variety of warm-blooded animals, including domestic and wild birds and mammals (e.g., humans, pigs, and horses). The natural reservoir for influenza virus is thought to be wild waterfowl, and genetic material from avian strains episodically emerges in strains infectious to humans. These human viruses continually circulate in yearly epidemics (mainly during the winter months in temperate climates), and antigenically novel strains emerge sporadically as pandemic viruses \[[@pbio-0030300-b01],[@pbio-0030300-b02]\]. In the United States, influenza is estimated to kill 30,000 people in an average year \[[@pbio-0030300-b03],[@pbio-0030300-b04]\]. Every few years, influenza epidemics boost the annual mortality level above this average, causing 10,000--15,000 additional deaths. Occasionally, and unpredictably, global pandemics of influenza occur, infecting 20% to 40% of the population in a single year and raising death rates dramatically above normal levels. Pandemic influenza A viruses emerged three times during the last century: in 1918 (H1N1 subtype), in 1957 (H2N2), and in 1968 (H3N2) \[[@pbio-0030300-b02],[@pbio-0030300-b05]\]. The recent circulation of highly pathogenic avian H5N1 viruses in Asia from 2003--2005 has caused at least 52 human deaths \[[@pbio-0030300-b06],[@pbio-0030300-b07]\] and has raised concern about the development of a new pandemic \[[@pbio-0030300-b05]\]. How and when novel influenza viruses emerge as pandemic strains and their precise mechanisms of pathogenesis are still not understood. While the risk of pandemic influenza poses a significant public health concern, inter-pandemic or epidemic influenza remains a major cause of morbidity and mortality. The influenza A surface glycoprotein hemagglutinin (HA) protein is under selective pressure for change in order to evade the host\'s immune system \[[@pbio-0030300-b08]\]. Antibodies against the HA protein inhibit receptor binding and are very effective at preventing reinfection with the same strain. However, HA can change to evade previously acquired immunity either by antigenic drift, whereby mutations of the currently circulating HA gene disrupt antibody binding, or by antigenic shift, in which the virus acquires an HA of a new subtype by reassortment of one or more gene segments. While it is generally accepted that drift is responsible for inter-pandemic influenza outbreaks and shift for pandemics, there are exceptions to this rule. For example, in 1977, an H1N1 virus re-emerged but failed either to cause a pandemic or to replace the prevailing H3N2 subtype \[[@pbio-0030300-b09]\]. The importance of predicting the emergence of new circulating influenza strains for subsequent annual vaccine development cannot be underestimated \[[@pbio-0030300-b10]\]. To this end, the global influenza surveillance network coordinated by the World Health Organization was established to select the candidate strains of influenza A and B for the yearly production of influenza vaccine in both the northern and southern hemispheres. The network characterizes the antigenic properties of influenza viruses using HA inhibition assays and sequencing of the HA1 domain (globular head) of HA of a select number of strains \[[@pbio-0030300-b11],[@pbio-0030300-b12]\]. Antigenic, genetic, and epidemiologic data are then examined to make recommendations of candidate vaccine strains. A number of retrospective studies have been performed using partial HA gene sequences to understand, and sometimes predict, the evolution of human H3N2 strains \[[@pbio-0030300-b13]--[@pbio-0030300-b17]\]. Accumulation of amino acid replacements in HA is clustered in five variable antigenic sites \[[@pbio-0030300-b18]\] around the receptor binding site. Phylogenetic analysis has revealed that 18 codons in the HA1 domain exhibit significantly more nonsynonymous nucleotide substitutions than synonymous ones, constituting a signature of strong, selectively driven, antigenic drift \[[@pbio-0030300-b15]\]. More recently the antigenic and genetic evolution of HA was compared through the construction of an antigenic map of H3N2 viruses \[[@pbio-0030300-b17]\]. Although antigenic evolution was found to be more punctuated than genetic evolution over the same time period, the two measures of HA drift were generally correlated. Despite the wealth of data on the molecular evolution of influenza viruses, how the entire genome of influenza A virus evolves during epidemic years is unclear, particularly as past sample sizes have been inadequate. While antigenic drift of HA is clearly of vital importance in the survival of an influenza strain, other factors, including HA receptor binding specificity \[[@pbio-0030300-b19]\], antigenic drift of neuraminidase (NA) \[[@pbio-0030300-b20]\], matched activity between HA and NA \[[@pbio-0030300-b21]--[@pbio-0030300-b23]\], and the interaction of the other influenza proteins with each other and their host cells, are all likely to affect viral fitness in a polygenic manner. Similarly, it is unclear how many lineages of influenza A viruses persist between seasonal epidemics, particularly in genes other than that encoding HA. To this end the National Institute of Allergy and Infectious Diseases of the National Institutes of Health has funded the Influenza Genome Sequencing Project with several partners \[[@pbio-0030300-b24],[@pbio-0030300-b25]\] including the Institute for Genomic Research. To date, 156 genomes of human H3N2 viruses collected between 1999 and 2004 from New York State have been completely sequenced and deposited in GenBank. We have performed an initial analysis of these viruses and have found evidence for both the existence of multiple clades of viruses co-circulating at the same time point and for multiple reassortment events among these clades. One of these reassortment events was the likely progenitor of the A/Fujian/411/2002-like drift epidemic of the 2003--2004 influenza season, in which there was a poor match between the vaccine strain and the predominant circulating viruses of that year \[[@pbio-0030300-b26],[@pbio-0030300-b27]\]. This report extends recent observations of reassortant H3N2 influenza A viruses from the southern hemisphere during this same time period \[[@pbio-0030300-b28]\]. Results {#s2} ======= Analysis of the Concatenated Complete Genome of the New York State Influenza Isolates {#s2a} ------------------------------------------------------------------------------------- Three major clusters of sequences were apparent in phylogenetic trees of the 156 complete genomes of H3N2 influenza A viruses sampled from New York State. These corresponded to particular influenza seasons (winter months): (a) 1999--2000, (b) 2001--2002 and 2002--2003 together (although only five members of latter season are present in these data), and (c) 2003--2004 ([Figure 1](#pbio-0030300-g001){ref-type="fig"}). Such temporal structure is commonly observed in trees of influenza A virus, and this is thought to be largely driven by positive selection acting on the HA gene \[[@pbio-0030300-b17]\]. However, a number of isolates did not fall into these three groups. First, three isolates from the 2002--2003 and 2003--2004 seasons---A/New York/32/2003, A/New York/198/2003, and A/New York/199/2003---formed a distinct phylogenetic group whose closest relationship was with those viruses sampled in the 1999--2000 season, rather than with those sampled during later seasons. We have denoted the two groups of viruses circulating after the 2001--2002 season as clade A (the major cluster after the 2001--2002 season) and clade B (the minor cluster comprising isolates 32, 198, and 199). Second, isolates A/New York/52/2004 and A/New York/59/2003 (designated as clade C in [Figure 1](#pbio-0030300-g001){ref-type="fig"}) occupied a position intermediate between clade A and the 2001--2002 group of viruses, as did isolate A/New York/11/2003. Finally, isolates A/New York/137/1999 and A/New York/138/1999 formed a group that was divergent from the main group of viruses sampled in the 1999--2000 season. ::: {#pbio-0030300-g001 .fig} Figure 1 ::: {.caption} ###### Evolutionary Relationships of Concatenated Major Coding Regions of Influenza A Viruses Sampled in New York State during 1999--2004 The maximum likelihood phylogenetic tree is mid-point rooted for purposes of clarity, and all horizontal branch lengths are drawn to scale. Bootstrap values are shown for key nodes. Isolates assigned to clade A (light blue), clade B (yellow), and clade C (red) are indicated, as are those isolates involved in other reassortment events: A/New York/11/2003 (orange), A/New York/182/2000 (dark blue), and A/New York/137/1999 and A/New York/138/1999 (green). ::: ![](pbio.0030300.g001) ::: Analysis of Individual Gene Segments {#s2b} ------------------------------------ To investigate the evolutionary history of the outlier viruses in more detail we inferred phylogenetic trees for each of the eight individual gene segments ([Figure 2](#pbio-0030300-g002){ref-type="fig"}). Strikingly, although the distinction between clades A, B, and C was apparent in seven of the eight genes, no such separation was seen in the HA phylogeny. In this case, clades B and C clearly clustered within clade A and with strong bootstrap support. The close phylogenetic relationship of these three groups of viruses in HA set against a background of genetic divergence in all other segments strongly suggests that these data contain evidence for at least two independent reassortment events, one involving clades A and B and another involving clade C and either clade A or clade B. In the case of the clade C viruses, the separate gene phylogenies also reveal that these isolates share a common ancestry with viruses first sampled during the 2001--2002 season, while the clade B viruses share a closer relationship with those viruses of the 1999--2000 season. ::: {#pbio-0030300-g002 .fig} Figure 2 ::: {.caption} ###### Evolutionary Relationships of Individual Major Coding Regions from Influenza A Viruses Sampled in New York State during 1999--2004 All maximum likelihood phylogenetic trees are mid-point rooted for purposes of clarity only, and all horizontal branch lengths are drawn to scale. Bootstrap values are shown for clades A, B, and C. Colors are as in [Figure 1](#pbio-0030300-g001){ref-type="fig"}. ::: ![](pbio.0030300.g002) ::: Two more major phylogenetic displacements suggestive of reassortment involving other segments were similarly identified. First, isolate A/New York/11/2003, which fell within clade A in seven of the gene trees (including HA), clustered with clade B viruses in PB2. Consequently, isolate A/New York/11/2003 represents a reassortment of two segments between clades A and B. Second, isolate A/New York/182/2000, which clustered with the main set of viruses sampled during the 1999--2000 season in most of the gene trees, was very closely related to the divergent A/New York/137/1999 and A/New York/138/1999 isolates in PA and M1, although the high degree of genetic similarity among all viruses in M1 precludes a further analysis of reassortment in this case. To determine the direction of the reassortment events in HA, we inferred phylogenetic trees of larger datasets comprising the New York State isolates and representatives of the other human and swine H3N2 viruses sampled during the same time period. Because sequences from the core genes have only been sporadically collected, this analysis necessarily focused on HA and NA. As expected from the phylogenetic analysis of the New York State viruses, the distinction between clades A, B, and C was apparent in the NA gene tree ([Figure 3](#pbio-0030300-g003){ref-type="fig"}) as well as the core genes (trees shown in [Figure S1](#sg001){ref-type="supplementary-material"}). Moreover, in the NA gene tree, viruses sampled from a variety of locations during 2000--2005 fell into clade B, including Europe (Denmark and Norway), Asia (China and Singapore), and the Americas (Argentina, Brazil, and the United States). Hence, although clade B was at low frequency in the New York State dataset, it represents a distinct lineage of H3N2 viruses globally circulating from at least 2000 to 2005. Similarly, a third North American virus, A/Charlottesville/03/2004, was designated as clade C. ::: {#pbio-0030300-g003 .fig} Figure 3 ::: {.caption} ###### Evolutionary Relationships of 197 NA Sequences Sampled from Mammalian Hosts during 1999--2005 The maximum likelihood phylogenetic tree is mid-point rooted for purposes of clarity only, and all horizontal branch lengths are drawn to scale. Bootstrap values are shown for clades A, B, and C. Colors are as in [Figure 1](#pbio-0030300-g001){ref-type="fig"}. ::: ![](pbio.0030300.g003) ::: A very different evolutionary history was revealed in HA. In this case, clade A of the New York State viruses expanded to contain the majority of viruses sampled after 2002 and from a variety of locations (Asia, Australasia, Europe, and North America), as well as a number of Asian viruses from 2002 ([Figure 4](#pbio-0030300-g004){ref-type="fig"}). This large group of viruses then clustered within clade B, such that most clade B viruses sampled from 2000 to 2002 formed a clear, but closely related, outgroup to the later clade A viruses, with the remaining clade B viruses falling within clade A ([Figure 4](#pbio-0030300-g004){ref-type="fig"}). The three clade C viruses also fell within this expanded clade A. Such a phylogenetic pattern strongly suggests that the HA from both the clade A and clade C viruses was acquired from that present in clade B through reassortment. In this context it is of interest to note that the HA of the A/Fujian/411/2002 virus and related isolates are present in this reassorted clade. Further, the fact that the clade A and B isolates closest to the phylogenetic location of the reassortment event were both sampled in 2002 suggests that the reassortment occurred in this year, although pinpointing the exact phylogenetic location of the recombinant event is difficult given the relatively small number of samples available from this critical time period. Similarly, the fact that these viruses had Asian origins is also compatible with the reassortment event occurring in this region, a hypothesis also supported by a recent analysis of comparable partial genomic analysis of H3N2 isolates from the southern hemisphere \[[@pbio-0030300-b28]\]. ::: {#pbio-0030300-g004 .fig} Figure 4 ::: {.caption} ###### Evolutionary Relationships of 256 Sequences of the HA1 Sequences Sampled from Mammalian Hosts during 1999--2004 The maximum likelihood phylogenetic tree is rooted on a divergent set of human and swine viruses for purposes of clarity only, and all horizontal branch lengths are drawn to scale. Bootstrap values are shown for key nodes. Colors are as in [Figure 1](#pbio-0030300-g001){ref-type="fig"}, with yellow indicating all those viruses identified as clade B viruses from the analysis of the expanded NA dataset. The phylogenetic location of the reassortment event between clades A and B was set at the position of the most basal clade A virus, defined on the basis of the NA analysis, within the older clade B. The phylogenetic locations of the two critical antigenic mutations at sites 155 and 156 are also shown. ::: ![](pbio.0030300.g004) ::: Analysis of the Coding Differences between Clades A, B, and C {#s2c} ------------------------------------------------------------- Because both clade A and clade B contain viruses sampled on a near global basis, it is important to determine possible phenotypic differences between them. [Table 1](#pbio-0030300-t001){ref-type="table"} shows the amino acid replacements that consistently distinguish the clade A and B viruses. These changes are not uniformly distributed among the seven segments (not including HA). Of the 48 amino acid differences between clades A and B, 14 fall in NA and nine in NP. PB1, PB2, and PA have five differences each, while M2 and NS1 have three differences, and M1 and NS2 have two. In contrast, there are only nine amino acid differences between the clades A and C: PB2---T9N; PA---A20T, L226I, and N272D; NP---G450S; NA---H40Y and V263I; and NS1---A56T and N143T. ::: {#pbio-0030300-t001 .table-wrap} Table 1 ::: {.caption} ###### Amino Acid Changes Between the Proteins Encoded by the Clade A and Clade B Viruses ::: ![](pbio.0030300.t001) ::: Discussion {#s3} ========== Our analysis of whole genomes of H3N2 influenza A viruses sampled during 1999--2004 has identified two key evolutionary patterns. First, although the majority of viruses isolated after 2002 fall into a single phylogenetic group (clade A), multiple, co-circulating viral lineages are present at particular time points. The genetic diversity of influenza A virus is therefore not as restricted as previously suggested, particularly when genes other than that encoding HA are analyzed. This co-circulation of lineages is most apparent with the identification of three clades of H3N2 viruses that appear to infect the same populations until 2002, after which they acquired a common HA gene through reassortment. Second, and more dramatically, these multiple, co-circulating lineages may have complex genealogical histories and interact through reassortment. Indeed, we have documented two reassortment events involving the HA gene of clade B: one in which it was acquired by the clade A viruses and another in which it was independently acquired by those isolates assigned to clade C. Two further reassortment events involving the PB2 and PA genes were also evident from our phylogenetic analysis. Given that we are only able to reliably detect reassortment when it is associated with major changes in tree topology, it is likely that reassortment among closely related lineages is also commonplace in influenza A viruses. Reassortment between influenza A viruses has been described in both human and animal viruses \[[@pbio-0030300-b01],[@pbio-0030300-b29]\]. Notably, antigenic shift by reassortment between human and avian influenza A viruses has been documented in the formation of the 1957 H2N2 and 1968 H3N2 pandemics \[[@pbio-0030300-b30]--[@pbio-0030300-b32]\]. Other recent examples of reassortment between human and animal influenza A viruses have resulted in the emergence of novel H3N2 and H1N2 swine viruses in North America and Europe \[[@pbio-0030300-b33],[@pbio-0030300-b34]\] and the evolution of H5N1 viruses in Asia from 1997 to the present \[[@pbio-0030300-b35]\]. Reassortment between co-circulating lineages of human influenza A and more recently influenza B viruses following mixed infection has also been described \[[@pbio-0030300-b36]--[@pbio-0030300-b41]\]. For example, human H2N2 viruses formed two distinct clades in the 1960s prior to the emergence of the 1968 H3N2 pandemic virus, with one virus a reassortant containing genes of both clades \[[@pbio-0030300-b42]\]. Similarly, the early H3N2 viruses (1968--1972) acquired the H3 HA and the PB1 gene via reassortment with an avian virus \[[@pbio-0030300-b30],[@pbio-0030300-b31]\]. Reassortment between H3N2 and H2N2 viruses may therefore have assisted successful cross-species transmission \[[@pbio-0030300-b42]\]. Reassortant viruses were also described following the re-emergence of the H1N1 subtype in 1977 that did not replace the previously circulating H3N2 viruses. In this case, co-circulation of influenza viruses of both subtypes continued, and co-infection with both subtypes was reported \[[@pbio-0030300-b43]\]. While reassortant H3N1 strains were not isolated, H1N1 strains containing reassorted internal protein-encoding gene segments from H3N2 viruses were observed \[[@pbio-0030300-b44],[@pbio-0030300-b45]\]. Occasional isolates of H1N2 viruses were also detected after the re-emergence of H1N1 \[[@pbio-0030300-b46],[@pbio-0030300-b47]\]. More recently, the widespread circulation of viruses with the H1N2 subtype has been documented \[[@pbio-0030300-b41]\]. These viruses contained the HA segment of contemporary H1N1 viruses reassorted onto an H3N2 background, a 7:1 reassortment pattern similar to that observed with the sporadically circulating H1N2 viruses of the 1980s and early 1990s \[[@pbio-0030300-b47]\] and to the dominant reassortments described in this analysis. Since the H1 and N2 subtype proteins were antigenically and genetically similar to co-circulating H1N1 and H3N2 subtype viruses, the emergence of this new subtype did not result in an epidemiologically significant event \[[@pbio-0030300-b41]\]. Reassortment among co-circulating clades of H3N2 viruses like that observed in the current study has also been previously described, including reassortment of the NA gene segment \[[@pbio-0030300-b48]\] and the core protein-encoding segments \[[@pbio-0030300-b49]\]. Most prior phylogenetic studies of human influenza A have suggested that inter-pandemic evolution may be essentially described as a series of successions by variants of the previous season\'s dominant strain. These successions are largely determined by strong positive selection acting on the abundance of mutational diversity in the HA of the dominant strain. However, we found that at least four reassortment events occurred among human viruses during the period 1999--2004 and that two of these involved a major change in HA. Recently, Barr et al. independently provided phylogenetic evidence of the clade A--clade B reassortment described here in an analysis of predominantly southern hemispheric influenza A H3N2 isolates collected during the same period \[[@pbio-0030300-b28]\]. To our knowledge, these analyses are the first demonstrations of the emergence of a major antigenically variant virus derived by reassortment between two distinct clades of co-circulating H3N2 viruses rather than by antigenic drift. These findings suggest that the ongoing evolution of human influenza A virus is likely to be more complex than depicted in standard models of antigenic drift; multiple lineages of antigenically distinct viruses can persist within populations and, through their reassortment, produce major changes in antigen space. Similarly, the persistence of multiple lineages of H3N2 within a single population indicates that human populations represent a larger reservoir of genetically distinct viruses than previously anticipated. Indeed, it is possible that key changes in antigen type, depicted as jumps between cluster types \[[@pbio-0030300-b17]\], could be strongly influenced by reassortment among co-circulating human strains. Crucially, the real importance of both lineage persistence and reassortment in influenza A virus evolution could not be determined until a representative sample of full-genome sequences was collected from a single population. In the 2003--2004 influenza season, a major drift variant emerged in both the northern and southern hemispheres \[[@pbio-0030300-b50]--[@pbio-0030300-b52]\]---the A/Fujian/411/2002-like variant. In the United States, the 2001--2002 influenza was an H3N2-predominant season, and all antigenically characterized isolates matched the A/Moscow/10/1999 vaccine strain \[[@pbio-0030300-b53]\]. In the 2002--2003 season, which was an H1- and influenza B--dominant season in the United States, a minority of antigenically characterized H3N2 isolates were different from the Moscow/10/1999-like vaccine strain \[[@pbio-0030300-b54]\], probably coinciding with the emergence of the Fujian strain. Indeed, in our phylogenetic analysis two of five H3N2 viruses from the 2002--2003 season fall into clade B, along with two newly sequenced isolates that were not included in our analysis (A/New York/201/2003 and A/New York/203/2003). Consequently, clade B viruses made up a significant fraction of the few H3N2 isolates in that season. In contrast, only a small minority of influenza H3N2 viruses sampled in Europe at this time were antigenically characterized as Fujian-like \[[@pbio-0030300-b50]\], while in the southern hemisphere\'s 2003 influenza season the Fujian strain was predominant \[[@pbio-0030300-b28],[@pbio-0030300-b51]\]. The 2003--2004 influenza season in the northern hemisphere was also dominated by the Fujian strain \[[@pbio-0030300-b55]\], although the vaccine contained the H3N2 (A/Panama/2007/1999) from the previous year. This strain was a poor antigenic match to the Fujian strain \[[@pbio-0030300-b26],[@pbio-0030300-b27]\], which in turn led to reduced vaccine effectiveness. Thirteen amino acid changes distributed across the five antigenic sites of the HA1 domain distinguish the A/Panama/2007/1999-like and A/Fujian/411/2002-like strains. While these replacement changes appear to have phenotypic consequences (e.g., replication efficiency in eggs), only two residues, 155 and 156, are responsible for the major antigenic differences between the strains \[[@pbio-0030300-b27]\]. Although data are insufficient for precise determination of the timing of these two critical mutations, the available data are most consistent with these changes occurring in a relatively short time period before the reassortment event. The histidine to threonine change at site 155 and the glutamine to histidine change at site 156 are present in all the clade A reassortant isolates as well as in clade B isolates from 2003--2004, thus suggesting that they occurred prior to the reassortment event. No available clade B isolates prior to 2002--2003 have either of these mutations, and we were able to identify only three "intermediate" isolates from 2002--2003 (A/Kwangju/219/2002, A/Kwangju/243/2002, and A/Cheonnam/340/2002) with the replacement at site 155 but not at site 156. Overall, the data presented here, coupled with those recently reported \[[@pbio-0030300-b28]\], reveal that the HA segment of the H3N2 clade B viruses, present in low frequencies at least since 1999, was reassorted into clade A of H3N2, probably in 2002 soon after the appearance of these two critical mutations, and that this reassortment was central to the production of antigenically variant strains that were poorly matched to the vaccine strain in the 2003--2004 season \[[@pbio-0030300-b27]\]. In addition, presumably because of the high rate of reassortment and the fitness advantage conferred by these two mutations, this clade B HA segment also appeared to replace the HA segment of the clade C strains. Finally, though previously present only in low frequencies, recent sequencing of HA and NA by investigators in Denmark \[[@pbio-0030300-b56]\] and published phylogenetic trees from Australia \[[@pbio-0030300-b28],[@pbio-0030300-b51]\] show the existence of clade B virus, suggesting that it continued to have a global distribution and sometimes at appreciable frequency. Several questions remain unanswered by our study. Since the HA donated by clade B led to a major expansion of the reassorted clade A, it is uncertain why clade B did not initially out-compete clade A without reassortment. One possibility is that the HA of clade B had an intrinsically higher fitness than other HAs circulating at the same time but was unable to reach a high frequency in the New York population owing to linkage to mutations located in other segments that reduced the overall fitness of this genotype. According to this hypothesis, it was not until it was placed by reassortment into a more favorable genetic background, in this case the clade A viruses, that its fitness advantage was realized. Since clade B itself appeared to proliferate in other regions, it will be useful to analyze whole-genome sequence from these isolates when they are available. More generally, it is clear that the genotypic basis to viral fitness has not been entirely elucidated. In particular, it is likely that interactions among viral proteins and between viral proteins and host factors play a key role. In this respect it is notable that of the 48 amino acid differences that distinguish the clade B viruses, nine fall in NP and 14 fall in NA (see [Table 1](#pbio-0030300-t001){ref-type="table"}). In NP, the replacement at residue 425 falls within an HLA-B35-restricted cytotoxic T lymphocyte (CTL) epitope and is not commonly seen in human populations \[[@pbio-0030300-b57]\]. Two other changes, at residues 27 and 197, have also been identified as being contained in two HLA-A11-restricted CTL epitopes \[[@pbio-0030300-b58]\]. Another unusual change in NP is M136I: a change to methionine at this site was proposed as one of six human adaptive changes distinguishing the 1918 NP from avian NPs \[[@pbio-0030300-b59]\]. Indeed, all 45 available NP sequences from human H2N2 viruses, and all but four of approximately 250 human H3N2 NP sequences, maintain this methionine. Only the three New York clade B viruses, as well as A/Taiwan/1/71, have a change to an isoleucine at this residue. Of the changes in NA, six of them lie in five regions previously identified as being phylogenetically important regions, residues 40, 42, 143, 199, 307, and 385 \[[@pbio-0030300-b60]\], and therefore play a role in virus--host interaction. Three further changes, at residues 172, 199, and 307, map to antigenic sites \[[@pbio-0030300-b20]\], while a change at residue 18 has been mapped as an HLA-A2-restricted CTL epitope \[[@pbio-0030300-b58]\]. Similarly, two changes in M2, at residues 51 and 56, map to an HLA-A11-restricted CTL epitope, while residue 82 of NS1 maps to an HLA-A2-restricted CTL epitope \[[@pbio-0030300-b58]\]. Another change between clade A and B viruses, at residue 226 of PA, also maps to an HLA-A2-restricted CTL epitope \[[@pbio-0030300-b58]\]. It is possible that some of the mutations that fall in CTL epitopes assist the persistence of clade B by elongating the viral infectious period \[[@pbio-0030300-b61]\]. However, any combination of the constellation of amino acid changes may have altered the fitness of the clade B viruses in a way that we do not have the ability to understand from sequence analysis alone. Interestingly, Gulati et al. \[[@pbio-0030300-b62]\] have recently shown that Fujian-like strains have a mismatch in their HA and NA activities that is probably the result of the reassortment event described in the current study. The significance of this for the pathophysiology of the virus is currently unknown. In summary, our study clearly demonstrates the utility of whole-genome analyses of influenza A viruses, and further makes clear that additional whole-genome analyses are required to understand fully the evolutionary mechanisms and epidemiological dynamics of this virus. While antigenic variance of HA is still the dominant selective pressure on human influenza A virus evolution, the finding that antigenically novel clades emerge by reassortment among persistent viral lineages rather than via antigenic drift is of major significance for vaccine strain selection. Materials and Methods {#s4} ===================== Influenza viruses used in this study {#s4a} ------------------------------------ The influenza virus isolates were collected as part of the diagnostic service provided by the Virus Reference and Surveillance Laboratory at the Wadsworth Center, New York State Department of Health. Viruses were received as part of outbreak investigations, through the reference function of the laboratory, and, since 2001, as part of a sentinel physician influenza surveillance program. Viruses were passaged minimally in primary rhesus monkey kidney cell culture and the RNA extracted from the clarified supernatant. Whole-genome sequence information was derived at the Institute for Genomic Research using methods described elsewhere (E. Ghedin, N. A. Miller, M. Shumway, J. Zaborsky, T. Feldblyum, et al., unpublished data). Use of the diagnostic samples in this study was approved by the New York State Department of Health Institutional Review Board. Sequences used in the analysis {#s4b} ------------------------------ Sequence data for 156 complete genomes of influenza A virus (H3N2) sampled from New York State during the period 1999--2004 were downloaded from GenBank (1,248 separate accessions, representing the eight gene segments of 156 individual influenza isolates; GenBank accession numbers available from <http://www.ncbi.nlm.nih.gov/genomes/FLU/FLU.html>). Separate sequence alignments were then manually compiled for the major coding regions of each segment: PB2, 2,277 bp; PB1, 2,271 bp (PB1 protein); PA, 2,148 bp; HA, 1,698 bp; NP, 1,494 bp; NA, 1,407 bp; MP, 756 bp (M1 protein); and NS, 690 bp (NS1 protein). An alignment of the concatenated major coding regions was also constructed for all 156 isolates (12,741 bp). To place the New York State viruses in the context of global H3N2 evolution, larger datasets were complied comprising the New York State isolates plus all other mammalian influenza A viruses sampled from 1999--2004 available on GenBank and the Los Alamos Influenza Sequence Database (<http://www.flu.lanl.gov/>). For the six core genes, few such background sequences were available, particularly from clades B and C (see below). However, far larger numbers of background sequences were available for the HA1 domain portion of HA and for sequence of NA genes. To facilitate the computational analysis of these background datasets, those New York State isolates with identical sequences were removed from the analysis. This resulted in alignments of the following sizes: PB2, 134 sequences, 2,277 bp; PB1, 135 sequences, 2,271 bp; PA, 140 sequences, 2,148 bp; HA, 256 sequences, 987 bp; NP, 117 sequences, 1,494 bp; NA, 197 sequences, 1,407 bp; M1, 73 sequences, 756 bp; and NS1, 75 sequences, 690 bp. A full list of the isolates used in this study is provided in [Table S1](#st001){ref-type="supplementary-material"}. Phylogenetic analysis {#s4c} --------------------- Phylogenetic trees were inferred for all of the datasets described above using the maximum likelihood method available in the PAUP\* package \[[@pbio-0030300-b63]\] (see [Table S2](#st002){ref-type="supplementary-material"}). In each case the general time-reversible model of nucleotide substitution was used also incorporating a proportion of invariable sites and a gamma distribution of rate variation among sites with four rate categories. All parameter values were estimated from the empirical data and are given in [Table S2](#st002){ref-type="supplementary-material"}. Tree bisection--reconnection branch-swapping was used in all cases apart from the expansive ("background") HA and NA datasets, which contained so many sequences that the analysis was restricted to subtree pruning--regrafting branch-swapping. To assess the reliability of key nodes on the phylogenetic trees, a bootstrap resampling analysis was also undertaken in each case. This involved the inference of 1,000 replicate neighbor-joining trees using the maximum likelihood substitution model inferred for each dataset. Supporting Information {#s5} ====================== Figure S1 ::: {.caption} ###### Phylogenetic Trees of New York State and Background H3N2 Strains Inferred from the Core Genes Maximum likelihood phylogenetic trees depicting the evolutionary relationships of the remaining six segments (major coding regions only) from the H3N2 influenza A viruses sampled in New York State during the period 1999--2004 (unique sequences only) and the "background" viruses taken from GenBank or the Los Alamos Influenza Sequence Database. All trees are mid-point rooted for purposes of clarity only, and all horizontal branch lengths are drawn to scale. Bootstrap values are shown for clades A, B, and C. Colors are as in [Figure 1](#pbio-0030300-g001){ref-type="fig"}. (715 KB EPS). ::: ::: {.caption} ###### Click here for additional data file. ::: Table S1 ::: {.caption} ###### Isolates of Influenza A Virus H3N2 Used in This Study (164 KB DOC). ::: ::: {.caption} ###### Click here for additional data file. ::: Table S2 ::: {.caption} ###### Parameter Values for Maximum Likelihood Phylogenetic Analysis (50 KB DOC). ::: ::: {.caption} ###### Click here for additional data file. ::: The authors wish to acknowledge the excellent technical assistance of Sara Griesemer and Matthew Kleabonas. Viruses described in this study collected after 2001 include some isolates collected as part of the Sentinel Physician Influenza Surveillance Program, which is supported by Cooperative Agreement Number U50/CCU223671 from the Centers for Disease Control and Prevention. The work at the Institute for Genomic Research and EG, NM, SLS, and CMF were supported in whole or in part with federal funds from the National Institute of Allergy and Infectious Diseases, National Institutes of Health, Department of Health and Human Services, under contract number N01-AI-30071. The contents of this manuscript are solely the responsibility of the authors and do not necessarily represent the official views of the Department of Health and Human Services or the Department of Defense. Competing interests. The authors have declared that no competing interests exist. Author contributions. ECH conducted the phylogenetic analyses. EG and NM developed the laboratory protocols and sequenced the genomes for the influenza sequencing project. JT and KS collected the clinical isolates, selected the subset for sequencing, and prepared the viral RNA. YB performed quality assurance and annotated the sequences. SLS managed development of bioinformatics software for assembly and data management for the influenza sequencing project. CMF contributed to overall project management. DJL and JKT analyzed the data. ECH, BTG, DJL, and JKT wrote the paper. Citation: Holmes EC, Ghedin E, Miller N, Taylor J, Bao Y, et al. (2005) Whole-genome analysis of human influenza A virus reveals multiple persistent lineages and reassortment among recent H3N2 viruses. PLoS Biol 3(9): e300. CTL : cytotoxic T lymphocyte HA : hemagglutinin NA : neuraminidase	PubMed Central	2024-06-05T03:55:59.597155	2005-7-26	{ "license": "Creative Commons Zero - Public Domain - https://creativecommons.org/publicdomain/zero/1.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180517/", "journal": "PLoS Biol. 2005 Sep 26; 3(9):e300", "authors": [ { "first": "Edward C", "last": "Holmes" }, { "first": "Elodie", "last": "Ghedin" }, { "first": "Naomi", "last": "Miller" }, { "first": "Jill", "last": "Taylor" }, { "first": "Yiming", "last": "Bao" }, { "first": "Kirsten", "last": "St George" }, { "first": "Bryan T", "last": "Grenfell" }, { "first": "Steven L", "last": "Salzberg" }, { "first": "Claire M", "last": "Fraser" }, { "first": "David J", "last": "Lipman" }, { "first": "Jeffery K", "last": "Taubenberger" } ] }
PMC1180518	Soon after a human baby emerges from the womb, its brain churns out new cells at an unbelievable rate of 250,000 neurons per minute. Each neuron typically acquires multiple dendrites to pick up electrical signals from other cells, and just one axon to transmit those signals to other cells. An axon can grow several feet in the service of relaying signals to its target tissue, but how it manages this growth has been an open question. Now Carlos Portera-Cailliau and colleagues use the real-time fidelity offered by two-photon microscopy to peer inside the brain of a living neonatal mouse and observe axons growing and reorganizing. By creating a window into the developing mouse brain, the authors could watch and record time-lapse images of living axons as they migrated through the neural landscape. [](#pbio-0030301-g001){ref-type="fig"} ::: {#pbio-0030301-g001 .fig} ::: {.caption} ###### Implanting "windows" into the skulls of newborn mice allowed researchers to observe axon development ::: ![](pbio.0030301.g001) ::: Researchers have long been aware that a complex process of axon growth and retraction occurs during the first few weeks of life, but until now have not been able to observe it for themselves in real time, at least not in mammals. Axon growth and guidance have been studied extensively in cultured neurons, and, along with time-lapse imaging of the brains of live frogs and fish, have provided insights into brain development in lower vertebrates. But because studies of mammalian axon growth have relied on measurements taken from fixed brain tissue samples, very little is known about the details of the growth and refinement of axonal projections in mammals. Portera-Cailliau et al. solved this problem by using two-photon microscopy, a state-of-the-art imaging technique that allowed them to watch the same developing axons in the cerebral cortex of live mice during the first two weeks of life. The group used a line of transgenic mice that expressed green fluorescent protein in small subsets of neurons. The procedure involved removing a portion of the skull and replacing it with a glass coverslip, thus creating a window for viewing live brain tissue at different developmental stages. Although traditional confocal microscopy can also be used to view fluorescently labeled cells, two-photon microscopy can image deeper into tissue without damaging the brain cells, which allows imaging repeatedly over long periods of time. Time-lapse images were generated from the mice over a period of several minutes to as long as three weeks, allowing the researchers to observe changes in neuronal structure throughout the critical period of cortical development. Portera-Cailliau et al. observed changes in two very different types of neurons: thalamocortical (TC) neurons and Cajal-Retzius (CR) neurons. Both of these neurons send axons to the outermost layer of the cortex, but CR neurons project only locally and TC neurons project over long distances. The structure and dynamics of axons from these neurons, it turns out, are also very different, indicating that axonal development is not homogeneous across cell types. For example, TC axons grew quickly in long, straight paths and added new branches frequently. By contrast, CR axons grew much more slowly, along tortuous paths. Additionally, TC axons exhibited both short branch retraction (tens of microns) and elimination of larger branches (hundreds of microns or more), while CR axons only used branch tip retraction for pruning. Thus, the structure and dynamics of axon elaboration are dependent on neuronal cell type, even when they grow side by side in the same environment. This suggests that different neurons may exhibit different axon elaboration programs and/or interpret differently cues from their surroundings. This study provides insight into the process of refinement and optimization of neuronal circuitry that occurs in mammals in the early stages of life, and begins to solve the mystery about how axons develop in the cortex. By observing axon growth in real time, scientists have taken the first step in understanding the cues that control each twist and turn of every axon in the cortex. Such efforts may one day suggest ways to prevent or treat the many types of cognitive disabilities that arise from abnormalities in brain development.	PubMed Central	2024-06-05T03:55:59.600229	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180518/", "journal": "PLoS Biol. 2005 Aug 26; 3(8):e301", "authors": [] }
PMC1180519	Last October, as Americans started lining up for flu shots, news broke that 48 million vaccine doses had been contaminated. With 100 million people considered at high risk and fears of a potentially deadly avian flu epidemic on the horizon, the shortage caused long lines, allegations of price gouging, and a new bill to bolster the nation\'s anemic vaccine manufacturing base. Influenza A viruses are RNA viruses that infect humans, pigs, horses, and birds, both wild and domestic. Flu infection relies on a viral glycoprotein, hemagglutinin (HA), that binds to receptors on a host cell and allows the virus to be internalized. If antibodies produced by host immunity recognize viral antigens (on the surface of the HA protein), HA binding is inhibited and infection prevented. A virus\'s best chance of gaining the upper hand in this evolutionary game of cat and mouse is to change its HA in a way that eludes antibody recognition. Typically the mutations are minor and the virus\'s antigens conserved enough for the host body\'s immune system to recognize them. On occasion, influenza can acquire an antigenically novel HA subtype, becoming a virulent pandemic strain that completely escapes immune surveillance and kills millions. Minimizing the effect of yearly influenza outbreaks---by developing effective matched vaccines---depends on predicting which flu strains are likely to evolve. Toward this end, Eddie Holmes and colleagues took the global approach afforded by genomics to explore the forces underlying viral adaptations. They found multiple flu strains circulating in the population at the same time, and a more complex evolutionary pattern than previously thought. They also showed that co-circulating viruses can exchange genes in a way that creates antigenically novel, epidemiologically significant strains---a process that humans may facilitate by simultaneously hosting more than one strain. [](#pbio-0030302-g001){ref-type="fig"} ::: {#pbio-0030302-g001 .fig} ::: {.caption} ###### A transmission electron micrograph of the influenza A virus. New evidence suggests that flu viruses can rapidly reshuffle genetic material and mutate into new strains capable of widespread infection. (CDC/Dr. Erskine Palmer) ::: ![](pbio.0030302.g001) ::: They analyzed the genomes of 156 influenza A viruses (serotype H3N2) collected by New York State public health officials between 1999 and 2004 in search of global patterns of viral evolution. Using the flu virus genome sequences produced at the Institute for Genomic Research (TIGR), funded by a National Institute for Allergy and Infectious Diseases (NIAID) initiative, the authors grouped the viral sequences according to sequence similarity. They also included partial flu sequences obtained from other studies in their analysis. These data are the initial output of the first large-scale effort to completely sequence influenza genomes. While most of the virus genomes sampled after 2002 fell into one group---which the authors called clade A---there were also other clades circulating at different times (called clades B and C). Gene trees, or phylogenies, constructed for each of the virus\'s eight genes all diverged according to their respective clades, except one---the HA gene. The HA gene cluster grouped all the clade A viruses that emerged after 2002 as well as both the clade B and C viruses from the same time period and viruses from multiple locations (in Asia, Australia, Europe, and North America). Altogether, these results indicate that different viral strains had circulated in the same populations until 2002 and then the clade A and C viruses acquired a common HA gene from clade B through reassortment. While reassortment between co-circulating human influenza strains has been previously described, this study is the first to examine in detail a reassortment event leading to an epidemiologically significant outcome, the emergence of the "Fujian" strain in the 2003--2004 season. Though it\'s not yet clear how variant clades manage to persist alongside dominant strains, the fact that they do suggests the influenza virus has multiple adaptive tools at its disposal. Luckily, the tools of genomics should help predict what evolutionary paths the virus might take and help in the process of selecting the most promising vaccines to contain it.	PubMed Central	2024-06-05T03:55:59.600795	2005-7-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180519/", "journal": "PLoS Biol. 2005 Sep 26; 3(9):e302", "authors": [] }
PMC1180707	Background ========== Mood disturbances are widespread in western societies and general practitioners (GPs) see many patients with depressive disorders\[[@B1],[@B2]\]. Only a minority of these patients consult directly for psychological problems. More often they consult their doctors because of bodily symptoms\[[@B3]\]. In addition, depressive disorders often present combined with anxiety states\[[@B4],[@B5]\] and a large proportion of the depressed patients in primary care also suffer from concurrent somatic diseases\[[@B6]\]. Consequently, when GPs see depressed patients, they often have to deal with complicated cases. The ways GPs handle these patients have been criticised because of deficient identification and insufficient treatment of the depressive disorders \[[@B7]-[@B11]\]. In two previous studies we explored Swedish GPs\' conceptions of depressive disorders and their ideas on how their working methods with depressed patients had been shaped. We found a considerable variation in their conceptions about depression, but a relative consensus concerning drug treatment of major depression. As regards manner of working, many interviewed GPs considered individual experiences from general practice and private life as more influential than academic education and professional literature\[[@B12],[@B13]\]. When further examining the geographic variation of sales levels of antidepressant agents (ADs) we found weak but statistically significant associations with some of local GPs\' conceptions of factors that may have formed the way they treat depressive patients\[[@B14]\]. Thus, high sales levels correlated positively with a high evaluation of ADs\' effectiveness in depression and panic disorders and were inversely correlated with the degree of appreciation of psychotherapy-based treatments. High sales levels were also associated with a high evaluation of GPs\' own clinical and private experience, with a positive appreciation of the work with depressed patients and with a high level of participation in the pharmaceutical companies\' activities. The purpose of the present study was to elaborate further the frequencies of Swedish GPs\' conceptions of depressive disorders and its treatment and of their ideas of factors that may influence their manner of work with depressive patients. Methods ======= Participants ------------ The present study made use of the same data collection as the previous study of the relations between GPs\' conceptions of treatment of depression and local sales levels of ADs\[[@B14]\]. For that purpose we had surveyed a stratified sampling of GPs based on the population\'s purchases of ADs in each GP\'s working area. Information on purchases of ADs for the year 2000 in each Swedish municipality was obtained from Apoteket AB (The National Corporation of Swedish Pharmacies). We selected three groups of municipalities; those with the highest, the average, and the lowest AD sales rates. For each sales group, municipalities were included according to their sales figures until the total number of GPs working in the municipalities of the group approximated 200. In that way we selected 617 GPs working in a total of 60 municipalities. The respondents of a pilot study were not included. The names and addresses were received from Läkarmatrikeln 2001, a national register of Swedish physicians. Instrument ---------- A postal questionnaire was designed based on the findings in the interview study\[[@B12],[@B13]\]. A five-point Likert scale ranging from \"Agree not at all\" to \"Agree completely\" and from \"Very little importance\" to \"Very great importance\" was used for most items. The questionnaire was tested in 20 GPs. That pilot study had a response rate of 75 per cent, no internal drop off, i.e. all responders answered all items, and the answers were spread over the alternatives of the Likert scale on most items. After a few modifications the questionnaire was finalized\[[@B15]\]. In the graphs, the GP\'s answers are combined to a three-point scale after merging alternatives 1 and 2 and alternatives 4 and 5 of the original five-point Likert scale. Performance ----------- The survey was carried out in spring 2001. The questionnaires, including a return envelope, were mailed from the NEPI Foundation, a non-profit organisation for studies on epidemiology of drugs, to all the 617 GPs working in the selected municipalities. No incentive for participation was offered. Two reminders were sent if no answer was received, the second of which included a new questionnaire. After the data collection, the work places of non-respondents were contacted to ascertain whether or not these GPs were still employed at the time of the survey. The present analysis of data was not reported to the Ethics Committee because our earlier studies had been approved on the administrative level alone. Results ======= After the second reminder 317 of the 617 mailed questionnaires were returned. Among non-respondents 82 had left their work places leaving 535 GPs eligible for the survey. The response rate of the eligible participants was 59.4 percent: 56 percent among the 339 men and 65 percent among the 196 women. Among respondents, the average age was 48.7 years and they had worked as GPs for 12.7 years on average. Sixty percent were male. Among non-respondents available for the survey, the average age was 59.0 years and 69 percent were male. To work with depressive patients was perceived a more positive than negative task by 69 per cent of the respondents, \"neither positive, nor negative\" by 27 per cent and mainly negative by four per cent of the respondents. When deciding treatment for the depressive patient, nearly all GPs considered that their own clinical experiences of the treatment had great importance, and, for a majority, also patients\' own preferences and clinical guidelines to treatment were of major value (Fig. [1](#F1){ref-type="fig"}). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Factors of importance for decision about antidepressive treatment.Mark which importance you attach to each of the mentioned factors when you decide about antidepressive treatment: a. The patient\'s own attitude towards different forms of treatment; b. Guidelines for treatment; c. Your previous clinical experience of the treatment. ::: ![](1471-2296-6-21-1) ::: Treatment of moderate depression includes mainly antidepressant medication and psychotherapy. A majority of the respondents agreed that medication with ADs is effective, and did not assume that psychotherapy can replace drug treatment. A majority agreed, fully or partially, that psychotherapy is needed as a complement to drug treatment (Fig. [2](#F2){ref-type="fig"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Psychotherapy and drug treatment.Give for each statement your degree of agreement: a. Psychotherapy is needed as a compliment to drug treatment in most cases. b. Drug treatment with antidepressive agents (ADs) may be replaced by psychotherapy, e.g. cognitive therapy, if the patient is prepared. c. Drug treatment with ADs is effective and sufficient in most cases. ::: ![](1471-2296-6-21-2) ::: Most responders had great confidence in ADs for treatment of panic disorder and thought they are effective also in obsessive-compulsive disorders and social phobias (Fig. [3](#F3){ref-type="fig"}), but the majority of the responders showed a limited reliance on ADs\' effectiveness to treat sleep disorders, psychosomatic illness, chronic pain, and premenstrual syndrome. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Use of antidepressive agents in other conditions than depression.Antidepressive agents can be used in other conditions than depression. Give for each statement your degree of agreement: a. ADs are effective in panic disorder. b. ADs are effective in obsessive-compulsive disorder. c. ADs are effective in social phobia. d. ADs are effective against insomnia. ::: ![](1471-2296-6-21-3) ::: Two thirds of the responding GPs considered ADs as effective for the treatment of old people as for people at younger ages. Their confidence in psychotherapy for old patients was low. The responders\' judgments on the statement that ADs are effective for the treatment of anxiety and behavioural disturbances in patients with dementia were distributed evenly from agree not at all to completely agreement (Fig. [4](#F4){ref-type="fig"}). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Depressive disorders in old people.Here follow some statements on depressive disorders and their treatment in old people. Mark for each statement which degree of agreement you have: a. Depressive disorders in old people have more of biological causes then depressive disorders in younger ages. b. ADs are as effective in old people as in middle age people. c. Psychotherapy is as effective in old people as in middle age persons. d. ADs are effective against anxiety and behaviour disorder in people with dementia. ::: ![](1471-2296-6-21-4) ::: Table [1](#T1){ref-type="table"} displays factors, which, according to the GPs, may have influenced their manner of working with depressed patients. Almost all regarded individual experiences from family medicine as very important, and a large majority also regarded experiences from private life to be of significant importance. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Factors that may have influenced on GPs\' manner of working with depressed patients. Share of respondents that have marked \"great\" or \"very great\" importance on the factor in question. ::: ------------------------------------------ ------ Experiences from general practice 96 % Experiences from private life 83 % Residency in psychiatry 57 % Professional reading 50 % Education on communication 50 % A mentor colleague 26 % Reading fiction literature 20 % Religious faith 17 % University education 12 % Popular reading on psychiatry/psychology 10 % ------------------------------------------ ------ ::: Some issues concerned continuing medical education (CME). Much the same number of GPs considered non-commercial courses to be of great importance as courses arranged by pharmaceutical companies. The respondents regularly took part in more lectures arranged by drug companies than in lectures without involvement of commercial interests (Table [2](#T2){ref-type="table"}). Commercial courses were generally regarded as meeting high standards. More than one third of the respondents thought they had received more knowledge from such courses than from commercially independent education. A majority of the respondents agreed partly or completely that commercially sponsored courses may influence physicians in an inappropriate way, but many GPs did not recognise that risk (Fig. [5](#F5){ref-type="fig"}). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Respondents\' reports of occasions of participation in education/lectures each year. Number of respondents. ::: Not regularly 1 -- 3 /year 4 -- 7/year \>7/year ------------------------------------------- --------------- -------------- ------------- ---------- Arrangements without industry involvement 74 127 71 43 Arranged by pharmaceutical industries 30 141 81 62 ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Pharmaceutical companies\' information and courses on depression.Here follow five statements about the pharmaceutical companies\' information and courses on depression and antidepressive agents. Mark for each statement your degree of agreement: a. The company representants\' information on antidepressive agents keeps high standards. b. Their courses on depression and depression treatment keep high standards. c. Their information and education on treatment of depression have given me more knowledge than commercially independent education. d. The pharmaceutical companies\' information and education pose a risk of non-appropriate influence in the selection of treatment. ::: ![](1471-2296-6-21-5) ::: Most respondents stated that they had good possibilities to discuss with their colleague GPs. In contrast, a majority claimed that psychiatrists do not provide them with the necessary support for patient care. Discussion ========== Our purpose was to investigate conceptions concerning depressive disorders among Swedish GPs. We found that GPs regarded experiences from family medicine and from private life more important than university and post-graduate education as influences on their work with depressive patients. Most GPs had great confidence in ADs for treatment of depressive disorders and panic disorder and thought that the drugs are effective also in obsessive-compulsive disorders and social phobias. A majority of GPs also agree that commercially sponsored courses may influence physicians in an inappropriate way, but many of them do not recognise that risk. Method considerations --------------------- For the purpose of an additional study utilizing the same data collection, the sampling procedure was not randomised but stratified. The participants were all GPs who worked in the selected municipalities. They constituted groups of GPs with different levels of prescribing ADs and were from all parts of Sweden. We assume the procedure resulted in an appropriate sample of Swedish GPs. A response rate of 59 percent among eligible GPs is well in line with comparable surveys among physicians, although such a rate means that the possible impact of selection bias must be considered \[[@B16]-[@B20]\]. However, there are solid arguments that the result of the questionnaire can be viewed as representative even if the response rate is moderate, when a homogenous professional group is surveyed on issues of central professional importance \[[@B22]-[@B24]\]. We assume that the results are representative of all the selected GPs even if there were minor differences in gender distribution and age between responders and non-responders. The questionnaire was not validated against any earlier instrument, but the items of the questionnaire were extracted from findings of the previous interview studies among GPs, a procedure that should ensure the relevance of the items \[[@B12],[@B13]\]. The questionnaire was tested by a pilot study that had a response rate of 75 per cent and an extensive spread of responses over the alternatives of the Likert scales. In the present study there were a few internal drop-offs, i.e. some responders did not answer all items. Altogether, we claim that the present study has acceptable representativeness and validity. Interpretation of the findings ------------------------------ The respondents opinions of the usefulness of ADs in treatment of moderate depression and anxiety disorders were in accordance with findings in our previous interview study and with actual recommendations\[[@B12],[@B25],[@B26]\]. Psychotherapy was considered most often as a complement to drugs in the treatment of depression. In the previous interview study, the majority of the interviewees presented a positive attitude to psychotherapy\[[@B12]\]. There is now good evidence for the effectiveness of several forms of psychotherapy against moderate depressions \[[@B27]-[@B34]\]. These findings appear not to have become generally accepted by Swedish GPs, probably partly due to the limited access to psychotherapists. A clear majority of the GPs in the present study perceived the work with depressed patients in a predominantly positive way. A notable finding was the great importance they attached to their own experiences from family medicine and from private life. That was valid for both treatment decisions and for their manner of working, and the same came out in the interview study\[[@B13]\]. Judging from these responses, GPs try to consider treatment guidelines as well as patients\' attitudes to treatment and their individual experiences of similar situations when they propose treatments. In such complex decision situations, the physicians seem to perceive their individual experiences as indispensable. That is in accordance with Schön, who reported that practitioners gather repertoires of understandings and actions from their experiences, and that they use their repertoires to guide their professional actions. The quality of this guidance varies. A practitioner must have worked through his/her experiences by discussions and reflection if his/her repertoire should serve as a valuable guide\[[@B35]\]. To a substantial degree, the outcome in drug treatment of depression is dependent on the therapeutic alliance between the physician and the patient\[[@B36]\]. This alliance is built up by the physician\'s ability to communicate, which is acquired mainly through individual experience, guiding and reflective reading. When professionals rely primarily on personal experiences as grounds for clinical assessment and intervention, they demonstrate preference for \"knowledge in action\" over inquiring scientific knowledge, and, tentatively, a scepticism regarding the usefulness of scientific literature\[[@B37]\]. Many respondents\' low or moderate evaluation of their undergraduate education indicates such scepticism. Lennarson Greer writes about \"the problem of uncertainty\" as a key problem of the medical profession\[[@B38]\]. For many physicians, trust in personal experiences and collegial support appear to be better ways to handle the worry of uncertainty than reading scientific literature\[[@B38],[@B39]\]. Thus, the respondents\' emphasis on their own experiences is understandable, but, on the other side, a low evaluation of scientific and evidence-based knowledge may imply insufficient capability to form an accurate opinion of a patient\'s state and to propose an optimal treatment. Even if the respondents ranked their own experiences of clinical work and private life as most important, they also indicated that lectures and professional reading are important sources of knowledge about depression. Further, the striking increase of prescriptions of ADs in the 1990s indicates that recognition and management of depressive disorders have changed\[[@B40],[@B41]\]. Reasonably, that is due to other influences than GPs\' own experience. Psychiatric specialists have probably had an impact on GPs\' prescribing, partly directly by initiating or recommending treatment for patients, partly indirectly by education and informal consultations. The pharmaceutical companies\' marketing of ADs is another powerful incentive to increase the use of ADs. In our previous study we learned that high sales levels correlated with more participation in the pharmaceutical companies\' activities\[[@B14]\]. Also other studies have documented the impact of commercial marketing on prescribing and professional behaviour of physicians \[[@B42]-[@B45]\]. GPs stand out as a cultural subgroup, where ongoing educational, collegial and marketing influences are obvious from outside but not quite clear for the members themselves. Many GPs appear to overlook these external influences on their thinking and professional behaviour, while they tend to over-emphasize individual experiences as a base of their clinical work. To some extent, GPs\' disappointment with psychiatric services may depend on unrealistic ideas about psychiatrists\' ability to solve GPs\' problems. It has been pointed out that GPs and psychiatrists have different preconditions in their clinical work, and that they meet different kinds of depressed patients\[[@B3],[@B46]\]. Projects to improve depression management have reported that simple guideline implementation and educational strategies were generally ineffective. In short term evaluation, effective strategies were those with broad interventions that incorporated clinical education, enhanced team work and a greater integration between primary care and psychiatry \[[@B46]-[@B50]\]. Conclusion ========== In their conceptions of factors that form their way to treat depressed patients, GPs tend to emphasize experiences, both from clinical work and private life, and overlook influences of collegial dealings and ongoing CME as well as the effects of the pharmaceutical companies\' marketing activities. Many GPs appear to need more evidence based knowledge about depressive disorders. Interventions to improve depression management have to be supporting and interactive and combined with organisational reforms to improve co-operation with psychiatrists. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= The authors have planned and carried out the study together. SJA wrote the first draft of the manuscript. GL organised data and constructed the graphs. MT was active in planning the study, organising the data and the linguistic formulation of the article. The authors revised the manuscript together. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2296/6/21/prepub> Acknowledgements ================ The authors thank the NEPI foundation for administrative support to the study.	PubMed Central	2024-06-05T03:55:59.601898	2005-5-16	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180707/", "journal": "BMC Fam Pract. 2005 May 16; 6:21", "authors": [ { "first": "Stig J", "last": "Andersson" }, { "first": "Margareta", "last": "Troein" }, { "first": "Gunnar", "last": "Lindberg" } ] }
PMC1180708	Background ========== Hereditary hemochromatosis (HH) is a common genetic disorder of iron metabolism that is usually inherited in an autosomal recessive pattern and associated with missense mutations in the HFEgene. This pathology displays a large genetic heterogeneity because several other types of hemochromatosis, associated with different genes and patterns of inheritance, have been reported \[[@B1]\]: HH type 2B is a juvenile form linked to the HAMPgene (encoding for hepcidin) \[[@B2]-[@B4]\], HH type 3 is linked to the TfR2gene (encoding for transferrin receptor 2) \[[@B5]-[@B7]\], HH type 4 is linked to the SLC11A3gene (encoding for ferroportin, an intestinal iron transporter) \[[@B8],[@B9]\]. and HH type 5 is linked to a gene encoding subunit H of ferritin \[[@B10]\]. Moreover, the gene responsible for juvenile hemochromatosis (HH type 2A), and whose protein product is called hemojuvelin, has recently been cloned \[[@B11]\]. The main form of HH (i.e. type I which is linked to the HFEgene) occurs predominantly in Northern European populations, with a prevalence of approximately 3 to 8 in 1000 \[[@B12]-[@B16]\]. It is characterised by excessive iron absorption, which progressively leads to the destruction of tissues in different organs of the body. After a phase of latency, the first signs of biochemical expression appear, generally around the age of 20. This is characterised by increases in serum iron parameters (transferrin saturation, ferritin). The clinical expression manifests later during adulthood, generally around the age of 40 in males and later in females, around the age of 50, because of the protective effects of menstrual blood loss and pregnancies \[[@B17]-[@B19]\]. The clinical picture may include at an early stage, non-specific symptoms such as persistent fatigue and arthralgias, and at a later stage, clinical signs such as skin pigmentation, hepatomegaly, arthropathy, cardiomyopathy, diabetes and cirrhosis \[[@B20]-[@B22]\]. Classically, this clinical expression occurs more frequently in males than in females (sex ratio of 3:1) \[[@B23]\]. HH can be treated or prevented by periodic phlebotomies. This simple and efficient treatment prevents iron accumulation and clinical complications. Without this early treatment, the disease may progress towards irreversible damage such as cirrhosis and hepatocellular carcinoma \[[@B17]-[@B19]\]. A candidate gene for HH type 1, HFE, was identified in 1996 on chromosome 6 and encodes the HFE protein, a transmembrane glycoprotein that is implicated in modulation of iron uptake \[[@B24],[@B25]\]. Currently, about twenty different mutations have been identified in this gene worldwide and, one of them, termed C282Y, is present at homozygous state in 80 to 95% of HH patients. This molecular anomaly corresponds to the substitution of a tyrosine for a cysteine at amino acid 282, which prevents formation of a disulfide bond \[[@B24],[@B26]\]. The two other most common mutations of the HFEgene are associated with milder forms of HH (H63D and S65C) \[[@B1],[@B21],[@B27]-[@B29]\]. The discovery of the HFEgene in 1996 and the fact that one of its mutations (C282Y) is responsible for the large majority of HH cases enabled the implementation of efficient strategies for molecular diagnosis \[[@B30]\], what has altered the way in which HH is diagnosed. Initially, the diagnosis relied on a high index of suspicion associated with evidence of elevated iron parameter values \[[@B18],[@B20],[@B22]\]. Following discovery of the HFEgene, a DNA test was proposed to confirm the diagnosis of HH. Such a test, which allows the detection of at least the C282Y mutation, is now widely available. This discovery has simplified the diagnostic strategy and enabled pre-symptomatic or earlier diagnosis in some patients. If phlebotomy treatment is implemented before the appearance of irreversible damage, the excess iron can be removed and patients have a normal life expectancy \[[@B20],[@B31]\]. In this study, we assessed the impact of HFEgenetic testing on the clinical presentation and epidemiology of HH in a cohort of 415 patients homozygous for the C282Y mutation who were followed in a blood centre in western Brittany, France. This report contains objective data to measure this impact. Methods ======= Study population ---------------- The present study was conducted in Brittany, a region of nearly three million inhabitants located in the north-western part of France, where HH is particularly frequent (carrier rate: 1 in 7) \[[@B32]\]. This disease presents a large genetic and allelic heterogeneity, but as the majority of patients are homozygous for the C282Y mutation, we decided to include in this study only the patients homozygous for this mutation. This study included all the C282Y homozygous patients who are or were included in a phlebotomy program in a blood centre of western Brittany and who presented with transferrin saturations of greater than or equal to 45%. The first patients began their treatment in the early eighties and the last date of patient entry for inclusion in this study was December 31^st^2003. The proportion of homozygous C282Y patients before and after the implementation of the genetic test did not change significantly. Clinical questionnaire ---------------------- A detailed clinical questionnaire was completed during the clinical exam performed at the first visit of patients to the blood centre (i.e. at entry in the phlebotomy program). Information contained in this questionnaire was previously described in detail \[[@B33]\]. Briefly, it provided information regarding socio-demographic characteristics of patients, their age at diagnosis, the circumstances of HH discovery, the biochemical parameters and the clinical signs associated at the time of diagnosis. This questionnaire also included data related to the treatment, such as the number and quantity of phlebotomies needed to reach depletion and the quantity of iron extracted. The date of the beginning of the treatment (i.e.entry into the phlebotomy program) was also documented. This date was used to determine if the treatment of patients began before or after the availability of the genetic test (i.e.1996). The intake of alcohol was assessed by a detailed item included in the questionnaire, which measured the number of glasses of alcohol drunk each day (including glasses of wine, beer and liquor). These data enabled the quantity of ethanol (in grams) consumed each day, by each patient of the cohort, to be determined. Excessive alcohol consumption was defined as a daily consumption greater than or equal to 60 grams of ethanol \[[@B33]\]. Determination of biochemical parameter levels and of HFEgenotype ------------------------------------------------------------------ Serum iron parameters (i.e.serum iron, serum ferritin and serum transferrin saturation) were determined by standard biochemical methods (including collection of serum after a 12-hour fast, confirmation by at least two measurements). Analysis of the C282Y mutation relies on amplification of the specific gene region by the polymerase chain reaction, followed by mutation detection using restriction enzymes \[[@B29],[@B32]\]. Recently, another method was adopted for this analysis: the denaturing high-performance liquid chromatography method \[[@B34]\]. If the DNA test confirms the presence of mutations in a patient, family testing is offered to the relatives of this newly diagnosed patient. Family testing combines the collection of biochemical and clinical evidence of iron overload for the patients relatives with analysis of the main HFEmutations. Before 1996, family testing was already possible by analysis of HLA haplotypes in families \[[@B35]\] and this was commonly practised in our region where the disease incidence is high. The genotype of the patients diagnosed before 1996 was retrospectively determined when the genetic test became available. Statistical analysis -------------------- Data were analysed using Epi-Info software (version 6.04; Centers for Disease Control and Prevention, Atlanta, Georgia) and SAS statistical package (version 8.2; SAS Institute). Quantitative variables, expressed as means and standard deviations, were compared with the Student\'s ttest or ANOVA, whereas qualitative variables, expressed as percentages, were compared with the Chi square test or the Fisher\'s exact test (in case of small sample size). A logarithmic transformation was performed for the serum ferritin variable which had a skewed distribution. A significance level of 5% was used for all of the analyses which were performed two-sided. The analyses consisted of determining the influence of the implementation of the genetic test on the sex ratio, the age at diagnosis, the circumstances of HH discovery, the levels of biochemical parameters and the frequency of clinical signs associated at the time of diagnosis. The comparisons of biochemical and clinical data, which were made separately for males and females, were adjusted for age at diagnosis and for alcohol consumption, because we showed previously that excessive alcohol consumption (\>60 g/day) increased HH expressivity \[[@B33]\]. The relation between the time of diagnosis (i.e. before or after availability of HFEgenotyping) and each of the different clinical symptoms was assessed by calculating the odds-ratio (OR) and its 95% confidence interval (CI). This study complied with French bioethical regulations. Informed consent of patients was obtained before blood samples were taken. Results ======= Description of the study population ----------------------------------- This study included 415 C282Y homozygous patients of whom 56.9% were male, resulting in a sex ratio of 1.3:1 (236/179). The age at diagnosis, ranging from 13 to 76 years, was significantly higher in females than in males (48.5 yrs (σ = 14.3) vs. 46.1 yrs (σ = 12.5), p = 0.037). Overall, the diagnosis was mainly made on the basis of clinical features (61.4%) or through family testing (30.9%). The circumstances of diagnosis remained unknown for one patient. Among these 415 patients, 21.9% began their phlebotomy program before the implementation of the molecular testing (n = 91, 72 males and 19 females) and 78.1% after this date (n = 324, 164 males and 160 females). Characteristics of patients before and after the implementation of the genetic test ----------------------------------------------------------------------------------- ### Socio-demographic characteristics of patients in relation to the type of diagnosis (based on HFEgenotyping or not) A significant decrease in the sex ratio was noted following development of the genetic test, as illustrated in figure [1](#F1){ref-type="fig"}. The sex ratio was 3.79 (72/19) before discovery of the gene and 1.03 (164/160) after this date (p \< 10^-5^), meaning that the proportion of females diagnosed since 1996 increased greatly: from 20.9% (19/91) to 49.4% (160/324). These females seemed to be diagnosed earlier. Their age at diagnosis tended to decrease from 52.9 yrs (σ = 9.8) before the implementation of the genetic test to 47.9 yrs (σ = 14.7) after this date. Nevertheless, this difference was not significant because a small number of women was diagnosed before 1996 (n = 19, p = 0.210). This pattern was not observed in males, their age at diagnosis increased from 42.7 yrs (σ = 10.4) to 47.6 yrs (σ = 13.1) after 1996 (p = 0.007). After introduction of molecular testing, the symptom of unexplained and persistent fatigue was more commonly present at diagnosis of HH. The frequency of this symptom increased from 51.2 to 68.0% following introduction of the DNA-based testing (OR = 2.03, 95% CI: 1.21, 3.41; p = 0.004), notably in males (from 47.1 to 58.4%). On the other hand, the proportion of patients detected by family testing was not significantly changed after 1996 (31.3 vs. 29.7%). ### Biochemical parameters in relation to the type of diagnosis (based on HFEgenotyping or not) Differences were also observed in biochemical parameter values. The results of the comparison of the biochemical data of patients before and after the development of the genetic test are presented in table [1](#T1){ref-type="table"}. Serum ferritin and serum iron values were significantly lower in patients diagnosed after implementation of the genetic test (serum ferritin: 810.4 vs. 1788.8 μg/liter, p \< 0.001; serum iron: 34.9 vs. 39.8 μmol/liter, p \< 0.001). These results were observed in males (serum ferritin: 1136.9 vs. 1886.6 μg/liter, p \< 0.001; serum iron: 36.7 vs. 40.7 μmol/liter, p \< 0.001) and in females (serum ferritin: 475.7 vs. 1418.0 μg/liter, p \< 0.001; serum iron: 33.1 vs. 36.6 μmol/liter, p = 0.052). As shown in table [1](#T1){ref-type="table"}, the results remained unchanged after adjustment for age at diagnosis and alcohol consumption. In this cohort, 8.0% of the patients declared having excessive alcohol consumption (≥ 60 g/day; n = 33) comprising 9.9% of the patients diagnosed before 1996 (n = 9) and 7.4% of those diagnosed after 1996 (n = 24). A decrease was not observed for the third iron parameter, transferrin saturation (79.5 vs. 79.4%, p = 0.930), for which an elevated value (\>45%) was used as a selection criterion for this study. ### Clinical signs in relation to type of diagnosis (based on HFEgenotyping or not) The frequency of the main clinical signs observed at the time of diagnosis before and after the availability of the molecular testing is shown in figures [2A](#F2){ref-type="fig"} and [2B](#F2){ref-type="fig"}. The clinical signs were less frequent in the subjects diagnosed after the development of the genetic test, particularly skin pigmentation (20.1 vs. 40.4%; OR = 0.37, 95% CI = 0.22, 0.63; p \< 0.001) and hepatomegaly (11.0 vs. 22.7%; OR = 0.42, 95% CI = 0.21, 0.83; p = 0.006). This change was more significant in women. Indeed, only the two signs mentioned above tended to be less frequently observed in men following the introduction of genetic testing although these changes did not reached statistical significance (skin pigmentation: 32.1 vs. 43.7%; OR = 0.61, 95% CI = 0.33, 1.13; p = 0.089 -- hepatomegaly: 14.7 vs. 22.5%; OR = 0.59, 95% CI = 0.27, 1.32; p = 0.160), whereas in women, four symptoms were significantly less frequent following implementation of molecular testing: skin pigmentation (7.6 vs. 27.8%; OR = 0.22, 95% CI = 0.06, 0.84; p = 0.006), hepatomegaly (7.3 vs. 23.5%; OR = 0.26, 95% CI = 0.06, 0.86; p = 0.028), arthritis (47.5 vs. 76.5%; OR = 0.28, 95% CI = 0.07, 0.98; p = 0.023) and diabetes (2.0 vs. 16.7%; OR = 0.10, 95% CI = 0.01, 0.71; p = 0.001). These results were similar in the sub-group of patients having no excessive alcohol consumption (data not shown). Discussion ========== The discovery of the HFEgene in 1996 constituted a considerable advance in the medical and scientific field. This discovery concerned one of the most common inherited disorders in white populations, HH -- a disorder that was complex to diagnose but for which a treatment existed -- and identified one of the few undiscovered genes that has an important impact on public health. The symptomatology of HH has evolved over the past years and it is now rare to diagnose severe forms of the disease, associated with diabetes, cirrhosis and darkened skin \[[@B36],[@B37]\]. Through a survival analysis based on a cohort of 251 patients diagnosed between 1947 and 1991, Niederau et al.showed that the percentage of patients with early diagnoses increased 3-fold during the period of 1970--1981 compared to the period of 1947--1969, and that there was a further 20--25% increase in the early diagnosis rate during the period of 1981--1991 \[[@B37]\]. These changes occurred before the discovery of the HFEgene, and were probably the consequences of improved education of physicians and the implementation of HLA testing for family members of probands. The current study highlights the importance of the discovery of the HFEgene in 1996 and demonstrates how the clinical presentation and epidemiology of HH have changed since the availability of the DNA test. Our results objectively measure these changes, and show that the sex ratio of this disease has altered: the proportion of females currently diagnosed has increased and has reached that of males. This study also highlights that the profile of HH patients has changed: the patients have lower iron parameter values (serum ferritin and iron) and a lower frequency of clinical signs at the time of diagnosis, notably skin pigmentation and hepatomegaly. This change is more pronounced in females in whom clinical manifestations of HH appears later than in males (around the age of 50 versus around the age of 40 in males). This study included all the C282Y homozygous patients who are or were included in a phlebotomy program in a blood centre of western Brittany. For the patients diagnosed before the implementation of the genetic test, the genotype was retrospectively determined in 1996 if they were still alive at this date. Consequently, the patients who died before 1996 were not genotyped and not included in this study. With this bias, some severe cases of the disease have been missed and the difference between the two groups should therefore be even greater than reported here. Identification of the HFEgene and of its main mutation (C282Y) has greatly simplified diagnosis of, and family testing for, HH \[[@B20],[@B31]\]. The fact that homozygosity for the C282Y mutation is responsible for the majority of HH cases has enabled use of the molecular test for this mutation as a diagnostic criterion for HH. Before the genetic test was available, diagnosis of HH required a high index of suspicion (as the clinical signs are non-specific) and evidence of elevated iron parameters \[[@B18],[@B20],[@B22]\]. Traditionally, diagnosis was based on the measurement of transferrin saturation. A liver biopsy then enabled confirmation of iron overload by detection of elevated hepatic iron levels \[[@B27]\]. The discovery of the HFEgene enabled molecular analysis to be included in the diagnostic strategy and thus genetic testing for confirmation of the diagnosis was proposed. In this way, HH could easily be differentiated from all other types of iron overload. Currently, the diagnosis combines molecular testing with traditional biochemical methods. The diagnostic strategy is as follows: 1) To suspect the diagnosis from non-specific symptoms (such as persistent fatigue, arthralgias), and not only when presented with classical signs of HH (such as skin pigmentation, diabetes and cirrhosis); 2) Once the disease is suspected, the second step is to determine the serum transferrin saturation; 3) If the value of this iron parameter is elevated, molecular analysis of the main HFEmutations (C282Y +/- H63D) must be done to confirm the diagnosis of HH \[[@B20]\]. The diagnostic strategy has changed, and as a consequence, patients with increased iron parameter values and a genotype of HH are now diagnosed as having HH. The molecular basis of the disease has been evidenced and inclusion of genetics in the diagnostic strategy has enabled detection of iron overload that is expressed only at a biochemical level. The HFEgene discovery has improved our knowledge of this complex disease. It has enabled the genotypes of patients to be determined, and by considering this information in relation to other factors such as age, gender and environment, elucidation of genotype/phenotype correlations has begun \[[@B33],[@B38],[@B39]\]. The HFEgene discovery also raised the complex issue of the penetrance, which is clearly incomplete \[[@B22],[@B40]-[@B44]\]. It is probable that some of the patients who exhibit biochemical evidence of iron overload and a genotype of HH would never progress towards the clinical manifestations of HH. Two studies reported that less than 1% of the C282Y homozygous subjects develops clinical hemochromatosis \[[@B40],[@B45]\]. Unfortunately, studies on penetrance generally suffer from bias that results in under or over-estimation of the frequency of the disease \[[@B22],[@B46]\]. Until more data are available on the penetrance of the C282Y homozygous state, screening using HFEgenotyping remains controversial \[[@B27],[@B47]\]. Nevertheless, looking beyond this complex issue of penetrance, the gene discovery has led to a better understanding of some of the phenotypic variability observed in HH. This improved knowledge has been conducive to better medical education of physicians, such that they now may more often suspect a diagnosis of HH when presented with non-specific symptoms (such as unexplained and persistent fatigue) than they did previously. This education is certainly not perfect at this time but we can observe in the present study that the proportion of patients, particularly males, diagnosed with the symptom of fatigue has already increased since the availability of HFEgenotyping. With astute clinical assessment and HFEgenetic testing, patients can be diagnosed and treated before the appearance of irreversible damage, and this therefore avoids development of severe forms of HH. In our study, an increase in the age at diagnosis after the introduction of the DNA test was observed in men. This could be explained by the fact that a diagnosis of HH has been done in some men older than 65 presenting with fatigue, arthralgia and a discrete ferritin elevation. Those C282Y homozygous patients would probably never have been diagnosed ten years ago. The inclusion of these old diagnosed C282Y homozygous men in our cohort significantly increased the age at diagnosis of HH during the last years. Family testing performed among the relatives of a newly diagnosed patient also enables detection of subjects in the pre-symptomatic phase. The discovery of the HFEgene has not significantly altered family testing for HH because, prior to 1996 it was already possible to analyze the transmission of HLA haplotypes in families (the HFEgene is located near the HLA complex). Such testing was commonly practiced in our region where HH is common. In our study, the proportion of patients detected by family testing before and after the introduction of HFEgenotyping was similar (29.7% versus 31.3%). Consequently, the inclusion, in the present study, of patients identified through family testing did not alter our findings. The impact of the HFEgene test on the identification of HH through family testing is expected to be higher in other regions where family testing was not practiced as systematically as in our region prior to 1996. This pre-symptomatic or early diagnosis of HH and follow-up phlebotomy treatment should prove efficacious in preventing organ damage and therefore aid in achieving normal life expectancy for patients \[[@B48]\]. Early detection can completely prevent premature death caused by HH. Illustrating this point, Milman et al.found, through analysis of a cohort of patients diagnosed in Denmark between 1945 and 1985, that the survival of HH patients without cirrhosis or diabetes mellitus was similar to that in the general population \[[@B48]\]. Conclusion ========== In conclusion, HFEmutation testing has supplemented the determination of serum iron parameters as a criterion for the diagnosis of HH. This has increased the proportion of women diagnosed with HH and has decreased the frequency of certain clinical signs at diagnosis. The method of diagnosis of HH has changed and this has contributed to modify the epidemiology of this disease, with the sex ratio reduced to close to 1.0 and a weaker clinical expression than observed previously. This study highlights an example of the progress enabled by Genomic Medicine \[[@B49]\] and shows that knowledge of the molecular basis of a disease (following identification of its gene and mutations involved) can lead to a change in the epidemiology of that disease. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= VS contributed to the conception and design of the work, analysed the data and wrote the paper. GLG and CM were involved in genetic analysis and revised the paper. MCM, AYM, BC and JBN helped in the acquisition of data. CF contributed to the conception and design of the work and supervised the study. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2350/6/24/prepub> Acknowledgements ================ The authors thank the reviewer Robert Britton for his helpful comments that improve the manuscript. This work was supported by grants from the Projet Hospitalier de Recherche Clinique \"Mise en place d\'un diagnostic phénotypique des surcharges en fer primaire\" and the Etablissement Français du Sang. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Evolution of the sex ratio: number of males and females diagnosed before and after the implementation of the genetic test. ::: ![](1471-2350-6-24-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### (A) Frequency of the main clinical signs and symptoms at the time of diagnosis in males, before and after the implementation of the genetic test. (B) Frequency of the main clinical signs and symptoms at the time of diagnosis in females, before and after the implementation of the genetic test ::: ![](1471-2350-6-24-2) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Biochemical characteristics of the C282Y homozygous patients before and after the implementation of the genetic test. ::: *Diagnosis based on HFEgenotyping* *Crude p-value* *Adjusted\ p-value ------------------------------ ---------------------------------------- --------------------- -------------------------- --------- ------------- --------- No. of patients 91 324 Gender (no. and % of males) 72 (79.1%) 164 (50.6%) p \< 10^-5^ \- Males Serum ferritin (μg/L) 1886.6 (1769.7) 1136.9 (951.0) \<0.001 \<0.001 Serum iron (μmol/L) 40.7 (6.9) 36.7 (6.7) \<0.001 \<0.001 Transferrin saturation (%) 79.9 (11.9) 83.0 (13.2) 0.090 0.134 Females** Serum ferritin (μg/L) 1418.0 (1457.9) 475.7 (744.6) \<0.001 \<0.001 Serum iron (μmol/L) 36.6 (8.2) 33.1 (7.4) 0.052 0.029 Transferrin saturation (%) 77.2 (13.4) 75.9 (14.4) 0.710 0.714 \* Adjusted for age at diagnosis and alcohol consumption :::	PubMed Central	2024-06-05T03:55:59.603851	2005-6-1	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180708/", "journal": "BMC Med Genet. 2005 Jun 1; 6:24", "authors": [ { "first": "Virginie", "last": "Scotet" }, { "first": "Gérald", "last": "Le Gac" }, { "first": "Marie-Christine", "last": "Mérour" }, { "first": "Anne-Yvonne", "last": "Mercier" }, { "first": "Brigitte", "last": "Chanu" }, { "first": "Chandran", "last": "Ka" }, { "first": "Catherine", "last": "Mura" }, { "first": "Jean-Baptiste", "last": "Nousbaum" }, { "first": "Claude", "last": "Férec" } ] }
PMC1180817	Background ========== Treatment of acute psychiatric illness in resident settings is expensive, albeit less so than in intensive medical care units \[[@B1]-[@B3]\]. The costs are to a great extent indispensable, but not of the same magnitude in all elements of the treatment chain. Allocation of patients in a treatment and cost-efficient way along this chain is a logistic challenge. Regression-based cost functions have been used to illustrate patient and system related cost elements\[[@B4]\]. The possibility of violence in acute psychosis is an important contributor to the costs\[[@B5]\]. Psychiatric acute wards in Norway are obliged by law, the Norwegian Act on Compulsive Psychiatric Care (ACPC), to accept persons who, after an examination by an external doctor are found to be in danger of severely damaging own life or other people\'s lives. Reasons for referrals are acute psychosis, mania, or severe suicidal conditions. A person may be referred voluntarily to a mental hospital, or for compulsory observation of up to 10 days or for compulsory treatment for a prolonged period of time. Not later than 24 hours after admission the consultant psychiatrist on duty has to make a legally binding decision on admission status. The patient or his relatives may appeal this decision to a legal body outside the hospital. The acute wards are often the first step in a chain of facilities that the patient may need in order to regain his functional ability. Such secondary facilities, sub-acute/intermediate wards, long term treatment, half-way houses or nursing care homes are, however, not obliged to accept patients at demand, but rather as empty places/beds become available. The result may be crowded acute wards. Either because too many patients are referred to the wards per time unit or because patients do not get another place to stay, if they are too sick to be transferred to community services of ambulatory type. Shortened duration of resident stays might be cost-effective treatment, although the clinical outcome may be variable. High rates of relapse may be counterproductive. In a Norwegian study relapses were shown not to be an indicator of efficiency, but rather of logistic planning of mental health services\[[@B6]\]. Waiting time whilst in resident acute care has not, to our knowledge, been studied from a clinical and economic standpoint using cross-sectional data. The decision on length of stay was taken on daily meetings of the treatment staff, where all the doctors, psychologist and social workers were present. The group of senior psychiatrists decided where the patient should be referred after acute care treatment. If no resident place was available at the desired time, the patient had to wait in the acute ward for such a slot. The aim of the study was thus to present a novel way of calculating the partial cost efficiency of waiting time in one of four acute wards in the city of Oslo (ca 500.000 inhabitants). Methods ======= All resident patients (N = 23) on a randomly chosen day in March 2002 were included. Day of entry, which was different from the chosen day, legal admission status, sex, age and number of previous resident stays were noted, see table [1](#T1){ref-type="table"} and figure [1](#F1){ref-type="fig"}. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Socioeconomic data and level of compulsion according to ACPC\* by entry for 23 patients who all were inpatients on a random chosen day in 2003 in the acute wards of a psychiatric hospital, (Standard deviation). ::: Waiting for further treatment Not waiting --------------------------------------------------------------------- --------------------------------- --------------- Men 7 5 Women 4 7 Mean number of earlier referrals 3.6 (3.2) 5.7 (6.2) Mean age 34.4 (10.4) 36.8 (8.6) Psychotic illness 11 (47.8%) 8 (34.8%) §according to Law: §2-1 (voluntarily) 1 3 §3-6 (compulsory observation) 5 3 §3-3 (compulsion) 5 6 §as decided by senior psychiatrist within 24 hours after admission: §2-1 1 3 §3-6 3 3 §3-3 7 6 \*) ACPC = The Norwegian Act on Compulsory Psychiatric Care of 1999. ::: ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Number of resident days for 11 out of 23 patients in an acute ward before a decision was made for further intramural treatment in a less costly facility, and number of waiting days after the decision day. Twelve patients did not wait for other intramural treatment, and did not accumulate waiting days. ::: ![](1744-859X-4-11-1) ::: Only one of the authors, JEB, a psychiatrist in training at the time of the study, was aware of the registration of time from the senior psychiatrists decision to the actual day of referral to the next step in the treatment chain, called decision days in figure [1](#F1){ref-type="fig"}. Waiting days were defined as the number of days from this decision date to the factual transfer. Patients were initially evaluated for some days before such a decision would be made. One patient was still waiting at study end. Waiting time for this patient was truncated at the end of the study period. Cost estimation --------------- Direct costs of treatment were calculated as the daily inpatient expenditures multiplied by number of days. The same costing procedure was used for cost of stay per day in the different secondary facilities. Cost of waiting in the acute ward was then calculated as the difference between cost of residency in the acute ward and the chosen secondary facility. This cost was withdrawn from the cost of the acute ward stay for each patient. The costs used in the calculations were taken from the hospital and secondary facility balances of 2001. Results ======= Twenty-three patients were in the acute wards at the chosen day. Twelve were men (52.2%) and 11 (47.8%) women. Mean age was 35.7 (SD = 9.3) with a range from 22 to 56, table [1](#T1){ref-type="table"}. Four patients were referred voluntarily to the acute ward, 8 were under compulsory observation, and 11 under compulsion for a prolonged period. Eleven patients (47.8%) waited for secondary resident treatment. All patients waiting for further intramural treatment suffered from a psychotic illness, whereas 8 of 12 not waiting had a psychotic illness. Duration of stay was composed of the days preceding the chosen day and the number of days of further treatment in the acute ward. Twenty-three patients had a total of 776 days, of which 425 (54.8 %) were waiting days as defined above. Waiting time for single patients varied from one day to 86 days. There were altogether 7925 resident days in the wards during the year 2001. Net running costs for the acute wards that year were 7,001,960 Euro (9,156,410 \$). Spurious incomes of 394,853 Euro (516,346 \$) and the acute ward\'s share of the food, cleaning, and computer services costs of the whole general hospital were not included. Net running costs of the largest secondary facility were 8,095,343 Euro (10,586,217 \$). The number of resident days were, respectively, 7919 and 14187 during 2001. Mean cost per day of treatment was 884 Euro (1156 \$) and 570 Euro (746 \$), respectively for the acute wards and the largest secondary facility. The cheapest secondary facility had a cost per patient day of 168 Euro (219 \$). Table [2](#T2){ref-type="table"} gives the percentage of patients waiting at randomly chosen days from each of ten other months of 2002. From 9.5% to 39.1% waited on the chosen days. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Patients referred to the acute wards during the first ten months of 2002 according to waiting status (%) and level of compulsion according to the ACPC. ::: Month Waiting (%) Level of compulsion (%) ------- ------------- ------ ------------------------- ------ ------ ------ 1 31.0 69.0 44.8 6.9 48.3 2 39.1 60.9 39.1 13.0 47.8 3 33.3 66.7 25.0 16.7 58,3 4 16.7 83.3 50.0 12.5 37.5 5 25.0 75.0 53.1 21.9 25.0 6 29.2 70.8 37.5 12.5 50.0 7 28.6 71.4 46.4 10.7 42.9 8 9.5 90.5 52.4 19.0 286 9 34.8 65.2 52.2 8.7 39.1 10 27.6 72.4 31.0 10.3 58.6 ::: Mean age varied from 34.9% (SD = 8,0) to 39.4% (SD = 10,1) on the chosen days. For those waiting mean age was 34.0 (SD = 9.0) and for those not waiting for further intramural treatment 38.0 (SD = 11.2). Net running cost was 686,132 Euro (897,250 \$) and the amount allocated to waiting 375,781 Euro (491,406 \$), i.e. 54.8 %. The net loss accrued in the acute wards was 197,693 Euro (258,522 \$), and constitutes 28.8 % of net running cost, i.e. the difference between waiting cost at the acute wards because of delayed further referrals and the net running cost at the secondary facilities. Discussion ========== The impression of many clinicians that patients are waiting unnecessarily in the acute wards is confirmed by the present study. The net loss to the chain of treatment facilities, regardless of where the loss is incurred, was 28.8% of total net running costs, as calculated for the 425 waiting days in resident treatment. A financial system exists that does not contribute to make these costs explicit. Neither the acute wards nor the secondary resident facilities were made economically responsible for the imputed loss. Cost containment would be attained more easily if the economic responsibility covered the complete chain of facilities. That would also give more efficient logistics of patients through the treatment chain. Patients had to wait in the acute ward because referral to ambulatory treatment or treatment at home was deemed clinically irresponsible by the senior psychiatrists. As shown in table [2](#T2){ref-type="table"}, the choice of one day in March represented a higher percentage of waiting patients than post hoc observed for the rest of the year. Length of waiting is, however, not causally related to number of waiting patients, but to factors inherent in the logistics of the psychiatric treatment sector. We have not tried to aggregate our results to an effect for a longer period of time, as this would demand more data on both costs and incomes to the whole treatment chain \[[@B7],[@B8]\]. The present study was not set up to comply with such a comprehensive cost utility or cost effectiveness analysis. Treatment planning for schizophrenia patients after hospitalisation and into more permanent care outside the hospital has been difficult to standardise, but studies show that this would be of importance \[[@B9],[@B10]\]. Early intervention in recently diagnosed schizophrenia has been advocated as a method to reduce future loss of functional abilities\[[@B11]\]. Early intervention is aimed at reducing duration of untreated psychosis, DUP time. Successful reduction of DUP time is also suggested to reduce costs of long-term treatment. Patients with substantial psychosocial problems are also frequent users of resident treatment\[[@B12]\]. The difference between ambulatory and resident psychiatric care was studied by Creed etal in a randomised controlled trial of 179 patients deemed to profit from either\[[@B1]\]. The authors found, as expected, that day treatment was cheaper than inpatient treatment for those patients who could be in day treatment. Inpatients improved significantly faster, but at 12 months the burden on families, also the economic burden, was equal in the two groups. Deinstitutionalisation has been studied in a cost effectiveness perspective\[[@B13]\]. Cost of treatment was lower in long-term patients discharged from hospital compared to those staying. Released patients turned out to be healthier along several dimensions of positive health. Patients in need of continued care are often best helped by treatment and rehabilitation efforts close to where they live. Such care can be sufficient and appropriate after acute care, but a priori it is not necessarily cost effective. If this really is the case, it would be even more important to use acute care resources in an efficient way, i.e. delivering the services needed at the right time in the right facility. Waiting time is not per se a waist of money \[[@B14]\]. Zero waiting time requires excess capacity, probably higher than necessary from a public point of view. The American health financial system is organised with a tilt towards shorter waiting times, and thus towards higher expenditure per capita. Waiting time should be the result of clinical judgments, not bad logistics. Direct costs, as daily inpatient expenditures multiplied by number of days, were used in a study of waiting time before relapse in schizophrenia in USA. Indirect costs of treatment were not estimated\[[@B15]\]. A similar approach, disregarding other ancillary costs, was used in the present study. Cost of illness analysis estimates running and maintenance costs as direct costs, and cost of mortality and morbidity as indirect costs\[[@B2],[@B16]\]. Dorothy Rice found that the direct costs of schizophrenia and affective disorders were twice the indirect costs. The share of the costs of these two illnesses was higher than the prevalence would suggest. However, cost of illness analysis tends to give huge cost estimates, disregarding any imputed gain for others. One group of patients was not considered as waiting in the present study. Several of our patients are referred to the acute ward partly because they also are homeless, and they often get their stay prolonged for social reasons. Homelessness is partly a social welfare problem, but might also be a corollary to the incumbent loosing his home because of psychotic acts (fires, non-payment of rent, destructive and disturbing behaviour). The patients waiting in the acute wards used the services of the department to the same extent as the other patients, they were not \"idly waiting\". Some established a close therapeutic relationship to therapists in the acute ward, who later could not follow up the patient due to other tasks. This may be detrimental to some patients, and could have been avoided if referrals to secondary institutions were smoother. The burden on the families would also be increased by the uncertainty\[[@B17]\]. These factors are not part of the calculations of the study, but would if entered have increased the imputed loss. In a study of re-entries, half the patients had previously been resident patients\[[@B6]\]. Fewer re-entries were observed in patients with long and planned stays, sufficiently organised end of resident stay and follow-up visits. The amount of follow-up by community centres did not improve outcome. Lack of beds in the acute wards due to waiting also affects the health of those seeking treatment, as they would have to wait longer and get their mental status deteriorated. The impact of waiting on staff performance and therapeutic efforts has not been studied here, but would perhaps also be of importance for sick leave and burnout. It is probably a questionable option to shorten resident stays in the acute wards without augmenting the quantity and/or quality of services outside the ward, but it would partly solve the lack of beds, before re-entries increases. Cost of new antipsychotic medication has been used as an argument for cost containment. But at Norwegian prices for medication, even the most expensive atypical antipsychotics in relevant doses for a patient year would not cost more than 7--8 days of resident treatment \[[@B18],[@B19]\]. Waiting for treatment may be rational. Absolutely no waiting for entry to a facility in the treatment chain would demand a capacity which would not be used cost efficiently, the case observed for instance in the airline business. If waiting is necessary, we should avoid waiting at the most expensive slots in the chain, i.e. in acute wards for psychiatric treatment\[[@B14]\]. The financial system of public health care treatment chains makes it difficult to discover where the financial loss accrues, as the acute wards in this case do not take regress on the extra expenses at the following secondary facility not offering a treatment option at the desired time point. The referring doctors outside the acute ward find it increasingly difficult to send patients to acute care, and the number of compulsory referrals is high, in this study 19 out of 23 patients. This would probably represent an economic burden on the families of severely ill psychiatric patients in the case of unduly delayed referrals\[[@B17]\]. The costs demonstrated in the present study should therefore be viewed as a minimum estimate. Modern psychiatric treatment should thus be given the possibility to use the given economic and clinical resources in a cost effective way. This would also include care given by municipalities and private contributors. Conclusion ========== A substantial part of the costs of running an acute psychiatric ward, 29% of running costs in this study, were allocated to waiting. Better logistics in the treatment chain could change this, and several economic incentives along the chain could be used. A treatment chain were only one link is obliged to accept patients without delay, would probably not be the ideal solution. This study indicates that participant observation and cost effectiveness analysis may be combined. Conflict of interest ==================== Both authors were salaried workers in the facility at the time of the study. No financial or other conflicts of interest were present.	PubMed Central	2024-06-05T03:55:59.606252	2005-5-27	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180817/", "journal": "Ann Gen Psychiatry. 2005 May 27; 4:11", "authors": [ { "first": "John E", "last": "Berg" }, { "first": "Asbjørn", "last": "Restan" } ] }
PMC1180818	Background ========== Epidemiological child health data are an essential driver for policy, advocacy, service design, health promotion and prevention programs. Such data can make population health strengths, deficits and inequalities explicit, provide evidence of the influence that the social and political contexts have on health \[[@B1]\] and provide evidence for improvement or worsening in health parameters over time. Data from population surveys may influence both the direction and content of policies and programs for government and non-government organisations, in areas such as education, health, transport, justice and the environment. In designing a population health survey for children, there is a potentially exhaustive list of areas of health that could be examined. It is important that a) the areas that are included reflect the changing mortality and morbidity patterns in children and the changing environments to which children are exposed and b) the results have the potential to inform policy and programs and are user-friendly. This paper will demonstrate a process of selecting areas of child health that meet these criteria, using preliminary work towards the development of a new survey on Victorian children as an example. A recent review of national and international epidemiological studies of children\'s health and wellbeing demonstrated that these studies included several different areas of child health such as chronic diseases, physical activity, oral health, development, services, and neighbourhood and family interactions \[[@B2]\]. In developing a new population survey for children, it is possible to simply apply an existing survey to a new population. However, given that population surveys are resource-intensive and time consuming for families, there is an ethical obligation and financial benefit for survey developers to collect only information relevant to the specific population and the rationale for that survey. Unfortunately, there does not appear to be any standard procedure for identifying the relevant areas of health for a population. A recent review of the major national and international population surveys of child health and wellbeing has demonstrated that researchers often fail to report how they identified the relevant areas of child health \[[@B2]\]. In this paper, we propose that the first step in identifying the relevant areas of child health is to consider a comprehensive intersectoral approach to child health and wellbeing, and its determinants. Theoretical models of health and development, such as the social health model \[[@B3],[@B4]\], ecological model \[[@B5],[@B6]\], Lynch\'s model \[[@B7]\], the National Child Health Performance Framework \[[@B8],[@B9]\] and the lifecourse perspective \[[@B10]\] are useful to understand the scope of children\'s health. For example, the ecological model proposes that children need to be considered within their family, school, neighbourhood and the larger social, structural, economic, political and cultural environment \[[@B5],[@B7]\]. The National Child Health Performance Framework, developed by the National Health Performance Committee, contains a set of indicators to measure trends in health status, determinants of heath and the use and delivery of service \[[@B8],[@B9]\]. Although these models are important in understanding the spectrum of child health, it is clearly not possible, and may not even be necessary, to include all of these indicators of child health in a population survey. The next step in developing a population survey is to prioritise the areas of child health relevant to the funders and users of population data. It is possible to apply some criteria to reduce the potential list of areas. Only one epidemiological study of children\'s health has developed and reported criteria to prioritise areas of child health. The New South Wales Child Health Survey \[[@B11]\] selected their areas of child health using the following criteria: 1) it is a priority for child health as documented in a state or national child health policy document; 2) it meets the information needs of the NSW Department of Health and Area Health Services in relation to child health; 3) the information is not readily available from other sources; 4) the estimated sample size is large enough to provide data that can be used to generalise responses to the NSW population of children; 5) the areas are not highly sensitive to respondents and likely to cause failure to complete the survey. The Strategic Plan Health Gain for Children and Youth in Central Sydney prioritised issues for children and youth by analysing information on prevalence, severity of a condition, community concern for the issue and efficacy of available interventions \[[@B12]\]. Other important criteria have been developed for use in adult population health surveys. According to the NSW Strategy for Population Health Surveillance, each area of health is considered in terms of its burden (ie. incidence, prevalence, mortality, years of potential life lost, hospitalisation rate), preventability, communicability, public interest, and legislative requirements \[[@B13]\]. In this paper we argue that the criterion through which all others need to be filtered is that the data have the potential to inform policy and programs. It is essential that survey developers consider dissemination and uptake of results, ensuring there is utility for researchers and policymakers alike. To address this criterion, it is recommended that survey developers consult with stakeholders and potential users of the data, and then apply the remaining criteria. Stakeholders can provide insights on policy and program decision making, the use of data in that process and recommendations for the best data. This paper aims to demonstrate a process by which survey developers can consult with stakeholders to determine the relevant areas of child health for a population survey of child health and wellbeing. Method ====== Participants ------------ Fifty four key stakeholders participated in this study. The sample was selected to represent the areas of health and development in the National Child Health Performance Framework \[[@B8],[@B9]\]. The National Child Health Performance Framework consists of three broad groups of indicators health status, risk and protective factors, and services and interventions. Health status has four subgroups: health and wellbeing, growth and development, mortality, morbidity and disability, and safety and security. The risk and protective factors group has three subgroups: social, cultural and environmental factors; biological and behavioural factors and health knowledge. The services and interventions group includes health services, health programs, health promotion and intervention, intersectoral services and community services. To identify the indicators of health status, we consulted with stakeholders with expertise in children\'s physical, social and emotional wellbeing, development, disability, mental health problems, illness, oral health problems, nutrition related problems, child abuse and parental health and wellbeing. To identify the risk and protective factors, we consulted with stakeholders with an understanding of the impact of the physical, family, economic, social and school environment. We also consulted with experts in the area of child diet, activity and overweight and obesity. To identify services and interventions, we consulted with stakeholders with knowledge about health service utilisation, maternal and child health programs, community services and health promotion programs. Stakeholders were selected to encompass a variety of organisations, including university and government departments within Victoria. The stakeholders were identified by the authors and though literature reviews. A snowball technique was also used where initial respondents were asked to suggest others whom they know are in the target group and who could be invited to take part, and so on. Materials --------- The stakeholders participated in one-on-one interviews. The interviews were semi-structured and the questions were adapted from those included in a quasi-delphi study for the Victorian Adolescent Health and Wellbeing Survey \[[@B14],[@B15]\]. The stakeholders provided advice on the area of child health that they have expertise in, what they thought were the most important areas of health and discussed how they would use the results (Refer to Table [1](#T1){ref-type="table"}). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Interview questions for stakeholders ::: ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- 1\) Which key areas of child health are you interested in? 2\) Thinking about current policy and programs within the key area, which specific aspects of children\'s health and wellbeing would you measure in a statewide survey? 3\) Would your organisation use the results of a statewide survey of children\'s health than measured these aspects? If so, how? What results are needed? 4\) In what format would you want to receive the results so that they were meaningful for you? ------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ::: Procedure --------- The interviews with the stakeholders generally lasted between 15--45 minutes. Interviewers recorded the major points of the interview, and produced a summary of each interview. The summaries were then sent to the interviewees to correct information and/or add further information. Once corrected, the responses for each question were entered into an Excel database and the data was coded by two researchers using open coding. This is the process of identifying persistent words, phrases, themes or concepts within the data so that the underlying patterns can be identified and analysed \[[@B16]\]. A coding framework was developed and two researchers coded each of the summaries using focused coding (EW, ED). Agreement on key themes was achieved by discussion. Results ======= Fifty-four stakeholders participated in this study. Stakeholders were asked which aspects of child health they were interested in. As demonstrated in Table [2](#T2){ref-type="table"}, their areas of interest could be mapped to the National Health Performance Framework. The total number of areas exceeds 54, because several stakeholders indicated more than one area of interest/expertise. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Stakeholders expertise ::: ------------------------------------------------------------------------------------------------------------------------------------------ Indicators of NHPF Subgroups of NHPF Stakeholders Areas of Expertise (numbers of stakeholders) ----------------------------- ------------------------------------- ---------------------------------------------------------------------- Health Status and Outcomes Life expectancy and wellbeing Child physical (4)\ Child social and emotional wellbeing (9) Mortality, morbidity and disability Child disability (3)\ Child mental health problems (5)\ Childhood injury (2)\ Child chronic illnesses (4) Risk and Protective factors Environmental factors Physical environment (2)\ Community environment (4)\ Exposure to tobacco smoke (4) Socioeconomic factors Economic environment (3)\ Child education (5)\ Parental employment (1) Community capacity Family environment (9)\ Social environment (7)\ Parental health (1) Health behaviours Health behaviours -- All (2)\ Child physical activity (4)\ Child diet and nutrition (3)\ Child oral health behaviours (2)\ Sun protection (1)\ Vaccinations (1)\ Injury prevention (2) Person-related factors Birth defects (1)\ Health behaviours during pregnancy (ie smoking, alcohol, folate) (2) Services and Interventions Health service utilization (3)\ Maternal and Child Health Programs (1)\ Community services (1)\ Health promotion programs (1) Socio-demographic factors Socioeconomic position (2)\ Socioeconomic inequalities (2)\ Family structure (1) Population groups Socioeconomically disadvantaged groups (4)\ Rural and remote area residents (1)\ Overseas born (1)\ Indigenous Australians (1) ------------------------------------------------------------------------------------------------------------------------------------------ NHPF -- National Health Performance Framework ::: Stakeholders were also asked what aspects of child health they would include in a population survey of Victorian children\'s health and wellbeing. As several different areas of health were identified, their responses were grouped according to major themes. These include obesity and determinants, social and emotional health and wellbeing, family environment, health service utilisation, illness, community, oral health, injury, pregnancy and breastfeeding and socioeconomic position. Table [3](#T3){ref-type="table"} demonstrates the overarching themes, the areas that represent the themes, and the specific data that are required by the stakeholders. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Priority Areas of Child Health Identified by Stakeholders ::: ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ Themes Areas of Child Health Specific data required by stakeholders ------------------------------------------- --------------------------------- ------------------------------------------------------------------------------------------------------------------------------------------------------------------ Obesity and determinants Physical activity\ 1\) Need epidemiological data on childhood obesity, physical activity, sedentary behaviours and nutritional intake in Victoria.\ Nutrition\ 2) Need data on mediating and psychosocial variables. Obesity Social and emotional health and wellbeing Social and emotional wellbeing\ 1\) Need data on the prevalence and distribution of mental health problems.\ Behavioural problems\ 2) Need data on the adequacy of mental health services and barriers to seeking help. Mental health Family Environment Family environment\ 1\) Families have undergone substantial changes, and we need data on how different family environments impact on children\'s health. Parenting style\ Reading\ Exposure to smoking Health service utilisation Health service utilisation 1\) We need data to ensure that our services are meeting the needs of the community, and ensure that people are satisfied with them. Childhood illness Chronic illness\ 1\) Need data on the prevalence of chronic illness and disability. \\par Disability\ Development Community Neighbourhood/Community 1\) The community environment impacts on children\'s health; to get a complete picture of children\'s health, need to examine the community environment. Oral health Oral health 1\) There are no population data on the oral health status of children, across this proposed age group. Injury Injury 1\) Need data on the prevalence of injuries and how they are treated.\ 2) Need data on whether families are reducing the risk of injuries by protecting their home. Pregnancy and breastfeeding Breastfeeding\ 1\) Need prevalence data on smoking, alcohol and folate intake during pregnancy. Smoking in pregnancy Socioeconomic position Health inequalities 1\) A statewide survey of child health should include the child\'s socioeconomic position to examine distributional effects of health and program effectiveness. ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------ ::: After gaining insight into the potential for each area of child health to aid policy and program decisions, the remaining criteria can now be applied. Based on the criteria developed for the NSW Child Health Survey and the criteria developed for the NSW Strategy for Population Health Surveillance, it is recommended that: 1\) The information is not being collected elsewhere (ie. databases, school records etc). 2\) The question can be answered using a population survey. 3\) The domain impacts on children\'s mortality or morbidity. 4\) The area of child health can be measured in a population survey (ie depending on data collection method and length). 5\) The data are user-friendly and the results have the potential to inform policy and programs Using these criteria, breastfeeding, development and parenting style were excluded. Breastfeeding is already being measured by maternal and child health centres and the Australian Bureau of Statistics. Although child development is important, assessments of children\'s developmental status are extremely resource intensive and therefore unable to be employed in a population data collection. Parenting style was assessed by the authors to be less useful for policy and program development. The remainder of the areas met the above criteria, and were therefore included. The stakeholders indicated that the results from a child population health survey could be used in the pursuit of evidence-based policies, practice and programs, for service planning, for advocacy, to develop networks across community, to support the generation of appropriate local responses, to develop interventions, to use in submissions for funding, and to use in publications. Some stakeholders indicated the need to localise data and to make it more powerful in action terms; other stakeholders suggested that rural/urban comparisons would be important. Stakeholders suggested that the results should be available from both a representative sample and also from key minority groups such as Indigenous children. The results should also contain some comparable measures to other work done elsewhere. Discussion ========== This study demonstrated the process by which areas of child health can be identified and prioritised for a population study of health and wellbeing. Conducting qualitative interviews with stakeholders is a useful and efficient method to identify current issues in a specific area, and to provide exposure to significant research papers and unpublished research. The areas of child health that were identified in this study are not only useful in developing a population survey of child health and wellbeing for Victorian children, they are also useful for researchers and practitioners in the field of child health, in terms of guiding research, policy and program development. The main themes of child health tended to reflect the changing patterns of morbidity, where there is increasing interest in the rising prevalence rates of obesity, mental health problems, and oral health problems. The emphasis on health service utilisation, disability and chronic illness is reflective of the costs that such children impose on the health care system. The emphasis on family health, exposure to tobacco smoke, community and socioeconomic position is indicative of the more recent emphasis placed on the wider community environment and influences, and recognition that children\'s environments have changed profoundly. In terms of the specific data that the stakeholders recommended for each area of health, there was a clear need for prevalence data and also for establishing and modelling the determinants of child health. The areas of child health that emerged from the interviews are consistent with the stakeholders\' areas of expertise. Although it seems likely that the exact sample of stakeholders will always influence the areas of child health that are identified, the selection of these stakeholders was based on the National Child Health Performance Framework, an acceptable indicator framework. In terms of the process by which researchers determine the priority areas of child health, we recommend that survey developers utilise a model of health and development, such as the National Child Health Performance Framework, to identify the possible areas of child health. To prioritise areas, it is recommended that survey developers consult with relevant stakeholders to ensure that the data are user-friendly and the results have the potential to inform policy and programs. The selection of the stakeholders\' areas of expertise should be consistent with a selected theory or framework of health. The areas of child health identified by the stakeholders can be further prioritised using the proposed criteria, which are based on the NSW Strategy for Population Health Surveillance. Limitations =========== This study has limitations in its sampling methodology. The stakeholders were identified by the authors and through the use of snowballing. This methodology does have potential for selection bias and thus may limit generalisability of results. Given that the stakeholders were selected to ensure that there were representatives from all areas of health, selective sampling was necessary. A further issue for discussion is the inclusion of children in such a study. Increasingly there is recognition that children and parents need to be included in program planning and policy development. Given the format of the questions and the aim of this study, children\'s and parent\'s perspectives were not obtained. It is recommended that when the questionnaire is established and parents and children can understand what is being measured, they should be consulted about the areas of health that are included in a population survey. This process is currently being undertaken with a diverse group of parents and children. Conclusion ========== Population child health data is important for informing policies, programs and services in a range of sectors. However, the process by which researchers determine the priority areas of child health remains largely un-defined. The phases of this study included a rigorous research process, including qualitative interviews with stakeholders in the area of child health. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= All authors made substantial contribution to the conception, design, analysis and interpretation of data. ED and JW conducted the data collection. ED and EW analysed the data for themes. All authors read and approved the final manuscript. Acknowledgements ================ We would like to thank all of the stakeholders who participated in this study; and the specific funding provided by the Victorian Health Promotion Foundation for the study and the Public Health Research Fellowship awarded to EW.	PubMed Central	2024-06-05T03:55:59.608383	2005-7-20	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180818/", "journal": "Aust New Zealand Health Policy. 2005 Jul 20; 2:16", "authors": [ { "first": "Elise", "last": "Davis" }, { "first": "Elizabeth", "last": "Waters" }, { "first": "Melissa", "last": "Wake" }, { "first": "Sharon", "last": "Goldfeld" }, { "first": "Joanne", "last": "Williams" }, { "first": "Ozlem", "last": "Mehmet-Radji" }, { "first": "Frank", "last": "Oberklaid" } ] }
PMC1180819	Introduction ============ Attention-deficit/hyperactivity disorder (ADHD) is the most commonly diagnosed psychiatric disorder of childhood \[[@B1],[@B2]\]. Children with ADHD are characterized by an inability to sit still, they have difficulty organizing tasks, they are forgetful, have a tendency to be easily distracted, fidget, they have difficulty with tasks that require sustained attention and are risk-takers \[[@B3],[@B4]\]. Their behaviour falls into two or three core clusters of symptoms, impaired sustained attention and hyperactivity/impulsiveness that develops gradually in familiar situations, with impairment manifesting before age 7 \[[@B3],[@B5]\]. The high population prevalence and heritability of ADHD agrees with ADHD being caused by multiple genes with small effect size \[[@B2],[@B6]\]. Associations have been found between polymorphisms in several monoaminergic genes and ADHD. These include the dopamine D1, D4 and D5 receptor (DRD1, DRD4, DRD5) genes, the α~2~-adrenoceptor gene, dopamine, norepinephrine and serotonin transporter (DAT1, NET1, SERT1) genes \[[@B7]-[@B19]\]. Contradictory negative findings have also been reported suggesting that different combinations of genetic factors may combine to produce individual clusters of behavioural characteristics of ADHD \[[@B20]-[@B23]\]. Different alleles of genes encoding proteins related to dopamine function differentially affect cognitive function \[[@B24]\]. The effect of a single gene on behaviour has been described as small causing a slight bias towards one end of a continuum \[[@B4],[@B24]\]. Numerous studies have found reduced brain volume in ADHD patients, particularly prefrontal cortex, cerebellum, corpus callosum, and basal ganglia \[[@B25]-[@B29]\]. Dopamine alters brain structure and function \[[@B24]\]. The DAT1 genotype preferentially influenced caudate volume; individuals homozygous for the 10-repeat allele which is associated with ADHD had smaller caudate volumes than individuals carrying the 9-repeat allele \[[@B24]\]. The DRD4 genotype influenced prefrontal gray matter. Individuals homozygous for the 4-repeat allele had smaller volumes than individuals carrying other variants of the gene \[[@B24]\]. Imaging studies have demonstrated functional abnormalities in striatum, frontal cortex and cerebellum of patients with ADHD \[[@B30]-[@B34]\]. Imaging studies have revealed robust increases in striatal DAT of up to 70% in ADHD children and adults \[[@B10],[@B15],[@B35]\]. Although not every study found increased DAT \[[@B36],[@B37]\], there is a strong possibility that the DAT1 gene is overexpressed in the striatum of ADHD subjects, and that this results in reduced synaptic dopamine. Psychostimulants are highly effective in ameliorating the three major clusters of behavioural symptoms of ADHD \[[@B38]\]. Methylphenidate produced improvements in spatial working memory, attentional set-shifting \[[@B39],[@B40]\], and inhibition of previously acquired behavioural responses to non-relevant stimuli \[[@B34],[@B41]-[@B43]\]. It increased the previously reduced striatal activity in patients with ADHD \[[@B34]\] and reduced cerebral blood flow in frontal and parietal cortex \[[@B39],[@B44]\]. Animal models of ADHD --------------------- Although non-human primate brains are closer to human brains than rodents, rodent models of ADHD have the advantage that they are genetically more homogeneous, they are less expensive to maintain, greater numbers of experimental animals are available so they are not used for multiple studies, and much more is known about their neurobiology than primates. The researcher also has better control over variables such as diet, environment, and learning history. Rodent models have simpler nervous systems, they cannot be used to study complex cognitive behaviour like language but their basic behavioural mechanisms are similar to humans. ADHD is a heterogeneous disorder with individual patients presenting with quite different behavioural symptoms probably as a result of different combinations of genetic and environmental factors. Animal models provide invaluable insight into the neurochemistry underlying specific aspects of ADHD behaviour, when compared to appropriate controls. Differences between the behaviour of an animal model and its control can be correlated with differences in their neurochemistry and behavioural pharmacology. A list of criteria for an optimal animal model of ADHD was recently suggested \[[@B45]\] (i) the model should mimic the fundamental behavioural characteristics of ADHD (face validity), impulsiveness should be absent initially and develop gradually over time, sustained attention-deficit should be demonstrated only when stimuli are widely spaced in time, hyperactivity should not be observed in a novel, non-threatening environment, it should develop over time; (ii) the model should conform to a theoretical rationale for ADHD (construct validity): the two main behavioural processes that are proposed to be major contributory factors in the aetiology of ADHD, altered reinforcement of novel behaviour and deficient extinction of previously reinforced behaviour, should be demonstrated; (iii) the model should predict novel aspects of ADHD behaviour, genetics, and neurobiology (predictive validity); and (iv) it should be neurodevelopmental, preferably a genetic model. We use the concept \'reinforcer\' strictly in a behavioural sense, without making any references to subjective or cognitive states. The alternative concept of \'reward\' is more cognitive and may connote several subjective states like pleasure as well as incentive and reinforcer \[[@B46]\]. Therefore, there is not a perfect overlap between reinforcer and reward. We prefer the more descriptive and less ambiguously defined concept of reinforcer rather than reward \[[@B4]\]. Spontaneously hypertensive rats (SHR) were found to be the best characterized and also currently the most appropriate model of ADHD \[[@B45]\]. SHR fulfil most of the validation criteria listed above and compare well with clinical cases of ADHD \[[@B45],[@B47]\]. Poor performers in the 5-choice serial reaction time (5-CSRT) task were suggested to be a useful model for the inattentive subtype of ADHD \[[@B45]\]. Other animal models were suggested to provide useful information concerning aspects of ADHD behaviour \[[@B45]\]. SHR --- SHR exhibit all the behavioural characteristics of ADHD: impaired sustained attention without obvious sensory problems, motor impulsiveness, and hyperactivity that is not present in novel, non-threatening situations but develops over time when reinforcers are infrequent \[[@B45],[@B47]\]. Similar to children with ADHD, SHR display increased behavioural variability, deficient response re-engagement, and make significantly more errors than controls \[[@B45],[@B47],[@B48]\]. Besides conforming to behavioural criteria for an animal model of ADHD, SHR fulfils the additional criterion that it is a genetic model of ADHD bred from progenitor Wistar Kyoto (WKY) rats \[[@B49]\]. WKY serve as a valid control for SHR since their behavioural characteristics are similar to those of other rat strains \[[@B47]\]. Three candidate dopamine genes (DRD2, DRD4, and DAT) were sequenced in SHR and WKY \[Mill et al., unpublished\]. No differences were found in DRD2 or DRD4 genes but a 160 bp insertion was found in the non-coding region upstream of exon 3 of the DAT1 gene which is of significance since variable number tandem repeats in the 3\'-untranslated region of the DAT gene has been associated with ADHD in several family studies \[[@B7],[@B9],[@B10],[@B14],[@B15]\]. A possible disturbance in the regulation of the DAT1 gene is in agreement with findings that DAT1 gene expression is transiently reduced in SHR midbrain during the first postnatal month and increased in adult SHR compared to controls \[[@B50],[@B51]\]. Alterations in DAT1 gene expression can affect dopamine uptake and reutilization. Decreased expression of DAT1 will reduce reuptake and increase metabolism of dopamine. Differences in dopamine metabolism have been reported for children and adults with ADHD \[[@B52],[@B53]\] which is consistent with developmental changes in DAT1 expression and consequent changes in dopamine uptake. DOPA decarboxylase activity was found to be increased in the midbrain of children and decreased in prefrontal cortex of adults with ADHD compared to controls \[[@B52],[@B53]\]. Reduced DAT1 expression at a young age would reduce dopamine reuptake, thereby reducing dopamine reutilization and necessitating increased synthesis of dopamine by DOPA decarboxylase. In adults, increased expression of DAT1 might be expected to increase reuptake of dopamine, thereby reducing the need for synthesis by DOPA decarboxylase. SHR appear to have higher extracellular tonic dopamine in the nucleus accumbens shell \[[@B54]\]. However, consistent with increased DAT1 expression in adult SHR striatum, extracellular dopamine levels are decreased in the caudate nucleus \[[@B55],[@B56]\] and d-amphetamine-stimulated release of dopamine via DAT1 is greater in SHR striatum than WKY \[[@B54],[@B57],[@B58]\]. Evidence suggests that DAT1 is hypofunctional in SHR, since despite the increased number of DATs, inhibition of dopamine uptake by low concentrations of methylphenidate or nomifensine increased the electrically-stimulated release of dopamine to the same extent in SHR and WKY nucleus accumbens and caudate-putamen \[[@B58],[@B59]\]. These findings suggest that increased expression of the DAT1 gene may reflect an attempt to compensate for increased tonic extracellular dopamine in the nucleus accumbens shell of SHR or increased DAT1 expression may occur in an attempt to compensate for decreased function of DAT1 in adult SHR striatum. Hypertension is a confounding factor in the SHR model of ADHD. However, SHR do not develop hypertension until they are adults, from 10 to 12 weeks of age, whereas hyperactivity is observed at 3 to 4 weeks of age before they enter puberty \[[@B55]\]. In an attempt to map quantitative trait loci for complex phenotypes, SHR were crossed with a Brown Norway congenic strain to generate a set of recombinant inbred strains \[[@B60]\]. Analysis of their behaviour revealed that locomotion mapped to chromosomes 3, 8 and 18 while hypertension exhibited multigenic complexity with both environment and genetic background as contributing factors. Elevated arterial blood pressure was higher when measured by direct catheterization compared to radiotelemetry suggesting that SHR hypertension is a product of stress-dependent trait expression \[[@B60]\]. SHR behaviour was suggested to result from an interaction between genetics and the environment \[[@B60]\], much like ADHD \[[@B4],[@B6],[@B23]\]. In addition to behavioural and genetic similarities to ADHD, SHR exhibit brain pathology similar to ADHD. SHR brain volumes, specifically prefrontal cortex, occipital cortex, and hippocampus, are smaller than controls \[[@B61]\]. MRI revealed significantly increased ventricular volume in SHR compared to WKY at 3 months of age \[[@B61]\]. There are fewer neurons in these brain areas compared to WKY \[[@B62]-[@B64]\]. Results obtained with SHR have predicted novel alternatives to existing theories concerning the aetiology of ADHD. A major second messenger system involving calcium signalling is dysfunctional in SHR \[[@B65],[@B66]\] suggesting that several neurotransmitter systems could be impaired in ADHD. Compared to WKY, SHR have lower brain Ca^2+^ATPase activity \[[@B67]\]. Because neurotransmitter release is dependent on calcium influx, a disturbance in the concentration gradient of calcium across the cell membrane may decrease the influx of calcium ions into the cell and impair neurotransmitter release \[[@B65]\]. Decreased calcium influx through NMDA channels would also impair intra- and intercellular signalling as well as LTP, the neuronal analogue of learning \[[@B4],[@B68]\]. Calcyon, a transmembrane protein involved in DRD1/DRD5 signalling, has been implicated in ADHD \[[@B69]\]. Polymorphisms of the calcyon gene (DRD1IP) have been associated with both inattentive and hyperactive/impulsive subtypes of ADHD \[Laurin et al., submitted\]. Calcyon enables the typically Gs-linked DRD1/DRD5 to switch from Gs to Gq coupling, thereby stimulating inositol 1,4,5-triphosphate (IP3) turnover with resultant release of calcium from intracellular stores. This effect requires a priming step involving heterologous Gq-linked G-protein-coupled receptor activation. Subsequent DRD1 activation causes elevation of intracellular calcium that triggers calcyon accumulation in the plasma membrane \[[@B70]\]. SHR appear to have a disturbance in calcium metabolism. Increased intracellular calcium could interfere with calcyon translocation to the cell membrane and thereby impair DRD1/DRD5 function. Alternatively, altered calcyon function could impair DRD1/DRD5 function by affecting the switch from Gs to Gq coupling, thereby also altering intracellular calcium concentration. These findings suggest that the primary disturbance in ADHD may be located in factors that regulate postsynaptic DRD1/DRD5 signalling mechanisms, hence, calcyon and other DRD1/DRD5 signalling-related proteins/peptides deserve further investigation \[Laurin et al, submitted\]. The hypothesis that calcyon may be a primary disturbance in ADHD could be tested in animal models, particularly SHR. ### Dopamine Dopamine neurons play an important modulatory role in the brain \[[@B71]\]. Neurons that release dopamine influence behaviour by exerting modulatory effects on the transfer of information through neuronal circuits that connect functionally distinct cortical areas to specific regions of the striatum in parallel cortico-striato-thalamo-cortical pathways. Dopamine assists in reprogramming the brain by selectively reinforcing the weights of the synapses that are active around the time of behavioural reinforcement \[[@B4],[@B72]\]. There are three major dopaminergic systems in the brain, the mesolimbic, mesocortical and nigrostriatal pathways. Mesolimbic dopamine neurons project from the ventral tegmental area of the midbrain (VTA) to limbic areas of the brain. The firing rate of dopamine neurons is increased in response to unexpected reward and decreased when a fully predicted reward is omitted \[[@B73],[@B74]\]. It has been suggested that deficient reinforcement of appropriate behaviour and/or deficient extinction of previously reinforced behaviour can give rise to ADHD symptoms of delay aversion, hyperactivity in a familiar environment, impulsiveness, deficient sustained attention, increased behavioural variability and failure to extinguish previously acquired behaviour \[[@B4]\]. The mesocortical dopamine system originates in the VTA and projects to cortical areas, including the prefrontal, parietal and temporal cortex. These dopamine projections modulate circuits that are known to play an important role in a variety of executive functions, including motor control, behavioural inhibition, attention, and working memory \[[@B4],[@B75]\]. Dopamine activation of DRD2 selectively modulates neural activities associated with memory-guided motor activity in delayed response tasks whereas DRD1 are responsible for memory-related persistent activation of prefrontal cortex neurons \[[@B76]\]. Deficient dopamine-mediated modulation of prefrontal cortical circuits has been suggested \[[@B4]\] to cause attention response deficiencies (impaired orienting responses, saccadic eye movements and responses towards a target) and impaired executive functions (poor behavioural planning). Nigrostriatal dopamine neurons project from the substantia nigra pars compacta to the dorsal striatum (caudate nucleus and putamen). Impaired dopamine modulation of cortico-striato-thalamo-cortical circuits can impair motor function and cause deficient habit learning i.e. impaired nondeclarative memory formation \[[@B4]\]. These impairments can give rise to apparent developmental delay, clumsiness, and neurological \"soft signs\" \[[@B4]\]. Dopamine uptake, storage and/or metabolism are disturbed in SHR \[[@B58],[@B77]\], as has been proposed for children and adults with ADHD \[[@B52],[@B53]\]. Although the tissue concentration of dopamine, which reflects vesicle stores of dopamine, is similar in SHR and WKY \[[@B78]\], dopamine turnover which reflects release and metabolism of dopamine was lower in adult SHR substantia nigra, VTA, striatum and frontal cortex \[[@B78],[@B79]\]. The dopamine metabolite, homovanillic acid, and the homovanillic acid /dopamine ratio are much lower in several brain areas of adult SHR compared to WKY. These findings are consistent with increased expression of DAT1 increasing dopamine re-uptake and reutilization thereby reducing metabolism. The results could also be interpreted to suggest that dopamine release is decreased and dopamine transmission is impaired in SHR. In vitrostimulation-evoked (electrical and/or exposure to high K^+^concentration) release of dopamine from terminals of mesocortical, mesolimbic and nigrostriatal dopamine neurons of SHR is significantly less than that of WKY \[[@B55],[@B56],[@B58],[@B59],[@B77],[@B80]\]. A similar decrease in stimulus-evoked release of dopamine was observed in SHR nucleus accumbens shell in vivo\[[@B54]\]. The elevated dopamine concentration in the nucleus accumbens shell may have increased activation of endogenous DRD2 autoreceptors and thus reduced stimulus-evoked dopamine release \[[@B54]\]. In vitroDRD2-mediated inhibition of dopamine release was greater in SHR caudate-putamen and nucleus accumbens than WKY, while DRD2 function was unchanged in prefrontal cortex of SHR \[[@B77]\]. Results are consistent with increased endogenous activation of DRD2 in the nucleus accumbens while increased efficacy of endogenous dopamine activation of DRD2 autoreceptors was suggested to account for the decreased stimulus-evoked release of dopamine in SHR striatum \[[@B77]\]. Increased DRD2 function was suggested to have occurred in a compensatory reaction to abnormally elevated dopamine levels at an early stage of development, perhaps as a result of exposure to stress and/or a genetic defect \[[@B80]\]. A likely explanation could also be that elevated extracellular dopamine levels could have resulted from decreased DAT1 expression during the first postnatal month of SHR \[[@B50]\] which caused compensatory upregulation of both DAT1 and DRD2 receptors \[[@B51],[@B58],[@B77]\]. Increased d-amphetamine stimulated, DAT-mediated, release of dopamine from SHR prefrontal cortex and striatum is consistent with upregulation of DAT1 in adult SHR \[[@B58]\]. In agreement with decreased stimulus-evoked release of dopamine, postsynaptic DRD1 are increased in the caudate-putamen and nucleus accumbens of SHR \[[@B51],[@B81],[@B82]\] suggesting compensatory upregulation of postsynaptic receptors. Dopamine dysfunction may contribute to the altered reinforcement processes of ADHD \[[@B4],[@B45]\]. DRD1 mediates reinforcement by strengthening synaptic connections between neurons (long-term potentiation, LTP) or weakening synaptic connections (long-term depression, LTD) in neural circuits that involve prefrontal cortex and/or striatum (e.g. cortico-striato-thalamo-cortical circuits) \[[@B74],[@B83],[@B84]\]. LTP is regarded as a neuronal correlate of learning \[[@B68]\]. It requires interplay between several factors. Among these is coincident glutamate stimulation of NMDA receptors and local membrane depolarization sufficient to remove the magnesium block of NMDA receptor channels. Glutamate activation of AMPA receptors allows influx of sodium ions and can thereby produce this depolarization. Activation of the DRD1-protein kinase A signalling pathway increases the mobilisation of AMPA and NMDA receptors to the cell surface, thereby promoting LTP \[[@B85]-[@B87]\]. Calcium enters the cell through the NMDA receptor channel, activates various protein kinases including calcium/calmodulin-dependent protein kinase II, and mobilizes \"silent\" AMPA receptors required for LTP to take place \[[@B68]\]. NMDA receptor-induced excitation is enhanced by DRD1 activation and attenuated by DRD2 activation \[[@B88]-[@B90]\]. Thus DRD1 activation may synergistically increase the excitatory actions of glutamate at NMDA receptors, increasing the open time of NMDA channels and therefore increase the calcium signal. Activation of DRD1 receptors is required for stimulation of cAMP formation, and subsequent activation of cAMP-dependent protein kinase, phosphorylation of CREB and gene transcription required for consolidation of memory traces \[[@B91]\]. NMDA receptors are required for LTP in the hippocampus \[[@B68]\], in cortico-striatal synapses \[[@B92]\], and in the nucleus accumbens \[[@B93]\]. Phasic application of dopamine in the prefrontal cortex facilitates LTP, suggesting that dopamine can promote reinforcement processes by strengthening (or weakening) network connections in the prefrontal cortex \[[@B94]\]. Bilateral infusion of a DRD1 agonist increased attentional accuracy and facilitated short-term spatial memory after delays of several seconds and impaired memory of the location of a visual target after short delays, thereby modulating short-term working memory in a delay-dependent manner \[[@B95]\]. Effective decision making requires the ability to adapt behaviour on the basis of changes in emotional significance. Rats with lesions of the orbitofrontal cortex showed increased preference for larger but delayed rewards whereas rats with lesions of the basolateral amygdala showed increased choice of small immediate rewards \[[@B96]\]. Within the striatum, LTP (and LTD) only occurs in the presence of dopaminergic input \[[@B97]\]. Consistent with decreased DRD1 activation in SHR giving rise to decreased facilitation of NMDA receptor function and impaired gene transcription, SHR have reduced expression of calcium/calmodulin-dependent protein kinase II and c-fos gene in the anterior striatum \[[@B98]-[@B100]\]. Decreased DRD1 function will alter transmission of signals in cortical and striatal circuits of SHR and impair dopamine-mediated cognitive function and reinforcement of appropriate behaviour. Deficient DRD1 function in SHR implies that only stimuli with strong reinforcer value will release enough DA to stimulate DRD1 sufficiently to facilitate NMDA receptor function and produce phosphorylation of CREB and other important proteins/peptides required for strengthening of synaptic connections in circuits representing goal-directed learning and memory consolidation. Weak stimuli will not cause behaviour to be reinforced in SHR because of reduced activation of the DRD1 signalling pathways. As suggested in the dynamic developmental behavioural theory (DDT) of Sagvolden et al \[[@B4]\], SHR have a steeper delay gradient, so decreased activation of DRD1 signalling pathways in response to an unexpected reinforcer would not strengthen synaptic connections in circuits that were activated by stimuli some time prior to the reinforcer with the result that more recently activated behavioural circuits would be preferentially strengthened and their memory consolidated. The striatum is central to behavioural control and receives the greatest density of dopamine input of all central nervous structures. Striatal dopamine hypofunction may be associated with subtle motor control problems in children with ADHD and SHR \[[@B4]\]. Impaired dopamine regulation of striatal function may contribute to the poor motor development associated with severe cases of ADHD \[[@B101]\]. Response \"disinhibition\" may be due to impaired motor control associated with dopamine hypofunction in the striatum rather than frontal lobe dysfunction \[[@B4]\]. Indeed, motor control problems may explain a number of effects including clumsiness, increased variability in speed, less accurate response re-engagement, \"failure to inhibit\" responses when quick reactions are required, impaired orienting responses, increased number of responses with extended reaction times, apparent developmental delay, neurological \"soft signs\", and language delays \[[@B4]\]. Papa et al. \[[@B100]\] found decreased calcium/calmodulin-dependent protein kinase II in the nucleus accumbens shell but not the core of SHR when compared to WKY. The mesolimbic dopamine projection to the shell subdivision of the nucleus accumbens is responsible for motivation and determines the amount of effort an animal is prepared to exert in order to obtain a reward. Hypofunction of the mesolimbic dopamine system will impair the function of the mesocortical and nigrostriatal dopamine systems by influencing dopamine release and the cortico-striato-thalamo-cortical circuits that dopamine modulates. This could impair learning and expression of goal-directed behaviour thereby contributing to the aspects of ADHD symptoms displayed by SHR \[[@B102]\]. ### Norepinephrine In addition to the hypothesis that dopaminergic systems are hypofunctional in ADHD, noradrenergic neurons have been suggested to be poorly regulated and hyperfunctional in the prefrontal cortex of children with ADHD \[[@B40],[@B103],[@B104]\]. Noradrenergic neurons appear to enhance the signal-to-noise ratio in prefrontal and parietal cortices, amplify responses to attended stimuli, and reduce responses to irrelevant stimuli \[[@B71],[@B105]\]. Both of these functions are defective in ADHD \[[@B71]\]. The locus coeruleus diffusely innervates diverse regions throughout the central nervous system including the entire cerebral cortex, various subcortical areas, cerebellum and spinal cord, and plays an important role in attention, arousal, orienting, and vigilance \[[@B40]\]. For example, locus coeruleus neurons respond selectively to attended (target) stimuli; tonic locus coeruleus activity corresponds to arousal state, and both very low and very high locus coeruleus activity are associated with impaired vigilance \[[@B103],[@B105]\]. Noradrenergic neurons that project from the locus coeruleus to the prefrontal cortex release norepinephrine which helps to guide behaviour by modulating the transfer of information through neuronal circuits that are responsible for selective and sustained attention \[[@B40]\]. Norepinephrine, like dopamine, alters the strength of neural connections leading to adaptive changes in behaviour. This occurs through activation of β-adrenoceptors that stimulate cAMP formation, with subsequent activation of cAMP-dependent protein kinase, phosphorylation of CREB and gene transcription required for consolidation of memory traces in several brain areas, including amygdala and hippocampus \[[@B91],[@B106]\]. Dopamine and norepinephrine act in concert to regulate prefrontal cortex function and thereby ensure appropriate behaviour \[[@B103]\]. α~2~-Adrenoceptors agonists enhance performance of tasks requiring prefrontal cortex function while α~1~-adrenoceptor agonists impair prefrontal cortex function \[[@B103]\]. Furthermore, α~1~- and α~2~-antagonists reverse these effects \[[@B103],[@B107]\]. Either excessive activation of α~1~-adrenoceptors or deficient α~2~-adrenoceptor mediated modulation of prefrontal cortical circuits can impair prefrontal cortex function \[[@B103],[@B108]\]. The prefrontal cortex projects to the VTA and locus coeruleus, thereby influencing the firing rate of both dopamine and norepinephrine neurons and impacting on many cognitive processes \[[@B103]\]. The highly specific antagonist of NET1, atomoxetine, is as effective as methylphenidate in treating ADHD \[[@B38],[@B109]\], further emphasizing an important role for the noradrenergic system in the disorder. However, atomoxetine also increases synaptic availability of dopamine in the prefrontal cortex \[[@B110]\] which may contribute to its beneficial effects. Drugs used to treat ADHD symptoms are likely to have different effects on different neurotransmitter systems. Drugs that act on the noradrenergic system, such as atomoxetine, tricyclic antidepressants like the NET1 blocker, desipramine, and α~2~-adrenoceptor agonists such as clonidine and guanfacine, have a different therapeutic time-course compared to psychostimulants. Methylphenidate produces amelioration of ADHD symptoms within 30 minutes and is short-acting whereas noradrenergic drugs have to be administered for longer periods of time before a therapeutic effect is observed, and improvement is sustained for several months \[[@B38]\]. This suggests that noradrenergic drugs cause long-term adaptive changes that are therapeutic. Chronic treatment with desipramine produces a series of changes in presynaptic and postsynaptic α- and β-adrenoceptors and causes long-term downregulation of cortical β-adrenoceptors \[[@B111]-[@B113]\]. This suggests that behavioural improvement can perhaps be achieved by decreased noradrenergic activation of cortical β-adrenoceptors, thereby decreasing noradrenergic function, which is consistent with a hyperactive noradrenergic system and the dopamine/norepinephrine imbalance hypothesis of ADHD. Disturbances in norepinephrine metabolism in SHR are suggested by the finding that tyrosine hydroxylase gene expression is higher in the ventrolateral medulla oblongata of SHR than WKY \[[@B114]\], consistent with elevated norepinephrine concentration in several brain areas of SHR including locus coeruleus, substantia nigra and prefrontal cortex \[[@B79]\]. Increased norepinephrine concentrations in SHR brain is consistent with downregulation of β-adrenoceptors in cerebral cortex of SHR \[[@B115]\]. Furthermore, NET1 function is increased in SHR cerebral cortex \[[@B115]\] which could increase uptake of dopamine into noradrenergic terminals and varicosities and deplete extracellular dopamine in the prefrontal cortex \[[@B116]\]. Evidence suggests that there is an imbalance between dopaminergic and noradrenergic neurotransmission in the prefrontal cortex of SHR \[[@B104]\]. While dopamine release is decreased in SHR prefrontal cortex, norepinephrine concentrations are elevated. The noradrenergic system appears to be hyperactive as a result of impaired α~2~-autoreceptor regulation \[[@B104]\]. Stimulus-evoked (electrically stimulated or K^+^-evoked) release of norepinephrine from prefrontal cortex slices of SHR was no different from that of WKY \[[@B117]\]. However, α~2A~-adrenoceptor mRNA levels were decreased in SHR compared to WKY and α~2~-autoreceptor-mediated inhibition of norepinephrine release was less efficient in SHR than in WKY suggesting that α~2~-adrenoceptor function is impaired \[[@B117]-[@B119]\]. α~2A~-Adrenoceptors are the subtype specifically expressed in the prefrontal cortex, so impaired α~2A~-adrenoceptor function would be expected to impair cognition \[[@B103],[@B120]\]. Decreased α~2~-autoreceptor-mediated regulation of norepinephrine neurons and impaired inhibition of norepinephrine release may be particularly disruptive to the function of target structures when the firing rate of locus coeruleus neurons is high, causing excessive spillover of norepinephrine. Repeatedly increased release of norepinephrine from sympathetic nerve terminals could give rise to the stress-dependent \[[@B60]\] development of hypertension in SHR. Expression of the gene encoding G~iα~the G-protein subunit that inhibits cAMP formation from ATP by adenylyl cyclase is increased in SHR aorta at 2 weeks of age, possibly reflecting an attempt by a target organ to decrease the effect of increased norepinephrine release from sympathetic nerve endings. Poorly controlled norepinephrine release could also give rise to excessive activation of α~1~-adrenoceptors in the prefrontal cortex impairing its function. Other noradrenergic terminal areas in the central nervous system may be similarly affected. These findings suggest that the noradrenergic system is hyperactive in SHR, particularly in response to stress, and supports the hypothesis that there is an imbalance between norepinephrine hyperfunction and dopamine hypofunction in ADHD. ### Serotonin Brain serotonin (5-hydroxytryptamine, 5-HT) function has been suggested to be altered in SHR \[[@B121]\]. Higher serum testosterone, and lower amygdala serotonin content has been associated with a mutation in the non-pseudoautosomal region unique to the Y-chromosome of SHR \[[@B122]\]. Administration of a serotonin transporter inhibitor, fenfluramine, evoked less prolactin secretion in SHR than WKY \[[@B121]\]. Acute administration of the selective serotonin reuptake inhibitor, citalopram, reduced hyperactivity of SHR in an elevated plus-maze \[[@B123]\]. However, there was no difference between SHR and WKY in midbrain, hippocampal, or striatal serotonin concentration or serotonin uptake kinetics \[[@B123]\]. In addition, stressors released serotonin in the locus coeruleus of SHR and WKY rats to the same extent \[[@B124]\] and 5-HT~2C~receptor function was reported to be unaltered in SHR compared to WKY \[[@B125]\]. These findings do not support a role for serotonin in the aetiology of ADHD symptoms in SHR. ### Glutamate In addition to decreased autoreceptor-mediated inhibition of norepinephrine release from SHR prefrontal cortex slices, glutamate activation of AMPA receptors caused greater release of norepinephrine from SHR prefrontal cortex slices than WKY \[[@B126],[@B127]\]. Glutamate is present in micromolar concentrations in the extracellular space outside the synaptic cleft and regulates tonic dopamine concentration in the extracellular fluid \[[@B128]-[@B132]\]. Dopamine release is increased by activation of AMPA receptors in rat striatum \[[@B133],[@B134]\]. Glutamate activation of NMDA receptors upregulates DRD1 function by a direct protein-protein interaction at the carboxy terminals of both receptors \[[@B135]\]. As suggested by Seeman and Madras \[[@B136]\], the common defect in ADHD could be decreased extracellular dopamine levels. This deficiency could result from increased expression of DAT1, impaired dopamine synthesis, impaired release, or impaired regulation of extracellular dopamine by glutamate afferents from the prefrontal cortex, hippocampus, or amygdala \[[@B102]\]. However, in vitroactivation of AMPA receptors caused similar fractional release of dopamine from SHR and WKY nucleus accumbens core \[[@B134]\]. Unlike WKY, glutamate-stimulated release of dopamine from the shell subdivision of SHR nucleus accumbens was significantly lower than from the core subdivision of SHR, suggesting that the shell may be particularly vulnerable to disturbances in dopamine release \[[@B134]\]. Neural circuits that use glutamate as a neurotransmitter are modulated by dopamine and norepinephrine. Future studies should be aimed at investigating glutamate function in the brains of SHR. In particular, measuring glutamate release in the prefrontal cortex and nucleus accumbens using in vivomicrodialysis during reinforcement and extinction (hypothesized to be the major causative factors underlying ADHD \[[@B4]\]) may provide useful information concerning afferent glutamate input to these brain areas as a result of the additional demands of such tasks. ### Psychostimulants Psychostimulants are the most effective drugs used in the treatment of ADHD and provide a powerful means to gain insight into the underlying disturbances of ADHD. d-Amphetamine and methylphenidate reduced the ADHD-like behaviour of SHR \[[@B137]\] \[Sagvolden, unpublished; Russell, unpublished\]. The increase in DRD1 density observed in SHR striatum is reversed by methylphenidate treatment suggesting that psychostimulants reduce ADHD-like behaviour of SHR by increasing dopamine activation of DRD1 \[[@B51],[@B81],[@B82]\] thereby enabling dopamine-mediated LTP and reinforcement mechanisms to take place. Psychostimulants changed the performance of SHR in fixed-interval schedules of reinforcement of bar-presses by lengthening the delay-of-reinforcement gradient. However, WKY performance changed to a greater extent than SHR, suggesting that the effect of psychostimulant drugs was less pronounced in SHR than in WKY \[[@B138]\]. The reduced reactivity to psychostimulants may be associated with abnormalities in DAT1 gene expression \[[@B50],[@B51]\] \[Mill et al., unpublished\] and dopamine hypofunction as a result of adaptation to increased availability of dopamine at an early stage of development with subsequent reorganization of neural mechanisms that control VTA dopamine neuron function \[[@B139]\]. Further support for the hypothesis that regulation of midbrain dopamine neurons is altered in SHR was provided by the fact that repeated administration of methylphenidate (2.5 mg/kg) elicited locomotor sensitization i.e. increased locomotor response to the same dose of methylphenidate 3 days after cessation of treatment in Sprague-Dawley and WKY rats but not in SHR who were unaffected by the drug \[[@B140]\]. In vitrofindings provided further support, where methylphenidate released significantly less dopamine from SHR nucleus accumbens slices than WKY \[[@B58]\], and chronic methylphenidate treatment (3 mg/kg for 2 weeks) increased endogenous dopamine activation of DRD2 in WKY striatum but did not alter DRD2 function in SHR probably because DRD2 were already up-regulated in SHR and no longer responsive to increases in extracellular dopamine \[[@B141]\]. These results suggest that neural circuits have been altered in SHR and that psychostimulant drugs affect SHR and WKY brains differently. This finding stresses the importance of studying animal models of ADHD as it shows that it cannot be assumed that drugs will have the same effect in children with ADHD and controls. Other animal models of ADHD --------------------------- Several other animal models of ADHD have been proposed. These models were developed through genetic manipulation, exposure to toxins, rearing in social isolation, or interference with neurochemical systems. However, several do not satisfy the criteria for animal models of ADHD \[[@B45]\] and have therefore been excluded from the present review. These include the Naples high-excitability rat (NHE), WKHA rat, acallosal mouse, hyposexual rat, PCB-exposed rat, lead-exposed mouse, and rat reared in social isolation \[[@B45]\]. The reasons for exclusion are briefly as follows: NHE are hyperreactive in a novel environment, they are not hyperactive or impulsive in a familiar environment, and they have not been shown to be impaired in sustained attention \[[@B142],[@B143]\]. WKHA rats are hyperactive but they are not impulsive \[[@B47],[@B144]-[@B146]\]. The acallosal mouse becomes hyperactive over time and shows impaired acquisition of conditioned learning tasks \[[@B147]\]. However, impulsiveness decreases over time with repeated testing which is not characteristic of ADHD. Polychlorinated biphenyls (PCBs) administered either pre- or postnatally cause hyperactivity in rats but do not impair sustained attention \[[@B148]-[@B150]\]. Postnatal exposure of infant mice to lead causes ataxia and hyperactivity \[[@B151]\] but lead produces many other complications that would exclude a diagnosis of ADHD. Rat pups reared in social isolation display hyperactivity in a novel environment and increased errors of omission and perseveration \[[@B152]\]. However, these rats are not impulsive and are unimpaired in measures of task acquisition in the 5-choice serial reaction time (5-CSRT) test of sustained attention. In addition, children reared in social isolation would not be diagnosed as ADHD. ### Coloboma mutant mouse The SNAP-25 deficient mouse mutant coloboma (Cm/+) is of interest to ADHD because SNAP-25 polymorphisms have been associated with the disorder \[[@B153],[@B154]\]. SNAP-25 regulates membrane trafficking and is involved in the release of all neurotransmitters as well as regulating translocation of proteins (e.g. NMDA receptor subunits) to the cell membrane. Altered expression of SNAP-25 will therefore have diffuse effects on neuronal function. The SNAP-25 deficient mouse mutant coloboma displays spontaneous hyperactivity \[[@B155]\] but lacks impulsiveness and has not been shown to have problems with sustained attention. Although the SNAP-25 deficient mouse does not model ADHD symptoms specifically, it may nevertheless serve as a useful model of non-specific brain dysfunction such as minimal brain disorder (MBD). Depolarization-evoked (K^+^-evoked) release of glutamate from cortical synaptosomes is reduced in the coloboma mouse \[[@B156]\]. DRD2 expression is increased in the VTA and substantia nigra, suggesting increased inhibition of dopamine neuron firing rate \[[@B157]\]. Dopamine release and dopamine metabolites (DOPAC and HVA) are decreased in the striatum of coloboma mice which is consistent with decreased dopamine release and turnover \[[@B156],[@B158]\] i.e. a hypofunctional dopaminergic system, similar to SHR. Striatal DRD1 and DRD2 expression is unaltered in the coloboma mouse \[[@B157]\]. Tyrosine hydroxylase expression is unaltered in VTA and substantia nigra whereas tyrosine hydroxylase and α~2A~-adrenoceptor expression is increased in the locus coeruleus of the coloboma mouse \[[@B157]\]. Noradrenergic function appears to be increased since experimental depletion of norepinephrine by DSP-4 reduces the hyperactivity of coloboma mice \[[@B159]\]. This suggests that motor activity in coloboma mice is caused by a hyperactive noradrenergic system but the hyperactivity is not completely abolished by depletion of norepinephrine, suggesting that additional factors contribute to the mutant phenotype \[[@B159]\], perhaps the imbalance between noradrenergic hyperfunction and dopamine hypofunction as suggested for SHR. ### 6-OHDA-Lesioned Rat Neonatal 6-OHDA-lesioned rats are not impulsive but they display hyperactivity and impaired learning in a spatial discrimination task, which improves after methylphenidate or d-amphetamine treatment \[[@B160]-[@B163]\]. Rat pups lesioned on postnatal day 1 displayed hyperactivity in adulthood \[[@B162]\]. They showed an initial decrease in spontaneous motor behaviour when placed in a novel environment, but after repeated testing their activity was increased relative to controls \[[@B162]\]. Hyperactivity was accompanied by decreased dopamine \[[@B162]\], increased DRD4 \[[@B164]\], and increased serotonin transporter binding in striatum but not cerebral cortex \[[@B165]\]. Hyperactivity was not altered by DAT1 inhibitors but was greatly reduced by DRD4 antagonists as well as inhibitors of SERT1 and NET1 \[[@B160],[@B161],[@B164],[@B166]\]. These findings suggest that psychostimulants reduce hyperactivity of 6-OHDA lesioned rats not by inhibiting DAT1 but by inhibiting norepinephrine and serotonin transporters. Inhibition of NET1 would reduce dopamine uptake into noradrenergic terminals in several brain areas including prefrontal cortex and nucleus accumbens. ### DAT-Knockout Mouse DAT-knockout (DAT-KO) mice lack the gene that encodes DAT. These mice have been suggested as a model for ADHD because they are hyperactive in novel situations \[[@B167]-[@B169]\], have impaired extinction of responses in operant food reinforcement tasks \[[@B170]\]. They are also impaired in learning and memory tasks \[[@B168],[@B169]\]. Impulsiveness has not been systematically investigated in DAT-KO mice. Although the absence of DAT is an extreme model of reduced midbrain DAT binding in adolescents with ADHD \[[@B36]\] it also contrasts with several studies that found increased DAT in striatum of ADHD children and adults \[[@B10],[@B15],[@B35]\]. The DAT-KO mouse nevertheless provides useful information concerning the neurobiological consequences of impaired DAT function. Released dopamine is cleared at a slow rate giving rise to a 5-fold elevation of extracellular tonic dopamine in the striatum of DAT-KO mice i.e. a hyperdopaminergic state \[[@B171]\]. Electrically stimulated release of dopamine is decreased, suggesting that phasic release of dopamine is reduced i.e. the dopamine system is hypofunctional \[[@B171]\] similar to SHR and the coloboma mouse. However, unlike SHR, striatal DRD2 autoreceptors controlling dopamine synthesis are nonfunctional while DRD1 and DRD2 are downregulated by approximately 50% in the striatum of DAT-KO mice \[[@B171]\]. Hyperactivity in the DAT1 knock-out mouse might be the result of increased dopamine tone or decreased phasic dopamine release with consequently impaired activation of postsynaptic DRD1 required for LTP (and LTD) to produce changes in synaptic strength necessary for associative learning and reinforcement of appropriate behaviour. Whereas specific inhibitors of NET1 or DAT1 did not affect DAT-KO hyperactivity, inhibitors of the SERT1 as well as drugs that activate the serotonergic system, such as serotonin receptor agonists and serotonin precursors, dramatically reduced hyperactivity \[[@B168]\]. DAT-KO mice provide convincing evidence that hyperactivity induced by high extracellular levels of dopamine can be reduced by enhancing serotonergic tone i.e. psychostimulants do not act via DAT1 to reduce hyperactivity in this model \[[@B168]\]. Interestingly, antagonists of the 5-HT~2A~receptor reverse the behavioural deficits of DAT-KO mice \[[@B172]\] and polymorphisms of the 5-HT~2A~receptor gene have been associated with ADHD \[[@B173],[@B174]\]. While this model provides invaluable insight into possible mechanisms of psychostimulant action, the relevance of these findings to ADHD is not clear since serotonin reuptake inhibitors are of limited value in treating ADHD as one of the side effects of serotonin uptake inhibitors is stimulation of motor activity \[[@B38],[@B168]\]. However, serotonin acts on a large number of receptor subtypes each with different spatial location and behavioural effects. Evidence obtained with the DAT-KO mouse suggests that specific antagonists of the 5-HT~2A~receptor deserve further investigation. Interestingly, the DAT-knockdown (DAT-KD) mouse has been suggested as a model for obsessive compulsive disorder and Tourette\'s syndrome \[[@B175]\]. DAT knockdown, achieved by reducing DAT promotor strength, reduces adult DAT expression to 10% of wild-type levels and raises extracellular dopamine levels in the striatum to 170% of wild-type controls \[[@B175]\]. Hyperdopaminergic DAT-KD mice displayed excessive sequential stereotypy reflected as a complex serial pattern of grooming actions becoming more sequentially rigid and persistent. This type of behaviour is not characteristic of ADHD but may serve as a model for Tourette\'s and obsessive compulsive disorder \[[@B175]\]. Consistent with enhanced dopamine activation of DRD1 receptors being responsible for the excessively rigid serial pattern of instinctive grooming behaviour, DRD1 agonists produced similarly enhanced sequential stereotypy of syntactic grooming chains \[[@B175]\]. DAT-KD mice also tend to be hyperactive, to walk in perseverative straight paths, and to over-pursue certain incentive stimuli, consistent with obsessive compulsive disorder. ### Poor 5-CSRT task performer Rats that are selected for poor performance when trained in the 5-CSRT task provide a useful model of ADHD in that they are selected for deficient sustained attention, they show poor choice accuracy towards the end of testing sessions, and they demonstrate impulsiveness (premature responding) \[[@B176],[@B177]\]. Methylphenidate treatment improved accuracy and reduced impulsiveness (at low doses) in poor performers \[[@B177]\]. Poor 5-CSRT task performers are not hyperactive and therefore may serve as a model of the inattentive subtype of ADHD. In normal animals, response accuracy is adversely affected by activation of serotonin 5-HT~1A~receptors \[[@B178]\], while activation of 5-HT~2A~receptors increases the number of premature responses, suggesting that increased serotonin tone could be responsible for impulsivity of poor performers \[[@B179]\]. This is consistent with 5-HT~2A~receptor antagonists reversing the behavioural deficits of DAT-KO mice \[[@B172]\]. Evidence supports a role for dopamine in regulating the level of performance in the 5-CSRT task. In normal animals, d-amphetamine-stimulated release of dopamine in the nucleus accumbens caused a dose-dependent increase in premature responding \[[@B178]\]. Microinfusion of a DRD1 agonist into the medial prefrontal cortex selectively impaired the accuracy of attentional performance in high performers in the 5-CSRT task \[[@B180]\]. In contrast, microinfusion of the DRD1 agonist into the medial prefrontal cortex of poor performers enhanced the accuracy of attentional performance; a low dose increased the speed of making correct responses \[[@B180]\]. This finding once again emphasizes the need to study animal models of ADHD rather than normal animals in order to gain insight into the mechanisms that underlie the beneficial effects of drugs used to treat children with ADHD. Evidence suggests that the nervous system is altered in the 5-CSRT model of ADHD. These results suggest that dopamine function is reduced in poor performers of the 5-CRST task and that 5-HT~2A~antagonists may be beneficial in the treatment of ADHD. ### Anoxia in Neonatal Rat Anoxia increases the risk of ADHD \[[@B181]\]. Neonatal anoxia caused a sequence of acute and persistent neurochemical changes in rat monoaminergic systems as well as transient hyperactivity and spatial memory impairment that persisted into adulthood \[[@B182]-[@B184]\]. Acute anoxia caused a transient decrease followed by an increase after 1 week in cerebellar norepinephrine levels \[[@B183]\]. At the same time, serotonin levels decreased while its metabolite, 5-hydroxyindoleacetic acid (5-HIAA), increased \[[@B183]\]. Striatal dopamine and metabolite concentrations decreased and then dopamine metabolites increased post ischaemia \[[@B183]\]. The increase in serotonin and dopamine metabolites persisted into adulthood, suggesting that dopamine turnover is increased. Tyrosine hydroxylase mRNA levels were increased in VTA and substantia nigra of perinatally asphyxiated rats suggesting increased dopamine synthesis consistent with increased turnover. However, DRD1 and DRD2 mRNA levels were increased in the striatum suggesting impaired release of dopamine \[[@B185]\]. These findings demonstrate the complex temporal sequence of compensatory changes that occur in monoaminergic systems following perinatal insult to the nervous system and implicate all three monoaminergic systems in spatial memory impairment. Insight provided by animal models of ADHD ----------------------------------------- One of the most important findings is the fact that animal models of both inattentive and hyperactive/impulsive subtypes of ADHD respond differently to psychomotor drugs when compared to controls suggesting that they have altered neurotransmitter systems in the brain. This emphasizes the need to study animal models of ADHD rather than normal animals in order to gain insight into the mechanisms that underlie the beneficial effects of drugs used to treat children with ADHD. Neuroadaptations leading to psychostimulant drug addiction involve the same glutamate-dependent cellular mechanisms that enable learning and memory \[[@B87]\]. Similarly, neural mechanisms implicated in behavioural disturbances of animal models of ADHD are consistent with altered dopamine and/or norepinephrine mediated modulation of glutamate-dependent cellular mechanisms that enable learning and memory. The various animal models of ADHD focus on different aspects of ADHD symptomatology and provide unique insights into ADHD neurobiology. They also emphasize the close interconnection between serotonergic, noradrenergic and dopaminergic systems. Changes in any one system can alter the function of the other monoaminergic systems and alter the underlying neural circuits that control behaviour. There is convincing evidence of a dysfunctional dopaminergic system in several models of ADHD. The dopaminergic system appears to be hypofunctional in SHR, the coloboma mutant mouse, 6-OHDA lesioned rat, DAT-KO mouse (although extracellular hyperdopaminergia may also contribute to its behaviour) and poor performers in the 5-CSRT task. There is no convincing evidence to suggest that the underlying disturbance is primarily located in the serotonergic or other neurotransmitter system, although amelioration of ADHD symptoms may be partly mediated by drugs acting on the noradrenergic and serotonergic systems. The dynamic developmental theory of ADHD \[[@B4]\] explains how behavioural changes associated with ADHD may result from deficient reinforcement and extinction processes as a result of a hypofunctional dopaminergic system and a steeper delay gradient in ADHD \[[@B4],[@B186]\]. Impaired function of the mesolimbic dopamine system can cause ADHD symptoms of delay aversion, hyperactivity in a familiar environment, impulsiveness, deficient sustained attention, increased behavioural variability and failure to extinguish previously acquired behaviour \[[@B4]\]. Deficient mesocortical dopamine-mediated modulation of prefrontal cortical circuits can impair behavioural planning (executive function). Hypofunction of the nigrostriatal system would impair dopamine modulation of cortico-striato-thalamo-cortical circuits that control motor function giving rise to apparent developmental delay, clumsiness, and neurological \"soft signs\" \[[@B4]\]. The development of ADHD symptoms could be the result of inappropriately increased levels of dopamine at an early stage of development causing compensatory changes that subsequently give rise to hypofunctional dopamine neurons and impaired reinforcement/extinction mechanisms \[[@B4],[@B139],[@B186]\]. Children who have been exposed to elevated levels of brain dopamine prenatally as a result of mothers taking drugs of abuse, exhibit ADHD-like behaviour \[[@B187]\]. Exposure to drugs of abuse increases the extracellular dopamine concentration which reduces autoreceptor inhibition of VTA dopamine neurons and increases glutamate-driven activity of dopamine-containing neurons \[[@B187]-[@B190]\]. The consequences of increased extracellular dopamine is demonstrated in the DAT-KO mouse which loses the normal inhibitory effect of DRD2 on dopamine neuron firing rate in the VTA and substantia nigra, as well as dopamine release-regulating DRD2 function in the striatum \[[@B190]\]. The mechanism is suggested to involve increased AMPA receptor-mediated excitatory transmission and decreased inhibitory metabotropic glutamate receptor function in VTA dopamine neurons \[[@B187],[@B188],[@B191]\]. Increased activation by glutamate initially causes sensitization of VTA dopamine neurons with subsequent adaptations in the nucleus accumbens \[[@B191]\]. The increased glutamate drive is suggested to lead to pathophysiological consequences resulting from the high intracellular concentrations of Ca^2+^which gives rise to impaired function of VTA dopamine neurons and adaptation \[[@B191]\]. Similarly, ADHD symptoms may result from adaptation to initially increased extracellular dopamine in the VTA of the midbrain as a result of genetic and environmental effects at a very early stage of development giving rise to increased glutamate drive and subsequent loss of function of dopamine neurons. The dynamic developmental theory of ADHD \[[@B4]\] explains how ADHD symptoms, including problems with sustained attention, can result from impaired dopamine function giving rise to a steeper delay gradient and poor stimulus control of behaviour when reinforcers are infrequent. Sustained attention is also controlled by noradrenergic projections from the locus coeruleus to the prefrontal cortex. There is considerable evidence to suggest that the noradrenergic system is poorly controlled by α~2~-autoreceptors in SHR, particularly at high norepinephrine release rates. This may be seen as hyperactivity of the noradrenergic system, especially when locus coeruleus neurons are stimulated in states of increased arousal. Increased release of norepinephrine from sympathetic nerve endings can give rise to the development of hypertension. Impaired regulation of norepinephrine release in the prefrontal cortex could give rise to ADHD-like symptoms. It is interesting to note that although the various animal models have quite different origins, they have in common either increased or decreased tonic dopamine and/or decreased phasic release of dopamine. Some models also display poor regulation of locus coeruleus neurons and noradrenergic hyperactivity. These alterations may be primary or may reflect compensatory changes in response to more basic disturbances in neurotransmission, such as deficient SNAP-25 or impaired Ca^2+^signalling. Future research should focus on determining the precise effects of psychostimulant drugs on the nervous system of animal models of ADHD. Psychostimulants affect animal models of ADHD and normal animals differently, suggesting that the nervous system of the ADHD model has undergone adaptive change which alters the effects of drugs used to treat ADHD. DAT-KO mice provide convincing evidence that psychostimulant drugs reduce hyperactivity by enhancing serotonin tone. Antagonists of the HT~2A~receptor are reported to reverse ADHD-like behaviour in some animal models. To test whether this is a general finding of relevance to ADHD, other animal models such as the SHR should be treated with HT~2A~antagonists to confirm whether their ADHD-like behaviour is similarly reversed. Once this is established, further testing to see if the beneficial effects of HT~2A~antagonists and psychostimulant drugs can be prevented by blockers of postsynaptic DRD1 receptors would be of interest, in order to test the dynamic developmental theory of ADHD \[[@B4]\] proposal that ADHD-like symptoms can be explained by deficient dopaminergic function and that psychostimulants enhance phasic dopamine release. ADHD is a heterogeneous disorder, suggested to result from combinations of genetic and environmental factors. Animal models can mimic only certain aspects of the complex symptomatology of ADHD and at best provide feasible hypotheses regarding the underlying causes of specific aspects of ADHD behaviour. These hypotheses can then be tested in the clinic. Animal models can also be used to test potential drugs for the treatment of ADHD. Future research on animal models of human disorders will undoubtedly promote a better understanding of the contribution of specific neurobiological factors to behavioural components like attention, reinforcement and extinction that seem to be important for understanding ADHD. Acknowledgements ================ This paper is part of the project \"Attention-Deficit/Hyperactivity disorder (ADHD): from genes to therapy\" conducted at the Centre for Advanced Study (CAS) at the Norwegian Academy of Science and Letters, in Oslo in 2004/05. The authors wish to thank CAS, the University of Cape Town and South African Medical Research Council for support.	PubMed Central	2024-06-05T03:55:59.610866	2005-7-15	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180819/", "journal": "Behav Brain Funct. 2005 Jul 15; 1:9", "authors": [ { "first": "Vivienne A", "last": "Russell" }, { "first": "Terje", "last": "Sagvolden" }, { "first": "Espen Borgå", "last": "Johansen" } ] }
PMC1180820	Background ========== The amount of scientific biomedical literature readable by computer programs is overwhelming. For example, PubMed \[[@B1]\] contains about 7.5 million article abstracts. Therefore automatic literature-mining methods can be exploited in order to retrieve relevant information (for recent thorough reviews of related work in Bio-NLP, see e.g. \[[@B2],[@B3]\]). For example, several algorithms have been developed for extracting information about protein-protein interactions from the biomedical literature \[[@B4]-[@B10]\]. The problem ----------- In order to find the relations between biological or chemical entities, first the names of the entities have to be recognized in a reliable way. There has been a significant amount of effort to do that automatically \[[@B11]-[@B20]\]. A large standardized domain corpus helps to consolidate the research efforts. The GENIA corpus \[[@B21]\] has been commonly used in biomedical named entity recognition. The state-of-the-art systems have recently been compared, for example, in Kim et al. \[[@B22]\] using the GENIA corpus. The task of named entity recognition can be divided in two subtasks, the identification of entities, that is, determining the boundaries of the named entities, and their classification into proper classes. The problem of finding the class the entity belongs to can be treated as a word sense disambiguation (WSD) task, which, on its own, is an essential part of natural language processing (see e.g. Manning and Schütze \[[@B23]\] for more information). The entities in biomedical text are highly ambiguous. For example, it is common that a gene has the same name as the protein it codes for. In the following three sentences from the GENIA corpus, the occurrences of BZLF1 are a protein, a gene, and an RNA, respectively: (1) Expression of either BZLF1 or BRLF1 triggers expression of\... (2) \... DNA in lymphoblastoid cell lines induced by transfection with BZLF1. and (3) \... lysis of certain HLA B8+ LCL targets was associated with the abundance of BZLF1 transcripts. Similar ambiguity is illustrated in the two sentences given by Hatzivassiloglou et al. \[[@B24]\]: By UV cross-linking and immunoprecipitation, we show that SBP2 specifically binds selenoprotein mRNAs both in vitro and in vivo. The SBP2 clone used in this study generates a 3173 nt transcript (2541 nt of coding sequence plus a 632 nt 3\' UTR truncated at the polyadenylation site). The occurrence of SBP2 is a protein in the first sentence, whereas the occurrence of SBP2 in the second sentence is a gene. In the same study, a domain corpus was annotated by three biology experts. The three experts unanimously agreed only in 78% of the cases, each name being classified as either a gene, protein or mRNA. This low rate of inter-annotator agreement suggests that the task is relatively difficult even for human experts, reflecting the inherent complexity of the domain. However, the study does not analyse more closely the reasons that lead to annotation disagreements. In this paper, we consider the disambiguation of the sense \"gene\" or \"protein\" when the name is not disambiguated explicitly by the author with the word \"gene\" or \"protein\" (e.g. \"SBP2 gene\"). This task is important, because the release of the human genome and large scale functional genomics studies and methods have made it important to be able to find information from literature specifically for proteins and the corresponding genes. However, database searches provide a lot of hits among which the correct and important articles have to be sorted manually. Therefore, for example, in data mining related to proteomics the scientists could save much time if they could direct their search only to proteins. Much of the ambiguity in biomedical text is caused by inconsistent or non-existent naming conventions. For example, there exist Drosophila gene names such as ringand arcthat can be confused with their ordinary meanings. Manual analysis of a small set of abstracts returned by PubMed for the query ring and drosophilashows that the word ringappears in its gene/protein sense in about 30% of the cases, both capitalized and non-capitalized. Similarly, the word arcis ambiguous and appears in about 75% of the cases in its gene/protein sense, again both capitalized and non-capitalized. In both cases, only abstracts regarding Drosophila were considered, thus the two example words retain their ambiguity even in the sublanguage of articles concerning Drosophila. In contrast, some other gene/protein names, such as, taxdo not retain their ambiguity in the sublanguage: all occurrences of taxin the GENIA corpus refer to its gene/protein sense. Another major source of ambiguity in scientific biomedical text are abbreviations, which are widely used and therefore are very important to be identified correctly in natural language processing applications \[[@B25]\]. There have already been applications of word sense disambiguation methods in the field of scientific biological text processing. Hatzivassiloglou et al. \[[@B24]\] disambiguated names of genes, proteins and RNAs using a Naive Bayes classifier. Previously we have developed a method named here Weighted Additive Classifier (WAC) and applied it to the problem of gene/protein name disambiguation \[[@B26]\]. Liu et al. \[[@B27]\] disambiguated abbreviations from Medline abstracts using a Naive Bayes classifier. Yu et al. \[[@B28]\] achieved better results for the same task using Support Vector Machines (SVMs) with the one sense per discourse hypothesis. Furthermore, a system developed by Podowski et al. \[[@B29]\] assigns gene names to their LocusLink IDs in previously unseen abstracts. Lee and Ng \[[@B30]\] performed a comparison of several supervised learning algorithms for WSD tasks and in their study, SVMs were confirmed to have the best performance. SVMs have also been applied in biomedical WSD (see e.g. Yu et al. \[[@B28]\]) as well as in biomedical named entity recognition \[[@B20],[@B31]-[@B35]\]. Furthermore, in the COLING-2004 JNLPBA shared task of Bio-Entity Recognition \[[@B22]\], five studies \[[@B36]-[@B40]\] used SVMs either alone or combined with other algorithms. In this paper, we apply SVMs and as baselines, we consider the Naive Bayes and WAC classifiers. The studies mentioned above use narrow context windows and focus mainly on studying different features such as orthographical, morphological, lexical, contextual, part-of-speech, head-noun, and name-alias features. We, in contrast, focus on context representations that use distance of the words from the ambiguous name in addition to the context words themselves. We evaluate the methods on the GENIA data set (see e.g. Collier et al. \[[@B21]\]), using the area under ROC curve (see e.g. \[[@B41]\]) as a performance measure. Support Vector Machines ----------------------- SVMs can be used to classify multidimensional data into two classes. SVMs were introduced by Boser et al. \[[@B42]\]. Thorough presentations of SVMs are given by Burges \[[@B43]\] and Vapnik \[[@B44]\], for example, and in the Methods section we give a concise introduction to SVMs. In a binary classification task, the training set consists of data points which are labeled as positive or negative. In our case, the training data points are the contexts of the ambiguous names. The positive and negative labels denote genes and proteins, respectively. In order to improve linear separability, the data points are mapped from the input space to a new feature space before they are used for training or for classification. The mapping is done implicitly by a so-called kernel function, which computes the similarity of two data points in the feature space. The choice of an appropriate kernel function is a nontrivial problem, but there are certain standard kernel functions which are frequently used. Kernels and the SVM itself also have certain parameters which have to be adjusted in order to make the SVM classifier work in the best possible way. Contribution of this work ------------------------- In short, we considered application of SVMs to the gene versus protein name disambiguation problem in abstracts of biomedical articles. While other studies focus mainly on studying different features, our work primarily considers context representations. We resolve the ambiguous names using their context which spans up to the whole abstract, in contrast to other previous applications of SVMs which typically use narrow context windows. To improve the performance of conventional SVMs and accommodate the wide context span, we adopted a weighting scheme introduced by Ginter et al. \[[@B26]\] that exploits the information about the distances of the words from the name to be disambiguated, and adjusted the scheme for the SVM classifier. We carefully searched for the best parameter values of SVMs and kernel functions using grid optimization as suggested by Hsu et al. \[[@B45]\], and we also performed a similar search for the parameters of the proposed weighting scheme. Finally, we measured the performance of both conventional and weighted SVMs together with two baseline methods, and showed that the performance improvement was statistically significant. Results and discussion ====================== We experimented with the protein versus gene name disambiguation problem using conventional SVMs with linear, Gaussian, as well as second and third degree polynomial kernels. Also, we tested SVMs using different kernels augmented with the proposed weighting scheme. As additional baseline classifiers, we used the Weighted Additive Classifier, which also uses contextual weighting, and the Naive Bayes classifier. These methods and their parameters to which we refer in this section are described in detail in the Methods section. In the following, we first discuss the weighting scheme and the reasons why its use is beneficial. Then, we present how the data was generated and preprocessed. Finally, we present the performance measure used in the experiments, describe the experimental setting and the results of the parameter estimation and the final validation. Contextual weighting -------------------- The training data points are vectors of word frequencies in the context in which the names to be disambiguated were found. The basic SVMs with any kernel use only the word frequencies and do not take into consideration the distances of the words with respect to the position of the name to be disambiguated. However, the distance information seems intuitively to be important, and therefore we apply a weighting scheme that incorporates this information into the context representation used by SVMs. The weighting scheme models the distances of the words from the ambiguous name, while the information whether the words are before or after the ambiguous name is not considered. The weight of a context word at the distance dis given by d^-λ^+ β, where the parameters λand βare used to control the effect of the distances of the words from the name to be disambiguated. A more detailed explanation of the weighting scheme is presented in the Methods section. We now discuss some possible reasons for the weighting scheme achieving a statistically significant gain in classification performance. Yarowsky \[[@B46]\] argues that the effect of context words is strongest for immediately adjacent words, and weakens with distance. This phenomenon is called the one-sense-per-collocationprinciple. Yarowsky also considers the one-sense-per-discourseprinciple, that is, all instances of an ambiguous word tend to have the same sense within one discourse unit, the article abstract in our case. In that case also distant words can help in disambiguation. One-sense-per-discourse is, however, presumed to be a weaker hypotheses, which should be overridden when the local evidence is strong. In order to study the tenability of these hypotheses in our data, we estimated the following conditional probabilities of the name to be disambiguated to have another instance of an ambiguous name in its context. For each distance from the name to be disambiguated, we estimated the conditional probability that there is a word in the context at that distance and the word is another ambiguous name with the same sense, as well as the conditional probability that the word is a name with the opposite sense. These probabilities are illustrated in Figure [1](#F1){ref-type="fig"}, where the solid line denotes the probability of an occurrence of a name with the same sense and the dashed line denotes the probability of the opposite sense. At close distances (\<6), the probability of the same sense turned out to be high and decreasing with distance, whereas the probability of the other sense behaved in the opposite way. In the proposed method, the words in the area of influence of the one-sense-per-collocation principle, that is, the words at close distances, have more weight than the long distance words and these weights are controlled by the parameter λ. On the other hand, at long distances, the probabilities of the same and the opposite senses settled down to 0.08 and 0.02, respectively, indicating that mostly the one-sense-per-discourse principle holds. Therefore, when the close context is unable to make a strong decision, the information of the long distance words may be useful. This effect is controlled by the βparameter, which balances the influence of the one-sense-per-discourse principle compared to the one-sense-per-collocation principle. Note that one-sense-per-discourse does not have to hold strictly, because the information can be useful if there are on average more instances of the names with the same sense than with the opposite sense in the far context. Both near and far context words are important when deciding the sense of a name. For example, verbs like \"activate\" or \"phosphorylate\" are often found around protein names, whereas verbs like \"express\" or \"transcribe\" may be found around gene names. Similarly, head nouns, such as expression, are also highly indicative of the sense. These words may be located near to the name to be disambiguated, being strong indicators of its sense. As shown above, ambiguous names in the abstract are more likely to be of the same sense and therefore the words around the other ambiguous names are partly indicative about the sense of the name to be disambiguated. Since other ambiguous names can occur at any position of the abstract, it is beneficial to use long context. Descriptions of experimental conditions can indicate one sense common to all of the names in the abstract. For example, \"yeast two-hybrid\" indicates protein-protein interaction finding, while \"microarray\" relates to gene experiments. Further, the occurrence of a distant coreference, for example, between the full form of an ambiguous name and its abbreviation, in the context of the ambiguous name may provide distant words indicative of the correct sense. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Conditional probabilities of having other instances of ambiguous names in the context of the name to be disambiguated.The X-axis denotes the distance in words from the name to be disambiguated and Y-axis is the probability. The solid line denotes the probability that if there is a word in the context at the given distance, the word would be another ambiguous name having the same sense with the name to be disambiguated and the dashed line denotes the corresponding probability for the opposite sense. ::: ![](1471-2105-6-157-1) ::: The phenomena discussed above are highly data dependent. However, the proposed weighting scheme models them if the weighting scheme is equipped with the optimal values of the parameters λand βfound in the training phase. Data and its preprocessing -------------------------- The data set considered in this paper is constructed as follows. We obtained the evaluation data set for the COLING-2004 JNLPBA shared task of Bio-Entity Recognition \[[@B22]\], which is derived from the GENIA corpus \[[@B21]\], a standard corpus for biomedical named entity recognition, by conflating the original 36 classes into five classes (DNA, RNA, protein, cell line and cell type) of which we only use two classes, namely, DNA and protein. The data set consists of 2000 hand-annotated abstracts and contains 30269 protein examples and 9533 DNA examples. Average number of words in the abstracts is 246. Naturally, not all names are truly ambiguous in all domains and corpora. In these cases a classifier could gain by simply memorizing the names. We made an experiment in which we used only the ambiguous gene and protein names as data points and the performance obtained was over 95%. Using context words as extra features improved the performance only by 0.3% percentage points, with optimal context span 1 found experimentally. The difference was not statistically significant. To assess the performance in a more general setting and to avoid the overfitting effect of memorizing, we do not include the instance of the term to be disambiguated into its context, as the purpose of this paper is to study context-based name disambiguation. Note also that memorizing the names does not help the classifier to disambiguate names that do not exist in training data, for example, names introduced only recently. The text was further preprocessed by removing stop words and stemming the words with the Porter stemming algorithm \[[@B47]\] (stemming and stop-word removal are discussed in the Methods section). Measure of performance ---------------------- The number of protein examples (30269) in our corpus is about three times greater than the number of gene examples (9533). Thus, we could achieve a classification accuracy of about 75% by always predicting the protein class. To cope with the imbalance in the data, we measure the performance of each classifier as the area under ROC curve (AUC). ROC curve is a relation between the true-positive rate (TPR) and the false-positive rate (FPR) at various classification thresholds: ![](1471-2105-6-157-i1.gif) where TP, FN, FP, and TNare true positives, false negatives, false positives, and true negatives, respectively. Unlike other popular measures such as accuracy and precision-recall analysis, the AUC measure is invariant to the prior class probabilities. AUC corresponds to the probability that given a randomly chosen positive example and a randomly chosen negative example, the classifier will correctly say which is which. For a thorough discussion of ROC curves and the AUC measure, see, for example, Fawcett \[[@B41]\], Maloof \[[@B48]\], and Bradley \[[@B49]\]. We cross-validate all AUC measurements using the 5 × 2 cross-validation scheme, which is an ordinary 2-fold cross-validation performed five times. To obtain a 2-fold cross-validated performance estimate, we randomly divide a set of abstracts into two equally-sized sets and average the two performance measurements obtained by training the classifier on one set and testing the classifier on the other set. To obtain a 5 × 2 cross-validated performance estimate, a 2-fold cross-validation is performed five times and the estimates are then averaged. To avoid indirect overlap between test and training sets, we form the sets so that examples originating in the same abstract always remain in the same set. To test for statistical significance, we use the robust 5 × 2-cv test \[[@B50]\]. The test avoids the problem of dependence between folds in N-fold cross-validation schemes and results in a more realistic estimate than, for example, the t-test. Experimental setup ------------------ We randomly divided the preprocessed set of 2000 abstracts into two equal-sized sets; 1000 abstracts for parameter estimation and 1000 abstracts for final validation of the methods. The 5 × 2 cross-validated AUC was used as the measure of performance in both parameter estimation and final validation. In all the experiments with SVMs, we normalized the word frequency vectors to unit length, because the sizes of the contexts varied considerably. We carried out the SVM experiments using the LIBSVM 2.6 software \[[@B51]\] and the Naive Bayes experiments using the Bow toolkit \[[@B52]\]. In the experiments, optimal parameter values for SVMs with different kernels and for the proposed weighting scheme must be searched (the exact definitions of the parameters and kernel functions are presented in the Methods section). Every SVM itself has always a penalty parameter C, linear kernel has no other parameters than C, and both Gaussian and polynomial kernels have an additional parameter λ. Adopting a contextual representation yields a context span parameter s. The SVM equipped with the weighting scheme has two additional parameters λand βby which we may control the effect of the distances of the words from the name to be disambiguated when weighting the context words. The performance of a classifier may be strongly influenced by the choice of the values for its parameters. For example, from Figure [3](#F3){ref-type="fig"} (discussed in more detail later) it can be observed that a wrong choice of the kernel parameters as well as the SVM penalty parameter Ccan lead to a severe loss in performance. Particularly when comparing the methods, the correct parameter setting for each of the compared methods is crucial, as only then a reliable estimate of the performance is obtained for each of the methods. The correct parameter values cannot be known in advance and the use of the default values may result in sub-optimal classification performance. Therefore, the parameter values are most commonly estimated from the data. Hsu et al. \[[@B45]\] recommend a grid-search on Cand γparameters using cross-validation and exponentially growing sequences of Cand γ. Since an exhaustive search for the parameters can not be done in the continuous space of SVM and kernel parameters, we performed a coarse preliminary search in order to find an auspicious region, and subsequently conducted a finer grid search. As can be observed from Figure [2](#F2){ref-type="fig"}, the choice of values of the weighting parameters λand βis as important for the classification performance as the choice of the other parameters. Moreover, as noticed by Ginter et al. \[[@B26]\], the optimal values depend on the task and are not known beforehand. Therefore, we use a grid-search also for finding the optimal values of the parameters λand β. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### *Parameter estimation: The performance of the weighted SVM with the linear kernel as a function of the weighting parameters λand β.The best performance is reached at λ= 1.5 and β= 0.025. ::: ![](1471-2105-6-157-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Parameter estimation: The performance of the weighted SVM with the Gaussian kernel as a function of the SVM penalty parameter Cand the kernel parameter γ.Both parameters are in a logarithmic scale. The best performance is reached at C= 1 and γ= 1. ::: ![](1471-2105-6-157-3) ::: In short, the parameter estimation for SVM classifiers was performed as follows. First we estimated the context span sfor the conventional SVM and the values of λand βfor the weighted SVM, using a grid search with the linear kernel function. We used the whole abstract to form the examples for the weighted SVM. With the s, λand βparameters fixed, we evaluated different types of SVM kernels, estimating the kernel parameters with a grid search. The SVM penalty parameter Cis optimized separately at each point of the context span, weighting and kernel parameter grids. In a similar manner, the optimal combination of λ, βand sfor the WAC classifier was found by performing a 3-dimensional grid search. For the Naive Bayes classifier, only the optimal value for smust be searched. For final validation, we chose the best performing kernel function for conventional and weighted SVMs. Using the parameter values found in the parameter estimation phase, we then measured the performance of each of the compared classifiers on the validation set, again using 5 × 2 cross-validation. A detailed explanation of the grid search for the parameter estimation described above is presented in the following sections. Parameter estimation for conventional SVM ----------------------------------------- First, we searched for the optimal context span sthat we will use in our experiments with the conventional SVM. We experimented with different context spans using the conventional SVM with the linear kernel, namely spans of 1, 2, 3, 4, 5, 10, 20, 30, 40, 50, 60, 70, 80, 90 and 100 words to both directions from the term to be disambiguated. The parameter Cfor the SVM with the linear kernel was searched with values 2^-5^, 2^-4^,\..., 2^3^for each of the different context spans. In these experiments, the context span of 60 words to both directions from the term to be disambiguated resulted in the highest performance for the conventional SVM. We used this context span when experimenting with Gaussian and polynomial (d= 2, d= 3) kernels, because simultaneous searching for optimal context span and kernel parameters Cand γfor the Gaussian kernel and polynomial kernels with degrees d= 2 and d= 3 would have been computationally impractical. The values of the Cand γparameters of the Gaussian kernel were 2^-5^, 2^-4^,\..., 2^2^and 2^-3^, 2^-2^,\..., 2^2^, respectively, and the values for the Cand γparameters of the polynomial kernels were 2^-10^, 2^-9^,\..., 2^-2^and 2^-2^, 2^-1^,\..., 2^5^, respectively. The results obtained with different kernel functions using the optimal context span s= 60 found with the linear kernel are shown in Table [1](#T1){ref-type="table"}. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Parameter estimation: The performance of conventional SVMs with different kernel functions, context span s= 60. ::: Kernel AUC Parameters --------------------- -------- -------------------- Linear 80.47% C= 2^-3^ Gaussian 80.62% C= 2^-2^, γ= 2 Polynomial (d= 2) 80.49% C= 2^-9^, γ= 8 Polynomial (d= 3) 80.53% C= 2^-7^, γ= 2 ::: Parameter estimation for weighted SVM ------------------------------------- With the weighted SVM, we always used the whole abstract as a context. For the weighted SVM, we estimated the best combination of the weighting parameters λand βwith the linear kernel. The parameter Cwas also separately searched with values 2^-5^, 2^-4^,\..., 2^3^for each of the different weightings. The comparison of the performance with different weightings using the linear kernel is illustrated in Figure [2](#F2){ref-type="fig"}. At this point, we found that the values λ= 1.5 and β= 0.025 performed best for the linear kernel (for illustration, see Figure [6](#F6){ref-type="fig"}). We used these parameters when experimenting with Gaussian and polynomial (d= 2, d= 3) kernels. The values of the Cand γparameters of the Gaussian kernel were 2^-4^, 2^-3^,\..., 2^3^and 2^-3^, 2^-2^,\..., 2^2^, respectively, and the values for the Cand γparameters of the polynomial kernels were 2^-7^, 2^-6^,\..., 2^-2^and 2^-2^, 2^-1^,\..., 2^2^, respectively. The performance of the weighted SVM with different Cand parameters of the Gaussian kernel is illustrated in Figure [3](#F3){ref-type="fig"}. The figure illustrates the importance of correct parameter selection. The weighted SVM performance with different combinations of γand the penalty parameter Cfollows the behavior described by Keerthi and Lin \[[@B53]\]: Areas of underfitting can be seen at the left, where the value of the Cparameter is low, and at bottom left where the values of both Cand γare low. On the other hand, SVM with Gaussian kernel overfits heavily if the value of γis too large, as can be seen at the top of the figure. Overfitting happens also with noisy data at the right part of the figure, where the value of Cis too large. The results obtained with different kernel functions using the best weighting parameters λ= 1.5 and β= 0.025 found with the linear kernel are shown in Table [2](#T2){ref-type="table"}. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Parameter estimation: The performance of weighted SVMs with different kernel functions, λ= 1.5 and β= 0.025. ::: Kernel AUC Parameters --------------------- -------- -------------------- Linear 85.31% C= 1 Gaussian 86.15% C= 1, γ= 1 Polynomial (d= 2) 86.09% C= 2^-3^, γ= 2 Polynomial (d= 3) 86.13% C= 2^-3^, γ= 1 ::: Parameter estimation for baseline methods ----------------------------------------- The only parameter of the Naive Bayes classifier is the context span s. We performed a search for s∊ \[5, 30\] with step 5. The performance reached maximum for s= 15. The WAC incorporates an identical weighting scheme as the weighted SVM. We found the optimal parameters by performing a 3-dimensional grid search for λ∊ \[0,3\] with step 0.1, β∥ \[0, 0.25\] with step 0.025 and context span in the interval \[5,70\] with step 5. The maximum performance was obtained for λ= 1, β= 0.025 and context span s= 55 (for illustration, see Figure [4](#F4){ref-type="fig"}). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Parameter estimation: The performance of WAC with s= 55 as a function of the weighting parameters λand β.The best performance is reached at λ= 1.0 and β= 0.025. ::: ![](1471-2105-6-157-4) ::: Final validation ---------------- The Gaussian kernel was found to be the best with both the conventional and weighted SVMs when tested in the parameter estimation. The best parameters for the Gaussian kernel were C= 0.25 and λ= 2 with the conventional SVM, and C= 1 and λ= 1 with the weighted SVM. The AUC results of the final validation are presented in Table [3](#T3){ref-type="table"} and the ROC curves are given in Figure [5](#F5){ref-type="fig"}. To test the statistical significance of AUC differences between the weighted SVM, the conventional SVM, WAC and Naive Bayes, we performed the robust 5 × 2-cv test on the validation data. Each of the pairwise differences were strongly significant with p-values below 0.01, except for the difference between the conventional SVM and the Naive Bayes classifier (p-value of about 0.1). The conventional SVMs performed poorly compared to the baselines, especially to the WAC classifier that takes advantage of the contextual weighting. Incorporation of the weighting scheme into SVMs, however, improved their performance by five percentage points. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Final validation: Averaged ROC curves of the classifiers.The averaged ROC curves were obtained from the folds of the 5 × 2 cross-validation using the vertical averaging method described by Fawcett \[41\]. ::: ![](1471-2105-6-157-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### A weighting example.Here is an example of a context (t~1~,\..., t~16~). The term tis the name to be disambiguated. Figure illustrates the weights of the context words t~1~,\..., t~16~with two different parameter value pairs of λand β. The weight values are represented as a continuous function, although they take discrete values. The parameter combination λ= 1.5 and β= 0.025 yields the best performance when using the linear kernel. ::: ![](1471-2105-6-157-6) ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Final validation: The performance of the classifiers. ::: Method AUC ------------------ -------- Conventional SVM 79.85% Weighted SVM 85.48% WAC 83.05% Naive Bayes 80.81% ::: Conclusion ========== In this paper, we show that SVMs can be successfully applied to gene versus protein name disambiguation. We demonstrate how their performance can be further improved by incorporating a weighting scheme based on the intuition that the words near the name to be disambiguated are more important than the other words. The weighting scheme results in a notable performance gain of five percentage points. We also study carefully the effects of different kernel functions and parameters and show that the proposed weighting scheme influences the performance even more than the selection of the kernel part of SVMs. The weighted methods statistically significantly outperformed their unweighted counterparts, the difference being particularly notable for SVMs. In Ginter et al. \[[@B26]\], we have shown that the optimal values for λand βare non-zero and differ substantially depending on the classification task at hand. This suggests that the extent to which the long distance words contribute to the classification is task-dependent and could reflect differing properties of the tasks. While finding correct values of these parameters is clearly important as shown by the experiments, an exact interpretation of the values remains speculative. However, we discuss several reasons why the use of the proposed weighting scheme is beneficial. Further study could bring a better insight into the underlying phenomena. The performance of the weighted SVM might be further improved, for example, by using collocations in order to capture the local syntax around the term to be disambiguated. However, the proposed weighting scheme uses the local information, and therefore it already captures the information represented by collocations to some extent. In addition, several special text kernels have successfully been applied to text classification as reported by Lodhi et al. \[[@B54]\] and by Cancedda et al. \[[@B55]\]. These kernels and different weighting methods based on the distances and also, for example, on the biological relevance of the words in the context, are still to be studied. Methods ======= In this section, we describe the concepts necessary to understand the application of a SVM classifier to gene versus protein name disambiguation. We start by giving a short introduction to SVMs. The training data points of the classifier are vectors describing the word frequencies in the context in which the names to be disambiguated were found. Then, we explain the general bag of words (BoW) approach that uses only the word frequencies. However, the distances of the words with respect to the word to be disambiguated seem intuitively to be important. We describe a weighting scheme (the weighted BoW) based on distances of the context words from the ambiguous name. Finally, we briefly introduce the two baseline methods, the Naive Bayes classifier and the Weighted Additive Classifier. Support Vector Machines ----------------------- Here, we give a brief description of SVMs. A more comprehensive treatment can be found, for example, in Burges \[[@B43]\] and Vapnik \[[@B44]\]. In a binary classification task mlabeled examples (x~1~, y~1~),\..., (x~m~, y~m~), where x~i~ε Xare training data points and y~i~ε {-1, +1} are the corresponding class labels, form the training set. In order to make the data linearly separable, data points are mapped from the input space Xto a feature space Fwith a mapping Φ : X→ F* before they are used for training or for classification. SVMs can be considered as a special case of the following regularization problem: ![](1471-2105-6-157-i2.gif) where iranges from 1 to m, lis the loss function used by the learning machine, f: X→ Yis a function which maps the input vectors x∊ Xto the output labels y∊ Y, C∊ ![](1471-2105-6-157-i11.gif)~+~is a regularization/penalty parameter, and \\|\\| · \\|\\|~k~is a norm in a Reproducing Kernel Hilbert Space defined by a positive definite kernelfunction k. The second term is called a regularizer. The loss function used by SVMs for binary classification problems is called linear soft margin loss or hinge loss and is defined as l(f(x), y) = max(1 - yf(x), 0). By the Representer Theorem, the minimizer of (1) has the following form: ![](1471-2105-6-157-i3.gif) where a~i~ε ![](1471-2105-6-157-i11.gif) and kis the kernel function associated with the Reproducing Kernel Hilbert Space mentioned above. The penalty parameter Ccontrols the trade-o3 between the complexity of the decision function and the number of wrongly classified training points the model will tolerate in the feature space. Minimizing number of training errors by selecting an appropriate parameter can sometimes lead to overfitting due outliers. On the other hand, too strong regularization (low penalization) underfits. A good insight of trade-o3 can be found, for example, in Hastie et al. \[[@B56]\]. There are several commonly used kernels (see Vapnik \[[@B44]\]). The ordinary inner product is called the linear kernel k(u, v) = \<u, v\> and the polynomial kernelis defined as k(u, v) = (γ\<u, v\> + 1)^d^ where dε ![](1471-2105-6-157-i12.gif) is the degree of the polynomial and γε ![](1471-2105-6-157-i11.gif)~+~. When the polynomial kernel is used, the datapoints are mapped into a feature space which contains all products of input vector elements up to d(see e.g. Vapnik \[[@B57]\]). The γparameter of the polynomial kernel controls the weight differences of the product features of different orders. Another widely used kernel function is the Gaussian kernel ![](1471-2105-6-157-i4.gif) whose width is determined by the γparameter. We refer to Keerthi and Lin \[[@B53]\] for more information of the behavior of the SVM with the Gaussian kernel with different combinations of and the penalty parameter C. Hsu et al. \[[@B45]\] suggested to use a cross-validation and a grid search in order to estimate the best combination of and the penalty parameter Cfor the Gaussian kernel. We adopt this procedure and also perform a similar search for the linear and polynomial kernels. A detailed description of the parameter selection is given in the Results and Discussion section. Representation of contexts -------------------------- In our experiments, we trained the SVM classifier to disambiguate the sense of a term between two possible senses based on its context. Let us denote by sthe context spanparameter controlling the lengths of the contexts. For a fixed s, we take such a context ![](1471-2105-6-157-i5.gif) = (t~1~,\..., t~l~) that both the number of words preceding and following tis maximal but at most s. The words which precede tare t~1~,\..., t~k~, 0 ≤ k≤ s, in the order they appear in the text, and correspondingly t~k+1~,\..., t~l~, 0 ≤ l- k≤ sare the words which follow tin the text. Hence, if there exist swords preceding and following the word to be disambiguated, then k= l- k= s. The contexts may be of different lengths, since the number of words from tto the beginning or end of the abstract may be smaller than s. The BoW approach ---------------- Let Cbe a set of all possible contexts and let V= {v~1~,\..., v~n~} be an ordered set of all distinct words of the contexts of C. We formed the set Vof all distinct words separately for each training-testing experiment from the words found in the contexts of abstracts of the training set. Let be the mapping, which maps contexts to BoW vectors, defined by ![](1471-2105-6-157-i6.gif) where ![](1471-2105-6-157-i7.gif), 1 ≤ i≤ n, is the number of occurrences of the word v~i~∊ Vin the context ![](1471-2105-6-157-i5.gif). Thus, only the frequency of a word in the context is recorded, but the information about the distances of the words from the ambiguous name is ignored. The BoW vectors can now be used as input space data points for SVM classifiers. The number of words in Vis usually large, and therefore the dimension of the input space is also high. Customary means to reduce the dimensionality are, for example, stop-word removal and stemming. Stop-words are words that occur very often in all documents, for instance, \'is\', \'the\', \'are\', \'a\'. Stemming combines words that only differ in suffix. For example, the stemming algorithm of Porter \[[@B47]\] removes all suffixes it recognizes. We have applied these dimensionality reduction techniques in our main experiments. However, the number of words still often remains in tens of thousands. In a separate experiment, we have estimated the effect of stop word removal and stemming using theconventional SVM with the linear kernel and grid search optimization of Cand context span parameters. We found that stemming results in a low increase in performance (0.59%), stop-words removal haspractically no effect on the performance (an increase of 0.01%). Neither of the differences were statistically significant. The weighted BoW approach ------------------------- The BoW approach does not preserve any information about the positions of the words in the context. Therefore, we use a particular weighting scheme based on the distances of the words from the term tto be disambiguated. The idea is that the words near tare more likely to be important than other words, and therefore they are given larger weights. The weighted vector space model of contexts can be formalized as follows. Let dist(j) denote the distance of the word t~j~from the term t, that is, the number of words between the word t~j~and tincluding the word t~j~itself. Let further Pos(v, ![](1471-2105-6-157-i5.gif)) = {j\\| v= t~j~ε ![](1471-2105-6-157-i5.gif)} denote a set of positions jfor each word vε Vin a particular context ![](1471-2105-6-157-i5.gif). The weight for the word t~j~is defined as ![](1471-2105-6-157-i8.gif) where λ, β≥ 0 are the parameters for the weighting. If λ\>0, the weights get a hyperbolic shape with highest values immediately around the term to be disambiguated (see Figure [6](#F6){ref-type="fig"}). The bigger λ is, the steeper the weight values grow towards the term t, and βis an offset of the values. The role of βis to reduce the ratio between the weights of the words that are near to the term tand the weights which are far from t. Let Ψ now be the function which maps contexts to weighted BoWs given by ![](1471-2105-6-157-i9.gif) Where ![](1471-2105-6-157-i10.gif) Note that the setting λ= β= 0 corresponds to the ordinary BoW approach (2). Baseline methods ---------------- The two baselines, the Naive Bayes classifier and the Weighted Additive Classifier, represent a family of linear classifiers based on aggregating the class-wise co-occurrence statistics of the words in the context. The Naive Bayes classifier (see, for example, Manning and Schütze \[[@B23]\]) evaluates the a posteriori conditional probability of a class by computing a product of the corresponding conditional probabilities of the context words obtained from their class-wise co-occurrence statistics. The Weighted Additive Classifier \[[@B26]\] considers co-occurrence statistics similar to that used in the Naive Bayes classifier. However, the decision rule is additive and incorporates a weighting scheme. The weighting can be defined in a manner identical to that described in the section on weighted BoW. Authors\' contributions ======================= TP carried out the experiments with SVMs and the weighting scheme as well as the sense distribution analyses. He also drafted the manuscript. FG designed and carried out all the experiments with Naive Bayes and WAC. JB, JJ and TS conceived the original design of the work, participated in the analysis and interpretation of data, and provided scientific guidance. All authors critically revised the manuscript and approved the final version. Acknowledgements ================ We thank Professor Mauno Vihinen, Institute of Medical Technology, University of Tampere, Finland, for the discussions and reasonings of the biological importance of the gene versus protein name disambiguation task. We also thank the anonymous reviewers for their insightful comments. This work was supported by Tekes, the National Technology Agency of Finland.	PubMed Central	2024-06-05T03:55:59.615571	2005-6-22	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180820/", "journal": "BMC Bioinformatics. 2005 Jun 22; 6:157", "authors": [ { "first": "Tapio", "last": "Pahikkala" }, { "first": "Filip", "last": "Ginter" }, { "first": "Jorma", "last": "Boberg" }, { "first": "Jouni", "last": "Järvinen" }, { "first": "Tapio", "last": "Salakoski" } ] }
PMC1180821	Background ========== Thyroid tumours show a good long-term survival, but, when we consider the poorly differentiated histotypes and in particular the anaplastic carcinoma, the prognosis is very poor \[[@B1],[@B2]\]. A major determinant of such an opposite behaviour is the lost ability to concentrate the radioiodine, used as very effective tool for diagnosis and treatment for both tumour remnants or distant metastases \[[@B3]\]. The loss of iodide uptake capacity in poorly differentiated thyroid carcinomas is mainly due to the reduced/lost functional expression of the sodium/iodide symporter (NIS), with a defect occurring mainly at gene expression level, but in some case involving post-transcriptional or other unknown alterations \[[@B4]-[@B7]\]. For this reason, many strategies have been used and are currently in progress in order to re-establish iodide uptake function by means of re-expressing the NIS in tumour cells. At present, two different strategies are based on: a. introducing NIS gene in cancer cells through a viral vector; b. stimulating endogenous NIS gene expression \[[@B5],[@B7]\]. The latter approach may take advantage from the description and characterization of the human NIS gene promoter as well as studies regarding NIS expression regulation \[[@B4]-[@B7]\]. Among the physiological stimulators of NIS gene transcription, a pivotal role is played by the thyroid-specific transcription factor Pax8 \[[@B8]-[@B10]\]. This homeobox protein is able to regulate, in normal thyroid cells, the expression of many thyroid specific protein and is often reduced in thyroid tumours, especially in the less differentiated histotypes \[[@B11]\]. In this study we analysed the effects of Pax8 overexpression on a human thyroid anaplastic cancer cell line ARO cells. It is a widely used cell line which resembles the behaviour of poorly differentiated thyroid tumours, including a very low iodide uptake function as well as lost of NIS and Pax8 gene expression, together with several other markers of thyrocyte differentiation \[[@B12]\]. By restoring Pax8 expression, we observed the recovery of NIS expression, together with other markers of thyroid differentiation, associated to partial ability to uptake the radioiodine. Methods ======= Recombinant plasmid construction -------------------------------- The plasmid vector pCMV and the full length cDNA fragment of Pax8 A (kindly provided by Dr. Di Lauro) \[[@B13]\], were cleaved by Kpn I and Eco RI. The cleaved products were ligated using T4DNA ligase into pCMV-Script cloning and expression vector (Stratagene, La Jolla, CA, USA). DH5 alpha cells were transformed using the recombinant product pCMV/Pax8 and then screened for the positive clones containing the inserted fragment by colony-PCR techniques. The positive colonies were amplified, extracted, purified and identified by endonuclease cleavage. The sequences of the inserted fragments were confirmed by automated sequencing using ABI Prism 7700 Sequence Detector (Applied Biosystems, Foster City, CA, USA). Cell cultures and stable transfection ------------------------------------- ARO cells were cultured in RPMI 1640 medium with 10% foetal bovine serum (Gibco, Milano, Italy) and containing penicillin/streptomycin and amphotericin B (Sigma-Aldrich S.r.l., Milano, Italy). The non-liposomal lipid mixture FUGENE 6 (Roche diagnostics, Monza, Italy) was used for transfection. ARO cells were plated at 3 × 10^5^cells/60 mm culture dish 24 h prior to transfection. 2 mg of the construct and 2 μg of empty vector in 200 μl of serum free medium and 12 μl of FUGENE 6 mixture each, were used for two different transfections. After 45 min of incubation at room temperature, the transfection mixture was added to the dishes with 2 ml of fresh complete medium. The cells were grown for 24 h before adding 400 μg/ml neomycin (Sigma-Aldrich S.r.l.) for selection and after 3 days the transfected cultures were split for single clone isolation. After propagation, total RNA was extracted from 24 isolated clones for screening of Pax8 gene expression by a quantitative RT-PCR. Cells transfected with the empty vector (ARO-pCMV) were used as control. FRTL-5 and CHO cells, used as additional control, were cultured as described previously \[[@B14]\]. RNA extraction, reverse transcription and quantitative PCR ---------------------------------------------------------- Total RNA was extracted from cells with using the Qiagen RNA/DNA kit (Qiagen S.p.A, Milano, Italy), according to manufacturer\'s instructions. First-strand cDNA synthesis was performed using 2 μg of each RNA sample primed with random hexamers with 200 U of Superscript II reverse transcriptase (Life Technologies S.r.l., San Giuliano milanese, Milano, Italy). Quantitative PCR analysis (Q-PCR) of Pax8, NIS, Thyroperoxidase (TPO), TSH-Receptor (TSH-R), Thyroid transcription factor-1 (TTF1) and pendrin (PDS) mRNA expression was performed on cDNA samples employing the primers indicated in Table [1](#T1){ref-type="table"}. Oligonucleotide primers and probe were purchased from Applied Biosystems as Assays-on-Demand Gene Expression Products for the study of Thyroglobulin (Tg), endogenous control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and β-actin were purchased from Applied Biosystems as TaqMan PDAR (VIC dye-labeled). A 25 μl reaction mixture containing 2.5 μl of cDNA template, 12.5 μl TaqMan Universal PCR master mix (Applied Biosystems) and 1.25 μl primer probe mixture was amplified using the following thermal cycler parameters: incubation at 50°C for 2 min and denaturing at 95°C for 10 min, then 40 cycles of the amplification step (denaturation at 95°C for 15 sec and annealing/extension at 60°C for 1 min). For each amplification run standard curve was generated using 6 serial dilutions of cDNA mix expressing all the genes analysed. All amplification reactions were performed in triplicate, and the averages of the threshold cycles were used to interpolate standard curves and to calculate the transcript amount in samples using SDS version 1.7a software. Levels of mRNA expression were expressed after normalization with those of two endogenous controls, GAPDH and β-actin. Protein extraction, western blot analysis and immunofluorescence ---------------------------------------------------------------- Total proteins were extracted from thyroid cell lines as described previously \[[@B15]\]. Briefly, confluent cells from 3 Petri dishes were collected and homogenized in 1 ml of buffer containing 250 mM sucrose, 10 mM HEPES-KOH (pH 7.5), 1 mM EDTA, 1 mM PMSF, 10 μg/ml leupeptin, 10 μg/ml aprotinin (Sigma-Aldrich S.r.l.). The homogenate was centrifuged at 14000 × g (4°C for 15 min) and the supernatant (which contained the whole cell lysate) was quantified spectrophotometrically using the Bradford method. Twenty micrograms of proteins were loaded on a 4--20% gradient SDS-polyacrylamide gel and subjected to electrophoresis at a constant voltage (110 V). Electroblotting to a Hybond ECL-PVDF nitrocellulose membrane (Amersham Pharmacia biotech., Milano, Italy) was performed for 2 h at 125 mA using a Mini Trans blot electroblotting system (Bio-rad Laboratories S.r.l, Milano, Italy). Blocking was done using TTBS/milk (TBS, 1% Tween 20, and 5% non fat dry milk) for 2 h at room temperature. The membrane was then incubated with a 1/500 dilution of affinity purified rabbit anti-NIS polyclonal antibody \[[@B16]\] or a 1/5000 dilution of mouse monoclonal anti-human beta-actin antibody, overnight at 4°C in TTBS/milk. After one 15 min and two 5 min washes in TTBS, the membrane was incubated with a 1/20000 dilution of a horse radish peroxidase conjugated anti-rabbit or anti-mouse antibody (Transduction Laboratories, Lexington, KY, USA) in TTBS/milk. After one 15 min and two 5 min washes in TTBS, the protein was visualized by an enhanced chemiluminescence Western blot detection system (ECL plus, Amersham Pharmacia biotech.). For immunofluorescence analysis, the cells were plated on glass coverslips and fixed using PBS containing 4% paraformaldehyde for 10 min at RT. Then the fixative was aspirated and the cells washed 3 times for 5 min each with PBS. The cells were incubated with blocking buffer (PBS 1× + BSA 1%) for 15 min at RT. Immunostaining was performed using 1:600 dilution of a rabbit polyclonal antibody anti h-NIS \[[@B15]\] for 1 h at RT. After three 5-min washes in PBS, the cells were incubated with secondary antibody conjugate with FITC (Sigma Aldrich S.r.l.) diluted 1:160 in PBS 1×, for 30 min at RT, re-washed in PBS and incubated in Hoechst 1× for 3 min at RT. After three final PBS washes, the cells were mounted with Vectashield (Vector Laboratories, Inc. Burlingame, CA, USA). Iodide uptake, efflux rate and cell growth rate assay ----------------------------------------------------- Uptake of ^125^I was measured as previously described \[[@B17]\]. Briefly, cells were splitted and seeded into 12-well plates and, after aspirating the culture medium, washed with 1 ml of Hank\'s balanced salt solution (HBSS) (Life technologies S.r.l) supplemented with Hepes 10 mM, pH 7.3. ^125^I uptake was initiated by adding to each well 500 μl buffered HBSS containing 0.1 μCi carrier-free Na^125^I and 10 μM NaI to obtain a specific activity of 20 mCi/mmol. In half of the wells, this assay buffer was supplemented with the NIS inhibitor KClO~4~(10 μM) to control for specific uptake. After 40 min at 37°C in a humid atmosphere, the radioactive medium was aspirated and cells were washed with 1 ml of ice-cold HBSS. One ml of 95% ethanol was added to each well for 20 min and then transferred into vials for counting with a gamma counter. Iodide uptake was expressed as picomoles per μg of DNA. Each experiment was done at least twice in quadruplicates and the FRTL-5 cells and CHO cells were used, respectively, as positive and negative controls. Iodide efflux in wild-type and transfected ARO cells, as well FRTL-5 cells, was evaluated as previously described by Arturi et al. \[[@B14]\]. For cell growth rate analysis, 10^5^cells were plated in 60 mm dishes. Seven hours were considered as extra time needed from cells to sediment and start to proliferate. Cells were harvested after 24, 48 and 72 hours and the number of cells was determined by Neubauer cell counting chamber. Results ======= ARO cells were transfected with the pCMV-Script-hPax8 construct or the expression vector alone. A series of 24 clones were grown and screened for the presence of the insert (data not shown). Pax8 mRNA levels were quantified by real time RT-PCR method and ARO-Pax8 clones 8 and 21 were selected in that they overexpressed different amounts of Pax8 transcript (Fig. [1](#F1){ref-type="fig"}). Since Pax8 has a key role in thyroid differentiation, we first checked in the ARO-Pax8 cells the presence of a series of thyroid specific genes. Fig. [2](#F2){ref-type="fig"} shows the levels of the transcripts detected in the two clones versus wild-type ARO cells or ARO transfected with the empty vector (ARO-pCMV). The highest mRNA expression levels were detected for Thyroglobulin gene (more than 20 fold in ARO-PAX8 clone 21), but also NIS, Pendrin, TPO and TTF1 gene were expressed in PAX8 clones with different levels of the transcripts depending on the different amount of Pax8 transcript (Fig. [2](#F2){ref-type="fig"}). All the transcript expression levels resulted lower (ranging between 10 and 100 fold) than those detectable in normal thyroid tissue (data not shown). The mRNA of the TSH receptor did not appear in the transfected cell clones (data not shown). Goal of the study, however, was to verify the possibility to recover the ability to concentrate the radioiodine. For this reason we evaluated the NIS protein levels by Western blot analysis and its localization by immunofluorescence, together with the ability of ARO-Pax8 cells to uptake the radioiodine. As shown in Fig. [3](#F3){ref-type="fig"}, NIS protein was detected in both clones and its amount paralleled that of the corresponding transcripts of NIS and Pax8. However, only few amount of the protein reached the plasmamembrane, as detected by immunofluorescence (Fig. [4](#F4){ref-type="fig"}). Accordingly, a low, but significant rate of radioiodine uptake reappeared in ARO-Pax8 cells with a similar pattern between the two clones and at much lesser extent than normal thyrocytes (Table [2](#T2){ref-type="table"}). It was inhibited by perchlorate (10 μM) (Table [2](#T2){ref-type="table"}) and not stimulated by activation of the cAMP pathway obtained by incubating the cells with 10 μM of Forskolin (data not shown). A high rate of efflux was also observed, at similar extent in both clones expressing Pax8 and earlier and much higher than that detected in FRTL-5 cells (Fig. [5](#F5){ref-type="fig"}). Finally, in the ARO-Pax8 clones, we observed a lower growth rate in comparison to ARO non transfected cells (Fig. [6](#F6){ref-type="fig"}). Discussion ========== A modified gene expression represents the basis for the development and acquisition of the abnormal growth and differentiation characteristics of tumour cells. Since the regulation of the gene expression takes place mainly at transcriptional level, structural or functional alterations of transcription factors, especially those specific for a given tissue, play a key role in the oncogenic process \[[@B18]\]. In thyroid cells, the simultaneous expression of a cohort of tissue-specific transcription factors, including TTF1, TTF2, Pax8 and HEX has been described \[[@B19]\]. Their expression and correct function is essential for the normal development of thyroid gland, and is often altered (reduced or lost) in thyroid tumours, especially in the poorly differentiated histotypes \[[@B11]\]. As a result, thyroid tumour cells present a dedifferentiated phenotype, including a reduced/lost ability to concentrate the radioiodine. This feature represents one of the major determinant of the poor prognosis of less differentiated thyroid carcinomas, since it does not allow the use of radioiodine as diagnostic and therapeutic tool for both tumour remnants or distant metastases. For this reason, attempts to reintroduce such a functional activity in tumour cells have represented the goal of many studies. The cloning and characterization of the protein responsible for iodide uptake, the NIS \[[@B20]\], has opened the way to novel therapeutic strategies based on induction of iodide uptake ability in tumour cells by recovery the expression of the NIS gene. Subsequently, a NIS-based gene therapy approach by using viral vectors has been exploited \[[@B4]-[@B7]\]: despite promising results obtained in both in vitro and in vivo experimental models, unresolved issues still remain before proposing such a treatment on humans, mainly a great efflux rate of the administered radioiodine which makes the effective dose of radioisotope (to be administered to destroy the tumour tissue) too high to be side effect free. A parallel approach to restore radioiodine uptake is to induce endogenous NIS expression and correct function by acting on its transcriptional regulation. Various molecules have been challenged, including retinoids, demethylating agents and inhibitors of histone deacetylation \[[@B21]-[@B25]\], with preliminary promising effects in various in vitro and in vivo experimental models. However, in poorly differentiated thyroid tumour cells, re-expression of NIS protein is not always accompanied by recovery of its function \[[@B26]\] and, at present, despite their unequivocal activity in vitro, retinoids did not show a sufficient effectiveness in clinical trials on human patients \[[@B27]\]. Probably, alterations of other proteins, including some involved in the intrathyroidal metabolism of iodine or in the correct folding, trafficking, polar localization and function of the NIS, do occur in tumour cells, making not fully effective the simple restoration of NIS expression. In this work, we explored an alternative way to recover radioiodine concentration ability in thyroid tumour cells. Indeed, we tried to activate the transcriptional machinery which physiologically control NIS gene expression by acting on its promoter, as well as other thyroid specific gene cooperating to express the differentiated thyrocyte phenotype. On this regard, the paired domain-containing protein Pax8 has been characterized for its physiological role in the development of thyroid tissue and maintenance of thyrocyte differentiation phenotype \[[@B28],[@B29]\]. Moreover, reduction/lost of Pax8 expression has been described in thyroid tumour tissues and cell lines \[\[[@B11]\], our unpublished observations\]. Pax8-binding sites have been described in the promoters of human TPO, Thyroglobulin and recently, also NIS gene \[[@B8],[@B9]\]. In addition, re-expression of Pax8 was associated with the recovery of the NIS, as well as TPO and Tg mRNA expression in a rat thyroid cell line \[[@B30]\]. An effect on Tg and TPO promoter activity had been previously described, in cooperation with TTF-1, also in human thyroid tumour cell lines \[[@B31]\]. Our study demonstrates that thyroid tumour cells expressing Pax8 after stable transfection become able to uptake the radioiodine, even if in absence of a full recovery of NIS protein expression on their plasmamembrane. In addition, other markers of thyroid differentiation were detectable after Pax8 expression, including proteins involved in the intrathyroidal metabolism of iodine such as TPO and pendrin. The expression levels of these thyroid-specific genes were strictly related to the amount of Pax8 expressed in the transfected clones, emphasizing the importance of the levels of this transcription factor for determining a differentiated phenotype \[[@B32]\]. It is of interest also our finding of the recovery of TTF-1 gene expression in the cells with higher levels of Pax8 which suggests a regulatory role for Pax8 at TTF-1 gene promoter level. It may contribute to the phenotypical effects observed and confirms the existence of various levels of cooperation among tissue specific transcription factors \[[@B33]\]. On this regard, few data are available about TTF1 gene promoter, although previous experimental data had suggested the TTF1 gene as a candidate target for regulation by homeobox proteins \[[@B34]\]. However, Pax8 and TTF1 do not determine a full recovery of normal thyrocyte phenotype, as demonstrated by the presence of lower expression levels of the transcripts in comparison with normal thyroid tissue and the absence of detectable levels of the TSH receptor even in the clones overexpressing Pax8, and, more important, the low intracellular retention of radioiodide. In ARO cells, a greater efflux rate had been also observed after overexpression of exogenous NIS gene inserted in a viral vector \[[@B35]\]; similarly, only a faint iodide uptake reappeared in BHP18-21v thyroid tumour cells expressing TTF1 \[[@B36]\]. Moreover, contrasting results have been reported after stimulation of endogenous NIS expression by inhibition of histone deacetylation: by using valproic acid, Fortunati et al. \[[@B26]\] did not obtain a correct localization of the NIS protein in the thyrocyte plasmamembrane, whereas Furuya et al. \[[@B25]\] described a full recovery of iodide uptake also in ARO cells using the histone deacetylase inhibitor depsipeptide. In addition, our observation of the slower growth rate of ARO-Pax8 cells suggest a role for Pax8 in the control of such a function, until now never detected. Altogether, these data indicate that in addition to act on NIS transcription, a more complex strategy including the correct targeting to the plasmamembrane is required to reach the goal of an effective radioiodine intracellular concentration. Conclusion ========== Our finding demonstrate that overexpression of thyroid-specific transcription factors may be a first step to obtain the radioiodine concentration by the transformed thyrocytes, by acting on a pool of differentiation markers including the NIS. Our data, therefore, represent a very promising starting point for further investigations and strategies exploiting this effect of Pax8, eventually in combination of other thyroid-specific transcription factors, for extending radioiodine treatment also to the less differentiated thyroid tumours. Competing interests =================== The author(s) declare that they have not competing interests. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2407/5/80/prepub> Acknowledgements ================ This work was granted by MIUR-Cofin 2003 to D. Russo, by Ministry of Health and MIUR-Cofin 2004 to Prof. S. Filetti. I. Presta was supported by a fellowship from AIRC/FIRC. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Levels of expression of Pax8 mRNA in the wild-type and transfected ARO cells. ARO cells were transfected with a pCMV expression vector empty or containing the hPax8 gene, as described in Methods. Total RNA was extracted from duplicate dishes of wild-type ARO cells (ARO), cells containing the empty vector (ARO-pCMV) and two selected clones expressing Pax8 gene (ARO-Pax8 cl. 8 and ARO-Pax8 cl. 21 (as described in Methods) and the levels of RNA were determined using the real-time PCR method. To normalize the different amount of total RNA added to the reaction, an amplification of GAPDH or β-actin was performed as endogenous control. The data are expressed as the mean ± SD of values obtained from at least three different experiments. Statistic analysis was performed using the one-factor ANOVA, followed by ttest. P\< 0.05 was considered statistically significant. ::: ![](1471-2407-5-80-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Levels of mRNA of various thyroid specific genes in wild-type and transfected ARO cells. Total RNA was extracted from duplicate dishes of wild-type ARO cells (ARO), cells containing the empty vector (ARO-pCMV), two selected clones expressing Pax8 gene (ARO-Pax8 cl. 8 and ARO-Pax8 cl. 21) and the levels of RNA were determined using the real-time PCR method. To normalize the different amount of total RNA added to the reaction, an amplification of GAPDH or β-actin was performed as endogenous control. The data are expressed as the mean ± SD of values obtained from at least three different experiments. Statistic analysis was performed using the one-factor ANOVA, followed by ttest. P\< 0.05 was considered statistically significant. ::: ![](1471-2407-5-80-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Expression of NIS protein in ARO cells expressing different levels of Pax8 gene. Western Blot analysis was performed as described in Methods. In upper panel/autoradiograph of a representative experiment. In lower panel/the staining intensity is expressed as mean ± SD of values obtained from three different experiments. The data are expressed relative to the control values (cells transfected with the empty vector, ARO-pCMV). Statistic analysis was performed using the one-factor ANOVA, followed by ttest. P\< 0.05 was considered statistically significant. ::: ![](1471-2407-5-80-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Immunofluorescent localization of NIS in ARO cells. ARO-pCMV and ARO-Pax8 cl. 21 cells were analyzed by immunofluorescence microscopy using rabbit polyclonal anti-NIS and anti-rabbit FITC secondary antibody. A: NIS (green) localization; B. Double immunofluorescence with blue that indicates DAPI nuclear staining and green that indicates the NIS localization. Magnification: 40×. ::: ![](1471-2407-5-80-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Iodide efflux studies in wild-type and transfected ARO cells. Wild-type and transfected ARO cells (black circles) and FRTL-5 (red squares) grown to 95% confluency in 35-mm diameter plates were incubated in shaking water bath for 60 min at 37°C in HBSS buffer containing 10 μM NaI and 1 μC ^125^I. To evaluate the iodide efflux, the medium was removed and replaced with 1 ml of non-radioactive medium every 2 min. After removal of the last medium (20 min) cells were extracted using 1 ml ethanol (see Methods). The values shown are the counts per min remaining (as percentage of the total) at the indicated times. The experiment has been performed two fold, being no differences in the obtained data. ::: ![](1471-2407-5-80-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Growth rate of wild-type and transfected ARO cells. 10^5^cells were plated in 60 mm dishes. Seven hours were considered as extra time needed from cells to sediment and start to proliferate. Cells were harvested after 24, 48 and 72 hours and the number of cells was determined by Neubauer cell counting chamber. \p value \<0.001 ::: ![](1471-2407-5-80-6) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Sequences of primer pairs and TaqMan probes ::: [Gene]{.underline} Primer forward* Primer reverse Probe -------------------- ------------------------------------ ----------------------------------- -------------------------------------- Pax8 5\'-CAACAGCACCCTGGACGAC-3\' 5\'-AGGGTGAGTGAGGATCTGCC-3\' 5\'-CTGACCCCTTCCAACACGCCACTG-3\' NIS 5\'-CCATCCTGGATGACAACTTGG-3\' 5\'-AAAAACAGACGATCCTCATTGGT-3\' 5\'-AGAACTCCCCACTGGAAACAAGAAGCCC-3\' TPO 5\'-ACGCCTCTGCGAGGTGC-3\' 5\'-TGCAAATCACCGTCGAGGT-3\' 5\'-TGCTGATCGGAGGCTTCGCAG-3\' TSH-R 5\'-CCTTCACCTCACACGGGCT-3\' 5\'-TGCTCTCATTACACATCAAGGACTC-3\' 5\'-TTTCTTACCCAAGCCACTGCTGTGCC-3\' TTF-1 5\'-CCTGTCCCACCTGAACTCCTC-3\' 5\'-GCCTTTGTGGTTTTTTGTTCC-3\' 5\'-CGGACTACGGCACCATGTCCTGC-3\' Pendrin 5\'-CATCAAGACATATCTCAGTTGGACCT-3\' 5\'-ACAGTTCCATTGCTGCTGGAT-3\' 5\'-TCTGAGCATGGCCCCCGACG-3\' For Thyroglobulin gene were used primers and probes provided by Applied Biosystems (Assays-on-demand gene products). ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Radioiodide uptake in wild-type and transfected ARO cells ::: Cell type Iodide uptake (cpm/well) ------------------- -------------------------- --------- ---------- ----------- ARO wild-type 133 ± 35 78 ± 26 \\0.09 ARO pCMV 137 ± 42 \0.9 82 ± 47 \\0.2 ARO-PAX8 clone 8 307 ± 43 \0.006 126 ± 27 \\0.003 ARO-PAX8 clone 21 257 ± 36 \0.01 104 ± 26 \\0.004 FRTL-5 4350 ± 320 CHO 40 ± 12 Iodide uptake was analyzed as described in Methods. The FRTL-5 cells and CHO cells were used as positive and negative controls, respectively. Iodide uptake was expressed as the mean ± SD of values obtained from three separate experiments. Statistic analysis was performed using the one-factor ANOVA, followed by ttest. P\< 0.05 was considered statistically significant. \ ARO PAX8 clone 8, clone 21 and ARO pCMV vsARO wild-type; \\ Uptake with vsUptake without KClO~4~ :::	PubMed Central	2024-06-05T03:55:59.619883	2005-7-19	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180821/", "journal": "BMC Cancer. 2005 Jul 19; 5:80", "authors": [ { "first": "Ivan", "last": "Presta" }, { "first": "Franco", "last": "Arturi" }, { "first": "Elisabetta", "last": "Ferretti" }, { "first": "Tiziana", "last": "Mattei" }, { "first": "Daniela", "last": "Scarpelli" }, { "first": "Emanuele", "last": "Tosi" }, { "first": "Angela", "last": "Scipioni" }, { "first": "Marilena", "last": "Celano" }, { "first": "Alberto", "last": "Gulino" }, { "first": "Sebastiano", "last": "Filetti" }, { "first": "Diego", "last": "Russo" } ] }
PMC1180822	Background ========== There is a considerable body of evidence that reproductive organ size contributes to male reproductive success. This mainly derives from interspecific comparisons that have found positive relationships between testis size and the risk of sperm competition \[[@B1]-[@B5]\]. In addition, the direct manipulation of sperm competition intensity under experimental evolution has been shown to cause correlated changes in testes size in two species of Diptera \[[@B6],[@B7]\]. However, few studies have addressed the importance of internal reproductive organ size to other components of male reproductive success, or the significance of size variation in accessory reproductive organs which are often vital for sperm transfer, fertility, and essential for success in sperm competition \[[@B8],[@B9]\]. In this paper, we investigate how reproductive organ size may limit male mating frequency under conditions where males encounter high numbers of mating opportunities and are thus potentially at risk of sperm or seminal fluid depletion \[[@B10]-[@B12]\]. Previous data support the hypothesis that male mating frequency can be limited by reproductive organ size in insects. For example, in dung flies, the length of the proximal section of the testis decreases with the number of copulations achieved in Scathophaga stercoraria\[[@B13]\] and increasing copula duration in Sepsis cynipsea\[[@B14]\]. Testis mass is also lower in mated than in unmated Dawson\'s burrowing bees Amegilla dawsoni\[[@B15]\]. In contrast, accessory gland size, but not testis size, is phenotypically correlated with male mating frequency in Drosophila melanogaster\[[@B16]\] and accessory glands become completely depleted and reduced in volume after 4--5 matings, leading to decreased fertility even though motile sperm remain in the seminal vesicles \[[@B17],[@B18]\]. The ability to replenish reserves of sperm and seminal fluid likely further constrains male mating frequency (reviewed in \[[@B10]\]). Mating stimulates the replenishment of accessory gland products in D. melanogaster\[[@B19]\]. This resynthesis reaches a maximum after 2--4 hours and decreases to basal levels after 48 hours in Drosophila funebris\[[@B20]\]. In this study, we used the stalk-eyed fly Cyrtodiopsis dalmannito test whether testis and accessory gland size are affected by mating. This is an ideal species, as males and females regularly mate at extremely high frequency \[[@B21]-[@B23]\]. Over 90% of matings occur in nocturnal aggregations which usually consist of a single male and a harem of several females \[[@B24]\] (up to 24 in the closely related species Cyrtodiopsis whitei\[[@B25]\]). Females join aggregations each evening and mate in the period immediately following dawn before dispersing \[[@B25],[@B26]\]. During copulation, males transfer a single small spermatophore composed of sperm from the testes enveloped in accessory gland secretions \[[@B27]\]. Previous work has shown that accessory gland length, but not testis length, is phenotypically correlated with male mating frequency \[[@B22]\]. Additionally, bidirectional artificial selection on male mating frequency resulted in a correlated response in accessory gland length but not in testis length \[[@B23]\]. While correlative evidence, whether phenotypic or genetic, indicates an association between accessory gland size and male mating frequency, it does not establish a direct physiological relationship between these two variables. In the current study, we provide direct evidence that mating induces a decrease in accessory gland, but not testis, size. Furthermore, we demonstrate that the timecourse of post-copulatory recovery of accessory gland size closely mirrors field observations of mating patterns in C. dalmanni. Results ======= We manipulated male mating status by providing males with the opportunity to mate with 6 virgin females for 60 minutes immediately following artificial dawn. Mated males were dissected at fixed times following this mating period (0 hours, 2 hours, 8 hours, 24 hours and 48 hours) and the sizes of their testes and accessory glands were compared to unmated control males. Mating resulted in a significant decrease in accessory gland length, but glands returned to their original size over the course of the next 8 to 48 hours. At average levels of male eyespan, included to as a measure of body size to control for allometric variation (F~1,185~= 5.25, p= 0.0231), mating status affected accessory gland length (F~5,185~= 4.72, p= 0.0004). Post-hoc Tukey HSD tests revealed that males dissected immediately after mating or 2 hours after mating exhibited significantly smaller accessory glands than unmated controls. Gland length began to recover after 8 hours and by 48 hours after mating the accessory glands were significantly longer than immediately following mating (Fig. [1](#F1){ref-type="fig"}). Removing unmated control males from the analysis revealed a positive effect of mating frequency on accessory gland length (b± s.e. = 0.0228 ± 0.0086, t~149~= 2.67, p= 0.0085) after controlling for the significant effect of recovery time (F~4,149~= 3.38, p= 0.0111). Males mated a mean ± s.e. of 3.79 ± 0.20 (range: 1--12) times during the course of the 60 minute observation period, and mating frequency did not vary between groups dissected at different times F~4,150~= 1.08, p= 0.3667). Identical results were obtained when accessory gland length was replaced with area, but are not included as accessory gland length and the square root of area were highly positively correlated (r~90~= 0.926, p \<0.0001). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Reduction and subsequent recovery of accessory gland length following mating.Mean accessory gland length decreased from 2.10 mm to 1.85 mm following mating and was restored to the original size within 8--48 hours. Controls (con) were unmated (virgin) males. Columns not marked with the same letter are significantly different (Tukey HSD). Values shown are least squares means ± s.e. at average values of male eyespan. ::: ![](1471-2148-5-37-1) ::: Mating did not result in a decrease in testis length compared to unmated controls (Tukey HSD, Fig. [2](#F2){ref-type="fig"}). However, significant differences in testis length were detected between males measured at different recovery times (F~5,185~= 3.10, p= 0.0102). Post-hoc Tukey HSD tests revealed that males allowed to recover for 48 hours exhibited shorter testes than males allowed to recover for 2 or 24 hours. Testis length scaled with male eyespan (F~1,185~= 1.71, p= 0.0054). Removing unmated males from the analysis failed to reveal any association between testis length and mating frequency (F~1,147~= 0.68, p= 0.4100) after controlling for recovery time F~4,147~= 3.66, p= 0.0071) and eyespan (F~1,147~= 4.33, p= 0.0392). Testis length and the square root of area were positively correlated (r~61~= 0.691, p\< 0.0001). As testis length explained less than half of the variance in testis area (r^2^= 0.477), we also directly compared testis area in males immediately after mating to that in unmated controls and detected no difference (mean ± s.e.: mated = 0.801 ± 0.021, unmated = 0.792 ± 0.020, t~65~= 0.276, p= 0.7834). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Response of testis length to mating.Males dissected 48 hours after mating exhibited smaller testes than males dissected at 2 hours and 24 hours post-mating. Controls (con) were unmated (virgin) males. Columns not marked with the same letter are significantly different (Tukey HSD). Values shown are least squares means ± s.e. at average values of male eyespan. ::: ![](1471-2148-5-37-2) ::: Discussion ========== Male accessory gland size in C. dalmannidecreased dramatically following copulation and slowly recovered over the next 8--48 hours. After removing the effect of recovery time, accessory gland length was positively correlated with male mating frequency. Neither testis length nor testis area appeared to be altered by copulation; no significant difference in testis length was observed between mated and unmated males in the 48 hours following copulation. Both male and female stalk-eyed flies mate frequently. In the current study, each male mated an average of 3.79 times (up to a maximum of 12) during the 60-minute observation period. Only 23.9% (37 out of 155) of males mated at least 6 times and therefore 76.1% (118 out of 155) of males failed to mate with all 6 virgin females provided. As females housed with three males will mate an average of 5.51 times during the 60 minutes following artificial dawn \[[@B28]\], it is clear that male mating frequency was limited by physiological ability rather than the availability of willing females. In the field, copulations occur primarily at dawn \[[@B24],[@B25]\]. Our observations of the recovery of the accessory glands match this behavioural pattern, as 24 hours after copulation (i.e. the subsequent dawn period), the accessory glands had recovered their original pre-mating size. We found that the accessory glands had partially recovered after 8 hours which is consistent with the lower frequency of mating observed at dusk \[[@B24],[@B25]\], whereas little recovery was observed in the hours immediately following copulation when flies leave mating aggregations to forage. The two most plausible physiological constraints on male mating frequency in C. dalmanniare the availability of accessory gland products and the availability of sperm, both of which are required to produce spermatophores \[[@B27]\]. Several lines of evidence indicate that accessory gland size is more likely to limit mating frequency than testis length. First, we have demonstrated a decrease in accessory gland size following copulation and the subsequent recovery closely mirrors mating behaviour in the field. No significant reduction in testis size was observed in mated males compared to unmated controls. Second, our study confirms the results of a previous experiment showing that accessory gland length, but not testis length, is phenotypically correlated with male mating frequency \[[@B22]\]. Third, bidirectional artificial selection on male mating frequency produced a correlated response in accessory gland length but not testis length \[[@B23]\]. However, we cannot exclude the possibility that some other currently unknown factor is the primary constraint on male mating frequency in C. dalmanni. The full importance of the accessory glands in stalk-eyed fly reproduction is poorly understood. Accessory gland products form the casing of the spermatophore and consequently are necessary for sperm transfer \[[@B27]\]. Furthermore, accessory gland products appear to be important in sperm competition as seminal fluid can decrease the viability of sperm from particular rival males in the female spermathecae \[[@B29]\]. However, in contrast to D. melanogaster\[[@B30]\], accessory gland products do not appear to play a role in sperm displacement \[[@B29]\], the inhibition of female remating \[[@B31]\] or the manipulation of female fecundity \[[@B28]\]. Consequently, the advantage of large accessory glands is likely gained through both increased mating frequency (by allowing males to produce more spermatophores over a given time period) and, potentially, greater success under sperm competition. Conclusion ========== When receptive females are not limiting, male mating frequency in C. dalmanniis likely constrained by accessory gland size. Copulation causes a significant reduction in accessory gland size and replenishment of the depleted accessory glands follows a time course that is consistent with the observed daily peak in male mating frequency at dawn. There was no reduction in testis size following mating and therefore testis size appears to be of less importance in determining male mating frequency in this species. Methods ======= General methods --------------- The base stock was an outbred laboratory population of the stalk-eyed fly, C. dalmanni, collected from Gombak, Malaysia in 1993. The stock was maintained in large cages at high population size (typically more than 200 individuals per cage) and with a 1:1 sex ratio. Flies were fed ground corn medium and kept at 25°C on a 12 h/12 h light/dark regime. The regime included a 15-min \"dawn\" period in which the culture room was illuminated by a single 60-W bulb. All observations of behaviour commenced at the start of this dawn period. Manipulation of male mating status ---------------------------------- Experimental flies were raised from eggs collected in groups of 13 from the population cages and allowed to hatch on moist cotton pads in Petri dishes containing at least 2 g of ground corn (maize). Upon eclosion, flies were segregated according to sex and raised to sexual maturity in groups of 10 housed in 1.5 L plastic pots on an ad libitumdiet of ground corn. Mating observations were conducted using virgin males aged 6 weeks post-eclosion and virgin females aged 6--8 weeks post-eclosion. Males were randomly assigned to 5 mating status groups: unmated controls (n = 36), 0 hours recovery (n = 38), 2 hours recovery (n = 29), 8 hours recovery (n = 30), 24 hours recovery (n = 30), and 48 hours recovery (n = 30). At artificial dawn, individual males were added to 1.5 L plastic pots containing 6 females, except for control males which were placed in empty 1.5 L pots. The number of copulations over 40 seconds in duration occurring during the subsequent 60-minute period was recorded. Males that failed to mate during this observation period were discarded. Unmated control males and 0 hour recovery males were immediately placed on ice and dissected. Males assigned to other recovery periods were moved individually to 500 ml plastic pots lined with a moist cotton pad and provided with ground corn until the appropriate time of dissection. Morphological measurements -------------------------- Males were dissected in a small amount of phosphate buffered saline on a microscope slide. Images of the accessory glands and uncoiled testes were captured using a monocular microscope connected via a video camera to a Macintosh computer with NIH Image (version 1.55). Length was measured by tracing a midline that longitudinally bisected each organ and the mean length of the two accessory glands or testes was used in analyses. Area was measured by tracing the outline of each organ and calculating the longitudinal surface area. Areas of both accessory glands were calculated and the mean used in analyses, but a single randomly chosen testis was measured per individual. Eyespan, was defined as the distance between the outer tips of the eyes. Statistical analyses -------------------- Unless otherwise indicated, general linear models were used to analyse the determinants of reproductive organ size. Initial models included an intercept, male eyespan, recovery time and the eyespan × recovery time interaction. Recovery time was coded into models as an ordinal categorical variable. Stepwise elimination was used to remove terms that failed to significantly improve the fit of the model. Secondary analyses extended the models to include the number of copulations observed which required the exclusion of control males that did not copulate. Data sets did not deviate significantly from the assumptions of general linear modelling. Authors\' contributions ======================= DWR conceived of the study, contributed to the design, carried out the experimental work and statistical analysis, and drafted the manuscript. TC, KF, and AP participated in the design and coordination of the study and helped to draft the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ The authors thank Matthew Denniff for assistance with fly rearing and behavioural observations, and two anonymous referees for their helpful comments on the manuscript. This research was funded by awards from the Natural Environment Research Council (TC, KF & AP), the Royal Society (TC) and the Association of Commonwealth Universities (DWR).	PubMed Central	2024-06-05T03:55:59.622798	2005-6-9	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180822/", "journal": "BMC Evol Biol. 2005 Jun 9; 5:37", "authors": [ { "first": "David W", "last": "Rogers" }, { "first": "Tracey", "last": "Chapman" }, { "first": "Kevin", "last": "Fowler" }, { "first": "Andrew", "last": "Pomiankowski" } ] }
PMC1180823	We recently received a query from a reader of our article \[[@B1]\], about the ICD-10-based definitions used to define the \"chronic obstructive pulmonary disease / asthma\" group of index admissions. Prompted by the query, we detected an error in the way the definition of this group, as well as the \"acute coronary syndrome\" group are described in the printed version of our article, which we wish to correct. The printed version of the article describes chronic obstructive pulmonary disease / asthma as composed of ICD-10 codes J44.0-45.9. This should have read: J40.0-J46 (i.e. diagnostic codes relating to Chronic Obstructive Pulmonary Disease, chronic bronchitis, emphysema and asthma), as well as J12.0-J18 (i.e. diagnostic codes relating to pneumonia).We have included asthma and pneumonia in the definition of this group because, based on experience, we felt that in adults over 50 who are admitted to hospital a primary diagnosis of either asthma or pneumonia is very likely to either represent miscoding of COPD for (asthma or pneumonia), and/or an exacerbation episode of underlying Chronic Obstructive Pulmonary Disease. We include the exact composition of this group, by diagnostic code in Table [1](#T1){ref-type="table"}. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### ICD-10 coding composition of the \"chronic obstructive pulmonary disease / asthma\" group of index admissions used in the study. ::: ICD-10 code Description Number of index admissions \% ------------- --------------------------------------------------- ---------------------------- ------ J13X Pneumonia due to Streptococcus pneumoniae 18 0.6 J14X Pneumonia due to Haemophilus influenzae 2 0.1 J151 Bacterial pneumonia, not elsewhere classified 3 0.1 J154 Bacterial pneumonia, not elsewhere classified 4 0.1 J157 Bacterial pneumonia, not elsewhere classified 10 0.3 J180 Pneumonia, organism unspecified 53 1.7 J181 Pneumonia, organism unspecified 450 14.8 J182 Pneumonia, organism unspecified 1 0.0 J188 Pneumonia, organism unspecified 1 0.0 J189 Pneumonia, organism unspecified 201 6.6 J42X Unspecified chronic bronchitis 2 0.1 J439 Emphysema 61 2.0 J440 Other chronic obstructive pulmonary disease 771 25.3 J441 Other chronic obstructive pulmonary disease 632 20.7 J448 Other chronic obstructive pulmonary disease 11 0.4 J449 Other chronic obstructive pulmonary disease 157 5.2 J450 Asthma 1 0.0 J459 Asthma 657 21.6 J46X Status asthmaticus 12 0.4 Total 3047 100 ::: Similarly, in the printed version of the article the group of \"acute coronary syndrome\" is defined as I20.0 and I21.0-9. This should have read: I20.0-9 (i.e. angina), I21.0-9 (acute MI) and R07.0-9 (i.e. \"non-specific pain in throat or chest\").Again we have decided to include non-specific chest pain along with angina and MI, due to potential for miscoding and/or true co-morbidity (of non-specific chest pain with acute ischaemic heart disease presentations). The exact composition of codes used for the \"acute coronary syndrome\" group is included in Table [2](#T2){ref-type="table"}. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### ICD-10 coding composition of the \"acute coronary syndrome\" group of index admissions used in the study. ::: ICD-10 code Description Number of index admissions \% ------------- ------------------------------------------------------------- ---------------------------- ------ I200 Unstable angina 367 8.6 I201 Angina pectoris with documented spasm 2 0.0 I208 Other forms of angina pectoris 3 0.1 I209 Angina pectoris, unspecified 721 16.8 I210 Acute transmural myocardial infarction of anterior wall 285 6.7 I211 Acute transmural myocardial infarction of inferior wall 388 9.1 I212 Acute transmural myocardial infarction of other sites 40 0.9 I214 Acute subendocardial myocardial infarction 90 2.1 I219 Acute myocardial infarction, unspecified 363 8.5 R072 Precordial pain 2 0.0 R073 Other chest pain 626 14.6 R074 Chest pain, unspecified 1396 32.6 Total 4283 ::: Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-227X/5/4/prepub>	PubMed Central	2024-06-05T03:55:59.624225	2005-7-18	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180823/", "journal": "BMC Emerg Med. 2005 Jul 18; 5:4", "authors": [ { "first": "Georgios", "last": "Lyratzopoulos" }, { "first": "Daniel", "last": "Havely" }, { "first": "Islay", "last": "Gemmell" }, { "first": "Gary A", "last": "Cook" } ] }
PMC1180824	Background ========== Canine mtDNA sequence comparisons have been used to clarify the origins of the domestic dog \[[@B1],[@B2]\], assess hybridisation with wild species \[[@B3],[@B4]\], and forensic analysis \[[@B5],[@B6]\]. Savolainen et al\[[@B1]\] and a recent revision aiming at nomenclature standardisation necessary for efficient data-basing \[[@B7]\], provided a good state of the art with a total of 1146 individuals studied, the majority assigned to a specific breed. However, the number of individuals per breed rarely exceeded 10. Most of the sequences reported surveyed the D-loop region between positions 15458 to 16039, a segment 582 bp long, of which 96 were polymorphic (72 transitions, 13 transversions and 14 indels), defining a total of 139 haplotypes. Six haplogroups were recognised (A to F), the first five interspersed with wolf lineages. Haplogroup A accounted for 70% of the lineages, and D to F just for 5%. As thorough as these studies were, there were still gaps and bias. In Europe, for instance, most studied breeds were British or Scandinavian, while mainland countries remained essentially unscreened. This study intends to fill one of these gaps, i. e. the analysis of mtDNA lineages of breeds in the south western edge of Europe. At present, 10 Portuguese dog breeds are officially recognised by the Fédération Cynologique Internationale. This study focused on 4 of these indigenous breeds: the Castro Laboreiro Dog (Cão de Castro Laboreiro) in the Northwest, the Portuguese Sheepdog (Cão da Serra de Aires) in the South, the Serra da Estrela Mountain Dog (Cão da Serra da Estrela) in the Centre, and the Azores Cattle Dog (Cão de Fila de S. Miguel), in the Azores Islands. The Castro Laboreiro Dog is a mastiff type lupoid breed traditionally used in sheep herding and guarding, especially against wolves. It is thought to be one of the most ancient dog breeds of Iberia. The region of Castro Laboreiro is a secluded and remote area of the country that was not easily accessed until recent times. Therefore, it is possible that the breed has existed under isolation for a long period. The Serra da Estrela Mountain Dog is a molossoid breed thought to have been used for farm and herd guarding against wolves since ancestral times and its origin is also unknown. Two types of coat (long and short) have developed over time. The Portuguese Sheepdogs are thought to descend from a pair of Briard Sheepdogs, imported 100 years ago from France to Portugal. They are used in the Alentejo region for guarding and driving herds. The Azores Cattle Dog originated in São Miguel Island, where it is still used in cattle and property guarding. They are thought to descend from dogs brought into the island by Portuguese, Spanish, Flemish and French colonisers and travellers since the XV century. The breed is referred to since the beginning of the 20^th^century. To different extents, all 4 breeds have experienced decreasing populations and have lost popularity over time. This is mostly due to the alteration of agricultural practices that rendered working dogs less useful, and to the introduction of competitive exotic breeds. These indigenous breeds are now winning back attention and their status is generally improving. Nevertheless, they still comprise relatively small population sizes, with registers at the Portuguese Kennel Club ranging between 96 Castro Laboreiro and 763 Serra da Estrela Mountain individuals in 2003. We tried to avoid one of the possible sampling biases by analysing a significant proportion of the animals registered in the studied breeds and we selected dogs from several kennels in order to minimise the relatedness of individuals and survey the internal diversity of the breed. Results & discussion ==================== MtDNA diversity in four Portuguese dog breeds --------------------------------------------- A total of 15 different haplotypes was found in 143 dogs, five of them newly described (Fig. [1](#F1){ref-type="fig"}). These haplotypes belong to the four major haplogroups A, B, C and D and their frequencies were 77%, 8%, 10% and 5%, respectively. The considerably higher frequency of haplogroup A in Portugal, when compared to the 68% observed in Europe by Savolainen et al\[[@B1]\], results from the 100% prevalence observed in the Castro Laboreiro Dog, as discussed below. The median-joining network (Fig. [2](#F2){ref-type="fig"}) displaying the Portuguese haplotypes shows high diversity between haplotypes belonging to different haplogroups, and also between haplotypes inside the same haplogroup, particularly in haplogroup A, thus indicating inputs from different founders rather than local population expansions. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Sequence alignments showing base substitutions and indels in the mtDNA D-loop region (582 bp) of 143 Portuguese dogs.Sequence codes and numbers are given in the first column, with asterisks for previously published sequences. Only variable sites are shown, with sequence positions given above. Nucleotide positions are numbered according to Pereira et al(2004). Identity with the reference sequence (A19) is denoted by a stop, substitution by a different base letter, and deletions by a dash. Last four columns indicate the number of individuals in each breed (CL-Castro Laboreiro Dog; SE-Serra da Estrela Mountain Dog; PS-Portuguese Sheepdog; AC-Azores Cattle Dog). ::: ![](1471-2156-6-37-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Median-joining network of mtDNA haplotypes observed in four Portuguese dog breeds.Circle sizes are proportional to the number of individuals they represent. White, light grey, dark grey, and black represent Castro Laboreiro Dog, Serra da Estrela Mountain Dog, Portuguese Sheepdog and Azores Cattle Dog, respectively. ::: ![](1471-2156-6-37-2) ::: Among the four breeds studied here (Table [1](#T1){ref-type="table"}), the Castro Laboreiro Dog (n = 58) presented the most differentiated history. This breed has the lowest level of diversity, as only two A haplotypes were found. Strikingly, a new A haplotype (IA), one step apart from haplotype A24 (a 15631A → G transition), is shared by 95% of the sample. Haplotype IA was also observed in two Serra da Estrela Mountain dogs. The remaining 5% display haplotype A33, also present in four Serra da Estrela Mountain individuals. Haplotype A33 was previously described only in one Sloughi dog from North Africa \[[@B1]\], two Retrievers and one unspecified dog \[[@B10]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Summary statistics of diversity measures for 13 dog breedsDiversity measures for 4 Portuguese breeds and 13 other for which at least 10 individuals have been typed in previous studies \[1,10,11,12\]. ::: Breed n Number of Haplotype diversity Mean n° of pairwise differences Nucleotide diversity ------------------------------- ---- ------------ --------------------- --------------------------------- ---------------------- --------------- haplotypes haplogroups Castro Laboreiro Dog 58 2 1 0.100 ± 0.052 0.299 ± 0.317 0.001 ± 0.001 Azores Cattle Dog 23 5 3 0.640 ± 0.065 7.012 ± 3.418 0.012 ± 0.007 Portuguese Sheepdog 28 5 3 0.484 ± 0.110 4.780 ± 2.408 0.008 ± 0.005 Serra da Estrela Mountain Dog 34 8 4 0.852 ± 0.030 9.802 ± 4.601 0.017 ± 0.009 German Shepherd ^\[1,12\]^ 16 3 2 0.575 ± 0.112 4.500 ± 2.338 0.008 ± 0.004 Retriever Labrador ^\[1,10\]^ 22 7 2 0.771 ± 0.060 4.554 ± 2.327 0.008 ± 0.004 Basenji ^\[1\]^ 10 3 1 0.644 ± 0.101 2.156 ± 1.304 0.004 ± 0.003 Akbasch^\[1\]^ 11 5 2 0.618 ± 0.164 3.455 ± 1.909 0.006 ± 0.004 Canaan Dog^\[1\]^ 17 6 2 0.831 ± 0.056 7.801 ± 3.821 0.013 ± 0.007 Saluki ^\[1,11\]^ 16 8 2 0.808 ± 0.093 7.875 ± 3.867 0.014 ± 0.007 Kangal^\[1\]^ 10 5 4 0.822 ± 0.097 9.622 ± 4.823 0.017 ± 0.009 Ryukyu ^\[11,12\]^ 14 7 3 0.879 ± 0.058 8.924 ± 4.377 0.015 ± 0.008 Shiba ^\[11\]^ 28 8 3 0.791 ± 0.049 6.717 ± 3.265 0.012 ± 0.006 ::: All major haplogroups were observed in Serra da Estrela Mountain dogs (n = 34). The breed presents the highest values of diversity measures, implying that it harbours a comparatively more variable gene pool. It shares five out of its eight haplotypes with other breeds, namely two with the Castro Laboreiro Dog (A33 and IA) and three with the Portuguese Sheepdog (A18, A24 and B1). The breed also displays a new haplotype (ID) belonging to haplogroup D, this latter being common in Scandinavia but rarely observed elsewhere. D haplotypes were previously described in eight Scandinavian breeds (n = 18), two Kangal Dogs (Turkey) and one Galgo Español (Spain) \[[@B1]\]. Newly reported ID was also found in seven Serra da Estrela Mountain dogs. However, it is closer to Turkish D5 and Spanish D6 (respectively 2 and 1 mutational steps apart) than to Scandinavian D1, D2, D3 and D5 (Fig. [3](#F3){ref-type="fig"}). Clearly, genetic inputs from diverged populations must have repeatedly enriched the genetic diversity of the Serra da Estrela Mountain dogs. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Median-joining network of dog mtDNA haplotypes belonging to haplogroup D.Haplotypes D1, D2, D3 and D4 were observed in Scandinavian dogs, while D5 was observed in Turkey, D6 in Spain and ID in this work. Circle sizes are proportional to the number of individuals they represent. ::: ![](1471-2156-6-37-3) ::: The Portuguese Sheepdog (n = 28) presented intermediate levels of diversity when compared to the Castro Laboreiro Dog and the Serra da Estrela Mountain Dog, mainly because one haplotype (A18) was present in 71% of the sample. This haplotype has a worldwide distribution and was also found in one Serra da Estrela Mountain dog. Haplotype A24 is also shared with the Serra da Estrela Mountain breed and was previously reported in British (n = 4) and Pyreneean breeds (n = 3). Haplotype B1, also present in the Serra da Estrela Mountain Dog, was observed in two individuals. A new B haplotype (IB), one mutational step apart from haplotype B1 (a 15651C → T transition), so far exclusive to the breed, was also observed in three dogs. A new C haplotype (IC), one mutational step apart from haplotype C3 (a 15672G → A transition), was displayed by one individual. The Azores Cattle Dog (n = 23), an insular Portuguese breed, presented the second highest levels for diversity measures following the Serra da Estrela Mountain Dog, but did not share lineages with the other breeds surveyed in mainland Portugal. A single individual presented a newly observed haplotype (IIB), one-step apart from haplotype B1 (a 15955T → C transition), which is widely found in Europe but was not observed in this breed. Nearly half of the individuals (n = 11) presented haplotypes that are more frequent in the northern region of Europe (A2, A5 and A17), and quite differentiated from the other A haplotypes found in the other Portuguese breeds (Fig. [2](#F2){ref-type="fig"}). The remaining individuals display haplotype C3, which was also reported in North Europe and in Asia \[[@B1],[@B6],[@B11]\]. Comparative analysis of mtDNA haplotype structure ------------------------------------------------- We collected published data on all breeds for which at least ten individuals were typed \[[@B1],[@B10]-[@B12]\]. Only 13 breeds met this criterion: two European (German Shepherd and Retriever Labrador), four Middle-Eastern (Akbasch, Canaan Dog, Saluki and Kangal), two Asian (Ryukyu and Shiba), one African (Basenji) and the four breeds here described. Although these numbers are still undesirably low, they can nevertheless provide insights into the apportionment of overall domestic dog mtDNA diversity in terms of intra-and interbreed components. The summary statistics of diversity measures are shown in Table [1](#T1){ref-type="table"}. Regarding haplotype diversity, we observed that all breeds but the Castro Laboreiro Dog and marginally the Portuguese Sheepdog, showed values above 0.5, which immediately reveals considerable intra-breed heterogeneity. This heterogeneity, however, could be due to slight nucleotide differences: it could be that different haplotypes within a breed were assigned to the same haplogroups, which in turn were not shared across breeds. This is not the case, for (a) only two breeds (Castro Laboreiro Dog and South African Basenji, although in the latter only 10 individuals were typed) show haplotypes belonging to the same haplogroup and (b) haplogroup A is present in all breeds and haplogroup B, the second most frequent haplogroup, is shared by several breeds from different continental areas. Even the relatively infrequent haplogroup C shows substantial sharing among breeds. In relation to haplogroups D and E, due to their overall rarity, it is premature to speculate. Still more remarkable is the fact that the number of haplotypes shared between breeds inside a region is generally lower than the number of haplotypes shared between breeds from different regions (Fig. [4](#F4){ref-type="fig"}). In Europe the figures are, respectively, 6 and 7; in the Middle East, 5 and 5; and in Asia, 0 and 8 shared haplotypes. These values are still debatable due to the low number of individuals typed per breed and the limited number of breeds analysed so far, particularly in Africa and America. Nonetheless, it indicates that many of the extant domestic dog breeds, despite their phenotypic homogeneity, are substantially heterogeneous in their maternal ancestry. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### MtDNA haplotype sharing between dog breeds and world regions.Thirteen breeds for which at least 10 individuals were typed are included. Squares represent world regions and state haplotypes observed therein. Figures inside brackets indicate the number of breeds where the particular haplotype was observed. Circles contain haplotypes shared between breeds from different regions; numbers enclosed in parenthesis are to be read from right top to left down, and correspond to the number of breeds in which a haplotype is observed in each continental region. For example, \"A18 (3:1:1)\" indicates that haplotype A18 was found in 3 European, 1 Asian and 1 Middle-Eastern breeds. ::: ![](1471-2156-6-37-4) ::: In order to quantify the partition of the overall diversity into separate components, we performed AMOVA in two ways: (a) when considering all the 13 breeds as a single group, 32.80% of the variation was found to be due to inter-breed diversity and 67.30% was attributable to the intra-breed component; (b) when considering the breeds grouped in 4 worldwide regions, the resulting values were remarkably similar: 29.51% for inter-breed; 66.02% for intra-breed; and 4.47% for diversity between regions. These values demonstrate that the geographic component of the domestic dog mtDNA diversity, at least at the broad continental level, is very low. Almost all pairwise genetic distances (measured as F~ST~s) between the 13 breeds were statistically significant (data not shown) except for the pairs Azores Cattle Dog (Europe)-Kangal (Middle East) (p= 0.063); Portuguese Sheepdog (Europe)-Ryukyu (Asia) (p= 0.063); Saluki (Middle East)-Canaan Dog (Middle East) (p= 0.396); Canaan Dog (Middle East)-Ryukyu (Asia) (p= 0.180); Kangal (Middle East)-Ryukyu (Asia) (p= 0.117), again evidencing no geographical correlation. Conclusion ========== This fine-scale survey undertaken on four autochthonous Portuguese dog breeds revealed different histories, even for geographically close breeds with similar selection characteristics. The unprecedented low mtDNA diversity observed in Castro Laboreiro dogs contrasts with the high diversity displayed by Serra da Estrela Mountain dogs, both of them herding breeds with putative centres of origin just 200 km apart. The Castro Laboreiro Dog has clearly suffered a population bottleneck (although the hypothesis of a low number of female lineages at its origin can not be excluded) that severely reduced its mtDNA diversity. In contrast, the Serra da Estrela Mountain Dog mtDNA diversity shows that the breed has not remained in isolation. This may also indicate that if population bottlenecks have occurred in this breed, subsequent expansions included dogs from other origins. A peculiar pattern of haplotype sharing between the Portuguese breeds was also detected: just between the Serra da Estrela Mountain Dog (with putative origin in the centre of the country), and the breeds to its north and south. The Serra da Estrela Mountain Dog presented two haplotypes in common with the north-western Castro Laboreiro breed (which behaves, at the mtDNA level, as a sub-sample of the Serra da Estrela Mountain breed) and three with the southern Portuguese Sheepdog breed. It would be interesting to compare the Portuguese Sheepdog mtDNA haplotypes with those of the Briard Sheepdogs, from which they are supposedly descending. Contrastingly, the insular Azores Cattle Dog did not share any haplotypes with continental Portuguese dogs, but mostly with dogs from Northern Europe. This suggests that the Azores Cattle Dog could descend maternally from Northern European rather than Portuguese mainland dogs, which is consistent with the historical reports. When comparing breeds from different regions at a worldwide scale, it is Asia that harbours most of the haplotype sharing, in accordance with the postulated domestication centre \[[@B1]\]. The broad geographic pattern of haplotype sharing is, however, paradoxical, because haplotype sharing is just as likely to occur between breeds from the same continent as between breeds from distant ones. The observation of molecularly distant haplotypes within most breeds indicates recent and polyphyletic origins for the majority of female gene pools of the current breeds. Although it is clear that mtDNA haplotypes have poor breed discriminating power, they are valuable for determining the origin and population history of dog breeds. We expect that the data we present here and further developments will allow for better resolution of the population histories of European dog breeds, since they are still poorly sampled. Methods ======= Sample collection and DNA extraction ------------------------------------ A total of 143 individuals were selected for the study based on breed assignment (Azores Cattle Dog, n = 23; Castro Laboreiro Dog, n = 58; Portuguese Sheepdog, n = 28; Serra da Estrela Mountain Dog, n = 34) excluding first order relatives in order to reduce sampling bias. A large number of individuals per breed (mean = 36) collected throughout the geographical areas of origin, also contributed to maximise the informative power of the sample. Samples consisted in whole blood on FTA (Whatman) cards or buccal epithelium cell swabs collected with cytology brushes. DNA from blood samples was extracted according to standard Chelex 100 or phenol-chloroform methods. Buccal swab samples were treated with a standard salting-out extraction protocol. DNA amplification and sequencing -------------------------------- Two overlapping fragments of the mtDNA D-loop region were amplified so that a 582-bp fragment between positions 15458--16039 was surveyed. Polymerase chain reaction (PCR) was performed in 35 cycles after an initial denaturing step of 95°C for 2 min. For primer pair A (5\'-TTA CCT TGG TCT TGT AAA CC-3\' and 5\'-CTG AAG TAA GAA CCA GAT GCC-3\'), conditions were 95°C for 30 s, 58°C for 30s, and 72°C for 30s. For primer pair B (5\'-CAT ACT AAC GTG GGG GTT AC-3\' and 5\'-CCA TTG ACT GAA TAG CAC CTT G-3\'), conditions were 95°C for 30 s, 60°C for 30 s, and 72°C for 30s. A final extension step of 72°C for 10 min was performed in both cases. PCR mixture consisted of 2 μL of DNA extract, 0.25 μM of each primer, 1.5 mM of MgCl~2~, 0.2 mM of dNTPs, 67 mM Tris-HCl pH 8.8, 16 mM (NH~4~)~2~SO~4~, 0.01% Tween 20 and 0.5 units of Taq DNA Polymerase (Bioron) in a total volume of 12.5 μL. Negative controls were included (one in each PCR) to monitor for contamination and resulted in no amplification product. The reaction was run on a GeneAmp PCR System 2400 (Perkin Elmer). The sequencing reaction was performed on forward and reverse directions using 2.5 μL of purified PCR product, 0.25 μM of primer (A and B) and 2 μL of big dye terminator (Abi Prism Big Dye Terminator Cycle Sequencing Kit v3.1) for a total reaction volume of 5 μL. The cycle program consisted of an initial step of denaturation (96°C, 2 min), 35 cycles (96°C, 15 s; 50°C, 9 s, 60°C, 2 min) and a final extension step (60°C, 10 min). The reaction was run on a GeneAmp^®^PCR System 2700 (Applied Biosystems). The cycle sequencing products were analysed on an ABI 3100 Automated Sequencer (Applied Biosystems) according to the manufacturer\'s directions. Data analysis ------------- The sequences obtained were compared by alignment (Fig. [1](#F1){ref-type="fig"}) to the reference sequence \[[@B8]\], indicated as haplotype A19 (GenBank accession entry: NC\_002008) and polymorphisms were numbered accordingly to Pereira et al\[[@B7]\]. Medium-spanning networks including previously published data were calculated using Network 4.0 <http://www.fluxus-engineering.com>, equally weighting polymorphisms. Diversity measures, AMOVA and F~ST~distances were obtained using Arlequin 2.0 software \[[@B9]\]. Authors\' contributions ======================= BA carried out most of the sample processing, participated in the sequence alignment and drafted the manuscript. LP participated in the conception and co-ordination of the study, performed the statistical analysis and helped with subsequent drafts. FP participated in the sample processing and the sequence alignment. PS-R and ML sampled the dogs studied and provided background information. AA conceived the study, participated in its design and co-ordination and helped to prepare the final manuscript. All authors read and approved the final manuscript. Acknowledgements ================ The authors would like to acknowledge the breed clubs, breeders and dog owners that provided samples and data, especially AAC-CCL (Associação dos Amigos e Criadores do Cão de Castro Laboreiro), Pedro Delerue (CPCSA-Clube Português do Cão da Serra de Aires) and Suzette Preiswerk da Mota Veiga (LICRASE-Liga dos Criadores e Amigos do Cão da Serra da Estrela). IPATIMUP is supported by POCTI-QCIII. This study is included in project GENCERT, financed by POCTI Medida 2.3 and POSI Medida 1.3 (PO-QCA III). The authors are grateful to three anonymous referees for comments and suggestions.	PubMed Central	2024-06-05T03:55:59.624951	2005-6-22	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180824/", "journal": "BMC Genet. 2005 Jun 22; 6:37", "authors": [ { "first": "Barbara", "last": "van Asch" }, { "first": "Luísa", "last": "Pereira" }, { "first": "Filipe", "last": "Pereira" }, { "first": "Pedro", "last": "Santa-Rita" }, { "first": "Manuela", "last": "Lima" }, { "first": "António", "last": "Amorim" } ] }
PMC1180825	Background ========== Upon completion and annotation of the nucleotide sequence of a bacterial genome, a great scientific challenge is to elucidate the regulation of expression of all the predicted genes and to deduce thereof the entirety of regulatory networks present in the respective microorganism. An important prerequisite in understanding regulation of gene expression on the global scale is the identification of the repertoire of regulatory proteins in a genome sequence \[[@B1]\]. DNA-binding transcription factors are key components in the regulation of gene expression because they rapidly respond to changes in the cellular environment by modulating the expression of relevant genes. Three-dimensional structure analyses and amino acid sequence comparisons of DNA-binding transcription factors enabled their allocation into distinct protein classes that apparently use specific structural motifs for DNA recognition and binding \[[@B2]\]. The helix-turn-helix (HTH) motif is obviously the most widely distributed DNA-binding domain in prokaryotic proteins and provides the structural basis for efficient protein-DNA interactions \[[@B3]\]. Although a considerable amino acid sequence divergence has been observed among HTH proteins, they generally share a site-specific DNA-binding domain that is composed of two α-helices separated by a short turn of variable length \[[@B4]\]. Both α-helices are involved in the DNA-binding and recognition process in such a way that the first helix associates non-specifically with the DNA molecule while the second helix recognizes and binds specifically to its cognate operator sequence \[[@B5]\]. Based on variations of the three-dimensional structure of the HTH motif, DNA-binding transcription factors can be subdivided into several DNA-binding domain types \[[@B6]\]. The canonical \'winged helix\' type, for instance, consists of two wings, three α-helices and three β-strands and is present in several families of DNA-binding transcription factors. In this case, the third α-helix is the DNA recognition helix, whereas the first and second ones are involved in stabilizing the DNA-binding and recognition process \[[@B3]\]. An interesting outcome of comparative studies between DNA-binding transcription factors was a position-function correlation such that repressor proteins usually possess the HTH motif within the N-terminal region, whereas activators tend to have the HTH motif close to the C-terminal end of the protein \[[@B1]\]. Moreover, the putative physiological role of a DNA-binding transcription factor can be deduced from its classification into an evolutionary regulatory protein family, since within many families the members are homogenous in respect of their regulatory role and the physiology of the regulated genes \[[@B1]\]. Because of the importance of corynebacteria in biotechnology as well as in human medicine they represent an attractive target to elucidate and compare the repertoire of DNA-binding transcription factors by bioinformatic approaches. Corynebacterium glutamicumand its closest phylogenetic relative Corynebacterium efficiensare both widely known for their capacity to produce amino acids by large-scale fermentation processes \[[@B7],[@B8]\]. On the other hand, Corynebacterium diphtheriaeis the etiological agent of the acute, communicable disease diphtheria and apparently the most important human pathogen of the genus Corynebacterium\[[@B9]\], which also includes a growing number of nosocomial pathogens, such as the multiresistant Corynebacterium jeikeium\[[@B10]\]. Accordingly, complete genome sequences of the four corynebacterial species have been determined and annotated recently \[[@B11]-[@B15]\]. Subsequent synteny analyses of the predicted coding sequences revealed that the four corynebacteria have largely maintained an ancestral genome structure \[[@B15],[@B16]\]. Therefore, we were not only interested in the identification and classification of the individual DNA-binding transcriptional regulatory repertoire of each of the four corynebacterial species by means of different bioinformatic tools but also in comparative genomic approaches to deduce thereof the common set of regulatory genes. The knowledge on conserved regulatory genes encoded by the corynebacterial core genome certainly provides a solid basis for the future analysis of transcriptional regulatory networks in pathogenic and non-pathogenic corynebacteria. Results and discussion ====================== The repertoire of DNA-binding transcriptional regulators in corynebacterial genomes ----------------------------------------------------------------------------------- To fully understand the regulation of gene expression in a bacterial cell it is necessary to identify and characterize the repertoire of DNA-binding transcriptional regulators with respect to regulatory and physiological properties. Therefore, we have screened the complete genome sequences of four corynebacteria by different bioinformatic tools to detect genes encoding DNA-binding transcriptional regulators. Other gene products typically exhibiting regulatory properties on the transcriptional level, such as sigma factors or two-component systems, were excluded from the present study. The work flow for the identification and analysis of transcriptional regulatory proteins is described in detail in the Methods section. Following three consecutive steps of data collection, a total number of 348 DNA-binding transcriptional regulators were identified in the genomes of C. glutamicumATCC 13032, C. efficiensYS-314, C. diphtheriaeNCTC 13129 and C. jeikeiumK411. A compilation of the data characterizing the DNA-binding transcriptional regulators of each species is provided as supplementary material (see Additional files [1](#S1){ref-type="supplementary-material"},[2](#S2){ref-type="supplementary-material"},[3](#S3){ref-type="supplementary-material"} and [4](#S4){ref-type="supplementary-material"}). An overview of the resulting data along with a genome comparison between the four corynebacterial species is presented in Table [1](#T1){ref-type="table"}. A collection of 127 DNA-binding transcriptional regulators was identified in the genome of C. glutamicum, whereas 103 transcriptional regulators were identified in the chromosomal sequence of C. efficiens, 63 in C. diphtheriaeand only 55 in C. jeikeium. Three additional genes encoding DNA-binding transcriptional regulators were detected on the endogenous plasmid pCE3 of C. efficiensYS-314 (data not shown). Considering the differences in genome size of the four corynebacterial species, it became apparent that a larger number of coding regions within a genome sequence requires more genes that encode DNA-binding transcriptional regulators (Table [1](#T1){ref-type="table"}). The identified regulatory proteins represent 4.2 and 3.5 %, respectively, of the predicted coding sequences in the genomes of the non-pathogenic corynebacteria C. glutamicumand C. efficiens, whereas the percentage is reduced to 2.7 and 2.6 %, respectively, in the pathogenic corynebacterial species C. diphtheriaeand C. jeikeium. These data are consistent with a previous observation that an increase of genomic complexity and physiological functionality is generally associated with a more complex regulation of gene expression since the additional genetic information has to be integrated into the existing regulatory networks that operate in a bacterial cell \[[@B17]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Comparison of sequenced corynebacterial genomes ::: Feature C. glutamicum C. efficiens C. diphtheriae C. jeikeium ---------------------------- ----------------- ---------------- ------------------ --------------- Genome size 3,282,708 bp 3,147,090 bp 2,488,635 bp 2,462,499 bp Number of coding sequences 3,002 2,950 2,320 2,104 Number of regulators 127 103 63 55 Percentage of regulators 4.2 % 3.5 % 2.7 % 2.6 % ::: Classification of corynebacterial DNA-binding transcriptional regulators into regulatory protein families --------------------------------------------------------------------------------------------------------- DNA-binding transcriptional regulators can be grouped into evolutionary regulatory protein families based on their amino acid sequence similarity \[[@B1]\]. According to amino acid sequence alignments performed with the CLUSTAL X program, the complete collection of corynebacterial transcriptional regulators fell into 25 protein families with only a small number of regulators that remained unclassified (Figure [1](#F1){ref-type="fig"}). Most of the identified regulatory protein families can be regarded as homogenous with respect to the size distribution of the assigned family members (see Addional files [1](#S1){ref-type="supplementary-material"}, [2](#S2){ref-type="supplementary-material"}, [3](#S3){ref-type="supplementary-material"} and [4](#S4){ref-type="supplementary-material"}). At least in some cases the high degree of amino acid sequence similarities along with the conserved protein size within a regulatory protein family indicate that the respective family members derived from a common ancestor. Therefore, one can assume that members of homogenous regulatory protein families tend to affect the expression of genes involved in related physiological functions of the cell \[[@B1]\]. On the other hand for instance, the HTH\_3 family contains several members of rather heterogenous size (see Addional files [1](#S1){ref-type="supplementary-material"}, [2](#S2){ref-type="supplementary-material"}, [3](#S3){ref-type="supplementary-material"} and [4](#S4){ref-type="supplementary-material"}) reflecting a lower degree of similarity of proteins within this family. Accordingly, a more diverse evolutionary history and thus physiological functionality in corynebacteria can be predicted for the members of the HTH\_3 family. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Classification of DNA-binding transcriptional regulators of corynebacteria into regulatory protein families. The identified regulatory protein families are indicated along with the number of assigned family members. The rightmost columns of the diagram comprise a small number of transcriptional regulators that remained unclassified. The regulatory protein families were named according to designations by the Pfam database. ::: ![](1471-2164-6-86-1) ::: The identified regulatory protein families vary significantly in their number of representatives, ranging from TetR, the largest family with up to 16 members, to protein families with a single member, for instance ArgR, FUR, HrcA, LexA and YbaD (Figure [1](#F1){ref-type="fig"}). The protein families ArgR, HrcA and LexA are characterized in bacterial genomes predominantly by single representatives that are involved in the regulation of arginine metabolism, the heat shock response and the SOS repair pathway of the cell respectively \[[@B18]\]. Most of the regulatory protein families with a few members have homologs in the four corynebacterial genomes with the exception of the PadR and ROK families that are absent in C. jeikeium(Figure [1](#F1){ref-type="fig"}). Further exceptions are single regulators grouped into the RpiR family of C. glutamicumand into the FIS family of C. efficiens. In principle, the overall composition of the repertoire of DNA-binding transcriptional regulators identified in the non-pathogenic corynebacteria C. glutamicumand C. efficiensseemed to be very similar, which was obvious when considering the close phylogenetic relationship of both species \[[@B8]\]. The average number of members per regulatory protein family was calculated as 4.1 for C. efficiensand 5.1 in case of C. glutamicum, suggesting that species-specific differences are mainly caused by slightly smaller numbers of regulatory proteins per family in C. efficiens. On the other hand, the average number of members per regulatory protein family is in the order of 2.6 for the pathogenic corynebacteria C. diphtheriaeand C. jeikeium. Remarkable differences between the transcriptional regulatory repertoire of pathogenic and non-pathogenic corynebacteria became apparent when comparing, for instance, the number of proteins grouped into the ArsR, IclR and LacI families (Figure [1](#F1){ref-type="fig"}). The ArsR family of transcriptional regulators contains metalloregulatory proteins that control genes whose expression is linked to stress-inducing concentrations of heavy metal ions \[[@B19]\], whereas members of the IclR and LacI families generally respond to environmental changes that affect the carbohydrate metabolism of the cell \[[@B20]\]. It certainly makes sense that soil bacteria have a large diversity of DNA-binding transcriptional regulators that function as metal sensors or respond to changes in the carbohydrate composition of the environment. The larger number of proteins grouped into the IclR and LacI families may provide these bacteria the ability to grow in the presence of several carbon sources and to rapidly adapt their gene expression to changing nutrient conditions. The observed differences in the number of transcriptional regulators involved in carbohydrate metabolism may also reflect the fact that pathogenic bacteria do not require a versatile sugar metabolism since only a limited range of carbohydrate nutrients might be present in their natural habitats. Variations in the number of metalloregulatory sensors may also be linked to the different environmental conditions such that pathogenic bacteria predominantly import metal ions directly from the host rather than from inanimate environment. It is noteworthy that the number of regulatory proteins of the TetR family is otherwise not reduced significantly in the pathogenic species (Figure [1](#F1){ref-type="fig"}). This observation implies that the TetR repressor family provides a very common switch for the regulation of gene expression in corynebacteria. Classification of corynebacterial transcriptional regulators according to DNA-binding domain types -------------------------------------------------------------------------------------------------- The transcriptional regulators were additionally grouped according to DNA-binding domain types that specify variations of the three dimensional structure of the HTH motif. The DNA-binding domain types were detected by using the domain assignment server SUPERFAMILY \[[@B21]\]. We found that only six DNA-binding domain types have representatives in corynebacteria with the \'winged helix\' type being the most prominent HTH motif (Table [2](#T2){ref-type="table"}). Especially the hairpin wings of the \'winged helix\' display considerable flexibility in their utilization for DNA-binding and alternative modes of DNA recognition \[[@B3]\]. Other DNA-binding domains were also identified in corynebacteria, but only in a few regulatory proteins (Table [2](#T2){ref-type="table"}). This includes DNA-binding transcriptional regulators of the WhiB protein family that were suggested to act as transcriptional activators \[[@B22]\]. Structure-function predictions for WhiB proteins indicated that the most C-terminal α-helix is a likely candidate for DNA-binding. Additionally, an aminoterminal Zinc β-ribbon domain combined with an ATP-cone domain was detected in the single member of the YbaD protein family, which is probably involved in a hitherto unknown mechanism of transcriptional regulation of ribonucleotide reductase expression in bacteria \[[@B23]\]. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Domain architecture of corynebacterial DNA-binding transcriptional regulators ::: DNA-binding domain type C. glutamicum C. efficiens C. diphtheriae C. jeikeium ----------------------------- ----------------- ---------------- ------------------ --------------- Winged helix 73 60 32 26 Homeodomain-like 20 17 14 14 λ repressor-like 19 12 8 3 Putative DNA-binding domain 5 5 4 6 C-terminal effector domain 3 3 1 1 FIS-like \- 1 \- \- C-terminal α-helix\* 4 4 3 4 Zinc β-ribbon 1 1 1 1 Unclassified 2 \- \- \- \* Putative DNA-binding domain named according to Soliveri et al.\[22\]. ::: Subsequently, the position of the DNA-binding domain within the transcriptional regulators in combination with the regulatory protein family membership was used to predict whether a protein is expected to act as repressor or activator of gene expression. The position of the DNA-binding motif was determined by both the HTH prediction tool \[[@B24]\] and the domain assignment server SUPERFAMILY \[[@B21]\]. This computational approach resulted in an average distribution of DNA-binding transcriptional regulators in the four corynebacterial species of 73.8 % repressors, 16.4 % activators and 9.8 % dual regulators. Dual transcriptional regulators are either activators of several genes and repressors of their own synthesis or activators and repressors of different sets of genes \[[@B1]\]. In particular, the members of the AsnC, Crp and LysR protein families represent dual regulators of gene expression whereas the AraC, LuxR, MerR, and WhiB protein families were regarded as activators of gene expression. However, one has to keep in mind that some regulatory protein families, for instance IclR, MarR and MerR, may include both activator and repressor proteins. Nevertheless, considering that approximately three-quarters of the DNA-binding transcriptional regulators are repressor proteins it is apparent that mechanistic requirements for repression are dominant in the architecture of regulatory networks in corynebacteria. A similar analysis of 314 regulatory proteins of Escherichia colirevealed a different picture in such a way that a quite even distribution of 42.8%, 34.8% and 22.3% of repressors, activators and dual regulators was predicted for this species \[[@B1]\]. The common repertoire of DNA-binding transcriptional regulators identified in four corynebacterial genome sequences ------------------------------------------------------------------------------------------------------------------- We were then interested to specify those DNA-binding transcriptional regulators that are common in the four sequenced corynebacterial genomes. For this purpose, orthologous proteins were identified in the complete collection of DNA-binding transcriptional regulators by using the BLASTP algorithm to detect amino acid sequence similarities and by performing synteny analyses of the respective genomic context. This comparative content analysis of transcriptional regulators allowed us to calculate the number of shared and species-specific regulatory proteins among the four corynebacteria. The resulting data were summarized in the Venn diagrams of Figure [2](#F2){ref-type="fig"}. It is striking that the genomes of C. glutamicumand C. efficiensshare 77 transcriptional regulators suggesting that many regulatory networks might be conserved in both species and controlled by homologous regulatory functions. This is consistent with the very similar distribution of DNA-binding transcriptional regulators of both species within the identified regulatory protein families (Figure [1](#F1){ref-type="fig"}). Approximately 75 % of the transcriptional regulators detected in C. diphtheriaeare shared with the repertoire of regulatory proteins of the non-pathogenic corynebacteria C. glutamicumand C. efficiens(Figure [2](#F2){ref-type="fig"}). Consequently, both species provide valuable model systems for investigating the regulation of gene expression in C. diphtheriaesince it is likely that not only the regulatory proteins but also the corresponding regulatory networks are somehow conserved. The composition of the small regulatory repertoire of C. jeikeiumis much more intriguing since virtually all of the DNA-binding transcriptional regulators are either species-specific or shared with the other three corynebacteria. This finding might reflect the distant phylogenetic relationship between C. jeikeiumand the investigated corynebacterial species \[[@B25]\]. By combining the data of the comparative content analyses, the common set of DNA-binding transcriptional regulators present in C. glutamicum, C. efficiens, C. diphtheriaeand C. jeikeiumwas finally confined to only 28 proteins (Table [3](#T3){ref-type="table"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Comparative content analysis of genes encoding DNA-binding transcriptional regulators in sequenced corynebacterial genomes. The Venn diagrams show the number of shared and species-specific genes among the four genomes. Abbreviations: Cg, C. glutamicumATCC 13032; Ce, C. efficiensYS-314; Cd, C. diphtheriaeNCTC 13129; Cj, C. jeikeiumK411. ::: ![](1471-2164-6-86-2) ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### The common set of DNA-binding transcriptional regulators in corynebacteria ::: Functional category CDS in C. glutamicumATCC 13032 Orthologous CDS in --------------------------- ---------------------------------- -------------------- ---------- ----------- ---------- Cell division & septation cg0878 whiB1 ce0783 dip0712 jk1618 cg0850 whiB2 ce0758 dip0684 jk1644 cg0337 whiB4 ce0283 dip0299 jk1976 SOS & stress response cg2109 oxyR ce1817 dip1421 jk1102 cg2114 lexA ce1823 dip1426 jk1106 cg2152 clgR ce1855 dip1456 jk1122 cg2516 hrcA ce2190 dip1721 jk0600 cg3097 hspR ce2626 dip2117 jk0184 cg1765 ArsR family ce1687 dip1296 jk0985 Macroelement & metal cg2103 dtxR ce1812 dip1414 jk1097 homeostasis cg2502 furB ce2180 dip1710 jk0612 cg3253 mcbR ce2788 dip2274 jk0101 cg1631 MerR family ce1574 dip1205 jk0904 cg1633 MerR family ce1576 dip1207 jk0906 Carbohydrate metabolism cg0350 glxR ce0287 dip0303 jk1972 cg0444 ramB ce0385 dip0369 jk1934 cg1738 acnR ce1663 dip1284 jk0970 cg2115 DeoR family ce1824 dip1427 jk1107 cg1486 IclR family ce1426 dip1126 jk1222 cg2910 LacI family ce2511 dip1969 jk0329 Biosynthesis pathways cg1585 argR ce1531 dip1172 jk0846 cg2112 YbaD family ce1820 dip1424 jk1105 Unknown cg3261 GntR family ce2809 dip2280 jk0088 cg2831 LuxR family ce2445 dip1889 jk0397 cg3001 MarR family ce2556 dip2008 jk0271 cg3315 MarR family ce2826 dip2296 jk2061 cg0454 TetR family ce0397 dip0937 jk1455 cg1053 TetR family ce0985 dip0888 jk1501 ::: According to functional assignments deduced from computational predictions and protein similarities, 22 of the common DNA-binding transcriptional regulators were grouped into five functional categories (Table [3](#T3){ref-type="table"}). The physiological role of the remaining six transcriptional regulators remained unknown since the potential targets of their regulation were difficult to predict from the existing data and the genome organization. The resulting functional classification is of particular interest since members of four categories are apparently involved in the control of fundamental processes of the bacterial life style, namely cell division and septation, SOS and stress response, carbohydrate metabolism and macroelement and metal homeostasis (Table [3](#T3){ref-type="table"}). A fifth category comprises only single representatives of the ArgR family and the YbaD family that are involved in the regulation of specific biosynthesis pathways \[[@B18],[@B23]\]. ### Cell division and septation The first functional category of common DNA-binding transcriptional regulators includes three members of the WhiB protein family. This regulatory protein family represents a specific group of transcriptional activators that appeared to be present in perhaps all actinobacteria while being absent from all other sequenced bacterial genomes \[[@B22]\]. Members of the WhiB protein family were postulated to function in cell division and septation of mycobacteria \[[@B26]\] and in differentiation of streptomycetes \[[@B27]\], possibly by sensing redox changes in the environment or internally during metabolic shifts that occur as inescapable part of alterations in the cellular metabolism. The genomes of C. glutamicum, C. efficiensand C. jeikeiumcarry four whiBhomologs while a whiB3ortholog is missing in the C. diphtheriaegenome. In this context it is noteworthy that the whiB3gene was shown to be dispensable for growth of Mycobacterium tuberculosisand Mycobacterium smegmatis\[[@B28],[@B29]\]. ### SOS and stress response The second functional category contains six DNA-binding transcriptional regulators that are apparently involved in the SOS and stress response of the cell. First of all, homologs of the LexA protein and the redox-responsive transcriptional regulator OxyR were identified in the four corynebacterial genomes. These proteins are well-known to control the SOS response and the oxidative stress response of the bacterial cell respectively \[[@B30],[@B31]\]. Moreover, the conserved HrcA, HspR and ClgR proteins were included in this functional category since they represent main components of the heat shock response of C. glutamicum\[[@B32],[@B33]\]. In general, the major function of the three responses is either the repair or the elimination of damaged macromolecules of the cell. A conserved member of the ArsR protein family was also included into this functional category of transcriptional regulators because its conserved genomic context suggested that it plays a role in the regulation of expression of the corynebacterial sufgene cluster. The respective genes of Escherichia colihave been implicated in the assembly of Fe-S clusters in proteins during oxidative stress conditions by encoding a specific sulphur transfer mechanism that limits the release of sulphide and thus the formation of highly damaging hydroxyl radicals \[[@B34]\]. ### Macroelement and metal homeostasis The third category of conserved DNA-binding transcriptional regulators is obviously involved in the regulation of macroelement and metal homeostasis of corynebacteria. This functional category of proteins is build up by the McbR, DtxR and FurB homologs and includes also two regulators of the MerR family which is a collection of metal-responsive transcriptional activators \[[@B35]\]. The McbR protein of C. glutamicumwas recently demonstrated to regulate a wide spectrum of genes comprising all aspects of transport and metabolism of the macroelement sulphur in this species as well as two transcriptional regulators that were classified into the ROK family \[[@B36],[@B37]\]. In this context it is noteworthy that AmtR, the master regulator of nitrogen metabolism of C. glutamicum\[[@B38]\], was not among the conserved set of DNA-binding transcriptional regulators since an orthologous gene is apparently absent in the genome of C. jeikeium. This striking difference in the equipment of corynebacterial genomes with global regulatory genes indicated that transcriptional regulation of sulphur metabolism is vitally important in corynebacteria in contrast to a conservation of transcriptional control mechanisms to provide an optimal supply of nitrogen for cellular processes. The DtxR repressor is known as global regulator of iron homeostasis in C. diphtheriae\[[@B39]\] but its regulatory network might also include genes whose expression is linked to protect the cell from oxidative stress \[[@B40]\]. The conserved FurB proteins were included in this functional category due to the observation that expression of the orthologous gene in Mycobacterium tuberculosiswas specifically induced by zinc \[[@B41]\]. ### Carbohydrate metabolism Our data suggested that six conserved genes are involved in the regulation of corynebacterial carbohydrate metabolism. The respective functional category of DNA-binding transcriptional regulators includes the AcnR repressor that controls the expression of the aconitase gene acn, thus representing an important control point of the tricarboxylic acid cycle of C. glutamicum\[[@B42]\]. Both RamB and GlxR apparently participate in the regulation of acetate metabolism and the glyoxylate bypass of C. glutamicum\[[@B43],[@B44]\]. The GlxR protein represses at least the expression of the aceBgene in the presence of cAMP, whereas RamB seems to be a more global regulator of gene expression. The observation that the glxRgene could not be mutated in C. glutamicumimplies that the common set of DNA-binding transcriptional regulators may include further members with essential regulatory functions. Three transcriptional regulators of the DeoR, IclR and LacI families were also grouped into this functional category due to the conserved physiological role of these homogenous regulatory protein families in the control of carbohydrate metabolism \[[@B20],[@B45],[@B46]\]. Conclusion ========== Based on genome analyses with different bioinformatic tools we have defined the individual set of DNA-binding transcriptional regulators in four sequenced corynebacterial genomes and deduced thereof the common repertoire of transcriptional regulators of these species. The data will provide valuable information on the corynebacterial biology in general and the ways these bacteria interact with different environments by modulating the expression of relevant genes. Only 28 DNA-binding transcriptional regulators have counterparts in the four corynebacteria and according to functional predictions most of these proteins are involved in the control of fundamental cellular processes encoded by the corynebacterial core genome. However, the majority of the genes belonging to the common set of transcriptional regulators has not been adequately studied yet, so further research and direct functional studies are necessary to determine the physiological role of these regulators and to precisely place them into the architecture of corynebacterial transcriptional regulatory networks. Further research is also necessary to determine the physiological function of species-specific and shared transcriptional regulators that might be involved either in the regulation of cellular processes relevant for biotechnological production or that might control the expression of genes involved for instance in virulence of pathogenic corynebacteria. With respect to a very recent study on the regulation of sulphur metabolism by the conserved transcriptional regulator McbR \[[@B37]\] we are confident that the combined use of classical genetic approaches along with post-genomic techniques, such as DNA microarrays and 2D gel electrophoresis, will provide the frame to decipher transcriptional regulatory networks and to build comprehensive regulatory models of corynebacterial cells. Methods ======= The general strategy to detect and classify DNA-binding transcriptional regulators of sequenced corynebacterial genomes was based on a combination of several bioinformatic tools. According to the work flow schematized in Figure [3](#F3){ref-type="fig"}, putative DNA-binding proteins were first searched for in the complete genome sequences of four corynebacterial species, comprising C. glutamicumATCC 13032 (GenBank accession number BX927147), C. efficiensYS-314 (BA000035), C. diphtheriaeNCTC 13129 (BX248353) and C. jeikeiumK411 (CR931997). These searches were performed by means of the genome assignment server SUPERFAMILY \[[@B21]\] that contains a library of hidden Markov models (HMMs) and the results of searches by these models against completely sequenced genomes. The HMMs of SUPERFAMILY are based on the sequences of domains collected in the Structural Classification of Proteins (SCOP) database \[[@B47]\] and are thus applicable for a structural classification of proteins. The search for corynebacterial DNA-binding proteins also included information deduced from the respective genome annotations deposited in public databases \[[@B12]-[@B14]\]. To identify among the DNA-binding proteins those potentially representing transcriptional regulators, 35 different HMM profiles of bacterial protein families with known function in transcriptional regulation of gene expression were downloaded from the Pfam database and used for searches against the predicted corynebacterial proteins by applying the HMMsearch module of the HMMER software package \[[@B48]\]. Verification of the results was performed by means of the BLAST algorithms \[[@B49]\], in particular by blastp and rpsblast along with the NCBI Conserved Domain Search program. Literature information was used to find additional coding sequences for DNA-binding transcriptional regulators in the corynebacterial genome sequences. During the final step of data analysis, the putative DNA-binding transcriptional regulators were grouped into regulatory protein families. This classification was performed by using HMM profiles of the Pfam database, CLUSTAL X alignments \[[@B50]\] of those proteins belonging to a distinct regulatory family and results of PROSITE pattern searches \[[@B45]\]. The HTH recognition tool designed by Dodd and Egan \[[@B24]\] was used to scan the putative DNA-binding transcriptional regulators for the presence and position of HTH motifs. By combining all these data, defined collections of putative DNA-binding transcriptional regulators were identified for each of the selected corynebacterial genomes (see Additional files [1](#S1){ref-type="supplementary-material"},[2](#S2){ref-type="supplementary-material"},[3](#S3){ref-type="supplementary-material"} and [4](#S4){ref-type="supplementary-material"}). To deduce thereof the common set of DNA-binding transcriptional regulators of the four sequenced corynebacterial species, comparative genomic analyses were performed. This approach included blastp searches with the identified regulatory proteins \[[@B49]\] against the predicted proteins of each genome sequence, using an evalue cut off smaller than 1 × e^-10^. To verify whether the orthologous regulatory genes are present in the same genomic context, neighbouring coding sequences were included into synteny analyses. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Work flow applied for the identification and classification of DNA-binding transcriptional regulators in corynebacterial genomes. The approach includes several methods and tools and consists of three consecutive steps indicated on the left. ::: ![](1471-2164-6-86-3) ::: Authors\' contributions ======================= IB carried out the bioinformatic analyses, comparative genomics studies and drafted the manuscript. KB performed all analyses related to the C. diphtheriaegenome sequence. JK participated in data evaluation and helped to draft the manuscript. AP participated in coordination and supervison and conceived of the design of the figures. AT conceived of the study and participated in its coordination. All authors read and approved the final manuscript. Supplementary Material ====================== ::: {.caption} ###### Additional File 1 classification and relevant molecular data of the DNA-binding transcriptional regulators identified in C. glutamicumATCC 13032. ::: ::: {.caption} ###### Click here for file ::: ::: {.caption} ###### Additional File 4 classification and relevant molecular data of the DNA-binding transcriptional regulators identified in C. jeikeiumK411. ::: ::: {.caption} ###### Click here for file ::: ::: {.caption} ###### Additional File 2 classification and relevant molecular data of the DNA-binding transcriptional regulators identified in C. efficiensYS-314. ::: ::: {.caption} ###### Click here for file ::: ::: {.caption} ###### Additional File 3 classification and relevant molecular data of the DNA-binding transcriptional regulators identified in C. diphtheriaeNCTC 13129. ::: ::: {.caption} ###### Click here for file ::: Acknowledgements ================ KB acknowledges the receipt of a grant from the International NRW Graduate School in Bioinformatics and Genome Research.	PubMed Central	2024-06-05T03:55:59.627407	2005-6-7	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180825/", "journal": "BMC Genomics. 2005 Jun 7; 6:86", "authors": [ { "first": "Iris", "last": "Brune" }, { "first": "Karina", "last": "Brinkrolf" }, { "first": "Jörn", "last": "Kalinowski" }, { "first": "Alfred", "last": "Pühler" }, { "first": "Andreas", "last": "Tauch" } ] }
PMC1180826	Background ========== The family Cichlidae is one of the most species-rich families of vertebrates \[[@B1]\]. More than 3,000 species of cichlid fishes are distributed from Central and South America, through Africa and Madagascar, to southern India \[[@B2]\]. Although cichlids are diverse and dominant components of the freshwater fauna of both the Old and New Worlds, it is in the lakes of East Africa that they have undergone the spectacular adaptive radiations for which the group is best known \[[@B3]\]. Cichlids are an emerging model system for studying a broad range of questions at the interface of organismal biology and genomics \[[@B4]\]. Tilapias (Oreochromisspp.) are cichlid fishes which have become one of the most important species in global aquaculture. Native to Africa, several species of tilapia have been introduced to tropical areas of Asia and the Americas to increase supplies of animal protein. World aquaculture production of tilapia is second only to carp, and now exceeds 1.5 million tons per year \[[@B5]\]. The Nile tilapia (Oreochromis niloticus) genome contains 1.06 gigabase pairs distributed over 22 chromosome pairs \[[@B6]\]. Several partial genetic linkage maps of tilapia have been produced \[[@B7]-[@B9]\]. The latest and most complete map orders 550 loci in 24 linkage groups spanning a total of 1311 cM \[[@B10]\]. Here we present a physical map of the tilapia genome based on restriction fingerprints of more than 35,000 large-insert bacterial artificial chromosome (BAC) clones. This physical map will help speed positional cloning in tilapia, and will facilitate the long-range assembly of a tilapia genome sequence. Results and discussion ====================== BAC fingerprinting ------------------ We processed 40,704 clones from libraries 3 and 4, and obtained valid fingerprints from a total of 35,245 clones (87% success; Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}). Library 3 has an average insert size of 145 kb, and produced an average of 53.9 valid bands per clone. Library 4 has an average insert size of 194 kb, and produced an average of 69.8 bands per clone. Figure [1](#F1){ref-type="fig"} shows the regression of fingerprint band number on clone insert size. Together, the fingerprinted clones represent an estimated 5.6-fold coverage of the tilapia genome. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### BAC libraries fingerprinted for the tilapia physical map. Construction of these BAC libraries is described in Katagiri et al.\[16\]. Copies of the libraries are available as plates and filters from <http://hcgs.unh.edu/BAC/>. ::: Library Cloning site Vector Mean insert size (kb) No. of clones fingerprinted Valid bands per clone Genome coverage ------------- ------------------ ------------ --------------------------- --------------------------------- --------------------------- --------------------- HCGS-03TI HindIII pBAC-lac 145 18,700 53.9 2.56 HCGS-04TI HindIII pBAC-lac 194 16,545 69.8 3.02 Total 182 35,245 61.4 5.58 ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Summary of the tilapia physical map ::: ----------------------------------------------- -------------- Number of clones processed 40,704 T3 library 20,736 T4 library 19,968 Number of clones used for contig assembly 35,245 T3 library 18,700 T4 library 16,545 Average success rate 87% Number of singletons 2,647 Number of contigs 3,621 2--4 clones 1,646 5--5 clones 973 10--25 clones 771 26--50 clones 188 51--100 clones 34 101--200 clones 8 \>200 clones 1 Physical length of the contigs 1.752 Mb ----------------------------------------------- -------------- ::: ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Relationship between number of fingerprint bands and clone insert size.Clones from the T3 library shown as circles, T4 library shown as triangles. The line shows the regression: number of bands = 22.37 + 0.238 \* insert size (kb). ::: ![](1471-2164-6-89-1) ::: Contig assembly --------------- Contigs were assembled from the fingerprint data using the computer program FPC version 6.0 \[[@B11],[@B12]\]. We estimated the sizing accuracy of the capillary sequencer by analyzing the size of the vector band in 200 clones. The mean size was 246.20, with a standard deviation of ± 0.253 bp. We therefore multiplied all fragments sizes by 10, and used a fixed tolerance of 5, corresponding to 0.5 bp, in the FPC analysis. Using a cutoff stringency of 1e-08, the number of contigs reached a plateau of approximately 3,500 after 20,000 clones had been fingerprinted. The number of contigs reached a maximum of 3,748 contigs at 30,000 clones, and dropped to 3,621 contigs in the final analysis of 35,234 clones (Fig [2](#F2){ref-type="fig"}). A total of 32,598 clones (92.5%) were placed in contigs. Only 2,647 clones remained as singletons (Table [2](#T2){ref-type="table"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Coalescence of contigs during the fingerprinting process.The number of contigs rises to a maximum of 3,748 contigs after fingerprinting 30,000 clones. With additional fingerprinting, it appears that the contigs are beginning to coalesce. All analyses performed with a tolerance of 5 and cutoff threshold of 1e-08. ::: ![](1471-2164-6-89-2) ::: The contigs contain an average of 9.0 clones each, and had an average estimated length of 389.9 kb. The assembled contigs have an estimated length of 1.752Gb, or about 1.65x the genome length. Half of the total assembly length is in the largest 1,054 contigs. The top half of the contigs (1,630 clones) contained 69% of the total length of the assembly. Contig reliability ------------------ We used several different approaches to assess contig reliability. The first was to determine the stability of contigs at different cutoff values. Increasing the stringency of assembly, from 1e-08 to 1e-09, increased the number of contigs from 3,621 to 4,008. This means that approximately 200 contigs were split at the higher stringency, which is less than 5% of initial total. FPC identified a total of 3,127 questionable clones (Q\'s) in the assembly, an average of 0.86 Q\'s per contig. However, the distribution was strongly skewed from Poisson expectations. 2,891 contigs (92.5%) had no Q\'s called. Most of the questionable clones were in a few large contigs (Table [3](#T3){ref-type="table"}). More than half of the Q\'s were in the 58 contigs with 10 or more questionable clones. Since the number of Q\'s was strongly correlated with the number of clones in the contig (Fig [3](#F3){ref-type="fig"}), we suspect this represents improper assembly of clones containing repetitive sequences. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Distribution of FPC questionable clones (Q\'s). Poisson expectations calculated from the average of 0.86 Q\'s per contig. ::: \#Q\'s \# Contigs Poisson ------------ ---------------- ------------- 0 2891 1526 1 328 1318 2 133 569 3 93 163 4 42 35 5 24 6 6 14 1 7 22 0 8 10 0 9 6 0 10 5 0 11 8 0 12 4 0 13 2 0 14 1 0 15 3 0 16 1 0 17 5 0 18 3 0 19 1 0 20+ 25 0 ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Q scores for contigs of different size.The number of questionable clones identified by FPC rises with the size of the contig. Very large contigs tend to have a disproportionate number of Q\'s, suggesting improper assembly of repetitive sequences. The line represents a least squares fit of y = 0.252x (r^2^= 0.54). ::: ![](1471-2164-6-89-3) ::: Cichlid fishes have an expanded set of opsin genes relative to tetrapods. Changes in the expression of these genes are responsible for differences in visual sensitivity among species \[[@B13]\]. In order to identify the regulatory regions for these genes, we isolated BAC contigs containing opsin genes. PCR screening of pooled BAC DNAs identified clones containing the SWS1, RH2 and LWS genes. The FPC database was then used to identify overlapping BACs at a tolerance of 5 and cutoff threshold of 1e-08. The SWS1 contig contained six clones, all of which were positive for the SWS gene by PCR. The RH2 contig contained 18 clones, 11 of which were positive for the RH2 gene. Probes derived from end sequencing of these BACs were used to verify that the remaining 7 clones were members of a genuine contig. The LWS contig contained 10 clones, 5 of which contained the LWS gene. Probes developed from the end sequences of these clones verified that four of the remaining five clones were members of an overlapping contig. The fifth clone should not have been included in this contig. The genes in this contig are homologous to the SWS2 and LWS genes located on scaffold 5 of Fugu assembly version 3.0. Two of the six BAC end sequences derived from the tilapia contig had BLAST hits to Fugu scaffold 5, providing strong evidence for homology of this contig to a 100 kb region of the Fugu genome. Finally, in the course of positional cloning a mutation for red body color in tilapia, we identified a BAC containing the tyrosinase-related protein 1 (trp1) gene. Using a reduced FPC stringency (tol = 5, 1e-06) this BAC was near one end of a contig of 70 clones which is estimated to span 1.97 Mb. An RFLP was identified from a clone at the opposite end of this contig, and was mapped 3cM from trp1in a large F2 progeny (Fig [4](#F4){ref-type="fig"}). This result emphasizes the utility of the fingerprint database, even at reduced stringencies of assembly. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### *Contig containing the trp1gene.PCR screening identified trp1sequences in BAC clone b03TI073AG01, near one end of this contig. A RFLP was developed by shotgun sequencing of clone b04TI008AG07, near the other end of the contig. Genetic mapping shows these markers are about 3 cM apart, confirming the utility of this contig spanning approximately 2 Mb. ::: ![](1471-2164-6-89-4) ::: Conclusion ========== East African cichlid fishes, including the tilapias as well as the closely related and highly diverse haplochromine cichlids, constitute more than 5% of vertebrate species. An international consortium has come together to develop genomic tools for studying these fishes <http://hcgs.unh.edu/cichlid/>. Resources already developed include a genetic map with more than 550 microsatellite markers \[[@B10]\], and a collection of more than 50,000 ESTs \[[@B14],[@B15]\]. The physical map described in this paper is a further step in building the infrastructure to support complete sequencing of the cichlid genome. Fingerprinting of additional clones from these libraries would undoubtedly allow further coalescence of contigs, but it is not clear how cost-effective this approach would be. The current set of 3500 contigs is a manageable number for anchoring to physical and comparative maps. A logical next step in this research would be analysis of the gene content of these contigs to relate the contigs to the sequences of other fish genomes. In the meantime, the physical map will facilitate the positional cloning of genes controlling economically important traits in tilapia, as well as the genes underlying the spectacular adaptive radiation of cichlids in the lakes of East Africa. Methods ======= Source BAC libraries -------------------- Four BAC libraries have been constructed for Oreochromis niloticus*\[[@B16]\]. All four libraries were constructed from the sperm of a single male (\#00-0135-EA1B) from a strain originating from Lake Manzallah, Egypt and maintained at the University of Stirling, UK. We fingerprinted clones from the two libraries with the largest average insert size (Table [1](#T1){ref-type="table"}). Insert sizes of 200 BACs from each library were determined by NotI digestion and comparison to a lambda PFG ladder (New England Biolabs, Beverly MA). Plates and filters of these clones are available on a cost-recovery basis from the Hubbard Center for Genome Studies <http://hcgs.unh.edu/>. BAC fingerprinting ------------------ BAC DNA was isolated using a modified alkaline lysis method \[[@B17]\]. Briefly, BAC clones were inoculated into 96-deep well plates. Each well contained 1.5 ml of 1x LB media with chloramphenicol at a concentration of 12.5 μg/ml. The plates were covered with Qiagen Airpore tape sheets (Cat\# 19571) and incubated at 37°C for 20--21 hours on a Bellco mini-orbital shaker. Restriction fingerprints were obtained following the approach of Ding et al. \[[@B18]\]. The DNA was double-digested with HindIII and HaeIII and the HindIII ends labelled with fluorescently labelled ddGTP in a fill-in reaction using the reagents from a Beckman DTCS sequencing kit. The fragments were sized on Beckman CEQ2000 capillary DNA sequencers using the CEQ-600 molecular weight standard (Beckman Coulter, Fullerton CA). BAC contig assembly ------------------- Every chromatogram was manually reviewed to confirm the peaks identified by the Beckman CEQ8000 software. Only the bands between 80 to 620 bp were used for contig assembly. The chromatograms and associated peak values were then stored in a MySQL database for further analysis. Contig assembly was done using the computer program FPC (vers. 6.0; <http://www.agcol.arizona.edu/software/fpc/faq.html>) \[[@B12]\]. The resulting contigs are displayed using a new www-based viewer which mimics the WebFPC interface <http://hcgs.unh.edu/fpc/image.php>. This viewer is written in PHP and generates html in response to queries of the database. DNA markers and BAC library screening ------------------------------------- To facilitate screening of the BAC libraries by PCR we constructed pools of the bacterial cultures. The pools were constructed from 252 96-well plates (144 from library T3 and 108 from library T4). This is equivalent to 2x coverage, or 2 Gb equivalents, from each library. We collected row and column pools from each plate using a Beckman Biomek2000 robotic pipettor. The row pools from each plate were pooled by hand to produce 252 plate pools. The plate pools were grouped into one of 10 arrays of either 4 × 6 or 5 × 6 plates. We then constructed pools from the rows and columns in each of these arrays. Finally, we constructed 10 superpools corresponding to the groups of plates in each array. This allowed us to identify positive clones by PCR in a sequence of 3 experiments. We first attempted amplification from each of the 10 superpools. We then analyzed the row and column plate pools for each positive superpool to identify the plate. Finally, we analyzed the 8 row and 12 column pools from each positive plate to identify the clone. Authors\' contributions ======================= TK developed the DNA extraction and fingerprinting protocols. CK, ET, JTD and CW obtained the fingerprints and imposed quality controls on data entering the analysis pipeline. JES developed the databases and software interfaces for displaying the FPC results on the www. KLC and AEH tested contig reliability by sequencing and probe hybridization. TDK conceived the project, supervised its execution and wrote the manuscript. Acknowledgements ================ We thank Drs. Takashi Aoki and Ikuo Hirono (Tokyo University of Marine Science and Technology) and Drs. Noboyushi Shimizu and Syuichi Asakawa (Keio University School of Medicine) for facilitating access to the tilapia BAC libraries constructed in their laboratories. Supported by the USDA/CSREES National Research Initiative Project \#00-03504. This is Scientific Contribution Number 2261 from the New Hampshire Agricultural Experiment Station.	PubMed Central	2024-06-05T03:55:59.630909	2005-6-9	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180826/", "journal": "BMC Genomics. 2005 Jun 9; 6:89", "authors": [ { "first": "Takayuki", "last": "Katagiri" }, { "first": "Celeste", "last": "Kidd" }, { "first": "Elizabeth", "last": "Tomasino" }, { "first": "Jesse T", "last": "Davis" }, { "first": "Cassandra", "last": "Wishon" }, { "first": "Justin E", "last": "Stern" }, { "first": "Karen L", "last": "Carleton" }, { "first": "Aimee E", "last": "Howe" }, { "first": "Thomas D", "last": "Kocher" } ] }
PMC1180827	Background ========== Transcription factor activities (TFAs) rather than levels of transcription factor expression mediate transcriptional regulations. Various post-transcriptional and post-translational modifications abolish significant correlations between TFAs and the level of transcription factor expression. Therefore, a strategy to determine the physiological functions and interactions of transcription factors based on the analysis of TFA profiles would be more fundamentally sound than one based on transcript levels per se. However, owing to post-translational modifications, TFAs are difficult to measure experimentally and therefore, many analyses infer TFAs by computational analysis of target gene expression levels either from single regulatory factors alone or in combination \[[@B1]-[@B10]\]. Among these, Network Component Analysis (NCA) \[[@B3]\] and generalized NCA (gNCA) \[[@B2]\] provide a robust framework for deducing TFAs based on DNA microarray data, promoter connectivity, and genetic regulatory constraints imposed by regulatory knock-out experiments. Using TFA profiles deduced by NCA and gNCA, we are now in a position to assign transcription factor functions and reconstruct functional interactions between transcription factors. The goal of this work is to demonstrate how we analyze TFA profiles to determine physiologically relevant characteristics of transcription factors. Specifically, by analyzing TFA profiles of the S. cerevisiaecell cycle we identify cell cycle regulators and putative interaction partners of two forkhead transcription factors (Fkh1 and Fkh2) that are responsible for expression of a gene cluster within the M/G1 interval and involved in mitotic exit \[[@B11]\]. Once TFA profiles are determined, the transcription factor regulators and interactions are deduced based on the following hypotheses: 1) Transcription factors with similar activity patterns function together; 2) Transcription factors involved in cell cycle regulation exhibit oscillatory activity patterns, 3) Activities of transcription factors which functionally interact with each other will be disturbed if one of the interacting partners is deleted. These hypotheses are in spirit similar to the traditional analysis of gene expression. In contrast to traditional analysis of gene expression, our analyses are constructed and applied directly to TFA profiles. Results ======= Overall strategy ---------------- A flow diagram demonstrating the relationships between each component of the methodology is presented in Figure [1](#F1){ref-type="fig"}. Given the gene expression data and connectivity network between transcription factors and gene expression, we first deduced TFAs from gene expression by using gNCA. The deduced TFAs were then analyzed by two complementary methods. First, we used cluster analysis \[[@B12]\] to group TFAs that behave in a similar manner. Second, a statistical analysis \[[@B4]\] was used to determine if the signal from each of the TFAs was periodic. The results from these two analyses were integrated to deduce putative cell cycle regulators. We hypothesized that putative cell cycle regulators should exhibit periodic activity profiles and cluster closely with known cell cycle regulators. Then TFA profiles from wild type and the fkh1 fkh2mutant strain under the same experimental conditions were statistically analyzed to deduce transcription factors whose profiles were significantly perturbed, thus identifying putative functional interaction partners of the fork-head transcription factors. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Flowchart summarizing our methodology. Flowchart of the method used to determine transcription factor functions and interactions. ::: ![](1471-2164-6-90-1) ::: Application of gNCA to the combined wild-type and mutant data set ----------------------------------------------------------------- We used gNCA to analyze both the wild-type \[[@B13]\] and the fkh1 fkh2mutant data \[[@B11]\]. The wild-type data and the TF-knockout data were organized in different columns of the Ematrix (expression data matrix where E~ij~corresponds to the log of gene expression ratio of gene ievaluated at experiment j, see Methods). The elements in P(TFA matrix where P~ij~corresponds to the log of relative transcription factor activity of TF iat experiment j, see Methods) corresponding to the specific knockout strain were kept at zero, in addition to the zero constraints placed in A(control strength matrix where A~ij~denotes the control strength of transcription factor jon gene i, see Methods) according to the transcription factor-gene connectivity information. The first step in gNCA was to select sub-networks based on available transcription factor-gene connectivity \[[@B14]\]. Each sub-network was constructed to satisfy the gNCA criteria. gNCA requires that the number of transcription factors (L) in each sub-network be less than that of data points (M) in the data matrix. In the combined cell cycle microarray data set, M= 69, and thus the number of transcription factor in gNCA in each analysis cannot exceed 69. Thus, subsets of the 104 transcription factors included in the genome-wide location analysis \[[@B14]\] were selected to form sub-networks for gNCA. Overlapping random sub-networks were generated and screened using the gNCA criteria. Among them, four that satisfy the gNCA criteria were selected in order to show that using multiple sub-networks makes it possible to determine TFAs more than the number of data points (M = 69). Each sub-network contained 40 transcription factors, but together, a total of 74 transcription factors were analyzed. The transcription factors involved in each of the 4 sub-networks are shown in Figure [2](#F2){ref-type="fig"}. Only 16 transcription factors among 74 transcription factors were fully overlapped among all 4 sub-networks. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Venn diagram of overlapped transcription factors among the four sub-networks analyzed. Each sub-network contains 40 transcription factors, but together, 74 Transcription factors can be used for determining TFAs. Sub-network 1, 2, 3, and 4 contains 1110, 847, 1015, and 793 genes, respectively. Total 1529 genes were finally selected for generating 4 multiple sub-networks from 1818 genes with full data points in the combined dataset. ::: ![](1471-2164-6-90-2) ::: After applying gNCA independently to the 4 sub-networks, 74 TFAs were determined. Figure [3](#F3){ref-type="fig"} shows the TFAs of the 11 known cell cycle factors from 4 different sub-networks. Qualitatively, these TFAs appear to show expected oscillatory behavior for 1, 2, and 3 cycles according to the method of synchronization (elutriation, α-factor arrest, and cdc15mutation). The dynamics of overlapping TFA profiles between the four different sub-networks are very similar. This indicates that the deduced TFAs were insensitive to the sub-network selection. In addition, combining the fkh1 fkh2mutant data with the wild-type data yields similar TFAs predicted from wild-type data alone \[[@B3]\], suggesting the consistency between the two data sets. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### *Comparison of TFA profiles of 11 major yeast cell cycle related transcription factors between wild type and fkh1 fkh2mutant. The first 3 columns corresponds to data deduced from yeast cultures synchronized by elutriation, α-factor arrest, and arrest of a cdc15temperature-sensitive mutant which can give one cell cycle, two cell cycles, and three cell cycles for given experimental measurements, respectively. The last column corresponds to fkh1 fkh2double knock-out mutant synchronized with α-factor arrest. Different stages in the cell cycle are indicated by the color code. Different colors in TFA profiles for each transcription factor represent TFAs from different sub-networks. S1, S2, S3 and S4 represent sub-networks 1, 2, 3, and 4, respectively. ::: ![](1471-2164-6-90-3) ::: It is recognized that most transcription factors are regulated in post-transcriptional processes. Thus, their activity may deviate significantly from the expression level as noted previously \[[@B3]\]. For 27 transcription factors with both gene expression level and TFA available, the correlation coefficients ranged from - 0.5 to 0.6. The wide range difference in the correlation coefficients confirms that TFAs and expression levels exhibit very different dynamic profiles and therefore TFAs cannot be substituted by expression levels in analysis. Sensitivity Analysis -------------------- To assess the effect of connectivity errors on our results, we altered up to 10% of the connectivity graph for every subnetwork, by randomly deleting and inserting connections between transcription factors and genes. This effectively simulates the presence of false negatives and false positives in ChIP-chip assays. Subsequently, we ran gNCA on our randomly perturbed graphs and compared the deduced TFAs to our reported results via the correlation coefficient. The procedure was performed 100 times and the TFAs deduced from each run were compared with the TFAs computed with unaltered connectivity using Pearson correlation coefficients (Table [1](#T1){ref-type="table"}). Transcription factors with high correlation coefficients suggest robustness of the results against connectivity errors. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Sensitivity of TFA profiles to errors in connectivity from ChIP-chip assay. Connectivity errors were simulated by randomly inserting and deleting connections in the connectivity graph generated from the ChIP-chip assay. gNCA was performed on the perturbed connectivity, and the resultant TFAs were compared to those from un-perturbed connectivity via the correlation coefficient. This procedure was performed on each of the 4 subnetworks 100 times. The average correlation coefficients for transcription factors in the 4 subnetworks are presented here. A low correlation coefficient (\<0.5) suggests sensitivity to network connectivity error. ::: Subnetwork 1* Mean Correlation Coefficient Subnetwork 2 Mean Correlation Coefficient Subnetwork 3 Mean Correlation Coefficient Subnetwork 4 Mean Correlation Coefficient ------------------ ---------------------------------- ------------------ ---------------------------------- ------------------ ---------------------------------- ------------------ ---------------------------------- ABF1 0.941 ABF1 0.951 ABF1 0.94 ACE2 0.912 ACE2 0.953 ACE2 0.857 ACE2 0.897 ARO80 0.657 ARO80 0.645 ARG81 0.286 CHA4 0.362 CIN5 0.601 CAD1 0.511 CAD1 0.659 CIN5 0.59 DIG1 0.718 CBF1 0.401 CHA4 0.448 CRZ1 0.313 FHL1 0.992 CIN5 0.651 DAL81 0.581 FHL1 0.993 FKH2 0.669 DIG1 0.686 FHL1 0.99 FKH2 0.657 GAT3 0.493 FHL1 0.992 FKH2 0.635 GAL4 0.492 GCN4 0.601 FKH1 0.614 GAT3 0.457 GAT3 0.511 GCR1 0.321 FKH2 0.535 GCR2 0.425 GCR1 0.306 GLN3 0.402 GAT3 0.446 GRF10(Pho2) 0.565 GCR2 0.446 GTS1 0.227 HAP4 0.706 HIR1 0.732 HAP4 0.769 HAP2 0.375 HIR1 0.472 HIR2 0.357 HIR1 0.465 HAP3 0.47 IXR1 0.511 IME4 0.357 HSF1 0.981 HAP4 0.856 MBP1 0.946 INO4 0.419 IME4 0.442 HIR1 0.32 MCM1 0.558 LEU3 0.575 MBP1 0.979 IXR1 0.475 MET31 0.436 MAC1 0.723 MCM1 0.572 LEU3 0.518 MET4 0.381 MBP1 0.985 MET31 0.435 MBP1 0.951 MIG1 0.336 MCM1 0.701 MIG1 0.354 MCM1 0.705 MSN4 0.444 MET31 0.507 NDD1 0.965 MIG1 0.33 MTH1 0.463 MOT3 0.413 NRG1 0.421 MSS11 0.404 NDD1 0.958 NDD1 0.971 PDR1 0.366 NDD1 0.962 NRG1 0.495 PHD1 0.541 PHD1 0.583 NRG1 0.563 PDR1 0.334 PHO4 0.599 RAP1 0.917 PDR1 0.372 PHD1 0.569 RAP1 0.856 RCS1 0.48 PHD1 0.64 RAP1 0.917 RME1 0.407 REB1 0.576 RAP1 0.897 REB1 0.643 RTG1 0.413 RME1 0.415 RCS1 0.424 RFX1 0.493 RTS2 0.302 RPH1 0.294 RLM1 0.553 RLM1 0.519 SFP1 0.564 SFP1 0.582 RME1 0.389 RME1 0.367 SKN7 0.536 SKN7 0.669 SKN7 0.609 SKN7 0.545 STE12 0.518 SMP1 0.619 SMP1 0.642 SMP1 0.614 STP2 0.322 SWI4 0.553 STB1 0.576 STE12 0.79 SWI4 0.592 SWI5 0.987 STE12 0.738 SWI4 0.747 SWI5 0.963 SWI6 0.61 SUM1 0.823 SWI5 0.991 SWI6 0.648 UGA3 0.392 SWI4 0.843 SWI6 0.606 UGA3 0.383 YAP5 0.512 SWI5 0.991 YAP5 0.863 YAP1 0.565 YAP6 0.404 SWI6 0.589 YAP6 0.56 YAP5 0.657 YFL044C 0.578 YAP5 0.572 YFL044C 0.583 YFL044C 0.546 ZAP1 0.442 YAP6 0.595 YJL206C 0.396 ZAP1 0.527 ZMS1 0.295 YJL206C 0.455 ::: Determination of cell cycle-dependent regulators based on TFA dynamics ---------------------------------------------------------------------- We hypothesized that transcription factors with similar activity patterns are involved in related processes. Thus, clustering of TFAs would allow identification of transcription factors related to cell cycle regulation. After hierarchical clustering (Figure [4](#F4){ref-type="fig"}), 11 transcriptional factors (Dal81, Dig1, Gat3, Hap2, Hir2, Mss11, Pdr1, Rlm1, Rph1, Yap5, and Yap6) were found to cluster closely with the 11 known cell cycle regulators. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Hierarchical clustering of TFAs of all 74 Transcription factors. Absolute correlation coefficient as a similarity measure and the average linkage method were used for clustering. Green, red, and black color represents negative log TFA ratios, positive log TFA ratios, and 0, respectively. The color intensity increases as the magnitude of each TFA value increases. The TFs denoted in blue are those TFs known to be involved in cell cycle. ::: ![](1471-2164-6-90-4) ::: In addition to cluster analysis, we performed a statistical test to identify the set of TFAs with a periodic profile. Figure [5A](#F5){ref-type="fig"} shows the power spectra of the 11 known cell cycle regulator. A power spectrum is a representation of a signal in the frequency domain. A dominant peak in the power spectrum corresponds to the fact that the underlying process has a principal oscillation frequency. Most of the power spectra to identify periodic patterns exhibit a single strong peak at low frequency. This suggests that the dynamic profiles appear to be periodic. To classify whether a signal is periodic or not, we employed a statistical criterion \[[@B4]\] of rejecting the null hypothesis of purely random process. The result suggests that 9 out of 11 known cell cycle regulators exhibit periodic behavior. This confirms that the activity profiles of cell cycle regulators are periodic. We then applied the same periodicity test to the rest of 63 transcription factors and found that about 16 of the deduced TFAs from elutriation and α-factor arrest were statistically periodic and 44 of the deduced TFAs from cdc15 were statistically periodic. Combining the results of the periodicity test and cluster analysis, we found that 5 (Dal81, Hap2, Hir2, Mss11, and Rlm1) out of the 11 transcription factors closely clustered with the known cell cycle factors were statistically periodic and were regarded as putative cell cycle-related regulators. ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Power spectra for selected TFAs from the 4 different sub-networks. (A) 11 known cell regulators. (B) The top 5 TFAs that exhibits periodic function. In both sub-figures, solid blue lines are the data deduced from yeast cultures synchronized by arrest of a cdc15temperature-sensitive mutant, dash green lines are data deduced from yeast cultures synchronized by α-factor arrest, and dotted red lines are data collected from yeast cultures synchronized by elutriation. ::: ![](1471-2164-6-90-5) ::: The power spectra of these 5 putative cell cycle regulators are shown in Figure [5B](#F5){ref-type="fig"}, and the TFA profiles of these regulators are illustrated in Figure [6](#F6){ref-type="fig"}. The clustered TFA pattern of all cell cycle-related transcription factors (Figure [8](#F8){ref-type="fig"}) shows that the peak activities of these transcription factors gradually change from one phase to another through the cell cycle. Among the 5 putative cell cycle-related regulators, Hir2 is a regulator of histones \[[@B1],[@B15]\], and it is therefore reasonable to expect it to have cell cycle related functions. It is not clear how the other 4 transcription factors are related to cell cycle regulation. Lee et al. \[[@B14]\] categorized Dal81 and Mss11 as metabolism regulators, Hir2 as a DNA/RNA/Protein biosynthesis regulator, and Rlm1 as environmental response regulators. These results suggested that transcription factors related to many other cellular processes may be involved in or dependent on cell cycle regulation to coordinate cellular processes \[[@B14]\]. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### TFA profiles of 5 putative cell cycle regulators. Details of the figure legends are the same as Figure 3. ::: ![](1471-2164-6-90-6) ::: ::: {#F8 .fig} Figure 8 ::: {.caption} ###### Pearson correlation coefficients and the deviation coefficient for 11 known cell cycle factors (empty symbols) and other remaining 63 TFs (solid symbols) under the release from α-factor arrest. The oval encloses TFAs with both low deviation coefficient. ::: ![](1471-2164-6-90-8) ::: Functional interaction of FKH1 FKH2 with other transcription factors -------------------------------------------------------------------- Comparing the TFAs derived from the wild-type strain and the fkh1 fkh2mutant under the same experimental conditions allows the determination of functional interactions between Fkh1 Fkh2 with other transcription factors. If such interaction exists, the activities of the interaction partners will change in the fkh1 fkh2mutant. In general the fkh1 fkh2mutant showed reduced oscillation amplitude as compared to the wild type (Figure [3](#F3){ref-type="fig"}). In particular, Ace2, Swi5, and Ndd1 showed significantly reduced oscillation in the fkh1 fkh2mutant, suggesting potential interactions between these transcription factors and the two fork-head transcription factors. Indeed, it was found that Fkh2 forms a complex with Mcm1 and Ndd1 \[[@B16],[@B17]\] which controls G2/M genes and also binds the promoters of Ace2 and Swi5 to activate genes at the following M/G1 phase. The reduced amplitude in TFAs of Ace2 and Swi5 in the fkh1 fkh2mutant support that there is a cascade interaction between the two forkhead transcription factors and Ace2 and Swi5 \[[@B16]\]. TFA profiles of the wild-type strain (Figure [3](#F3){ref-type="fig"}) show that Ace2 and Swi5 were most active around M/G1 phase while Fkh1, Fkh2, Mcm1, and Ndd1 were most active at G2/M, G2, M/G1, and G2/M phase, respectively. These results are well consistent with previous results that Mcm1, Ace2, and Swi5 are M/G1 transcriptional regulators while the complex of Mcm1, Fkh2 and Ndd1 is a G2/M activator. Therefore, this result shows that TFA in a regulator knockout strain can be used to determine functional interactions. By utilizing the Pearson correlation coefficient and deviation coefficient defined in the Methods section, Figure [8](#F8){ref-type="fig"} shows that most transcription factors were aggregated in one cluster. Outside of this cluster are transcription factors which potentially were affected by the fkh1 fkh2mutation and thus are functional interaction partners of the forkhead transcription factors. These interaction partners include Ace2, Ndd1, and Swi5, which are known to interact with Fkh2, and Hap2, Rts2, Cha4, and Fhl1, which were not known to interact with the forkhead transcription factors. In particular, Hap2 was also identified in the above analysis as a putative cell cycle-related regulator, supporting their functional interaction with the forkhead transcription factors. Cha4 and Hap2 were classified as metabolism-related factors, Fhl1 DNA/RNA/Protein synthesis, and Rts2 cell cycle and data processing \[[@B14]\]. Their modes of interaction with the forkhead transcription factors remain unknown. Discussion ========== Transcriptional regulators are commonly modified at the post-transcriptional level, and consequently their biological activities do not correlate significantly with expression levels. Previous work infer TFAs from expression levels of genes regulated by single factors or in combination \[[@B2],[@B5]-[@B8],[@B10]\]. In general, the major difference between previous work and that of NCA is that the former require an explicit quantification of the control strengths (the A matrix in NCA) a priori. Bussemaker et al. \[[@B7]\] defined the control strength as the motif copy number in corresponding promoters and found that there was no statistical benefit to model expression with more than single factors and thus deduced single TFAs. Wang et al. \[[@B6]\] considered an expression-weighted motif to find potential target genes of single transcription factors. Gao et al. \[[@B18]\] used ChIP-chip log occupancy ratios as a surrogate for transcription factor binding affinity. In contrast, NCA explicitly models combinatorial regulation of gene expression, and allows both the control strengths and the TFAs to be deduced simultaneously with given network connectivity. In this approach, the lack of connectivity in specific pairs of transcription factor and promoter is used to provide constraints for data decomposition in order to obtain unique solutions when specific criteria are satisfied \[[@B3]\]. gNCA expands these capabilities by allowing incorporation of constraints onto the deduced TFAs, such as transcription factor knockout experiments, which offer a rich source of data and biochemical information. Development of these methodologies significantly expands the capabilities of transcriptional regulation analysis. With gNCA, we analyzed the combined wild-type and fkh1 fkh2mutant data set and showed that gNCA can be used to identify TFAs which are consistent with cell physiology, transcription factors with potential cell cycle dependent roles, as well as interactions between transcription factors. On the basis that transcription factors exhibiting similar activity patterns function together, we identified 11 transcription factors that clustered closely with the known cell cycle regulators. We performed a periodicity test to determine the TFA profiles that exhibit periodic behavior, and by combining the sets of transcription factors collected by these two methods, we identified 5 putative cell cycle-related regulators: Dal81, Hap2, Hir2, Mss11, and Rlm1. These transcription factors may participate in functions driven by cell cycles, or may regulate cell cycle directly or indirectly. Our comparison between the wild-type TFAs and the mutant TFAs confirmed that the forkhead transcription factors interact with Ace2, Ndd1, and Swi5. This result is consistent with previous reports \[[@B16],[@B17]\]. Using this approach, we identified 4 additional transcription factors that may functionally interact with Fkh1 Fkh2 directly or indirectly: Hap2, Rts2, Cha4, and Fhl1. Most of these transcription factors are not known to be related to cell cycle, suggesting that cell cycle regulation interacts with other physiological functions. It is worth noting that our analysis can be sensitive to errors in the connectivity graph. Through a sensitivity analysis we determined that all the known cell cycle regulators and all the known forkhead interaction partners have TFAs that exhibit low sensitivity to the connectivity network when using 0.5 as a correlation coefficient threshold. These results suggest that our analysis is robust to errors in connectivity. With the same sensitivity criterion, 2 (Dal81 and Rlm1) of the 5 putative cell cycle-related regulators and 1 (Fhl1) of the 4 putative forkhead interaction partners were determined to be robust to connectivity errors. The lack of sensitivity to error increases confidence in these predictions. A total of 1529 (out of 6200) genes and 74 (out of 104) transcription factors were analyzed from 69 microarray experiments. Both limited connectivity information from ChIP-chip and missing data points in DNA microarray expression attribute to the limited number of genes and TFs that can be studied in the current investigation. On the other hand, the results suggest that our analysis on this limited set of genes and TFs appears to be sufficient: our analysis strategy recovers nearly 90% of the known cell cycle regulators. On the basis of the TFA profiles, the time series of the key cell cycle regulators are summarized in Figure [9](#F9){ref-type="fig"}. A dominant feature in this map is the overlapping TFAs among known cell cycle regulators. This is common among some transcription factor complexes, including SBF (Swi4/Swi6) and MBF (Mbp1/Swi6), as well as transcription factors known to regulate the same phase in the cell cycle including Ace2, Swi5 and Mcm1 (Figure [9A](#F9){ref-type="fig"}). In addition, overlapping TFAs were also observed among different phases of cell cycle: transcription factors during one stage regulate transcription factors that function in the next stage. It is well known that serial regulation among transcription factors forms a connected regulatory network \[[@B17]\]. From the TFA profiles, we also observe the intrinsic property of the connected regulatory network among cell cycle factors in cell cycle regulation (Figure [9A](#F9){ref-type="fig"}). ::: {#F9 .fig} Figure 9 ::: {.caption} ###### Phase diagrams of TFAs. (A) 11 known Transcription factors and (B) 5 deduced cell cycle-dependent factors. ::: ![](1471-2164-6-90-9) ::: Among the putative cell cycle-related transcription factors (Figure [6](#F6){ref-type="fig"}), Dal81 is phosphorylated by Cdk1 \[[@B19]\] which is considered to be involved in G2/M transition \[[@B20]\]. The activity of Dal81 peaked over the G2/M phase is also in agreement with its regulatory role (Figure [9B](#F9){ref-type="fig"}). Hir2 functions as a transcriptional repressor of histone gene expression during the cell cycle \[[@B15]\]. Histone synthesis is triggered at the beginning of the S phase. So Hir2 is expected to have the lowest TFA around S phase. As such, TFA of Hir2 from NCA showed that it indeed has the lowest TFA around S phase (Figure [7](#F7){ref-type="fig"}). Most well characterized transcription factors showed biologically relevant activity profiles at specific cell cycle phases, suggesting that TFAs deduced from gNCA are biologically meaningful and a good predictor of transcription factor function and interaction. ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Hierarchical clustering of TFA profiles of the 11 known cell cycle factors (blue) and the 5 putative cell cycle-dependent factors (black). TFs denoted in blue are those TFs known to be involved in cell cycle. ::: ![](1471-2164-6-90-7) ::: Conclusion ========== Protein function and interaction are often deduced computationally through analysis of genomic sequence, protein sequence, domain architecture, phylogenic profile, or gene expression level. Gene expression analysis represents the only current technique that takes into account dynamic behavior to assign interaction and function. However, it has been shown that for proteins significantly regulated post-transcriptionally (e.g. transcription factors) their gene expression levels do not correlate well with their activity. Here a method for screening the physiological roles of transcription factors and their functional interactions based on their dynamic activity profiles was introduced. Our method first determines TFAs, which were then further analyzed with cluster and periodicity analysis to determine transcription factor function. By combining the results from both cluster analysis and periodicity analysis, we recovered more than 90% of transcription factors that are proposed to be involved in cell cycle regulation. In addition, we discovered 5 putative transcription factors that may be related to cell cycle regulation. Functional interactions between transcription factors were determined by isolating statistically significant perturbations in TFA patterns between mutant Δfkh1Δfkh2and wild-type experiments carried out under the same experimental conditions. This method allowed for the identification of 4 novel and 3 previously verified fork-head transcription factor interaction partners. We recognize that the Chip-chip data may be condition-dependent and noisy. However, together with microarray data, they provide useful information for functional deduction of transcription regulators. Methods ======= Gene expression data and connectivity information ------------------------------------------------- In this study, we utilized microarray data sets which were taken from wild-type S. cerevisiaecultures synchronized by three independent methods, α-factor arrest, elutriation, and arrest of a cdc15temperature-sensitive mutant \[[@B13]\] and a mutant that lacks in two forkhead transcription factors \[[@B11]\], Fkh1 and Fkh2 synchronized by α-factor arrest. The two data sets were combined after imputation \[[@B21]\] to fill in missing points, and common genes without missing data points between the two were selected for NCA. The wild type and the mutant data were organized side-by-side in different columns of the data matrix. The total number of microarray experiments in the combined dataset was 69. The connectivity information between transcription factors and their regulated genes comes from the genome-wide location or ChIP-chip assay \[[@B14]\]. In this study, the p-value threshold used was 0.001 for determining connectivity. As with any other method for determining genome wide transcriptional networks, the ChIP-chip assay suffers from miss-connectivity issues. Both under- and over-prediction errors in connectivity can be rationalized to occur in genome wide location analysis. Since NCA deductions are dependent on the connectivity graph generated by ChIP-chip assays, we endeavoured to determine the effects of connectivity errors on our TFA profiles. gNCA ---- NCA and its extension, gNCA, are a recent-developed technique that is capable of deducing TFAs based on gene expression. Compared to NCA, a distinct feature of gNCA is the capability of imposing constraints on both the regulatory network and regulatory signals. This considerable expands the range of network topologies capable of being analyzed with NCA, and thus allows application of the method to a much larger expanse of scenarios. The analysis starts with defining a model that expresses gene expressions as a function of TFA and the corresponding control strengths (CS). The resulting model can be written in a log-log form. In matrix notation, this model is: E = A × P + Γ, (1) where E~ij~= log(g~i~(t~j~) / g~i~(t~0~)) and g~i~(t~j~) is the expression of i-th gene evaluated at time t~j~, Ais the regulatory network of control strengths where A~ij~denotes the control strength of transcription factor jon gene i(A~ij~= CS~ij~), P~ij~= log(TFA~i~(t~j~) / TFA~i~(t~0~)) with TFA~i~(t~j~) being the i-th transcription factor activity evaluated at time t~j~, and Γrepresents external stimulus and stochastic background noise from the DNA microarray. The dimensions of E, Aand Pare (N× M), (N× L) and (L× M), respectively, where Nis the number of genes in the network, Mis the number of data points or experiments conducted, and Lis the number of transcription factors used in the study. Since most genes are regulated by only a subset of transcription factors, Ais generally sparse. This sparse connectivity pattern in Ais defined by Z~A~, which can be obtained from ChIP-chip assay or existing databases. In addition, gNCA can impose constraints on P. For example, when a gene that codes for a specific transcription factor is deleted, the TFA profile (represented in the P matrix) is set to zero. The corresponding zero pattern of Pis denoted as Z~P~. If Z~A~and Z~P~satisfy a given set of conditions, then gNCA can decompose the data Einto Aand Pup to a diagonal scaling factor. This condition is called essential uniqueness\[[@B3],[@B22]\]. Additional technical details may be found elsewhere. Selection of multiple sub-networks for gNCA ------------------------------------------- Among the set of required conditions for an essentially uniquedecomposition, a necessary condition is that the number of transcription factors (L) must be less than the number of data points (M) in a given network. Therefore, to analyze a large network in yeast, we dissected the regulatory network into multiple sub-networks and recombine them at the end of the analysis. Each sub-network was constructed such that the number of transcription factors was less than the number of time data points. Satisfaction of the NCA identifiablity criteria \[[@B3],[@B22]\] for each sub-network requires that each sub-network contain some overlapping TFs to ensure consistencyamong the computed dynamics. This is possible because the nature of the transcriptional regulatory network as determined by genome-wide location analysis data reveals that some transcript factors highly overlap the set of genes on the microarray \[[@B14]\]. Including these transcript factors in each sub-network decomposition step achieves the required consistency. First we select randomly ktranscript factors from the column of A as an initial sub-network (S\'). If this sub-network fails a rank check we then remove the lTFs that cause the rank deficiency, and replace them with lnew TFs that were not used to create S\'. The top lTFs are chosen by ranking the TFs by the total number of genes they control in descending order. If this continues to fail greater than a specified number of iterations, then we change k(usually to a smaller number) and repeat the process. For large enough k(40 TFs) it is likely that the many genes overlap because of the a priorinetwork structure. This is by no means the only way to construct sub-networks, nor is it the optimal way. We do not guarantee that this will work in all cases, but for our purposes it is sufficient to generate consistent TFAs. Among the combined data set with imputed missing data points, a total of 1818 genes with connectivity information of 104 TFs from genome wide location analysis were used to generate random sub-networks. Sub-networks with 40 transcription factors from total 104 TFs were generated and examined for gNCAuniqueness conditions. Four gNCA-compliant sub-networks, with 1110, 847, 1015, and 795 genes respectively, were chosen such that total number of TFAs deduced from 4 random sub-networks could be greater than the number of data points (M = 69). For each sub-network containing Fkh1 or Fkh2, the corresponding TFAs in the Pmatrix corresponding to the deletion mutant were constrained to zero. Cluster analysis of TFA profiles -------------------------------- The identification of potential new cell cycle factors was conducted by using hierarchical clustering on the TFA profiles. Specifically, hierarchical clustering based on absolute value of Pearson correlation coefficient, which measures the strength and direction of a linear relationship between two variables, was applied to TFA profiles deduced by gNCA. We hypothesize that transcription factors with similar activities function together. Therefore, transcription factors which are clustered together with 11 known cell cycle factors could be considered as possible cell cycle-dependent transcription factors. Publicly available software packages, Cluster and Treeview \[[@B12]\] were used for this test. Periodicity analysis of TFA profiles ------------------------------------ We identify periodic functions from stochastic background noise by employing a statistical analysis technique described previously \[[@B4]\]. This technique is based on rejecting the null hypothesis of a purely random process through considering the power spectrum of a time dependent signal. First, the technique computes the power spectrum of a given signal and evaluates the g-statistics expressed as the contribution of the power spectrum at a specific frequency to the total intensity of the power spectrum. Large values of g-statistics suggest that the underlying process is periodic. Then, the signals are screened for periodic motion using multiple testing under the criterion of false discovery rate (FDR). The R package GeneTS. \[[@B4]\] was used for this study. Statistical analysis for interaction determination -------------------------------------------------- The effect of fkh1and fkh2mutations on TFAs is elucidated by examining the correlation between TFAs from the wild type and the mutant under the same experimental condition. If there is a significant difference in TFAs between the wild type and mutant, the correlation coefficient should be low. However, low correlation coefficients may be caused by low overall activities, since in those situations the TFAs are dominated by noise. Therefore, we need to define a new deviation coefficient of TFAs between two strains to systematically identify transcription factors whose activity profiles are affected by fkh1 fkh2mutation. This new deviation coefficient is defined as, ![](1471-2164-6-90-i1.gif) where superscripts wtand mtrepresent wild type and fkh1 fkh2mutant, respectively, and the overhead bars represent the vector magnitude of TFA. Small values of the deviation coefficient suggest activities dominated by noise. While large values suggest significant activities and significant differences in TFA profiles between mutant and wild-type. Authors\' contributions ======================= YLY performed the analysis, evaluated the results, and drafted the manuscript. JS carried out spectral analysis of TFAs. MPB performed the network sensitivity investigation and participated in the design of this study. SJG developed the subnetwork generation algorithm. JCL initiated the project, helped with evaluation of the results and the manuscript, and provided mentorship. All authors read and approved the final manuscript. Acknowledgements ================ This work is supported by NSF grant BES-0120359, and Center for Cell Mimetic Space Exploration (CMISE), NASA University Research, Engineering and Technology Institute (URETI), under award number \#NCC 2-1364.	PubMed Central	2024-06-05T03:55:59.633598	2005-6-10	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180827/", "journal": "BMC Genomics. 2005 Jun 10; 6:90", "authors": [ { "first": "Young-Lyeol", "last": "Yang" }, { "first": "Jason", "last": "Suen" }, { "first": "Mark P", "last": "Brynildsen" }, { "first": "Simon J", "last": "Galbraith" }, { "first": "James C", "last": "Liao" } ] }
PMC1180828	Background ========== Tuberculosis (TB) is a leading cause of morbidity and mortality worldwide. According to the World Health Organization, about one-third of the world\'s population is infected with Mycobacterium tuberculosis, and about 8 million new cases of TB occur each year. Despite this large burden and intensive control efforts, only about 46% of the new infectious TB cases are detected each year \[[@B1]\]. Conventional TB diagnostics include sputum microscopy and culture of M. tuberculosis. Although microscopy is simple, specific and rapid, it suffers from low sensitivity (30--70%) \[[@B2]\]. Microscopy is particularly insensitive in HIV-infected populations. The laboratory turn around time for M. tuberculosisgrowth on solid culture media is around eight weeks. Cultures on liquid media are more rapid. Although the cost of liquid cultures has recently been reduced, most hospitals in developing countries may not find the test affordable. Low-income countries needs diagnostic tools that are sensitive, specific, cost effective, easy to perform, and easy to implement within the current infrastructure \[[@B3]\]. Among the various alternative diagnostic tests being evaluated, tests based on mycobacteriophages have shown promise \[[@B4]\]. Phage-based tests are relatively easy to perform, but requires the type of laboratory infrastructure that is needed for routine mycobacterial cultures. The turnaround time of phage-based tests is 2 days compared to about 2 hours (microscopy) or up to 2 months (culture). Two large-scale studies have shown that the test detected 65--83% of the confirmed TB cases within 48 hours; the specificity of the tests in each of the studies was \>95% \[[@B5],[@B6]\]. The assay could offer new tools for the rapid diagnosis of TB in both the developed and developing world \[[@B3]-[@B18]\]. However, it is important to systematically review the literature and summarize the current evidence on these new assays. Two main phage-based approaches are used to detect M. tuberculosis(Figure [1](#F1){ref-type="fig"}) \[see Additional file\]: (i) amplification of phages after their infection of M. tuberculosis, followed by detection of progeny phages using helper cells (plaque formation); and (ii) detection of light produced by luciferase reporter phages (LRP) by live M. tuberculosis\[[@B3]\]. Phage-based tests are available as commercial kits (e.g. FASTPlaque-TB^®^and PhageTek MB^®^, a name variant of the FASTPlaque-TB, Biotec Laboratories Ltd, UK) \[[@B14]\] and as in-house (laboratory-developed) assays \[[@B15]\]. In-house tests use either amplification technology (e.g. phage amplified biologically \[PhaB\]) or LRPs. Some of the phage-based tests are designed to rapidly detect rifampin resistance (e.g. FASTPlaque-MDRi) in culture isolates. Existing studies show that the accuracy of phage based tests might vary \[[@B18]\]. Because traditional reviews are often subjective, less comprehensive, and rely on qualitative methods \[[@B19]\], we conducted a systematic review to evaluate the overall accuracy of phage-based tests for the direct detection of M. tuberculosisin clinical specimens. We addressed the following questions in our review: ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Bacteriophage-based assays for diagnosis of tuberculosis. ::: ![](1471-2334-5-59-1) ::: 1\. What is the overall accuracy of phage-based tests compared to the gold standard, culture? 2\. What is the overall accuracy of phage-based tests compared to smear microscopy? 3\. How does the accuracy of phage-based tests vary by sputum smear status? Methods ======= Search strategy and selection criteria -------------------------------------- We searched the following electronic databases: PubMed (1985--2004), Web of Science (1985--2004), EMBASE (1988--2004), and BIOSIS (1993--2004). All searches were up to date as of November 2004. The search terms included \"tuberculosis,\" \"Mycobacterium tuberculosis,\" \"mycobacteria,\" \"bacteriophages,\" \"mycobacteriophage,\" \"phage,\" \"FASTPlaque,\" \"phage amplification,\" \"phage-based,\" \"bacteriophage-based,\" \"sensitivity and specificity\", \"accuracy\" and \"predictive value\". We also searched the reference lists from the primary studies and review articles, sought help from experts in the field, and obtained lists of studies from companies that manufacture commercial rests. Although we did not impose language restriction while searching, we included only English language articles for our review. We followed a written protocol and explicit study selection criteria. Studies were included in the review if they met the following criteria: comparison of phage test against a reference standard, data necessary for the computation of both sensitivity and specificity, and culture (either liquid or solid media) as the reference standard. Two reviewers (SK and MP) independently screened the titles and abstracts to identify eligible studies. Disagreements between the reviewers were resolved by consensus. A list of excluded studies, along with the reasons for exclusion is available from the authors on request. Data abstraction and quality assessment --------------------------------------- The final set of included articles was assessed by one reviewer (SK), who extracted data from all the studies using a piloted data extraction form. A second reviewer (MP) independently extracted data from a subset of the included studies to evaluate reproducibility of data extraction. The inter-rater agreement between the two reviewers was 100%. Data retrieved from the reports included methodological quality, participant characteristics, laboratory methods, and outcome data (sensitivity and specificity). We assessed the quality of the studies by using criteria selected from the QUADAS checklist \[[@B20]\] for assessment of quality of diagnostic studies: study design (cross-sectional versus case-control), blinding (single/double blind versus unblinded interpretation of index test and reference standard results), and potential for verification bias (complete versus partial/differential verification of index test results by reference standards). To compute sensitivity and specificity of phage-based assays, we chose sputum culture on solid and/or liquid media (e.g. BACTEC 460) as the reference standard in our review. To compare the accuracy of phage based assays with sputum microscopy, we evaluated studies that reported head-to-head comparisons of sputum microscopy with phage based assays (against a common reference standard) and computed the diagnostic yield of phage based assays over and above sputum microscopy. To evaluate the effect of sputum-smear status on accuracy, we conducted subgroup analyses for smear-positive and smear-negative specimens. Statistical analysis -------------------- We used standard methods recommended for meta-analyses of diagnostic test evaluations \[[@B21]\]. Data were analyzed with Meta-Disc (version 1.1.1) software \[[@B22]\]. Our analyses focused on the following measures of diagnostic accuracy: sensitivity (true positive rate \[TPR\]), and specificity (1-false positive rate \[FPR\]). Each study in the meta-analysis contributed a pair of numbers: TPR and FPR. Since these measures are correlated and vary with the thresholds (cut points), it is customary to analyze TPR and FPR proportions as pairs, and to also explore the effect of threshold on study results \[[@B23]\]. We summarized the joint distribution of sensitivity and specificity using the Summary Receiver Operating Characteristic (SROC) curve. Unlike a traditional Receiver Operating Characteristic plot that explores the effect of varying thresholds on sensitivity and specificity in a single study, each data point in the SROC space represents a separate study. The SROC curve is obtained by fitting a regression curve to pairs of TPR and FPR. The SROC curve and the area under it present an overall summary of test performance, and display the trade off between sensitivity and specificity. A symmetric, shoulder-like SROC curve suggests that variability in thresholds employed could, in part, explain variability in study results \[[@B19]\]. The area under the SROC curve is a global measure of overall test accuracy. An area under the curve of 100% indicates perfect discriminatory ability. Heterogeneity in meta-analysis refers to a high degree of variability in study results \[[@B24]\]. Such heterogeneity could be due to variability in thresholds, disease spectrum, test methods, and study quality across studies \[[@B24]\]. In the presence of significant heterogeneity, pooled, summary estimates from meta-analyses are not meaningful. We investigated heterogeneity using subgroup analyses. Meta-regression was not attempted because of the small number of studies identified. Results ======= Study selection --------------- Figure [2](#F2){ref-type="fig"} describes the study selection process. Thirteen studies from 12 articles \[[@B5]-[@B13],[@B15]-[@B17]\], of the 532 articles identified, met our eligible criteria, and were included in our final analyses. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Study flow. ::: ![](1471-2334-5-59-2) ::: Description of included studies ------------------------------- Studies included in the final analyses are summarized in Table [1](#T1){ref-type="table"}. The total number of samples analysed in all the studies were 5820 (mean 448); 1330 (23%) of the samples were culture positive. Most studies enrolled subjects with suspected pulmonary tuberculosis, as suggested by history, physical examination or chest radiograph. Three studies \[[@B8],[@B16],[@B17]\] included non-respiratory specimens (urine, CSF, pleural fluid and lymph node aspirate\], in addition to sputa. Except one \[[@B9]\], no study provided data on age range, sex ratio and prevalence of HIV. In half the studies, patients had received some treatment before they were evaluated. The study design was cross-sectional in 11 of 13 (85%) \[[@B5]-[@B13],[@B15]\], and case-control in 2 studies (15%) \[[@B11],[@B17]\]. Only 2 \[[@B13],[@B15]\] of 13(15%) studies reported blinded comparison of the phage based assays with the reference standard. No study had potential for incorporation bias. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Study characteristics and methodological quality of included studies ::: Study Characteristics Frequency -------------------------------------------------------------------- --------------- Study Design Cross-sectional 11 Case-control 2 Verification of phage tests with reference standard Complete 13 Partial 0 Blind assessment of phage tests and reference standard results Yes 3 Unclear 10 Test before treatment Yes 6 No 5 Unclear 2 Year of Publication Before 2002 4 After 2002 9 Study size \< 20% positive specimens 2 \> 20% positive specimens 11 Reference standard BACTEC 460 3 LJ medium 6 Bactec and LJ\* 3 LJ/AMTD\\ 1 Type of assay Commercial 11 In-house 2 \LJ: Lowenstein Jensen medium \\AMTD: Amplified Mycobacterium tuberculosisDirect Test ::: Assay characteristics --------------------- All studies (except one that classified any visible plaques as positive \[[@B15]\] used the same cut point for the phage amplification assay: 20 or more plaques were reported as a positive test. No study used luciferase reporter phages to detect M. tuberculosisin clinical specimens. Sputum samples were processed by a standard N-acetyl-L-cysteine-sodium hydroxide (NALC-NaOH) method in all studies evaluating commercial phage-based assays. Only three studies \[[@B6],[@B7],[@B13]\] reported the proportion of the phage tests or the reference tests that were contaminated. Non-tuberculous mycobacteria (NTM) were grown in 4 studies \[[@B6],[@B9],[@B10],[@B17]\], but most studies provided no data on what proportion of the NTM isolates were positive by the phage test. Six studies \[[@B6],[@B9],[@B11]-[@B13],[@B15]\] had used LJ cultures as the reference standard; three studies \[[@B5],[@B7],[@B10]\] used BACTEC 460; three studies \[[@B8],[@B17]\] used LJ and BACTEC methods and one study \[[@B13]\] used LJ and AMTD tests. The index test was FASTPlaque in 10 studies \[[@B5]-[@B8],[@B10]-[@B13],[@B16],[@B17]\], PhageTek in 1 study \[[@B9]\] and in-house amplification assays in 2 studies \[[@B13],[@B15]\]. None of the studies used LRP assays. Most studies reported sensitivity and specificity data with specimens as the unit of analysis (not individuals). Overall accuracy of phage assays -------------------------------- We identified 13 studies that assessed the sensitivity of phage-based assays (Table [2](#T2){ref-type="table"}, Figure [3](#F3){ref-type="fig"}). Sensitivity is the proportion of patients with culture-proven tuberculosis who are positive by phage-based assays. Overall, as seen in Figure [3](#F3){ref-type="fig"}, the sensitivity of the phage based assays varied widely, from 0.21 to 0.94 (test of heterogeneity: p \< 0.001). Because of the significant heterogeneity, sensitivity estimates were not pooled. Three studies \[[@B9],[@B11],[@B13]\] reported very low sensitivity. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Description of studies in the meta-analysis and measures of test accuracy ::: Source Country Study Design Blinding Complete verification Specimen Treatment status Test Patients With TB (No./Overall) Reference Standard Sensitivity (95%CI) Specificity (95%CI) ----------------------- -------------- -------------- ---------- ----------------------- --------------- ------------------ -------------- -------------------------------- -------------------- --------------------- --------------------- Commercial assays* Albay (2003) Turkey CS Unclear Yes Sputum Untreated FASTPlaque 64/192 BACTEC 0.88 (0.77, .94) 0.97 (0.92, 0.99) Albert (2002) South Africa CS Unclear Yes Sputum Untreated FASTPlaque 207/1618 BACTEC 0.72 (0.66,0.78 0.99 (0.98,0.99) Alcaide (2003) Spain CS Unclear Yes Sputum+ other Some treated FASTPlaque 144/2048 B+LJ 0.58 (0.50,0.66) 0.99 (0.99,0.99) Bellen (2003) Philippines CS Unclear Yes Sputum Some treated Phage Tek 103/206 LJ 0.31 (0.22,0.41) 0.86 (0.78,0.92) Butt (2004) Pakistan CS Unclear Yes Sputum Untreated FASTPlaque 60/160 BACTEC 0.77 (0.64,0.87) 0.96 (0.90,0.99) Cavusoglu (2002) Turkey CC Unclear Yes Sputum Some treated FASTPlaque 33/63 LJ 0.30 (0.15,0.49) 0.97 (0.83,1.00) Marei (2003) Egypt CS Unclear Yes Sputum Untreated FASTPlaque 60/160 BACTEC 0.77 (0.64,0.87) 0.96 (0.90,0.99) Mbulo (2004) Zambia CS Yes Yes Sputum Untreated FASTPlaque 29/115 LJ 0.21 (0.08,0.40) 0.91 (0.82,0.96) Muzaffar (2002) Pakistan CS Unclear Yes Sputum NR FASTPlaque 103/1209 LJ 0.82 (0.76,0.86) 0.98 (0.95,0.99) Shenai (2002) India CS Unclear Yes Sputum+ other Some treated FASTPlaque 62/90 BACTEC+LJ 0.76 (0.60,0.89) 1.00 (0.74,1.00) Shenai (2004) India CC Unclear Yes Sputum+ other Some treated FASTPlaque 103/129 BACTEC+LJ 0.94 (0.79,0.99 0.83 (0.52,0.98) In- house assays McNerney (2004) Zambia CS Yes Yes Sputum NR In-house 220/496 LJ 0.44 (0.37,0.51) 0.92 (0.89,0.95) Mbulo (2004) Zambia CS Yes Yes Sputum Untreated In -- house 245/514 LJ+AMTD 0.45 (0.32,0.60) 0.95 (0.85,0.99) Abbreviations: CS, cross-sectional studies; CC, case-control studies; NR, Not recorded. ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Forest plots of estimates of sensitivity and specificity for commercial and in-house assays. circle: commercial assays; rectangles: in-house assays. Error bars represent 95% CIs. ::: ![](1471-2334-5-59-3) ::: Thirteen studies assessed the specificity of phage-based assays (Table [2](#T2){ref-type="table"}, Figure [3](#F3){ref-type="fig"}). Specificity is the proportion of individuals without tuberculosis (culture-negative) who are negative by the phage- based assay. The specificity estimates were high and fairly consistent across the studies (range: 0.83 to 1.00). Except two studies \[[@B9],[@B17]\] that reported false- positive rates of 14% and 17% respectively; all other studies reported \>90% specificity. The in-house assays were as specific as commercials assays. Figure [4](#F4){ref-type="fig"} presents the sensitivity and specificity estimates in a SROC space. The curve shows that most studies had high specificity with lower and highly variable estimates of sensitivity. Although the area under the curve was 0.95, significant heterogeneity in the sensitivity estimates precluded the determination of clinically meaningful summary estimates of accuracy. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Summary receiver operating characteristic (SROC) curves for commercial and in-house assays. Each circle (commercial assay) and rectangle (in-house assay) represents an individual study. ::: ![](1471-2334-5-59-4) ::: Effect of sputum smear status on accuracy of phage tests -------------------------------------------------------- Because phage-based assays have threshold of detection of about 100 viable bacilli, they are expected to perform better in smear-positive samples compared to smear- negative ones. We performed subgroup analysis to evaluate this hypothesis. Five studies \[[@B5],[@B6],[@B8]-[@B10]\] provided estimates of accuracy, stratified by smear status (Table [3](#T3){ref-type="table"}, Figure [5](#F5){ref-type="fig"}). Smear-positive specimens tended to yield higher estimates of sensitivity than smear-negative specimens. The sensitivity ranged between 0.29 and 0.87 in smear positive and 0.13 to 0.78 in smear negative specimens. The specificity ranged between 0.60 and 0.88 among smear positive specimens and 0.89 to 0.99 in smear negative specimens. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Sensitivity and specificity of phage assays in studies that reported results stratified by smear microscopy status ::: Sputum smear microscopy status ----------------- -------------- ------------ ------------------------------------ ------------------ ------------------ ------------------ Albert (2002) South Africa 1618 (207) 0.87 (0.79,0.92) 0.83 (0.69,0.93) 0.49 (0.37,0.60) 0.99 (0.99,1.00) Alcaide (2003) Spain 2048 (144) 0.75 (0.66,0.83) 0.76 (0.55,0.91) 0.13 (0.04,0.27) 0.99 (0.99,1.00) Bellen (2003) Philippines 206 (103) 0.29 (0.20, 0.40) 0.83 (0.67,0.93) 0.45 (0.17,0.77) 0.89 (0.78,0.95) Butt (2004) Pakistan 160 (60) 0.76 (0.60, 0.88) 0.60 (0.15,0.95) 0.78 (0.52,0.93) 0.98 (0.93,1.00) Muzaffar (2002) Pakistan 514 (245) 0.87 (0.82,0.92) 0.88 (0.63,0.98) 0.67 (0.55,0.78) 0.98 (0,96,1.00) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Forest plots of estimates of sensitivity and specificity for smear positive and smear negative specimens. ::: ![](1471-2334-5-59-5) ::: Head-to-head comparisons against sputum microscopy -------------------------------------------------- Ten studies -- 8 commercial \[[@B5]-[@B10],[@B12],[@B13]\] and 2 in-house \[[@B13],[@B15]\] -- provided data on head-to-head comparisons of sputum microscopy with phage based assays, against a common reference standard viz. culture \[Table [4](#T4){ref-type="table"}; Figure [6](#F6){ref-type="fig"}\]. Eight studies reported staining sputum smears with Ziehl -- Neelsen technique \[[@B6],[@B7],[@B9]-[@B12],[@B16],[@B17]\], whereas four studies used fluorescent microscopy \[[@B5],[@B8],[@B13],[@B15]\]. Except three studies \[[@B5]-[@B7]\], other studies did not report clinically meaningful higher sensitivity for the phage assays. Interestingly, the sensitivity of phage based assays was lower than that of microscopy in four studies \[[@B8],[@B9],[@B13],[@B15]\]. The SROC plots \[Figure [6](#F6){ref-type="fig"}\] suggest that phage assays have a slightly higher accuracy than microscopy (area under the SROC curve was 0.95 for phage assays, whereas it was 0.86 for microscopy). ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Head-to-head comparison between smear microscopy and phage assays ::: Smear microscopy Phage test Difference (Phage -- Microscopy) ----------------------- ------ ------ ------------------- ------------------- ---------------------------------- ------------------- --------- --------- Commercial assays Albay 2003 192 0.58 (0.44,0.70) 1.00 (0.97,1.00) 0.88 (0.77,0.94) 097 (0.92,0.99) 0.30 \- 0.03 Albert 2002 1618 0.62 (0.55,0.69) 0.97 (0.96, 0.98) 0.72 (0.66,0.78) 0.99 (0.98,0.99) 0.10 0.02 Alcaide 2003 2048 0.73 (0.65,0.80) 0.99 (0.98,0.99) 0.58 (0.50,0.66) 0.99 (0.99,0.99) -0.15 0 Bellen 2003 204 0.89 (0.82,0.95) 0.60 (0.50,0.70) 0.31 (0.22,0.41) 0.86 (0.78, 0.92) \- 0.58 0.26 Butt 2004 160 0.70 (0.57.0.81) 0.95 (0.89,0.998) 0.77 (0.64, 0.87) 0.96 (0.90,0.99) 0.07 0.01 Marei 2003 38 0.64 (0.31,0.89) 0.93 (0.76, 0.99) 0.55 (0.23, 0.83) 1.00 (0.87, 1.00) -0.09 .07 Mbulo 2004 496 0.45 (0.32, 0.60) 0.98 (0.91, 1.00) 0.45 (0.32, 0.60) 0.91 (0.82, 0.96) 0 -.07 Muzaffar 2002 514 0.71 (0.65,0.77) 0,94 (0.90,0.96) 0.82 (0.76,0.86) 0.98 (0.95,0.99) 0.11 0.05 In- house assays Mbulo 2004 115 0.48 (0.29, 0.60) 0.90 (0.81, 0.95) 0.21,(0.08,0.40) 0.91 (0.82,0.96) -0.27 0.14 McNerney 2004 496 0.60 (0.53,0.66) 0.87(0.83,0.91) 0.44 (0.37, 0.51) 0.92 (0.89,0.95) \- 0.16 0.05 ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Summary receiver operating characteristic (SROC) curves for sputum microscopy and phage-based assays. Each solid circle represents a study in the meta-analysis. ::: ![](1471-2334-5-59-6) ::: Discussion ========== Overall accuracy of phage-based tests ------------------------------------- Our results indicate that phage-based assays are highly specific but not sensitive enough to be equivalent to culture. Phage-based assays detected M. tuberculosisin one-half to two thirds of sputum samples with specificity that ranged between 0.83 and 1.00. On the other hand, sensitivities varied between 0.21 and 0.94. Ten of the 13 studies used FASTPlaque-TB kit; the two studies that used in-house assays were less accurate than the commercial assays. Because of heterogeneity in estimates of accuracy across studies, we did not calculate pooled estimates of sensitivity and specificity. What factors might explain the observed heterogeneity? The differences in the extent of treatment given to patients prior to collecting specimens, the potential differences in timeliness of specimen transport and processing may have had impacts on the number of viable bacilli in the specimens. Since the phage-based assay detects only viable mycobacteria, and has a threshold of detection of about 100 viable bacilli, any factor that impacts viability of TB bacilli can affect sensitivity of the assay. Of the 5 studies of commercial assays in patients who were not treated prior to specimen collection, 4 studies reported sensitivities of 0.72 or greater. Conversely, only 2 of the 5 studies whose patient population included those on anti-TB therapy had sensitivities greater than 0.72. It is likely that anti-TB therapy in these studies decreased the number of viable bacilli in specimens. Further assessments of this issue should be made. Another important issue that may have been responsible for heterogeneity in sensitivity estimates was specimen transport and the potential impact of environmental conditions on specimen viability. Rapid processing of specimens and timely initiation of the phage assay is important: the sensitivity of the phage assay was 72% when the specimens were processed immediately and tests were performed daily \[[@B5]\] compared to the sensitivity of 29% when the processing of the specimens was delayed and the assays were run twice a week \[[@B9]\]. The specificity of phage-assays for detecting acid fast bacilli is likely to decrease in settings in which infection with mycobacteria other than M. tuberculosisare common. Sodium hydroxide, used to decontaminate the specimens, may damage the acid fast bacilli and can reduce the sensitivity of phage-based assays for detecting tuberculosis. Use of gentler decontamination techniques can reduce the specificity of the test: this approach protects the acid fast bacilli but fails to prevent contamination from other microorganisms. Contamination of LJ slants with micro-organisms is a common problem in certain settings (40.4% in Zambia \[[@B13]\] and 18.2% in Pakistan \[[@B6]\]). The exclusion of contaminated results may result in biased estimates of sensitivity and specificity. This approach (partial verification) can lead to bias if systematically more abnormal than normal test results are subjected to the reference standard. Although no study reported use of smear instead of contaminated cultures, such a strategy could weaken the reference standard and result in differential verification bias. The impact of HIV infection on test accuracy could not be determined in this review because none of the studies reported the proportion of HIV infections among the tested population. Because HIV infection may impact the extent of viable mycobacteria present in sputum specimens, studies of populations with defined HIV status should be performed. Also, the varying sensitivity of the reference standard might have had an impact on the accuracy of the phage-based assays and could have contributed to the heterogeneity. Because grouping the studies in four sub-groups (LJ media, BACTEC media, LJ and BACTEC and AMTD and LJ) would have resulted in small numbers of studies in each of the category, we did not attempt statistical comparisons. Accuracy of phage-based tests compared to smear microscopy ---------------------------------------------------------- Although some phage-based studies proved to be more sensitive than smear microscopy \[[@B5]-[@B7],[@B10]\], others were not \[[@B8],[@B9],[@B12],[@B13],[@B15]\]. Differences in performance of phage-based tests compared to smear microscopy were also determined for smear-positive and smear-negative, culture-positive specimens. Overall, sensitivity in smear-positive specimens appeared to be higher than in smear-negative specimens, but among smear -negative specimens, phage- based assays had high specificity. To evaluate the accuracy of smear microscopy for detecting M. tuberculosis, except one study \[[@B5]\] in which patients provided two sputum specimens each, all studies used only a single sputum sample. This approach differs from the current World Health Organization (WHO) and International Union against Tuberculosis and Lung Disease (IUATLD) recommendations, which state that at least three sputum samples must be examined for each patient \[[@B25]\]. Use of a single sample might make the smear microscopy look less sensitive than what it actually it is because the sensitivity may go up with greater number of smear tests. It would be interesting to evaluate the additional yield from repeated sputum examinations by microscopy and culture and do a head-do-head comparison of 3 sputum smears with phage-based assays. To our knowledge, this has not been done in any of the available studies. Non- tuberculous mycobacteria are also known to contribute to false positive phage test results. The phage can be amplified by almost any mycobacteria present in the sputum and it is important to verify that the sputum contains Mycobacterium tuberculosiscomplex. The effect of non-tuberculosis mycobacteria (NTM) might be minimal in places where true M. tuberculosisis very common (high incidence settings). In settings with low incidence, NTM might have a greater impact. A requirement for a confirmatory test would make the overall testing strategy more expensive, delay the diagnosis, and result in an additional patient visit to the laboratory. Clearly, these factors could adversely impact the practical applicability of the test in resource -limited settings with high burden of tuberculosis. Spectrum and selection bias are known to affect sensitivity and specificity of a diagnostic test. Spectrum bias may occur when the test is assessed in a study population with a different clinical spectrum than will be found among those in whom test is to be applied in clinical practice. Specimens in several reviewed studies were collected from patients reporting to tertiary care centres \[[@B10],[@B11],[@B16],[@B17]\]. Such patients are more likely to have advanced disease with a large bacillary load. Selection bias could also have played a role in influencing the diagnostic properties of phage-based assays -- patients referred to chest clinics and those with abnormal chest radiographs are more likely to have a higher bacillary load. The referral bias can make the phage-based assays appear more sensitive than they actually are. Overall, our review suggests that when a patient\'s phage test is negative, there is roughly one in three probability that patient has tuberculosis. A negative test, therefore, does not exclude tuberculosis in patients suspected to have tuberculosis. Strengths and weaknesses of the review -------------------------------------- Our review has several strengths. We followed a written protocol, performed a very comprehensive search of several databases and sources to identify studies. We assessed the quality of included studies by using established criteria \[[@B20]\]. We also analyzed commercial and in-house assays separately, to take into account the potential differences between the assay techniques. Lastly, we used summary ROC curves to take into account the impact of varying thresholds and also the interdependence of sensitivity and specificity. Our review has some limitations. Most studies provided no data on prevalence of TB and/or HIV in their study population. Several studies did not specify whether the specimens were collected before or after anti-TB therapy was started. Most studies did not specify the volume and quality of sputum specimens, the proportion of contaminated phage tests and indeterminate results. Most studies did not specify if non-tuberculous mycobacteria were isolated in the study. We could not analyse such factors as laboratory infrastructure and expertise with phage assays on the accuracy of phage assays. Although we explored the issue of heterogeneity using subgroup analyses, techniques such as meta-regression may be useful in the evaluation of heterogeneity. However, we were unable to perform metaregression because we had only 13 studies in our systematic review; with only 13 data points, it is difficult to fit and interpret a regression model. Also, because of the small number of data points, we were unable to perform statistical comparisons between subgroups. Finally, bias may have been introduced by the exclusion of non-English language studies. Clinical applicability and implications --------------------------------------- Since many countries with a high TB burden do not use culture, an inexpensive alternative diagnostic test is necessary. One-third to two-thirds of all cases of pulmonary tuberculosis are not being detected by the commonly used smear microscopy \[[@B25]\]. Can phage-based tests be used to diagnose pulmonary tuberculosis in high burden and limited resource settings? Although our review shows that phage-based tests have high specificity, their sensitivity is lower and highly variable. Also, they do not have substantially higher accuracy than sputum microscopy. In addition, phage-based assays are much more complicated and labor-intensive than microscopy and thus cannot be performed in primary-care settings; they require a laboratory infrastructure similar to that required for performing cultures. Although it can be argued that the rapid and specific diagnosis of 50% or more of TB patients can be made within 2 days, leading to reduced potential for ongoing transmission and more effective management of individual patients \[[@B26]\], our review suggests that phage-based assays need to have higher sensitivity before such potential advantages can be realized. Mbulo et al \[[@B13]\] point out that the main disadvantage of phage-based test compared to sputum smear microscopy is that specimens need to be transported to a specialist reference laboratory. The fact that phage-based assays are more expensive than smear microscopy, need a trained microbiologist makes them less suitable in resource-poor settings. Also, it remains to be seen how the test performs outside reference laboratories. Molecular strain typing tests such as polymerase chain reaction (PCR) are competitors for phage-based assays. Of the two studies \[[@B7],[@B12]\] that reported head to head comparisons of molecular tests with phage -- based assays against a common reference test (i.e. culture), one study \[[@B7]\] found no difference in test accuracy between the phage-based assay and the polymerase chain reaction; the other study \[[@B12]\] found that phage-based assay was less sensitive (64% vs. 82%) but more specific (93% vs. 85%) compared to the polymerase chain reaction. Conclusion ========== Our review suggests that phage-based assays have high specificity, but modest and variable sensitivity. Their accuracy is slightly higher than smear microscopy in head to head comparisons. However, because of the overall low sensitivity, the similarity of phage-based assays to sputum microscopy with respect to accuracy, and the need for a fairly advanced laboratory infrastructure, phage-based assays cannot replace conventional diagnostic tests at this point. Further research is required to identify methods that can enhance the sensitivity of phage-based assays, without compromising the high specificity. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= SK formulated the research question, searched the databases, extracted the data, performed the statistical analysis and wrote the manuscript. MP participated in the study design, extracted a part of data, and helped to draft the manuscript. LP contributed to the development of the study protocol, provided critical inputs in laboratory associated issues and reviewed the manuscript. LR and AR contributed to the review of the manuscript. All authors read and approved the final manuscript. Ethical approval ================ Not required. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2334/5/59/prepub> Supplementary Material ====================== ::: {.caption} ###### Additional File 1 Permission to reproduce figure 1 from Biomedica. ::: ::: {.caption} ###### Click here for file ::: Acknowledgements ================ This work was supported by the Fogarty AIDS International Training Program (1-D43-TW00003-16), and the National Institutes of Health NIH/NIAID grant R01 AI 34238. We are grateful to Ruth McNerney for sending us additional information and representatives of the test manufacturers (Richard Mole and Heidi Albert) for sending us a list of studies and hard copies of published studies. We would like to thank Cooker Perkins for helping us retrieve references and to Dr. P. Narang for reviewing the manuscript. We are also grateful to Manzour Hazbon for providing us figures of phage-based assays.	PubMed Central	2024-06-05T03:55:59.637957	2005-7-16	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180828/", "journal": "BMC Infect Dis. 2005 Jul 16; 5:59", "authors": [ { "first": "Shriprakash", "last": "Kalantri" }, { "first": "Madhukar", "last": "Pai" }, { "first": "Lisa", "last": "Pascopella" }, { "first": "Lee", "last": "Riley" }, { "first": "Arthur", "last": "Reingold" } ] }
PMC1180829	Background ========== Any reader of the scientific medical literature knows that research prose is a very specialized use of language, distant from the general intellectual prose. Unlike writing in the humanities, the single most important function of scientific writing is the transfer of exact information and explicitly stated ideas. The writer of scientific reports attempts to convey the most precise meaning, in a logically coherent order, and in as few words as possible. Despite the obvious importance and distinction of the scientific writing when compared to other types of writing, no previous articles have described software solutions specifically designed to facilitate the process of manuscript writing (scientific articles and grant proposals) in virtual interdisciplinary groups. The fundamental purpose of scientific discourse is not only the presentation of information, but rather its communication. It does not matter how well authors might regard their own writing; what really matters is whether the large majority of the audience understands what the authors wanted to communicate. Achieving simplicity in research texts is a complex task, the failure to write the final manuscript being one of the most common reasons for completed research projects not being published in peer-review journals \[[@B1],[@B2]\]. The reason for this complexity is usually misunderstood. Most people assume that the difficulties in scientific writing are inherent to the scientific concepts, data, and analysis. However, Gopen \[[@B3]\] has argued that complexity of thought does not necessarily lead to a difficult text. In our manuscript, we argue that a software solution may assist researchers in simplifying this task. It has been shown that information is interpreted more easily and more uniformly if it is placed where most readers expect to find it \[[@B3]\]. These needs and expectations of readers affect the interpretation not only of tables and illustrations but also of the text itself. Scientific readers have relatively fixed expectations about where in the text structure of scientific manuscripts they will encounter particular concepts. If writers can become consciously aware of these locations, they can better control the degrees of recognition and emphasis a reader will give to the various pieces of information being presented. Experienced researchers are aware of these expectations, but this skill is acquired usually after many failed attempts to learn the \"unwritten rules\" of scientific writing, which usually go far beyond the common IMRaD guidelines \[[@B4]\]. This underlying concept of reader expectation is perhaps most immediately evident at the level of the largest units of discourse, a unit of discourse being defined as anything with a beginning and an end (e.g., a clause, a sentence, a section, an article, etc.). A research article, for example, is generally divided into recognizable sections, sometimes labeled Introduction, Methods, Results and Discussion. When the sections are confused in situations where too much experimental detail is found in the Results section, or when discussion and results are mixed together, readers are often equally confused. If these structural expectations are continually violated, readers are forced to spend a considerable amount of effort to understand its structure, taking time way from simply understanding the underlying message. In our article, we argue that a software solution might assist researchers in meeting readers\' expectations in terms of text structure and its order, therefore substantially improving the final transmission of information. Existing approaches: To our knowledge, no previous applications have explored the ability to use a Web application to allow interdisciplinary virtual groups to work synchronously and asynchronously on the same manuscript. Currently, word processors are the most common tools used for scientific writing. Also more formally known as document preparation systems, word processors are computer applications used for the production (including composition, editing, formatting, and possibly printing) of any sort of viewable or printed material. Word processing was one of the earliest applications for the personal computer in office productivity, prior to the widespread use of the World Wide Web. Therefore, the text content is most often stored in local computers and then exchanged through e-mails. Although electronic files are efficient for texts produced by single writers, interdisciplinary collaborations require more sophisticated exchange systems. The objective of this article is to describe the Web application Manuscript Architect, designed to assist virtual interdisciplinary groups in the writing of scientific manuscripts (research papers or grant proposals). Implementation ============== Goals ----- The primary goal of the Manuscript Architect application is to simplify the process of scientific writing. Manuscript Architect divides the writing process into the following steps: (1) list main concepts (text blocks), (2) establish hierarchical order of text blocks, (3) connect text blocks, (4) ensure consistency across text blocks, (5) use previous examples of text blocks with a similar focus, (6) facilitate the translation when manuscripts are not written in the language used by the target audience (readers and journals). Design objectives ----------------- The overall objective of the Manuscript Architect project was to build a Web-based tool that would allow virtual groups to work on scientific manuscripts using synchronous and asynchronous methods. Before designing the application, we analyzed similar target applications in a diverse set of domains, including project management tools, text editors, tools for voice communication, and software tools for sharing of computer screens. Our search can be summarized into the following list of technical requirements that would be highly desirable for the application: • Hierarchical navigation of text blocks, allowing authors to easily locate their current block within the entire manuscript. The concept of text blocks has been previously described by linguists such as Hoey \[[@B5]\]. A text block is a unit of text with a single focused content. In the Manuscript Architect application, we use the term text block to refer to a large unit of text, not necessarily related to the linguistic concept of text block. Similar approaches have been used by other applications such TuxCards \[[@B6]\]. • Fast screen upload for easy transition between text blocks • Block text saving diversified among multiple application activities to avoid loss of text • Primary author should be able to assign text blocks to co-authors and track their progress • A full view of the manuscript should be available, displaying subsections as required • Print to PDF or easy transference to commonly used electronic formats • Word count by block texts and groups of block texts (e.g., abstract) • Automated e-mails when a text block is assigned • Color coding to identify the status of text blocks. For example, it should be clear for co-authors which text sections have been assigned to them and which sections have been completed • Support for insertion of hyperlinks, figures, and other electronic files • Robust security • Scalability in the associated management system, including full compatibility with our existing project management application Research Manager The Web application we have implemented to date meets all of these requirements. Other requirements will be met in our planned enhancements, as described in the Conclusion section below. Software architecture for Manuscript Architect: The language of choice for the application development was JAVA, since it facilitates the integration of the Manuscript Architect application with the other existing applications in our suite \[[@B7]\]. We chose the classic life cycle development method \[[@B8]\], which includes the following steps: (1) the analysis and specification of pre-requisites using prototyping methods, (2) project, (3) implementation and unit testing, (4) system integration and testing, and (5) operation and maintenance. The Manuscript Architect application works integrated with Research Manager, an application developed by our group for project management \[[@B7]\]. As the user logs into Research Manager, the first screen of Research Manager (Figure [3](#F3){ref-type="fig"}) displays a list of manuscripts named \"My Manuscripts\". This list represents all manuscripts where the user is either the primary author or has at least one text block assigned to her/him. Having the name of the manuscript listed under \"My Manuscripts\" serves as a reminder that a writing task is to be performed in that manuscript. Once all text blocks assigned to the co-author are completed, the manuscript name no longer appears in the \"My Manuscripts\" list. It can, however, be accessed by co-authors through the Research Manager screen available for individual projects (Figure [4](#F4){ref-type="fig"}). Only project participants have password-protected access to the manuscript. If the manuscript is accessed either by the \"My Manuscripts\" list or directly through the project itself, the interface page for Manuscript Architect is the same (Figure [5](#F5){ref-type="fig"}). ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Manuscript Architect interface. ::: ![](1472-6947-5-15-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Manuscript Architect interface. ::: ![](1472-6947-5-15-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Manuscript Architect interface. ::: ![](1472-6947-5-15-5) ::: Interface --------- The interface of Manuscript Architect is divided into two separate portions. The left portion of the screen contains a hierarchical tree with all text blocks displayed in order of appearance. The hierarchical level is represented by the left indentation. In other words, a text block of level two is contained within the text block of level one right above it. The entire manuscript is contained under the root level. Text blocks where the user is currently working are highlighted in orange to facilitate identification. The title of each text block is identified by colors: red is a text block that has been assigned (the name of the assigned co-author and date of assignment appearing on the right side of the text block title; green is a text block that has been returned from a co-author to the primary author, and blue is a text block that has been marked as completed by the primary author. A button below the hierarchical tree assembles all text blocks into a single document, displayed on the right side of the screen. The right portion of the Manuscript Architect application can display two types of screens. First, when the user has ownership over the text block, the block is editable and the window can be used as a regular word processor. A list of available editing tools include the ability to turn the title of the text block into a heading in the final manuscript, font size, bullets, numbered items, bold/italics/underlined, text alignment, table insertion, link insertion, and file insertion. Inserted pictures are displayed directly on the screen. Links to files, such as full-text documents, are displayed as hyperlinks. At the bottom of the screen a drop box allows the primary author to assign the text block to individual co-authors, save or remove the text block, and mark the text block as completed. Co-authors only have the save and completed options. Of importance, the text block is saved every time the user moves from one block to another. When the user does not have ownership of the text block, the right portion of the screen will be visible with read-only privileges. Primary authors have the ability to revoke the text block if necessary. Usability --------- Usability was evaluated at two distinct levels: (1) Formal usability analysis and (2) field observations. Formal usability analysis was conducted with ten different users with no previous experience with the Manuscript Architect. Five users had previous experience with scientific writing, while five were novice researchers. Formal usability tests \[[@B9]\] followed a protocol where users were observed by one evaluator (RP) and had to complete scientific writing using the notes and previous examples. Users were free to ask questions at any point in time. Tasks included the writing of different portions of scientific articles, assigning different text blocks to different users, count words, observe the document in full-view, recover text blocks that had been initially assigned, insert links, insert documents (pdf files), and change font and alignment characteristics. Each participant answered a questionnaire at the end of the formal usability analysis with items about interface problems, missing features, and suggestions for overall improvement. Field observations were comprised by observation of researchers using Manuscript Architect for the writing of scientific articles during an online writing workshop where over twenty different manuscripts were being written at the time by 14 different researchers (for a current list, please refer to <http://www.ceso.duke.edu/servlet/URMController>). Junior researchers include medical students, graduate students, residents, fellows, and junior faculty. During these workshops, Manuscript Architect is coupled with Voice over the Internet Protocol (VoIP) applications. Most of the manuscripts include co-authors from different disciplines located in at Duke University as well as other academic institutions in the U.S. and abroad. Similar to the formal usability tests, the evaluator (RP) took notes of problems encountered during the writing session regarding interface, missing features, and overall suggestions. Results ======= Manuscript Architect was designed to separate all activities involved in the production of a scientific text into a number of sequential steps. Following a sequence of steps facilitates the writing process since it eases the number of constraints that must be satisfied at one time and also increases the likelihood of satisfying any particular constraint. This sequence can be described as follows: 1\. The first author designs the overall structure of the manuscript. This structure includes the inclusion of specific text blocks in a hierarchical order. For example, the first author may decide that the Introduction (level 1) should have the following text blocks under it (level 2): significance, knowledge lag, brief review of the status of knowledge, and objective. The words describing the central objective of each text block becomes its title. 2\. Each text block is supplemented by a note explaining in detail the focus of the block. Notes are marked as meta-data, not being part of the main text of the manuscript. This explanation complements and extends the description already given by the title of the text block. 3\. Each text block receives a previous example of a block with a similar focus. Similar to notes, previous examples are also marked as meta-data. Examples are chosen based on their structure rather than their topic. 4\. Administrator assigns text blocks to users. Users are all co-authors of the manuscript, which has already been defined in the Research Manager application. Co-authors are chosen based on their expertise for the text block at hand. Therefore, a statistician might be assigned to the text block associated with statistical methods, while a medical student might be assigned a block where a literature review is required. When a text block is assigned, co-authors receive an automated e-mail. They are also notified of their text block assignment by a line under the \"My Manuscripts\" heading on the first page of the Research Manager application. 5\. Co-authors write their text blocks in accordance with the note guidelines and the previous example. Once completed, co-authors mark the text block as \"completed\", which automatically returns the text block to the administrator. 6\. Several iterations are made to ensure that the text block is adequately written. These iterations may be accomplished asynchronously or synchronously during writing workshops performed using VoIP (Voice of the Internet Protocol) tools. 7\. When the text block is written in a language other than the language of the target journal, the first author might assign the block to the translator. 8\. Once the block is translated, the text block is returned to the first author for revision. The version in the original language is kept in a separate meta-data field. 9\. Once all text blocks are completed, the manuscript is formatted according to the guidelines of the target journal and submitted. Usability --------- The results for the formal usability analysis revealed that users were satisfied with the speed of the application, which they considered as an important factor in the transition from a word processor writing environment to a Web writing environment. Other characteristics were positively ranked by users in the formal usability analysis (Table [1](#T1){ref-type="table"}). Two users requested an additional feature to highlight the title of the text block (Table [2](#T2){ref-type="table"}). Results from the field analysis demonstrated that first authors were satisfied with the ability to assign text blocks to other investigators with greater expertise in certain areas. Many suggestions were made regarding the ability of track co-author\'s writing performance. This additional feature was judged particularly helpful in the tracking of graduate students. In response to their concern, we have added the following measures of performance: Total number of words written on the current day, average number of words written over the last seven days, and total number of words ever written by the researcher while using Manuscript Architect. A ranking of the maximum average number of words among all users was also added to the first page of the Research Manager application, from which all manuscripts are accessed. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Satisfaction rating in formal usability analysis ::: Item construct\* Evaluation rating --------------------------------------------------------------------------------------------------------------------------- ------------------- ----------------------- MA speed is excellent Strongly disagree 0/8 Disagree 0/8 Neutral 0/8 Agree 1/8 Strongly agree 7/8 MA is extremely easy to learn Strongly disagree 0/8 Disagree 0/8 Neutral 2/8 Agree 4/8 Strongly agree 2/8 MA is extremely easy to use Strongly disagree 0/8 Disagree 0/8 Neutral 1/8 Agree 5/8 Strongly agree 2/8 It is very easy to find the writing functions (e.g., assigning text blocks, viewing previous examples) associated with MA Strongly disagree 0/8 Disagree 0/8 Neutral 1/8 Agree 5/8 Strongly agree 2/8 The navigation in MA is highly intuitive Strongly disagree 0/8 Disagree 0/8 Neutral 0/8 Agree 5/8 Strongly agree 3/8 \* MA = Manuscript Architect ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Free text recommendations and observations after formal and field usability analyses ::: Suggestions for improvement of interface ------------------------------------------------------------------------------------------------------------------------ Missing features that should be added to MA Ability of track co-author\'s writing performance Suggestions for overall improvement of the MA application Translator features were not considered to provide adequate assistance in the process of converting texts to English Use of non-scientific translators were deemed to add information noise to the scientific communication Features that you consider as an important additional value offered by MA in contrast with regular word processors Ability to assign text blocks to co-authors ::: Although we had high expectations regarding the possibility having a non-scientist translator converting text written by researchers whose first language was not the same as the target journal -- most commonly English -- the system was not efficient. The lack of efficiency resulted from (1) the difficulty in finding translators that were proficient in both the primary and target language and (2) the \"communication noise\" resulting from having a non-researcher translating research concepts. The translating tools were therefore only used when the first author was bilingual. In this situation, the system was effective since it allowed the first author and co-authors performing parallel tasks in the same manuscript. Although shared screen applications were also available during workshops, we have found that they are unnecessary. Discussion ========== To our knowledge, this is the first description of a Web application to facilitate the writing of scientific manuscripts by virtual interdisciplinary groups. Manuscript Architect simplifies the writing of scientific manuscripts by decomposing them into smaller units, thus making the writing task cognitively simpler. Although at this point Manuscript Architect has only been used within the Center for Excellence in Surgical Outcomes, the free distribution of its source code under the GNU Public License is expected to spread its use among other institutions. Much of the difficulty of writing stems from the large number of constraints that must be concomitantly satisfied. In expressing a research idea the writer must consider at least four structural levels at once: (1) overall text structure, (2) paragraph structure, (3) sentence structure (syntax), and (4) word structure (spelling). Clearly, the attempt to coordinate all these requirements is a very difficult task which, over time, leads researchers to discouragement. Another great difficulty for writers is maintaining connective flow. In other words, the relationship between ideas must be made clear while the idea must be expanded downward into paragraphs, sentences, words, and letters. Sometimes writers, and in particular the novice ones, become lost is the process of downward expansion and lose sight of the high-level relationships they originally wanted to express. Down sliding, the phenomenon of getting pulled into lower and more local levels of task processing is a very common problem in scientific writing. If a researcher decides to focus on accuracy in spelling and grammar, it will reinforce the natural tendency toward down sliding \[[@B10]-[@B18]\]. The enrichment provided by virtual peer-review is perhaps one of the most appealing areas in the field of virtual group writing. With the advent of new software tools, many of the tasks associated with scientific writing may be better distributed, in contrast with the now prevalent model of one researcher in charge of the manuscript while others simply review it for intellectual content. The socialization of this process has many consequences in terms of productivity and quality. From the perspective of peer-review, virtual group writing emphasizes the social construction of knowledge, a theoretical perspective characterized by the assertion that knowledge is created through social interaction \[[@B19]\]. Facilitating the exchange of information with other researchers will most likely improve quality, since it moves to an earlier stage a peer-review process that in the classic writing models is characteristically intensified only after the manuscript submission. Conclusion ========== Although Manuscript Architect was primarily designed for writing within interdisciplinary virtual teams, it can also be used in a variety of other circumstances such as: (1) use by single writers who would like to maintain their manuscript in an on-line environment while taking advantage of the stepwise approach of building a manuscript, and (2) educational writing workshops \[[@B20],[@B21]\]. The main features planned for future versions of Manuscript Architect include integration with other tools previously developed by our group \[[@B7]\]. First, Manuscript Architect will be linked to literature maps, a method developed for structured literature reviews. This method will enhance the accuracy of literature citations as well as allow the literature search to run in parallel with other project activities. For example, a senior researcher can work on a text block of the discussion, while a junior researcher conducts an in-depth review of references to be later used in another text block of the same discussion. Second, links will be created between Manuscript Architect and the method of layers of information. The method of layers of information allows researchers without a formal graduate degree in statistics to understand complex statistical techniques by explaining them in a hierarchical fashion. Third, Manuscript Architect will be integrated with QuestForm, a Web application designed to formulate research questions based on existing data sets. The integration with QuestForm will allow for manuscripts to receive the description of databases, the variables used to answer the research question, ICD9 and CPT codes (when applicable), and tables and graphics resulting from the statistical analysis. Finally, text maps will be included to orient writers in the co-dependency across different text blocks. For example, if the Results section of a manuscript describes a certain patient outcome, a corresponding block should exist in the Methods section describing how the data describing that outcome was collected, validated, and analyzed. Text maps will enforce the checking of accurate links among different text blocks. In conclusion, Manuscript Architect has proven to be a useful tool for the interdisciplinary collaboration among clinical researchers in academic centers. Future investigations should evaluate its role in undergraduate education as well as translational research. Availability and requirements ============================= The Manuscript Architect application is available at <http://www.ceso.duke.edu>. The Manuscript Architect application is distributed under the GNU General Public License. This license ensures that the source code can be freely distributed, modified, or even sold, as long as the source code is provided with every copy of the application. The source code for the application is available at no charge for download at <http://www.ceso.duke.edu> Abbreviations ============= VoIP: Voice of the Internet Protocol GNU: GNU\'s Not Unix ICD9: International Classification of Diseases, 9th Revision CPT: Current Procedural Terminology GPL: General Public License Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= RP was primarily responsible for the application design, performed the usability testing, and drafted the manuscript HM assisted in the application design, wrote the source code for the application, wrote the sections on software architecture, and reviewed the manuscript for intellectual content AM assisted in the application design, assisted in the usability testing, wrote portions of the usability assessment, and reviewed the manuscript for intellectual content KN, SS, and DJ assisted in the application design, assisted in the usability testing, and reviewed the manuscript for intellectual content ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Class Components. ::: ![](1472-6947-5-15-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Manuscript Architect architecture. ::: ![](1472-6947-5-15-2) ::: Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6947/5/15/prepub> Acknowledgements ================ We would like to thank Juliana and Guilherme Serzedello for article formatting	PubMed Central	2024-06-05T03:55:59.642587	2005-6-16	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180829/", "journal": "BMC Med Inform Decis Mak. 2005 Jun 16; 5:15", "authors": [ { "first": "Ricardo", "last": "Pietrobon" }, { "first": "Karen C", "last": "Nielsen" }, { "first": "Susan M", "last": "Steele" }, { "first": "Andreia P", "last": "Menezes" }, { "first": "Henrique", "last": "Martins" }, { "first": "Danny O", "last": "Jacobs" } ] }
PMC1180830	Background ========== Low back pain (LBP) is a very common ailment in the Western World and musculoskeletal disorders are leading causes of long-term sick leave \[[@B1]\]. Thus, it has a major social and economic impact on society. Although elimination of LBP is only wishful thinking, prevention of chronicity may be within the scope of reality -- with obvious gains for society. To obtain this goal, a thorough understanding of the etiology is necessary. Presently, etiological research has a broad approach to LBP, based on the bio-psycho-social model and several studies have been performed in which risks associated with physical characteristics \[[@B2]\], psychological characteristics \[[@B3]\], lifestyle factors \[[@B4]\], employment \[[@B5]-[@B8]\], social factors \[[@B9],[@B10]\] and genetic components \[[@B11],[@B12]\] were investigated. However, these studies are difficult to interpret due to the close relationship between social factors, intellectual capacity, coping strategies, education, profession etc., causing a mesh of interactions and confounding. Traditionally such studies have been carried out on working-age populations, but there is growing evidence that LBP has an earlier onset than hitherto thought. In a study of Danish twins, aged 12 to 41, the one-year prevalence of LBP was found to rise from 7% in those aged 12 to above 50% in those aged 22 and to increase only slightly thereafter to reach 56% in those at the age of 41 \[[@B13]\]. Furthermore, earlier studies have shown that the risk of LBP at age 30 is significantly increased for individuals with LBP at age 18 \[[@B14]\]. Obviously, to obtain a thorough understanding of the etiology of LBP, young populations should be investigated. The Danish Army has a system of mandatory conscription for six months of service. Usually, there is an intake in August and one in January, with most high school graduates commencing in August after finishing high school, whereas the January-intake mainly consists of those with no academic education beyond primary school. However, due to special circumstances, there was no intake in January 2000 and therefore the August-intake of that year represented the social variety of the Danish society, as reflected by education. During the first three months of service all the recruits receive the same training, housing etc. Therefore, we had a unique opportunity to investigate the development of LBP in young persons (primarily males) of different educational backgrounds, who were fit for military service, at a time when they were submitted to the same living and working conditions. This made it possible to investigate the impact of various social background factors on the occurrence of LBP, without results being confounded by differences in lifestyle and type of work. By surveying these conscripts on their first day of enrollment, we can estimate how their background influences their prevalence of LBP, reflecting the influence of their usual environment. After three months of being subjected to unusual stressors in a different environment, like performing physically demanding tasks whilst being deprived of their normal personal freedom, we can estimate how the occurrence of LBP is influenced by their social background as well as their baseline LBP-status. Thus, the present study has two objectives: 1\) To investigate possible associations between social background factors and the prevalence of LBP in young persons, both in their usual environment and in a new setting. 2\) To identify possible predictors, both social and LBP-related, for reporting LBP during military service. Methods ======= Subjects -------- During the year of 2000, 2,343 persons, aged 18--24, were enrolled as military conscripts at 15 locations across Denmark. Upon reporting for duty, the recruits had a medical examination and during the course of this, a research questionnaire was administered. The medical officer filled in the first part, which contained the army\'s own information: personal identification number, years of schooling, the score from the military\'s intelligence test, and type of work. The second part was filled in by the conscripts themselves and contained questions relating to their LBP-history and the education of their parents. After three months of basic training the conscripts were given different tasks, thus conditions were no longer the same for all and our follow-up period ended. At this time they were given a second questionnaire relating to LBP during their first three months of service. An information letter accompanied the questionnaires, explaining the purpose of the project and assuring the confidentiality of the informant. It was stressed, that the information would not be available to the army, as only staff at the Medical Research Unit in the county of Ringkøbing would have access to the information. Variables --------- ### Social background variables The following background information was collected routinely by the military on all potential recruits: • Years of school: Years of schooling from 1^st^grade till joining the military. Treated as a continuous variable. • IQ-score: Result of an intelligence test performed when examined for liability for military service. Divided into three categories with cutpoints close to the 25- and the 75-percentile (low, medium, high) and treated as a categorical variable. • Occupation: Each profession has its own code in the military. These codes were translated and dichotomized into sedentary or manual work. The questionnaires contained the following information: Parents\' education: The level of the highest educated parent. Choices were: grade school only, high school level, or \>12 years (university/other type of higher education). ### LB-related variables • LBP: Questionnaire 1: Pain or discomfort in the lower back during the past year (yes/no). Questionnaire 2: Pain or discomfort in the lower back during military service (yes/no). This was accompanied by a drawing, showing the lower back as the area between the gluteal folds and the 12^th^ribs. • Duration: Questionnaire 1: Number of days with LBP during the past year. Questionnaire 2: Number of days with LBP during military service. This was treated as a categorical variable (0 days, 1--7 days, 8--30 days, \>30 days) and as a dichotomous variable (long-lasting LBP (LLBP) = LBP\>30 days, yes/no). • Leg pain: Questionnaire 1: Leg pain radiating from the low back during the past year (yes/no). Questionnaire 2: Leg pain radiating from the low back during military service (yes/no). • Exemption: Questionnaire 2: Number of days exempted from duty due to LBP during the follow-up period. Due to the small number of exemptions, this was formed into a dichotomous variable (Exemption yes/no). The questions of one-year prevalence of LBP, duration of LBP, and leg pain were modeled on the Nordic low back pain questionnaire \[[@B15]\] of which the LBP-questions have been validated in previous studies \[[@B16],[@B17]\]. Analysis -------- The analysis is divided into two parts: 1) a cross-sectional study investigating associations between LBP and the described social variables (IQ, years of schooling, occupation and parents\' education) and 2) a three-months follow-up study investigating how LBP during military service is related to the social background factors and baseline low back related (LB-related) factors: LBP, long-lasting LBP and radiating pain to the leg. All outcome variables were analyzed seperately for the total sample, i.e. leg pain against no leg pain, regardless of LBP-status. Since the social background variables are closely related, they were all tested for interactions by means of the Mantel-Haenzel homogeneity test. All analyses were performed with Stata 7.0. Statistical significance was defined at the 5%-level. However, since multiple associations were tested, one might choose to be restrictive in the interpretations of significance. Therefore, statistical significance has been indicated in the relevant tables at the 1%-level as well. ### Cross-sectional study The baseline outcome variables were analyzed with respect to the social background variables by means of multivariate logistic regression. Following multivariate analyses, forward exclusion was applied to decide which variables to keep in the models for the different outcome variables. The variables were kept in the models if they were statistically significant or if they had a bearing on the estimates of the variables that were statistically significant in the multivariate analyses. They were included if they changed the estimates of the statistically significant variables by 25% or more, if they changed the corresponding p-values with 25% or more, or if they changed the significance status of the relevant estimates. Results are presented as odds ratios with 95% confidence intervals (CI). ### Follow-up study The same type of regression-analyses were used to analyze the follow-up outcome measures with respect to the social background variables. The most relevant baseline LB-related variable(s) was/were determined through univariate logistic regression, and a final model for predicting LB-related trouble during military service was constructed by combining social and LB-related variables. ### Post-hoc analyses Due to the low response rate at baseline, it was decided to compare the prevalence rates of the baseline LBP-variables to those of the background population. This was done by means of data from the Danish Twin Register, which previously has been shown to be representative of the general Danish population and is described in detail elsewhere \[[@B18]\]. The twins had answered a questionnaire containing the same LBP-questions in 1994. Therefore, a nested cohort of 18--22 year old males was selected for comparison to the soldiers in the present study. Finally, the consequence in relation to exemption from duty were demonstrated by calculating the reduction in intake of recruits and the corresponding reduction in exemptions, if the identified significant factors were to be used as part of the inclusion criteria for military service. In this analysis, only subjects who have answered all the relevant questions are included. Results ======= Response rate ------------- Of the originally planned enrollment of 2,343 individuals, 1,711 (73%) returned the baseline questionnaire with valid answers to the LBP-questions. The follow-up response rate is 58% (985/1711). The numbers of valid responses to the individual variables are shown in Table [1](#T1){ref-type="table"}. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Number of responses to the questionnaires. ::: Years of school. n: 1,657 ------------------------ ---------------------- ------------------- ------------------------- ------------ ----------- ------- IQ Low score Medium score Highest score n 351 682 406 1,439 Occupation Sedentary occupation Manual occupation n 840 807 1,647 Parental education Grade school only High school \>12 years of education n 474 521 461 1,456 LBP Yes No n Questionnaire 1 712 999 1,711 Questionnaire 2 345 639 985 Duration: 0 days 1--7 days 8--30 days \>30 days n Questionnaire 1 948 243 259 158 1,608 Questionnaire 2 637 140 119 49 945 Leg pain Yes No n Questionnaire 1 82 1589 1,671 Questionnaire 2 66 912 982 Exemption Yes No n 65 806 871 ::: Gender and age -------------- The study sample is predominantly male with only 4% females. The mean age is 20.57 years with a standard deviation of 2.16 years, ranging from 18 to 30 years of age. Quality of data --------------- Double entry of a 10%-subsample revealed an error-rate of less than 0.6%, which we consider satisfactory. External validity (post-hoc analysis) ------------------------------------- The one-year prevalence rates of LBP, leg pain and long-lasting LBP (LLBP) are shown in Table [2](#T2){ref-type="table"} for 18--22 year old male twins from the Danish Twin Register and for the soldiers at baseline of the present survey. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### One-year prevalence rates with 95% confidence intervals of the soldiers at baseline in the present study and in an age-matched cohort, representative of the background population (18--22 year old male twins, surveyed in 1994). ::: Cohort LBP LLBP leg pain ---------------- --------------- ------------ ------------ 2,213 twins 43% (41--45%) 7% (6--9%) 5% (4--6%) 1,711 soldiers 42% (39--44%) 7% (6--9%) 5% (4--6%) ::: The present sample of soldiers is perfectly comparable to an age- and almost sex-matched sub-sample from the Danish Twin Registry with regard to the outcome variables: one year prevalence of LBP, long-lasting LBP and leg pain. Analysis of non-responders at follow-up --------------------------------------- There was no difference in responders and non-responders at follow-up with regard to any of the social background factors or the one-year prevalence of LBP at baseline. However, non-responders had more days with LBP during the past year than the responders (mean 16.11 (SD 59.45) and mean 10.74 (SD 37.24), respectively, p = 0.000), and there was also a statistically significant difference in leg pain (p = 0.027) with 6% (5--8%) of non-responders and 4% (3--5%) of responders complaining of leg pain, indicating that the LBP in non-responders might be more severe in nature. Interactions between social background variables ------------------------------------------------ All the background variables were positively interrelated (p ≤ 0.08), but tests for homogeneity showed no interactions (p ≥ 0.72). Cross-sectional analysis ------------------------ ### Model for associations between LB-related variables at baseline and social background variables The statistically significant factors found in the multivariate analysis were parents\' education for LBP, occupation and parents\' education for long-lasting LBP, but none for leg pain. The forward exclusion method left parents\' education, sedentary job and IQ in the model for LBP and all four variables in the model for long-lasting LBP (removing years of schooling or IQ both lead to a decrease in the p-value of the effect-estimate of occupation of more than 25%). Having a sedentary job showed a reduced association with LBP. Although this was not statistically significant for simply having LBP, it was highly significant for long-lasting LBP. Having the highest educated parent at high school level lead to a statistically significant higher prevalence of LBP and long-lasting LBP whereas the findings for those with a parent with \>12 years of education were not statistically significant and no gradient across groups was detected. The results are presented in Table [3](#T3){ref-type="table"}. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Final model for associations between LBP-variables at baseline and social background variables (odds ratios with 95% confidence intervals). Significant findings are indicated with bold typing. ::: N = 1708 LBP n = 710 LLBP n = 157 ------------------------------ ------------------------- ------------------------- IQ-score low 1.00 1.00 IQ-score medium 1.09 (0.81--1.48) 1.09 (0.69--1.73) IQ-score high 1.19 (0.84--1.68) 0.95 (0.54--1.68) Sedentary occupation 0.84 (0.65--1.08) 0.55 (0.33--0.90)\* Parents grade school 1.00 1.00 Parents high school 1.39 (1.05--1.84)\* 1.67 (1.05--2.64)\* Parents \>12 years education 1.23 (0.93--1.64) 1.33 (0.82--2.17) Years of school \- 0.98 (0.89--1.08) \: p \< 0.05, \\: p \< 0.01 ::: Follow-up analysis ------------------ ### Model for associations between social background variables and LBP during military service The multivariate analysis showed no statistically significant findings for long-lasting LBP. Parents\' education was found to be statistically significant for LBP during service with an increasing gradient, i.e. higher education associated with higher prevalence of LBP. IQ was inversely related to both leg pain and exemption during service with decreasing gradients, i.e. the higher the score, the lower the risk. Years of schooling could be excluded from the model for LBP and leg pain without any bearing on the estimates, and type of occupation could be excluded from the model for exemption from duty. The exact estimates are shown in Table [4](#T4){ref-type="table"}. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Final model for associations between social background variables and LBP during military service (odds ratios with 95% confidence intervals). Statistically significant findings are indicated with bold typing. ::: N = 982 LBP n = 345 Leg pain n = 66 Exemption n = 65 ------------------------------ --------------------------- ------------------------- ------------------------- IQ-score low 1.00 1.00 1.00 IQ-score medium 1.02 (0.68--1.52) 0.72 (0.37--1.42) 0.56 (0.27--1.17) IQ-score high 0.99 (0.64--1.53) 0.33 (0.13--0.88)\ 0.26 (0.09--0.79)\* Parents grade school 1.00 1.00 1.00 Parents high school 1.58 (1.06--2.34)\* 1.25 (0.62--2.54) 1.17 (0.54--2.55) Parents \>12 years education 1.81 (1.20--2.72)\\** 1.19 (0.56--2.49) 1.22 (0.54--2.76) Sedentary occupation 0.84 (0.59--1.20) 0.69 (0.36--1.33) \- Years of school \- \- 0.89 (0.71--1.10) \: p \< 0.05, \\: p \< 0.01 ::: ### Model for associations between LB-related trouble at baseline and LB-related trouble during military service All the LB-related variables were significantly associated with all the outcome variables. We decided to use number of days with LBP during the past year rather than LBP at all during past year to obtain more detailed analyses. There is a high correlation between baseline leg pain and the outcome variables (73 of the 82 (89%) persons with leg pain also had LBP), and adding leg pain to the models does not alter the estimates significantly. With regard to leg pain at follow-up, baseline leg pain is the most important factor. The results of the bivariate logistic regression analyses are shown in Table [5](#T5){ref-type="table"}. ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Associations between baseline LB-related variables and LBP during military service (odds ratios with 95% confidence intervals) as assessed by univariate logistic regression analyses. Statistically significant findings are indicated with bold typing. ::: N = 982 LBP n = 345 Leg pain n = 78 LLBP n = 56 Exemption^1^n = 74 ----------------------------------- ---------------------------- ---------------------------- ---------------------------- ---------------------------- Baseline LBP 3.52\\* (2.68--4.63) 1.71\\ (1.15--2.54) 1.58\* (0.01--1.28) 2.89\* (1.70--4.89) Baseline leg pain 2.22\* (1.15--4.30) 8.78\\ (4.23--8.21) 4.92\\ (2.03--11.89) 3.61\\ (1.50--8.69) 0 days with LBP previous year 1.00 1.00 1.00 1.00 1--7 days with LBP previous year 2.04\\ (1.39--2.99) 0.90 (0.36--2.24) 1.11 (0.44--2.84) 1.96 (0.95--4.06) 8--30 days with LBP previous year 5.42\\ (3.63--8.08) 2.57\\ (1.28--5.14) 2.47\* (1.14--5.32) 1.44 (0.63--3.30) \>30 days with LBP previous year 6.45\\ (3.86--10.79) 5.51\\ (2.73--11.12) 4.80\\ (2.13--10.84) 6.13\\ (3.05--12.50)** \: p \< 0.05, \\:p \< 0.01 ::: ### Final models for predicting trouble related to LBP during military service combining social background variables and LB-related baseline-variables When the models for associations between social background and LB-related trouble during military service were combined with LB-related baseline variables, the following models could be derived: ### LBP during military service The model for the association between social background variables for LBP during service included IQ and parents\' education. When this was combined with the number of days with LBP the year prior to service, parents\' education remained statistically significant with an increase in education resulting in an increase in odds ratio. Number of days with LBP showed the strongest correlation with odds ratios above 6. IQ-level could not be excluded from the model without altering the significance level of parents\' education ### Long-lasting LBP during military service None of the social background variables had significant influence on the development of long-lasting LBP, thus only the LB-related baseline variable, duration of LBP, was included. The effect of LBP-duration was statistically significant, demonstrating a positive association (OR = 4.8 for LBP\>30 days at baseline). ### Leg pain related to LBP during military service IQ, parents\' education and type of occupation were included in the model for leg pain during service. Adding leg pain during the year prior to service and number of days with LBP during the same period, leave the same variables as significant, but excluding parents\' education from the model would change the significance of both IQ and duration of LBP. A high IQ-level showed a negative association with leg pain during service, whereas long duration of LBP and previous leg pain both showed positive associations (OR = 3.3 and 3.1, respectively). ### Exemption from duty due to LBP during military service Including the relevant factors in the model (IQ, parents\' education, years of schooling and number of days with LBP), left IQ and number of days with LBP to be significant. However, once more, excluding parents\' education or years of schooling, changed the significance level for IQ and it was therefore kept in the model. A high IQ lowered the risk of exemption (OR = 0.2) whereas long-lasting LBP the previous year increased the risk (OR = 5.9). The results of the final models for all outcome measures are shown in Table [6](#T6){ref-type="table"}. ::: {#T6 .table-wrap} Table 6 ::: {.caption} ###### The final models for outcome measures during service, combining social and physical factors (odds ratios with 95% confidence intervals). Statistically significant findings are indicated with bold typing. ::: N = 982 LBP n = 345 LLBP n = 56 Leg pain n = 66 Exemption^1^n = 65 ----------------------------------- ---------------------------- ---------------------------- --------------------------- ---------------------------- IQ-score low 1.00 \- 1.00 1.00 IQ-score medium 1.14 (0.73--1.79) \- 0.75 (0.35--1.61) 0.56 (0.25--1.25) IQ-score high 1.00 (0.61--1.63) \- 0.24 (0.08--0.74)\\* 0.24 (0.08--0.76)\* Parents grade school 1.00 \- 1.00 1.00 Parents high school 1.56 (1.01--2.42)\* \- 1.05 (0.48--2.27) 1.14 (0.48--2.71) Parents \>12 years education 1.90 (1.21--2.99)\\ \- 1.06 (0.47--2.38) 1.40 (0.59--3.35) Sedentary occupation \- \- 0.87 (0.42--1.79) \- Years of school \- \- \- 0.91 (0.72--1.14) Baseline Leg pain \- \- 3.14 (1.13--9.85)\* \- 0 days with LBP previous year 1.00 1.00 1.00 1.00 1--7 days with LBP previous year 2.31 (1.49--3.59)\\ 1.11 (0.44--2.84) 0.79 (0.28--2.22) 2.33 (0.94--5.79) 8--30 days with LBP previous year 6.09 (3.78--9.84)\\ 2.47 (1.14--5.32)\* 2.52 (1.28--6.20)\\ 1.78 (0.64--4.94) \>30 days with LBP previous year 6.18 (3.42--11.17)\\ 4.80 (2.13--10.84)\\ 3.28 (1.31--8.25)\\ 5.93 (2.37--14.83)\\** \: p \< 0.05, \\:p \< 0.01 ::: Consequences ------------ For practical purposes, in the process of selecting military recruits for army, the outcome of interest is exemption from duty. For this outcome, the most interesting factors, of the ones investigated in this study, are the number of days with LBP during the year prior to service and the IQ-level. Table [7](#T7){ref-type="table"} presents an estimation of the reduction in the number of soldiers exempted from duty and the corresponding reduction in intake if various combinations of those two factors were used as inclusion criteria. The practical significance must then be judged in the actual situation, for example the number of available potential recruits. ::: {#T7 .table-wrap} Table 7 ::: {.caption} ###### Consequence in reduction of exemptions and intake-size when excluding subgroups of potential recruits from service, on basis of the two factors found to influence the rate of exemption from duty significantly (IQ-score and duration of prior LBP). ::: Included* \# exempted from duty Total intake ---------------------------------------- --------------------------- ------------------ Full sample 45 (100%) 664 (100%) IQ-score=high 8 (18%) 187 (28%) IQ-score=high or medium 29 (64%) 512 (77%) LBP-days\<30 34 (76%) 607 (91%) IQ-score=high & LBP-days\<30 6 (13%) 169 (25%) IQ-score=high or medium & LBP-days\<30 21 (47%) 467 (70%) ::: Discussion ========== The major advantage of our cohort is that people of similar age from backgrounds reflecting the variety of society are all exposed to the same living and working conditions for the same period of time when doing their military service. Obviously, only subjects fit for military service are included, thus excluding the most intellectually and physically disadvantaged. It has, however, previously been established, that the prevalence of LBP in Danish conscripts is similar to the rest of the Danish population of that age \[[@B19]\]. This we confirmed, as the present cohort does have similar prevalence rates of the LBP-variables as a comparable background population. The major problem of our study is the low response rate (58%). There might have been whole platoons/units (rather than individuals) that did not return the questionnaires. However, we were unable to confirm this, as all the questionnaires were collected in one box upon arrival at the research centre and the place of duty was not stated in the questionnaire. Furthermore, the personal identification number of each soldier was used to combine the first and second questionnaires. A large number of soldiers returned the second questionnaire without stating this number, therefore leaving us with a large number of un-identifiable, although otherwise completed, questionnaires. Thus, most of this is likely to reflect an administrative error (whole platoons not returning the questionnaires) and poor design of the questionnaires (the first and second questionnaire could not be linked) and therefore one would not expect any disease-specific bias as a result of non-response. This is also confirmed by the similarity of prevalence rates between our study sample and the sample from the Danish Twin Register. Anyway, the problem of non-responders having more severe LBP at baseline still remains. This could be related to termination of service prior to follow-up due to LBP of some individuals with a high level of baseline-LBP. Unfortunately, we did not have access to such information. In the protocol, the army staff was asked to administer the second questionnaire to the recruit in case of rejection during service, but this did not happen in any instance. This bias does not directly influence the results relating to the social background variables, but it might weaken the results relating to baseline LBP-status with associations actually being stronger than our findings indicates, if the rejected recruits were the ones with the more severe LBP-symptoms at baseline. As for the influence of social background factors on the presence of LBP at baseline in our population, we found a statistically significant increase in odds ratios if parents belonged to educational group 2 (high school-level). This is an illogical finding that might be an effect of multiple testing and since there is no gradient from the lowest to the highest group, the practical significance of this result is limited. However, when analyzing the occurrence of LBP and related trouble during military service, parents\' education does show an increasing gradient for LBP lending support to the credibility of the baseline-findings. A previous study found no association between parents\' education and back pain in Danish adolescents \[[@B20]\]. The most important factor for LBP at baseline was found to be type of occupation. This is in line with a previous study, showing sedentary occupations to have a \"protective effect\" on LBP \[[@B21]\]. Interestingly, there was no effect of pre-service occupation on LBP and related troubles during military service -- those with sedentary occupations were as likely as the rest to develop LBP, when they were exposed to the same stress. This implies that the association with a sedentary occupation is direct, rather than a proxy of intelligence level, social class or other similar confounders/interactions. Intelligence level, as measured by the army\'s IQ-test, did not influence any of the LBP-variables at baseline, and did not influence the occurrence of LBP during service. A high test-result did, however, limit the occurrence of leg pain and exemption from duty. Both sedentary occupation and parents\' education are strongly associated with intelligence level (p = 0.000 and p = 0.008, respectively). Thus, by taking our results a step further, it could be speculated that the group with above-average intelligence, well-educated parents and a sedentary occupation has a low occurrence of LBP at baseline, an increased risk of LBP during the physical demands of service, but seem to suffer less from the consequences (leg pain and exemption from duty). This might be a sign of better coping strategies in this group. Coping strategies as they relate to dealing with chronic pain has been a focus of research in recent years \[[@B22]-[@B24]\], but coping abilities as part of the etiology of back pain has only received limited attention yet. In a recent study of a general population, high levels of passive coping were found to be associated with disabling neck and back pain \[[@B25]\]. When the LBP-variables at baseline are added to the model for LBP-related trouble during service, the duration of LBP the previous year is the important variable. LBP in itself does not predict LBP during service but a longer duration of LBP at baseline increased the risk of long-lasting LBP, leg pain and exemption from duty during the study period. This might indicate that short-lasting LBP is more co-incidental and a normal life experience. This is consistent with the literature, where the duration of LBP is considered to be an important prognostic factor \[[@B26]-[@B29]\]. Furthermore, we showed that leg pain at baseline is a strong predictor for continued leg pain. As a screening tool to predict which recruits will develop LBP during military service, models containing the investigated variables are not sufficiently accurate. If used to exclude vulnerable subjects from military service, many would be excluded unnecessarily. Key points ========== • Having a sedentary occupation is negatively associated with long-lasting LBP and leg pain in this population, irrespective of intelligence level and education. • Having parents with a high education may increase the risk of developing LBP during military service, but not of its consequences. • A high IQ seems to protect against leg pain and exemption from duty during military service. • Long-lasting LBP at baseline increases the risk of long-lasting LBP, leg pain and exemption during the course of duty. • The predictive value of the investigated factors is too low to be used as a screening tool for the military. Conclusion ========== With regard to the social variables (IQ, occupation, and parents\' education), type of occupation is the most important factor at baseline with a sedentary occupation having a negative association with LBP. This protective effect disappears, when the person actually enrolls and becomes more physically active. Then, a higher level of parents\' education seems to predispose for simple LBP, whereas a high intelligence level is a protective factor for the development of leg pain and exemption from duty. With regard to the LBP-related variables (LBP at all, long-lasting LBP, and leg pain), longer duration of LBP during the previous year increases the risk of all LBP-related problems during military service. Finally, the presence of leg pain at baseline predicts leg pain during service. Our findings provide important information in relation to etiology, but a suitable screening instrument will require further refinement. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= LH participated in the design and coordination of the study, double-entered a sub-sample of data, performed the statistical analyses and drafted the manuscript. KL participated in the design and coordination of the study and confirmed the statistical analyses. FW provided contact to the army practitioners and participated in the coordination of the study. CL-Y participated in the design of the study and in drafting the final manuscript. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2474/6/25/prepub> Acknowledgements ================ We would like to thank all the soldiers, who participated in the surveys, the medical officers and other army personnel, who helped to fill in, collect and mail the questionnaires. Furthermore, we would like to acknowledge Heidi Hensgen, The Medical Research unit in Ringkjøbing Amt, and Ulla Hansen at Backcenter Funen, the two secretaries, who entered all the data. Finally, The Foundation for Chiropractic Research and Postgraduate Education is gratefully appreciated for financing the first author.	PubMed Central	2024-06-05T03:55:59.645109	2005-5-26	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180830/", "journal": "BMC Musculoskelet Disord. 2005 May 26; 6:25", "authors": [ { "first": "Lise", "last": "Hestbaek" }, { "first": "Kristian", "last": "Larsen" }, { "first": "Flemming", "last": "Weidick" }, { "first": "Charlotte", "last": "Leboeuf-Yde" } ] }
PMC1180831	Background ========== Diabetic nephropathy is the most common cause of end-stage kidney disease (ESKD) in the United States and worldwide. Most diabetic patients with ESKD have type 2 diabetes. Since only a minority of type 2 diabetic patients develop kidney disease, predisposing factors for development of the disease are operative. In addition, once clinical kidney disease is evident, the rate of decline of glomerular filtration rate (GFR) is highly variable, ranging from 2 to 20 ml min-1 yr-1 \[[@B1]\]. The reasons for these differences in the rate of disease progression are multifactorial, including both non-modifiable and modifiable factors \[[@B2],[@B3]\]. Blood pressure control is known to be important in preventing adverse cardiovascular and renal outcomes in diabetic patients with hypertension \[[@B4]\]. However, it is not clear whether or not blood pressure is an important predictor of GFR decline in diabetic patients with CKD in whom blood pressure is controlled. The purpose of the present study was to determine the factors independently associated with chronic kidney disease (CKD) progression assessed by rate of decline of GFR in a cohort of male predominantly elderly veteran patients with type 2 diabetes. In this population, in whom BP was generally well controlled, hypoalbuminemia, black race, and degree of proteinuria, but not blood pressure, were associated with disease progression. Methods ======= Subjects -------- A search of the computerized patient record system at Hines VA Medical Center, Hines, Illinois was performed to find all patients seen in the outpatient clinics between January 1993 and September 2002 with type 2 DM and clinical evidence of CKD. Potential subjects were included if they had all of the following: a) type 2 DM \[clinical diagnosis and/or glycated hemoglobin \> 6.5%\]; b) chronic kidney insufficiency, defined by a persistent elevation of serum creatinine (SCr) level; at least 3 serum creatinine values of \> 124 μmol/L with at least a six-month interval between the first and last serum creatinine value were required for inclusion; c) proteinuria \[positive urine dipstick or urine protein \> 0.15 g/d or urine albumin/creatinine ratio ≥ 30 mg/g\]. GFR was calculated using the simplified MDRD formula \[GFR (ml min-1 1.73 m^2^-1) = 186 × (SCr)^-1.154^× (age)^-0.203^× (0.742 if female) × (1.210 if black)\]\[[@B5]\]. The serum creatinine concentration was not calibrated to MDRD laboratory creatinine values. Patients known to have another cause of kidney disease (e.g., polycystic kidney disease, ischemic nephropathy, or biopsy-proven non-diabetic kidney disease) were excluded. In order to determine rate of CKD progression, individual GFR values were calculated from every serum creatinine value between the time when the serum creatinine was first noted to be \> 124 μmol/L (baseline value) and either the time of the last available serum creatinine or the commencement of kidney replacement therapy. The rate of GFR decline (slope) (ml min-1 1.73 m^2^-1 yr-1) for each patient during the study period was then determined by linear regression analysis using all calculated GFR values. Decline in GFR was represented as a positive value. The mean duration of the study period was 36 months (range 7--149). The mean number of GFR values utilized for slope determination was 14 (range 3--69). Co-variates ----------- Information on age, race, weight, height, use of angiotensin converting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs), blood pressure, glycated hemoglobin, serum cholesterol, hemoglobin, serum albumin, and urine protein excretion was obtained from the computerized patient record system. Age, weight (kg) and height (m) were defined as the recorded values at the time of the first laboratory creatinine measurement \> 124 μmol/L. Body mass index (BMI) (kg/m^2^) was then calculated using the weight and height values. Race was self-reported. Blood pressure was measured at each clinic visit by a medical technician or a nurse (generally every 3--6 months) in the seated position using a mercury sphygmomanometer. Glycated hemoglobin was measured by high performance liquid chromatography. Hemoglobin values were measured by Coulter counter. Serum values of creatinine, cholesterol, and albumin were measured by automated analyzer. Urinary protein was measured by the pyrogallol red-molybdate method and urinary albumin by immunoturbidemetry. The degree of proteinuria was classified as mild, moderate, or heavy. Mild proteinuria was defined as dipstick trace-0.3 g/L and/or protein 0.15--0.5 g/d and/or spot urine albumin/creatinine ratio of 30--300 mg/g; moderate proteinuria was defined as dipstick 1.0 g/L and/or protein 0.5--3.0 g/d and/or spot urine albumin/creatinine ratio of 300--2000 mg/g; and heavy proteinuria was defined as dipstick 3.0 g/L and/or protein \> 3.0 g/d and/or spot urine albumin/creatinine ratio of \> 2000 mg/g. Statistical analysis -------------------- Mean values of systolic blood pressure, diastolic blood pressure, glycated hemoglobin, and serum cholesterol during the study period were computed for each patient and utilized for analysis. Because degree of proteinuria was a categorical variable, maximum degree of proteinuria recorded during the study period was utilized. Since hemoglobin and serum albumin values were expected to decline as kidney disease progressed (see Discussion), initial hemoglobin and serum albumin values (i.e., values obtained closest in time to the beginning of the study period for each patient) were used. Univariate analyses were performed using t-test for continuous variables and Chi-square analysis for categorical variables. Multivariate linear regression analyses using rate of decline (slope) of GFR as the dependent variable were then performed in both age-adjusted and non-age-adjusted fashion using an automated backwards elimination procedure to delete factors with a P-value \> 0.15 from the model one at a time. Multivariate analyses were performed using all variables as well as using only those variables found to be significantly associated with progression on univariate analysis. Data are expressed as mean ± SD. All P values were two sided and a P value of \<0.05 was used to indicate statistical significance. Statistical analyses were performed using Systat software (SPSS, Chicago, IL). This study complied with the recommendations of the Declaration of Helsinki and was approved by the local Institutional Review Board. Results ======= Demographic and clinical variables in the 343 patients who met study criteria are given in Table [1](#T1){ref-type="table"}. The mean study period was 36 months (range 7--149). The mean age was 69 (range 43--92) with 73% of patients aged 65 or older. The majority of patients (77%) were Caucasian. Mean values for blood pressure, glycated hemoglobin, and serum cholesterol during the study period were 138/72 mmHg, 8.1%, and 4.8 mmol/L, respectively. The majority of patients (71%) were treated with ACE inhibitors or ARBs. Mean rate of decline of GFR during the study period was 4.5 ml min-1 1.73 m^2^-1 yr-1 (range -14 to +32). The variables most highly associated with progression on univariate analysis were lower initial serum albumin and greater degree of proteinuria (p \< 0.001), followed by younger age and black race (p = 0.002), higher glycated hemoglobin (p = 0.007), higher mean diastolic blood pressure (p = 0.012), lower initial hemoglobin (p = 0.014), and higher systolic blood pressure (p = 0.048) (Table [2](#T2){ref-type="table"}). Figures [1](#F1){ref-type="fig"} and [2](#F2){ref-type="fig"} depict the association between decline in GFR and degree of proteinuria and serum albumin levels, respectively. Patients with greater degrees of proteinuria not unexpectedly had higher rates of disease progression. In addition, the most rapid disease progression was seen in patients with the most profound hypoalbuminemia (serum albumin level \< 25 g/L). Rate of progression was only approximately 3 ml min-1 1.73 m2-1 yr-1 with no relationship between serum albumin and progression once serum albumin levels were \> 35 g/L. 138 patients (40%) had systolic hypertension (BP = 140 mmHg systolic). There was no association between systolic blood pressure and GFR decline when systolic BP was \< 140 mmHg. However, above that level, rate of GFR decline increased with increasing systolic BP (Figure [3](#F3){ref-type="fig"}). Only 49 patients (14%) had diastolic BP readings of = 80 mmHg. The relationship between diastolic BP and GFR decline was U-shaped, with the lowest rate of progression in the 75--79 mmHg range (Figure [4](#F4){ref-type="fig"}). The following independent variables were considered for selection in the stepwise linear regression, using backwards selection: age, race, systolic BP, diastolic BP, glycated hemoglobin, degree of proteinuria, initial hemoglobin, and initial serum albumin level. Because of the U-shaped relationship between diastolic BP and GFR decline, (diastolic BP)^2^was also evaluated. When both diastolic BP and (diastolic BP)^2^were entered into the regression equation, neither was significantly associated with BP decline (p = 0.943 for diastolic BP and p = 0.935 for (diastolic BP)^2^). In addition, systolic blood pressure and glycated hemoglobin were not independently associated with progression in this analysis. Lower initial serum albumin, black race, and degree of proteinuria were the only variables independently associated with progression at the p \< 0.05 level (Table [3](#T3){ref-type="table"}). Similar results were seen using a non-age-adjusted model (Table [3](#T3){ref-type="table"}) and when all variables regardless of their association with progression were entered into the regression equations. As regression diagnostics for data from two of the patients indicated high leverage and one was an outlier (with a Studentized residual of 5.071) in the linear regression model, further analyses were performed excluding these cases. The results were not changed after exclusion of these cases, with initial serum albumin (p = 0.001), proteinuria (p = 0.002) and black race (p = 0.003) the only significant variables independently associated with GFR decline. Discussion ========== The most important finding of this study is that the potentially modifiable factors hypoalbuminemia and proteinuria were strongly and independently associated with kidney disease progression in a mostly elderly male type 2 diabetic population with generally well-controlled blood pressure. As shown in Figure [4](#F4){ref-type="fig"}, there was no effect of systolic BP on progression with systolic BPs \< 140 mmHg, and diastolic BP had a U-shaped relationship with GFR decline (Figure [4](#F4){ref-type="fig"}). Therefore systolic BP and both diastolic BP and (diastolic BP)^2^were examined in the multivariate regression analyses. In these analyses, there was no independent association between systolic BP or either diastolic BP or (diastolic BP)^2^and rate of GFR decline in this population. Blood pressure control is known to be important in preventing adverse cardiovascular and renal outcomes in diabetic patients \[[@B4]\]. However, prospective interventional studies demonstrating slowing of kidney disease progression by blood pressure control were mostly performed in type 1 diabetic patients in whom blood pressure was generally poorly controlled by current standards \[[@B6]-[@B12]\] (see Table [4](#T4){ref-type="table"}). More recent studies in type 1 and type 2 diabetic patients designed to evaluate the effects of blood pressure control on progression have found lesser effects, either because they were underpowered \[[@B13]\] or possibly because of lower pre-treatment and on-treatment blood pressures \[[@B14],[@B15]\]. In support of the latter possibility, in the Reduction of Endpoints in NIDDM with the Angiotensin II Antagonist Losartan (RENAAL) trial, proteinuria, increased serum creatinine, hypoalbuminemia, and anemia at baseline but not baseline blood pressure were predictors of progression \[[@B16]\]. In a recent analysis of RENAAL data, last systolic (but not diastolic) BP prior to end points predicted a higher risk for progressive nephropathy; however, this effect was only seen when systolic BP was \> 140 mmHg \[[@B17]\]. In a multivariate analysis of data from the Irbesartan Diabetic Nephropathy Trial (IDNT), another large clinical trial in type 2 diabetic patients, hypoalbuminemia, increased serum creatinine, albuminuria, decreased hemoglobin, and increased blood pressure were predictors in that order of importance \[[@B18]\]. In the ramipril trial of Lewis et al. there was no significant difference in rate of decline in GFR between two groups of type 1 diabetics with mean arterial blood pressures of 98 vs. 92 mmHg, respectively \[[@B14]\]. Similarly, in the ABCD trial, a type 2 diabetes trial, there was no benefit of decreasing diastolic blood pressure from 80--89 mmHg to 75 mmHg (difference in actual BPs 138/86 vs. 132/78, or mean arterial blood pressures of 103 vs. 96 mmHg) \[[@B15]\]. A recent observational study from Japan investigated factors affecting progression of kidney failure in type 2 diabetic patients \[[@B2]\]. These investigators found that lack of insulin therapy, hypoalbuminemia, higher mean blood pressure, and lower hemoglobin at baseline were independent predictors of kidney failure. However, this study involved only 85 patients and a relatively short survey time (14 months), and disease progression was assessed by time to doubling of serum creatinine. Moreover, the mean arterial blood pressure at baseline was 105 mmHg, which is higher than that seen in most recent studies (mean blood pressures during the survey period were not given). In a larger analysis of 227 Caucasian patients with diabetic nephropathy from the Steno Diabetes Center in Denmark, in whom the mean follow-up time was 6.5 years, rate of decline of GFR was predicted by higher values of baseline and mean albuminuria, systolic blood pressure, hemoglobin A1c, and degree of diabetic retinopathy, as well as by baseline GFR, age, lower hemoglobin values, and smoking status during follow-up \[[@B3]\]. In this study, two-thirds of the patients had mean systolic BPs during follow-up of \> 149 mmHg. In contrast, in our observational study, the mean BP was 138/72 mmHg (mean arterial blood pressure 94 mmHg) during the study (follow-up) period. This level of BP control was substantially better than that achieved in the Danish study and was also better than that obtained (i.e., 146/76 mmHg) in a recent clinical trial in which a stepped-care algorithm designed to meet American Diabetes Association blood pressure goals of \< 130/80 was employed \[[@B19]\]. Thus, taken together, the findings of interventional and observational studies suggest that BP control becomes less important in delaying progression when BP is not very elevated. The results of our study indicate that hypoalbuminemia independently predicted GFR decline. In a study investigating the relationship between blood pressure and diabetic kidney disease progression, Dillon \[[@B20]\] found that mean serum albumin was associated with decline in GFR. However, since serum albumin could decrease as kidney disease progresses as a consequence of anorexia and malnutrition, this association could be attributed to reverse causality. Using a similar approach to that employed in our study, Ueda et al. \[[@B2]\] reported that baseline serum albumin was a predictor of the development of ESKD in Japanese patients with type 2 diabetes. This study used reciprocal creatinine and/or the development of ESKD as endpoints rather than GFR decline; moreover, the population was quite different from the predominantly Caucasian U.S. male veterans examined in our study. Nevertheless, these data, in conjunction with the RENAAL and IDNT data, in which hypoalbuminemia was also an independent predictor of progression in type 2 diabetes, support the relationship between hypoalbuminemia and GFR decline. Why would hypoalbuminemia be predictive of kidney disease progression? One possibility is that the serum albumin level reflects the degree of proteinuria, a known risk factor for progression in most kidney diseases, including diabetes \[[@B3],[@B21]-[@B27]\]. However, this is not the sole explanation for the association between serum albumin level and disease progression, because hypoalbuminemia is significantly associated with GFR decline after adjustment for the effects of proteinuria. Since many patients with CKD have evidence of chronic inflammation, as reflected by elevated C-reactive protein (CRP) levels, anemia of chronic disease, and elevated D-dimer and vonWillebrand factor levels \[[@B28]-[@B30]\], it is possible that hypoalbuminemia may reflect an inflammatory state, which may increase kidney disease progression \[[@B30]\], possibly by causing oxidative stress \[[@B31]\]. In addition, decreased GFR may result in accumulation of proinflammatory compounds (such as advanced glycated end-products and oxidation products) with resultant hypoalbuminemia \[[@B32]\] and kidney injury \[[@B33]\]. The hypothesis that low serum albumin reflects an inflammatory state, which contributes to progression of kidney disease is supported by recent studies in both non-diabetic \[[@B34],[@B35]\] and diabetic \[[@B36]\] populations. As has been done by other investigators \[[@B2],[@B16],[@B18]\], we utilized initial serum albumin levels in our analyses rather than mean values during the treatment period. Of note, when mean serum albumin values were utilized, there was also a strong association between serum albumin and decline in GFR (data not shown). However, we believed that this could be due to a \"reverse\" association, as progression of renal failure could result in decreased protein intake, malnutrition, and a decrease in serum albumin level. We also chose to use initial values for hemoglobin rather than mean values during the treatment period. This was done because of the well known direct association between anemia and lower GFR due to erythropoietin deficiency. Mean values of other continuous variables (such as blood pressure) were utilized in order to better capture treatment effects during the observation period and to enhance the precision of the analysis by utilizing many measurements of each variable. We included patients who had the diagnosis of diabetes mellitus and/or mean glycated hemoglobin values \> 6.5% in our analysis. We did not utilize blood glucose values for study inclusion or for analysis since it was not always possible to determine if a sample was a fasting or random sample. We did not find an association between blood glucose control (as reflected by glycated hemoglobin) and progression in our study. The role of hyperglycemia in progression of nephropathy in type 2 diabetes is not clear, in large part because many studies have lacked statistical power. However, in post-hoc analyses of two large clinical trials, RENAAL \[[@B16]\] and IDNT \[[@B18]\], baseline glycated hemoglobin was not an important predictor. However, in their observational study, Rossing et al. \[[@B3]\] recently reported that mean glycated hemoglobin was associated with GFR decline. The lack of association between serum cholesterol and progression in our study is consistent with these other studies \[[@B3],[@B16],[@B18]\]. The observational nature of our study does have limitations. For instance, there was no control over the type of medication used for hypertension, nor any control over diabetes management or dietary intake. We utilized a formula to estimate rate of GFR decline instead of performing direct clearance measurements. The formula utilized, namely the four-variable MDRD estimate, was developed from a largely non-diabetic patient population. Moreover, this formula loses accuracy when the GFR is \> 60 ml min-1 1.73 m^2^-1 and/or if the serum creatinine concentration is not calibrated to MDRD laboratory creatinine values \[[@B37]\]. However, our analysis is based on rate of decline of GFR (slope analysis) and not absolute values of GFR and most patients had baseline GFR values of \< 60 ml min-1 1.73 m^2^-1. Moreover, the National Kidney Foundation currently recommends using this prediction equation to calculate GFR and monitor progression of kidney disease \[[@B5]\]. Indeed, using rate of GFR decline as our outcome variable has advantages over the doubling of serum creatinine utilized in many prospective clinical trials. This is because doubling of serum creatinine reflects a progressively smaller absolute GFR decline as the serum creatinine increases. Finally, the nature of the kidney disease in our mostly elderly diabetic patients probably represented a mix of diabetic glomerulosclerosis, hypertensive nephrosclerosis, and, in some cases, possibly undiagnosed ischemic nephropathy, either alone or in combination. We did not analyze data on retinopathy since it was not always available in our computerized database. We also did not analyze the relationship between kidney disease progression and cardiovascular disease. Despite these limitations, we believe that our findings are relevant to usual clinical practice. Conclusion ========== The results of our study may have important clinical implications. Although blood pressure control is undeniably important in delaying progression of kidney disease in hypertensive diabetic patients, attention to other potentially remediable factors may be more important in delaying progression once BP is effectively treated. Antiproteinuric agents (such as ACE inhibitors and ARBs) may exert beneficial actions beyond those due to BP control. In addition, the finding that hypoalbuminemia is independently associated with progression of CKD due to type 2 diabetes after adjustment for the effects of proteinuria suggests the possibility that inflammation (and/or malnutrition) may be a progression promoter. In particular, further understanding of the role of inflammation in CKD may lead to therapies which favorably affect kidney disease progression in type 2 diabetic patients. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= DJL conceived and designed the study, supervised data collection, carried out the statistical analyses, and prepared the manuscript. HJK assisted in statistical analyses and manuscript preparation. TMD, MPC, and MAI collected data and prepared the spreadsheet for data analysis. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-2369/6/8/prepub> Acknowledgements ================ The authors would like to thank John Rotta and Guichan Cao for their assistance with informatics and statistical analysis. Domenic Reda, Ph.D., Acting Director of the Hines Cooperative Studies Program Coordinating Center, gave helpful advice on presentation of the results of this study. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Relationship between degree of proteinuria and rate of GFR decline (1 = mild proteinuria; 2 = moderate proteinuria; 3= heavy proteinuria; see text for definitions) ::: ![](1471-2369-6-8-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Relationship between initial serum albumin level and rate of GFR decline. ::: ![](1471-2369-6-8-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Relationship between mean systolic blood pressure during the study period and rate of GFR decline. ::: ![](1471-2369-6-8-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Relationship between mean diastolic blood pressure during the study period and rate of GFR decline. ::: ![](1471-2369-6-8-4) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Demographic and clinical variables in 343 study patients ::: -------------------------------------- -------------------------------------------- Age (years) 69 ± 9 (43--92) Race 80 black; 263 Caucasian Body mass index (kg/m2) 30 ± 5 (16--50) Initial serum creatinine (μmol/L) 150 ± 35 (124--389) Initial GFR (ml/min/1.73 m^2^) 10 (13--77) Degree of proteinuria (see text) Group 1= 118; Group 2 = 100; Group 3 = 125 Mean systolic blood pressure (mmHg) 138 ± 14 (96--189) Mean diastolic blood pressure (mmHg) 72 ± 8 (49--97) Treatment with ACEI/ARB No = 99; Yes = 244 Mean serum cholesterol (mmol/L) 4.8 ± 1.0 (2.3--8.9) Mean glycated hemoglobin (%) 8.1 ± 1.6 (5.5--17) Initial hemoglobin (g/L) 128 ± 19 (67--179) Initial serum albumin (g/L) 34 ± 6 (17--53) -------------------------------------- -------------------------------------------- ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Univariate association of demographic and clinical variables with decline in GFR ::: Variable r p ------------------------------- -------- --------- Age -0.192 0.002 Race \* 0.002 Body mass index -0.010 0.859 Initial serum creatinine -0.137 0.055 Degree of proteinuria \* \<0.001 Mean systolic blood pressure 0.107 0.048 Mean diastolic blood pressure 0.136 0.012 Treatment with ACEI/ARB 0.039 0.482 Mean serum cholesterol 0.014 0.797 Mean glycated hemoglobin 0.147 0.007 Initial hemoglobin -0.136 0.014 Initial serum albumin -0.295 \<0.001 \categorical independent variables ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Variables independently associated with rate of GFR decline in multivariate linear regression models ::: Age-adjusted:* ----------------------- --------- --------- Variable F ratio P value Initial serum albumin 14.5 \<0.001 Black race 13.2 \<0.001 Degree of proteinuria 10.2 0.002 Non-age-adjusted: Variable F ratio P value Initial serum albumin 15.6 \<0.001 Black race 14.4 \<0.001 Degree of proteinuria 12.8 \<0.001 ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Summary of Interventional Studies in Diabetic Patients Examining Effect of Blood Pressure Control on Kidney Disease Progression ::: -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- AUTHOR STUDY POPULATION STUDY SIZE/ DURATION RESULTS ---------------------------------------------------- --------------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------ ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Mogensen et al. Br Med J 1982 \[6\] Type I DM\ N = 5\ BP decreased to 144/95 mmHg with treatment\ Mean BP 162/103\ 73 month treatment period compared to previous 20--31 month untreated period Rate of decline of GFR decreased from 14.8 to 5.9 ml min-1 yr-1 in treated period Urinary albumin 3.9 g/d\ GFR 86 ml min-1 1.73 m^2^-1 Parving et al. Lancet 1983 \[7\] Type I DM\ N = 10\ BP decreased to 128/84 mmHg with treatment\ Mean BP 144/97 mmHg\ 39 month treatment period compared to 29 month untreated period Rate of decline of GFR decreased from 0.9 to 0.39 ml min-1 mo-1 in treated period Proteinuria \> 0.5 g/d\ GFR approx 77 ml min-1 1.73 m^2^-1 Bjorck et al. Br Med J 1986 \[8\] Type I DM\ N = 14\ BP decreased to 155/94 mmHg with treatment (p \< 0.02)\ Mean BP 163/97 mmHg\ 4.8 yrs (2.8 yrs untreated followed by 2 yrs treated with captopril) Urinary protein decreased from 2.9 to 2.8 (NS)\ Mean proteinuria 2.9 g/d\ Rate of decline of GFR decreased from 10.3 to 5.5 ml min-1 yr-1 (p \< 0.02) Mean serum Cr 2.1 Parving et al. Br Med J 1987 \[9\] Type I DM\ N = 11\ BP decreased to 129/84 mmHg with treatment (p \< 0.01)\ \> 200 ug/min albuminuria\ 32 mo. untreated followed by 6 yrs treated) Albuminuria decreased from 1038 to 504 ug/min (p \< 0.01)\ BP 143/96 mmHg pre-treatment Mean GFR approx. 103 ml min-1 1.73 m^2^-1 Rate of decline of GFR decreased from 0.89 to 0.22 ml-1 min-1 mo-1 during treatment (p \< 0.01) Parving et al. BMJ 1988 \[10\] Type I DM treated with captopril or placebo\ N = 31\ MAP decreased by 8.7 mmHg to 136/85 by captopril; increased by 6.6 mmHg to 145/98 mmHg in controls (p \< 0.001) Albumin excretion decreased to 390 ug/min in captopril, increased to 1367 ug/min in controls (p \< 0.001)\ \> 0.3 g/d albumin (mean approx. 0.95)\ Captopril (N = 18) Control (N = 13)\ GFR decline 5.8 ml min-1 yr-1 in captopril vs. 10 ml min-1 yr-1 in controls (p \< 0.01) Baseline BP 146/93 mmHg and 137/95 mmHg in 2 groups, respectively\ 2.5 yrs Serum Cr \< 120 μmol/L\ GFR mean approx. 97 ml min-1 1.73 m^2^-1 Sawicki et al J Diabetes Complications 1995 \[11\] Type I DM\ N \~ 100\ Primary study endpoints (need for dialysis or death) decreased in IT group Proteinuria \> 0.5 g/d\ 5 yrs Serum Cr \< 265 μmol/L\ Diabetic retinopathy\ BP \> 140/90 mmHg or need for BP meds\ Intensive treatment including BP self-monitoring with goal BP \< 140/90 mmHg (IT) vs. routine treatment (RT) UKPDS 38 1998 13 Type II DM\ N = 31 with overt nephropathy (\> 300 mg/l albuminuria) 8.4 yrs Insufficient numbers to determine if BP control benefitted patients with overt nephropathy (P = 0.06) Serum Cr \< 177 μmol/L\ BP \<150/85 mmHg vs. \< 180/105 mmHg Trocha et al. J Hypertension 1999 \[12\] Type I DM\ N = 91\ Primary study endpoints (need for dialysis or death) decreased in IT group Proteinuria \> 0.5 g/d\ IT (N = 45)\ Cr \< 265 μmol/L\ RT (N = 46)\ Diabetic retinopathy\ 10 yrs BP \> 140/90 mmHg or need for BP meds\ BP self-monitoring with goal BP \< 140/90 (IT) vs. routine care (RT) Lewis et al. Am J Kidney Dis 1999 \[14\] Type I DM with overt nephropathy\ N = 129\ MAP 92 mmHg in Group I vs. 98 in Group II\ Serum Cr \< 354 μmol/L\ N = 63 in Group I\ No significant difference in rate of decline of GFR\ MAP of 92 mmHg (Group I) vs. 102--107 mmHg (Group II) N = 66 in Group II\ Significant decrease in albuminuria in intensive group 2 yrs Estacio et al. Diabetes Care 2000 \[15\] Type II DM\ N = 470 total\ BP decreased significantly with IT vs. MT (132/78 mmHg vs. 138/86 mmHg)\ DBP \> 90 mmHg\ N = 83 with overt proteinuria\ Pts with overt albuminuria demonstrated a steady decline in creatinine clearance of 5--6 ml min-1 1.73 m^2^-1 yr-1 throughout the follow-up period in both IT and MT groups (pNS) CCr of 76 ml min-1 1.73 m^2^-1 Randomized to intensive therapy (IT) (DBP \< 75 mmHg) or moderate therapy (MT) (DBP 80--89 mmHg) 5 yrs -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- \*categorical independent variables :::	PubMed Central	2024-06-05T03:55:59.649355	2005-6-28	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180831/", "journal": "BMC Nephrol. 2005 Jun 28; 6:8", "authors": [ { "first": "David J", "last": "Leehey" }, { "first": "Holly J", "last": "Kramer" }, { "first": "Tarek M", "last": "Daoud" }, { "first": "Maninder P", "last": "Chatha" }, { "first": "Majd A", "last": "Isreb" } ] }
PMC1180832	Background ========== Quality of life (QOL) issues have historically played an important part in the nursing role of patient advocacy. Ropka et al. found in the Year 2000 ONS Research Priorities Survey that quality of life ranked second among the top 20 research priorities of sampled nurses \[[@B1]\]. Other priorities included neutropenia/immunosuppression (fifth), as well as depression and stress-coping adaptation in thirteenth and fourteenth place respectively. These research priorities and the realization that accurate assessment of QOL in patients can provide both physicians and nurses with important clinical information \[[@B2]\] have provided the framework for initiating this research on the correlation between chemotherapy-induced neutropenia and QOL. The term quality of life has been defined and measured in many different ways in the literature. Haas \[[@B3]\] offers the following definition to differentiate QOL from closely related concepts such as well-being, satisfaction with life, and functional status: \"QOL is a multidimensional evaluation of an individual\'s current life circumstances in the context of the culture and value systems in which they live and the values they hold. QOL is primarily a subjective sense of well-being encompassing physical, psychological, social and spiritual dimensions\...\" Complications and side effects of cancer treatment can not only adversely affect clinical outcomes, but also negative influence a patient\'s quality of life. Chemotherapy specifically can impact the patient\'s QOL by significantly impacting social, physical and global functionality \[[@B4]\]. Many cancer chemotherapies work by suppressing fast-dividing cells like cancer cells; however, the fast-dividing cells of the bone marrow, which are responsible for producing the cells involved in the first line immune defense such as neutrophils and other white cells, can also be negatively affected. Neutropenia, i.e., decreased white blood cell counts, is the name given to this side effect of such myelosuppressive chemotherapies. The National Cancer Institute\'s Common Toxicity Criteria provides for 5 grades of neutropenia (Grade 5 = death) \[[@B5]\]. Grade 4 neutropenia is defined by an absolute neutrophil count (ANC) of less than 0.5 × 10^9^cells/L, commonly referred to as an ANC of less than 500 cells/mL. By comparison, grade 0, which is considered a normal count, varies from 3 to 6 × 10^9^cells/L. When patients develop febrile neutropenia (grade 4 neutropenia plus fever \[i.e., a body temperature of greater than 38.5°C), they are often hospitalized and placed on antibiotics to prevent life-threatening sepsis \[[@B6],[@B7]\]. In addition to the clinical consequences of febrile neutropenia, hospitalization and possible isolation can have a significant impact on a patient\'s quality of life. Little is actually known about the impact of neutropenia on quality of life, which has only occasionally been examined as a secondary outcome in the context of side effects of chemotherapies. Fortner \[[@B8]\] reported on a cohort of 62 patients receiving chemotherapy with no growth factors. Patients completed different scales (SF-36, Cancer Care Monitor, Psychosocial Adjustment to Illness Scale (PAIS)) at weekly intervals before having their absolute neutrophil count (ANC) measured. The results of this study provided preliminary evidence of the association between CIN and QOL impairments. Oncology nurses are increasingly playing a major role in the management of neutropenia. Recent nursing publications have reported on the suitability of the nurse, as a direct caregiver, to assess patients\' potential risk factors for neutropenia and implement guidelines for management \[[@B9],[@B10]\]. Maxwell et al \[[@B11]\] showed how the utilization guidelines for neutropenia management could not only save nursing staff considerable time, but also improve the overall flow of patient care for clinical practices. An increased knowledge of how CIN affects QOL would assist nurses in their patient education efforts and allow them to direct patients\' preventative care measures during myelosuppressive chemotherapy regimens. The purpose of this preliminary investigation was to obtain detailed information from patients, describing the experience of becoming neutropenic. In particular, by conducting qualitative interviews during neutropenic periods, patients were assessed on the impact it had on their quality of life and their routine functioning. Whereas we value and have, in fact, deployed standardized, quantitative measures of quality of life, this report focuses on qualitative information. Such qualitative information is often obtained to assist in the development of more finely-tuned, quantitative measures of a given condition. The pilot study reported here has resulted in a prospective study, results of which are going to be reported in the near future. Methods ======= Design and participants ----------------------- A convenience sample of adult cancer patients with grade 4 neutropenia was evaluated in the pilot study to determine the impact of neutropenia on patient psychological adjustment and functioning. The prospective design restricted participation to adult cancer patients from the West Clinic (Memphis, Tennessee) who were scheduled to receive the first cycle of one of the following chemotherapy regimens, at the full dose, given in either a 21- to 28-day cycle: (1) docetaxel, (2) AT (doxorubicin and paclitaxel), (3) CHOP (cyclophosphamide, doxorubicin, vincristine, and prednisone), (4) ESHAP (etoposide, methylprednisolone, cisplatin, and cytarabine), (5) carboplatin-paclitaxel, (6) carboplatin-docetaxel, and (7) carboplatin-gemcitabine. These chemotherapies were selected because they were judged to be among those that carried a significant risk of producing grade 4 neutropenia \[[@B7],[@B12]-[@B15]\]. There was no restriction on cancer type. Patients were excluded if they were already participating in another formal clinical trial; were scheduled to receive supportive treatment with colony-stimulating factors (eg, filgrastim or sargramostim); or had a life expectancy less than three months. Patients visited the clinic on days 0, 7, 14, and 21 of the chemotherapy cycle for assessment of vital signs, complete blood count (CBC) with differential, and QOL assessment. For this analysis, 34 patients who developed grade 4 neutropenia (ANC \< 0.5 × 10^9^/L) at any point during their first cycle of chemotherapy were interviewed and completed a total of 100 interviews. Demographic and medical characteristics of the 34 participants are shown in Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. Seventy percent or more were white, married women who had high school or greater level of education. Consistent with first cycle treatment in an outpatient setting, most participants had a good (0 to 1) Eastern Collaborative Oncology Group (ECOG) performance rating \[[@B16]\]. Over two thirds of patients received docetaxel, and nearly half of the patients had either breast cancer or lung cancer. Only 5 of the 34 patients that were interviewed exhibited a significant fever (≥101.4°F) that required hospitalization. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Demographic Characteristics ::: --------------------------------------------- ----------------- Number of Patients with Grade 4 Neutropenia 34 Mean age in years (range) 61.4 (27 -- 76) Women 71% Ethnic background Black/African American 18% White/not Hispanic 82% Education ≥12 years 76% Marital status Married 70% Single 6% Divorced 3% Widowed 21% Family income (\$) Under 14,600 11% 14,601 to 29,999 19% 30,000 to 49,999 18% 50,000 to 79,999 29% 80,000 to 99,999 15% 100,000 to 220,000 8% --------------------------------------------- ----------------- ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Medical Characteristics of 34 Adult Cancer Patients with Grade 4 Neutropenia ::: ------------------------------------------------------ ----- Good ECOG performance status^a^ 79% Cancer diagnosis Breast 28% Lung 24% Lymphoma 15% Ovarian 9% Prostate 6% Other diagnoses 18% Metastatic disease 37% Type of chemotherapy Docetaxel 70% CHOP 12% Carboplatin-docetaxel 15% Carboplatin-gemcitabine 3% Hospitalization due to neutropenia and grade 3 fever 15% ------------------------------------------------------ ----- ^a^Good performance status is defined as ECOG performance scale score of 0 to 1 on a 5-point scale, where 0 is fully active, and 4 is completely disabled \[13\]. ::: Procedure --------- Participants completed informed consent procedures approved by Western Institutional Review Board (Olympia, Washington). Participants were compensated for their participation. ### Interview schedule Participants were scheduled to be interviewed on days 7, 10, 14, and 21 of their respective chemotherapy cycle, but the interviews only began once a patient had an ANC \< 0.5 × 10^9^/L (grade 4). Nursing staff in consultation with treating physicians examined the laboratory results and contacted research staff to come and interview the patient. For example, a patient who first developed grade 2 neutropenia on day 7 and grade 4 neutropenia on day 10 would have contributed only 3 interviews (days 10, 14, and 21). Thirty-three of the 34 patients who developed grade 4 neutropenia did so on or before day 10 of chemotherapy; one patient who was receiving carboplatin-gemcitabine chemotherapy first developed grade 4 neutropenia on day 21. ### Interview protocol The interviews were conducted by trained interviewers during clinic visits. Interviewers were 4 advanced graduate students in clinical psychology who were highly trained and experienced in structured clinical interviewing techniques with special emphasis on employing empathic reflection, asking open-ended questions, and using techniques to promote patient elaboration of their own experiences. These interviewers were part of the clinics staff in clinical research assistant positions. The interview protocol was designed to obtain the participant\'s perspective of the impact of neutropenia within the context of an empathic conversation. Although not administered in this study, the interviewers had also been trained in the Structured Clinical Interview to Diagnose DSM-IV Axis I Disorders (SCID) and were familiar with the criteria needed to identify individuals who might need special assistance and referral for severe psychiatric difficulties \[[@B17]\]. Medical staff (either physicians or nurses) informed participants of their current ANC and, if appropriate, explained that the condition was called neutropenia. Patients were instructed in how to keep a diary of symptoms and how to monitor temperature with a thermometer that was provided for home use. Patients were then interviewed by the clinical assistant. Interviews lasted 20 minutes and were retrospective accounts of the impact of neutropenia covering the period between the interview and the preceding visit. The following topics were covered, and all were phrased in terms of changes since the previous visit: physical feelings and sensations, reduction and/or cessation of activities, interactions with others, financial impacts, ability to work, sex life, emotions, satisfaction with medical care, thoughts about disease, thoughts about treatment procedures, and overall quality of life. These areas were covered because they represent aspects of quality of life that might be affected by grade 4 neutropenia. In addition, interviewers encouraged participants to talk freely and to describe in their own words how they were feeling and how their daily living had been different after developing neutropenia. Every interview was recorded and transcribed verbatim. Preliminary analyses -------------------- Two raters characterized the information in the interviews by reviewing all transcripts and inductively developing descriptive categories for the nature of patient concerns expressed in the transcript \[[@B18]-[@B20]\]. Raters resolved discrepancies by mutual agreement. This procedure yielded 80 unique, descriptive problem domains, distributed over 5 broad categories as follows: Physical Complaints (n = 29), Daily Routine Disruptions (n = 14), Negative Thought and Self Evaluation (n = 17), Negative Emotions (n = 10), and Social Relationships (n = 10). Two raters then counted the instances of each problem domain within the 100 interview transcripts, and any discrepancies were resolved by mutual agreement. Table [3](#T3){ref-type="table"} illustrates the 5 broad categories along with selected unique problem domains and an illustrative exemplar for the displayed problem domain. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Categories and Exemplars of Patient-nominated Effects of Neutropenia ::: Major Category Specific Problem Domain Illustrative Exemplars in Patients\' Own Words -------------------------------------------------------- ------------------------------------------------------------------------------------------------------ Physical Complaints Tired, fatigued, exhausted, weak, feeling \"drained\" If I walk from one room to the next, I\'m worn out. Reduced sexual activity It has definitely slowed down. I don\'t have a great deal of interest. Daily Routine Disruptions Loss of daily routine I have stopped going to church and taking my son to school. Difficulties with household activities I have stopped most everything. I am not doing my housework. Negative Thought and Self Evaluation Worrying about paying for extra care needed If I don\'t work, I don\'t get paid. It\'s getting very scary. Feeling useless/helpless, letting people down I feel guilty when I see other people doing work I should be doing. Negative Emotions Feeling down I feel like giving up. I just wish it would all end. Feeling anxious I am scared. I hope I don\'t have to go through this (neutropenia) the next three times I come here. Social Relationships Decreased social contacts and activity I can\'t spend time with other people because I am sick. Avoiding crowds I don\'t go to church because someone might hug me or sneeze near me. Note: This table contains selected exemplars of the problem domains presented by patients in the five broad categories identified as follows: Physical Complaints, Daily Routine Disruptions, Negative Thought and Self Evaluation, Negative Emotions, and Social Relationships. ::: Results ======= Tables [4](#T4){ref-type="table"} through [8](#T8){ref-type="table"} illustrate the results of the tabulations of only the unique problem domains that were nominated in 5% or more of the interview. This restriction permitted a balance between rich description and characteristic or representative patient reactions to grade 4 neutropenia. These tables are designed to show how often a given problem was mentioned within the 100 interviews and also how frequently an individual patient mentioned the given problem, and data are presented in order of descending frequency. Altogether, the tables present information about 35 of the 80 specific problem domains identified in the content analysis of the interview transcripts. ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Patient-nominated Physical Complaints During and After Grade 4 Episodes of Chemotherapy-induced Neutropenia ::: Patient-nominated Physical Complaint Frequency Per Interview (N = 100) n (%) Frequency Per Patient (N = 34) n (%) ----------------------------------------------------------- ----------------------------------------- -------------------------------------- Tired, fatigued, exhausted, weak, feeling \"drained\" 69 (69%) 31 (91%) Sleep interruption (eg, onset insomnia, waking) 19 (19%) 14 (41%) Muscle aches, feeling achy, swollen joints 16 (16%) 13 (38%) Pain 15 (15%) 10 (29%) Cough, sore throat, mucous in throat, problems swallowing 10 (10%) 8 (24%) Reduced sexual activity (eg, too tired for sex) 9 (9%) 6 (18%) Change in taste of foods 7 (7%) 6 (18%) Decreased appetite 7 (7%) 7 (21%) Diarrhea 7 (7%) 6 (18%) Stomach upset, cramping 7 (7%) 7 (21%) Mouth Sores 6 (6%) 6 (18%) Nausea 5 (5%) 5 (15%) Skin problems (rash, skin breaking out, dry skin) 5 (5%) 3 (9%) Note: Thirty-four patients completed 100 interviews with each patient contributing between 1 and 4 interviews. To be included in the table, the complaint had to have been nominated within at least 5% of the interviews. ::: ::: {#T8 .table-wrap} Table 8 ::: {.caption} ###### Patient-nominated Social Relationship Changes During and After Grade 4 Episodes of Chemotherapy-induced Neutropenia ::: Patient-nominated Social Relationship Changes Frequency Per Interview (N = 100) n (%) Frequency Per Patient (N = 34) n (%) ----------------------------------------------- ----------------------------------------- -------------------------------------- Decreased social contacts and activity 38 (38%) 20 (59%) Avoiding crowds 17 (17%) 11 (32%) Other people getting on my nerves 10 (10%) 8 (24%) Feeling lonely and isolated 6 (6%) 6 (18%) Withdrawing and keeping to myself 6 (6%) 4 (12%) Note: Thirty-four patients completed 100 interviews with each patient contributing between 1 and 4 interviews. To be included in the table, the complaint had to have been nominated within at least 5% of the interviews. ::: Table [4](#T4){ref-type="table"} shows the frequency of the occurrence of patient-nominated problem domains broadly categorized as physical complaints. The most frequent problem mentioned by these patients with grade 4 neutropenia was fatigue, which was expressed using terms such as feeling tired, weak, and drained. Fully 91% of the participants complained about fatigue and its interference with their daily lives. In terms of frequency in the interviews, no other physical complaint matched fatigue, which was a complaint in nearly 70% of the interviews compared with less than 20% for other problems across interviews. Sleep interruption, muscle aches, pain, and sore throat were the next most frequently mentioned physical complaints, and these affected between 41% and 14% of the patients interviewed. With the exception of skin irritations, which affected 9% of the patients, the remaining problems listed in Table [4](#T4){ref-type="table"} affected between 15% and 21% of the participants. It should be noted that, in part because of the nature of the data, while we have an association between grade 4 neutropenia and these complaints, we do not have a causal link. It is entirely likely that some of these complaints such as fatigue and exhaustion are endemic to chemotherapy and would occur at a comparable rate among similar patients with similar cancers undergoing the same chemotherapy, only without the condition of grade 4 neutropenia. Given what little is known about the experience of neutropenia from the patient\'s perspective, this may be a reasonable start to finding out how the physical impact of neutropenia affects patient functioning and quality of life. The top five complaints listed in Table [4](#T4){ref-type="table"} certainly could be complaints that are exacerbated, if not caused, by neutropenia. Table [5](#T5){ref-type="table"} lists the most frequently mentioned effects of neutropenia on patients\' daily routine. Over half of the patients noted that their daily routine was interrupted, preventing them from leading normal lives. This effect was especially noted in terms of physical activity and may simply be a correlate of the severe fatigue that nearly all of the patients reported. In terms of quality of life, this interference with the activities of daily living prevented these neutropenic patients from engaging in the physical and social activities that brought them pleasure and made life enjoyable. ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Patient-nominated Daily Routine Disruption Problems During and After Grade 4 Episodes of Chemotherapy-induced Neutropenia ::: Patient-nominated Daily Routine Disruption Problems Frequency Per Interview (N = 100) n (%) Frequency Per Patient (N = 34) n (%) ------------------------------------------------------------- ----------------------------------------- -------------------------------------- Loss of daily routine (\"Cannot live life like I used to\") 42 (42%) 19 (56%) Unable to engage in normal physical activities 39 (39%) 19 (56%) Difficulties with household activities (eg cooking) 34 (34%) 19 (56%) Unable to work like before 27 (27%) 13 (38%) Laying around more and napping 25 (22%) 18 (53%) Unable to engage in normal non-physical hobbies 23 (23%) 16 (47%) Unable to attend religious services 21 (21%) 12 (35%) Difficulties with childcare 9 (9%) 6 (18 %) Note: Thirty-four patients completed 100 interviews with each patient contributing between 1 and 4 interviews. To be included in the table, the complaint had to have been nominated within at least 5% of the interviews. ::: Table [6](#T6){ref-type="table"} covers the broad category of negative thinking and negative self-evaluation. The experience of becoming neutropenic was associated with a sense of dread and increased fear about continuing further chemotherapy treatment for nearly one fourth of these patients. Interviewers were struck by the fact that some patients seemed to begin to question their own self worth as they became increasingly isolated and restricted because of lack of energy and the necessary medical prohibitions to avoid social contact during the period of high risk for infection (ie, severe neutropenia), even though this type of complaint was less frequent than physical complaints like fatigue. This negative self-evaluation was evident in approximately 1 out of 5 patients who developed grade 4 neutropenia. ::: {#T6 .table-wrap} Table 6 ::: {.caption} ###### Patient-nominated Negative Thought and Self Evaluation Changes During and After Grade 4 Episodes of Chemotherapy-induced Neutropenia ::: Patient-nominated Negative Thought and Self Evaluation Changes Frequency Per Interview (N = 100) n (%) Frequency Per Patient (N = 34) n (%) ---------------------------------------------------------------- ----------------------------------------- -------------------------------------- Feeling of dread about coming for chemotherapy 10 (10%) 9 (27%) Worrying about paying for extra care needed 9 (9%) 5 (15%) Feeling useless/helpless, letting people down 7 (7%) 6 (18%) Sense of loss of independence 6 (6%) 5 (15%) Worrying about being around other sick people 5 (5%) 5 (15%) Note: Thirty-four patients completed 100 interviews with each patient contributing between 1 and 4 interviews. To be included in the table, the complaint had to have been nominated within at least 5% of the interviews. ::: Table [7](#T7){ref-type="table"} details the most frequent negative emotions mentioned during interviews. Not surprisingly, nearly one fourth of the patients complained that they felt down or blue. This may be a natural consequence of the fact that over half of all the participants reported a restriction in their daily activities that brought them pleasure and a sense of accomplishment. A bit more surprising was the fact that over 20% of participants reported feeling more irritable and short tempered after episodes of neutropenia. ::: {#T7 .table-wrap} Table 7 ::: {.caption} ###### Patient-nominated Negative Emotions During and After Grade 4 Episodes of Chemotherapy-induced Neutropenia ::: Patient-nominated Negative Emotions Frequency Per Interview (N = 100) n (%) Frequency Per Patient (N = 34) n (%) ------------------------------------- ----------------------------------------- -------------------------------------- Feeling down 21 (21%) 8 (24%) Feeling irritable and frustrated 12 (12%) 7 (21%) Crying more than usual 5 (5%) 4 (12%) Feeling anxious 5 (5%) 5 (15%) Note: Thirty-four patients completed 100 interviews with each patient contributing between 1 and 4 interviews. To be included in the table, the complaint had to have been nominated within at least 5% of the interviews. ::: Table [8](#T8){ref-type="table"} illustrates social relationship changes associated with grade 4 neutropenia. Most patients (59%) reported a decrease in social contacts and social outings. This most likely resulted from the combination of the reported severe fatigue and the doctors\' orders to refrain from social contact during the period of high infection risk. This experience of being isolated from customary social contacts and activities was uniformly experienced as a negative aspect of neutropenia. Conclusion ========== The picture of neutropenia that emerges from these structured clinical interviews with grade 4 neutropenic cancer patients undergoing chemotherapy illustrates their diminished quality of life, wherein fatigue, negative emotion, and a sense of isolation and reduced self worth are featured. To be sure, not every patient studied experienced the full spectrum of these problems, and, even among those who did, the intensity of the experience varied from mild to rather severe, especially for the 5 individuals who were hospitalized with fever. Although neutropenia is characterized by the lab value ANC, neutropenia clearly has a broader consequence, considering how it can impact the quality of life of the affected patient and potentially endanger life. Results from these interviews may be useful in constructing measures of quality of life outcomes in chemotherapy patients who develop neutropenia. Physical symptoms are clearly important for any such measure, and inclusion of some measure of fatigue or exhaustion is mandatory. As one might expect from compromised immune functioning, patients mentioned some obvious physical symptoms such as sore throat, mouth sores, and swollen joints. Less intuitively obvious, but nevertheless important, were other symptoms such as sleep interruption and reduced sexual activity. These suggest that even the physical symptoms associated with neutropenia are not as straightforward as one might expect. The impact of neutropenia goes beyond physical symptoms to affect social, cognitive, and emotional functioning of patients. We found that most patients reported significant disruptions to their daily routine. The degree to which their fatigue and exhaustion interfered with their ability to live a normal life was striking. Instead of being able to carry on as before, these patients reported that they were upset and often felt guilty about not being able to do those routine tasks that gave them a sense of personal fulfillment and pleasure. Being in this debilitated state and unable to carry on with their usual routines also led to negative self-appraisals and negative emotions. It should be noted that we do not regard these reactions to be unusual or out of the ordinary. In fact, we consider these reactions to be normal, considering the very difficult physical conditions of undergoing chemotherapy with the added burden of developing neutropenia and the fact that we did not observe psychiatric disorders among these patients. Continued research is needed to establish specific correlations between physical symptoms caused by neutropenia and their psychosocial effects. This will further assist nurses and physicians to understand the impact of neutropenia on patients and allow them to institute specific interventions to decrease the severity of side effects from myelosuppresive chemotherapy as well as to ameliorate their influence on the patients\' quality of life. As a qualitative and preliminary study of the quality of life implications of neutropenia, this investigation has some important limitations. We focused on those patients who became neutropenic as defined by grade 4 neutropenia. Future research should investigate the human qualitative impact of lesser grades of neutropenia. We began with this extreme group to get as clear a picture as possible for this condition. Furthermore, it is also important to discover how these difficulties develop as patients progress from a normal ANC through various grades of neutropenia. For example, there may be subjectively reported precursors of grade 4 neutropenia that patients can identify when they have grade 3 neutropenia. This might be clinically important to facilitate identifying those patients who will be at greater risk for developing severe neutropenia that may require hospitalization. Another important limitation of this investigation is the lack of a comparison group, e.g., patients who were receiving chemotherapy but who did not experience significant neutropenia. When one looks at the content of the complaints that patients presented in this investigation, some of those seem like they would be complaints likely to surface for any patient undergoing chemotherapy. For example, a change in the taste perception and nausea are unlikely to be unique to neutropenia, and may reflect common complaints for a range of chemotherapy patients. Considering the major physical complaints of fatigue and a sense of exhaustion, it would be important to determine to what extent that complaint was generally true of chemotherapy patients versus being a complaint unique to, or especially exacerbated among, chemotherapy patients who developed significant neutropenia. It should also be noted that in this prospective study we did not attempt to correlate additional laboratory values such as hemoglobin that might be related to fatigue associated with anemia. Such important factors should be addressed in more controlled studies where patients with particular cancers undergoing specific chemotherapy regimens can be followed prospectively to make the appropriate comparison. In addition, experimental investigations where patients are randomly assigned to receive neutropenia preventive interventions such as colony stimulating factors could help to clarify the extent to which such fatigue is caused by neutropenia as distinct from caused by chemotherapy alone. Neutropenia is certainly a serious medical condition that many cancer patients will have to face. This has been known for quite some time. What has been less obvious is that neutropenia is a medical condition with widespread cognitive, emotional, and social impacts for patients. In addition to developing treatment protocols for preventing and treating neutropenia from a medical standpoint, it is also important to consider the human experience side from the point of view of the patient. Both febrile and afebrile neutropenia present formidable challenges to chemotherapy patients, and their caregivers need to be aware of these challenges to design treatments that adequately address them. Competing interests =================== Funding for this research was provided by Supportive Oncology Services, Ted Okon, CEO, and by an unrestricted educational grant from Amgen Inc. (Thousand Oaks, California). Amgen is the manufacturer of Neulasta^®^(pegfilgrastim) and NEUPOGEN^®^(Filgrastim), indicated for reducing the incidence of infection associated with chemotherapy-induced neutropenia in cancer patients with nonmyeloid malignancies. Authors\' contributions ======================= BF conceived of the study and participated in the design and coordination of the study. KT participated in the design and coordination of the study. TO participated in the coordination of the study and draft of the manuscript. AH participated in the design of the study, conducted the summary of data findings and drafted the manuscript. LS participated in the design of the study and development of the study protocol. All authors read and approved the final manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1472-6955/4/4/prepub> Acknowledgements ================ We thank Heith Durrence, Adrienne Kovacs, Daniel Taylor, and Kathryn Wheetley for assistance with structured clinical interviews. These individuals were supported by research stipends from SOS and The Department of Psychology of the University of Memphis, Memphis, TN. Lori Reynolds assisted with transcription of the interviews and was supported by a research stipend from SOS. We also thank the patients and staff of West Cancer Clinic for their cooperation.	PubMed Central	2024-06-05T03:55:59.652324	2005-7-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180832/", "journal": "BMC Nurs. 2005 Jul 8; 4:4", "authors": [ { "first": "Barry V", "last": "Fortner" }, { "first": "Kurt W", "last": "Tauer" }, { "first": "Ted", "last": "Okon" }, { "first": "Arthur C", "last": "Houts" }, { "first": "Lee S", "last": "Schwartzberg" } ] }
PMC1180833	Background ========== Bisdioxopiperazines, including ICRF-154, razoxane (ICRF-159, Raz); ICRF-186 and ICRF-187), two stereo-isomers of Raz, and ICRF-193, developed in the UK, were some of the earliest agents found against a murine spontaneous metastatic model (Lewis lung carcinoma) in 1969 \[[@B1]\]. Many papers and projects have dealt with their potential use and mechanisms since that time. Three main mechanisms of bisdioxopiperazine action have been investigated, including assisting in radiotherapy, \[[@B2],[@B3]\] overcoming multi-drug resistance (MDR) of daunorubicin and doxorubicin to leukemias \[[@B4],[@B5]\] and inhibiting topoisomerase II \[[@B6],[@B7]\]. More importantly, Raz has been licensed for cardioprotectant of anticancer anthrocyclines in more countries. Since bisdioxopiperazines represents a unique family of antimetastatic agents that are structurally conservative in their pharmacological actions, two new derivatives, probimane \[1,2-bis (N^4^-morpholine-3, 5-dioxopeprazine-1-yl) propane; AT-2153, Pro\] and MST-16, 1, 2- bis (4- isobutoxycarbonyloxymethyl- 3, 5- dioxopiperazin-1- yl) ethane were synthesized at this institute in Shanghai, China. \[[@B8],[@B9]\]. Apart from data of anti-tumor activity \[[@B10]-[@B12]\], the pharmacological mechanisms of Pro as Raz, like the detoxication of Adriamycin(ADR), induced cardiotoxicities and synergism with ADRagainst leukemias were reported at Henan Academy of Medicine, Henan, China \[[@B13]\]. As the main researchers of Pro, we reported some novel biological actions of Pro, including the inhibition of the activity of calmodulin (CaM), a cell-signal regulator, which can explain anticancer actions and the combined cytotoxic effect of Prowith ADR\[[@B13],[@B14]\] inhibiting lipoperoxidation (LPO) of erythrocytes \[[@B15]\], down-regulating sialic acid synthesis in tumors \[[@B16]\] and blocking the binding of fibrinogen to leukemia cells \[[@B17]\]. MST-16, as a licensed drug in Japan since 1994, was permitted for direct use in leukemia chemotherapy, mainly against adult T-cell leukemia treatment \[[@B18]\]. Structural formulae of the three bisdioxopiperazines are represented in Figure [1](#F1){ref-type="fig"}. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Structural formulae of three bisdioxopiperazines ::: ![](1471-2210-5-11-1) ::: As a new bisdioxopiperazine, the pharmacological characters and features of Pro are intriguing. Increased understanding of the advantages and disadvantages of the two compounds is a first step for promoting applications of Pro and MST-16. Therefore, in depth pharmacological evaluation was carried out. Tumors studied are from 7 different organs of origin -- two gastric tumor cell line (SCG-7901, MKN-28), a lung tumor cell line (A549), a colon cancer cell line (HCT-116), two mammary tumor cell lines (MDA-MB-435, MDA-MB-468), one hepatic tumor cell line (BEL7402), two leukemia cell line (HL-60 and K562) and an uteric cervical tumor cell line (HeLa). In addition, time- and concentration-dependent relations to classify the effectiveness of different therapeutic schedules and schemes of Pro and MST-16 therapy have been addressed. Results ======= Cytotoxic effects of Pro and MST-16 against human tumor cell lines ------------------------------------------------------------------ Data on the anticancer effects of Pro using 10 human tumor cell lines in vitroare showed in Figure [2](#F2){ref-type="fig"} and Table [1](#T1){ref-type="table"}. Pro had anticancer effects in vitroat clinical acceptable concentrations (IC~50~values \< 50 μM) by MTT methods. The IC~50~values of Pro are 1.3672 ± 0.6230 μM, 24.314 ± 5.465 μM, 14.476 ± 3.085 μM, 45.325 ± 5.335, 22.169 ± 1.250, 0.02947 ± 0.02456 μM, 5.3417 ± 1.245 μM, 4.786 ± 1.556, 42.457 ± 2.325 μM and 18.238 ± 1.112 μM representing tumor cells of SCG-7901 and MKN- 28 (two human gastric tumor cell lines), HCT-116 (a human colon tumor cell line), MDA-MB-435 and MDA-MB-468 (two human mammary tumor cell lines), A549 (a human lung tumor cell line) and HL60 and K562 (two human leukemia cell lines), BEL-7402 (a human hepatic tumor cell line) and HeLa cell (a human uteric cervical tumor cell line) respectively (Figure [2](#F2){ref-type="fig"}). Among these tumor cell lines, Pro is more effective to SCG-7901 (a gastric cancer cell line), A549 (a lung cancer cell line) and HL60 and K562 (two leukemia cell lines), the IC~50~values being ≤10 μM. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Anticancer activities of probimane in vitro. MTT method is used. Pro exposures for 48 h at 5 different concentrations. n = 3 and in 2 independent tests. ::: ![](1471-2210-5-11-2) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### The IC~50~values of Pro in different human tumor cell lines for 48 h. MTT method was used. ::: Cell origin Cell types IC~50~μM; mean ± SD ----------------- ------------ --------------------- Gastric SCG-7901 1.3672 ± 0.6230 MKN-28 24.314 ± 5.465 Colon HCT-116 14.476 ± 3.085 Mammary MDA-MB-435 45.325 ± 5.335 MDA-MB-468 22.169 ± 1.250 Pulmonary A549 0.02947 ± 0.02456 Leukemia HL-60 5.3417 ± 1.245 K562 4.786 ± 1.556 Uteric cervical HeLa 18.238 ± 1.112 Hepatic BEL-7402 42.457 ± 2.325 ::: Comparison of the cytotoxic effects of bisdioxopiperazines with other drugs --------------------------------------------------------------------------- The cytotoxic effects against tumor cell lines (p388, HL-60 and HeLa cells) are included in Table [1](#T1){ref-type="table"}. Although IC~50~s of Dox, VCR and 5-Fu are lower than that of Pro, the greatest inhibitory rates of Pro at high concentrations are seen (Table [2](#T2){ref-type="table"}). No inhibitory difference between low and high concentrations of Dox, VCR and 5-Fu was observed (Table [3](#T3){ref-type="table"}). Generally, the LD~50~of VCR and Dox in experimental animals and humans are dramatically lower than Pro. These results suggest more difficult management and wider toxicities of these drugs in their application in the clinics, suggesting Pro may avoid these drawbacks. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Cytotoxic effects of anticancer drugs against tumor cell lines in vitro; drug exposure for 48 h ::: Compounds IC~50~μM ---------------- ----------- ------- ------ Doxorubicin 11.7 0.005 1.12 Vincristine No effect 0.05 4.56 5-fluorouracil 22.6 0.04 0.23 Probimane 64.6 1.97 5.12 ICRF-187 64.0 3.73 129 MST-16 5.23 33.4 26.4 ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Dose- response relations between anti-cancer drugs for cytotoxic effect against human leukemia cell line HL-60 for 24 h; \* P \< 0.01; n = 3; ::: Compounds Concentrations OD values Mean ± SD Percentage inhibition % ---------------- ---------------- --------------------- ------------------------- Control \-- 1.229 ± 0.125 \-- Probimane 10.0 0.298 ± 0.010\* 75.6 2.0 0.260 ± 0.005\* 78.9 0.4 1.142 ± 0.010 7.1 0.08 1.199 ± 0.012 2.4 Doxorubicin 10.0 0.256 ± 0.021\* 79.2 2.0 0.266 ± 0.013\* 78.3 0.4 0.312 ± 0.016\* 74.5 0.08 0.408 ± 0.031\* 66.9 5-Fluorouracil 5.0 0.421 ± 0.021\* 65.6 1.0 0.518 ± 0.012\* 57.9 0.2 0.585 ± 0.025\* 54.4 0.04 0.892 ± 0.038\* 27.5 Vincristine 5.0 0.425 ± 0.010\* 65.4 1.0 0.423 ± 0.009\* 65.6 0.2 0.401 ± 0.009\* 67.4 0.04 0.394 ± 0.012\* 68.0 ::: Comparison of anti-tumor effects of probimane and MST-16 and their time- response relationships ----------------------------------------------------------------------------------------------- Cytotoxic effects (IC~50~) of probimane and MST-16 against tumor cells were compared (Figure [3](#F3){ref-type="fig"}). ::: {#F3 .fig} Figure 3 ::: {.caption} ###### The IC~50~of probimane and MST-16 on 3 human tumor cell lines (SCG-7901, HCT-116 and MDA-MB-468) for 48 h exposure. MTT method was used. ::: ![](1471-2210-5-11-3) ::: In addition, the time- response curves indicate that the anti-tumor effects of Pro increase to a climax over 3 days of drug exposure (Figures [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"} and [6](#F6){ref-type="fig"}). The cytotoxic effects of Pro persist or rise with time, whereas those of VCR, Dox and 5-Fu decrease after 24 h (Table [4](#T4){ref-type="table"}). IC~50~of both Pro and MST-16 reduces dramatically by 72 h from 48 h. (Figures [7](#F7){ref-type="fig"} and [8](#F8){ref-type="fig"}). The reductions of IC~50~for both agents Pro or MST-16 depend on the cell line. IC~50~ratios of Pro and MST-16 for 3 days relative to 2 days against the most metastatic phenotype tumor cell line MDA-MB-435, are 8.9 and 7.5 times higher, and 2.6 times for the medium metastatic phenotype of MDA-MB-468 cells. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Time -- response curve of probimane inhibiting a human mammary cell line (MDA-MB-468). MTT method was used. A: Pro concentration at 5 μM (dark); B: Pro concentration at 0.5 μM (purple). ::: ![](1471-2210-5-11-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Time- response curve of probimane inhibiting a human gastric tumor cell line SCG-7901. MTT method was used. A: Pro concentration at 5 μM ::: ![](1471-2210-5-11-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Time- response curve of probimane inhibiting a human mammary cell line (MDA-MB-435). MTT method was used. A: Pro concentration at 50 μM (dark); B: Pro concentration at 5 μM (purple). ::: ![](1471-2210-5-11-6) ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### The time- response relations between different anticancer drugs for cyto-toxic effects against leukemia cell line HL-60; N = 3, probimane, Pro; 5-fluorouracil, 5-Fu; doxorubicin, Dox; vincristine, VCR; ICRF-187, (+) stereo-isomer of razoxane ::: Compounds Concentrations Percentage inhibition % ----------- ---------------- ------------------------- ------ ------ Pro 10 75.6 78.5 75.9 5-Fu 2 65.8 53.5 51.1 Dox 4 78.3 72.4 72.3 VCR 2 65.4 59.0 57.1 ICRF-187 10 47.6 37.8 42.9 MST-16 10 47.8 5.6 0.0 ::: ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Differences of anticancer effects (IC~50~) of probimane for different exposure intervals by a MTT method, n = 3 for 2 independent tests. ::: ![](1471-2210-5-11-7) ::: ::: {#F8 .fig} Figure 8 ::: {.caption} ###### Differences of anticancer effects (IC~50~) of MST-16 for different exposure intervals by a MTT method, n = 3 for 2 independent tests. ::: ![](1471-2210-5-11-8) ::: G~2~and M phase arrests induced by Pro or MST-16 ------------------------------------------------ Our data shows that both probimane (Pro) and MST-16 can arrest tumor cells in G~2~and M phases of the cell cycle. Dose- and/ or concentration- dependency are observed in G~2~and M arrests (Figures [9](#F9){ref-type="fig"} to [12](#F12){ref-type="fig"}), and the arresting effect of Pro on MDA-MB-435 and HCT-116 is only 2 times higher for MST-16 at equivalent concentrations. Pro at 4 μM can increase G~2~/M accumulation from 16.8 % (vehicle control) to 86.4 % after 24 h (p \< 0.001, n = 3). ::: {#F9 .fig} Figure 9 ::: {.caption} ###### G~2~/M phase arrests of human mammary tumor cell line (MDA-MB-435 cell) exposed to probimane at different concentrations for 20 h. ::: ![](1471-2210-5-11-9) ::: ::: {#F12 .fig} Figure 12 ::: {.caption} ###### G~2~/M phase arrests of human mammary tumor cell line (MDA-MB-435 cell) exposed to probimane for different times. A: vehicle control; B: Pro 6 h; C: Pro 12 h; D: Pro 24 h ::: ![](1471-2210-5-11-12) ::: Chromosome segregation inhibition by Pro and MST-16 --------------------------------------------------- Chromosome linkages, aggregations and segregation in tumor cells were blocked by both Pro and MST-16. Figure [13](#F13){ref-type="fig"} and [14](#F14){ref-type="fig"} show linkages and segregation blockade of chromosomes in cells treated with Pro and MST-16 at 4 μM. Despite this, chromosomes began to separate with each other, and their morphology became slimmer at lower concentration 1μM in both human mammary tumors of MDA-MB-436 cells and MDA-MB-468 cell lines in vitro. This chromosome poisoning action of Pro, MST-16 and ICRF-187 was seen at 1- 4 μM. In vehicle control group, chromosomes of tumor cells separated from each other very well. Although we only show typically one or two cells, the chromosomal characteristics in each group have an overall consistency (\> 80 %) in each piece of preparation from cell treated with Pro, MST-16 and ICRF-187. They are common characteristics and phenotypes induced by the three compounds. In addition, there seems no difference in overall chromosome effects of Pro and MST-16 at equivalent concentrations (Figures [13](#F13){ref-type="fig"} and [14](#F14){ref-type="fig"}), suggests that Pro and MST-16 act equally in this pathway. ::: {#F13 .fig} Figure 13 ::: {.caption} ###### Chromosomal behaviors of human mammary tumor cell line- MDA-MB-435 incubated with bisdioxopiperazines. A: control; B: probimane 4 μM; C: MST-16 4μM; D: ICRF-187 4μM. ::: ![](1471-2210-5-11-13) ::: ::: {#F14 .fig} Figure 14 ::: {.caption} ###### Chromosomal behaviors of human mammary tumor cell line- MDA-MB-468 incubated with bisdioxopiperazines. A: control; B: Probimane 4 μM; C: MST-16 4μM; D: Probimane 1μM. ::: ![](1471-2210-5-11-14) ::: Discussion ========== Increased understanding over the mechanisms of bisdioxopopiperazines can greatly improve their indications and narrow down contraindicates in clinical practice. The explanations for the anticancer actions of bisdioxopiperazine are currently focusing on anti-angiogeneses \[[@B19],[@B20]\] and tumor cell DNA alterations caused by topoisomerase II. Generally speaking, most angiogenesis inhibitors often have low cytotoxicity and are ineffective against larger tumor masses, and are better combined with cytotoxic drugs clinically \[[@B21],[@B22]\]. This work on the anticancer activity of Pro and MST-16 shows that they act through the blocking of chromosomal segregation and G~2~/M phase arrests, causing complete inhibition of tumor cell division. Pro, MST-16 and ICRF-187 play similar roles at equi-molar concentrations. This pathway may be related to topoismerase II inhibition \[[@B23]\] as a possible mode of tumor growth inhibition, but is not suggested as a systematic approach through a cascade series. Two findings in this study need further discussion; (i) the effective ranges of Pro and MST-16 in the blocking of chromosome segregation, and causing G~2~/M phase arrests are 1- 4 μM, similar for Pro and MST-16. This suggests the two processes operate in the same course or cascade, and most possibly are directly linked; (ii) cyto-toxicity test (MTT) showed that Pro was more effective than MST-16. Lacking parallels in the effective dose ranges of Pro and MST-16 between cyto-toxicities and chromosome segregation -- induced tumor inhibition can be explained by the fact that these effects of Pro and MST-16 do not strictly follow the same pathway given in Figure [15](#F15){ref-type="fig"}. Stronger cytotoxic effects of Pro against many other human tumor cell lines than original bisdioxopiperazines derivatives, especially on solid tumors, suggest some as yet undiscovered mechanism that Pro may have, and Pro may have better applications and require fewer drug combinations in the future. ::: {#F15 .fig} Figure 15 ::: {.caption} ###### Proposed mechanism of anticancer effects for bisdioxopiperations. ::: ![](1471-2210-5-11-15) ::: This work shows that anticancer activities of Pro against lung cancer and leukemia are relatively greater than against other tumor typies. Cytotoxic and antimetastatic activities of Pro against lung tumor models in vivohave also been found \[[@B24]\]. Lung cancer is the most prevalent among all cancer categories, and is one of the deadliest cancers in the clinics. Targeted at lung cancers, Pro may offer better medical and economic benefits in the future. For clinical chemotherapy, the paramount task is the balancing between treatment outcome and risks \[[@B25]\]. To optimize chemotherapeutic protocols containing bisdioxopiperazines, knowledge of its pharmacological parameters in terms of concentration- and time- responses are prerequisites. We found that Pro and MST-16 might act and accumulate longer in tumor cells than most of anticancer drugs. The peak of cytotoxicity of both Pro and MST-16 is on day 3, and not usually on day 2. This result and our early work of auto-radiography that Pro \[[@B26]\] persists longer in tumor tissues suggest that longer intervals may be used between treatments and less nursing responsibilities may arise, while maintaining high levels of tumor growth inhibitions. The long- term cytotoxic effects of Pro and MST-16 are more obvious in high metastatic tumor cell lines, which can explain the selective effects of compounds to tumor metastases. Early reports suggest that MST-16 needs to transform into ICRF-154 to exhibit its anticancer effects \[[@B27]\]. This work proves that MST-16 does not degraded to ICRF-154, and has a lower cytotoxic effects against tumor cells than Pro. Yet MST-16 can maintain a high activity in the cascade of the proposed mechanism -- chromosome segregation blockage and cellular G~2~/M phase arrest, leading to inhibition of cell division (Figure [15](#F15){ref-type="fig"}). It further suggests this mechanism is not a pivotal pathway for cytotoxic activity against tumors. Conclusion ========== We suggest that Pro has a wider spectrum of cytotoxic effects against human tumor cells than other bisdioxopiperazines, especially on solid tumors. The cytotoxic pathway of Pro and MST-16 appears to be through chromosome segregation blocking and G~2~/ M phase arrests. Pro may be more potent than MST-16. High dose- and time- related responses of Pro than VCR, 5-Fu and Dox are seen that suggest a selectivity by Pro against tumor growth. It suggests that the family of bisdioxopiperazines may sustain their cytotoxic effects longer than other anticancer drugs. Methods ======= Pro and MST-16 were synthesized in this institute. Other chemical agents were purchased from sources stated below. The tumor cell lines were obtained from various sources and serially passaged in this lab. MTT method ---------- The cells were maintained in RPMI 1640 (Gibco, Invitrogen Corporation, NY, USA) medium supplemented with 10 % FCS, streptomycin (100 μg/ml) and penicillin (100 units/ml). A density of 10^5^tumor cells /ml (90 μl) were seeded in 96-well plates for 24 h. Pro or MST-16 (10 μl), final concentrations indicated below, were added to each well for incubating for the next 48 h. 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl tetrazolium bromide (MTT) (Sigma Company, USA) (5 mg/ ml, 20 μl) was added to each well. Four h later, 50 μl compound solution (10 % SDS- 5 % isobutyl alcohol-1 N HCl) were added and incubated under 5 % CO~2~atmospheric condition for 24 h. Optical density at 570 nm was measured with a tunable microplate reader, VERSAmax, USA, each group was in triplicate samples and Pro or MST-16 were divided into 5 concentrations. Cell cycle analysis by cytometry -------------------------------- Tumor cells in exponential phase were exposed to Pro or MST-16. After 6 -24 h, cells were collected (300 × g, 10 min) and incubated with ice-cold PBS. Then fixed with ethanol and collected and washed with PBS by centrifugation (300 × g, 10 min). Cell deposition was added with PBS 1 ml and RNAse (5 μl) at 37°C bath for 15 min. Cells were dyed with 5 μl PI (2 mg/ml) in dark. Cells were measured for their DNA content by cytometry (Becton/Dickinson -- FACS Calibur) after passing through a cell filter. Chromosome preparation protocols -------------------------------- Cell chromosome preparation was by a routine procedure. Human mammary tumor cells (MDA-MB-435 and MDA-MB-468) were seeded into a 6-well plate and maintained under an atmosphere of 5 % CO~2~condition. When tumor cells covered about 60- 70 % of the surface, bisdioxopiperazines were added. Drug -- treated cells were treated with hypotonic KCl, 0.075 M at 37°C for 30 min. Cell nuclei were fixed with fresh-prepared fixative solution \[methanol/acetic acid, 3:1\] for 5 min. Cell nuclei were collected by centrifugation (900 × g 15 min) and washed with fixative solution by centrifugation (1500 × g 20 min). Cell nuclei were dropped onto a cooled glass plate and placing overnight under a dehydrogenated atmosphere. The scattered chromosomes were dyed with a Giemsa solution for 15- 20 min and washed with tap water. Chromosomal behaviors were viewed and photographed by microscopy with an oil-lense (LEICA, Qwin image processing analysis system, Germany). Statistics ---------- IC~50~of agents were calculated by software in this lab and X ± SD was calculated from data of two groups. List of abbreviations used are ============================== ADR (Dox), adriamycin; VCR, vincristine; 5-Fu, 5-fluorouracil; CaM, calmodulin; LPO, lipoperoxidation; Raz (ICRF-159 or ICRF-187), razoxane; Pro, probimane; PI, propidium iodide; MTT, 3-(4,5-dimethylthiazol-2-yl)- 2,5- diphenyl tetrazolium bromide; Authors\' contributions ======================= The work was designed by Da-Yong Lu and Jian Ding. The manuscript was written by Da-Yong Lu. Cytotoxic effects of compounds against human tumors was evaluated by Da-Yong Lu, Min Huang, Chen- Hui Xu, Wei- Yi Yang, Mei- Hong Li. Cell cycle phase determination and plotting were completed by Da-Yong Lu and Lin- Jiang Tong. Chromosome morphology was prepared and observed by Da-Yong Lu and Chao- Xin Hu. The project was partly administered by Li Ping Lin and Xiong Wen Zhang. Anticancer bisdioxopiperazines were provided by Wei Lu. ::: {#F10 .fig} Figure 10 ::: {.caption} ###### G~2~/M phase arrests of a human colon tumor cell line (HCT-116 cell) exposed to probimane at different concentrations for 20 h. ::: ![](1471-2210-5-11-10) ::: ::: {#F11 .fig} Figure 11 ::: {.caption} ###### G~2~/M phase arrests of human mammary tumor cell line (MDA-MB-435 cell) exposed to MST-16 and probimane for 20 h. A: vehicle control; B: MST-16 0.8 μM; C: MST-16 4.0 μM; D: Pro 2.0 μM ::: ![](1471-2210-5-11-11) ::: Acknowledgements ================ All the experiment was supported by National Foundation of Science and Technology in China and completed in Professor Jian Ding\'s lab, Shanghai Institute of Materia Medica, Chinese Academy of Sciences, PR China.	PubMed Central	2024-06-05T03:55:59.655602	2005-6-20	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180833/", "journal": "BMC Pharmacol. 2005 Jun 20; 5:11", "authors": [ { "first": "Da Yong", "last": "Lu" }, { "first": "Min", "last": "Huang" }, { "first": "Cheng Hui", "last": "Xu" }, { "first": "Wei Yi", "last": "Yang" }, { "first": "Chao Xin", "last": "Hu" }, { "first": "Li Ping", "last": "Lin" }, { "first": "Lin Jiang", "last": "Tong" }, { "first": "Mei Hong", "last": "Li" }, { "first": "Wei", "last": "Lu" }, { "first": "Xiong Wen", "last": "Zhang" }, { "first": "Jian", "last": "Ding" } ] }
PMC1180834	Background ========== The completion of human and rodent genome sequences \[[@B1]-[@B3]\] has brought the post-genomic era to the field of endocrine research. Detailed genetic maps of the main endocrine models can now be used to study the molecular basis of endocrine disease and the molecular mechanisms of hormone actions. The possibility to explore expression data of thousands of genes across multiple experimental paradigms promise to rapidly increase our understanding of biological systems \[[@B4],[@B5]\]. The acquisition of experimental data at a genomic scale requires high throughput technologies such as DNA microarray analysis. Microarrays enable the simultaneous assessment of expression levels of tens of thousands of gene products in an ease to perform assay. Microarrays are especially attractive to the field of endocrine research because regulation of gene expression is an important mechanism whereby hormones exert their physiological actions. This is obvious in the case of steroid and thyroid hormones, which use intracellular receptors belonging to the nuclear receptor family of transcription factors \[[@B6]\]. Peptide hormones also regulate gene expression after activating complex cascades of intracellular signaling events upon binding to transmembrane receptors \[[@B7]\]. If the relation between hormones and the expression of different genes could be annotated, the abundant knowledge concerning endocrine physiology might be used to clarify the biological function of those genes. On the other hand, because expression profiles are rich in information, they are suitable to study the complex and pleitropic actions of hormones. Here we analyzed a compilation of rat liver expression profiles from experiments designed to study gender and hormone actions in order to provide novel insight into the mechanisms of action of specific hormones. The dataset used in this study comprises the actions of thyroid hormone (T3), 17-α-ethinylestradiol and GH in liver. The data is freely available from the Endocrinology Gene Expression Database -- <http://www.cmm.ki.se/EndoGED> and have also been deposited in Gene Expression Omnibus <http://www.ncbi.nlm.nih.gov/geo/>. Using this collection of microarray data, we analyzed the differential contribution of estrogens and GH to the regulation of gender differentiated liver gene expression. We also compared the actions of GH and T3 in liver and found a small overlap comprising genes involved in lipogenesis suggesting the common regulation of SREBP transcription factors. The regulation of SREBP1 by GH and thyroid hormone was analyzed. Results ======= Exploring hormonal regulation of gene expression in liver --------------------------------------------------------- Several expression profiles where compiled and used to obtain insight into the hormonal regulation of liver physiology. The following six experiments were analyzed together (Table [1](#T1){ref-type="table"}): 17-α-ethinylestradiol treatment of male rats, infusion of bGH in young (3 month) male rats \[[@B8]\], infusion of hGH in old (2 years) male rats \[[@B9]\], bGH treatment of primary hepatocytes isolated from young male rats, comparison of female and male rats \[[@B8]\], and the rapid effects of T3 treatment of hypothyroid mice \[[@B10]\]. The following questions were formulated: Can expression profiling be used to clarify the physiological actions of hormones in the liver? Can promoter analysis of hormonally regulated genes provide novel insight into the mechanism of hormone actions? It should be noted that our intentions were not to exhaustively explore the data set but rather to illustrate the utility of microarray data mining in endocrine research. The experiments included in this analysis were not specifically designed to answer the questions formulated in the present study although they are sufficient to test our hypotheses. This mimics the situation when the experimental biologist try to derive knowledge from a set of disparate experiments performed in different laboratories, using different experimental designs and microarray technologies. Importantly, we have taken all possible measures to minimize systematic experimental errors. All the arrays used for analysis have been fabricated in house from a unique set of PCR products and these have been validated in numerous studies \[[@B7]-[@B15]\]. The protocols for labeling and data analysis were also similar along all the experiments. Within each of the experiments included in this analysis, we have accounted for biological variability by independent replication of the measurements using RNA from individual animals. Unsupervised clustering algorithms were first used on the entire dataset to gain a global view of different hormone actions in liver. Average-linkage hierarchical clustering (using Euclidean distance as measurement of similarity) was used to evaluate the relation between the expression profiles. As shown in Figure [1a](#F1){ref-type="fig"}, the effects induced by GH treatment in young males were similar to those induced by GH treatment in old animals. This observation supports the robustness of the GH effects since the designs of the two experiments differ not only regarding the age of the animals but also in the dose and type of hormone administered (bovine versus human GH). A calculation of the correlation coefficients between bGH-induced expression changes and the rest of the experiment groups supports this conclusion (Figure [1b](#F1){ref-type="fig"}). Interestingly, the expression changes induced by GH treatment of primary rat hepatocytes cultured on matrigel show a positive, although small, correlation with the effects observed in vivo(Figure [1a](#F1){ref-type="fig"} and [1b](#F1){ref-type="fig"}). This suggests that hepatocytes significantly contribute to the expression changes measured in intact liver upon GH treatment. The fact that well-known GH-regulated genes: insulin-like growth factor 1(IGF-1) and CYP2C12 \[[@B16]\] are also regulated in hepatocytes helps to substantiate our conclusion. The similar effects on gene expression after in vivoor in vitroGH treatment were confirmed for some genes using quantitative real-time PCR (Tables [2](#T2){ref-type="table"} and [3](#T3){ref-type="table"}). As shown in figure [1b](#F1){ref-type="fig"}, the correlation between GH treatment and thyroid hormone or estrogen treatment in young males was rather low. This is not surprising since both hormones have distinct liver functions not always overlapping those of GH. Differences in expression can also arise from the choice of treatment duration, dose and mode of hormone treatment and this could result in the underestimation of commonly regulated genes. This ambiguity can only be resolved by measuring more expression profiles in experiments specifically designed to study hormonal interactions. GH and estrogen contribute to gender differences in hepatic gene expression --------------------------------------------------------------------------- When we compared the gender-related expression differences to the rest of the experiments (Figure [1c](#F1){ref-type="fig"}, Table [4](#T4){ref-type="table"}), it was evident that as many as 48% of the female-enriched transcripts were also up-regulated by continuous GH infusion in 3 month old males. Similarly, around 34% of the male-enriched transcripts were down-regulated in males by the same treatment. Few genes (less than 4% of the male-enriched and none of the female-enriched) were affected in the opposite direction by continuous treatment with GH. On the other hand, estrogen treatment in 3 month old male rats induced the expression of 27% of the female-enriched genes, and repressed the expression of 27% of the male-enriched ones (Table [5](#T5){ref-type="table"}). Again, few genes (less than 7% of the gender differentiated) were affected in the opposite direction by treatment with estrogen. This data strongly suggest that both estrogen and GH significantly contribute to the gender differences in adult rat liver. Although, as illustrated in Figure [1c](#F1){ref-type="fig"}, one week of continuous infusion with GH is more efficient than an injection with 17-alpha-ethynilestradiol to feminize the adult male rat liver expression profile. T3 and GH regulate lipogenic genes in liver ------------------------------------------- Both GH and T3 are required for longitudinal growth. Therefore, we expected these hormones to have some overlapping effects on liver gene expression. Nevertheless, we found that similarities were limited to a small number of genes (indicated with a \# in Table [6](#T6){ref-type="table"}). A comparison of frequency distribution among gene ontology (GO) categories related to the biological function between the whole set of expressed genes and those up-regulated by T3 and GH revealed a statistically significant overrepresentation (p \< 0.05) of regulated genes in the category of lipid metabolism. Similar overrepresentation of genes involved in lipogenesis was identified among the genes induced by thyroid hormone. The SREBP family of transcriptional regulators plays an essential role in the regulation of lipogenesis and is known to regulate several of the genes found to be induced by thyroid hormone (Table [6](#T6){ref-type="table"}) \[[@B17]\]. Because the transcriptional effects of T3 were rapid, we hypothesize the existence of a direct mechanistic crosstalk between SREBPs and T3 signaling in mouse liver. Therefore, we analyzed the promoter regions of T3-induced genes to find out whether any consensus SREBP binding sites could be found in promoter regions. The genes that were included in this analysis are indicated with a \* in Table [5](#T5){ref-type="table"}. Interestingly, we could identify a clear overrepresentation of putative SREBP binding sites around the transcriptional start site in the T3-regulated group compared to the control group (Figure [2](#F2){ref-type="fig"}). To make sure that this result was not only due to higher phylogenetic conservation in the T3-regulated group, a basic statistic analysis was performed: the region from -500 to +100 relative the transcriptional start sites (TSS) was picked out of each alignment in both groups. A simple two-sided t-test of the mean conservation in both groups showed that they were not significantly different (p = 0.12). In fact, the T3-regulated group had a slightly lower (but not statistically significant) mean degree of conservation than the control group. We tested if the distribution of SREBP sites was significantly different between the two groups using Chi-square analysis. As the number of genes in the control group was higher, this distribution was normalized to correspond to 30 genes instead of the original 300 controls. The test demonstrates conclusively (p = 0.0001) that the two distributions are different and thus that there is a clear overrepresentation of putative SREBP binding sites near the TSSs in the T3-regulated group (Figure [2](#F2){ref-type="fig"}). In contrast, we were not able to localize a significant overrepresentation of thyroid hormone response elements (TREs) in the close promoters of the T3-regulated genes. We can not exclude that TREs exist in the promoters of these genes, but either they are located further away from the transcriptional initiation sites or in areas that are not phylogenetically conserved between mice and humans. Three different SREBP isoforms, 1a, 1c and 2, encoded by two distinct genes have been described. SREBP1c is the predominant form in adult liver and adipocytes \[[@B17]\]. SREBPs are translated as large precursors tethered to the endoplasmic reticulum and nuclear membrane where they, in response to sterol depletion, are proteolytically cleaved into mature, transcriptionally active factors that migrate to the nucleus and bind sterol regulatory elements of specific genes \[[@B18]\]. The mechanisms whereby T3 regulate SREBP actions have been previously studied but remain unclear. Long-term treatment with T3 has been shown to induce SREBP2 expression in hepatocytes \[[@B19]\]. We tested whether rapid transcriptional induction could account for the observed up-regulation of lipogenic genes. As shown in Figure [3](#F3){ref-type="fig"}, T3 treatment had no effect on SREBP1c or SREBP2 mRNA levels whereas the expression of SREBP1a was significantly reduced. The results demonstrate that rapid transcriptional induction of SREBPs is not the mechanism behind the T3 regulation of lipogenic genes. Since the set of genes overlapping between T3 and GH contained mostly genes involved in lipid metabolism, we also analyzed the transcriptional induction of SREBP1a, 1c and 2 after long term GH treatment, but failed to detect any significant effect that could explain the up-regulation of lipogenic genes (Figure [3](#F3){ref-type="fig"}). We next explored whether early events in SREBP activation, such as the proteolytic processing of endoplasmic reticulum (ER) resident SREBPs could be regulated by T3. The ER and nuclear forms of SREBP1 were measured before and after T3 treatment. The analysis was extended to study the effects of GH treatment. The Western blots in Figure [4](#F4){ref-type="fig"} show a decrease in the concentration of ER bound (high molecular weight) SREBP1, indicating a rapid activation of its proteolytic processing by T3. No significant effect of T3 could be detected for SREBP2 (data not shown). Interestingly, the concentration of the nuclear (short) form of SREBP1 was also reduced indicating that the total nuclear concentration of SREBP1 poorly reflects its activity as judged by the transcriptional induction of target genes. In contrast to the findings with thyroid hormone, the effect of GH doesn\'t seem to be exerted through similar mechanisms since no significant changes in ER or nuclear SREBP levels were evident after GH treatment (Figure [4](#F4){ref-type="fig"}). It is therefore possible that alternative mechanisms such as RNA stability or even the activation of other transcription factors account for the GH effects on lipogenic genes. Discussion ========== Here, we have attempted an analysis of a compilation of expression profiles to gain insight into the hormonal regulation of liver gene expression. We demonstrated that a positive correlation exists between the effects of GH treatment in primary hepatocytes cultured on matrigel and those detected in vivo. Nevertheless, the correlation was not very high despite the care taken of cultivating the hepatocytes on matrigel to avoid de-differentiation. We know from previous studies that hepatocytes cultured on matrigel express GH receptors, that GH signalling through the JAK2/STAT5 pathway is functional and that GH induces IGF-1, a well- known GH regulated gene in vivo\[[@B25]\]. Therefore, the differences between the in vivoand in vitromodels are likely due to structural and systemic factors found in intact liver which would be required for the full extension of GH actions. On the other hand, our data demonstrate that primary hepatocytes cultured on matrigel do provide a model to study some GH activated mechanisms; those directly related to the activation of the GH receptor and its signaling molecules. The newly described GH regulated genes in hepatocytes (Table [3](#T3){ref-type="table"}): phytanoyl-CoA hydroxylase (Phyh), hydroxysteroid 11-beta dehydrogenase 1 (Hsd11b1), the catalytic subunit of protein phosphatase 3, alpha isoform (Ppp3ca) and fatty acid translocase/CD36 antigen (FAT/CD36) constitute new target genes that could be used to study the basis of transcriptional regulation by GH in hepatocytes. Our global assessment of gene expression, demonstrates that estrogen and to a larger extent, the female-like continuous pattern of GH secretion are important for the maintenance of the gender differences in liver gene expression (Tables [4](#T4){ref-type="table"} and [5](#T5){ref-type="table"}). These data support the existence of a cross talk in the hepatic actions of GH and estrogens for the regulation of a subset of the female-enriched genes. The exact mechanism of this cross talk cannot be extracted from the data, but previous findings offer a possible explanation. It is well known that the liver expresses relatively low levels of estrogen receptor and that some estrogen-induced effects in liver, including the expression of the estrogen receptor itself, may be secondary to its feminizing effect on GH pituitary secretion \[[@B20],[@B21]\]. The role of GH secretory patterns in determining gender differences in rat liver expression of genes involved in sterol and drug metabolism has been described before \[[@B16],[@B22]\]. Here and in recent studies by Ahluwalia et al \[[@B23]\] and Stahlberg et al \[[@B8]\], the significance of this regulation is demonstrated at a more comprehensive level and novel hepatic gender-regulated genes are identified. We don\'t know yet how gender determines the transcription of these genes. To the date, only a few gender-predominant and GH-regulated transcription factors have been identified, including female predominant HNF-6 \[[@B24]\] and male predominant STAT5b \[[@B25]\]. Nevertheless, an analysis of promoter sequences of the genes identified in Table [4](#T4){ref-type="table"} has failed to identify any significant overrepresentation of HNF-6 or STAT5 consensus binding sites (data not shown). Therefore, further analyses are required to better understand the molecular mechanism behind the feminizing actions of GH in liver. Both GH and thyroid hormone are required for longitudinal growth \[[@B26]\]. The coordinate actions of these hormones are achieved through multiple mechanisms. Thyroid hormone is a key activator of GH secretion in the pituitary gland while GH promotes the formation of T3 from less active T4 in peripheral tissues \[[@B27]\]. Through the comparisons of expression profiles, we could identify a small group of genes that were up-regulated by both hormones. Most of the genes were assigned to the category of lipid metabolism by unbiased classification based on the current gene ontology. This is in agreement to the known lipogenic effects of both hormones in liver \[[@B7],[@B10],[@B28]\]. The correlation between GH and T3 effects was small which could be due to important differences in experimental design as well as to mechanistic differences. Therefore, the study of the long term effects of T3 treatment has the potential to identify more overlapping effects. The effects of T3 described herein are produced after 2 hours of treatment and are likely to be direct hormonal effects on the liver. Therefore, we analyzed the proximal promoter of genes rapidly regulated by T3 to gain further insight into the mechanisms of hepatic T3 actions. Interestingly, we were not able to localize a significant overrepresentation of TREs in the close promoters of the T3-regulated genes. On the other hand, we found a clear overrepresentation of putative SREBP binding sites; in agreement with the lipogenic nature of the T3-regulated genes. Further analysis did not detect any transcriptional induction of SREBPs by T3. Instead, we showed that T3 decrease the concentration of ER-bound SREBP1, probably due to induction of its proteolytic processing. Regulation of SREBP activity by T3 seems to be complex and involves multiple mechanisms. The thyroid hormone receptor (TR) and active SREBP1c can cooperate to activate the transcription of a single gene even when their response elements are situated very far apart \[[@B29]\]. Moreover, direct interaction between TR alpha and the active form of SREBP1 has been demonstrated when their binding sites are closely located in the promoter of the chicken acetyl-CoA carboxylase-α gene \[[@B30]\]. Direct interactions between TR and ER-resident SREBPs are unlikely to be responsible for the effect observed in this study since the two proteins have different intracellular localizations \[[@B31]\]. Nevertheless, this possibility can not be completely discarded since 10% of TRs are found in the cytosol both in the absence or presence of T3 \[[@B31]\]. Non-genomic effects of T3 such as the activation of PI-3 kinase, MEK and STAT transcription factors have been reported in several systems \[[@B32]\]. Whether the rapid induction of SREBP1 processing by T3 is due to a non-genomic mechanism of T3 action, or not, remains to be demonstrated. SREBP proteolyticactivation in the Golgi is regulated by its interaction with SCAP and Insig-1 and -2 \[[@B33]\]. When sterols are present at high concentrations, the SCAP/SREBP complex is retained in the ER. When the sterol concentration is reduced, SCAP does not interact with Insig and the SCAP/SREBP complex exits the ER and is delivered to the Golgi where it is proteolytically cleaved. A role of SCAP as a monitor of the composition of the cytoplasmic leaflet of the ER membrane has been proposed \[[@B33]\]. Since T3 has been shown to bind the outer half of the lipid bilayer in reconstituted microsomes \[[@B34]\], there is a hypothetical possibility that T3 could bind ER membranes and regulate SCAP activity. It is important to notice that the concentration of the nuclear (short) form of SREBP1 was also reduced upon T3 treatment. Why this occurs simultaneously to the transcriptional induction of SREBP target genes is unknown but a recent publication indicates that transcriptionally active SREBP1 is rapidly targeted for proteosomal degradation \[[@B35]\]. If this mechanism is at play, one would expect the effects of T3 on lipogenic genes expression to be transient. This is indeed the case, most of the T3 effects on lipogenic genes can not be found after 5 days of hormonal treatment \[[@B10]\]. Future studies will clarify the importance of SREBPs for thyroid hormone liver actions. Conclusion ========== In summary, we have analyzed six different experiments concerning the hepatic actions of GH, T3, estrogen and gender. We could conclude that GH and estrogen are both important determinants of gender-related differences in hepatic gene expression, that GH and T3 have overlapping effects on the regulation of several lipogenic genes, and that some T3 effects in the liver may be mediated through the induction of proteolytic processing of SREBP1. Through EndoGED and its web-based interface, we have made available a large data set of transcript profiles related to the actions of several hormones in different in vivoand in vitromodels. This resource is of special interest for the endocrine researcher offering the possibility of in depth exploration of hormonal transcriptional actions and interactions. In the same way, here exemplified by the analysis of hormone actions in liver, the actions of other hormones can be explored to generate testable hypotheses of relevance to endocrine research. Methods ======= Experimental design ------------------- We have studied the effects of different hormones on liver gene expression patterns in various rat and mouse models. The experiments included in this study are listed in Table [1](#T1){ref-type="table"}. They were conducted separately in different groups of animals, and by different persons in the laboratory. Some of the animal experiments have been described in previous publications: infusion of bovine growth hormone (bGH) in young male rats \[[@B8]\], infusion of human growth hormone (hGH) in old male rats \[[@B9]\], comparison of female and male rats \[[@B8]\], and the rapid effects of thyroid hormone (T3) treatment of hypothyroid mice \[[@B10]\]. Total hepatic RNA from individual animals was isolated using Trizol (Invitrogen, CA) and microarray hybridizations were performed on arrays larger than the ones described in the previously published studies. At least 4 statistically independent microarray measurements were used to characterize each physiological situation; with the exception of the experiment where thyroid hormone actions were studied where RNA pooled from 5 different mice was used. Other experiments included in this study have not previously been described. Primary hepatocytes were isolated from young male rats and cultured on matrigel as described previously\[[@B16]\]. The cells were grown in serum-free William\'s media E (Invitrogen, CA) supplemented with 55 μg/ml ascorbic acid (Sigma-Aldrich, MO), 100 IU/ml streptomycin and 1 μg/ml insulin (Sigma-Aldrich, MO) for two days before adding 100 ng/ml bGH (National Hormone and Peptide Program, A.F. Parlow, USA) to the media. GH-treated and untreated cells were harvested 24 hours later in Trizol (Invitrogen, CA), and RNA was purified according to the manufacturer\'s protocol. The procedure was repeated with cells isolated from different rats. In another experiment, 3 months old male rats were injected with 17-α-ethinylestradiol (5 mg/kg body weight; Sigma-Aldrich, MO) or vehicle 24 hours before sacrifice. Total RNA was isolated using Trizol (Invitrogen, CA). All animal experiments used in these studies were approved by the local ethical committee. cDNA microarrays, probe preparation and hybridization ----------------------------------------------------- The cDNA microarrays used in this study were produced in our lab, as described previously \[[@B7]\]. They have, however, been extended to comprise about 6200 clones, including clones from the TIGR Rat Gene Index, Research Genetics (Invitrogen, CA), and our own obtained through differential cloning experiments. The arrays were pre-hybridized in 1% BSA, 5XSSC and 0.1% SDS at 42°C for 1--2 hours, washed in milli-Q water, and dried immediately before the probe was applied. Total RNA was reverse-transcribed in the presence of Cy3- or Cy5-conjugated dUTP (PerkinElmer, MA) and purified as described previously \[[@B13]\]. In all studies except the one with T3-treated mice, RNA samples originating from livers of individual animals were labeled and each tester sample was hybridized against a control sample. In the T3-study, a pool of tester samples (from 6 different T3-treated animals) was hybridized against a pool of control samples due to limited availability of RNA. Dye-swaps were used in all studies to reduce systematic errors \[[@B36]\]. The final volume was adjusted to 25 μl with hybridization buffer consisting of 3.4XSSC, 0.3% SDS, 20 μg mouse Cot1 DNA (Invitrogen, CA), 20 μg polyA RNA, and 20 μg yeast tRNA. After heating at 98°C for 2 min and cooling to room temperature, the probe was added to the array and covered with a plastic cover slip (Grace Bio-Labs, OR). Hybridization took place at 65°C for 15--18 hours. The array was then washed and scanned with a GMS 418 scanner (Affymetrix, CA). Data processing and analysis ---------------------------- Data processing was performed essentially as described previously \[[@B13]\]. The software GenePix Pro (Axon Instruments, CA) was used to quantify the fluorescence intensity of each spot and the surrounding background. Automatic and manual flagging were used to localize absent or very weak spots (less than 2 times above background), which were excluded from analysis. The signal from each spot was calculated as the average intensity minus the average local background. We used a normalization method that takes into account and corrects for intensity-dependent artifacts in the measurements, the locally weighted linear regression (Lowess) method in the SMA package (Statistics for Microarray Analysis, available at <http://www.stat.berkeley.edu/users/terry/Group/software.html>) \[[@B36]\]. SMA is an add-on library written in the statistical language R. We next used EndoGED to extract all expressed hepatic genes from the included experiments concerning hormone treatment of rodents. In total, 6096 transcripts were expressed in one or more hybridizations. Hierarchical clustering using the TIGR Multiexperiment Viewer (MeV) software (available at <http://www.tigr.org>\[[@B37]\]) was performed to explore and compare the different hormone treatments in the generated gene expression matrix. The euclidean distance was used as distance metric. We first compared the usage of ratios from all hybridizations with the usage of just the median log2 ratio for each set of replicated measurements (experiment group). These strategies gave similar results in the clustering, grouping the replicated measurements closely together. The median rather than the mean ratio was used to minimize the influence of outliers. Therefore, we calculated the median log2 ratio for each transcript within each experiment group, and used this ratio for further analysis. Genes that were detected in only half of the replicated measurements, or less, were excluded since we did not consider these measurements reliable. Included in the hierarchical clustering were only genes that had four or more median expression ratios from the six experiment groups (2518 genes). Using the same dataset with median ratios from each experiment group, we also calculated the correlation coefficient between the different experiment groups. A statistical evaluation of differentially expressed genes was performed using the SAM (Significance Analysis for Microarrays) statistical technique \[[@B38]\]. This was done for each experiment group separately. A 5% false discovery rate (FDR) was used as cutoff. A gene expression matrix containing only genes affected by GH in vivoin a statistically significant manner was extracted from EndoGED. The T3-regulated genes in this list were identified by applying a log2 ratio cutoff of ± 1 (2-fold regulation) in each of the T3 hybridizations. We next used the web-based tool eGOn (developed at the Norwegian University of Science and Technology, available at <http://nova2.idi.ntnu.no/egon/>) to functionally classify the transcripts. With the two-sided one-sample binomial test implemented in eGOn, we compared the list of differentially expressed genes to all genes expressed in the T3 and GH experiments. The same software was also used to classify all T3-regulated genes (at least 2-fold up-regulation), and to compare them to all genes expressed in the experiments regarding T3. For the comparisons between gender differentiated genes and the effects of GH and estrogen, a mean ratio cutoff (log2 ratio treated/untreated \>0.58, corresponding to at least 1.5-fold difference) was applied on top of the SAM statistical criteria. Brief description of EndoGED ---------------------------- The EndoGED system includes a Lab Information Management System (LIMS) to manage array fabrication, and modules to collect, store and process gene expression data concerning the actions of hormones. The system offers a flexible solution to integrate external analysis tools, including possibilities to store transformed expression values (e.g. normalized ratios) and associated parameters derived from statistical evaluation. The systems run on Microsoft operating system and have a client-server architecture implemented in SQL-Server as the database engine and Borland Delphi for the program modules. In the database design, we have taken into consideration the latest MIAME recommendations from the MGED Society regarding microarray data description \[[@B39]\]. We have implemented a detailed description of biomaterials and treatments used in the experiments, carefully considering what information would be significant to the endocrine researcher. The exploration tools have been implemented to allow easy retrieval of relevant data through a multilevel search engine. Furthermore, we have made part of our collected data available through Internet for easy access to researchers worldwide <http://www.cmm.ki.se/EndoGED>. The database structure and software are available for free to academic and other nonprofit researchers. Real Time-PCR ------------- The expression of some genes from the array experiments were verified using quantitative real-time PCR. The Dynamo kit (Finnzymes Oy, Finland) containing SYBR Green was used for quantification. The primers and applied annealing temperatures are listed in Table [2](#T2){ref-type="table"}. The expression of all genes was normalized to glyceraldehyde-3-P dehydrogenase (GAPDH), which was always measured in parallel to the other genes. Promoter analysis ----------------- We undertook a promoter analysis of a group of genes that were all up-regulated more than 2-fold two hours after T3 treatment in hypothyroid mice. We sought to find out if there was a statistically significant overrepresentation of SREBP and TRE binding sites in the T3-regulated group of genes compared to a control group, unaffected by the treatment. We used pair wise cross-species comparison (phylogenetic footprinting) as described by Lenhard et al \[[@B40]\] to identify putative transcription factor binding sites in the regions upstream of the transcriptional start site (TSS). Comprehensive reviews covering the field of transcriptional regulation bioinformatics are available \[[@B41],[@B42]\]. As the study concerned mice, we used mouse orthologs to the rat genes printed on the arrays. To locate mouse orthologs for the rat sequences, we aligned the rat sequences to the mouse genome (NCBI build 30) using BLAT \[[@B43]\] and then searched for mouse cDNA sequences aligning to the same loci as the rat sequences and indicating similar gene structures. Human cDNA sequences orthologous to the mouse cDNA sequences were identified using the GeneLynx database \[[@B44],[@B45]\]. Guided by genomic mappings of the cDNA sequences from the UCSC Genome Browser Database \[[@B46]\], we retrieved human (NCBI build 33) and mouse gene and promoter sequences. Since cDNA sequences are often truncated, we used consensus exon-intron structures derived from overlapping and similar cDNA mappings. We aimed to retrieve sequence from -5000 to +100 relative to TSSs. However, to ensure that the corresponding human and mouse genomic regions were extracted, we extended this region if, in an alignment of the human and mouse gene sequences, the TSSs could not be align. In addition, to avoid inclusion of other nearby genes in the retrieved sequences, we truncated the sequences at the border of any multi-exon cDNA mapping upstream of a TSS. If such a mapping was encountered within 100 bps of a TSS, no sequence was retrieved. Genes which could not confidently be mapped onto either mouse or human genomes were excluded from the analysis. We analyzed 30 genes rapidly up-regulated by T3 (indicated with a \* in Table [6](#T6){ref-type="table"}) and, as a control group, 300 expressed but unaffected genes that were randomly selected from the array. We used a 50 bp sliding window and 70% sequence conservation for the alignment windows. The binding model was constructed by merging two matrix models (M00220 and M00221) from the TRANSFAC database \[[@B47]\]. The score cutoff for this model was set to 75%. Western blots ------------- Whole cell protein extracts were prepared from frozen liver tissue by homogenization in RIPA buffer. Protein extracts were resolved by SDS-PAGE and transferred to PVDF membranes with a Trans-Blot SD semi-dry transfer cell (Hoefer, Pharmacia Biotech, Sweden). The filters were blocked for 1 h at RT in TTBS (20 mM Tris-HCl, 150 mM NaCl, Tween 0.1%, pH 7.4) containing 5% skim milk powder. Membranes were incubated overnight at 4°C with an anti-SREBP1 or antiβ-actin, as a loading control (SantaCruz Biotechnology, USA). After three 10 min washes in TTBS, binding of primary antibody was visualized using horseradish peroxidase-conjugated secondary antibodies, and the immunolabeling was detected by an enhanced chemoluminescence (ECL) method according to the manufacturer\'s instructions (Pierce Chemical Company, USA). Authors\' contributions ======================= NS, LH, P.T-E, LFP and A.F-M carried out the microarray studies, NS, A.F-M and RM implemented the EndoGED database. NS carried out the expression measurements by RT-PCR. AS, BL and PS carried out the promoter analysis. AF-M carried out the Western Blot analysis. All authors read and approved the final manuscript. Acknowledgements ================ We are grateful to Dr. Vennstrom for providing tissues from T3 treated mice. This work was supported by grants from the Swedish Medical Research Council, Wallenberg Consortium North, the Swedish Society of Medical Research, the Fredrik and Ingrid Thuring Foundation, the Tore Nilsson Foundation for Medical Research, the Magnus Bergvall Foundation, the Swedish Medical Association, the Loo and Hans Osterman Foundation and the Åke Wiberg Foundation, the MECD (PM98-033 to L.F.), the Ministerio de Sanidad y Consumo (FIS 1/1000 to L.F.), and the MCYT (SAF2003-02117 to L.F.). L. H-H is recipient of predoctoral fellowship from the Ministerio Educación, Cultura y Deportes. Figures and Tables ================== ::: {#F1 .fig} Figure 1 ::: {.caption} ###### A comparison of all experiments included in the study. a) Hierarchical clustering of hepatic gene expression profiles using average linkage analysis and the Euclidean distance metric clusters together the in vivoexperiments concerning GH. The included experiments were: E\_male\_5d: 17-α-ethinylestradiol treatment of male rats for 1 day, young\_GH\_7d: infusion of bovine growth hormone (bGH) in young (3 months) male rats for 7 days, old\_GH\_21d: infusion of human growth hormone (hGH) in old (2 years) male rats for 3 weeks, hep\_GH\_24h: bGH treatment of primary hepatocytes from young (2 months) male rats for 24 hours, F/M\_rat: comparison of untreated female and malerat livers, and T3\_2h: thyroid hormone treatment of hypothyroid mice for 2 hours. b) Correlation coefficients of the different expression profiles to the profile of young GH-treated male rats. The strongest correlation is found between GH-treated young and old male rats, and also between young males treated with GH and female rats. The in vitroGH-treatment of isolated primary rat hepatocytes is also similar to the in vivosituation. c) Correlation coefficients of the different expression profiles to the female/male comparison. The strongest correlation is found between the female profile and GH-treated young and old male rats. Estrogen treatment of male rats shows lower correlation to gender. ::: ![](1472-6793-5-8-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Promoter analysis of T3-regulated genes. The promoter analysis revealed an overrepresentation of putative SREBP binding sites in the T3-regulated group compared to the control group. The figure shows the expected number of SREBP binding sites in each region of the promoter as estimated from the control group (solid line), and the observed number in the T3-regulated group (dotted line). A Chi-square test showed very significant (p = 0.0001) overrepresentation of SREBP sites in the region surrounding the transcriptional start site. The genes included in this analysis are indicated with a \* in Table 6. ::: ![](1472-6793-5-8-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Effects of GH and T3 on SREBP expression measured by real-time PCR. Two hours after thyroid hormone treatment of hypothyroid mice, SREBP1a expression is down-regulated, while SREBP1c and 2 are unaffected. Expression of SREBP1a, 1c and 2 is not affected after 7 days of bGH infusion in young male rats. \* p \< 0.005. ::: ![](1472-6793-5-8-3) ::: ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Effects of GH and T3 on SREBP1 processing. A Western blot showing hypothyroid mice livers with or without a T3-injection 2 hours prior to measurement. In the T3-treated samples, there is a decreased concentration of high molecular weight (ER bound) SREBP1, indicating an activation of its proteolytic processing by T3. ::: ![](1472-6793-5-8-4) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Experiments included in the study. ::: species age sex tissue Control sample Tester sample (hormone, time, dose) ------------- ---------- --------- ------------- -------------------- --------------------------------------------------- rat 3 months M liver c.i. of vehicle c.i. of bGH, 1 week, 5 μg/h rat 2 years M liver c.i. of vehicle c.i. of hGH, 3 weeks, 0.34 μg/g body weight/day rat 2 months M hepatocytes medium bGH added to medium, 24 hours, 100 ng/ml rat 3 months M/F liver male female rat 3 months M liver vehicle 17-α-ethinylestradiol, 1 day, 5 mg/kg body weight mouse 3 months M liver hypothyroid injection of T3 & TT4, 2 hours, 5 μg T3 + 5 μg T4 Expression profiles from six independent studies regarding gender differences and hormonal regulation of hepatic gene expression were included in the study. bGH = bovine growth hormone, hGH = human growth hormone, T3 = triiodothyronine, T4 = thyroxine, c.i. = continuous infusion, hepatocytes = primary rat hepatocytes. All rat experiments used Sprague Dawley rat strain. The mice strain used to study thyroid hormone actions was a hybrid of 129/Sv X C57Bl76J. ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### The primers, amplicon sizes and annealing temperatures used for gene expression measurements by real-time PCR. ::: gene left primer right primer size temp ---------- ---------------------- ---------------------- ---------- ---------- FAT/CD36 GCAACAACAAGGCCAGGTAT TGTGGCTGAGCAGAAAGAGA 200 54 Ppp3ca GCAGGCTGGAAGAAAGTGTC AAGGCCCACAAATACAGCAC 200 54 Hsd11b1 TTTTGCAGAGCGATTTGTTG TGCTCAGGACCACATAGCTG 200 54 Phyh TACGTGGAGTGCTTCACTGG CCATTGTTCCTGTCGATGTG 200 54 Srebp1a GCGCCATGGACGAGCTG TTGGCACCTGGGCTGCT 200 57 Srebp1c GGAGCCATGGATTGCACATT GCTTCCAGAGAGGAGGCCAG 200 54 Srebp2 CCCTTGACTTCCTTGCTGCA GCGTGAGTGTGGGCGAATC 200 ^54^ The expression of phytanoyl-CoA hydroxylase (Phyh), hydroxysteroid 11-beta dehydrogenase 1 (Hsd11b1), the catalytic subunit of protein phosphatase 3, alpha isoform (Ppp3ca), fatty acid translocase/cd36 antigen (FAT/CD36) and sterol regulatory element-binding proteins (SREBPs) 1a, 1c and 2 were measured using the primers shown in the Table. To allow comparison between samples, the expression data for each of the genes was normalized to the expression of glyceraldehyde-3-phosphate dehydrogenase (GAPDH), which was always run in parallel. ::: ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Comparisons of GH response in liver and in primary hepatocytes ::: Genes in vivo hepatocytes --------- --------- ------------- ------ ------ Phyh 0.34 0.6 0.54 0.6 Hsd11β1 0.31 0.28 0.63 0.5 Ppp3ca 1.33 1.43 1.48 1.33 CD36 2.39 2.78 1.26 2.07 Microarray and RT-PCR expression measurements for phytanoyl-CoA hydroxylase (Phyh), hydroxysteroid 11-beta dehydrogenase 1 (Hsd11β1), the catalytic subunit of protein phosphatase 3, alpha isoform (Ppp3cα) and fatty acid translocase/CD36 antigen (FAT/CD36). The real-time PCR results were normalized to GAPDH. The ratios between treated and untreated samples in vivo(young male rats treated with bGH or vehicle for 1 week via minipumps, n = 4 per group), and the ratios between GH-treated (24 hours) and untreated primary hepatocytes are shown. ::: ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Genes with a gender-differentiated expression pattern that were also affected in male rats continuously infused with GH. ::: Unigene Accession Name F/M GH/untreated ------------------------------------------------------------------ ------------ -------------------------------------------------------------------------------------------------- ------ -------------- Female predominant and up-regulated by continuous infusion of GH Rn.3211 AW917574 similar to TNF ligand superfamily member 12 4.61 8.14 Rn.14535 AI070587 carboxylesterase 2 (intestine, liver) 3.54 4.57 Rn.53990 AJ302031 alpha-1-B glycoprotein 7.08 3.60 Rn.2586 NM\_031572 Cytochrom P450 15-beta gene 4.72 3.08 Rn.115975 AW916713 EST sequence 5.61 2.85 Rn.2011 AA818134 peroxiredoxin 3 8.75 2.78 Rn.107116 AW142960 EST sequence 4.66 2.77 Rn.101709 AA819200 similar to Alcohol sulfotransferase (Hydroxysteroid sulfotransferase) 9.73 2.71 Rn.92406 AA819605 rat senescence marker protein 2A gene, exons 1 and 2 9.17 2.67 Rn.11377 AA875291 HRAS like suppressor 2.13 2.61 Rn.3790 L19658 cd36 antigen 5.81 2.39 Rn.91378 AA818024 sulfotransferase, hydroxysteroid preferring 2 7.53 2.35 Rn.32282 AI045872 arginine vasopressin receptor 1A 1.86 1.99 Rn.1292 AA858662 tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide 2.20 1.99 Rn.102325 AW917611 EST sequence 5.41 1.94 Rn.4000 X74402 guanosine diphosphate dissociation inhibitor 1 2.38 1.84 Rn.91122 AA858966 Cytochrome P450, subfamily IIC6 2.28 1.84 Rn.6946 AW140722 ferredoxin 1 2.28 1.79 Rn.8195 AW917572 EST sequence 3.61 1.68 Rn.106771 AA996745 similar to stromal interaction molecule 1 1.94 1.55 Rn.2180 CF108424 Atp5g3: ATP synthase, H+ transporting, mitochondrial F0 complex, subunit c (subunit 9) isoform 3 1.86 1.53 Male predominant and down-regulated by continuous infusion of GH Rn.1647 AF037072 carbonic anhydrase 3 0.06 0.09 Rn.37424 X79320 testosterone 6-beta-hydroxylase 0.10 0.15 Rn.23348 NM\_031332 solute carrier family 22, member 8 0.10 0.16 Rn.103016 J00737 alpha-2u globulin PGCL1 0.01 0.18 Rn.106677 AA851893 similar to nucleoporin 37; nucleoporin Nup37 0.15 0.18 Rn.103770 CA504514 similar to Gelsolin precursor, plasma 0.19 0.22 Rn.888 AA819595 hydroxysteroid 11-beta dehydrogenase 1 0.12 0.31 Rn.7279 AF121345 phytanoyl-CoA hydroxylase (Refsum disease) 0.44 0.34 Rn.11320 CF110333 phosphoribosyl pyrophosphate synthetase 2 0.33 0.35 Rn.20403 CB805116 ectonucleotide pyrophosphatase/phosphodiesterase 2 0.22 0.36 Rn.106064 AW141056 similar to another partner for ARF 1 0.61 0.43 Rn.43232 AI045953 cysteine-sulfinate decarboxylase 0.27 0.54 Rn.22952 AW140875 putative homeodomain transcription factor 1 0.59 0.59 Rn.801 AW144321 EST sequence 0.48 0.59 Rn.6835 AA963739 similar to Putative lysophosphatidic acid acyltransferase 0.48 0.59 Rn.12345 AW916917 similar to RIKEN cDNA 6330575P11 0.48 0.64 Rn.93760 AA998734 glutathione S-transferase, mu 1 0.45 0.65 Male predominant but up-regulated by continuous infusion of GH Rn.29771 AA900486 ATP citrate lyase 0.46 2.61 Rn.10992 AA964628 glucose-6-phosphatase, catalytic 0.48 2.10 Rn.98269 NM\_145878 fatty acid binding protein 5, epidermal 0.63 2.03 Rn.9486 X62888 fatty acid synthase 0.56 2.36 Differentially expressed genes were defined using SAM statistics, with a 5% false discovery rate as cutoff. An additional criterion was for the gene to have a mean ratio of at least 1.5. The table shows UniGene ID, GenBank accession number, gene name, and the median expression ratio (female/male or GH treated/untreated male). ::: ::: {#T5 .table-wrap} Table 5 ::: {.caption} ###### Genes with a gender differentiated expression pattern that were also affected in male rats treated with 17-α ethinyl-estradiol (EE). ::: UniGene GBAccession Gene Name F/M EE/untreated ----------------------------------------------------------- ------------- ------------------------------------------------------------------------------------------ ------ -------------- Female predominant and up-regulated by ethinylestradiol Rn.40365 AA819200 hydroxysteroid sulfotransferase subunit 9.73 3.39 Rn.40124 AA819605 Rat hydroxysteroid sulfotransferase a (STa) mRNA, complete cds 9.17 3.13 Rn.2011 AA818134 Peroxiredoxin 3 8.75 4.34 Rn.2151 AA818024 Rat hydroxysteroid sulfotransferase mRNA, complete cds 7.53 3.93 Rn.53990 AJ302031 Rattus norvegicus mRNA for putative alpha 1B-glycoprotein (ORF1) 7.08 1.53 Rn.102325 AW917611 Similar to Kruppel-like factor 7 (ubiquitous) 5.41 1.50 Rn.2586 NM\_031572 Rattus norvegicus Cytochrom P450 15-beta gene (Cyp2c12) 3.89 1.74 Rn.4000 X74402 R. norvegicus rab GDI alpha mRNA 2.38 1.77 Rn.1247 AA818043 cytochrome P450, 2c39 2.31 5.72 Rn.7245 AA858966 Rat cytochrome P450 PB1 (PB1 allele) mRNA 2.28 3.39 Rn.17105 AI029316 ESTs, Highly similar to tetrahydrofolylpolyglutamate synthase 1.88 3.74 Rn.32282 AI045872 R. norvegicus mRNA for V1a arginine vasopressin receptor 1.86 1.79 Rn.2382 AA964489 R. norvegicus mRNA for C-CAM2a isoform 1.79 3.47 Rn.13801 AW142659 Amphoterin induced gene and ORF 3 1.69 1.78 Rn.23741 AA900073 ESTs, Weakly similar to Ser/ Thr protein phosphatase 5 1.59 2.35 Female predominant and down-regulated by ethinylestradiol Rn.1292 AA858662 Tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide 2.20 0.18 Rn.756 AA859785 Rat alcohol dehydrogenase (ADH) mRNA, complete cds 1.88 0.61 Rn.19721 U73174 glutathione reductase mRNA 1.64 0.43 X12367 glutathione peroxidase I 1.69 0.56 Male predominant and down-regulated by ethinylestradiol Rn.1647 AF037072 carbonic anhydrase III 0.06 0.06 Rn.37424 U09742 CYP3A2 (testosterone 6-Beta-hydroxylase) 0.12 0.39 Rn.106677 AA851893 Similar to nucleoporin 37 0.15 0.16 Rn.102461 X16417 beta-globin 0.21 0.66 Rn.107334 NM\_013096 Hemoglobin, alpha 1 (Hba1) 0.24 0.64 Rn.107335 AW142257 2-alpha globin; alpha-2-globin chain; hemoglobin alpha chain 0.28 0.59 Rn.100762 XM\_235562 platelet-derived endothelial cell growth factor 1 0.42 0.33 Rn.7279 AF121345 peroxisomal phytanoyl-CoA hydroxylase (PHYH) 0.44 0.35 Rn.29771 AA900486 Rat ATP citrate-lyase mRNA, complete cds 0.46 0.32 Rn.15755 AW918421 EST 0.47 0.45 Rn.17644 AA817759 Peroxisomal Ca-dependent solute carrier-like protein 0.47 0.24 Rn.1086 AA817745 Adenylate kinase 4 0.52 0.57 Rn.5819 AA900928 Glutamic-oxaloacetic transaminase 1, soluble 0.53 0.58 Rn.9486 X62888 fatty acid synthase (EC 2.3.1.85) 0.56 0.34 Rn.48821 NM\_012624 pyruvate kinase (L-type) 0.61 0.48 Rn.106064 AW141056 Similar to another partner for ARF 1 0.61 0.66 Rn.15739 M13508 apolipoprotein A -- IV 0.62 0.53 Rn.10985 AW140851 choline kinase R; choline kinase R1 0.63 0.53 Rn.10389 AI058887 Rat p450Md mRNA for cytochrome P-450 0.65 0.54 Male predominant and up-regulated by ethinylestradiol Rn.66254 AA817793 Glucose-6-phosphatase catalytic subunit 3 0.68 1.55 Rn.888 AA819595 Hydroxysteroid dehydrogenase, 11 beta type 1 0.12 3.17 Differentially expressed genes were defined using SAM statistics, with a 5% false discovery rate as cutoff. An additional criterion was for the gene to have a mean ratio of at least 1.5. The table shows UniGene ID, GenBank accession number, gene name, and the median expression ratio (female/male or EE/Control male). ::: ::: {#T6 .table-wrap} Table 6 ::: {.caption} ###### T3 and GH have overlapping effects on gene expression in liver. ::: Accession Name T3/untreated young\_GH old\_GH --------------- ------------------------------------------------------------------------------------ ------------------ --------------- ------------- ---- ---- AW140999 cytochrome P450 4A3 9.77 0.88 0.53 AA956687 malic enzyme 1 7.86 2.21 2.64 \* \# BC029693 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1 6.82 1.42 1.79 NM\_145878 fatty acid binding protein 5, epidermal 6.59 2.02 1.46 \* \# K01934 thyroid hormone responsive protein 5.69 1.50 1.31 AW142176 similar to Igh-6 protein 5.20 NA 1.70 AW140621 similar to coenzyme A diphosphatase 4.94 0.90 1.37 \# AA925003 cytosolic acyl-CoA thioesterase 1 4.64 1.12 1.63 \* AW916988 sterol-C5-desaturase (fungal ERG3, delta-5-desaturase)-like 4.51 1.43 1.50 \# AW142891 EST 4.41 1.10 1.18 AA924590 FXYD domain-containing ion transport regulator 2 4.36 1.02 0.69 BF281499 signal transducer and activator of transcription 5B 4.29 NA 1.07 AA819712 EST 4.22 NA 1.13 X16417 hemoglobin beta chain complex 4.13 0.91 1.25 M34477 farensyl diphosphate synthase 3.96 2.01 1.50 \# NM\_013096 hemoglobin, alpha 1 3.79 0.92 1.39 \* AW916676 EST 3.79 0.91 1.53 AW140468 hypothetical protein LK44 3.70 1.37 1.01 \* AA923919 cathepsin E 3.63 NA NA AW916686 similar to HT021 3.63 0.89 1.68 \* AW142204 peptidyl arginine deiminase, type 4 3.59 0.92 1.44 \* AA860063 similar to Glutathione S-transferase, theta 3 3.56 1.14 0.75 X62888 fatty acid synthase 3.55 2.18 1.74 \* \# AW142682 podocalyxin-like 3.52 NA NA AA957248 fatty acid elongase 1 3.50 1.08 1.20 \* AA818305 similar to putative NAD(P)H steroid dehydrogenase 3.44 1.26 1.45 \# AA956747 fatty acid desaturase 2 3.33 1.11 1.51 \* AW918434 hypothetical protein RMT-7 3.31 0.90 0.39 \* AA997956 2,3-oxidosqualene: lanosterol cyclase 3.29 1.63 1.43 \* CB736793 NADH dehydrogenase (ubiquinone) 1 alpha subcomplex 5 3.24 1.59 1.23 \# AW141045 multiple inositol polyphosphate histidine phosphatase 1 3.21 0.92 1.38 \* AW142257 similar to hemoglobin alpha chain 3.19 0.93 1.45 \* AW142786 Na+ dependent glucose transporter 1 3.17 0.98 1.22 AA858817 EST 3.05 NA 0.94 M31672 insulin-like growth factor binding protein 2 2.95 NA 1.08 \* AA998020 protein O-mannosyltransferase 1 2.95 NA NA AW913874 adipose differentiation-related protein 2.93 1.61 0.95 \* AA924800 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1 2.88 NA NA AA819200 similar to Alcohol sulfotransferase (Hydroxysteroid sulfotransferase) (ST) (ST-60) 2.86 3.01 2.44 \# AI045953 cysteine-sulfinate decarboxylase 2.83 0.54 0.61 NM\_012598 lipoprotein lipase 2.78 NA NA \* AI029316 folylpolyglutamyl synthetase 2.72 1.31 1.07 \* AA817840 opioid receptor, sigma 1 2.68 1.19 1.89 \* \# BF282623 EST 2.63 1.01 1.06 AI137633 Nrp: neuropilin 2.59 NA 1.16 AW914056 growth response protein (CL-6) 2.58 2.30 1.48 \# AI059131 2-4-dienoyl-Coenzyme A reductase 2, peroxisomal 2.56 NA NA M73714 aldehyde dehydrogenase family 3, subfamily A2 2.55 0.96 1.20 \* J02585 stearoyl-Coenzyme A desaturase 1 2.53 1.35 1.14 NM\_007760 carnitine acetyltransferase 2.51 NA 1.19 \* AA819496 hypoxia induced gene 1 2.49 NA 0.79 AW916795 occludin 2.49 0.94 1.01 BF281801 kinesin family member 1B 2.44 0.68 1.28 \* AF007775 aquaporin 8 2.42 1.21 1.86 \* \# AW916443 EST 2.32 0.95 0.92 AW915619 inositol 1,4,5-triphosphate receptor 1 2.32 0.90 1.32 \* AB012933 fatty acid Coenzyme A ligase, long chain 5 2.30 2.55 1.79 \# AA964037 phospholipase A2, group VI 2.26 NA NA AI044427 similar to myotonic dystrophy protein kinase 2.17 NA 1.20 \* AW913988 GTP cyclohydrolase 1 2.17 0.83 1.29 \* M95591 farnesyl diphosphate farnesyl transferase 1 2.16 1.02 1.15 \* AA998726 nasal embryonic LHRH factor 2.12 NA 0.92 \* AW916626 EST 2.09 1.07 0.90 AW140633 hemoglobin Y, beta-like embryonic chain 2.08 0.97 1.37 AA926032 enoyl coenzyme A hydratase 1 2.05 0.91 1.19 \* AA925091 fatty acid binding protein 4 2.01 NA NA AA925731 isocitrate dehydrogenase 1 2.00 NA NA \* All of the genes in this list were up-regulated by T3 in hypothyroid mice (ratio repeatedly larger than 2). Clones that did not map to any Unigene cluster were removed. The genes that were included in the promoter analysis, where mouse and human orthologs could be found, are indicated with a \. The genes that were up-regulated by continuous infusion of GH in at least one of the in vivo*models in a statistically significant manner are indicated with a \#. The table shows Unigene ID, GenBank accession number, gene name, and the median ratio (treated/untreated). \"NA\" means that the gene was flagged away in the analysis due to low expression in two or more hybridizations. The shown experiments are: T3: thyroid hormone treatment of hypothyroid mice for 2 hours; Young GH: infusion of bovine growth hormone (bGH) in young (3 months) male rats for 7 days; Old GH: infusion of human growth hormone (hGH) in old (2 years) male rats for 3 weeks. :::	PubMed Central	2024-06-05T03:55:59.658326	2005-6-13	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180834/", "journal": "BMC Physiol. 2005 Jun 13; 5:8", "authors": [ { "first": "Nina", "last": "Ståhlberg" }, { "first": "Roxana", "last": "Merino" }, { "first": "Luis Henríquez", "last": "Hernández" }, { "first": "Leandro", "last": "Fernández-Pérez" }, { "first": "Albin", "last": "Sandelin" }, { "first": "Pär", "last": "Engström" }, { "first": "Petra", "last": "Tollet-Egnell" }, { "first": "Boris", "last": "Lenhard" }, { "first": "Amilcar", "last": "Flores-Morales" } ] }
PMC1180835	Background ========== The dimensional structure of personality disorders is a growing area of research. Several studies have presented factor models over the past decade \[[@B1]-[@B4]\]. Advanced statistical methods have been applied, but a serious limitation is that it is impossible to disentangle the effects of the instrument used, sample characteristics, and measurement error. Any given study may find a particular factor structure, but will be unable to determine whether this factor structure is a reflection of the instrument\'s characteristics, the sample\'s particular characteristics, random error, or a general factor structure underlying personality disorders. These same limitations have applied to studies that have attempted to link personality disorders with models of normal personality \[[@B5],[@B6]\]. In order to overcome these limitations, O\'Connor \[[@B7]\] conducted a meta-analysis of personality disorders in order to produce a consensual factor structure based on different measures. He included 33 different published correlation matrixes of personality disorders in his analysis, using a wide array of measurement instruments and samples. He produced aggregated factors to produce what he called a consensus structure of personality disorders. He reported both 2, 3 and 4-component consensus solutions. The 2-component solution appeared to represent an internalizing/externalizing dimension, the 3-component solution represented low agreeableness, low extroversion and neuroticism, whereas the 4-component solution represented these three plus conscientiousness (represented only by obsessive-compulsive personality disorder). In the following, I shall refer to the low agreeableness factor as antagonism, following Blackburn \[\[[@B8]\], p. 61\], and to the low extroversion factor as introversion for simplicity. Substance abusers are a good target for the study of personality disorders, because of the high comorbidity of personality disorders and substance abuse \[[@B9]\]. The focus of the present study is validation of the DSM-IV section of the Danish translation of the DSM-IV and ICD-10 Questionnaire \[[@B10]\]. Validation in this case requires that its factor structure is identical to the factor structure of the Swedish original, and identical, or at least highly similar, to the factor structure of personality disorders in general. A secondary purpose of the study is to assess the latent dimensions underlying the relationship between self-reported traits representing personality disorder and observer reported traits representing the same personality disorders. Method ====== Participants ------------ A sample of clients from several Danish substance abuse treatment settings was included, with both inpatient and outpatient settings, rural settings and urban settings, methadone and none-methadone settings. A total of 238 different substance abusers (hereafter referred to as \"clients\") in treatment were administered the DIP-Q and rated by staff members. Of the clients, 28% were women, 68% were men, and for 4%, gender was not reported. The mean age of clients was 33.4 years (standard deviation \[SD\] = 9.4). Of the clients, 50% were from inpatient settings for substance abuse treatment, and 50% were from outpatient treatment settings for substance abuse. Each client was rated by a mean of 2.0 staff members (SD = 1.8). Clients who had only been seen in individual counseling or psychotherapy constituted 37% of the sample, 8% was rated only by staff members who had not provided individual counseling to the client, and the remaining 55% had been rated by staff members who had seen the client in a variety of settings, including both individual counseling and group settings. A total of 59 raters participated, 17 men and 40 women. Most ratings took place at face-to-face meetings with the author, but 36 clients were rated on forms, and subsequently mailed to the author by letter or conducted as telephone interviews. Most patients from inpatient settings (60%) and 17% from outpatient settings were rated by more than one staff members. In the canonical correlation analysis, the mean of the different raters\' ratings was used. The Swedish sample consisted of follow-up samples from two different treatment institutions (N = 133). One sample consisted of 60 substance abusers admitted for treatment around 1989 who were assessed at a 15 year follow-up in 2004, whose mean age in 2004 was 44.5 years (range:35 -- 66). The second sample was from a five-year follow-up of 54 alcohol abusers from a private inpatient treatment centre in Sweden. Gender and age was not available for this subsample. The remaining 19 subjects were adult males from a project for criminal youth (mean age: 30, range: 21--45). Instruments ----------- ### The DIP-Q The DSM-IV and ICD-10 Personality Questionnaire \[DIP-Q\] was used as the measure of personality pathology. The DIP-Q is a self-report questionnaire for screening for DSM-IV and ICD-10 PDs, plus schizotypal disorder in ICD-10 \[[@B10],[@B11]\]. The instrument is highly similar to other questionnaires measuring PDs, such as the SCID-IIQ, and the PDQ-R. It consists of 151 statements that must be rated as true or false and three self-rating scales: severity of current events, global assessment of functioning, axis V of the DSM-IV for past year and global assessment of functioning for recent weeks. The DIP-Q was constructed through a consensus process. First, four psychiatrists worked together select a range of statements considered representative of the diagnostic criteria for each personality disorder. These statements could answered in a true/false format. The representative statements were then reviewed and validated by a second set of independent psychiatrists \[[@B12]\]. A translation and an English version was made available from the Swedish authors. No details of this translation were available. Therefore, a new Danish translation was made based on the English and Swedish versions, and then compared with the original translation. Any discrepancies in the items were checked against the English and Swedish versions, and revised if necessary. Studies show indications of concurrent \[[@B11]\] and predictive \[[@B13]\] validity of the instrument. Responses are used to determine the presence of criteria for each PD. In the present study criteria counts are used as measures of personality disorder features. Clients (N = 238) filled in the DIP-Q typically after at least 1 week of treatment. The DIP-Q questionnaires were sent to the Centre for Alcohol and Drug Research, where they were scored by the author. ### Staff ratings A subset of the sample (N = 152) were also rated by staff members involved in their treatment. Rating scales were used for staff ratings, rating the degree of personality pathology on each of the 10 PDs in the DSM-IV on a scale from 0 to 100. Each PD was assessed with only one scale. Three keywords were added for each disorder as prompts (e.g.: Paranoid: Guarded, jealous, expecting malevolence). Inter-rater agreement for this instrument has been reported \[[@B14]\]. Raters rated the patients at staff meetings with the author present, or filled in the rating scales on their own and mailed them to the author. Raters were not shown the results of the questionnaire before they had completed their rating of the client. Thus, raters were blind to the results of the questionnaire. Statistical analyses -------------------- Following O\'Connor \[[@B7]\], principal components analysis was conducted for 4 components, and Varimax rotations were performed. The coefficient of congruence (congruence) was calculated for the entire factor solutions with both O\'Connor\'s consensus model and between the Danish and Swedish samples. The congruence is a commonly used statistic in factor analysis to test replicability of components, and it indexes the proportional similarity between two sets of loadings, and yields values that are essentially identical to the values produced by other indices. Standard interpretation of the congruence coefficient suggests that 0.90 indicates acceptable fit. In the O\'Connor meta-analysis, the range of congruence with consensus solution for each individual dataset went as low as 0.60, but the mean congruence was around 0.80 for individual datasets with the consensus solution. This, however, included instruments that were based on quite different conceptualizations of personality disorders, such as the first version of the MCMI, and some very small datasets. Based on these reports, it was decided that an congruence of 0.80 or higher was an acceptable replication of the factor structure for the present report. However, a congruence of 0.90 or higher is desirable. The same factor analysis was conducted on the staff ratings, and congruence between the solution found for the staff ratings and the DIP-Q, and between the solution found for the staff rating and the consensus model were calculated. However, beyond the factor structure, an equally important point is to what degree the same latent variables link the self-reported personality disorder traits and the staff reported severity of personality disorder. A canonical correlation analysis was conducted to assess the latent dimensions connecting self-reported personality disorder on the DIP-Q and staff reported personality disorder severity \[[@B15]\]. Canonical correlation analysis is an approach that assesses the overall relationships between 2 sets of variables. It is similar to factor analysis, in that it identifies synthetic variables (or latent dimensions) underlying observed variables, but does so by producing maximally correlated variables across to sets of variables. The canonical correlation analysis produces several statistics for the relationships between synthetic variables, and between observed and latent or synthetic variables: the canonical correlation coefficientrepresents the Pearson r relationship between the two synthetic variables on a given canonical function. A canonical function is a set of standardized canonical function coefficients (from two linear equations) for the observed predictor and criterion variable sets. This function is analogous to the components in a principal components analysis, in that it represents a dimension in the data that is uncorrelated with other dimensions. Standardized canonical function coefficientsare the standardized coefficients, and are similar to beta coefficients in regression analysis. A structure coefficient(r~s~) is the correlation between an observed variable and a synthetic variable. The analyses planned were as follows: 1\. A 4-component factor analysis was constructed for the Swedish and Danish samples of drug abusers. 2\. The congruence scores across samples of each of the components and the whole models were analyzed. First, each pair of components were compared, and then the pair with the maximum congruence was selected. Based on this, components received labels. 3\. The congruence of the loadings between the Danish sample and the consensus model were calculated, for both the whole model and each of the factors. 4\. A 4-component factor analysis was constructed for the staff ratings on the subset of the Danish sample of substance abusers who were rated by staff members. 5\. Canonical correlation analysis was used to assess the latent dimensions linking staff members\' reports with DIP-Q scales. The canonical correlation analysis was conducted along the guidelines of Sherry and Henson \[[@B15]\]. Results ======= Factor analysis --------------- The factor loadings as well as congruence statistics from the factor analysis are summarized in table [1](#T1){ref-type="table"}. Readers are referred to O\'Connor for the factor loadings from the consensus model \[[@B7]\]. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Factor loadings for the Danish DIP-Q and congruence scores ::: Antagonism Introversion Neuroticism Conscientiousness h^2^ ----------------------------- ---------------- ------------------ ----------------- ----------------------- ---------- Paranoid 0.46 0.51 0.24 0.06 0.53 Schizoid -0.07 0.86 0.09 0.10 0.76 Schizotypal 0.49 0.63 0.07 0.13 0.66 Antisocial 0.81 0.10 -0.04 -0.13 0.69 Borderline 0.70 0.29 0.39 0.07 0.73 Histrionic 0.74 -0.08 0.20 0.10 0.60 Narcissistic 0.67 0.17 0.01 0.40 0.65 Avoidant 0.05 0.44 0.74 0.18 0.78 Dependent 0.18 -0.02 0.91 0.04 0.86 Obsessive-compulsive 0.07 0.14 0.14 0.93 0.91 Congruence coefficients With Swedish sample 0.93 0.96 0.95 0.89 With Consensus model 0.96 0.89 0.93 0.90 Notes:Loadings greater than \\|.45\\| are in boldface. ::: The four-components solution showed good congruence between the Danish and the Swedish version (0.93), and good congruence between the Danish DIP-Q and the consensus solution (0.92). The congruence of the individual factors were mostly above the 0.90 cut-off. Two exceptions were noted: the conscientiousness factor was just under adequate congruence between the two samples, and the introversion factor was just below cut-off in similarity between the Danish DIP-Q and the consensus model. The lack of congruence between the DIP-Q introversion and the consensus sample was due mainly to the low negative loading of histrionic personality disorder on introversion. In the consensus model, histrionic PD had a high negative loading on introversion (-0.53). The factor structure for the staff ratings is presented in table [2](#T2){ref-type="table"}. Congruence with neither the DIP-Q nor the consensus model was as high as that found between the two versions of the DIP-Q, or between DIP-Q and the consensus model. The antagonism and conscientiousness components were both congruent with the predicted models, but the neuroticism and introversion factors had much lower congruence scores. The main differences were that the narcissism scale had a high negative loading on the neuroticism component, and that avoidant personality disorder had a lower loading on introversion, while antisocial had an unexpected loading on the introversion component. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Factor loadings for the staff ratings and congruence scores ::: Antagonism Neuroticism Introversion Conscientiousness h^2^ ------------------------- ---------------- ----------------- ------------------ ----------------------- ---------- Paranoid 0.46 0.40 0.49 0.30 0.70 Schizoid -0.03 0.36 0.76 0.14 0.73 Schizotypal 0.21 0.02 0.83 0.08 0.73 Antisocial 0.67 -0.13 0.44 0.01 0.67 Borderline 0.81 0.30 -0.03 -0.01 0.75 Histrionic 0.72 -0.12 0.07 0.13 0.55 Narcissistic 0.56 -0.47 0.28 0.27 0.69 Avoidant -0.30 0.76 0.26 0.10 0.75 Dependent 0.22 0.81 0.12 -0.06 0.73 Obsessive-compulsive 0.12 0.00 0.16 0.96 0.96 Congruence coefficients With DIP-Q 0.95 0.85 0.89 0.91 With consensus model 0.92 0.84 0.78 0.92 Notes:Loadings greater than \\|.45\\| are in boldface. ::: Canonical correlation analysis ------------------------------ Canonical correlation analysis was conducted to assess the dimensions that underlie the relationship between the DIP-Q and the staff ratings. The correlations between the individual DIP-Q scales and staff members\' ratings of patients are summarized in table [3](#T3){ref-type="table"}. The correlations are quite modest in size, which is typical of associations between ratings based on clinical impression and self-report \[[@B16],[@B17]\], and between self-report and observations from other informants \[[@B18]\]. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Canonical Solution for staff ratings predicting DIP-Q profiles ::: Function 1 (agreeableness) Function 2 (neuroticism) Function 3 (extroversion) h^2^ ---------------------- -------- ---------------------------- -------------------------- --------------------------- ------- ---------- ------------ ------- ----------- ------------ ---------- Variable Conv r Coef r~s~ r~s~^2^(%) Coef r~s~ r~s~^2^(%) Coef r~s~ r~s~^2^(%) DIP-Q Paranoid 0.15 -0.01 -0.23 5 0.05 0.39 15 0.23 -0.03 0 0.21 Schizoid 0.21 0.06 0.11 1 -0.11 0.19 4 -0.91 -0.82 67 0.72 Schizotypal 0.19 0.29 -0.14 2 0.19 0.35 2 0.12 -0.23 5 0.20 Antisocial 0.49 -0.65 -0.87 76 0.08 0.21 4 0.04 -0.06 0 0.80 Borderline 0.44 -0.42 -0.57 32 0.29 0.57 32 -0.25 -0.12 1 0.66 Histrionic 0.21 -0.13 -0.42 18 -0.36 0.06 0 0.25 0.22 5 0.23 Narcissistic 0.22 -0.23 -0.49 24 -0.02 0.10 1 -0.29 -0.15 2 0.27 Avoidant 0.29 0.25 0.21 4 0.55 0.76 58 -0.17 -0.11 1 0.63 Dependent 0.27 0.15 0.08 1 0.46 0.75 56 0.45 0.30 9 0.66 Obsessive-compulsive 0.03 0.01 0.08 1 -0.40 -0.04 0 0.30 0.04 0 0.01 R~c~ 0.61 0.55 0.47 Staff ratings Paranoid 0.15 -0.35 12 0.67 0.53 28 0.15 -0.03 0 0.40 Schizoid 0.02 -0.11 1 -0.31 0.04 0 -0.58 -0.54 29 0.31 Schizotypal 0.03 -0.29 8 -0.16 -0.06 0 -0.55 -0.46 21 0.30 Antisocial -0.69 -0.85 72 0.01 0.03 0 0.18 0.06 0 0.73 Borderline -0.48 -0.70 49 0.32 0.40 16 -0.18 0.03 0 0.65 Histrionic 0.19 -0.35 12 -0.13 -0.15 2 0.51 0.40 16 0.31 Narcissistic -0.30 -0.65 42 -0.35 -0.43 18 -0.22 0.01 0 0.61 Avoidant 0.08 0.32 10 0.30 0.53 28 -0.16 -0.32 10 0.49 Dependent 0.03 -0.03 0 0.22 0.62 38 0.26 0.00 0 0.39 Obsessive-compulsive 0.11 -0.08 1 -0.23 -0.23 5 0.54 0.33 11 0.17 Notes.Conv. r:Convergent correlation coefficient (simple bivariate correlation). Coef= standardized canonical function coefficient; r~s~= structure coefficient; r~s~^2^(%) = squared structure coefficient; h^2^= communality coefficient. R~c~is the canonical correlation coefficient. Structure coefficients (r~s~) greater than \\|.45\\| are in boldface. Communality coefficients (h^2^) greater than \\|.45\\| are in boldface. ::: A canonical correlation analysis was conducted using the 10 rating scales as predictors and the 10 DIP-Q scales as dependent variables. The analysis yielded 10 functions with squared canonical correlations (r~c~^2^) ranging from 0.38 to less than 0.001 for each successive function. These 10 functions will be referred to as functions 1--10 in the following. The full model across all functions was statistically significant using the Wilks\'s criterion (λ = .21, F(100, 971.3) = 2.35, p \< .001). Because Wilks\'s λ represents the variance unexplained by the model, 1 -- λ yields the full model effect size in an r^2^metric. Thus, for the set of 10 canonical functions, the r^2^type effect size was .79, which indicates that the full model explained nearly all, about 79%, of the variance shared between the variable sets. Significance testing of functions later than function 4 showed that these were not significant (F(36,608.8) = 1.26, p = 0.144). Only functions 1--3 explained more than 10% of the variance each (and 20% of remaining variance after the extraction of prior functions), and these are the only ones retained for further analysis. Inspection of table [3](#T3){ref-type="table"} shows functions 1--3 are highly similar in content to the three first factors derived from the principal components analysis, and can be labelled agreeableness, neuroticism and extroversion, respectively. The agreeableness function yielded a strong correlation between staff ratings and DIP-Q (R~c~= 0.61). The structure coefficients of the agreeableness factor are high for antisocial, borderline and narcissistic personality disorder in both the DIP-Q and the staff ratings (all\>0.45). The standardized coefficient was quite small for narcissistic personality disorder, indicating that this variable is somewhat redundant in this context. The canonical correlation coefficient for the neuroticismfactor was 0.55. In the neuroticism function, the dependent and avoidant personality disorder had strong structure coefficients with both methods, and borderline personality had a strong structure coefficient for the DIP-Q. The canonical correlation coefficient for the extroversionfactor was 0.47. This function had only one strong structure coefficient, schizoid personality disorder on the DIP-Q (coefficient = -0.82). Small positive coefficients were found for dependent and histrionic personality disorders, and in the staff reported version, schizoid and schizotypal personality disorder have strong negative loadings, a small negative loading was seen for avoidant personality disorder, and small positive loadings were found for histrionic and obsessive-compulsive personality disorder. Discussion ========== The present study showed that the factor structure of the DIP-Q is invariant over the Danish and Swedish version for samples of drug abusers. It has adequate congruence with a consensus factor structure derived from a meta-analysis of several different instruments. It was also found that a similar set of latent functions explain the relationship between the DIP-Q scales and staff reported personality pathology, although the observed correlations between the personality disorder scales and the staff ratings were small. These associations are important, because they link the factor structure observed to external and independent ratings of the subjects\' pathology. The factor structure of staff reported personality pathology was not as close to the consensus model as was the factor structure of the DIP-Q. The factor structure of staff reported pathology was only a close match on antagonism and conscientiousness. Staff reported neuroticism had a strong negative loadings on narcissism (curiously, similarly to the MCMI\'s high correlation between narcissism and neuroticism \[[@B19]\]. However, the congruence for neuroticism and introversion was still close to the mean of the values for congruence with varimax rotation reported in O\'Connor\'s meta-analysis \[\[[@B7]\], table 4\]. Similarity and differences between the canonical solution and the factor analysis solutions ------------------------------------------------------------------------------------------- Overall, the factor loadings and the canonical correlations are quite similar. The most striking difference is, of course, that the conscientiousness factor did not emerge in the canonical correlation analysis. However, obsessive-compulsive personality disorder, the only disorder with a strong loading on this factor in the presented factor analyses and in O\'Connor\'s meta-analysis, had only very weak inter-rater agreement \[[@B14]\], thus making it very hard for the two datasets to converge on this dimension. Also, this factor is statistically weak, because the latent dimension, maladaptive conscientiousness, is covered by only a single personality disorder in the DSM-IV system. It is, however, conceptually important. Generally, the principal components analysis extracted most of the variance in all 10 personality disorders, as is evident in the high communalities in table [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}, whereas much variance was unaccounted for in the canonical correlation analysis, as reflected in the low communalities in table [3](#T3){ref-type="table"}. This overall difference may reflect overall significant and possibly systematic differences between what people observe themselves and what others see in them \[[@B18]\]. In other words, much of what we experience may not be reflected in what others see, and much of what others see may not be reflected in how we describe ourselves. This may derive from lack of insight into one\'s behaviour: a severely narcissistic person may experience himself as healthy and extroverted; it may also stem from the difficulties inherent in inferring motives from overt behaviour: e.g., a person with schizoid personality disorder and a person with avoidant personality disorder may both appear withdrawn and become uncomfortable with close contact, but their inner world and perception of other people may differ substantially. Limitations ----------- The present study is limited to drug and alcohol abusers, and is limited by being based on convenience samples. However, previous research has not shown meaningful differences in the factor structure of personality measures, including personality disorder measures, between various types of samples \[[@B20]\]. Substance abusers have the advantage that all personality disorders are more common among substance abusers than in the general population, leading to more variance in such a sample \[[@B9]\]. Another limitation is sample size. With a larger sample, approximately 1500, factor analysis could be done on the items of the DIP-Q rather than the composite scales. However, while this is a limitation, it does allow comparison of the present analysis with the many analyses that have been conducted on personality disorders in a similar manner. Conclusion ========== The Danish DIP-Q has a latent factor structure in drug and alcohol abusers that is consistent with the original Swedish, and similar to the general factor structure in personality disorders. The latent structure is valid, in as far as it converges with staff members\' impression of the personality disorders of respondents. Abbreviations ============= DIP-Q: The DSM-IV and ICD-10 Personality Questionnaire PD: Personality Disorder SCID-II: The Structured Clinical Interview for the DSM, Axis II section Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= MH planned, designed and conducted the study, and wrote the manuscript. Pre-publication history ======================= The pre-publication history for this paper can be accessed here: <http://www.biomedcentral.com/1471-244X/5/24/prepub>	PubMed Central	2024-06-05T03:55:59.667168	2005-5-24	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180835/", "journal": "BMC Psychiatry. 2005 May 24; 5:24", "authors": [ { "first": "Morten", "last": "Hesse" } ] }
PMC1180836	This book by Leah Edelstein-Keshet, \"Mathematical Models in Biology,\" is a discovery that delighted me at once. It is simple to read and well organized, with basics of mathematics given in chapters separated from applications and examples. At first glance it might be taken more as a book for researchers already involved in the fields of biology, biophysics, and physical medicine rather than a text book for graduate students. To be sure, many of the biologists who are slowly discovering that mathematics is essential to understand modern biology and medicine, will find this book to be a guided tour to many of the mathematical fields connected to the life sciences. However, also students in the mathematical and physical sciences will find in this book a detailed guide to the field of mathematical modeling biological processes. Although some of the newest fields in mathematical modeling of biological systems are only cited in this text, the way the basic ideas in modeling are presented is an essential step for future studies. This major goal of the book is reached by always giving the basic tools of mathematics and calculus in short separate chapters or in boxes in the text. Few mathematical concepts are needed for a reader who is new to this field. Moreover, the author expends much effort in presenting and explaining the methods of mathematical modeling of biological problems. Many examples are discussed in the text, and numerous others are given as guided exercises at the end of each chapter. These features would actually allow the book to be used as an excellent text book in courses for biophysics or biology curriculums. But more than this, the text offers deep insights into higher mathematical concepts and methods that are extremely useful for active researchers. In summary, \"Mathematical Models in Biology\" by Leah Edelstein-Keshet is an essential step for students and researchers active in a wide variety of fields; from cell and molecular biophysics to classical biology.	PubMed Central	2024-06-05T03:55:59.671899	2005-6-7	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180836/", "journal": "Biomed Eng Online. 2005 Jun 7; 4:35", "authors": [ { "first": "Giberto", "last": "Chirico" } ] }
PMC1180837	The popular book by Drs. Enderle, Blanchard and Bronzino was now published in a second edition. The first edition, published on 1999, became the major textbook in many introductory courses given in the first or second year of Biomedical Engineering (BME) undergraduate programs. The second edition is aimed at serving the same purpose as the first one, i.e. to provide in depth and in breadth overview of the continuously growing field of BME. The dynamics of the field since the first edition was released is reflected in major changes in the second edition. Specifically, the authors kept the division into two gross parts: (i) core biomedical engineering areas (chapters 4--10: biomechanics, rehabilitation, biomaterials, tissue engineering, bioinstrumentation, sensors and biosignal processing), and (ii) biomedical technology (chapters 12--17: modeling, genomics, computational biology, imaging, lasers and optics). The additions to part (ii), biomedical technology, are a chapter on genomics and bioinformatics (chapter 13, which replaces \"biotechnology\" in the first edition), and a chapter on computational biology and complexity (chapter 14). The new chapter on genomics (chapter 13) is motivated by the recent sequencing of the human genome (as well as numerous viruses, microbes, eukaryotes, yeast and rice). The second new chapter on computational biology and complexity (chapter 14) includes examples of cellular process models in individual cells, as well as in cell populations and systems. On the other hand, the texts on imaging were reduced and condensed, so that the technologies of ultrasound and MRI, each occupying a separate chapter in the first edition, are now surveyed under one chapter of \"medical imaging\". The major parts (i, ii) follow background of basic anatomy, physiology and cell theory (chapter 1, which, as in the previous edition, serves a limited purpose of providing the terminology used in later chapters), and of moral and ethical issues (chapter 2). The chapter on moral and ethical issues had been extended and now includes practical sections on marketing medical devices in the US, and on the role of biomedical engineers in the process for FDA approval. Real and hypothetical case studies were also added here, to illustrate ethical issues, patient privacy concerns, and medical liability questions. Overall, this remains an excellent textbook for BME students, and the progress in the field over the last 6 years is well reflected. Each chapter includes example problems with solutions and some 10--30 exercises. The list of suggested additional reading material, which concludes each chapter, was updated to cover literature published since the first edition was released. Figures are of good quality and are informative. Particularly useful is the new appendix on Matlab and Simulink software tools, which are required for solving some of the problems and exercises in this book. This not only contributes to the completeness, but also focuses the students on the computational abilities of these powerful software tools which are commonly used in BME work (e.g. solving polynomial and differential equations, plotting data, and simulating dynamic systems). My only reservation relates to the level of mathematics and basic engineering sciences (e.g. solid and fluid mechanics, electrical circuit analysis etc.) which is expected from students in the tasks provided. Given that an introduction to BME course is offered in many undergraduate programs during the first year of studies, students may be frustrated by not having the necessary background. In my introduction to BME course at Tel Aviv University, Israel, which was based on the first edition, I had the impression that students do not take full advantage of what this book has to offer, simply because they did not yet study differential equations, numerical methods, statistics, solid and fluid mechanics, and electrical circuits. In BME programs where an introduction to BME course is offered in the second year of studies, this issue may be resolved, but often the motivation in teaching an introduction to BME course during the first year is to provide students with the \"taste and flavor\" of BME while they are dedicating most of their time to mathematics, physics, biology, and basic engineering science courses. The second edition does not solve this conflict. In closure, despite the above reservation, this is certainly the most comprehensive textbook of its kind, and is recommended not only for undergraduate BME students but also for BME engineers in the industry or at the graduate level in the academia, as a reference book for a quick dive into new topics, or for an up-to-date survey of recent developments in this field.	PubMed Central	2024-06-05T03:55:59.672390	2005-7-19	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180837/", "journal": "Biomed Eng Online. 2005 Jul 19; 4:44", "authors": [ { "first": "Amit", "last": "Gefen" } ] }
PMC1180838	The health of the central nervous system (CNS) is a subject that impacts on all people and especially on those who have the misfortune to be afflicted by CNS injury or disease. The cerebrovascular barriers at the CNS capillaries and the choroid plexus play a vital role in maintaining the microenvironment of the cells in the brain and spinal cord. The stated aim of this book is to act as a stimulus for research directed towards minimizing the effects of CNS stress, injuries and insults. With seven sections that include 27 chapters from 59 contributors and 605 pages in total, this book covers many topics including: basic aspects of the barriers, in vitromodels, physiological transport mechanisms, neurochemical mediators, stress situations, changes in disease conditions and clinical aspects. Each chapter consists of a review, together with up-to-date original data. The chapters are laid out clearly in a consistent format with informative illustrations. This is an important wide-ranging book aimed at researchers in neuroscience, drug discovery and development, and neuropathology. Although large, it contains a wealth of information on contemporary research with an impressive 29-page index from arachadonic acidto zonula occludens proteins. The bulk of the text deals with research relating to the cerebrovascular barriers in the brain. However, in contrast with past books on the blood-brain barriers, this one contains several chapters concerned with the blood-spinal cord barrier. Considering the volume of research on spinal cord injury in recent decades, this serves as a useful source of information and a reminder that barrier function is important throughout the CNS. Furthermore, it reinforces the probability that information obtained from research on barriers in the brain can be applied to the therapeutic treatment of spinal cord injury. One rather disappointing aspect of this otherwise excellent book is the small amount of space devoted to the choroid plexus and the cerebrospinal fluid (CSF). This is the alternative route to the brain, the blood-CSF barrier with somewhat different properties to the blood-brain barrier. Only two chapters contribute to this very important aspect of brain homeostasis and to the potential for manipulation of the CSF for diagnostic and therapeutic purposes. Overall, \'Blood-Spinal Cord and Brain Barriers in Health and Disease\' is a contemporary and very informative volume that should be a ready source of reference for all researchers and clinicians concerned with the CNS in health and disease. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= Sole author	PubMed Central	2024-06-05T03:55:59.673106	2005-6-12	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180838/", "journal": "Cerebrospinal Fluid Res. 2005 Jun 12; 2:1", "authors": [ { "first": "Hazel C", "last": "Jones" } ] }
PMC1180839	Introduction ============ Attention-deficit hyperactivity disorder (ADHD) is characterized by a persistent and developmentally inappropriate pattern of inattention, hyperactivity, and/or impulsivity \[[@B1]\]. Children with ADHD tend to have difficulty organizing tasks and sustaining attention during schoolwork or play activities. Typical disruptive behaviors include failing to remain seated, talking excessively, playing noisily, and blurting out answers before questions have been completed. ADHD is relatively common, with prevalence rates among school-age children in the United States ranging from roughly 4% to 12%, depending on the diagnostic approach \[[@B2]-[@B4]\]. Furthermore, the percentage of children treated for ADHD in the United States increased dramatically from the 1980s to the 1990s \[[@B5],[@B6]\]. ADHD is associated with impairment in many areas of children\'s lives, including academic performance, social functioning, and overall quality of life \[[@B7]-[@B12]\]. Children with ADHD are frequently rejected by their peers as early as the first day of contact, likely as a result of their tendency toward disruptive and aggressive behavior \[[@B13]\]. ADHD also has long-term negative outcomes for many children, including decreased educational attainment, work performance, and occupational stability compared to individuals without ADHD \[[@B14],[@B15]\]. Because of the broad impact of ADHD, the disorder is likely to have serious economic implications for children, families, and society. Research has only recently begun to examine these economic costs, but the initial studies suggest that ADHD leads to increased costs in healthcare and other domains. The purpose of this paper is to review and summarize available literature on the economic costs of ADHD, as well as potential economic benefits of treating this condition. In addition, recommendations for additional research on the economic implications of ADHD are provided. Methods ======= A literature search was performed using MEDLINE, accessed through PubMed, to identify all published articles on the economic implications of ADHD. Searches were conducted using both the term and medical subject heading \"ADHD\" (in addition to the full term \"attention-deficit/hyperactivity disorder\" as well as individual words including \"attention\" and \"hyperactivity\"), along with economic terms including \"cost,\" \"costs,\" \"economic,\" and \"economics.\" Reference sections of relevant articles, including one review article that summarized eight economic studies \[[@B16]\], were reviewed to identify additional studies that may not have been included in MEDLINE. Finally, authors were contacted to locate conference abstracts and articles in press that were not yet indexed. In total, 22 relevant items were located including published original studies, unpublished conference presentations, and reports available on the Internet. Articles that discussed resource use patterns for ADHD but did not attach costs to these patterns \[[@B17]\] are not included in this review. All costs were updated to year 2004 US dollars based on the medical component of the Consumer Price Index \[[@B18]\]. Results ======= Direct Medical Costs of ADHD in Children and Adolescents -------------------------------------------------------- Formal cost of illness studies measure the \"economic\" burden resulting from disease and illness across a defined population (e.g., patients in the US), including both direct medical costs and indirect (e.g., lost productivity) costs \[[@B19]\]. Although no formal cost of illness estimates incorporating both direct and indirect costs of ADHD have been published, several studies have presented estimates of the direct medical costs of treating children and adolescents with ADHD (see Table [1](#T1){ref-type="table"}). In three of the studies \[[@B20]-[@B22]\], annual costs of care for children with ADHD were compared to annual costs for matched (on age and sex) controls. Three other studies included a similar comparison, but utilized non-matched controls \[[@B23]-[@B25]\]. Studies by Chan et al. \[[@B26]\] and Kelleher et al. \[[@B27]\] compared treatment costs for childhood ADHD with childhood asthma. Leslie et al. \[[@B28]\] examined trends over time in costs for children with ADHD, and Marchetti et al. \[[@B29]\] compared costs by type of ADHD treatment. Birnbaum et al. \[[@B30]\] used the data set previously analyzed by Swensen et al. \[[@B21],[@B22]\] to estimate total excess costs for the US population, defined as the difference between ADHD patients and matched controls. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Studies of the Direct Costs of ADHD ::: Citation Sample Data Sources (Time Period) Findings^1^ ----------------------- -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- ------------------------------------------------------------------------------------------- ---------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Birnbaum et al. 2005 Treated ADHD patients aged 7--44 (N = 1219) and their family members under age 65 (N = 3692); controls without ADHD matched to both patients (N = 1219) and family members (N = 3692) on age, gender, employment status, geographical location age, gender, state of residence, and employment status) Claims data from large Fortune 100 company (1996--1998) This study estimated total excess costs for the US population, defined as the difference between ADHD patients/family members and controls. Annual mean direct ADHD treatment costs (2004 US \$) were \$674/\$745 for girls/boys with ADHD (excess costs = \$0.53/\$1.06 billion) and \$412/\$529 for female/male adults with ADHD (excess costs = \$0.13/\$0.40 billion). Annual other direct treatment costs (2004 US \$) were \$865/\$990 for girls/boys (excess costs = \$0.80/\$2.0 billion) and \$2609/\$3022 for female/male adults with ADHD (excess costs = \$0.67/\$1.46 billion). Burd et al. 2003a Children aged 0--21 years with ADHD (N = 3,872) and non-matched controls (N = 95,119) without ADHD North Dakota Health Claims Database (1996--1997) Annual, mean direct treatment costs (2004 US \$) were \$870 for ADHD patients versus \$663 for controls; 1.9% of total annual health expenditures in North Dakota attributable to ADHD. Chan et al. 2002 Children aged 5--20 years with ADHD (N = 165), asthma (N = 322) or neither diagnosis (N = 4,952) Nationally representative household survey data (1996) Annual, incremental direct treatment costs (2004 US \$) were \$661 for children with ADHD (P\< 0.001) and \$603 for children with asthma (P\< 0.01) (relative to costs for children with neither diagnosis) Guevara et al. 2001 Children aged 3--17 years with ADHD (N = 2992) and matched (on age and sex) controls (N = 11,968) without ADHD Health maintenance organization in western Washington State (1997) Annual, incremental direct treatment costs (2004 US \$) were \$503 (95% CI: \$450--552) for children with ADHD alone and \$1088 (95% CI: \$899--\$1,304) for children with ADHD plus coexisting mental health disorders (relative to costs for children with no ADHD) Kelleher et al. 2001 Children aged 7--20 years with ADHD (N = 1,602) and with asthma (N = 1,411) Medicaid claims data for patients in Pittsburgh, PA and surrounding counties (1994--1995) Annual, mean (± SD) direct treatment costs (2004 US \$) were \$2,567 ± \$2,959 for the ADHD group versus \$2,382 ± \$2,664 for the asthma group (difference not statistically significant) Leibson et al. 2001 Children aged 5--19 years with ADHD (N = 309) and non-matched controls (N = 3,810) without ADHD Rochester, Minnesota medical facility-linked billing data system (1987--1995) Long-term (9 year), median direct treatment costs (2004 US \$) for ADHD patients compared with those without ADHD were more than double (\$6,158 vs. \$2,780; P\< 0.001), even for the subset with no hospital or emergency room admissions (\$183 vs. \$93; P \< 0.001) Leslie et al. 2001 Children aged ≤ 17 years with use of mental health services, including patients with hyperactivity (N\~10,000) Health care claims for privately insured population (MarketScan^®^) (1993--1996) Annual inpatient costs (2004 US \$) per treated hyperactive patient declined from \$26,550 in 1993 to \$8,644 in 1996 (P\< 0.001); there was also a significant decline in outpatient treatment costs (\$931 and \$659, respectively, P\< 0.001) Mandell et al. 2003 Children aged 3--15 years with ADHD (N = 4,306) and with no psychiatric disorder (N = 60,975) Medicaid claims data for patients in Philadelphia, PA (1993--1996) Long-term (3 year), mean direct treatment costs (2004 US \$) were \$4,891 for ADHD patients versus \$221 for patients with no psychiatric disorder Marchetti et al. 2001 Hypothetical cohort of school-aged children with ADHD Literature review, managed care survey, clinical experts (2000--2001) Average, total annual expected cost (2001 US \$) per treated patient was \$1,710 for Metadate CD, \$1876 for Concerta, \$2,061 for methylphenidate immediate-release/extended release (MPH IR/ER), \$2,122 for MPH IR, \$2,392 for Ritalin, and \$2,567 for Adderall. Secnik et al. 2005b Adults aged 18--64 with ADHD (N = 2,252) and matched controls without ADHD (N = 2252; matched on age, gender, metropolitan statistical area, and type of insurance coverage) Health care claims for privately insured population (MarketScan^®^) (1999--2001) Controlling for the impact of comorbidities, adults diagnosed with ADHD had significantly (P\< 0.0001) higher outpatient costs (\$3,009 vs. \$1.491), inpatient costs (\$1,259 vs. \$514), prescription drug costs (\$1,673 vs. \$1,008) and total annual medical costs (\$5,651 vs. \$2,771) compared with matched controls without ADHD Swensen et al. 2003 Children aged 0--18 years with ADHD (N = 1,086) and matched (on age, gender, and state of residence) controls (N = 1,086) without ADHD Claims data from large Fortune 100 company (1996--1998) Annual, mean (± SD) direct treatment costs (2004 US \$) were \$2,046 ± \$3,474 for the ADHD group versus \$703 ± \$2,215 for matched controls without ADHD (P\< 0.0001). Swensen et al. 2004 Individuals aged 0--64 with ADHD (N = 1,308) and matched (on age, gender, state of residence, and employment status) controls (N = 1,308) without ADHD Claims data from large Fortune 100 company (1996--1998) Annual, mean direct treatment costs (2004 US \$) were \$1,797 for children with ADHD versus \$577 for matched controls without ADHD (P\< 0.05); \$2,230 for adolescents with ADHD versus \$783 for matched controls without ADHD (P\< 0.05); and \$4,929 for adults with ADHD versus \$1,473 for matched controls without ADHD (P\< 0.05). ^1^All costs updated to Year 2004 US \$ using the Medical Services component of the Consumer Price Index (for US-based studies). For non-US studies, all country-specific costs first updated to Year 2004 currency values based on country-specific inflators; and then converted to Year 2004 US\$ based on currency exchange rates. MPH = methylphenidate IR = immediate release ER = extended release ::: A majority of the medical cost studies used insurance claims data from private insurers \[[@B20]-[@B22],[@B28],[@B30]\], from state Medicaid agencies \[[@B25],[@B27]\], or from both sources \[[@B23],[@B24]\]. The cost estimates reported by Chan et al. \[[@B26]\] were based on nationally representative household survey data, while the study by Marchetti et al. \[[@B29]\] relied on literature review and clinical expert opinion. The time period for the health care resource data used in the various studies ranged from 1987 to 1998. The results of the medical cost studies were consistent in indicating that children with ADHD had higher annual medical costs than either matched controls (difference ranged from \$503 to \$1,343) or non-matched controls (difference ranged from \$207 to \$1,560) without ADHD (Table [1](#T1){ref-type="table"}). The higher costs for ADHD patients were due to increased use of hospitalizations, primary care office visits, outpatient mental health visits, and pharmacy fills. For example, children with ADHD were 9.02 and 8.75 times more likely than other children (matched on age and sex) to have outpatient mental health visits and pharmacy fills, respectively \[[@B20]\]. When compared with cohorts of children with asthma (who were more likely to be female and African-American), children with ADHD had slightly higher annual treatment costs; but the differences were not statistically significant \[[@B26],[@B27]\]. When Birnbaum and colleagues \[[@B30]\] used claims data to estimate excess costs of ADHD across the US population, the excess ADHD-related treatment costs were \$0.53 billion for girls and \$1.06 billion for boys, and the excess overall healthcare costs were \$0.80 billion for girls and \$2.00 billion for boys. Marchetti et al. \[[@B29]\] examined the costs of six different ADHD drug therapies: generic and branded (Ritalin) MPH immediate release (IR) therapies, two branded MPH extended release (ER) therapies (Concerta and Metadate CD), generic MPH IR/ER, and a combination therapy of amphetamine salts (Adderall). The annual expected cost of treatment with each drug therapy, including costs incurred with physician visits and lab exams, was highest for Adderall at \$2,567 (2004 US \$) and lowest for Metadate CD at \$1,710. The cost for the other therapies ranged from \$2,061 (Concerta) to \$2,392 (Ritalin). Direct Medical Costs of ADHD in Adults -------------------------------------- Although ADHD is often perceived as a disorder of childhood and adolescence, there is growing awareness that many children continue to demonstrate symptoms in adulthood, although many adults with ADHD remain undiagnosed and untreated \[[@B31]\]. Limited prevalence data on adult ADHD are available, but estimates generally indicate that 30% to 70% of children with ADHD continue to have symptoms in adulthood \[[@B32],[@B33]\]. Adult symptom presentation is likely to be somewhat different from the common symptoms of childhood, with less hyperactivity, but with continued problems with organizational tasks and distractibility. ADHD is known to have a broad range of negative outcomes for adults, including relatively high rates of criminality, poor job performance, lower occupational status, problems in social skills, poor driving records, and comorbid psychiatric disorders \[[@B14],[@B15],[@B34]\]. Despite the continuing impact of ADHD in adulthood, only three studies were located that examined economic costs of adults with ADHD (Table [1](#T1){ref-type="table"}) \[[@B22],[@B30],[@B35]\]. Secnik et al. and Swensen et al. compared costs of care for adults with ADHD with matched controls (matched on age, sex, metropolitan statistical area, and type of insurance coverage), and both utilized privately insured claims data for the analysis. The two studies both indicated that adults with ADHD have significantly higher annual medical costs (ranging from \$4,929 to \$5,651) than matched controls (ranging from \$1,473 to \$2,771), even after controlling for patient comorbidity \[[@B35]\]. Birnbaum and colleagues used the same sample as Swensen et al. to estimate total excess costs for adults in the US population (i.e., the difference between ADHD patients and matched controls). The excess ADHD-related treatment costs were \$0.13 billion for women and \$0.40 billion for men, and the excess overall healthcare costs were \$4.79 billion for women and \$8.51 billion for men. Costs to Families ----------------- The relationship between children and their families is complex and bi-directional, involving simultaneous mutual influence \[[@B36],[@B37]\]. Children are shaped by their experiences with parents, while simultaneously influencing their parents\' behavior and emotions \[[@B38],[@B39]\]. This mutual influence between parents and children has been well-documented among families with a child who has been diagnosed with ADHD. Family environment and parent-child interaction has been shown to be a key causal factor in the development of ADHD and related conduct problems in longitudinal studies \[[@B40]\]. Conversely, the symptoms of ADHD have profound effects not only on the child, but also on the child\'s parents. For example, children\'s ADHD is frequently linked with strain in the parent-child relationship, disturbance in parents\' marital functioning (e.g., less marital satisfaction and more conflict than parents of children without ADHD), and extremely high parental stress \[[@B40],[@B41]\]. A recent study conducted by Swensen and colleagues \[[@B21]\] suggests that childhood ADHD also places an economic burden on parents and other family members. This analysis was conducted using 1996--1998 data from a national sample of over 100,000 beneficiaries of a large US company that include industrial, service, and professional employees. Family members of individuals affected with ADHD had 1.6 times as many medical claims as matched control individuals without a family member diagnosed with ADHD (matching based on age, gender, geographical location, and employment status). This greater healthcare utilization resulted in increased costs. Annual direct per-capita medical costs were twice as much for family members of ADHD patients (\$2,740) than for family members of control patients (\$1,365). Indirect costs related to disability and absenteeism followed a similar pattern (family members of ADHD patients, \$888; family members of controls, \$551). Birnbaum et al. \[[@B30]\] used the same data set to estimate excess healthcare costs across the US population (i.e., the difference between family members of ADHD patients and family members of matched controls), which were \$6.78 billion for family members of children with ADHD and \$12.10 billion for family members of adults with ADHD. There are several possible reasons for the higher indirect costs of parents whose children have ADHD. For example, children with ADHD are likely to require energy and attention that might otherwise be focused on work-related responsibilities. Furthermore, these parents may often be required to miss work in order to meet with teachers or take their children to appointments with physicians or mental health professionals. These high indirect costs suggest that ADHD has a financial impact not only on family members, but also on employers who might be affected by family members increased disability and absenteeism. A nationally representative US survey conducted in 1998 suggests that mothers of children with ADHD perceive the financial impact of their children\'s disorder \[[@B42]\]. Compared with mothers whose children were not diagnosed with ADHD, mothers of children with ADHD were 3.3 times as likely to say that the family could not afford prescription medication for the child and 7.4 times as likely to say the family could not afford mental-health care for the child. Another study of parent perceptions clearly demonstrates the substantial economic burden of ADHD \[[@B43]\]. In this study, parents were asked whether they perceive their child\'s hyperactivity as a serious problem. The strongest predictor of whether parents considered ADHD to be a serious problem was the financial impact related to work, defined as the impact of the child\'s behavior on either parent\'s employment patterns or chances of a career (e.g., leaving work to pick up the child). Compared with parents who did not think ADHD was a serious problem, parents who perceived their child\'s ADHD to be a serious problem were 17.6 times more likely to say that there child\'s ADHD had a financial impact related to their work. Taken together, Swensen\'s study of family medical costs and these two studies of parent perceptions indicate that ADHD has a substantial impact on family finances. Given the impact of a child\'s ADHD on parents\' absenteeism and productivity, it may be beneficial for employers and human resource specialists to consider strategies for minimizing this impact. A recent survey of 41 employers in four American cities found that the participants knew little about ADHD prevalence or its potential effects on parents, despite their responsibility for purchasing employees\' health insurance \[[@B44]\]. However, employers did offer several benefits and policies that could be helpful to parents whose children have ADHD, including on-site parent training programs, assistance with child care, flexible work/leave policies, and referral services that linked parents with community programs. When asked about benefits that could be targeted specifically for families of children with ADHD, employers suggested lunch seminars about ADHD and flexible hours for employees to meet with schools or physicians. As employers gain greater awareness of the economic impact of ADHD, it is hoped that such programs and benefits may be implemented at more companies. Costs of Criminality -------------------- Several longitudinal studies have shown that childhood ADHD is associated with criminality in adolescence and adulthood. For example, a study conducted in Los Angeles found that children diagnosed with ADHD between the ages of 6 and 12 years old had significantly higher juvenile (46% versus 11%) and adult (21% versus 1%) arrest rates compared to normal control subjects \[[@B45]\]. A similar study conducted in New York found that children with ADHD were more likely than controls to later be arrested (39% versus 20%), convicted (28% versus 11%), and incarcerated (9% versus 1%) \[[@B46]\]. Another study, conducted with 17--18 year old adolescents in San Francisco, found that the ADHD group was more likely than the control group to be on probation, in jail, or assigned to a social worker by the court \[[@B47]\]. One study has estimated the economic impact of criminality associated with ADHD \[[@B48]\]. Data were from a sample of children (4--12 years old) identified in 1979 and 1980. Follow-up interviews were conducted between 1991 and 1996 with 149 children diagnosed with ADHD and 76 control children when a sample ranged in age from 19 to 25 years. Criminal history was assessed through self-report, including crimes (e.g., stealing, assault), juvenile detention, probation, and jail. The costs of crimes incurred by victims and costs to the criminal justice system were estimated based on information from the Bureau of Justice Statistics, the Federal Bureau of Investigation, and the Criminal Justice Institute. Compared with the control group, the ADHD patients were more than twice as likely to have been arrested (48% versus 20%). The mean total criminal costs were dramatically greater for ADHD patients than for controls (\$12,868 versus \$498). All differences were statistically significant. Although this study should be considered a rough estimate, findings strongly suggest that criminality associated with ADHD results in a significant cost to society. Costs of Comorbidities ---------------------- Children with ADHD tend to have elevated rates of other psychiatric conditions \[[@B21],[@B49],[@B50]\] For example, about 30% of children with ADHD meet criteria for an anxiety disorder, compared to about 10% of the general population. Behavioral problems such as conduct disorder and oppositional defiant disorder are particularly common among children with ADHD, with comorbid rates of roughly 50%. Other conditions that are commonly comorbid with ADHD include learning disabilities, depression, and possibly bipolar disorder. When estimating the impact of ADHD, it is important to consider these comorbidities because comorbid conditions can influence children\'s long-term course and response to treatment. One study has estimated the incremental increase in costs of treatment for ADHD with comorbid conditions, compared with treatment of ADHD alone \[[@B51]\]. Analyses were conducted using 1996 and 1997 data from the North Dakota Department of Health\'s Claims Database. Generally, comorbid psychiatric disorders substantially increased the costs of treating children with ADHD. For example over the two years, comorbid depression increased costs by an average of \$358 per patient per year. Increases were also observed with oppositional defiant disorder (\$258), bipolar disorder (\$541), conduct disorder (\$488), anxiety (\$499), nondependent drug use (\$868), tics (\$198), and personality disorders (\$247). Non-psychiatric medical disorders also resulted in increased costs, including respiratory illness (\$630), acute sinusitis (\$670), general injuries (\$972), and allergies (\$507). In the North Dakota sample, the only comorbid conditions that were associated with lower costs compared to children without comorbid conditions were learning disabilities (-\$759) and epilepsy (-\$777). However, another recent study focusing on comorbid epilepsy in an administrative claims database from 1998 to 2001 found contrasting results \[[@B52]\]. In a sample, the average annual treatment costs for children with ADHD and comorbid epilepsy were \$5,194, compared with \$4,246 for children with ADHD but not epilepsy. Despite the mixed results relating to learning disabilities and epilepsy, initial findings suggest that the common comorbid conditions of ADHD may contribute to elevated treatment costs among this population. Costs of Accidents ------------------ Children with ADHD have been shown to be more accident prone than other children \[[@B53]\], likely because of their tendencies toward impulsive, overactive behavior. They are also more likely than other children to experience injuries due to accidents, such as broken bones, lacerations, head injuries, bruises, lost teeth, or accidental poisonings \[[@B54],[@B55]\]. One study has estimated the incidence and cost of accidents among individuals with ADHD using an administrative database of medical, pharmaceutical, and disability claims for national manufacturers\' employees, spouses, dependents, and retirees \[[@B22]\]. Analyses were conducted for the whole population, adults alone, children under age 12, and adolescents aged 12 to 18 years. ADHD patients in all age groups were more likely than a matched control group to have at least one accident claim: children, 28% versus 18%; adolescents, 32% versus 23%, and adults, 38% versus 18%. Among adults, the accident-specific direct medical costs were significantly higher among ADHD patients than among the control group (\$642 versus \$194). Among children and adolescents, there were not significant differences in accident-specific costs between the ADHD groups and the control groups. Costs of Work Loss ------------------ ADHD is associated with work-related problems in adulthood such as poor job performance, lower occupational status, less job stability, and increased absence days in comparison to adults without ADHD \[[@B15],[@B34],[@B35],[@B56]\]. This poor performance and work loss is likely to have profound economic implications. One study quantified this impact by estimating the excess costs (i.e., the difference between adult ADHD patients and matched controls) related to work loss \[[@B30]\]. Indirect work loss costs were calculated based on employer payments for disability claims and imputed wages for medically-related work absence days (e.g., days in the hospital, physician visits). The excess costs were \$1.20 billion for women with ADHD and \$2.26 billion for men with ADHD. Cost-Effectiveness of Treatments for ADHD ----------------------------------------- Three published studies have utilized decision analytic modeling techniques to assess the cost-effectiveness of drug therapy (i.e. methylphenidate \[MPH\]) for ADHD \[[@B57]-[@B59]\] (see Table [2](#T2){ref-type="table"}). Each of these studies represented \"complete\" economic evaluations (i.e. estimated both the incremental costs and incremental effects associated with treatment). In two of the studies \[[@B57],[@B58]\], the effectiveness of treatment was measured in terms of \"quality-adjusted life years\" (QALYs), an outcome measure that incorporates quality of life benefits and time \[[@B60]\]. These quality of life benefits are quantified using utility scores, which have been shown to be feasible and valid for assessment of children with ADHD \[[@B61]-[@B63]\]. In the third study \[[@B59]\], treatment effectiveness was based on gains in the Conners Teacher Rating Scale, a commonly used teacher-report questionnaire for assessing children\'s classroom behavior \[[@B64]\]. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Studies of the Cost-Effectiveness of Treatment for ADHD ::: Citation Comparators Study Design Findings^1^ ----------------------------------------------------------------- ----------------- -------------------------------------------------------- --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- Gilmore & Milne 2001 MPH, placebo • Decision analytic model assessing cost-utility Cost per each additional QALY gained with MPH treatment (versus no treatment) ranged from \$15,509 to \$19,281 when considering short-and medium-term benefits of MPH. Cost per QALY gained ranged from \$9,850--\$59,101 in sensitivity analyses Novartis data on file (2000; referenced in Lord & Paisley 2000) MPH, placebo • Decision analytic model assessing cost-utility Cost per each additional QALY gained with MPH treatment (versus no treatment) was \$27,766. Zupancic et al. 1998 MPH, placebo • Decision analytic model assessing cost-effectiveness Cost per each additional point in the Conners Teacher Rating Scale gained with MPH treatment (versus no treatment) was \$93, or \$560 for a 6-point (1 SD) gain. ^1^All costs updated to Year 2004 US \$ using the Medical Services component of the Consumer Price Index (for US-based studies). For non-US studies, all country-specific costs first updated to Year 2004 currency values based on country-specific inflators; and then converted to Year 2004 US\$ based on currency exchange rates. MPH = methylphenidate QALY = quality-adjusted life years ::: Overall, results of the three modeling analyses indicate that MPH is a cost-effective treatment option for children with ADHD. The cost per QALY gained in the Gilmore and Milne \[[@B57]\] study ranged from \$15,509 to \$19,281 when considering the short-and medium-term benefits of MPH. The authors note that evidence of cost-effectiveness beyond 6 months is poorer, and it is uncertain whether the effects of MPH persist into adolescence and adulthood. In the Novartis study, the cost per QALY gained was \$27,766. CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE RESEARCH =================================================== A growing body of literature has demonstrated that ADHD places a substantial economic burden on patients, families, and third-party payers. However, all available published studies on the economic implications of ADHD are relatively recent, and there are many additional questions to be examined in future research. Thus far, studies have identified several aspects of ADHD that are likely to have economic implications, including direct treatment costs, increased rates of comorbid psychiatric disorders, high accident rates, work loss, and criminality. There are many other well-documented outcomes of ADHD that are likely to have economic implications. For example, ADHD frequently has detrimental effects on a child\'s academic performance and behavior in school. These difficulties are likely to place an economic burden on school systems, as there is an increased need for school-based services such as in-school medication administration; special education services; child and possibly parent counseling; educational testing; development of individualized educational programs; and efforts to address disruptive classroom behaviors \[[@B65],[@B66]\]. Research is needed to quantify these costs and identify strategies for implementing the most cost-effective services. Furthermore, it is well known that adults with ADHD tend to have poor driving records and relatively high rates of traffic accidents \[[@B14],[@B15],[@B34]\]. These driving problems are also likely to present a significant cost which is not yet been examined. The international literature on cost of ADHD could also be expanded. Nearly all published studies identified for the current review were conducted in the US. Given international differences in medical care systems and practice patterns, it is difficult to apply the direct treatment costs from US studies to countries outside the US. It is likely, however, that the indirect cost burden of ADHD identified in US studies may be similar to the burden in other countries, although it may not be recognized to the same extent. Recognition, diagnosis, and treatment of ADHD are increasing in Europe and Australia, and future studies may document the economic burden of ADHD in these areas. Finally, more work is needed on determining the potential cost-effectiveness of the various treatment options for ADHD. The initial research on cost-effectiveness of treatment has focused on MPH, and results generally indicate that treatment of ADHD is cost-effective. As new treatments are introduced (e.g., the new non-stimulant atomoxetine), it is important to evaluate their cost-effectiveness to provide an indication of their potential value to clinicians, patients, families, and third-party payers. Effective treatments, while possibly increasing direct medical costs, are likely to reduce the overall burden of ADHD by controlling symptoms, improving children\'s functioning, and substantially reducing indirect costs to families. List of Abbreviations ===================== Attention-deficit hyperactivity disorder (ADHD) extended release (ER) immediate release (IR) methylphenidate (MPH) quality-adjusted life year (QALY) Competing interests =================== All three authors of this article are employed at the MEDTAP Institute at UBC, which is an independent health services research organization. Funding for the current review was provided by Eli Lilly & Co. Authors\' contributions ======================= LM was the principal investigator. He conducted the literature search, designed the structure of the manuscript, reviewed and integrated much of the literature, and wrote and conceived most of the manuscript. CP and MP had primary responsibility for the two sections that review, integrate, and summarize studies of direct costs and treatment cost-effectiveness. They reviewed the relevant literature, wrote these two important sections of the paper, and created Tables [1](#T1){ref-type="table"} and [2](#T2){ref-type="table"}. Acknowledgements ================ The authors thank Jodi Shorr for production and editorial assistance as well as Kristina Secnik for assistance in obtaining unpublished conference presentations.	PubMed Central	2024-06-05T03:55:59.673854	2005-6-9	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180839/", "journal": "Cost Eff Resour Alloc. 2005 Jun 9; 3:5", "authors": [ { "first": "Louis S", "last": "Matza" }, { "first": "Clark", "last": "Paramore" }, { "first": "Manishi", "last": "Prasad" } ] }
PMC1180840	Background ========== Tuberculosis (TB) is among the top ten causes of global mortality and morbidity accounting for about 26% of all preventable deaths \[[@B1],[@B2]\]. In Tanzania, more than 60,000 new TB patients are notified annually and TB is the third leading killer of adults behind Malaria and Acquired Immune-deficiency syndrome (AIDS) \[[@B3],[@B4]\]. The rapid annual increase of TB cases in the country is largely attributed to Human immunodeficiency Virus (HIV). The importance of TB/HIV co-infection as a major public health problem in Tanzania is not only due to its high incidence and mortality but also to its social and economic consequences. Tanzania is one of the several countries experiencing a reversal in human development mainly due HIV/AIDS and TB epidemics \[[@B5]\]. The majority of people with TB/HIV co infection are in the economic productive age group and the World Bank estimates that the gross domestic product (GDP) will be 15--20% lower in 2015 than it would have been without the AIDS pandemic \[[@B3],[@B5]\]. The internationally recommended DOTS (Directly Observed Treatment-short course) strategy has been implemented in the country for more than 18 years mainly in public health facilities with treatment directly observed by trained health staff. However, fewer than half of all health facilities in the country provide TB services. Efforts to expand DOTS to more health facilities have been limited by scarce resources and understaffing. This leads to overcrowding of patients especially in urban settings like Dar es Salaam, which notify about a quarter of all TB patients in the country. The rising number of TB cases and critical shortage of skilled staff has put considerable strain on the public health system. With a public health budget of about US \$ 6 per capita per year, the TB epidemic significantly influences the Tanzanian health system performance. Therefore, identifying new approaches to treatment that maintain effectiveness and put less demands to the meagre health resources by reducing pressure and lowering cost of delivering TB treatment becomes a priority. One such approach is community based direct observed treatment (DOT). Supervision of TB patients in the community has been piloted in a variety of settings with successful results \[[@B6]-[@B9]\]. However, little is known about the cost and cost effectiveness of this option in Tanzania. The economic analysis is important to assist policy makers in decision making and rational allocation of scarce health resources. We conducted the study to determine the cost and cost effectiveness of health facility and community based DOT in an urban setting in Tanzania. Methods ======= Study setting ------------- The study was conducted in Temeke district in Dar es Salaam city. Temeke is an urban district with low socio-economic indicators characterised by rapid population growth and overburdened public health services. The district has a population of about 700,000 people with a health facility to population ratio of about 1 per 5000. Tuberculosis and AIDS are leading causes of death in the district, representing 36.5% of years of life lost for all age groups \[[@B10]\]. Alternative strategies ---------------------- Two treatment options were compared; health facility based DOT option and community based DOT option using treatment supervisors. A brief description of the alternatives is given below, as full details are available elsewhere \[[@B6]\]. The conventional (health facility based DOT) approach to tuberculosis control in Tanzania presupposes initial diagnosis and treatment of patients on ambulatory basis, provided the patients are strong enough to follow their medication and are free from complications that need hospital care\[[@B11]\]. During the intensive two months period, patients visit health facility daily for direct observation of treatment by trained health personnel. This is followed by a 6-month continuation phase period where patients visit a health facility once per month. Before 2003 and during the study period, smear negative and extra pulmonary TB patients received intermittent drug regimen three times weekly. Currently all patients receive daily treatment In the new treatment option (community based DOT) direct observation of treatment was supervised by treatment supervisors rather than health facility personnel as in the conventional treatment. The supervisors were guardians and former TB patients. A guardian was defined as a family member or a close relative living with patient. Patients were assisted in the selection of a responsible, trusted guardian. Willing former TB patients who had successfully completed treatment and lived close to the patients\' home were selected by the district tuberculosis and leprosy co-ordinators (DTLC). Treatment supervisors collected drugs from health facilities once per week during the entire two months intensive period. During this period a health worker also made surprise visits to patients\' homes to check for adherence of treatment by reviewing treatment cards and pill count. Patients were also requested to report to the TB clinics fortnightly for clinical evaluation and monitoring of their progress. After the first two months, treatment was exactly as in the conventional health facility option. Study participants ------------------ The study was conducted alongside a randomised-controlled study \[[@B6]\]. To select patients to be included in this study, a systematic sampling method was used. A list of all patients who participated in the randomised trial was drawn from the TB registers. Patients were listed consecutively as they appeared in the TB register. Separate lists were prepared for patients who received community DOT and health facility DOT. From the lists every fifth patient was picked to participate in the present study. The corresponding supervisors of the patients allocated to community DOT were also interviewed. The study population consisted of tuberculosis patients over 5 years old who started treatment in the district. New smear positive and smear negative pulmonary tuberculosis as well as extra pulmonary tuberculosis patients were included in the study. Patients were excluded from the study, if they had been previously treated for TB and if they had severe illness that precluded ambulatory treatment. Patients were recruited from five diagnostic centres in the district with high caseload (more than 100 patients per year). The study was conducted between December, 2001 and January 2003. Costing ------- Costs were assessed from a societal perspective as advocated by current standards for cost-effectiveness analysis \[[@B12]-[@B14]\]. Average costs of each component of care and treatment was calculated as the quantity of resources used multiplied with their unit price. Costs associated with diagnosis were not included in the analysis, since approaches to diagnosis were the same for all options. Furthermore, costs associated with TB diagnosis would have been difficult to accurately establish because of the long delays in diagnosis and the fact that these costs were incurred before TB was confirmed. The exclusion of the cost of diagnosis in the study does not however influence the relative ranking of community DOT and health facility based DOT, but means that comparison with other health interventions such as malaria control cannot be done. Broadly, the costs were categorised as \"provider\" and \"personal\" or \"community\". Provider cost is associated with developing and operating a health care service. They include staff costs, supplies and equipment. Provider\'s costs were incurred by the health system, tuberculosis programme and the community TB project. Personal or community costs are those incurred by patients and treatment supervisors. They include direct and indirect costs. Direct costs are non-medical costs related to visiting the TB clinic such as transport costs, buying food and drinks when visiting a TB clinic. Direct medical costs such as drug costs are included as provider\'s costs because they were incurred by the TB programme. Indirect costs refer to the value of lost time by the patients and treatment supervisors to follow up TB treatment. This should not be confused with overheads of fixed costs used in accounting practices as pointed by some authors \[[@B12]\]. Costs of a visit to a TB clinic were calculated from one health facility (Mbagala) in Temeke. Joint costs (costs items which were used for more than one activity) were allocated according to the proportion of the time the cost item was used for TB activity. Capital costs were annualised using a discount rate of 10%. This has been used in other studies in Tanzania and has been recommended by the Bank of Tanzania in 2002 \[[@B15],[@B16]\]. Annualisation was done on the assumption that the expected useful life of buildings was 30 years, for vehicles and equipment 10 years and for motorcycles 5 years. The base year for valuing costs was 2002, and exchange rate applicable at that time was 967 Tanzanian Shillings to US \$1. Sources of data included budget and expenditure files for Temeke district, health facilities, the community TB project in Temeke and the National Tuberculosis and Leprosy Programme (NTLP). Others include salary scales for established positions, reports from Temeke municipal, Ministry of Health and the Bank of Tanzania. Face to face interviews were conducted with the staff of the health facilities, Temeke Municipal and Ministry of Health. Patients costs were estimated using a structured questionnaire. Patients were interviewed about travel and time costs associated with a visit to the TB clinic, their average monthly income and other costs associated with TB treatment and care. Time costs were converted to a monetary value based on the average reported income among interviewed patients (including all sources of incomes). Costs incurred by the treatment supervisors were assessed by a structured questionnaire by asking about the time and travel costs to collect drugs as well as supervising a patient to take medication, and other costs (incurred) including training and supervision of community DOT. Costs per patients treated in different treatment options (community VS health facility) were calculated as a weighted average of the costs according to TB type (smear positive, smear negative and extrapulmonary) Effectiveness ------------- Measure of effectiveness used was treatment success. Data on treatment outcomes were obtained from the randomised-controlled trial and operational definitions used in this study are explained in detail elsewhere \[[@B6]\]. Briefly, treatment success included patients who were cured and those who completed treatment. Cured patients were those with positive sputum smear before starting treatment and confirmed to be sputum negative at 7(or 8) months and at least one previous occasion. Completed treatment applied to: patients who had positive pre-treatment results, negative results at 2 months, and no end of treatment results; patients who had negative pre-treatment results and had been placed on treatment for clinical reasons, and patients who completed the full course of treatment, but had no pre-treatment or end-of-treatment bacteriological results. The term treatment success is used in routine practice to refer to smear positive patients who are cured and have completed treatment. It is worth noting that we have expanded this definition to include smear negative and extra pulmonary tuberculosis patients who completed treatment as well, since our aim is to compare alternative treatment strategies regardless of the patient\'s TB type. This operational definition has also been used by other researchers \[[@B17]\]. Estimates of cure rate that would apply without treatment (null intervention scenario) was calculated assuming a self cure rate of 20% in the absence of treatment for individuals who are not otherwise immuno-compromised, and a self cure rate of 0% for those who are HIV+\[[@B13],[@B18]\]. This can be calculated from the formula: {(estimated percentage of patients who are HIV + x 0) + (estimated percentage of patients who are HIV- x 20) }/100 Cost-effectiveness ------------------ Cost effectiveness was calculated as the average cost per patient treated successfully. This was done by dividing the total cost and patients successfully treated. Sensitivity analysis -------------------- One way sensitivity analyses were undertaken to assess the robustness of the results to changes in key input variables. The uncertainty analyses were based on most likely, minimum and maximum values. For valuing time cost, average reported income of the patients was used as the base-case estimates. Sensitivity analysis was conducted when time was valued as zero for all patients and when time was valued for patients with income only. For the effectiveness data, lower and upper boundaries of 95% confidence interval of the treatment outcome was used. Sensitivity analysis was undertaken with and without \'no treatment \'option (Table [3](#T3){ref-type="table"}) ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Influence of changes in key input variables on cost-effectiveness ratio in US \$ per patient successful treated ::: Input variable Input value Average CER community DOT^a^ Average CER health facility DOT^a^ --------------------------------------- ------------- ------------------------------ ------------------------------------ ------ ------ ------ Min. Max. Min. Max. Min. Max. Patient time loss^b^ 0 US\$0.29 117 130 168 203 Supporter time loss^b^ 0 US \$ 0.48 123 129 196 196 Discount rate 0 3% 125 125 186 188 Success rate(community DOT)^c^ 69% 78% 136 120 Success rate (health facility DOT)^c^ 62% 74% 281 235 Success rate(community DOT)^d^ 81% 89% 116 106 Success rate (health facility DOT)^d^ 79% 87% 184 167 ^a^The sensitivity analyses are based on base-case calculations ^b^time loss: minimum=assuming time is valued as zero; maximum= average income per hour among those with income only ^c^Success rate taking in consideration cure rate in \'no treatment\' scenario ^d^Success rate without cure rate in \'no treatment\' scenario ::: Results ======= Study participants ------------------ A total of 103 TB patients and 42 treatment supervisors were enrolled in the study. Among the patients 45 (44%) received DOT in the community and 58 (56%) received DOT in health facility. The majority of the patients were male 63 (61%) and the mean age of the patients was 32 years (median 30). Most of the patients 67 (65%) completed primary education and worked in informal sector or self-employed 43 (42%). The average reported income of the patients was USD 43. No significant difference (p \> 0.05) was found between patients treated in community and health facilities on social-demographic variables (Table [1](#T1){ref-type="table"}). The majority of the treatment supervisors were guardians (39, 91%) and females (27, 63%). Their mean age was 37 years (SD 12, median 38). ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### General characteristics of the patients ::: Community DOT n (%) Health facility DOT n (%) -------------------------------------------- ------------------------- ------------------------------- Sex Male 26 (57.8) 37 (63.8) Female 19 (42.2) 21 (36.2) Age (in years) Mean (SD) 31 (9) 33 (12) Education status No education 5 (11.1) 4 (6.9) Not completed primary school 9 (20.0) 7 (12.1) Completed primary school 25 (55.6) 42(72.4) Secondary school and above 6 (13.3) 4 (6.9) Others 0 1(1.7) Occupation Informal and self employed 21 (46.7) 22 (37.9) Employed 14 (31.1) 14 (24.1) No employment 5 (11.1) 7 (12.1) Housewife 5 (11.1) 15 (25.9) Home-TB clinic distance Below 1 km 29 (64.4) 46 (79.3) 1--2 kms 10 (22.2) 7 (12.1) Above 2 kms 6 (13.3) 5 (8.6) TB type Smear- positive PTB 25 (55.6) 33 (56.9) Smear- negative PTB and extra-pulmonary TB 20 (44.4) 25 (43.1) ::: Costs ----- The average cost of treating patients under the alternative DOT options is shown in figure [1](#F1){ref-type="fig"}. Overall, treating a patient under community DOT was less costly by 35% compared to health facility based DOT. The total cost of treating a patient under health facility DOT was US 145 compared to USD 94 for treating a patient under community DOT. Costs were reduced from all perspectives: providers (TB programme and community project) and community (patients and treatment supervisors). Community DOT reduced providers cost by 27%, patients cost by 72%, and combined patients and treatment supporter cost (community cost) by 55%. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Average cost per patient (all patients). ::: ![](1478-7547-3-6-1) ::: Providers cost was about 70% (USD 102) of the total health facility DOT cost compared to 80 %(USD 75) of community DOT total cost (Table [2](#T2){ref-type="table"}). A visit to the TB clinic was the main cost item of health facility DOT, constituting about 55% (USD 56) of the providers cost. This cost item included all recurrent and capital clinic costs excluding drugs and investigations. Overall management and supervision was the main cost item under community based DOT constituting to about 53% (USD 40) of providers cost (Table [2](#T2){ref-type="table"}). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Average cost per patient for different items of alternative DOT options ::: Health facility DOT Community DOT ---------------------------------------------- --------------------- --------------------- ----------------------------------- ---------------- --------------------- ----------------------------- ---------------- ---------- ---------------- ------------ Provider\'s cost Smear positive Smear negative&EPTB *Total health facility USD\ Smear positive Smear negative&EPTB *Total community USD\ quantity Unit price USD quantity Unit price USD quantity Unit price USD quantity Unit price USD Visit TB clinic 58 1.12 24 1.90 56.0 10 1.12 10 1.90 15.0 Drugs 1 20.90 1 12.10 17.0 1 20.90 1 12.10 17.0 NTLP management & supervision district level 1 1.50 1 1.50 1.5 1 1.50 1 1.50 1.50 NTLP management & supervision regional level 1 0.93 1 0.93 0.9 1 0.93 1 0.93 0.90 Community project management 1 26.97 1 26.97 27.0 1 26.97 1 26.97 27.0 Community project supervision 1 13.16 1 13.16 13.0 Community project training 1 0.90 1 0.90 0.9** Total providers cost *102* *75* Patient\'s cost Direct 58 0.51 24 0.51 *22.0* 10 0.51 10 0.51 *5.0* Indirect 58 0.50 24 0.51 *21.0* 1 7.54 1 5.90 *7.0* Total patient\'s cost *43* *12* Treatment supporter\'s cost Direct 8 0.24 8 0.24 *1.92* Indirect 1 6.18 1 4.54 *5.43* Total treatment supporter\'s cost *7* *Total cost* *145* *94* \*Weighted average of smear positives(56%) and smear negatives and extra- pulmonary (44%) ::: Patients costs was about 30% (USD 43) of all health facility DOT cost compared 13 % (USD 12) of community DOT cost. The combined patients and treatment supervisors cost (community cost) was 20 %(USD 19) of all cost under the community DOT option (Figure [1](#F1){ref-type="fig"} and Table [2](#T2){ref-type="table"}). Indirect patients cost under health facility DOT was as important as direct cost, accounting for about 49% of patients\' cost. Indirect cost was attributed mainly to the time lost from work. Patients lost an average of 2 hours daily to follow up TB treatment in the health facility. Separate analysis of the cost according to TB type, showed a marked reduction in total cost among smear positive patients treated under community DOT compared to smear negative and extra pulmonary treated under the same option. Cost fell by 45% and 19% respectively (Figures [2](#F2){ref-type="fig"} and [3](#F3){ref-type="fig"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Average cost per patient (smear positive). ::: ![](1478-7547-3-6-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Average cost per patient (smear negative and extra pulmonary). ::: ![](1478-7547-3-6-3) ::: Effectiveness ------------- A total of 587 patients were recruited to study the effectiveness of TB treatment. Among them 260 were randomised to community based DOT and 327 to health facility based DOT. Both DOT options gave similar treatment outcomes. Treatment success rates among patients under community and health facility based DOT were 85 %(221 patients) and 83 %(271 patients) respectively (OR 1.17, 95% CI 0.75--1.83) \[[@B6]\]. The upper and lower confidence intervals for treatment success in community DOT were 89% and 81% while that of the patients in health facility DOT were 87% and 79%. For smear positive patients the cure rates were 78% for community based patients and 79% for health facility based patients. Given, HIV prevalence among TB patients in Tanzania of 44% \[[@B19]\], and using the formula to calculate cure rate without treatment, it is estimated that, about 11.2% of patients will be cured without TB treatment. When taking into consideration the \'no treatment\' scenario, the treatment successes are 74% and 72% for community and health facility based DOT patients respectively. Cost-effectiveness ------------------ Figure [4](#F4){ref-type="fig"}, shows cost-effectiveness of patients treated under community DOT and health facility based DOT. Community based DOT was more cost effective with USD 128 per patient successfully treated, compared to USD 203 for patient successfully treated with health facility DOT. The implementation of community based DOT has improved cost-effectiveness by 37%. For smear positive patients, Community based DOT was cost effective with USD 145 per patient cured while health facility was USD 258 per patient cured. Community based DOT improved cost effectiveness by 44% among smear positive patients. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Average cost per patient successfully treated under alternative DOT options. ::: ![](1478-7547-3-6-4) ::: Sensitivity analysis -------------------- The sensitivity analysis revealed that changes in key input variables did not change the cost-effectiveness ratio of community based DOT in favour of health facility based DOT. Community based DOT remained less costly and more cost-effective (Table [3](#T3){ref-type="table"}) Discussion ========== The key finding of this study was that community based DOT was less costly and more cost effective compared to health facility based DOT. This was true from all perspectives: provider\'s, patients\' and treatment supervisors\'. Community based DOT maintained programme effectiveness and improved cost-effectiveness by 37%. Implementation of community based DOT reduced cost of treating a patient by one third. This implies that with same level of resources, more TB patients can be successfully treated. The main reason for the substantial reduction of cost under community DOT was fewer number of visits to the TB clinic. A smear positive patient under health facility DOT made a total of 58 visits to a health facility while a smear negative and extra-pulmonary patient had a total of 24 visits. On the other hand, a patient under community DOT made a total of 10 visits. The study has a number of limitations, emanating from methodological and operational issues. First, since the study aimed at determining the costs and cost effectiveness of DOT options regardless of the smear status, we combined treatment outcomes and cost data of smear positive, smear negative and extra pulmonary TB patients. In routine practice, the treatment outcomes are reported separately. This may have disproportionately underestimated the cost of treating a smear positive patient under health facility compared to community DOT because, smear negative and extra pulmonary patients under health facility DOT had fewer visits to the TB clinic than smear positives. On the other hand all patients under community DOT had the same number of visits. However, this could not have changed the conclusion of the study since community based DOT remained the less costly option even after these assumptions were tested in sensitivity analysis. Furthermore, patients\' treatment outcome and cost data were analysed separately according to TB type and presented as weighted average in the final analysis. Second, the analysis focussed only on the direct benefit of TB treatment to the individual being treated. We calculated neither the secondary benefit of TB treatment accrued by reduction of transmission in the community nor indirect benefit of community based treatment to the community. Treatment of TB has been shown to have secondary benefit by averting three deaths for every case of TB cured \[[@B20]\]. By considering the benefit of treatment only to the index case, our estimates of the benefit of TB are conservative. However, quantifying the secondary benefits of TB treatment involves complex assumptions, and the inclusion of smear negative and extra pulmonary TB patients would have made such calculations more difficult. Moreover, since TB treatment is already considered an important health intervention \[[@B21]\], and implemented throughout Tanzania, our main interest was to determine cost-effectiveness of alternative TB treatment delivery options. Finally, health facility cost information was obtained from only one out of five health facilities studied. However, patients and treatment supervisors data were collected from all five-health facilities. The chosen health facility was viewed as broadly representative of the other health facilities in the district providing TB services. Furthermore the health facility had good record keeping of budget and financial details. Indirect cost in the study has been estimated from time lost due to a visit to TB clinic. There is no agreement among experts on proper way of measuring and valuing indirect cost due to lost time \[[@B12]-[@B14],[@B22]\]. We used average reported income among TB patients as baseline value of indirect cost due to lost time. The disadvantage of this method is that it may overestimate or underestimate costs \[[@B13]\]. However, sensitivity analysis was conducted when indirect cost due to lost time was valued to zero and when indirect cost was valued among people with income only. This did not change cost-effectiveness ratio in favour of alternative option. The main strength of the study is that costing was conducted prospectively alongside a randomised-controlled trial. Quantification and analysis of cost and effectiveness information was carried out in real time and the two groups in the alternative options were comparable. Although TB services in Tanzania are provided free of charge to the patients, the study showed that patients incurred considerable cost to follow their treatment at the health facility. Patients cost were 30% of the total cost, amounting to \$ 43. This is equivalent to their monthly average reported income. These costs did not include expenses used for seeking diagnosis, which have been shown by other studies to be substantial \[[@B23],[@B24]\]. Worth noting also is the fact that indirect costs contributed to about 49% of the total patient\'s cost. The findings that community DOT is less costly and more cost effective both from providers and community perspective is consistent with studies conducted elsewhere \[[@B25]\]. Community DOT reduced cost in our study by 35%. Studies in Malawi and Kenya showed that community DOT reduced cost by 50% and 65% respectively \[[@B26],[@B27]\]. Cost reduction was much higher in these settings than in our study. The main reason given is reduction in hospital stay. This is not surprising since we did not include hospitalised TB patients in our study. In Tanzania TB patients are treated on ambulatory basis and admitted only when they are seriously sick. In order to implement community based DOT, additional cost for initial investments has to be incurred. These include cost for activities such as community supervision and management. This constituted about 53% of providers cost under the community based DOT option. The increased investment, which is minor in absolute term, is nevertheless offset in the long run by savings accrued by implementation of community DOT. Community based DOT has the potential of increasing the number of TB patients treated without significantly increase in resources. With annual reported TB cases of about 4000, it can plausibly assumed that 2000 more TB patients could be successfully treated in Temeke district, using the current level of resources. This might have wider policy implications in Tanzania and beyond. If community DOT is less costly and is not in any way inferior to conventional health facility treatment, the question is how relevant are these results in other areas beyond the study area. Temeke district represents the growing challenge of controlling TB in urban settings where the numbers of TB patients are increasing and the health system is over-stretched. Our findings should be broadly generalisable in similar settings, since the cost of main items such as staff and drugs are fairly standard. The health facility costs are generally comparable across the country as shown in previous studies \[[@B4],[@B15]\]. The main determinants of patients\' cost are not different in many urban centres. The extent to which these findings can be generalised to rural areas is debatable. Evidence suggests that community based DOT is effective even in these settings \[[@B7]\]. Whether it is cost-effective as well is uncertain. The main factors for patients cost; low-income levels and greater distance to health facilities are different from urban settings. Therefore, potential difficulties with widespread implementation of this model must not be underestimated. Conclusion ========== Community DOT presents an economically attractive option to complement health facility based DOT. It improves the affordability and cost-effectiveness of TB treatment. Community DOT might also help to reduce congestion in TB clinics, which are increasingly overburdened by the rising number of patients. This is particularly important in urban centres where TB clinics are working beyond capacity. TB programmes need to consider community DOT as an economically sound and viable option and implement it in the framework of routine programme activities. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= All authors participated in designing of the study. EW supervised data collection. EW, OM and BR participated in data analysis, writing and editing of the paper. All authors read and approved the final manuscript. Acknowledgements ================ The authors thank the Ministry of Health in Tanzania, Temeke municipal authorities and Jamii TB research group for their contributions. This work was supported by the Norwegian Association of Heart and Lung Patients (LHL).	PubMed Central	2024-06-05T03:55:59.677513	2005-7-14	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180840/", "journal": "Cost Eff Resour Alloc. 2005 Jul 14; 3:6", "authors": [ { "first": "Eliud", "last": "Wandwalo" }, { "first": "Bjarne", "last": "Robberstad" }, { "first": "Odd", "last": "Morkve" } ] }
PMC1180841	Introduction ============ I had two great teachers in epidemiology: Len Syme and Geoffrey Rose. One had his thinking shaped by the insights of Durkheim, a great sociologist; the other by Pickering, a great hypertension specialist. One helped lay the foundations for social epidemiology; the other, if heeded, could change the way we think about public health. Both came to the conclusion that society mattered for health and that one could not understand the social rate of disease simply by studying individuals. The story of laying the foundations for social epidemiology is engagingly set out in the accompanying paper by Len Syme.\[[@B1]\] Geoffrey Rose\'s credo is beautifully articulated in his last book.\[[@B2]\] Scientist as teacher and mentor =============================== Syme\'s memoir of his vital role in the origins of social epidemiology is characteristically modest. He alludes to his students but his paper does not convey the half of it. I have never come across anyone in the academic world who had quite the powerful influence on students that Syme did. With cohort after cohort of students he developed special relationships, he with them and they with him. He was stimulating, challenging, encouraging, sensitive, hands off when appropriate, guiding when needed, critical when constructive, praising when spirits required i.e. more or less always. I suspect he had few failures, in part because there was \"assortative mating\" -- a matching of teacher and student that had the right chemistry, but also because he was simply so good at bringing the best out of people and then helping them with their subsequent careers. Not all students came under his magic spell -- some had different interests and sought different mentors -- but my guess is that he helped everyone who came into his orbit. For some of us the relationship, begun at Berkeley, became one of life long friendship and collegiality. I knew how much he influenced me while I was at Berkeley as a graduate student and young teacher, and then, having left, I discovered all over again how many of my \"original\" thoughts had their origins in Berkeley. Syme encouraged interdisciplinary work. He taught the importance of asking the right question. We were encouraged to seek insight wherever it may be found -- and on the Berkeley campus there was no shortage of insight. And he showed what it meant to consider social causes. I touch on each of these below. Analysis ======== \'Doctors have no special insight on causes of ill-health\' ----------------------------------------------------------- That was really shocking. How dare he? But Syme did say that to his students. Like most physicians I assumed that only doctors could understand the causes of illness. It was we, after all, who laid claim to some understanding of biology and pathology. Giving primacy to biology betrayed a classic misunderstanding of the notion of cause. It may be useful to think of accumulation of lipid in the endothelium of coronary arteries as \"the cause\" of coronary heart disease -- a good pathologist\'s perspective on cause -- but what causes the accumulation of lipid? I now use Geoffrey Rose\'s phrase: the causes of the causes\[[@B3]\] (it has less tendency to frighten the horses than other terms that I still use) but my first introduction to such thinking was from Syme. Syme\'s argument was that the reading of a medical textbook conferred no unique insight into the social causes of disease. Just as medical training, or its equivalent, may be necessary to understand the mechanisms of disordered pathology so is understanding of society necessary to understand the social causes of illness. He was not a social scientist who considered biology unimportant and social science paramount. Both were needed. That lesson has stayed with me. \'Three questions\' ------------------- Actually four. There was a meta-level. Syme taught that the first, perhaps most important, question was: \"what\'s the question?\" This then resolved into three separate questions one should ask about the occurrence of disease: Why does one group have a higher rate of disease than another? Why does one individual in a group get sick and another not? When an individual succumbs to illness what determines which illness s/he gets? Syme pointed out that most \'current\' (1970s but still the case) epidemiological research pursued the second of these questions: the search for risk factors. In the 1970s it was blood pressure and cholesterol. In the 2000s it is C Reactive Protein. The thinking is the same: will a risk factor discriminate those who will get sick from those who won\'t. Syme refers in his paper to John Cassel. Both Syme and Cassel were active in pursuing the argument that individuals\' risks of disease seemed to vary in parallel, i.e., people at risk of one disease seemed to be at risk of others that, according to conventional wisdom, had different pathology. Hence two different questions (the second and third in the above list): why did an individual get sick and what determined the particular disease from which he suffered. Syme and Lisa Berkman wrote an important review paper: Social Class, Susceptibility and Sickness\[[@B4]\]. In it they pointed out that low social class was associated with high risk of a range of apparently unrelated diseases. That is to say, they built on a notion of susceptibility that may relate to more than one illness. It is a notion that makes many uncomfortable in the world of medical research. The hypothesis of general susceptibility leads to a clash of ideas. One view is that low status people, for example, have a high risk of lung cancer because of smoking; of stomach cancer because of diet; of stroke because of diet; of heart disease because of diet, smoking, and sloth; of renal disease because of infection; of suicide because of \... I\'ll think of something; and so on. Syme threw out the challenge that there may be a common reason underlying these various ways of becoming ill -- one should search that out. Separately, one should enquire why one person got heart disease and another, tragically, took his own life. Syme spells out how an understanding of Durkheim leads to the other crucial distinction; between enquiring after the causes of variation among individuals and variation among groups. The reason why one young man, in a deprived inner city neighbourhood, gets shot and not another is the individual difference question. The reason why young deprived inner-city men have a higher rate of violent deaths than men in the affluent suburbs is a different sort of question. The answers to these two questions, the first and second of the triad, may not be the same. Theory? ------- Syme is disarming about his \'misguided\' counsel to ignore theory. I discovered, on leaving Berkeley for the UK, that the two worst criticisms levelled by social scientists against epidemiologists was that they were positivists and atheoretical. It can\'t get harsher than that! An extreme form of reaction to lack of theory and positivism is social constructionism. There is no objective reality -- all is a social construction. My first encounter with extreme relativism -- no objective truth just better or worse theories -- came, at Berkeley, from Paul Feyerabend\'s philosophy of science seminar.\[[@B5]\] Turning up there was the result of Syme\'s general encouragement to \'go forth, young man\' and explore. Nothing was ruled out. Quite magic! Feyerabend was working on his book, Against Method, and his catch phrase was: anything goes. It was simply hubris, said Feyerabend, to think that modern science was in some way better than pre-Socratic philosophy. But, I protested, with modern science, we are eliminating smallpox, and putting men in space. \"Simple technology! Judge the science by how it performs according to its own lights,\" was Feyerabend\'s response. In other words, just because a theory led to elimination of a disease responsible for a scourge on mankind was, in itself, no proof that it was a better theory than one that was internally consistent. I was only partly convinced.\[[@B6]\] This is, perhaps, a caricature. But not much. Faced with that sort of theory, one can see why Syme might have sought facts. But if a theory is a way of making sense of observations about the world then, I would argue, Syme did have us groping towards theory. It is true that when I first started to study the social gradient in health in Britain\[[@B7]\] I thought it did not much matter how one classified people. Occupational status, education, income all predicted differences in health and life expectancy. Social position was clearly important. What did it matter if the investigator were a Marxist or a Weberian? Surely theory was beside the point. That said, I did acknowledge that we have theory all the time. It permeates our very observations. Recognising that is not to subside into relativism. It is, however, a call to stand back and ask questions about what we are doing and what it means. Let me illustrate. Syme suggested that \'control of destiny\' was a fundamental influence on disease.\[[@B8]\] Separately Karasek and Theorell had been developing their job strain model.\[[@B9],[@B10]\] They proposed that it was not simply job demands that were important for health, but the imbalance between demands and control. Workers faced with high demands but who had little control were at increased risk of cardiovascular and other diseases. Was that theory? It certainly was a way of making sense of some disconnected thoughts and observations. People, going back to Sir William Osler, thought that high status individuals had more heart attacks than lower status because, among other influences, of the high stress that their exalted positions entailed. Yet, in the 1970\'s, we showed that in British civil servants, the lower their status the higher the mortality from coronary heart disease.\[[@B7]\] This was not a phenomenon unique to British civil servants but was true of the country as a whole.\[[@B11]\] Did that refute the proposition that stress was an important factor in cardiovascular disease? Karasek and Theorell, and Syme, provided a framework, theory even, for why it did not. Control was likely to be lower in lower status groups. Indeed, that is what we showed.\[[@B12]\] If anyone is \'doing theory\', surely it is philosophers. When I first came to read Amartya Sen\'s work -- Sen is as much philosopher as economist -- it was a revelation.\[[@B13]\] He was an empirical economist who examined the theoretical basis of what he did. One of his fundamental contributions was the idea of capabilities. What was important to social inequalities was not so much what one had but what one could do with what one had. This \'theory\' provided a way of thinking about the importance of control -- it was part of capability, what you could do with what you had. At the same time as Sen was dazzling with his insights, another philosopher and a political scientist, Doyal and Gough, elaborated a theory of human need, in which they argued that autonomy was a fundamental human need that cut across cultures and was therefore a way of rising above relativism.\[[@B14]\] This human needs approach gave me a way of thinking about my own struggle to understand the reasons for the social gradient in health.\[[@B15]\] This by any other name is theory, and Syme contributed to it actively. He is, once again, too modest about his own role here. Conclusion ========== Social causes ------------- It is, still, an unusual idea that diseases have social causation and that the remedies for social causation might be social in nature. They need not, of course. Syme spent a good part of his life trying to encourage individual people to change their diets and give up smoking. He came away from that experience claiming that it was too difficult to change individuals. He was going to try something easier: change society. Syme\'s enduring contribution has been as an advocate for the idea that social processes may be as much causes of illness as are biological processes. If the remedies of the social causes of health should be social, what should we do? I am now up to my ears in a new Commission on Social Determinants of Health.\[[@B3]\] We are trying to take a social approach to reducing inequalities in health between and within countries. The emphasis is on action. In my own mind, there is a continuous line between Syme\'s teachings and my own current activities. I blame Len Syme.	PubMed Central	2024-06-05T03:55:59.681198	2005-6-2	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180841/", "journal": "Epidemiol Perspect Innov. 2005 Jun 2; 2:4", "authors": [ { "first": "Michael", "last": "Marmot" } ] }
PMC1180842	Background ========== A growing literature on the social determinants of health, suggests explanations for many population and individual level health outcomes are not attributable to biology. Income, income inequality, social connectedness, and social capital all show some association with health and illness \[[@B1]-[@B6]\]. This paper explores the meaning of gender as another of these social determinants of health, and proposes an epidemiological framework for including gender as an independent variable in future research. Sex versus Gender ================= The usefulness of distinguishing between sex and gender, a common practice in the social sciences, has begun to percolate into the language of prevention, etiology and causation within health care. Sociologists describe sex as the relatively unchanging biology of being male or female, while gender refers to the roles and expectations attributed to men and women in a given society, roles which change over time, place, and life stage. Genetic makeup and hormone profile are both examples of sex, that is, of biologic characteristics, which tend to be constant across societies. Gender is a social, rather than a biological construct, and varies with the roles, norms and values of a given society or era. Being able to bear a child is, fundamentally, a function of biology, while expectations about the imperative to bear children, the nature of parenting, or the status associated with being a mother are more closely linked to gender roles and expectations. Gender has an impact on health in a variety of ways. Powerlessness and lack of control underlie much of the exposure to HIV/AIDS amongst women in Africa. Disproportionate barriers (that is, relative to men) in access to resources such as food, education, and medical care, disadvantage women throughout the developing world. Risk taking behavior is the norm amongst males throughout the world. Socially defined traits often stereotype men and women as having fixed and opposite characteristics such as active (male)/ passive (female), rational (male)/ emotional (female) \[[@B7],[@B8]\]. The language of medicine and its underlying philosophy have, and may still equate male with normal, leaving female to be considered as \"other\" or, perhaps, abnormal \[[@B9]\]. Both women\'s and men\'s occupational and behavioural roles, constrained by social norms, can result in hazardous, though different exposures to dangers and illness \[[@B10]\]. Any of these aspects of gender may intercede in the pathway from an individual to his or her health. Differentiating Gender and Women\'s Health ========================================== While the content of women\'s health is quite clearly restricted to women, both men and women are subject to the health effects of gender. Perhaps because the deleterious impact of gender appears to burden women disproportionately, the literature on gender as a determinant of health that has begun to appear is generally about women. At a population level, in most countries of the world, women have more limited access to, and less control over, resources, and over their bodies and lives, than do men. Women appear to suffer more negative health consequences of inequalities between the sexes. The very recent increase in research on sex and gender differences, with it\'s focus on women, has begun to correct a history of generalization from the male subject to the female population \[[@B11]-[@B13]\]. Government funding agencies have given priority to research in women\'s health in recent decades to correct these past imbalances in research. Although efforts to report sex differences in studies and clinical presentations are not always optimal, they are well underway. There remains, never the less, much work to be done in delineating the associations between gender and health, particularly the health of men. The philosophy of the diverse and extensive literature on women\'s health published over the preceding twenty years diverges into four groups over the question of gender. The first group does not question biology as the sole determinant of health but calls for an end to the discrimination that made the biology of women invisible. For these authors optimizing women\'s health requires attention to the diseases that are unique to, more common amongst, or present differently in women but accepts, as a biologic imperative, the male, female differential in incidence of these diseases. The limitations of such an approach are two-fold. Firstly, women continue to be defined in terms of men. Secondly, biology defines being, while social determinants vanish. A second group introduces the term gender but uses it interchangeably with sex \[[@B14]-[@B17]\]. The third group links the health of women as patients to the well being of women as health care providers in a traditionally patriarchal health care delivery system. Much influenced by \"the glass ceiling\" literature describing the challenges women face in advancing to leadership positions, this group suggests that only after models of care change will health outcomes for women improve. Of most relevance to this paper is the fourth group, which defines gender as a social determinant of the health of both men and women, but grapples with how sex and gender interact, and with the specifics of how gender shapes individual health \[[@B14],[@B18]\]. The remainder of this paper is concerned with these questions. Interactions Between Sex and Gender =================================== Neither sex nor gender can adequately be examined as variables associated with health but independent of each other. Gender interacts with biology, in every society, although the results of that interaction vary from setting to setting. Sex differences in life expectancy offer one of many examples of the interaction between sex and gender in health outcomes. In most of the world women outlive men, the exceptions being the least developed countries, where extremely high birth rates and significant human rights inequalities disproportionately disadvantage women. Despite having had most of the social determinants of health in their favor, men have higher mortality rates for all 15 leading causes of death, and a life expectancy about seven years shorter than women\'s in the United Kingdom, as in most Western countries \[[@B19]\]. Mortality data from 1995 for Canada\'s largest province, Ontario, show the same contradiction \[[@B20]\]. Government documents, however, report the numbers but do not explore the striking sex differences in longevity. Sub-analysis of the data explain the interconnectedness of sex and gender and shed light on the apparent contradiction inherent in both the UK and Canadian data. Between birth and age 45, there are 1,812 male deaths, of which 1,372 (76%) are due to motor vehicle accidents, suicide, and AIDS, leaving 440 deaths unrelated to behavior. Although the male excess of deaths from car accidents may, in part be attributable to greater distances driven and not behavior while driving, the male \"relationship\" with the automobile is almost certainly another aspect of gender roles. Only 308 (33%) of the 936 female deaths are explained by such behavior. When non-risk taking causes of death are isolated from the data, women under age 45 have a mortality which is 1.43 times that of men\'s. Over age 45 the leading causes of death for both men and women are chronic diseases. Men die of heart disease in equal numbers but at a younger age than do women. With increasing age the number of deaths for women creeps upward to equal that of men. Male gender roles as manifest by risky behavior around drinking, driving, and sex, account for virtually all excess male mortality below age 45, and approximately 50% of the excess below age 60. This data suggest that gender is a central, although not the sole cause of the shorter life expectancy of men in western society. How much of this gendered behavior actually arises from a male, historic, biological imperative to dominate other males and thereby win a mate to procreate is unknown. Therefore, a sex difference in longevity appears to result, in part, from a gender difference in behavior, which may, in turn, arise from the biology of sex. Defining which aspects of female advantage in life expectancy arise purely from the fixed biological attributes of sex and which are due to gender and amenable to change is challenging. Redefining Gender ================= Given the interconnectedness of the biological and the social, it might prove pragmatic to consider that gender encompasses both sex differences and the social constructs that give rise to gender differences. For example, in most developed countries myocardial infarcts are a leading cause of death for both men and women. Male risk of coronary artery disease at a younger age exceeds female risk for what appear to be biologic reasons. Because, however, the \"prototype\" of the heart disease patient has been male, the most common presentation amongst men, that of crushing chest pain, has become the most important diagnostic clue to the presence of angina. Recent evidence demonstrates that fatigue, rather than chest pain, is, in fact, the most common symptom of angina in women \[[@B21]\]. The gender biases inherent in the underlying assumption that coronary artery disease is primarily a male affliction, and that those women who suffer from it will present as men may explain why women have a significantly higher post MI mortality than do their male counterparts \[[@B22]\]. There is no practical advantage to disentangling where sex ends and gender takes over as a cause of the sequelae of coronary artery disease. Never the less, for a physician to properly diagnose angina requires an understanding of sex differences in the symptoms associated with coronary ischemia, and a rejection of the gender stereotype that men, and not women, are the bearers of this illness. By defining gender as the composite of both social and biological health effects associated with being either male or female, researchers may more easily move on to studying those effects, without getting stuck at enumerating sex differences. Measuring gender effects: epidemiology challenges ================================================= Defining gender is a necessary prerequisite but not a sufficient solution to the problem of developing a methodology of its measurement. The impact of gender as a social determinant of health is likely a composite of the effects of relative power, autonomy, poverty, and marginalization, within, and across, societies and cultures. As gender is, by definition, a social or population level determinant of health, its consequences at an individual level are less tangible. There are no randomized controlled clinical trials (RCTs), the \"gold standard\" of individual level research, that measure the health effects of gender. Gender defies \"packaging\" as an etiologic agent of disease nor could it appear in a list of differential diagnoses for a set of clinical findings. How should the associations between gender and illness be studied? The process could involve identifying a cohort from numerous countries, prospectively tracking health outcomes such as longevity, mental health, or incidence of a variety of morbidities, and identifying which non-biological inputs (the usual ones being age, sex, education, income, lifestyle risk factors like smoking, alcohol and drug use) are associated with adverse outcomes. Using regression analysis, the effect of socioeconomic factors on health could be identified and isolated from other inputs, rather than controlled for and eliminated (as happens in an RCT where the randomization equalizes the effect of these social phenomena between the study and control group, but precludes analysis of the association of these phenomena with the outcomes in question since they are effectively deleted prior to analysis). The challenge of how to insert gender as an independent variable, that is, into the left side of such a regression analysis, would, however, still remain. The interconnection between gender and socio-economic status necessitates addressing both in analyzing associations with health and illness. In thinking about this, the epidemiologic constructs of \'within group\' and \'between group\' variation may be useful. The concept of gender could include differences in socio-economic and cultural determinants of health between men and women. If the groups being studied or compared are men and women, the between group variations would then be summed up by gender. There remain, however, within group variations because not all women are the same. Data examining the percentage of births attended by trained personnel and aggregated by the level of the mother\'s education consistently favour the more educated and demonstrate variation in access to care \[[@B23]\]. Within the grouping \'women\' social determinants such as education or income often account for differences in power or access to care, and, ultimately, to health. A gender coefficient ==================== Specific coefficients, or composites of several variables have been developed to measure some of the social determinants of health. Levels of income or education can be categorized. Income inequality, another of the well-studied social determinants of health can be quantified in several ways. Human rights, although more vague, can be measured via proxy variables such as participation of women in government and public life. As yet, no one has proposed a proxy measure for gender. Indicators of human rights may approximate gender when the health of women is the outcome being examined. Similarly income, income distribution, and access to education or health care are likely colinear with gender in the measure of women\'s health. Even more important may be the interaction between these variables in answering questions such as how the health of women in a relatively wealthy, but repressive country compares to women\'s well being in a less wealthy, but more egalitarian country. Multilevel analyses may capture individual variations in wealth or freedom within population level research. More problematic is identifying proxies for gender when examining men\'s health. Perhaps variables that measure acceptance of violence in a society (e.g. prevalence of gun ownership) touch on gender as a determinant of the health of men. Unfortunately there is, at present, no gender equivalent of the Gini co-efficient, the summative measure of income inequality in a community. Summary ======= Within the past two decades, in response to the historic under- or mis-representation of women in the research that shapes medical practice, major funders such as the NIH in the US, developed more inclusive guidelines for research methodology and funding. Concurrent with this, and likely as a result of it, a literature on women\'s health began to appear. Subsequently, the construct of gender was borrowed from the social sciences to broaden etiologic concepts in the area of women\'s health beyond biologic differences between the sexes. Although authors often use the terms women\'s health and gender interchangeably, gender has a wider scope, which facilitates discussion of the effects of social norms and expectations on the health of both women and men. Perhaps, because to date, no gender index or composite of variables that, together, would be a measure of gender, has been proposed, there are few, if any studies of the associations between gender and health. In the context of health outcomes, defining gender to include both biologic and social aspects of being male or female, and considering which measurable variables could form a gender co-efficient should enable research to move forward. Using some of the proposals from this paper, indices of gender may, in the future, become part of analyses of how social factors impact on health.	PubMed Central	2024-06-05T03:55:59.682383	2005-7-13	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180842/", "journal": "Int J Equity Health. 2005 Jul 13; 4:11", "authors": [ { "first": "Susan P", "last": "Phillips" } ] }
PMC1180843	Introduction ============ This paper describes a set of fundamental concepts from causality theory that can be used to critically analyze summary measures of population health. It then uses these concepts to argue that health measures based on hypothetical outcome removal \[[@B1],[@B2]\] are ambiguous and potentially misleading. Thorough analyses require a multivariate framework to capture what is known about sources of morbidity and mortality. Because of the unavoidable shortcomings of single-valued summaries, multidimensional indices should be considered for summary measures of population health. The first task is to define cause and effect in a manner precise enough for logical manipulation and quantification. Three types of models have achieved widespread usage: • Counterfactual or potential-outcome models. The counterfactual conceptualization of causality was glimpsed by Hume \[[@B3]\] and was operationalized by statisticians in the 1920s and 1930s in the form of potential-outcome models \[[@B4]\]. Related counterfactual models have also received much attention in philosophy \[[@B5],[@B6]\], and potential outcome models are now common in social sciences \[[@B7],[@B8]\] and epidemiology (see citations below) as well as in statistics. • Structural-equations models. These models can be traced to early work in path analysis in the 1920s and were given full expression by econometricians in the 1940s. • Graphical models(causal diagrams). These models also originated in path analysis, but were not fully developed until the early 1990s \[[@B9]-[@B11]\]. In his book on causal theory, Pearl \[[@B11]\] details the above approaches and their histories, emphasizing that they are logically equivalent (isomorphic) for all practical purposes. This means that an analysis using one approach can be translated into either of the other two approaches, while maintaining logical consistency. Greenland and Poole \[[@B12]\] and Greenland and Brumback \[[@B13]\] describe the connections between potential outcome models and the sufficient-component cause models familiar to epidemiologists, and Greenland \[[@B14]\] discusses potential outcome models in relation to more traditional approaches to causal inference. Because of the emphasis on counterfactual models in the literature on measures of cause and effect, the following development is based on them. Further details of counterfactual theory for health science research are described elsewhere \[[@B12],[@B14]-[@B26]\]. For details on its relation to missing-data models, see Rubin \[[@B27]\]. The present paper concerns only issues of defining effects and their implications for policy formulation. Equally important is quantitative accounting for study biases in estimating effects; see Greenland \[[@B28],[@B29]\], Lash and Fink \[[@B30]\], and Phillips \[[@B31]\] for recent developments in that topic. Basic concepts of counterfactual causality ========================================== Actions, outcomes and counterfactuals ------------------------------------- To minimize ambiguity, a counterfactual model requires reasonably precise definitions of the following model ingredients: • At least one targetsubject of interest in whom or which causation is to be studied (e.g. a specific person or population). • A list of two or more possible alternative actions, x~0~, x~1~, etc., that could have been applied at or over a span of time, one of which may be the actual action taken. • An outcome measure, Y, taken at a point in time or over a period of time, following completion of the action. As an example, the subject could be Kosovo, the action x~1~could be revocation of its autonomy by Yugoslavia in 1988, an alternative x~0~could be that revocation never occurred, and the outcome measure could be the mortality rate of pre-1999 residents during 1999. Because only one of the possible actions x~0~, x~1~, etc. can take place, all but one of the actions must become counterfactual, or contrary to fact. In the example, the actual (or factual) action was x~1~= \"revocation of autonomy\"; thus, the action x~0~= \"no revocation\" is counterfactual. (This example also illustrates the truism that to do nothing is an action, corresponding to zero on an action scale.) If x~a~is the actual action taken (from the list x~0~, x~1~, etc.), we may observe the outcome Y(x~a~) that follows that action. A counterfactual outcome modelposits that, for any counterfactual action, x~c~(from the same list), there is also a well-defined outcome, Y(x~c~), that would have followed thataction. The entire list of such outcomes Y(x~0~), Y(x~1~), Y(x~2~), \... is called the set of potential outcomesof the subject; it includes both the actual outcome Y(x~a~) as well as all the outcomes of counterfactual actions. In the example, the actual action, x~a~, was x~1~and was eventually followed by a mortality rate, Y(x~1~), in 1999. Y(x~1~) is difficult to assess but nonetheless exists as a matter of fact; it is part of what occurred subsequently to the action x~1~. A counterfactual model for 1999 mortality posits that, had the counterfactual action x~0~been taken instead (that is, if no revocation had occurred), the mortality would have equaled some number, Y(x~0~). Given that x~0~is counterfactual, it is logically impossible for us to observe Y(x~0~). Nonetheless, the counterfactual model treats this number as a precise, though unknown, quantity. The idea of treating the outcome, Y(x~c~), under a counterfactual action, x~c~, as a precise quantity has been a source of much controversy and misunderstanding (e.g. \[[@B23],[@B32],[@B33]\]). Some major misunderstandings are addressed below: 1\) The counterfactual approach does notrequire that the outcome, Y(x~c~), be precisely defined for every possible counterfactual action, x~c~. In the above example, if we are interested only in contrasting revocation (x~1~) with no action (x~0~), our model need not mention any other actions. That is, we can limit the action list to just those actions of interest. 2\) The counterfactual approach is notinherently deterministic in either the classical or quantum-mechanical sense \[[@B15],[@B34]\]. The potential outcomes, Y(x~0~), Y(x~1~), etc., may represent different values for a statistical parameter in a classical probability model. For example, they may be expected rates in Poisson models. Alternatively, they may represent different mixtures of superposed states (different wave functions) in quantum models. Indeed, some theoreticians regard counterfactuals as essential for formulating coherent explanations of quantum phenomena \[[@B35]\]. 3\) The counterfactual approach extends the partial quantification of outcomes, Y(x~c~), under counterfactual actions embedded in ordinary discourse. In the example, some (though not all) observers of the events in Kosovo 1988--1999 speculated that the actual 1999 mortality, Y(x~1~), was probably greater than Y(x~0~), the mortality that would have occurred had autonomy never been revoked. This speculation arises from the following tentative explanation of actual events in Kosovo: revocation of autonomy, (x~1~), caused Albanian resistance to increased Serb authority, which in turn caused Serbian leaders to extend their \"ethnic cleansing\" policy to Kosovo. Had there been no revocation, (x~0~), this tragic causal sequence of events would not have occurred. Cause and effect ---------------- The speculative explanation in the third bulleted item above is an example of an informal causal hypothesis. Consideration of such hypotheses has led to the following definition: An effectof taking an action, x~j~, rather than another action, x~k~, on an outcome measure, Y, is a numerical contrast of that measure (e.g. the difference or ratio) under the two different actions. The contrast is called an effect measure. In the example, the contrast Y(x~1~) - Y(x~0~) is an effect of revocation x~1~versus no revocation x~0~; this effect measure is known as the mortality difference due to x~1~versus x~0~. Similarly, Y(x~1~) / Y(x~0~) is the effect measure known as the mortality ratio due to x~1~versus x~0~. Many common ideas and a few surprises follow from the above definitions, among them: 4\) An effect is a relationbetween the outcomes that would follow twodifferent actions, x~j~and x~k~, in just onesubject (a population or single person). It is thus meaningless to talk of (say) \"the effect of smoking a pack a day\"; one must at least imply a reference (baseline) action for \"the effect\" to have meaning. While smoking a pack a day can cause lung cancer relative to no smoking, it can also prevent lung cancer relative to smoking two packs a day. 5\) If Y(x~j~) = Y(x~k~), we say that having x~j~happen rather than x~k~had no effect on Yfor the subject; otherwise, we say that having x~j~happen rather than x~k~caused the outcome to be Y(x~j~) and prevented the outcome from being Y(x~k~). For example, we may say that smoking prevents survival past age 70 years just as surely as it causes death by age 70. Similarly, we may say that not smoking causes survival past age 70 just as surely as it prevents death by age 70. Thus, the distinction between causation and prevention is merely a matter of whether we are talking of an action, x~j~, and itsconsequence, Y(x~j~) (causation of Y(x~j~)), or an action, x~j~, and a consequence, Y(x~k~), of an alternative action x~k~≠ x~j~(prevention of Y(x~k~)). 6\) At least one of the actions, x~j~, x~k~, in an effect measure must be counterfactual. Thus, we can never observe an effect measure separate from an outcome measure. In the example, we observed the mortality Y(x~1~) = Y(x~0~) + \[Y(x~1~) - Y(x~0~)\], so the mortality difference, Y(x~1~) - Y(x~0~), is mixed with the reference (baseline) mortality rate, Y(x~0~), in our observation. The best we can do is make an informed estimate of Y(x~0~), which is the outcome that would have happened under the counterfactual action x~0~, and from that estimate deduce an estimate of the effect measure (by subtraction, in this example). Causation, confounding and association -------------------------------------- Problem 6 is considered a fundamental problem of all causal inference. It was recognized by Hume \[[@B36]\] and is now known as the identification problemof cause and effect. All causal inferences (and hence all intervention plans) depend on accuracy in estimating or predicting at least one unobserved potential outcome following one counterfactual action. We ordinarily make this prediction based on observations of other subjects (controls) who experienced actual actions different from the subject of interest. For example, we might estimate that the mortality Kosovo would have experienced in 1999 had there been no revocation, Y(x~0~), would equal the mortality it experienced in 1988, before violence began growing. In making this estimate, we run the risk of error because, even under the actionx~0~(no revocation), Kosovo mortality might have changed between 1988 and 1999. If so, we say that our estimate is confoundedby this unknown change. Denote by Y~1988~the mortality experienced by Kosovo in 1988. We can then restate the last problem as follows: we do not observe Y(x~0~), so we cannot directly compute a measure of the effect of x~1~versus x~0~. If, however, we believed the speculative explanation given in the above third bulleted item, we might also think that Y~1988~is not too far from Y(x~0~), and so substitute Y~1988~for Y(x~0~) in our measures. Thus, if we also observe Y(x~1~), the actual 1999 mortality, we would estimate the effect measure Y(x~1~) - Y(x~0~) with the observed mortality difference Y(x~1~) - Y~1988~. The latter observed difference is called a measure of association, because it contrasts two different subjects(Kosovo in 1999 versus Kosovo in 1988), rather than one subject under two different actionsin our list (Kosovo in 1999 under revocation versus Kosovo in 1999 with no revocation). Because of the identification problem (we cannot see Y(x~0~)), we must substitute a measure of association for the measure of effect. In this usage, the observed difference will misrepresent the effect measure by an amount equal to the difference of the two: \[Y(x~1~) - Y~1988~\] - \[Y(x~1~) - Y(x~0~)\] = Y(x~0~) - Y~1988~ This quantity measures the amount of confoundingin the association measure (the observed difference) when it is used as a substitute for the effect measure. Like the effect measure itself, the confounding measure contains the unobserved Y(x~0~) and so can only be estimated, not observed directly. Suppose, however, that we know of reasons why Y(x~0~) and Y~1988~would differ, such as changes in age structure over time. We can then attempt to adjust Y~1988~for these suspected differences, in the hopes of getting closer to Y(x~0~). Standardizationis probably simplest example of such adjustment \[[@B18],[@B26]\]. The presumption underlying use of an adjusted effect measure is that it accounts for all important differences between the unobserved (counterfactual) reference outcome, Y(x~0~), and the substitute, Y~1988~, in the above example. The presumption is debatable in most applications; for example, some would argue that \"ethnic cleansing\" would have spread to Kosovo even without autonomy revocation and Albanian resistance. This problem of uncontrolled confounding is but one of many methodological problems in estimating effects that are discussed in textbooks (e.g. \[[@B26]\]). The effects of outcome removal ============================== Consider a question that asks about the health burden attributable to y~1~versus y~0~, where y~1~and y~0~are not actions in the earlier sense, but are themselves alternative outcomessuch as AIDS death and CHD death. For example, y~1~could be \"subject dies of lung cancer\" and y~0~could be \"subject does not die of lung cancer\". As in the earlier framework, these outcomes are mutually exclusive possibilities for just one subject at any one time; hence, at least one must become counterfactual. Because they are not interventions, however, there is severe ambiguity in any definition of another outcome, T, as a function of the potential outcomes, y~1~and y~0~, because Tdepends in a critical fashion on how y~1~and y~0~are caused. To see this, suppose Tis years of life lived beyond age 50 (which is age at death minus 50). How would one have brought about y~0~(prevented the lung-cancer death) if the subject were a male lifelong heavy smoker who developed lung cancer at age 51 and died from it at age 54 (and so had T(y~1~) = 4 years of life after age 50)? If y~0~had been achieved by convincing the subject to never start smoking, T(y~0~) could be much larger than T(y~1~), because the risks of many other causes of death (e.g. CHD) would have been much lower as a consequence of never smoking. But if y~0~had been achieved via an unusually successful new chemotherapy for lung tumors, T(y~0~) might be little changed from T(y~1~). This would occur if, shortly after remission, the subject had a fatal myocardial infarction whose occurrence was traceable to smoking-induced coronary stenosis. The problem just described has long been recognized in discussions of estimating the impact of \"cause removal\" or \"removal of competing risks\" when the \"causes\" or \"risks\" at issue are outcomes rather than actions or treatments \[[@B37]\]. These outcomes are not subject to direct manipulation independent of the earlier history of the subject. Therefore, any realistic evaluation of the impact of their removal must account for other effects of the meansof removal. A similar problem arises in the evaluation of ordinary treatments whenever noncompliance can occur. In general, only advice or prescriptions are under control of the health practitioner; what a patient actually receives is affected not only by advice or prescription, but also by the many complex social and personality factors that influence compliance. This leads to manifold problems in evaluating the effects of received treatment \[[@B38]\], for then the treatment a subject receives is only an outcome, Y(x~j~), of an earlier prescriptive action, x~j~. In most cases, however, this initial action is unambiguous. Suppose we could avoid the ambiguity problem by introducing a pair of well-defined alternative actions, x~1~and x~0~such that x~1~causes y~1~and prevents y~0~relative to x~0~. That is, suppose y~1~will follow x~1~, whereas y~0~will follow x~0~, so that we have y~1~= Y(x~1~) and y~0~= Y(x~0~) with y~1~≠ y~0~. We may still face a serious confounding problem in the form of \"dependent competing risks\". Consider again the heavy smoker who develops lung cancer at age 54, with treatment, x~0~, being successful chemotherapy. It could be a mistake to calculate this subject\'s life expectancy, T(x~0~), from that of other heavy smokers of the same age and sex who had not developed lung cancer, because such smokers may differ in ways that render them not only less susceptible to smoking-induced lung cancer, but also less susceptible to other smoking-induced cancers (perhaps because they have better DNA-repair mechanisms). More generally, even if the means of removal is precisely defined, feasible and has no side effects, there is rarely a basis to believe, and often good reason to doubt, that removal of a particular outcome (such as a particular cause of death) would be followed by risks similar to risks among persons who, in the absence of intervention, do not experience the removed outcome \[[@B37],[@B39]\]. Unfortunately, standard statistical procedures for projecting outcomes under cause removal (such as Kaplan-Meier/product limit methods and traditional \"cause-deleted\" life tables) are based on this similarity assumption. In view of the problems just described, it is reasonable to conclude the following: 7\) Projections of the impact of outcome removal (e.g. removal of a particular ICD9 cause of death \[[@B1]\]), rather than an action that brings about outcome reduction, may not be useful for program planning. Except perhaps in some unusually simple cases (e.g. smallpox eradication), the effects of actions and policies do not correspond to simple cause removal. 8\) Even when we have a treatment that specifically and completely prevents an outcome, biased effect estimates are likely if one simply projects the experience of those who naturally lack the outcome onto those who avoid the outcome because of the treatment. Only ongoing follow-up of successfully treated subjects can reliably identify the impact of outcome removal. Problem 7 implies that summary measures for policy formulation should refer to effects of operationalizable actions (e.g. anti-smoking campaigns, food-distribution programs), rather than effects of removing the outcomes targeted by those actions (e.g. smoking, cancer, malnutrition). Only rarely will the two effects coincide. Focusing on the outcome removal presents a grossly overoptimistic picture of what can actually be accomplished, since the latter is determined by what feasible interventions are available. Focusing on outcome removal has the potential of diverting resources away from where it will do the most good -- outcomes with feasible and effective preventives -- toward outcomes that, while more common and costly, have less hope of successful and economical prevention. Finally, a focus on outcome removal diverts attention from assessing and comparing the full impacts of interventions. For example, even partial reduction in tobacco use will result in a broad spectrum of outcome prevention, from heart disease to lung cancer, whereas an effective treatment for lung cancer would only reduce the burden from that disease while raising the burden from tobacco-related risks. The preceding consideration raises another point: Because any action will have multiple consequences, a thorough analysis must consider outcomes in a multivariateframework that accounts for the multiple effects of actions and the competition among various outcomes. This multivariate perspective raises serious questions about the value of univariate summaries, which will be taken up after the next section. Are socioeconomic indicators causes? ==================================== The theory outlined above is strictly a theory of effects of actionsor interventions. It does not formalize all ordinary-language or intuitive uses of the words \"cause\" and \"effect\". Two naïve extreme reactions to this limitation have been common: one that denies it is meaningful to talk of causes that are not actions and so restricts \"causes\" to interventions \[[@B33]\], and one that rejects counterfactual theory outright (see the discussion of Maldonado and Greenland \[[@B23]\]). But two types of constructive reactions have also appeared. The first type generalizes the theory to encompass nonactions as causes, a prime example being the many-worlds theory \[[@B5]\]. While this generalization may best capture ordinary language, it is very controversial and not suitably operationalized for everyday use. The second constructive response accepts the limitations of the restricted theory and instead seeks to identify potential actions within ordinary events. This approach recognizes that certain \"causes\" are best treated as intermediate outcomes; one then traces the etiology of such \"causes\" back to events with intervention potential, or else treats such \"causes\" as conditioning events and searches for actions that modify or prevent the ultimate outcomes. Earthquakes, which cause extensive destruction, provide neutral examples of unmodifiable causes. An earthquake, y~1~, is the outcome of a long and complex chain of events with little intervention potential under today\'s technology. Perhaps someday we will be capable of interventions that lead to dissipation of crustal stresses with less ensuing damage. But for now, mere prediction would be a major achievement and would facilitate actions to prevent damage when an earthquake occurs. An example of such an action is the enforcement of strict building codes in earthquake-prone regions. Less neutral examples are provided by common measures of education, such as the classification \"No high-school diploma\", \"High-school diploma\", \"Some college, no degree\", \"Two-year degree\", \"Four-year degree\" and \"Graduate degree\". People believe that education leads to more income and health. But how well do differences in the observed education measure predict the effects of real interventions such as affirmative action, public-school improvements, or scholarship programs? For policy purposes, it is the implementation and evaluation of such programs that matter; the ensuing changes in education measures are only intermediates between the programs and the ultimate goals of improved social, economic and health outcomes. The value of restricting counterfactual models to interventions is that it forces us to explain observed associationsbetween risk factors and health as outcomesof potentially changeable events. Consider a highly charged example, \"race\", which when measured as \"white\" vs. \"black\" is strongly associated with many health events. People talk of race as a \"cause\". But to do something about racial disparities in health outcomes (which is to say, to eliminate the observed association of race and health), we must explain their origin in terms of changeable causes, such as disparities in school funding, availability of individual college funding, prevalence of racist attitudes, etc. Finding feasible interventions and estimating their costs and benefits is required to address observed disparities; asserting or denying that \"race\" is a cause does not help this endeavor. Should different outcomes be summarized in a single number? =========================================================== Two distinct connotations of summary measure appear extant: the first and most common presumes that the measure summarizes a single outcome variable with a single number. Classic examples include the mortality rate and the life expectancy. The second connotation, largely confined to statistics and physical sciences, allows a summary to be a vector, that is, an ordered list of numbers that summarize different dimensions of a system. An example of such a multidimensionalor multivariatepopulation summary would be the list containing life expectancy, health expectancy, health gap and the proportions of deaths due to various causes (e.g. starvation, violence, infectious disease, heart disease, stroke, cancer). It should first be noted that all the earlier concepts and discussion apply equally to any action or outcome, whether unidimensional or multidimensional. In particular, the potential outcomes, Y(x~j~), may represent outcome vectors and the alternative actions, x~0~, x~1~, etc., may also be vectors; for example, x~0~could specify that 30%, 40% and 30% of a fixed budget be allocated to family planning, sanitation and medical supplies, respectively, and x~1~specifies a different allocation scheme. The chief problem in expanding to the multidimensional perspective is the limited number of dimensions that the human mind can contemplate at once. Because that limitation is a key motive for summarization, it is essential to keep track of what is lost in the dimensionality reduction that defines summarization. It also is essential to keep track of the valuesthat influence (or should influence) what is kept and what is lost. Summary measures of population health serve no good purpose when they strongly confound valuations, which vary by individual preference, culture, etc., with measures of occurrence and effect (which are presumably matters of scientific fact, albeit subject to uncertainty). For example, many individuals, in continuing to smoke, explain their behavior as stemming from a conscious preference to die sooner from cardiovascular disease or cancer than survive until mental or neurological deficit is nearly inevitable. For such individuals, measures such as healthy years of life lost due to smoking represent a conflation of someone else\'s values with the factual risks of smoking, because that summary ignores preferences among various morbidity and mortality outcomes affected by smoking. To give the individual the information necessary for personal choice, we must supply a multidimensional summary that includes lifetime risks of different diseases. Moving to the societal level, healthy years of life lost due to smoking not only neglects the differences in resource allocation that must exist between present (actual) society and a counterfactual tobacco-free society, it also neglects differences in absolute and proportional morbidity and mortality with and without tobacco use. This neglect is addressed by measures of the economic cost of tobacco use, and by absolute and proportional morbidity and mortality comparisons. By providing all these measures, we shift to a multidimensional summary of tobacco impact. My intention in raising these issues is not to offer a solution to a specific summarization problem. Rather, it is to remind those facing a choice among measures that candidates need not (and, for policy purposes, should not) be limited to unidimensional summaries. While our ability to think in several dimensions is limited, it can be improved with practice. That practice has proven crucial in attacking problems in physics and engineering, and there is no reason to suppose it is less important in tackling more complex social policy issues. In instances in which many different people must make informed choices based on the same scientific data, but with different values, multidimensional measures are essential if we are to provide each person and each executive body with sufficient information for rational choice. Conflicts of interest ===================== The author(s) declare that they have no competing interests. Acknowledgements ================ This article is an updated version of Greenland, S. (2002), Causality theory for policy uses of epidemiologic measures, which appeared as Ch. 6.2 in Murray et al. \[[@B2]\].	PubMed Central	2024-06-05T03:55:59.684371	2005-5-27	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180843/", "journal": "Emerg Themes Epidemiol. 2005 May 27; 2:5", "authors": [ { "first": "Sander", "last": "Greenland" } ] }
PMC1180844	Background ========== Diabetes is one of the most common chronic diseases of childhood. Adjustment to the disease and the demands of its complex regimen are formidable tasks even for adults. Children face these demands in the context of already challenging normative developmental tasks. Understanding children\'s diabetes-related problems can inform intervention designed to improve medical outcomes and quality of life for this population. Response shift is a theoretical construct that provides a framework for this investigation. In essence, it posits that people can adjust how they think about their quality of life when they encounter relevant new information. In this model, antecedents(e.g., demographics, personality), interact with a catalyst(intervention or change in health status) to elicit psychological mechanisms(e.g., social comparison) in order to accommodate the catalyst. Response shiftthen influences one\'s quality of life evaluation (see figure [1](#F1){ref-type="fig"}). According to Schwartz & Sprangers \[[@B1]\], response shift per se refers to a change in one\'s evaluation of quality of life as a result of: (a) a redefinition of the target construct (i.e., reconceptualization); (b) a change in values (i.e., the importance of component domains constituting the target construct), or (c) a change in internal standards of measurement (scale recalibration in psychometric terms). A simple example using children with diabetes may illustrate some aspects of this model. There is a 15-year old boy with diabetes who enjoys soccer (antecendents). His diabetes is treated with multiple injection therapy and he considers his quality of life quite good, about 8/10. Then, he joins a diabetes support group and meets a girl using an insulin pump (catalyst). Because of her pump therapy, she has greater flexibility with sports and recreation than he does. He compares his lifestyle to hers (social comparison). He starts to value flexibility in lifestyle more than he used to (change in values), and begins to consider that his diabetes-related quality of life is dependent not only on adequate glucose control, but also on how flexible his lifestyle is (reconceptualization). If asked to rate his quality of life, he would now say that it was really a 6/10, not 8/10 as he had originally estimated (scale recalibration). Since learning about how a pump might better accommodate his athletic lifestyle, he has recalibrated the scale he uses to evaluate his quality of life, resulting in response shift. Simply put, scale recalibration is a cognitive reappraisal process that occurs after an experience such that the reappraisal differs from the original appraisal before the experience. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Response Shift Theoretical Model. Reprinted with permission from Sprangers & Schwartz (1999). Reprinted from Social Science and Medicine, 48, 1507--1515, copyright 1999, with permission from Elsevier Science. ::: ![](1477-7525-3-38-1) ::: The model specifically allows response shift to vary in direction and magnitude. For example, one can imagine that the teenager described above might recalibrate his quality of life as betterthan originally determined, if he compares himself to someone worse off, such as someone with serious complications of diabetes. Further, the model is dynamic in that a feedback loop allows the response shift to affect the mechanisms that were activated in the production of the initial response shift. Study designs that model the relationships among these constructs are ultimately desirable. However, Brossart, Clay and Willson \[[@B2]\] have stated that given the lack of response shift research with pediatric populations, investigations that simply try to detect a response shift are necessary. The current study did this by investigating scale recalibration in children attending a summer camp for children with diabetes. Among the many response shift assessment approaches available, the thentest design approach is one of the most commonly used \[[@B2]\]. It is a well-established method in the education discipline that is also gaining wider use in the social sciences. The \'thentest\' captures changes in internal standards of measurement, or scale recalibration by using a retrospective pre-test design. At the posttest session, participants fill out the self-report measure twice. First, they report how they perceive themselves at the present (conventional posttest). Immediately after, they also provide a renewed judgment about their baseline level of functioning (thentest). By taking the posttest and thentest in close succession, it is assumed that these measures will be completed with respect to the same internal standard of measurement. The comparison of the original and reconsidered quality of life scores reflects scale recalibration. In the current study, the catalyst was attendance at a two-week summer camp specifically designed for children with diabetes. While summer camp is not a treatment for diabetes per se, children who attend do have diabetes specific psychosocial experiences that may serve as a catalyst for response shift. These include psychoeducation, exposure to positive role models, skill development, symptom reduction, as well as emotional support for the camper and family. Camps provide a unique \'diabetic world\' in which diabetes is the norm and children have the opportunity to communicate with others similar to themselves, view others living well with diabetes, learn about their illness, increase their independence, and make self-management mistakes in a safe environment. Diabetes camp may thus activate some of the response shift mechanisms of coping, social comparison, social support, goal reordering, and reframing of expectations that have the potential to profoundly influence children\'s perceived diabetes-related problems. Indeed, two reviews of psychosocial interventions for children with chronic health conditions have discussed the value of the summer camp experience \[[@B3],[@B4]\]. The purpose of this study was to investigate response shift in children. Given the lack of research on response shift in both children and persons with diabetes, we conducted an exploratory study. Questions of particular interest were: 1) would children show evidence of scale recalibration? And 2) if scale recalibration does occur, is it related to diabetes control? Methods ======= Sample ------ Participants were children attending an overnight summer camp for children with diabetes, and their families. The camp draws mostly northern New England families. Each year campers age 8--15 attend a 2-week session. The majority of the staff also has diabetes. ### Procedures ### Year 1 One week prior to the two-week camp session, a letter was sent to the parents of campers, describing the study. Parents were sent a consent form for themselves, an assent form for their child, a survey of disease and demographic data, and a questionnaire for them to review and administer to their child. Parents were asked to let the child complete the questionnaire as independently as possible. Upon their arrival at camp, the materials were collected from parents. At the end of the two-week camp session, child participants were asked to complete the assessments again as a traditional posttest. They were also asked to complete the questionnaire using the thentest approach. They were given the instructions \"I would like you to answer this questionnaire based on how you now think you were doing before camp. In other words, now that you have been to camp, how do you think you were reallydoing before?\" It was emphasized that they were not to recall their original responses, but rather to provide a renewed judgment. All children heard instructions in which there were examples of no scale recalibration, as well as positive and negative scale recalibration. All children claimed to understand the task. If the investigator suspected poor comprehension, the child was asked to retell the directions to the investigators. ### Year 2 Just prior to the camp session the following year, the disease and demographic surveys were sent to parents whose children had participated in year 1. Upon their arrival at camp, the materials were collected from parents. See figure [2](#F2){ref-type="fig"} for study design timeline. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Study timeline. ::: ![](1477-7525-3-38-2) ::: Measures ======== Problem Areas in Diabetes (PAID) -------------------------------- The PAID is a 20-item questionnaire that taps into patient\'s subjective feelings about difficulties with their diabetes \[[@B5]\]. Problem areas include difficult feelings about diabetes, interpersonal problems, and frustration with aspects of the regimen. Items are rated on a 6-point Likert scale from \"No problem\" to \"Serious problem\". Examples include \"Worrying about low blood sugar reactions\" and \"Feeling burned out by the constant effort needed to manage diabetes\". This scale has been shown to have adequate construct and discriminant validity \[[@B5]\], high internal consistency \[[@B5],[@B6]\], cross cultural validity \[[@B7]\] as well as sensitivity and sound test-retest reliability over 2 months (r~t-rt~= 0.83) \[[@B8]\]. While the measure has been used primarily with adults, a Spanish version used with children showed good criterion related validity with higher PAID scores related to poorer glycemic control \[[@B9]\]. In examining change in scores, higher thentest than pretest scores are viewed as positive scale recalibration, because respondents raise their scores retrospectively, endorsing more diabetes-related problems. The corollary is that lower thentest than pretest scores are considered negative scale recalibration because respondents lower their scores retrospectively, endorsing fewer diabetes-related problems. Demographic and disease variables --------------------------------- Parents completed a survey of demographic and disease variables including age, duration of disease, number of unscheduled doctor and emergency medical appointments in the past year, and years spent at diabetes camp. Parents also reported their child\'s most recent HbA1c, the average blood glucose concentration over the preceding 6--10 weeks. Children were required to have an HbA1c test prior to coming to camp, and parents were required by the camp to bring the lab results with them to the first day of camp. Normal values are \<6.0, and the goal for people with diabetes is generally \<7.0. HbA1c is the gold standard indicator of blood sugar, or glycemic control \[[@B10]\]. Small differences in HbA1c are clinically meaningful. Across prospective trials, every 1-point decrease in HbA1c is associated with a 30--35% decreased risk for long-term vascular complications that lead to blindness, kidney failure, and amputation \[[@B11]\]. Furthermore there is no clinical threshold for decreased risk; anydecrease in HbA1c leads to decreased risk for complications \[[@B10]\]. Data Analysis ------------- Internal consistency reliability of the PAID with this sample was investigated by calculating Cronbach\'s coefficient alpha. Differences between PAID pretest, posttest, and thentest scores were analyzed with paired samples t-tests. Group differences were analyzed with independent t-tests. The relationship between HbA1c and scale recalibration was investigated by calculating zero order and partial correlations. Results ======= Thirty-two percent of campers (n = 38) and parents (n = 38) handed in completed questionnaires on the first day of camp. There were no apparent differences between responders and non-responders in age, sex, HbA1c, duration of diabetes, and type of treatment regimen, the data for which were available to the investigator in aggregated form. At year 1, on average participants were 12 years old, had diabetes for approximately 6 years and had been attending diabetes camp for 3 years. Glycemic control was suboptimal, HbA1c M = 8.2. At year 2, 40% of year 1 campers participated, yielding a subset of n = 15. Approximately half of the attrition was due to lack of interest in completing the study, and the other half was due to families not returning to camp the second year. The subset of 15 participants was very similar to the larger group at year 1, except of course for being slightly older because they had aged 1 year. Mean HbA1c did not change from year 1 (M = 8.2, SD = 1.2) to year 2 (M = 8.1, SD = 1.5). See Table [1](#T1){ref-type="table"} for means and standard deviations for year 1 and year 2. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Means and standard deviations of demographic and disease variables ::: Year 1 n = 38 Mean (SD) Year 2 n = 15 Mean (SD) --------------------------------------------- ----------------------------- ----------------------------- Sex % (n) Male 39.5 (15) 46.7 (7) Female 60.5 (23) 53.3 (8) Age 11.9 (1.8) 12.3 (1.5) Age at diagnosis 6.2 (3.6) 6.3 (3.5) Years since diagnosis 5.8 (3.2) 6.3 (3.6) Most recent HbA1c 8.2 (1.2) 8.1 (1.5) \# Injections/day 3.3 (1.1) 3.3 (1.2) \# Children on CSII % (n) 24 (9) 20 (3) Diabetes sick days from school in last year 5.7 (10.1) 5.7 (15.1) Diabetes hospitalizations in last year 0.32 (1.1) 0.13 (0.5) DKA episodes in last year 0.95 (2.4) 0.4 (1.1) Hypoglycemic episodes in last month 5.9 (6.2) 4.3 (3.2) Years at camp 2.7 (1.6) 3.3 (1.9) \# of Siblings 1.5 (1.0) 1.2 (0.7) Parent education (in years) 14.6 (2.4) 15.5 (2.4) Parent marital status % (n) Single/separated/divorced & living alone 18.4 (8) 20.0 (3) Single/separted/divorced & cohabitating 5.3 (2) 0 (0) Married 73.7 (28) 80.0 (12) School performance % (n) Very poorly 2.6 (1) 6.7 (1) Poorly 10.5 (4) 6.7 (1) Ok 13.1 (5) 6.7 (1) Well 15.8 (6) 26.7 (4) Very well 57.9 (22) 53.3 (8) ::: PAID scores showed good internal consistency, baseline PAID Cronbach\'s alpha =.92, posttest PAID Cronbach\'s alpha = .94, and thentest PAID Cronbach\'s alpha = .96. These coefficients are similar to those found with adults (e.g., Chronbach\'s alpha = .95) \[[@B5]\]. Observed PAID means in our sample were baseline M = 39.8, posttest M = 34.9, and thentest M = 43.9. Overall, PAID scores in our sample were higher than published means for adults with type 1 diabetes (e.g., M = 32.9) \[[@B5]\]. This indicates that the children in our sample endorsed more diabetes-related problems than have been observed among adults. PAID data were analyzed with a paired samples t-test. Perceived problems decreased significantly from baseline (M = 39.8, SD = 16.8) to posttest (M = 34.9, SD = 14.9), t(38) = 3.12, \p \<.01. See table [2](#T2){ref-type="table"}. Comparing thentest to posttest, an even larger difference was found between thentest and posttest scores, suggesting scale recalibration. On average, participants\' new judgment (M = 43.9, SD = 20.9) was that they had had more problems at baseline than they originally endorsed (M = 39.8, SD = 16.8). See Figure [3](#F3){ref-type="fig"}. Not all participants showed the same direction of scale recalibration. As reflected in the group mean, two thirds of participants indicated that they had had more problems at baseline than they originally endorsed (positive scale recalibration; Δ M = 13.2, SD = 13.3). However, one-third of participants indicated that they had had fewerproblems at baseline than they were originally aware of (negative scale recalibration; Δ M = -13.2, SD = 12.2). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Means and (SD) for PAID time1, time2, and thentest for total sample and for children over 11 ::: PAID time1* PAID time2 PAID thentest ----------------------- ---------------- ---------------- ------------------- Total sample (n = 38) 39.8 (16.8) 34.9 (15.9) 43.9 (20.9) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### PAID Thentest Results. ::: ![](1477-7525-3-38-3) ::: There was no association between age or number of years at diabetes camp on the one hand and baseline PAID, posttest PAID, thentest PAID, absolute value of scale recalibration, or direction of scale recalibration on the other hand. The association between HbA1c and scale recalibration was investigated. Baseline HbA1c levels were significantly positively correlated with thentest scores, r = .35, \p \< .05. Higher HbA1c was associated with higher thentest scores. Neither PAID baseline scores (r = -.10, p = .60), nor PAID posttest scores (r = -.06, p = .74) were related to HbA1c. Groups who reported a positive vs. negative scale recalibration effect were compared. Those with a positive scale recalibration had nonsignificantly higher HbA1c compared to those with a negative scale recalibration (M = 8.5 vs. 7.7, p = .09). One-year follow up data were available for a subset of 15 children. PAID thentest scores were associated with year 2 HbA1c. Partial correlation analysis between PAID thentest scores and year 2 HbA1c, controlling for year 1 HbA1c, showed that higher PAID thentest scores were correlated with higher year 2 HbA1c, r = .58, \p \< .05. Neither PAID pretest scores (r = .33, p = .30) nor PAID posttest scores (r = .16, p = .62) were correlated with year 2 HbA1c. Statistical comparison of groups who reported a positive vs. negative scale recalibration was not possible due to small n in each group and very low statistical power. Nonetheless, it is worth noting that means were in the same direction and of similar magnitude to those seen at baseline. Those with a positive scale recalibration had nonsignificantly higher year 2 HbA1c compared to those with a negative scale recalibration (M = 8.2 vs. 7.7). Discussion ========== This study explored response shift in children attending diabetes summer camp. Specifically, it asked whether children with diabetes evidence scale recalibration, and if so, whether scale recalibration is related to glycemic control. Children with diabetes did in fact show scale recalibration, suggesting that response shift occurs in children with diabetes. Children provided renewed judgment of their pretest functioning, reporting that, on average, they had been experiencing more diabetes-related problems than they were originally aware of. Furthermore, scale recalibration was related to glycemic control. In year 1 cross sectional analysis, children with higher thentest scores had higher HbA1c. Higher thentest scores were also related to higher HbA1c at one-year follow up, even after taking into account baseline HbA1c. Furthermore, there was a trend for an association between the direction of scale recalibration and glycemic control. Children whose retrospective assessment of diabetes related problems increased showed nonsignificantly higher HbA1c at both baseline and at one-year follow up, compared to children whose retrospective assessment of diabetes related problems decreased. The high level of diabetes-related problems observed in these children relative to adults, and their relatively poor glycemic control relative to clinical guidelines, speaks to the need for investigation of this population. These results raise as many questions as they answer. First, what are the mechanisms that could cause such a scale recalibration? Sprangers and Schwartz \[[@B11]\] suggest that coping, social support, goal reordering, reframing expectations, and social comparison may all be active mechanisms in response shift. In the context of the current study, one might hypothesize that social comparison, social support, and reframing expectations could be important mechanisms. Participants were in surrounded by other children with diabetes, and following the structured diabetes regimen at camp. Secondly, why do some children show a positive scale recalibration, and others a negative recalibration? Perhaps antecedents influence this, or mechanisms work differently in different individuals. For example, one might hypothesize differential effects of upward and downward social comparison for those in adequate vs. suboptimal glycemic control. Another question that arises is how the scale recalibration influences subsequent glycemic control. One might hope that children who had this experience -- a realization that their diabetes was more problematic than they had originally perceived -- would make healthy behavior changes such as testing blood sugar more frequently and come back to camp the next year in tighter glycemic control. However, this was not the case. It is well documented that having the awareness of a health related problem is not in itself enough to induce behavior change. Knowledge of an unsatisfactory state of affairs is a necessary, but insufficient, condition for behavior change. Other factors are also needed such as skills, problem solving ability, readiness, self-efficacy, and the belief that such behavior changes will make a difference in health outcomes. In children, parental involvement is also key -- a child\'s awareness and behaviors occur in the context of parental control. Furthermore, in diabetes, it is documented that increased adherence to the regimen does not always lead to a direct benefit in glycemic control. There are numerous reasons, not measured in this study, that may have prevented renewed judgment about diabetes-related problems from translating into glycemic improvement. It is important to note that age and the number of years that children have attended diabetes summer camp was not associated with scale recalibration. Returning campers presumably had had reasonably similar prior camp experiences. Yet, the scale recalibration still occurred. It is possible that response shift in general, and scale recalibration specifically, occur repeatedly over time, with each significant disease related experience. This would certainly concur with anecdotal accounts that repeated years at camp serve as \'booster sessions\' that reinforce previous experiences and benefits. If replicated, these findings may point to an opening for intervention. These data suggest that children in poor diabetes control who participate in an intervention can reflect upon their previous functioning and provide a renewed, and perhaps more accurate, judgment regarding diabetes problems. This may be a good time to intervene on skills, problem solving, motivation, self-efficacy, and health beliefs. Perhaps children who see their diabetes in a new light will be primed to receive an intervention that will produce health behavior change and subsequent glycemic control. Anecdotally, campers report that when they leave camp they are very motivated to improve diabetes self-management at home in order to maintain gains made at camp. However, they also report that after several months, the motivation declines to baseline levels, and self-management relapses. Campers who return annually describe the need for \'booster\' camp sessions. Irrespective of any treatment implications, this study highlights the importance of response shift in research using self-report measures with children. Attention to response shift in research with adults has been advocated for several years, but to date the phenomenon of response shift has not been investigated in children. The importance of these findings is not only the relationship of scale recalibration to glycemic control, but the evidence of scale recalibration at allin children. This finding is important to two lines of investigation \[[@B12]\]. Observational studies of the natural course of living with chronic illness may benefit from studying response shift explicitly, as the subject of investigation. Such studies could describe whether and how quality of life or diabetes-related problems change over the lifespan, and how response shift affects these changes. Response shift may also be important for pediatric treatment outcome research. Outcome studies may benefit from taking response shift into accountwhen detecting treatment effects. However, neither approach is feasible until it can first be adequately demonstrated that response shift occurs in children. Limitations =========== These data need to be interpreted with caution, given several limitations of this study. First, there was a low response rate (32%) which may reflect selection bias. However, comparison of responders with non-responders showed no differences in age, sex, HbA1c, duration of diabetes, or diabetes treatment regimen. Second, these children were from White, predominantly middle class families in New England, which certainly limits the generalization of its findings. Third, disease variables were reported by parents. However, each child was required to have the written results of a physical exam performed prior to camp that included the most recent HbA1c information. Thus, parents had accurate HbA1c information available to them when completing the parent questionnaire, decreasing potential unreliability of HBA1c data. Fourth, the PAID has not typically been used with children and it is not known how much parental help was necessary to complete the baseline PAID. However, high internal consistency coefficients similar to those found with adult type 1 diabetes patients suggest that the PAID performed well. A control group was not simultaneously studied, so alternative explanations for PAID changes cannot be ruled out. Finally, despite its increasingly popular use, the thentest has limitations that warrant attention. These limitations are primarily related to the difficulty in interpreting observed differences between pretest and thentest scores. That is, what appears to be a thentest effect could also be attributed to memory difficulties, social desirability, recall bias, effort justification, or unreliability of the measure. There are several ways to increase the confidence with which one can interpret thentest results. First, keeping the timeframe of recall to the minimum necessary to answer the research question reduces the possibility of memory difficulties \[[@B13]\]. Second, respondents should be given instructions for how to answer (or not answer) items to which they cannot recall their previous functioning.Third, asking specific rather than general questions may reduce recall bias. Considerable research has shown that specific questions are answered more reliably and with greater validity than general questions \[[@B14]\]. Fourth, the effects of social desirability and effort justification can be mitigated by the nature of the instructions given to the respondents for the thentest. Finally, a reliable measure with acceptable test-retest coefficients should be used. Each of these techniques to increase the reliability and validity of the thentest were employed in the present study. Conclusion ========== Children with diabetes exhibit scale recalibration in their reporting of diabetes-related problems after a 2-week summer camp experience. The small sample and uncontrolled design pose limitations, but results suggest that scale recalibration is related to glycemic control, both cross-sectionally and prospectively at 1 year follow-up. The children in this study endorsed more diabetes-related problems than published adult samples, and have suboptimal glycemic control, underscoring the need for further investigation of disease-related quality of life in this population. Future directions ================= Further research should specify conditions under which response shifts would be expected to occur and those in which they would not be expected to occur, and matched samples from each circumstance should be compared for response shift. Other core constructs of response shift such as reconceptualization and change in values should also be investigated. Individual differences such as family variables, or personality traits such as optimism may influence the magnitude and direction of response shift. The mechanisms of response shift should be investigated. These areas are ripe for investigation.	PubMed Central	2024-06-05T03:55:59.687502	2005-6-14	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180844/", "journal": "Health Qual Life Outcomes. 2005 Jun 14; 3:38", "authors": [ { "first": "Julie A", "last": "Wagner" } ] }
PMC1180845	Background ========== Urinary tract infections (UTIs) are one of the most common conditions seen by general practitioners, representing a significant healthcare cost burden \[[@B1]\]. In the USA alone there are an estimated seven million physician visits and one million emergency department visits each year specifically for UTIs \[[@B2],[@B3]\]. Although it has been reported that half of all women experience at least one UTI in their lifetime \[[@B2],[@B3]\], the incidence of UTI is difficult to assess accurately because UTI is not well reported. UTIs may be either \'uncomplicated\' or \'complicated\'. Complicated UTIs are associated with metabolic, functional, or anatomic abnormalities. An acute uncomplicated UTI (uUTI) (also referred to as cystitis, acute cystitis, and dysuria-frequency syndrome) has been defined in a number of ways. In women, it includes a clinical syndrome characterised by various combinations of dysuria (painful urination), frequency, urgency, gross haematuria, lower back and/or abdominal/suprapubic discomfort with pyuria and bacteriuria \[[@B4],[@B5]\]. As many as half of all patients with uUTI may not have bacteriuria according to established criteria, and have no known underlying renal or urologic dysfunction or obstruction \[[@B6]\]. While uUTI is generally considered a benign condition, it is associated with significant short-term morbidity. The nature and extent of these limitations has led to this benign classification being questioned \[[@B7]\]. Malterud and Baerheim \[[@B8]\] recently demonstrated a wide range of different symptoms in patients with uUTI. The authors concluded that traditional medical knowledge about symptom presentation in uUTI is insufficient, and that awareness of symptom diversity and highly individual presentation may lead to better recognition of the atypical patient, and to improved diagnosis, treatment, and care \[[@B9]\]. While there has been some research on the impact of uUTI on everyday activities, little is known about the impact of UTI on everyday routine. Foxman and Frerichs \[[@B10]\] reported that each episode of UTI results in an average of 6.1 symptomatic days and 2.4 restricted-activity days, as well as time lost from work. In a recent Canadian survey \[[@B11]\], almost half (48%) of Canadian women who had experienced a UTI reported that it affected at least one of their daily activities. Additionally, Decima Research Incorporated (2002) reported on the modification of the Williamson Functional Impairment Scale to measure outcomes of impairment due to UTI, such as time lost from usual daily activities. The Activity Impairment Assessment (AIA) was developed for the purposes of the present study by the authors in 2003 and was based on an existing work productivity measure (the Stanford Presenteeism Scale (SPS-6) \[[@B12]\] to evaluate the impact of health problems on individual performance and productivity. For the AIA measure, the concept of work was broadened to include other activities and social activities. The AIA includes 5 items, the content of which can be seen in table [1](#T1){ref-type="table"}. The present study was designed to validate the AIA in the assessment of activity impairment associated with uUTI. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Distribution of AIA responses to each individual item ::: AIA items and total score Mean (SD) Median Range --------------------------------- --------------- ------------ ----------- Cut down on time at work 1.52 (1.07) 2 0--4 Accomplished less 1.67 (1.10) 2 0--4 Limited in kind of work 1.44 (1.20) 2 0--4 Difficulty performing work 1.55 (1.20) 2 0--4 Interfered with social activity 1.38 (1.29) 2 0--4 Total score 7.56 (5.22) 2 0--4 ::: The data for the validation study came from a clinical trial of the effectiveness of CIPRO^®^XR (extended-release ciprofloxacin) in the treatment of uUTI. The trial was designed to assess time to improvement of symptoms, the size of any improvements, and the extent to which women can resume their daily activities following treatment. The psychometric properties of the AIA were assessed primarily by the pattern of associations between AIA scores and scores on the validated King\'s Health Questionnaire (KHQ). The assessed properties included internal consistency, convergent and divergent reliability, discriminant validity, and responsiveness. Methods ======= Assessment of reliability and validity -------------------------------------- Reliability and validity assessment was conducted in the context of a 3-day prospective, open-label, non-comparative, multi-centre clinical trial of CIPRO^®^XR 500 mg. Women with uUTI were recruited for entry to this trial between June 18^th^2003 and January 23^rd^2004. At first visit, prior to receiving the first dose of study medication, patients gave written informed consent, a urine sample for dipstick biochemical analysis for nitrites or leukocyte esterase, and a clear-catch midstream urine specimen for culture and sensitivity. Patients provided demographic and medical history details (age, ethnicity, years of education, employment status, previous history of uUTI, number of days since onset of uUTI before seeing physician) and completed the patient-reported questionnaires. Patient reported outcomes were recorded in electronic format using PalmOS™ palm pilots, with information transferred to a host computer. Activity Impairment Assessment (AIA) ------------------------------------ The AIA is a self-administered 5-item questionnaire assessing the amount of time, over the previous 24 hours, that an individual\'s work or regular activities have been impaired as a result of their UTI. Patients respond to AIA items on a 5-point Likert-type scale, with the response options \'none of the time\', \'a little of the time\', \'some of the time\', \'most of the time\' and \'all of the time\', scored 0--4. The AIA was administered on day 2, then every 24 hours until regular daily activities had been unimpaired by UTI for 24 hours. Administration was stopped after the patient indicated at two consecutive administrations of the questionnaire (over at least 24 hours) that she had no activity impairment. The measure was completed at Test of Cure (TOC) visit if the UTI symptoms had persisted or if regular daily activities were impaired. King\'s Health Questionnaire (KHQ) ---------------------------------- The KHQ is a self-administered questionnaire designed to assess the impact of urinary incontinence on QoL in women. The measure contains 21 questions that are scored in nine domains (general health perception, incontinence impact, role limitations, physical limitations, social limitations, personal relationships, emotions, sleep/energy, severity of urinary symptoms) \[[@B13]\]. Weighted summary scores in each domain range from 0 to 100, with higher scores indicating greater impairment. Part III of the questionnaire is a list of 10 bladder problems plus an \'other\' category. These items are not summed to form a domain score. The KHQ was chosen as the primary instrument in the validation of the AIA because it has been validated for use in the assessment of women with urinary problems \[[@B14],[@B15]\] and it contains questions on \"bladder problems\" (i.e., frequency, urgency, bladder pain etc.) and how much these problems presently affect the subject. The KHQ was administered at day 1, day 3, when the subject indicated that the UTI symptoms had resolved (no symptoms for 24 hours, or over three data capture points, whichever was longer), and at the TOC visit. Clinical evaluation involved an assessment of five UTI symptoms (dysuria, frequency, urgency, suprapubic pain, gross haematuria) rated \'none\', \'mild\', \'moderate\', or \'severe\', and scored 0--3. The clinical evaluations took place at visit 1, at any premature discontinuation (day 1 to day 3), and at the TOC visit. Statistical methods ------------------- The data were analysed primarily in SPSS version 12.0 \[[@B16]\]. The Normality of the distributions of the AIA and KHQ scores was assessed both visually (using the histogram and superimposed Normal curve) and numerically using the relation between the distribution mean and the standard deviation and Kolmogorov-Smirnov statistics. Where data were found to be non-normally distributed, non-parametric tests were used. Analysis of variance with post-hoc Tukey tests (or non-parametric Kruskal-Wallis tests with Bonferroni-corrected post-hoc Mann-Whitney tests) were used to compare group scores. Associations between two continuous variables (absolute or change scores) were assessed using Pearson\'s r, (or non-parametric Spearman\'s r~s~) correlation coefficients. Missing values were handled by excluding cases pairwise. Examination of the dimensionality of the AIA, and the functioning and fit of individual items, was undertaken, as described below, by fitting a Rasch unidimensional measurement model in RUMM2010 \[[@B17]\]. Statistical significance throughout was taken at the 5% level (P \< .05). AIA factor structure -------------------- Exploratory factor analysis using principle components extraction and varimax rotation was performed on the 5 AIA items to explore the structure of the measure. Factors were extracted if their eigenvalue was \>1. Domain scores of any resulting factors, or of a total score, were calculated as a sum of the component item scores. Item functioning ---------------- Because the AIA does not directly assess symptoms, it was appropriate to explore the performance of the AIA with the use of item response theory (IRT). IRT allows an understanding of how people respond to items and how items measure an underlying dimension \[[@B18]\]. One of the simplest IRT models is the one-parameter Rasch model. The Rasch model assumes that the probability of a respondent giving a \'correct\' answer (or responding at a given level on the scale) to a particular item is a logistic function of the relative distance between the item location parameter and the respondent location parameter \[[@B19]\]. The item-trait interaction Χ^2^statistic was used to assess the fit of the data to the model, and the item-person interaction mean and SD values (where perfect fit is indicated by a mean of zero and SD of 1.0) to assess the degree of consensus displayed collectively by all items of the instrument across persons located at differing \'ability\' (i.e., activity impairment). The divergence of each item from the Rasch model was assessed using the individual item fit Χ^2^statistic and residual (a residual \>3.0 is generally taken to indicate misfit and \<-3.0 that the item fits the model too closely), with item thresholds used to assessed whether, as a person\'s impairment increases, the probability of a maximum score on the item increases. The person-item threshold map was used to assess whether the questionnaire items represent respondents of all levels of impairment, and item characteristic curves whether any item significantly over- or under-estimates the impairment level. Internal consistency reliability -------------------------------- Cronbach\'s alpha statistics were calculated to assess the internal consistency reliability of the AIA scores. Convergent validity ------------------- The convergent validity of the AIA was assessed using Spearman\'s correlation coefficients between the AIA score and similar individual symptom and domain scores of the KHQ. The AIA score was expected to be significantly associated with the \'role limitations\', \'social limitations\', and \'physical limitations\' domain scores of the KHQ. Divergent validity ------------------ The divergent validity of the AIA was assessed using Spearman\'s correlation coefficients between the AIA score and dissimilar individual symptom and domain scores on the KHQ. Because short-term UTI is not expected to have a major impact on one\'s personal relationships, divergent validity was assessed by calculating the Spearman correlation coefficient between the AIA score and the KHQ \'Personal Relationships\' domain. Discriminant validity --------------------- The discriminant validity of the AIA was assessed by using Kruskal-Wallis analysis of variance to compare AIA domain scores (at first administration) between the initial clinical ratings for dysuria, frequency, urgency, suprapubic pain, and gross haematuria. It was hypothesised that AIA scores would differ significantly between the clinical groups. Responsiveness -------------- It was hypothesised that the change in AIA score from visit 1 to the time when symptoms were no longer present would be related to improvements in the clinical evaluation of UTI. Change scores for the AIA and clinical evaluations were calculated by subtracting the TOC scores from the initial scores. Spearman correlation coefficients were used to assess the degree of association between the measures of change and Kruskal-Wallis analysis of variance for comparisons of group mean change scores. Results ======= Patient characteristics ----------------------- Two hundred and seventy six women were recruited to the study. The mean (SD) age of the women was 33.0 (11.46) years (range 18--78 years). Although the sample was ethnically diverse \[12% (n = 34) black, 4% (n = 12) American Indian, 2% (n = 5) Hispanic\], the majority of the women (70%, n = 193) in the study were white. Thirty-two women (12%) did not give any details of their ethnic origin. Five women (1.8%) reported that they had attended grade school only, while around half of the women (52.4%, n = 143) completed high school. A further 40.3% (n = 110) attended college and 15 (5.5%) attended graduate school. In response to the question \'which best describes what you were doing in the past 6 months\', 49.5% (n = 135) reported that they had more than one role. One hundred and seventy five women (64.5%) were working full time, 65 (23.8%) were working part time, while 107 (39.2%) women were looking after the house and/or children full time. Fifty-three (19.4%) women were studying at university either full or part time, and 46 (6.8%) reported that they were engaged in some other role. The majority of women were considered to have moderate or severe dysuria (n = 208, 75.4%), frequency (n = 238, 86.2%), and urgency (n = 237, 85.9%) at baseline clinical evaluation. Suprapubic pain was less common, with the majority of women (n = 179, 64.9%) rated as having either mild or moderate subrapubic pain. Haematuria was reported as absent in 151 women (54.7%). Factor structure of the AIA --------------------------- On unforced principal components factor analysis on the AIA scores at baseline, one component was extracted that explained 78.6% of the variance in the data (Table [1](#T1){ref-type="table"}). The loadings for each item on the single component were = 0.85. This supported the use of the AIA items as a single scale (called the AIA total score), scored as the sum of the 5 individual item scores. Data distribution ----------------- Patients used the full range of responses for each AIA item, with the median response for each item being 2.0 (\'some of the time\') (Table [1](#T1){ref-type="table"}). The full range of scores was also used for the overall scale, with scores ranging from 0 (no limitations) to 20 (maximum level of limitations). Twenty-six patients (9.4%) reported no limitations at baseline with the distribution at lower levels of limitation appearing fairly uniform, but with smaller numbers of patients reporting extreme degrees of limitation (kurtosis = -0.98, SE = 0.29). The mean (SD) AIA total score was 7.56 (5.22) with a median of 7.00, also indicating non-Normality which was confirmed on Kolmogorov-Smirnov test (P \< .001). Association of AIA score with demographic variables --------------------------------------------------- There were no statistically significant associations between AIA scores and respondent age, ethnic group, or work status. Item functioning ---------------- Because factor analysis showed that the AIA is a uni-dimensional scale the one-parameter Rasch model was fitted to the data. Although the model fitted the data well (item-trait interaction Χ^2^= 20.32, P = .16), there was some suggestion of a small amount of misfit of the items and persons to the model (item-person interaction mean = -0.64, SD = 1.42). However, in terms of individual item fit, no Χ^2^statistic reached statistical significance and no residual was \>\\|3\\| (Table [2](#T2){ref-type="table"}). In addition, all items displayed ordered thresholds. The person-item threshold distribution map showed that the item thresholds cover the majority of respondent \'abilities\' (i.e., activity impairment) in the sample. Only four thresholds were positioned at the same severity level, suggesting that all the items and response options are valuable in measuring the full range of activity impairment levels. Finally, the item characteristic curves demonstrated that the AIA performs very well across all activity impairment severity levels and does not significantly overestimate or underestimate any group\'s activity level. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Individual item fit for the AIA items ::: AIA Item Residual Score Chi-square P-value -------------------------------------------------------------------------------------------- -------------------- ---------------- ------------- 1\. Cut down on the amount of time you spent on work or other activities 1.735 6.899 0.075 2\. Accomplished less than you would like -1.789 3.356 0.340 3\. Were limited in the kind of work or other activities -2.118 3.827 0.281 4\. Had difficulty performing work or other activities (for example, it took extra effort) -0.652 0.545 0.909 5\. Interfered with your social activities (like visiting friends, relatives, etc.) 2.577 5.696 0.127 ::: Internal consistency -------------------- The Cronbach\'s alpha coefficient for the AIA scale score was 0.93, confirming the internal consistency of the 5-item AIA. All items contributed equally to the reliability of the scale as all \'alpha if item deleted\' coefficients were between 0.90 and 0.92. Convergent and divergent validity --------------------------------- The convergent validity of the AIA was demonstrated by high and statistically significant correlations between the AIA score and the KHQ \'role limitations\' (r~s~= 0.76, P \< .001), \'social limitiations\' (r~s~= 0.71, P \< .001), and \'physical limitations\' (r~s~= 0.73, P \< .001) domains. Divergent validity was demonstrated by a low and non-significant correlation between the AIA score and the KHQ \'personal relationships\' domain (r~s~= 0.08, P = .201). Discriminant validity --------------------- The discriminant validity of the AIA was demonstrated by statistically significant differences in AIA scores between the clinical evaluations of dysuria, frequency, urgency, and suprapubic pain (Table [3](#T3){ref-type="table"}). In particular, the greater the clinical severity, the greater the AIA score (P \< .001). No significant difference in AIA score was found between the clinical evaluations of haematuria. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Discriminant validity of the AIA -- mean scores by clinical evaluation of 5 symptoms ::: Symptom Mean AIA total score by clinical evaluation Χ^2^ ----------------- ------------------------------------------------- ---------- ------------ ------------- ------------- Dysuria 3.00 5.97 7.19 10.23\\\* 22.04\\\* Frequency 3.71 3.03 7.50\\\* 9.32\\ 38.96\\\* Urgency 3.50 4.37 6.88\* 9.68\\\* 37.39\\\* Suprapubic pain 6.24 6.39 7.97 10.70\\ 22.46\\\* Gross hematuria 6.96 7.49 8.26 9.59 7.07 Mann Whitney U tests used to test differences between clinical evaluations ^a^None and Mild, ^b^Mild and Moderate, ^C^Moderate and Severe \p \< 0.05, \\p \< 0.01, \\\p \< 0.001 ::: Responsiveness -------------- The AIA score was responsive to changes in the clinical evaluation of \'Dysuria\' (r~s~= .28, P \< .001), \'Frequency\' (r~s~= .31, P \< .001), \'Urgency\' (r~s~= .31, P \< .001), \'Suprapubic pain\' (r~s~= .27, P \< .001), and \'Gross Haematuria\' (r~s~= .16, P = .008). In terms of mean AIA change by degree of improvement in clinical evaluation, clear trends were evident for increase in AIA improvement with increased degree of clinical evaluation improvement (Table [4](#T4){ref-type="table"}). ::: {#T4 .table-wrap} Table 4 ::: {.caption} ###### Mean (SD) AIA change scores by degree of symptom change ::: Clinical Evaluation ------------------------- ----------- ----- --------------- -------- Dysuria 0, 1 76 -5.32 (4.59) \<.001 2 138 -6.53 (4.95) 3 53 -10.08 (5.46) Frequency -1, 0, 1 48 -3.75 (3.73) \<.001 2 132 -6.84 (4.79) 3 87 -8.69 (5.72) Urgency 0, 1 51 -4.67 (4.03) \<.001 2 129 -6.34 (5.18) 3 87 -9.00 (5.18) Suprapubic pain -2, -1, 0 54 -5.30(5.24) \<.001 1 86 -6.09 (4.75) 2 89 -7.19 (5.18) 3 38 -10.24 (4.84) Gross hematuria 0 145 -6.19 (4.86) .003 1 54 -7.22 (5.46) 2 38 -7.08 (5.49) 3 30 -9.40 (5.48) ^a^positive score = improvement; negative score = deterioration ::: Discussion ========== This study has reported on the validation of the Activity Impairment Assessment (AIA) questionnaire in measuring the activity impairment associated with uncomplicated urinary tract infection (uUTI). Strengths of the study include the fact that the validation was conducted on a large sample of 276 ethnically and socio-economically diverse women with uUTI in the context of a (non-comparative) clinical trial. The trial was specifically designed for the validation of the new instrument and thus incorporated a validated measure (the KHQ) as well as a clinical evaluation of symptoms. On unforced principal components factor analysis, one component explained a high percentage (79%) of the variability in the AIA data. This component was named the AIA total score and was calculated as the sum of the five individual AIA items. The total score was found to have good internal consistency (Cronbach alpha = 0.93). However, it must be noted that Cronbach alpha scores \>0.90 may indicate that items are too similar \[[@B20],[@B21]\]. Mean scores for each AIA item indicated that feeling a lack of accomplishment and reducing time at work were the most frequently reported impairments in uUTI; interference with social activity was least commonly reported. The scores were significantly non-Normally distributed indicating that non-parametric tests should be used in the analysis of the AIA. The AIA total score was found to have excellent psychometric properties. Convergent validity was indicated by high and statistically significant correlations between the AIA score and related KHQ items and domains. Divergent validity was indicated by the small and non-significant correlation between the AIA score and the (unrelated) KHQ \'personal relationships\' domain. The AIA total score displayed excellent discriminant validity in relation to clinical evaluation. It discriminated well between all clinical evaluation groups with the exception of the \'gross haematuria\' evaluation, which might be expected to be less strongly related to UTI impairment. The AIA total score was also found to have very high levels of responsiveness, with strong associations existing between changes in AIA score and changes in clinical evaluation at TOC visit. These associations were all in the hypothesised direction and were again less associated with the \'gross haematuria\' evaluation. Conclusion ========== The AIA questionnaire enables the assessment of the amount of time an individual\'s work or regular activities have been impaired as a result of their UTI. The questionnaire is composed of one domain, a total score, which has been found to have excellent psychometric properties. While the measure was designed specifically for use in a clinical setting, it is likely that it will also be suitable for use in an epidemiological context and beyond. Furthermore, although the AIA is validated here in the context of UTI it could also easily be adapted and used to assess activity impairment in other disease areas. Authors\' contributions ======================= DJW drafted the manuscript. DJC performed the statistical analysis. KK and KG participated in the study\'s design and coordination and helped draft the manuscript. Acknowledgements ================ Bayer Healthcare Pharmaceuticals for sponsoring the study.	PubMed Central	2024-06-05T03:55:59.690072	2005-7-15	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180845/", "journal": "Health Qual Life Outcomes. 2005 Jul 15; 3:42", "authors": [ { "first": "Diane J", "last": "Wild" }, { "first": "Darren J", "last": "Clayson" }, { "first": "Karen", "last": "Keating" }, { "first": "Kathleen", "last": "Gondek" } ] }
PMC1180846	Background ========== Breast cancer is the most common cancer among women (excluding non-melanoma skin cancers). Early detection is the primary way to control breast cancer since survival drops sharply for late stage diagnoses\[[@B1]\] Since the proportion of late stage diagnoses in a geographic area can be viewed as a proxy for screening efficacy, this study determines whether the observed variations in the proportion of late stage cases is simply random or is statistically significant in space-time areas. A previous study looked at this same issue in Massachusetts using cases diagnosed between 1982 and 1986\[[@B2]\] It analyzed these data in aggregate and as a space-time model finding a single area with a significantly higher proportion of late stage cases than the rest of the state. No other studies cited in PubMed have included Massachusetts in a spatial or space-time proportion of late-stage breast cancer analysis. The objective of this study was to examine spatially the proportion of breast cancer cases diagnosed at late stage in Massachusetts from 1988 through 1997. It is not known whether the observed variation in geographical and temporal variations in the proportion of late stage cases is random or represents statistically significant excesses. This study examines whether there is excess variation, high or low, and whether such excesses are temporary or stable, and also examines the role of socioeconomic status (SES) and urban/rural status as covariates. Several studies have shown that low SES is a risk factor for diagnosis of breast cancer at late stage \[[@B3]-[@B8]\] Gregorio et al. found an increased likelihood of women in low-to-moderate income census tracts in Connecticut from 1986 to 1990.\[[@B7]\] However, for 1990--95, this disparity in SES and late stage diagnosis was greatly decreased from the previous time period. Living in a rural area as opposed to an urban area has also been shown to be associated with higher percentages of late stage diagnosis \[[@B9]-[@B12]\] This study analyzed surveillance data to identify those geographic areas that warrant closer attention because of their high or low proportion of late stage breast cancer. The department of public health can use this information to assess the effectiveness of screening and other programs. Methods ======= Ten years of data from the Massachusetts Cancer Registry (MCR) included 46,666 female invasive breast cancer cases diagnosed between 1988 and 1997. This study period was chosen since the previous study of the proportion of late stage breast cancer in Massachusetts \[[@B2]\] studied a period prior to our study period, 1982--1986. Also, at the time the study was initiated, 1997 was the most recent data available for analysis. For space-time analyses, we wanted 10 years to study, which made the study period 1988 to 1997. It should be noted that there is a lag of several years for cancer registries to verify and clean registry data prior to it being available for analysis. For each case, the record was designed to include information on place of residence classified according to the minor civil division (town code), ZIP Code, and census tract. The record also included the age at diagnosis, date of diagnosis, race, and stage of breast cancer where stage was the historical Surveillance, Epidemiology and End Results (SEER) summary stage: local, regional, distant and unknown. Distant stage alone was considered late stage. Aggregation unit ---------------- Census tracts were used as the geographic aggregation unit to conduct analyses. However, 12.5% (n = 5832) of the cases diagnosed in 1988--1997 could not be assigned a reliable residential census tract because of inaccuracies or omissions in the address information provided to the MCR. In most of these cases, a mailing address had been provided and, even after extensive research, MCR staff could not assign a reliable residential address for these patients at the time of diagnosis. Town and census tract boundaries were compared to assign the unassigned cases to tracts. For a town containing two or more census tracts, the cases missing census tracts were randomly assigned to tracts within the town based on the proportion of the town\'s female population each tract contributed. There were 4440 such cases, or 9.5% of all cases. The allocation of cases should therefore be free from systematic error and any error should be localized to a particular town, while the overall patterns remain correct. The proportion of late stage cases in each census tract was computed by dividing the number of late stage cases by total number of cases in that tract. Spatial analyses ---------------- The spatial scan statistic software SaTScan\[[@B13]\] was used to perform the space-time analyses. The spatial scan statistic assumes that the proportion of late stage cases follows a Bernoulli distribution. Since the Bernoulli probability model in SaTScan does not allow for covariate adjustment, the Poisson probability model was used. All incident cases of breast cancer in Massachusetts during the study period were used as the denominator for the Poisson probability model in SaTScan. The Poisson model was tested and found to be a very good estimate of the Bernoulli model, which found the same areas significant with the same risk ratios, but elevated p-values. Therefore, use of the Poisson model was more conservative than the Bernoulli model. According to the null hypothesis, the proportion of late stage cases in a particular location is equal throughout the state. Space-time analyses were performed so that the regional variations over the entire time period, 1988--1997, could be analyzed in a single model. The data were provided by MCR in one-year intervals. The data were left in one-year intervals for the space-time analyses to allow SaTScan to determine if a particular elevation or low was significant for all years, a group of consecutive years, or even just a single year. Purely spatial analyses were also performed, which do not take time into account. The maximum spatial cluster size was first set to include up to 50% of the cases, testing for both high and low excesses. Then it was set at 25% to test for high and low excesses separately and to discover smaller, more defined areas of excess and low proportion of late stage. Each area had a likelihood that was compared to the 9,999 likelihoods from the Monte Carlo trials based on a maximum cluster size that included 50% of the cases. Those areas with likelihoods within the top 500 likelihoods from the Monte Carlo trial were statistically significant at p \< 0.05. The results of the 25% spatial maximum analyses are presented in the Results section. The maximum temporal cluster size was set at 90% and also includes purely spatial clusters (temporal size = 100%). The overall SES and urban/rural status of each tract were determined and included as covariates separately and together to determine if SES or urban/rural status could account for the high or low areas by making them disappear. An SES index was created using the method of Yost et al in a principal component analysis using varimax rotation\[[@B14]\] When deciding on a method to measure SES, indices that included several variables was preferred in order to have a well rounded understanding of SES. Indices that used indicators available in the U.S. Census were examined. Some researchers chose only to look at one indicator from the census. \[[@B15]\] Yost \[[@B14]\] and Krieger \[[@B16]\] examined several indicators, therefore we evaluated their methods. Since the variables that Yost included in her studies accounted for more of the variance in a principal component analysis, her method was chosen for use in this study.\[[@B17]\] Two components accounted for 80% of the variance among seven economic measures obtained from the census. The first component explained 49.1% of the variance and was made up of median income, median rent, median house value, and percent with at least a high school diploma from the 1990 census\[[@B18]\] The first component is referred to as wealth. The second component explained 31.0% of the variance made up of the percent unemployed, percent working class, and percent below the poverty level. The second component is referred to as poverty. These principal component results were presented in an earlier study of breast cancer incidence in Massachusetts. \[[@B17]\] The two scores from the principal component analysis were first tested in a Poisson regression to determine the direction and statistical significance of their association with the proportion of late stage breast cancer in census tracts. They were then included in purely spatial and space-time models as covariates along with urban/rural status. Urban/rural status was obtained from the Massachusetts Institute for Social and Economic Research (MISER) website.\[[@B19]\] Towns were either classified as urban or rural. Tracts within a town were assigned the same classification as those towns. Tracts that included several towns were classified as rural since all towns within a tract were also classified as rural. This urban/rural classification was included in SaTScan analyses as a covariate by itself and along with the SES scores. The use of this binary classification of urban/rural status is consistent with other published studies of breast cancer incidence\[[@B7],[@B11],[@B20],[@B21]\] Although race was available for most patients, the non-white cells within tracts were very small, especially by year. After evaluation of its usefulness, it was determined that race would not produce meaningful results, and was therefore not included as a covariate in this study. Poisson regression was performed using the SES scores and urban/rural status as predictors of the number of cases diagnosed at late stage within tracts. This analysis was performed using PROC GENMOD in SAS\[[@B22]\] SES scores were categorized into equal quintiles. By dividing the parameter estimate of one category by another category within a SES score or urban/rural status, the percent increase or decrease of the number of cases diagnosed at late stage from one category to the next was determined. The figures were created using Maptitude software.\[[@B23]\] Results ======= Poisson regression ------------------ The Poisson regression of proportion of late stage diagnoses within each census tract on the SES and urban/rural status of each tract uncovered an increasing trend of diagnosis of late stage from the lowest category of the poverty component of 1 to the highest category, 5. The lowest poverty category has 38% fewer cases diagnosed at late stage than the highest category. Thus, the proportion of late stage cases increases with increased poverty across all five categories of poverty. The wealth component is not so clear-cut. There was not a constant increasing or decreasing trend from the lowest to the highest categories of wealth. However, the lower four categories of wealth all had higher estimates of late stage than the highest category of wealth, the variable was dichotomized where categories 1 to 4 equal 0 and category 5 equals 1. Late stage diagnosis was also affected by urban/rural status with urban tracts having about 12% more late stage diagnoses than rural tracts. Both SES scores were significant predictors of the number of cases diagnosed at late stage with p-values of 0.002 for wealth and less than 0.0001 for poverty. Urban/rural status was just significant with a p-value of 0.048. Table [1](#T1){ref-type="table"} displays the percent change of late stage diagnosis relative to the wealth and poverty SES categories of the census tracts. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Wealth and poverty SES. Relative changes in the proportion of late stage breast cancer associated with 2 levels of wealth and five levels of poverty. For wealth, category 2 represents the highest level of wealth. For poverty, category 5 represents the highest level of poverty. For example, the women in the highest poverty level, 5, had a proportion of late stage 37.4% lower than those in the lowest poverty level, category 1. ::: Categories Compared \% Change ------------- --------------------- ----------- Wealth SES 1--2 +17.3 Poverty SES 1--5 -37.4 2--5 -17.4 3--5 -16.3 4--5 -14.3 ::: Purely spatial analyses ----------------------- The geographic unit of analysis for all analyses was census tracts. The purely spatial scans analyzed the proportion of late stage cases with the maximum spatial window including up to 25% of the population at risk, which included all invasive breast cancer incidence diagnosed between 1988 and 1997. The observed count, relative risk (RR), and p-value for each area of excess or low proportion of late stage breast cancer can be found in Table [2](#T2){ref-type="table"} for each level of adjustment in the purely spatial analyses. The scan without covariates uncovered a large area north of Rhode Island with a statistically significantly lower proportion of late stage cases than expected. This area, Low A in Figure [1](#F1){ref-type="fig"}, was 19% lower than expected with 440 cases diagnosed at late stage where about 544 cases were expected. ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Purely spatial analysis. Proportion of late stage female breast cancer statistics for the purely spatial analyses, Massachusetts, 1988--1997. \* Area not significant for this analysis. ::: Low A Low B --------------------------------------- ------- ------- ------- ---- ------- ------- Not Adjusted 440 0.81 0.004 \* \* \* Adjusted for Urban/Rural Status 440 0.81 0.002 \* \* \* Adjusted for SES \* \* \* 94 0.63 0.009 Adjusted for SES & Urban/Rural Status \* \* \* 94 0.635 0.016 ::: ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Purely spatial, age-adjusted.Purely spatial analysis results for age-adjusted Massachusetts proportion of late state breast cancer diagnoses, 1988--1997. ::: ![](1476-072X-4-15-1) ::: The analyses were then adjusted for covariates. The analysis including urban/rural status found the same low area with no changes from the analysis without covariates. When the scan included wealth and poverty SES components as covariates, Low A was no longer significant. However, a new low area on Cape Cod and the islands, Low B, appeared with 37% fewer late stage cases than expected. This same area, Low B in Figure [2](#F2){ref-type="fig"}, persisted when the analysis was adjusted for both urban/rural status and SES. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Purely spatial, multiple adjustments.Purely spatial analysis results for socioeconomic status- and urban/rural status-adjusted Massachusetts proportion of late state breast cancer diagnoses, 1988--1997. ::: ![](1476-072X-4-15-2) ::: Space-time analyses ------------------- Space-time analyses used spatial windows that could include up to 25% of the population at risk, as well as varying periods of time. The maximum temporal window of 90% was used, and included purely spatial clusters, as well. The time frame, observed count, RR and p-value for each area of excess or low proportion of late stage cancer can be found in Table [3](#T3){ref-type="table"} for each level of adjustment in the space-time analyses. The time frame in the second column of the table applies to all levels of adjustment unless otherwise noted. The analysis without covariates identified one high and 2 low areas, shown in Figure [3](#F3){ref-type="fig"}. High 1 was significant from 1988 to 1990 and was 52% higher than expected with 236 late stage cases when about 155 were expected. Low B covering Cape Cod and the islands was statistically significant from 1993 to 1997 with 55% fewer cases than expected. The other low area west of Boston, Low C, had 40% fewer cases with 106 late stage cases when about 178 cases were expected. ::: {#T3 .table-wrap} Table 3 ::: {.caption} ###### Space-time analysis. Proportion of late stage female breast cancer statistics for the space-time analyses, Massachusetts, 1988--1997. ^a^Time frame for High 1 was 88--90. ^b^Time frame for Low A and Low B was 93--97. ^c^Observed count in all tracts for the area. ^d^Relative Risk Ratio. ^e^Time frame for Area 1 when adjusted for SES is 1989 alone. \Area not significant for this analysis. ::: High 1^a^ Low A^b^ Low B^b^ --------------------------------------- ----------- ---------- ---------- ---- ------ -------- ----- ------ -------- Not Adjusted 236 1.52 \<0.0001 39 0.45 0.002 106 0.60 0.001 Adjusted for Urban/Rural Status \ \* \* 39 0.46 0.008 106 0.59 0.0003 Adjusted for SES 103^e^ 1.85 0.006 39 0.42 0.0003 \* \* \* Adjusted for SES & Urban/Rural Status 103^e^ 1.85 0.008 39 0.42 0.0006 \* \* \* ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Space-time, age-adjusted.Space-time analysis results for age-adjusted Massachusetts proportion of late stage breast cancer diagnoses, 1988--1997. ::: ![](1476-072X-4-15-3) ::: When the analysis was adjusted for urban/rural status, High 1, seen in the analysis without covariates, was no longer significant. The lows west of Boston and on Cape Cod and the Islands remained significant. When the analyses were adjusted for wealth and poverty SES components, Low C, west of Boston, was no longer statistically significant. When the analyses were adjusted for SES and urban/rural status together, both Low B on Cape Cod and High 1 remained statistically significant. However, High 1 was only statistically significant for a single year: 1989. Discussion ========== In a previous study investigating the proportion of breast cancer cases diagnosed at late stage in Massachusetts between 1982 and 1986, a non-significant area of excess late stage was found east of Rhode Island, west of Cape Cod\[[@B2]\] Most of this same area was found to be statistically significantly high in the current analysis in the first three years of the study period. This area does not seem to persist as an area of significant excess after 1990, as shown in the unadjusted space-time analysis. The earlier study also found an area of significant excess in western Massachusetts. No statistically significant excesses or lows in Western Massachusetts were found in the current study. The concentrated area of excess proportion of late stage diagnoses in Figure [3](#F3){ref-type="fig"} existed only during the first 3 years of the study period. There do not appear to be any areas of significant excess proportion of late stage diagnoses in Massachusetts after 1990. The low areas identified in this study were low fro the last five of the study years. The Poisson regression showed an inverse relationship between the number of cases diagnosed at late stage and SES. This corresponds to past published studies, which found low SES as a risk factor for late stage diagnosis of breast cancer\[[@B3]-[@B7],[@B24]\] When the space-time analyses were adjusted for urban/rural status alone, the high area found in the unadjusted analysis was no longer significant. Most of the tracts included in High 1 were classified as urban. This is contrary to previous findings where rural areas were associated with higher percentages of late stage diagnosis \[[@B9]-[@B12]\] In an earlier study of breast cancer incidence in Massachusetts \[[@B17]\], the northwestern section of High 1 in the current study was found to have a high rate of breast cancer incidence. However, the area in the previous study was high from 1992 to 1997, which is a later time frame than the 1988 to 1990 period in the current study. Therefore, it seems as though more cases were being diagnosed and at an earlier stage in the 1990\'s as compared to the end of the 1980\'s. Eastern sections of Low A in the current study were found to have high incidence in the breast cancer incidence study of Massachusetts. \[[@B17]\] Although this area had a high rate of breast cancer incidence, the diagnoses are being made at an earlier stage. This seems as it might also be true for Low B on Cape Cod and western portions of Low C west of Boston. Researchers have reported that the following factors were associated with an increased risk of late stage diagnosis of breast cancer: old age \[[@B25]-[@B27]\]; foreign nationality and racial and ethnic minorities \[[@B4],[@B25],[@B26],[@B28],[@B29]\]; living in large households, non-participation in general health check-ups, low interest in health care \[[@B25]\]; a body mass index greater than or equal to 25 \[[@B30]\]; lack of health insurance or insured by Medicaid \[[@B6],[@B31]\]; low socioeconomic status \[[@B4],[@B6],[@B26],[@B28],[@B29]\]; and delay in seeking care \[6, 32\]. Arndt and colleagues also found that when tumors were detected by screening, the risk of late stage diagnosis was decreased \[[@B25]\]. Although data on these factors were unavailable for the current study, they might explain why certain areas have higher or lower proportions of late stage breast cancer. MCR collects the patient\'s usual address of residence when diagnosed, which is in the patient\'s medical records; this address is used when cases are aggregated to a geographical unit. It must be noted that the address of the patient at the time of diagnosis may not be related to the cause of their cancer since this is a slowly progressing disease; the patient may have moved from the place that caused the disease. However, the place where someone is living when diagnosed might be a factor in determining if the patient is diagnosed during early or late stage of the disease. Some areas are targeted for breast cancer screening programs by public health officials. If a woman with breast cancer lives in one of those areas, she may be more likely to be diagnosed at an earlier stage than if she lived in an area where there is not an intense breast cancer screening campaign. Since urban/rural status of census tracts was determined by the town the tracts were within or the towns that were within the tract, there is a potential weakness introduced. It is possible that towns classified as urban may have some rural sections, which may be separate tracts than the urban tract(s) of the town. The same could also be said for rural towns that may have urban tracts within the towns. However, the town level status was the smallest available geographic level with information on the urban/rural status. There were 9.5% of the total incident cases that were randomly assigned to a census tract since they could not be geocoded to a tract by MCR as discussed in the methods section. Because of the method used to assign these cases to tracts, there should not be any systematic error and any bias would be localized within a town. Since all significant areas are larger than a town, we can conclude that these assignments did not affect the overall patterns detected. Conclusion ========== Only one area of excess late stage breast cancer diagnoses was identified in the space-time analyses for the first three years of the study period and remained statistically significant after covariate adjustment. A low area, Low A, was found to be statistically significant in the purely spatial analyses was no longer significant after adjusting for SES. Low B on Cape Cod and the islands was not statistically significant in the purely spatial model until the model was adjusted for SES. In the space-time analysis, Low B is significant from 1993 to 1997 with or without covariate adjustment. Low C, west of Boston, was statistically significantly low for the last five years of the study period before the model was adjusted for SES. Authors\' contributions ======================= TJS: PI, responsible for design, funding, of project with overall responsibility for implementing the project, including the final paper. LMD: Principal data analysis, responsible for final checks on accuracy of data and all analyses, including their written interpretation. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Space-time, multiple adjustments.Space-time analysis results for socioeconomic status- and urban/rural status-adjusted Massachusetts proportion of late stage breast cancer diagnoses, 1988--1997. ::: ![](1476-072X-4-15-4) ::: Acknowledgements ================ This study was supported in part by a grant from the National Cancer Institute, CA81763-02 and the Massachusetts Department of Public Health.	PubMed Central	2024-06-05T03:55:59.692590	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180846/", "journal": "Int J Health Geogr. 2005 Jun 8; 4:15", "authors": [ { "first": "T Joseph", "last": "Sheehan" }, { "first": "Laurie M", "last": "DeChello" } ] }
PMC1180847	Background ========== Cutaneous T cell lymphoma (CTCL) is a malignant expansion of mature, clonal CD4 T cells with an affinity for epidermal localization \[[@B1]\]. The tumor cells proliferate in the epidermis around a central Langerhans cell (LC) and previous studies have demonstrated that immature DC play a crucial role in the life cycle of the malignancy \[[@B2]\]. The final stages of CTCL are characterized by systemic spread, immunosuppression and a poor prognosis. Despite the malignancy\'s dependence on immature DC for proliferative support, DC immunotherapy has been of benefit in this disease \[[@B3],[@B4]\]. Two strategies for the treatment of CTCL, extracorporeal photopheresis (ECP) and transimmunization, have been used to successfully treat this aggressive malignancy \[[@B4],[@B5]\]. The underlying principle of these treatments is extracorporeal establishment and re-infusion of malignant T cell-loaded DC \[[@B6]\]. In both therapies, a leukapheresis product is treated with the drug 8-methoxypsoralen (8-MOP) and passed through a plastic ultraviolet light (UVA) exposure plate. The 8-MOP intercalates in the DNA of nucleated cells and is cross-linked to adjacent pyrimidine bases by UVA light activation. The cross-link formation is a lethal defect and replicating cells are rendered apoptotic. At the same time, monocytes are activated by adherence and release from the plastic exposure plate surface and begin to transition into immature DC \[[@B6]\]. In the ECP treatment, both apoptotic CTCL cells and transitioning DC are re-infused into the patient immediately and association of the DC and apoptotic tumor cells occurs inefficiently in vivo. The transimmunization procedure was devised as a more effective modification of ECP and named to designate the transfer of tumor antigens to competent antigen presenting cells (APC) that could display the full complement of tumor antigens in the context of co-stimulatory and adhesion molecules. In the transimmunization procedure, the apoptotic malignant T cells and the transitioning DC are co-cultured overnight enabling the up-take of the apoptotic cells by the avidly phagocytic immature DC \[[@B6]\]. The activated monocytes produce cytokines that comprise the constituents of monocyte conditioned media thereby, potentiating the maturation of the malignant T cell-loaded DC \[[@B3]\]. The differentiating DC are re-infused the next day into the patient where they can further mature and have the potential to migrate to lymph nodes and induce anti-tumor immunity. In the current studies, we sought to explore the role of physical perturbation in the monocyte to DC transition by examining whether passage through a separation column that contains a porous matrix is sufficient to induce overnight DC differentiation from monocytes. Studies \[[@B7]\] suggest that trans-migrating monocytes passing through the small spaces of an endothelial cell layer become activated and assume the phenotype of immature DC. This monocyte-to-DC transition can be preserved by phagocytosis of particulate material such as zymosan \[[@B7]\]. We have also previously demonstrated that CD3-binding renders antigen-experienced proliferating CTCL cells apoptotic \[[@B2]\]. We therefore sought to take advantage of the dual observations of the role of physical stimulation in DC maturation and the rapid apoptotic cell death mediated by CD3-binding to develop in one day a clinically practical vaccine. We demonstrate that a simple one-step procedure using CD3-magnetic beads to render the malignant T cells apoptotic and the separation column matrix to simultaneously activate monocytes results in overnight production of apoptotic cell-loaded DC. These immature DC generated in the absence of cytokines could be driven to differentiate further when exogenous cytokines were added. Functional evaluation of the malignant T cell loaded DC, developed by this methodology, demonstrated a significantly enhanced stimulatory capacity in mixed leukocyte culture and the ability to promote CD8 T cell expansion and cytolytic capacity. Therefore, this approach yields malignant cell loaded DC in a rapid time-frame without extensive cell culture, exogenous factors or cell isolation and manipulation. This method may provide a clinically practical means for the production of immunogenic DC for cancer vaccine therapy. Materials and methods ===================== Patient Population ------------------ Therapeutic leukapheresis specimens were obtained from 7 CTCL patients (in accordance with the guidelines of the Yale human investigation committee). All patients had advanced disease with clonal CD4^+^T cell populations present in the peripheral circulation as determined by immunophenotyping with antibodies to the clonotypic variable region of family-specific T cell receptor (TCR) or polymerase chain reaction to detect rearrangements of the beta or gamma chain of the TCR. All patients were undergoing treatment with standard ECP. Cell Isolation -------------- Mononuclear cells (MNC) were isolated by centrifugation over a ficoll-hypaque gradient followed by two washes in RPMI 1640 (Gibco, Gaithersburg, MD) containing 10% AB serum and 2 mM EDTA. MNC (2 × 10^7^) were incubated with 40 μl Macs α-human CD3 microBeads (Miltenyi Bioteck, Auburn CA) following the manufacturer\'s directions. The cells were separated by passage through a Macs Separation Column (Miltenyi Bioteck) consisting of a magnetized iron matrix. CD3 positive and negative cells were counted, re-mixed together and incubated overnight. As a control, MNC (2 × 10^7^) were also incubated with 40 μl Macs α-human CD4 microBeads. After treatment, the cells were incubated in 3 ml RPMI 1640 containing 15% AB serum and 15% autologous plasma in one well of a 12 well tissue culture plate (Falcon). In some experiments half of the recombined cells obtained after CD3 column passage were incubated overnight in RPMI containing 10% FCS (Gibco) in the presence of the cytokines GM-CSF 800 U/ml and IL4 1000 U/ml (R & D Systems, Minneapolis, MN). Day 0 baseline cells were immediately removed for immunostaining while Day 1 cells were incubated overnight. Immunophenotyping ----------------- In order to monitor DC differentiation, the cells were stained by two-color immunofluorescence with a panel of antibodies to monocytes, DC and apoptotic cells. Cells (1 × 10^6^) were incubated with 10--20 μl of fluorocrome conjugated monoclonal antibody for 30 minutes in the dark at 4°C. The antibodies were directly conjugated to fluorescein (FITC) or phycoerythrin (PE) and included: CD14-FITC (monocytes) + CD86-PE (co-stimulatory molecule highly expressed on DC); HLA-DR-FITC (anti-class II MHC molecule) and CD83-PE (DC maturation marker); and their isotype matched controls (Beckman Coulter Immuno-Tech, Hialeah, FL). Cells were washed once and suspended in PBS and read on a XL flow cytometer (Beckman Coulter) within 24 hours. Combined membrane and cytoplasmic staining was performed following manufacturers instructions (Intraprep kit, Beckman Coulter). Antibody combinations included: membrane CD36-FITC (receptor for apoptotic cells) + cytoplasmic CD83 PE; DR-FITC + cytoplasmic CD83-PE; and isotype controls (Beckman Coulter). To detect apoptotic cells, lymphocytes were stained with: membrane HLA-DR-FITC (class II MHC) + cytoplasmic Apo2.7-PE (apoptotic cells); and isotype controls. Data was analyzed using the CXP software (Beckman Coulter). Confocal Microscopy ------------------- Cells were double-stained for membrane HLA-DR-FITC + cytoplasmic Apo2.7-PE following the manufacturer\'s instructions for combined membrane and cytoplasmic staining (see immunophenotyping). In addition, cells were double stained for cytoplasmic LAMP-2 FITC (lysosomal marker, Research Diagnostics) and HLA-DR-PE. Cells were prepared for microscopy following the instructions for Molecular Probes \"Slow Fade Light\" anti-fade kit (Molecular Probes Inc, Eugene, OR). Specimens were kept in the dark at 4° until microscopy was performed on a Zeiss confocal microscope. Mixed leukocyte culture assay ----------------------------- The mixed leukocyte culture assay was performed by isolating control leukocytes from two normal donors. Control T cells were purified with CD4 magnetic beads and the column effluent containing monocytes and B cells was γ-irradiated to prevent differentiation and used as a source of stimulators. Transitioning DC from CTCL patients were obtained one day prior to the normal control cells and cultured overnight without cytokines, γ-irradiated and used as stimulators for the control lymphocytes. The cells were adjusted to 4 × 10^6^/ml and 50 μl of responding cells and 50 μl of stimulating cells co-cultured in round bottom microtiter wells with the addition of 100 μl of RPMI 1640 containing 15% AB serum and 15% autologous plasma for 6 days at 37°C under a 5% CO~2~atmosphere. The wells were pulsed with 1 μCi/well ^3^\[H\]-thymidine 16 hours prior to harvest (PhD harvester, Cambridge Tech., Cambridge, MA). The incorporation of the isotope was evaluated in a liquid scintillation counter. CD8 T cell purification and expansion ------------------------------------- CD8 T cells were purified with CD8-magnetic beads (≥96% purity) and suspended in RPMI 1640/15% autologous serum and IL2 and added to DC that had been column eluted from the same CTCL patient. The cells were co-cultured overnight with 1.1 × 10^6^CD8 T cells/well added to CD3-bead rendered apoptotic CTCL cells or viable CTCL cells (4 × 10^6^/well). After overnight culture, the cells were harvested, counted, and immunophenotyped for markers of T cells (CD3, CD4, CD8) and apoptosis (Apo2.7). Tumor necrosis-α(TNF-α) ELISA ----------------------------- The production of TNF-α was measured in an ELISA assay (R&D Systems, Minneapolis, MN) essentially as described by the manufacturer. Statistical evaluation ---------------------- The expression of DC markers and the MLC response was evaluated statistically by the student\'s t test or if the data was not normally distributed the Mann-Whitney Rank Sum Test using the Sigma Stat analysis program. Results ======= Passage of monocytes through a separation column induces monocyte to DC transition ---------------------------------------------------------------------------------- Monocytes were obtained from a leukapheresis harvest performed therapeutically on CTCL patients and were cultured overnight with and without passage through a magnetic bead separation column. Monocyte differentiation into semi-mature DC was monitored by 2-color immunofluorescence. In a representative experiment, (Fig. [1](#F1){ref-type="fig"}, gated on the monocyte population as identified by co-expression of CD14 and CD86), the loss of monocyte membrane marker CD14 is revealed by a decrease in the mean fluorescence intensity (MFI) of the CD14 fluorochrome. CD14 expression declined as the degree of manipulation of the cells increased from primary isolation (Fig. [1a](#F1){ref-type="fig"}) to simple overnight culture of the leukapheresis product (Fig. [1b](#F1){ref-type="fig"}), compared to the addition to the differentiating DC of CTCL cells that were selected by the CD4 antibody, (Fig [1c](#F1){ref-type="fig"}) to the maximum reduction in monocyte CD14 expression found when the activated transitioning DC were cultured with CTCL cells rendered apoptotic by CD3 antibody (Fig. [1d](#F1){ref-type="fig"}). In total, (Fig [1e](#F1){ref-type="fig"}) the expression of CD14 was reduced by 54%, from a mean fluorescent intensity (MFI) of 13 on primary isolation to 5.89, when the column separated monocytes were co-cultured with the CD3-treated apoptotic CTCL cells. As the differentiating DC lost the monocyte marker, a 3-fold increase in expression of CD86, a co-stimulatory molecule, was found ranging from an MFI of 1.98 on Day 0 to 6.4 after passage through the CD3-magnetic bead column and overnight incubation (Fig. [1f](#F1){ref-type="fig"}). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### DC differentiation from monocytes induced by column activation. CTCL cells and DC were isolated from a leukapheresis by CD4 or CD3-antibody conjugated to magnetic beads. The cells were separated by passage through a column placed in a magnetic field and the purified CTCL cells were re-added to the column activated monocytes and cultured overnight. Binding of fluorochromes was analyzed using flow cytometry and 2-color quadstats were gated on the monocyte population. The results demonstrate membrane CD14-FITC and CD86-PE co-expression on cells obtained a: Day 0, primary isolation; and after overnight culture of b: leukapheresis cells; c: cells obtained by CD4-magnetic bead isolation and re-cultured overnight with column activated monocytes; and d: cells obtained from CD3-magnetic bead isolation and re-cultured overnight with column activated monocytes. e:Bar graph showing the reduction in mean fluorescent intensity (MFI) of CD14 expression on primary isolation (Day 0) and after overnight incubation of the leukapheresis (leuk) or column passaged and recombined cell populations using CD4 or CD3-magnetic bead isolation (negative control isotype staining is presented in the first bar). f: Bar graph showing the increase in MFI of CD86 expression (as described in e). ::: ![](1476-8518-3-4-1) ::: Transitioning DC increase their expression of the maturation marker, CD83 ------------------------------------------------------------------------- In Fig. [2A](#F2){ref-type="fig"} &[2B](#F2){ref-type="fig"}, the differentiation of monocytes into semi-mature DC is demonstrated by an increase in the percentage of cells that exhibit reduced fluorescent intensity of membrane CD36 (receptor for up-take of apoptotic cells, a marker that is lost as DC mature) and increased expression of cytoplasmic CD83 (DC maturation marker). Fig. [2A-a](#F2){ref-type="fig"}, demonstrates that only 4% of the cells co-express membrane CD36 and cytoplasmic CD83 on primary isolation. When the cells were cultured overnight, the percentage of cells co-expressing CD36/CD83 increased as the level of manipulation rose from 25% in the overnight culture of the leukapheresis (Fig. [2A-b](#F2){ref-type="fig"}) and in cells separated with a CD4-magnetic bead control antibody and re-added to the column effluent (Fig. [2A-c](#F2){ref-type="fig"}) to the maximal differentiation of 34% found when apoptotic CD3-treated CTCL cells were re-added to the activated transitioning DC (Fig. [2A-d](#F2){ref-type="fig"}). In Fig [2A-e](#F2){ref-type="fig"}, the reduction in CD36 MFI is shown by a decline from a MFI of 34 on primary isolation to 7.7 (77% reduction) in the monocyte/DC population activated by passage through the separation column and recombined for overnight culture in the presence of CTCL cells rendered apoptotic with CD3 antibody. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### DC maturation induced after column separation and overnight incubation. Fig. 2A: CTCL cells and monocyte/DC isolated as described in Figure 1 were fixed and permeabilized and stained with CD36-FITC (membrane) and CD83-PE (cytoplasm). The results show 2-color quadstats gated on the monocyte population of cells obtained from a: Day 0, primary isolation; after overnight culture of b: leukapheresis cells; c: CD4-magnetic bead isolation and re-addition to column activated monocytes; d: CD3-magnetic bead purification and re-addition to column activated monocytes; e: Bar graph of the MFI of membrane CD36 expression on the cell populations. Fig. 2B: Demonstration of cytoplasmic CD83 expression in the monocyte/DC population gated by side-scatter (SS) on 100% of the monocyte population. Cell treatment a--das described for Fig 2A. ::: ![](1476-8518-3-4-2) ::: The increase in cytoplasmic CD83 expression is shown in Fig. [2B](#F2){ref-type="fig"}. As expected only a small percentage of cells express the DC differentiation marker, CD83 on primary isolation (0.5%, Fig. [2B-a](#F2){ref-type="fig"}). Overnight incubation of the leukapheresis (Fig [2B-b](#F2){ref-type="fig"}) increases CD83 expression to an equivalent degree as CD83 expression detected after passage through a CD4-magnetic bead column (Fig. [2B-c](#F2){ref-type="fig"}). More than one third of the monocytes transitioned into semi-mature DC as shown by the increased expression of cytoplasmic CD83 (Fig. [2B-d](#F2){ref-type="fig"}) found when CD3-separated apoptotic CTCL cells were added to the column activated monocytes. Induction of simultaneous DC differentiation and CTCL cell apoptosis and engulfment ----------------------------------------------------------------------------------- Further confirmation of enhanced differentiation of monocytes to DC was found when membrane class II expression (HLA-DR) was measured and the up-take of apoptotic CTCL cells was assessed. In figure [3A](#F3){ref-type="fig"}, the percentage of DR-positive transitioning monocytes containing apoptotic cells was determined by measurement of the cytoplasmic expression of the early apoptotic marker APO2-PE. On primary isolation (Fig. [3A-a](#F3){ref-type="fig"}), or after overnight incubation of the leukapheresis without further processing (Fig. [3A-b](#F3){ref-type="fig"}), only a small percentage of the monocyte-DC population contained apoptotic material in the cytoplasm. CD4-treatment and column passage damaged enough cells to increase the number of apoptotic CTCL cells ingested by the activated monocyte-DC population (Fig. [3A-c](#F3){ref-type="fig"}). As previously reported \[[@B2]\], CD3-binding to CTCL cells rendered the malignant T cells apoptotic and material from the damaged and dying CTCL cells could be detected inside the developing DC population (Fig. [3A-d](#F3){ref-type="fig"}). While only 19% of the transitioning DC were reactive with DR/APO2-PE, this probably represents only a minimal level of engulfed apoptotic cells since processing and degradation of the apoptotic blebs during overnight incubation could have reduced the detectable expression of APO2-PE positive material. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Increased class II expression on semi-mature DC after ingestion of apoptotic CTCL cells. Fig. 3A: CTCL cells and DC prepared as described in Figure 1 were fixed and permeabilized and stained with DR-FITC (anti-class II MHC antibody, membrane) and APO2-PE (cytoplasm). The results present 2-color quadstats gated on the monocyte population of cells obtained from a: Day 0, primary isolation; b: leukapheresis cells; c: CD4-magnetic bead isolation and re-additon to column activated monocytes; d: CD3-magnetic bead purification and re-addition to column activated monocytes. Fig. 3B: Membrane DR staining on the monocyte/DC population gated on the total monocyte population by SS. Cell treatment a--das described for Fig 3A. ::: ![](1476-8518-3-4-3) ::: Differentiation of the DC population was also demonstrated by the increase in expression of membrane class II MHC molecules. Physical manipulation did not increase class II expression from the primary value obtained on initial isolation (Fig. [3B-a](#F3){ref-type="fig"}), when leukapheresis cells were cultured overnight (Fig. [3B-b](#F3){ref-type="fig"}). No enhancement of class II expression was noted even when the column activated monocytes were co-cultured overnight with CD4-bead separated CTCL cells (Fig. [3B-c](#F3){ref-type="fig"}). However, the overnight addition of apoptotic CTCL cells, obtained after CD3-binding, to transitioning DC increased class II expression from 55% (Day 0, Fig. [3B-a](#F3){ref-type="fig"}) to 72% (Fig. [3B-d](#F3){ref-type="fig"}). Statistical evaluation of the enhanced expression of DC differentiation markers ------------------------------------------------------------------------------- We evaluated the overall increase in markers of DC differentiation from monocytes in leukocytes obtained from seven CTCL patients. While substantial variation in the expression of several antigens precluded analysis, the results showed that overall expression of class II MHC antigen was significantly up-regulated in differentiating DC obtained after column passage with (P ≤ 0.005) and without (P ≤ 0.002) the addition of apoptotic CTCL cells (Fig. [4a](#F4){ref-type="fig"}). In addition, CD86 (P ≤ 0.025) expression was significantly increased when CTCL cells were co-cultured with column passaged transitioning DC loaded with apoptotic CTCL cells and CD83 (P ≤ 0.001) was enhanced irrespective of the presence of apoptotic CTCL cells (Fig. [4b](#F4){ref-type="fig"} &[4c](#F4){ref-type="fig"}). These results confirm that the physical perturbation encountered after passage through the small spaces of separation column significantly enhances the entry of monocytes into the DC pathway. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Statistical analysis of DC differentiation markers. The expression of markers of DC differentiation were compiled from the overnight culture of DC induced by column passage with and without apoptotic cell loading that had been obtained from 7 CTCL patients, averaged and analyzed for significance in comparison to the values obtained on primary isolation. a:Mean fluorescence intensity (MFI) of class II expression on Day 0, primary isolation (Pre Tx; pre-treatment), or Day 1 column activated cells loaded with apoptotic CTCL or co-cultivated in the presence of viable CTCL cells (Mann-Whitney Rank Sum Test). b:Percent of monocytes expressing CD86 on primary isolation, or after column activation and overnight culture with and without apoptotic cell ingestion (t test). c:Percent of monocytes expressing cytoplasmic CD83 on primary isolation or after column activation and overnight cultivation with and without apoptotic cell up-take (Mann-Whitney Rank Sum Test). ::: ![](1476-8518-3-4-4) ::: Demonstration of DC loading with apoptotic cells by confocal microscopy ----------------------------------------------------------------------- In Fig. [5A](#F5){ref-type="fig"}, CTCL cells were rendered apoptotic with CD3-magnetic bead conjugated antibody (Fig. [5A a--c](#F5){ref-type="fig"} ) or as a control treated with CD4-magnetic bead conjugated antibody (Fig. [5A d--f](#F5){ref-type="fig"}), run through the separation column and co-cultured with the simultaneously activated differentiating DC. The activated monocyte/DC population was double-stained for expression of membrane class II (green) and the marker of early apoptotic cells, intracellular APO-2 (red). Representative class II-positive cells (green fluorescence) are seen in Figures [5A-a](#F5){ref-type="fig"} and [5A-c](#F5){ref-type="fig"}. In Figure [5A-b](#F5){ref-type="fig"}, three cells that were rendered apoptotic after CD3-binding, were identified (white arrows) and material from one of these cells is contained in a class II positive cell (merge, Fig. [5A-c](#F5){ref-type="fig"}). In Figure [5A-e](#F5){ref-type="fig"} (CD4-treatment), only a small amount of apoptotic material is found and none of this material is associated with the class II positive cell (Fig. [5A-f](#F5){ref-type="fig"}, merge). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Confocal microscopic demonstration of apoptotic cell ingestion and class II localization in lysosomal compartments in differentiating DC. Fig. 5A:Cell populations prepared as described in Figure 1 were evaluated by confocal microscopy after fixation and permeabilization and staining. A representative activated monocyte/DC is shown after CD3 column passage and recombination with the apoptotic CTCL cells as detected by a: membrane class II-FITC (green); b: cytopolasmic APO2-PE (red, white arrows) and c: merged image demonstrating internalization of apoptotic material in a class II positive cell. A representative activated monocyte/DC is shown after CD4 column passage and recombination with viable CTCL cells as detected by d: membrane class II-FITC (green); e: cytopolasmic APO2-PE (red, white arrow) and f: merged image demonstrating absence of internalization of apoptotic material in a class II positive cell. Fig. 5B: Cells prepared as described in Fig. 5A were passed through the CD3 column and stained for a: membrane class II-PE (red); b: lysosomal membrane marker, LAMP (green); and c: merged image showing co-localization of class II molecules in lysosomal compartments. Cells obtained after passage through the CD4 column were stained for d: membrane class II-PE (red); e: lysosomal membrane marker, LAMP (green); and f: merged image showing an absence of co-localization of class II molecules in lysosomal compartments. ::: ![](1476-8518-3-4-5) ::: To confirm that class II molecules co-localized in lysosomal compartments in a pattern found in semi-mature DC \[[@B8]\], cells were stained with a lysosomal marker LAMP2 and an antibody to class II MHC molecules (Fig. [5B](#F5){ref-type="fig"}). In Fig. [5B-a](#F5){ref-type="fig"}, a cell that has been activated by passage through the separation column and co-cultivated overnight with CTCL cells rendered apoptotic by CD3-magnetic bead binding was stained with an anti-class II antibody (red). In Fig. [4B-b](#F4){ref-type="fig"} lysosomal compartments were visualized with an antibody that binds to the lysosomal membrane (LAMP2, green). Merging of the 2 fluorochromes (Fig. [5B-c](#F5){ref-type="fig"}, yellow) demonstrates colocalization of class II MHC molecules in lysosomal compartments. When class II staining was monitored on column activated transitional cells that had been co-incubated with control CTCL cells selected by CD4-magnetic bead separation (Fig. [5B-d](#F5){ref-type="fig"}, red), strong membrane staining was found. Weak lysosomal staining was localized beneath the plasma membrane (Fig [5B-e](#F5){ref-type="fig"}, green). When the pictures were merged, class II MHC molecules did not exhibit entry into the lysosomal compartment (Fig. [5B-f](#F5){ref-type="fig"}). The presence of class II MHC molecules in lysosomes is consistent with differentiation into semi-mature DC \[[@B8]\], and suggests that class II molecules have migrated to lysosomal compartments where they would have the opportunity for loading with peptides derived from processed apoptotic material. The addition of supportive cytokines enhances monocyte to DC differentiation ---------------------------------------------------------------------------- We sought to maximize induction of maturing DC loaded with apoptotic malignant T cells through the addition of exogenous cytokines known to be important for DC differentiation \[[@B9]\]. To study the effect of supportive cytokines on the phenotype of the developing DC, we divided the column separated cells in half and co-incubated them overnight with CD3-bead rendered apoptotic CTCL cells with and without GM-CSF and IL-4. The addition of cytokines to the co-cultured apoptotic CTCL cells and column activated transitioning monocytes increased the overall maturation of the DC. In Figure [6](#F6){ref-type="fig"}, the level of CD14 expression is reduced as shown by an increase in the CD14 negative population (Gate AA1) from 4.8% at baseline (Fig. [6a](#F6){ref-type="fig"}) to 10% when the transitioning DC were incubated with apoptotic cells without cytokines (Fig. [6b](#F6){ref-type="fig"}). The addition of cytokines enhanced the loss of CD14 expression resulting in 36% of the cells becoming CD14-negative after overnight culture (Fig. [6c](#F6){ref-type="fig"}). As the differentiating monocytes lost CD14 expression, a concomitant increase in CD86 expression was noted. CD86 expression rose from a baseline level of 61% (Fig. [6d](#F6){ref-type="fig"}) to more than 80% CD86-positive transitioning DC after column separation and co-cultivation with CD3-rendered apoptotic cells without cytokines (Fig. [6e](#F6){ref-type="fig"}) or in the presence of exogenous cytokines (Fig. [6f](#F6){ref-type="fig"}). ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Exogenous cytokines enhance DC differentiation from monocytes activated by column passage. Monocyte/DC populations isolated as described in Figure 1 were stained for membrane co-expression of CD14-FITC and CD86-PE. The results present 2-color quadstats gated on the monocyte population of cells obtained from a: Day 0, primary isolation; b: CD3-magnetic bead purification and re-addition to column activated monocytes; c: the same CD3 column purified and activated recombined cell population cultured with the cytokines GM-CSF and IL4. Demonstration of membrane CD86 expression on the monocyte/DC population gated by side-scatter on 100% of the monocyte population. d: Day 0, primary isolation; e: CD3-magnetic bead purification and re-addition to column activated monocytes; f: the same CD3 column purified and activated recombined cell population cultured with cytokines. ::: ![](1476-8518-3-4-6) ::: Cytokines enhance DC maturation ------------------------------- The percentage of semi-mature DC differentiated after overnight co-culture that co-expressed membrane CD36 and intracytoplasmic CD83 was enhanced by the addition of cytokines. In Fig. [7a](#F7){ref-type="fig"}, on primary isolation the monocytes expressed intermediate levels of CD36 and did not contain cytoplamic CD83 (Fig. [7a](#F7){ref-type="fig"}). Co-expression of CD36/CD83 (Fig. [7b](#F7){ref-type="fig"}) rose to 50%, after overnight culture in the absence of cytokines, on differentiating DC that had passed through the separation column and were recombined with CD3 rendered apoptotic CTCL cells. This increased expression of a receptor for apoptotic cells may have been driven by the presence of very high levels of apoptotic material in the co-cultures (Fig. [8](#F8){ref-type="fig"}). Further maturation was observed in the presence of cytokines (Fig. [7c](#F7){ref-type="fig"}) leading to 53% CD36 expression on the transitioning DC and the identification of 7% CD36-negative cells that contained CD83 in the cytoplasm. The percentage of differentiating DC that expressed cytoplasmic CD83 rose from 0% at baseline (Fig. [7d](#F7){ref-type="fig"}) to 49% after column separation and co-incubation with CTCL cells rendered apoptotic by CD3-magnetic bead binding (Fig. [7e](#F7){ref-type="fig"}) to 59% when cytokines were added to the cultured cells (Fig. [7f](#F7){ref-type="fig"}). ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Exogenous cytokines increase DC maturation induced after column separation and overnight incubation. Monocyte/DC populations isolated as described in Figure 1 were fixed and permeabilized and stained for expression of membrane CD36-FITC and cytoplasmic CD83-PE. The results present 2-color quadstats gated on the monocyte population of cells obtained from a: Day 0, primary isolation; b: CD3-magnetic bead purification and re-addition to column activated monocytes; c: the same CD3 column purified and activated recombined cell population cultured with the cytokines GM-CSF and IL4. Demonstration of cytoplasmic CD83 expression in the monocyte/DC population gated by side-scatter on 100% of the monocyte poulation. d: Day 0, primary isolation; e: CD3-magnetic bead purification and re-addition to column activated monocytes; f: the same CD3 column purified and activated recombined cell population cultured with cytokines. ::: ![](1476-8518-3-4-7) ::: ::: {#F8 .fig} Figure 8 ::: {.caption} ###### Exogenous cytokines enhance ingestion of apoptotic material in differentiating DC. CTCL cells and monocyte/DC populations isolated as described in Figure 1. The cells were fixed and permeabilized and stained for expression of membrane DR-FITC and cytoplasmic APO2-PE. The results present 2-color quadstats gated on the monocyte population of cells obtained from a: Day 0, primary isolation; b: CD3-magnetic bead purification and re-addition to column activated monocytes; c: the same CD3 column purified and activated recombined cell population cultured with the cytokines GM-CSF and IL4. ::: ![](1476-8518-3-4-8) ::: Class II expression and up-take of apoptotic material is enhanced in the presence of cytokines ---------------------------------------------------------------------------------------------- The baseline expression of class II MHC molecules on the cell membrane of monocytes on primary isolation is shown in Fig. [8a](#F8){ref-type="fig"}. Freshly isolated monocytes express a reduced intensity of class II expression and contain a small percentage of cytoplasmic apoptotic material. After column separation and co-incubation with CD3-magnetic bead treated apoptotic cells, membrane class II expression is enhanced (Fig. [8b](#F8){ref-type="fig"}) and large amounts of apoptotic material can be detected in the cytoplasm of the transitioning DC. Exogenous cytokines further increase the percentage of class II-positive cells that contain apoptotic material (Fig. [8c](#F8){ref-type="fig"}). Therefore, the addition of exogenous cytokines enhances both the differentiation of immature DC and the ingestion of apoptotic material improving the overnight yield of maturing apoptotic T cell loaded DC. Functional analysis of the differentiating DC obtained after column passage --------------------------------------------------------------------------- Transitioning DC obtained after column passage were evaluated for their stimulatory capacity in MLC (Fig. [9](#F9){ref-type="fig"}). The results demonstrate that DC induced by column passage of leukocytes from two normal controls were significantly better stimulators (P ≤ 0.034 & P ≤ 0.036) in MLC than autologous monocytes irrespective of apoptotic cell loading. These results confirm that column activation of monocytes and overnight culture enhances the membrane expression of class II MHC molecules recognized by alloresponsive CD4 T cells. Therefore, DC harvested after physical activation and overnight culture could expand CD4 T cells and potentially provide the help required for licensing of anti-tumor CD8 T cell responses \[[@B10]\]. ::: {#F9 .fig} Figure 9 ::: {.caption} ###### Mixed leukocyte culture response of normal T cells to monocytes or column activated loaded and non-loaded DC. Normal CD4 T cells were magnetic bead column purified from the peripheral blood of 2 controls and stimulated with column effluent monocytes that had been γ-irradiated to prevent differentiation, or γ-irradiated CTCL cell monocytes that had been column purified and either loaded or not with apoptotic malignant T cells and cultured overnight one day prior to the normal T cell isolation. The results are presented as delta CPM (less background proliferation obtained by autostimulation) of ^3^\[H\]-thymidine incorporation measured at day 6. Significance was evaluated with a student\'s t test. ::: ![](1476-8518-3-4-9) ::: We have begun to investigate the capacity of the DC harvested after column perturbation and apoptotic malignant T cell loading to induce and expand an anti-tumor CD8 T cell response. In these initial studies (Fig. [10A](#F10){ref-type="fig"}), we have found that the percentage of CD8 T cells (purified CD8 T cells ≥96% positive, obtained from the leukapheresis of a patient responsive to ECP) increased by 38% in the presence of DC fed apoptotic CTCL cells (Fig. [10A-a](#F10){ref-type="fig"}) compared to the percent of CD8 T cells found after overnight incubation with DC exposed to viable CTCL cells (Fig. [10A-b](#F10){ref-type="fig"}). In addition, the absolute number of CD8 T cells recovered from the overnight culture of differentiating column passaged DC loaded with apoptotic malignant T cells increased by 22% when compared to the initial number of CD8 T cells while the absolute number of CD8 T cells present in cultures of DC and viable CD4 T cells fell by 12% (Fig. [10A-c](#F10){ref-type="fig"}). Therefore, exposure of CD8 T cells to malignant T cell loaded DC increases both the percentage and absolute number of potential anti-tumor responsive T cells. ::: {#F10 .fig} Figure 10 ::: {.caption} ###### CD8 T cell response to column-activated malignant T cell loaded DC. Fig. 10A:CD8 T cells were magnetic-bead enriched (≥96% CD8^+^T cells) from the leukapheresis of a CTCL patient and added to column-activated DC with and without apoptotic malignant T cell loading. The percentage of CD8 T cells was identified by immunophenotyping and flow cytometry and the results presented as 1-color histograms. a: Percentage of CD8 T cells found after overnight culture with column-activated DC pulsed with CD3-magnetic bead rendered apoptotic CTCL cells. b: Percentage of CD8 T cells identified after overnight culture with DC co-cultivated with viable CTCL cells. c: Absolute number of CD8 T cells after overnight cultivation with DC loaded with apopototic malignant T cells or DC co-incubated with viable CD4-bead isolated CTCL cells. Horizontal line indicates the initial number of CD8 T cells added to the co-cultures on Day 0. Fig. 10B: The percentage of apoptotic cells was determined in the co-cultures by staining for APO2.7 and flow cytometry. The quadstats are gated on the lymphocyte population by side scatter (SS) and represent 100% of the lymphocyte population. Percent apoptotic cells found in co-cultures of a:CD8 T cells and DC loaded with apoptotic malignant T cells; b:CD8 T cells, DC and viable CTCL cells; c:DC loaded with apoptotic CTCL in the absence of CD8 T cells; d:DC cultured with viable CTCL without CD8 T cells. e:Culture supernatants were obtained from CD8 T cells cultured overnight alone, or in the presence of DC loaded with apoptotic CTCL cells or DC cultured with viable CTCL cells and the secretion of TNF-α determined in an ELISA assay. The results are presented as pg/ml and significance determined with a student\'s t test. ::: ![](1476-8518-3-4-10) ::: The level of apoptosis found when CD8 T cells were cultured overnight with column activated-DC loaded with CTCL cells doubled (56%, Fig. [10B-a](#F10){ref-type="fig"}) in comparison to the level of apoptosis present when CD8 T cells were added to non-loaded DC that had been cultivated with viable CTCL cells (27%, Fig. [10B-b](#F10){ref-type="fig"}). The baseline level of apoptosis was 24% when malignant T cell loaded DC (Fig. [10B-c](#F10){ref-type="fig"}), or non-loaded DC (Fig. [10B-d](#F10){ref-type="fig"}) were cultured in the absence of CD8 T cells. These results indicate that residual CTCL cells may be lysed in the presence of CD8 T cells stimulated with DC that have ingested apoptotic malignant T cells. Finally, further support for the contention that functional CD8 T cells were expanded by overnight exposure to column-activated DC loaded with malignant T cells was obtained by evaluation of the levels of TNF-α found in the culture supernatants. In Figure [10B-e](#F10){ref-type="fig"}, supernatants from CD8 T cells cultured overnight alone contained minimal levels of TNF-α. CD8 T cells stimulated with column differentiated DC loaded with CD3-bead rendered apoptotic malignant T cells or not loaded both significantly (P ≤ 0.001 & P ≤ 0.014) stimulated release of TNF-α. However, DC that had engulfed apoptotic cells caused the release of three fold more TNF-α than non-loaded DC, indicating that CD8 T cell activation had occurred and the release of a molecule that promotes tumor cytolysis was present. Discussion ========== Development of effective DC based cancer vaccine technology has been limited by the extensive manipulation and extended period of in vitroculture required for generation of mature DC loaded with the appropriate tumor antigens. We have circumvented some of these limitations through modification of a successful technology that permits both DC differentiation from peripheral monocytes and simultaneous loading of DC with apoptotic malignant T cells containing the full complement of potential tumor antigens \[[@B6]\]. DC are the most potent APC displaying when mature high levels of co-stimulatory, adhesion and MHC molecules which can present peptides derived from apoptotic cells to the immune system \[[@B9]\]. Therefore, the development of a simple rapid means of generating malignant cell-loaded DC could advance the immunotherapy of CTCL and perhaps other malignancies. Immunotherapy has played a major role in the treatment of CTCL since the introduction of ECP by Edelson and colleagues in 1987 \[[@B5]\]. The mechanism underlying the success of ECP treatment was defined by the demonstration that the simultaneous introduction of apoptotic malignant T cells and the differentiation of monocytes into DC resulted in patients receiving CTCL cell-loaded DC that have the capacity to present antigen, derived from the CTCL cells, to cytotoxic lymphocytes and initiate an immune response towards the malignant CD4 T cells. Previous studies had demonstrated that despite the clonal expansion of CD4^+^malignant T cells in the peripheral blood of CTCL patients, circulating populations of CD8 T cells that retained the capacity to lyse autologous malignant T cells \[[@B11]\] could be identified. One antigen that served as an immunogen recognized by cytotoxic T cells in CTCL was determined to be peptides derived from the beta chain of the TCR that was clonotypically displayed on the malignant T cells \[[@B12],[@B13]\]. Therefore, the potential for development of an anti-malignant T cell immune response exists in CTCL patients and immunotherapeutic approaches designed to expand anti-tumor CD8 T cells could be effective in this disease. We sought to exploit our understanding of the mechanism of ECP to develop more efficient, rapid, clinically practical means for producing malignant T cell-loaded DC. In the current study, we demonstrate that DC loaded with apoptotic cells can be produced in one day without extensive manipulation or the use of exogenous cytokines. The use of CD3-antibody to render CTCL cells apoptotic and passage of the treated MNC through the small pores of the iron matrix of a separation column followed by overnight co-incubation resulted in the generation of DC containing material derived from apoptotic CTCL cells. DC differentiation was demonstrated by both the reduction in monocyte markers and the significant increase in class II MHC molecules and co-stimulatory molecules, as well as the increase in CD83, a marker of maturing DC. The internalization of apoptotic blebs was confirmed by localization of the apoptotic material in the cytoplasm, indicating that processing of the apoptotic CTCL-derived material could make peptides available for MHC loading and transport to the cell membrane \[[@B14]\]. The ability to increase the number of maturing CTCL cell-loaded DC by the addition of exogenous cytokines demonstrates that this technique can produce cell populations that can be manipulated to maximize the production of DC that contain apoptotic material thereby providing access to a spectrum of CTCL cell-derived epitopes, without the requirement for identification or isolation of individual peptides that may be relevant for induction of an anti-CTCL cell immune response. Furthermore, we show that DC produced in this fashion are effective stimulators of alloproliferation in MLC confirming the significant up-regulation of class II MHC molecules. The malignant T cell loaded DC stimulated CD8 T cell expansion and an increase in apoptotic cell death and the significantly enhanced release of TNF-α. These results indicate that CD8 T cells that have been stimulated by malignant T cell loaded DC, produced by this methodology, may develop the ability to mediate tumor cell cytolysis. The current studies support our previous results demonstrating that monocyte differentiation into DC could be driven by increasing levels of physical perturbation \[[@B6]\]. We confirm that leukapheresis alone generates modest monocyte activation and conversion into immature DC that can be enhanced by further manipulation and the addition of apoptotic cells. We also demonstrate that CD3-binding is a potent means of rendering CTCL cells apoptotic \[[@B2]\] even when the CTCL cells are not cultured but directly isolated from the patients. The current study combines and extends these two previous observations into a format for simple, rapid, clinically practical DC vaccine generation. Current approaches to DC vaccine technology include peptide pulsing \[[@B15]\], one week or longer of culture with cytokines \[[@B16]\], cell fusion with tumor cell partners \[[@B17]\], and the use of a variety of vectors designed to introduce tumor antigens into the DC \[[@B18]\]. These methods are generally cumbersome, require extensive in vitromanipulation, and are limited to a small set of known tumor epitopes that may be lost from the patient\'s tumor, due to immuoselective pressures. Clinical results with these techniques have been variable and seldom provide long-term responses \[[@B19]\]. In contradistinction, treatment of CTCL patients with ECP has demonstrated an excellent safety profile and in multiple studies in the literature an overall response rate for all stages of the disease of 55.7% and a complete response rate of 17.6% \[[@B20]\]. Pilot studies using transimmunization to enhance the interaction of apoptotic tumor cells and differentiating DC through simple overnight incubation has demonstrated encouraging results in some patients \[[@B4]\], that suggest that the therapy retains the safety profile of ECP but may be more potent and effective in a shorter time course. The technology proposed in this study is likely to be as safe as transimmunization and ECP since it retains the same features of limited cellular manipulation and culture. The replacement of 8-MOP with CD3 antibody should not lead to significant apoptotic cell death and potential tumor lysis syndrome since CD3-binding renders only 30% of the CTCL cell population apoptotic \[[@B2]\]. Since the CD3 antibody is conjugated to the magnetic beads any free antibody could be removed by a second passage through the magnet prior to re-infusion, thereby, limiting the induction of anti-CD3 antibodies. However, presentation of portions of the CD3 antibody after DC ingestion may provoke an immune response that could prevent further therapy. These potential safety issues will require careful monitoring in future clinical trials. The current results demonstrate that further development of this technology through passage over a column that permits the one-step apoptotic cell death of CTCL cells, sparing of normal cells and activation of monocytes into the DC pathway may further improve the immunogenicity of the reinfusate. Since only proliferating tumor cells are rendered apoptotic by the CD3 antibody, normal resting lymphocytes will not be impacted which is in contrast to the use of 8-MOP/UVA that targets the DNA of all nucleated cells. This preservation of normal T cells may serve to improve the induction of anti-CTCL immune responses to the re-infused apoptotic cell-loaded DC by preventing damage to by-stander normal cells and precluding their uptake that could lead to tolerance induction \[[@B21]\]. Using a peristaltic pump it should be possible to rapidly flow a leukapheresis product through a magnetic separation column. Due to the concentration of MNC obtained with the leukapheresis procedure, high yields of monocytes approaching 10^8^cells could be obtained and activated by this procedure \[[@B6]\]. Since CTCL patients have large populations of circulating malignant T cells (approaching \>90% of the lymphocyte population), CD3-treatment would provide substantial apoptotic tumor cells for DC loading. Because both activated monocytes and apoptotic malignant T cells are obtained individually and can be re-added after treatment, the optimal conditions for apoptotic T cell and DC co-cultivation can be determined empirically. This access to both cell populations would permit the opportunity for loading DC with other tumor antigens, including solid tumors rendered apoptotic by irradiation or other methods. Other studies have determined that physical separation of DC clusters by simple pipetting \[[@B22]\] or cell transfer \[[@B8],[@B23]\] is among the most potent means of inducing DC maturation. Furthermore, even semi-mature DC are effective at cross-priming peptide \[[@B22]\] derived from exogenous material into the class I pathway for presentation to CD8 T cells. Our simple approach to rapid DC vaccine construction takes advantage of both physical stimulation and production of apoptotic material providing access to a broad spectrum of CTCL antigens for cross-priming into the class I pathway. Further studies to determine the functional ability of the CTCL cell-loaded DC produced by this methodology will be required to confirm the immunogenicity of the proposed vaccine components. We have already demonstrated that DC loaded with apoptotic malignant T cells are potent immunostimulators in mixed leukocyte culture \[[@B6]\], can provoke positive clinical results in treated patients \[[@B4]\] and that responsive patients treated by standard ECP develop increased levels of circulating CD8 T cells \[[@B24]\]. The current results indicate that the development of DC loaded with apoptotic cells for use in immunotherapy can be performed in a rapid, simple, clinically practical manner that provides ready access to the major cell types so that additional strategies to optimize the vaccine components can be implemented and monitored prior to re-infusion. Competing interests =================== Drs Berger, Edelson and Yale University hold patents pertaining to the transimmunization procedure. Authors\' contributions ======================= Maria Salskov-Iverson has performed the majority of the experiments presented in this manuscript and prepared the primary draft of the paper. Drs Berger and Edelson have defined the preliminary observations upon which this manuscript is based and provided intellectual guidance and supervision for the reported work and manuscript. Acknowledgements ================ The authors wish to acknowledge research support from: The Danish Cancer Society, M. S.-I. and the NY Cardiac Association, C.L.B. & R.L.E	PubMed Central	2024-06-05T03:55:59.694749	2005-7-19	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180847/", "journal": "J Immune Based Ther Vaccines. 2005 Jul 19; 3:4", "authors": [ { "first": "Maria", "last": "Salskov-Iversen" }, { "first": "Carole L", "last": "Berger" }, { "first": "Richard L", "last": "Edelson" } ] }
PMC1180848	Background ========== The human major histocompatibility antigens, HLA, are generally cell bound, but trace amounts exist in soluble form \[[@B1]-[@B3]\]. These soluble HLA (sHLA) molecules may have an immunomodulatory function \[[@B4]-[@B6]\]. The known linkage dysequilibrium between class I and class II antigens at the cell surface may have pathophysiological significance \[[@B7]\]. It has been reported that the presence of soluble HLA can be explained, at least in part, by the shedding of cell bound HLA \[[@B8]\]. We have observed no correlation between sHLA-I and sHLA-II levels in the sera of normal individuals \[[@B9]\]. sHLA-I was either non-detectable, or present in very low quantities, in the urine, sweat, saliva and tears of normal individuals. sHLA-I is highly elevated in the saliva of patients with autoimmune rheumatic diseases \[[@B2],[@B10]\]. sHLA-II is routinely detectable in the urine, tears, sweat and saliva of normal individuals, but concentrations of sHLA-II are not observed to be elevated in rheumatological diseases \[[@B10],[@B11]\]. In the neurological realm, there is a possible alteration of sHLA-I and/or sHLA-II levels as a reflection of disease activity in multiple sclerosis (MS). Clinical and brain magnetic resonance imaging (MRI) disease activity in MS is associated with fluctuations in sHLA-I and sHLA-II levels in the serum and cerebrospinal fluid (CSF) of patients with MS \[[@B12]-[@B14]\]. However, the published reports are somewhat in conflict. There has been reported elevation of serum sHLA-II, but not of serum sHLA-I, and an increase in CSF sHLA-I, but not CSF sHLA-II concentrations, in patients with MS \[[@B12],[@B13]\]. However, an elevation of CSF sHLA II and I as well as an increase in serum sHLA-I, but not in serum HLA-II levels, in MS has been reported \[[@B14]\]. Fainardi et al \[[@B15]\] reported a decrease in sHLA-I concentrations during exacerbations in MS, but an increase in CSF sHLA-I was observed in patients with lesional activity by MRI brain scan. The variability in the studies, to date, could possibly be explained by variability in phenotypic expression in genetically susceptible individuals as well as in assay methodology. Recent studies have demonstrated that variations in sHLA concentrations are due, at least in part, to the HLA allospecificities \[[@B16]-[@B18]\]. Racial-ethnic factors may also have an influence on sHLA levels \[[@B18],[@B19]\]. Thus, it appears advantageous to assess sHLA measurements in subjects with a similar racial-ethnic background. Theoretically, we would expect that measurement of sHLA in CSF would be most likely to reflect central nervous system (CNS) disease activity if indeed such measurement could serve as a monitor of a disorder such as MS. However, CSF exams are invasive and not without potential complications. Therefore, we sought to determine whether more readily accessible body fluid, specifically saliva, might provide correlative sHLA measurements in an autoimmune-mediated CNS disease such as MS. Methods ======= We analyzed CSF and saliva from thirteen consecutive Caucasian patients with relapsing-remitting form of MS (RRMS) defined by the McDonald criteria \[[@B20]\]. None of these patients was on immunomodulating therapy for at least six months prior to entrance into the study. We also studied saliva from fifty-three healthy subjects with no history of autoimmune disease for the purpose of comparison. Because there is a high degree of racial variation in the gene frequencies of HLA \[[@B7]\], we limited study participation to Caucasians born in the United States and residing in Louisiana. Saliva samples were collected through expectoration preceded by rinsing of the mouth with sterile water. The resultant salivary samples were collected into test tubes and stored at -20°C until subsequent assay. CSF was collected by standard sterile lumbar puncture technique after the informed consent was reviewed with the patient and signed. Brain MRI was performed using a 1.5 T machine with a standard quadrature head coil. The imaging protocol included sagittal T1-, axial T1-, T2-weighted, and fluid attenuated inversion recovery (FLAIR) images. All MRI scans were performed before and after (Gd-DTPA) infusion. Axial T2-weighted and pre- and post-contrast T1-weighted images were used for assessment of MS plaques. The images were independently interpreted using inspection and computer-assisted techniques by a neuroradiologist. Detection of lesions, compatible with MS, was made by visual inspection as was determination of the absence or the presence of contrast enhancing lesion. Computer based software allowed comparison of the lesions among different groups. Comparisons were made between all 13 MS patients who were subgrouped into either those with and those without enhancing plaques in their brain MRI scans. Hybridoma cell lines W6/32 (anti-HLA-A, B, C), L368 (anti-human B~2~-microglobulin), Ab2.06, L203 and IVA-12 (anti-HLA-DR) were obtained from the American Type Culture^®^collection (Rockville, MD). These lines were expanded and the mAbs were produced in BALB/c mice as described previously \[[@B21]\]. Anti-class I HLA-monoclonal antibody W6/32 detects a common determinate on the a-chain of all HLA class I molecules. Monoclonal antibody L368 detects B~2~-microglobulin, which is a constituent of all HLA class I molecules \[[@B22],[@B23]\]. Anti-class II HLA-monoclonal antibodies Ab2.06, L203 and IVA-12 react with non-competing epitopes in the constant domain of HLA-DR molecules \[[@B24]-[@B26]\]. The solid-phase ELISA for sHLA-I has been previously described \[[@B3],[@B10],[@B16],[@B21]\]. The levels of sHLA-II were determined using a previously described assay \[[@B9],[@B11]\] with minor modifications. Briefly, test samples were added to appropriate wells containing an anti-Class I (W6/32) or anti-Class II (Ab2.06) monoclonal antibody (Mab) coated beads. The reaction proceeded for 30 minutes for sHLA-I and for two hours for sHLA-II at 45°C. The beads were then washed × 3 with distilled water and 200 νl of peroxidase-labeled anti-B~2~M monoclonal antibody (L368) for sHLA-I or L 2.03 Mab for sHLA-II were added to each bead and incubated for an additional hour at 45°C. After additional washes, the color reaction was started by adding the appropriate substrate. Absorbance was measured at 492 nm. Each assay included a standard curve derived from positive and negative controls. Negative controls consisted of 2% BSA and human serum, free of sHLA-I and sHLA-II. Positive control standards were prepared by chromatography of pooled serum over a CL-6B Sepharose Mab W6/32 gel column. The sHLA-I captured by the Mab column was eluted with glycine HCL buffer (0.1 M glycine, pH 2.5). Fractions rich in sHLA-I were neutralized immediately with dibasic sodium phosphate pooled and dialyzed against saline. Total protein was quantitated with the Pierce BCA kit^®^(Rockford, IL), which was assumed to be pure sHLA-I. With each assay, a standard curve was established by including, in duplicate, seven sHLA-I standards (100, 50, 25, 12.5, 6.25, 3.1 and 0 ng/ml) of pure sHLA-I protein. The test values were calculated from the curve described by these standards. All sHLA-I assays were standardized with dilutions of banked standard serum. Measurements were reproducible. For sHLA-II values, the wide range of sHLA-II concentrations reported from studies of serum of normal individuals \[[@B3],[@B9],[@B17],[@B26]-[@B30]\] seem likely to reflect the use of various standards or to the different characteristics of the monoclonal antibodies used in described techniques. Although initial standardization of the sHLA-II assay has been made previously and reported by us \[[@B9]\], in this study, for greater precision of analysis, the amount of sHLA-II was inferred directly from the ELISA absorbance value (OD) within each sample of body fluid tested. The OD of studied samples corresponding to sHLA-II values were compared with the OD values of 5% BSA that had been utilized as dilution factor and negative control within each procedure. Comparisons of mean values for sHLA in study subjects and controls were made with the two-tailed t-test for the means of independent samples. However, this only applied for sHLA-II measurements in our study. P values \< 0.05 were considered significant. Results ======= All normal individuals tested had measurable amounts of sHLA-II in the saliva with a range of 186--362 unit/ml and a mean of 222 ± 18 unit/ml (Table-[1](#T1){ref-type="table"}). In saliva, sHLA-I levels ranged from 0.86 to 100 ng/ml. In five subjects, measurements were below the sensitivity of the assay and thus were non-detectable. These results are in agreement with our previous measurements of sHLA in normal saliva where we found that seven of thirty-seven subjects did not have detectable levels of sHLA-I in this body fluid \[[@B11]\]. However, in this study we raised the question as to whether these individuals represent a population with no sHLA-I. This is apparently not the case, as all saliva samples (n = 13) that were passed over a monoclonal antibody w6/32 column yielded the presence sHLA-I, regardless of detectability by assay. Thus, sHLA-I is present in the saliva in some quantities, however these values are too low to be distinguished from zero in the test system. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Concentrations of sHLA-II in cerebrospinal fluid (CSF) and saliva in multiple sclerosis patient subgroups and controls (unit/ml) ::: CSF Saliva ------------- -------- ---------------- ---------------- --------- ----------- Study group Number Mean ± std dev Mean ± std dev t-value p-value RRMS-total 13 369 ± 16 386 ± 52 .70 .70 RRMS-C(+) 5 356 ± 26 380 ± 51 .44 .67 RRMS-C(-) 8 377 ± 18 390 ± 77 .56 .59 controls 52 \-\-\-- 222 ± 18 8.68 \<.0005\* RRMS = relapsing remitting multiple sclerosis (+) = contrast enhancement by MRI of brain (active plaque formation) (-) = without contrast enhancing plaques by MRI (inactive) \Comparison of saliva sHLA-II in patients (386 ± 52) vs. controls (222 ± 18) ::: For the 13 patients with RRMS, two had a relatively low concentration of sHLA-II (172 unit/ml and 276 unit/ml, respectively) in saliva, while the remaining eleven had relatively high amounts of sHLA-II, ranging from 329 unit/ml to 470 unit/ml with a mean value of 386 ± 52 unit/mL (Table-[1](#T1){ref-type="table"}). This value was highly significant when compared to those of normals (t = 8.68, P \< .0005) (Figure [1](#F1){ref-type="fig"}). Of interest, each patient with RRMS had elevated levels of sHLA-II in the CSF, with a mean of 369 unit/ml, and this was essentially equivalent to the mean sHLA-II concentration in saliva (mean = 386 unit/ml, t = -.70, P = 0.5). In addition, it was noted that CSF and saliva sHLA-II distribution curves were fairly equivalent, except for two outliers (Figure [2](#F2){ref-type="fig"}). sHLA-II concentrations in the CSF and saliva of MS patients were further analyzed by subgrouping them into those with enhancing lesions vs. those without enhancing lesions on brain MRI, with the understanding that contrast enhancement tends to reflect disease activity. Comparison of sHLA-II concentrations in the patients with enhancing lesions (N = 5) to the patients with non-enhancing lesions (N = 8), revealed no significant CSF (356 vs. 377 unit/ml, t = 1.49, P = 0.16) or saliva (380 vs. 390 unit/ml, t = 0.2, P = 0.84) differences. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Demonstration of soluble HLA class II levels in the saliva of multiple sclerosis patients versus controls. The mean ( ± S.D.) values, denoted by \, are 386 ± 52 unit/ml for patients and 222 ± 18 unit/ml for controls (P \< .0005). ::: ![](1742-2094-2-13-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Demonstration of the distribution curves for soluble HLA-II levels in the cerebrospinal fluid (unbroken line) and saliva (dotted line) in multiple sclerosis patients. Despite two outlying values, denoted by \*, there is a fairly equivalent distribution with a mean sHLA-II concentration of 369 unit/ml for cerebrospinal fluid and 386 unit/ml for saliva (t = -.70, P = 0.7) ::: ![](1742-2094-2-13-2) ::: The measurements of the saliva and the CSF HLA-I demonstrated the following: sHLA I was highly elevated in the CSF and saliva of only two patients with RRMS, during an exacerbation (mean = 854 ng/ml), while the remaining eleven patients had no detectable sHLA-I in CSF or saliva. This was also true for those patients with or without contrast enhancement by MRI brain scan. Discussion ========== There is considerable interest in the apparent ability of HLA complex to release molecules, identified as sHLA proteins, into the surrounding fluids as this may translate into a biological monitor of autoimmune disease activity. However, the pathways responsible for, and the potential pathophysiological significance, of sHLA material in different body fluids have not been determined. Active secretion of sHLA-I by liver cells and activated immunocompetent cells \[[@B31],[@B32]\] are suggested sources for its production in serum. A small number of studies have shown that sHLA-I molecules appearing in serum are heterogeneous in molecular mass and multiple molecular forms of sHLA may have different physiological roles \[[@B33]-[@B35]\]. It has been proposed that sHLA-I can appear in the serum as a result of shedding from the cell membranes, can be a product of proteolysis, or can be secreted by an alternative splicing pathway \[[@B8],[@B33],[@B36]\]. It is possible that serum HLA-II may be derived from similar processes. However, there is no supportive data for this assumption. Biochemical studies of sHLA-II in the synovial fluid of patients with rheumatoid arthritis revealed a preferential release of high-molecular-weight (1000 kDa) sHLA-II in the inflamed synovium, but not in the serum \[[@B27]\]. In addition, attempts to induce production of similar material from a cell line expressing HLA-II on a cell surface have failed, indicating that release of sHLA-II is an active process. Of interest, sweat has been shown to possess polymorphic structures identical to those of serum HLA-I. However, excretion of sHLA-I in sweat has been found to be in markedly lower quantities than in serum \[[@B11],[@B38]\]. We reported the occurrence of 39 kDa sHLA-I in saliva as well as in serum during active Sjögren\'s disease and systemic lupus erythematosus, and the presence of 35--37 kDa HLA-I in both body fluids when the disease was relatively inactive \[[@B10],[@B35]\]. Taken together, it appears that the presence of sHLA in different body fluids has physiological relevance. However, it remains to be determined in which body fluids sHLA production reflects immunoreactivity, if indeed this is the case. We reported a substantial elevation of saliva sHLA-I in patients with autoimmune rheumatic diseases, when the saliva sHLA-II concentrations were in normal range \[[@B10]\]. sHLA-I concentrations in saliva were observed to be related to the activity or clinical course of rheumatological diseases. The present study indicates correlative elevation of sHLA-II in the saliva and CSF of patients with RRMS. Of particular interest, the great majority of sHLA-II measurements were equivalently distributed in both body fluids. In contrast, sHLA-I was undetectable in most specimens, with only occasional elevation, possibly associated with some sub-clinical episodes of the disease. It is possible that sHLA-I and sHLA-II are selectively altered by the immunological process and are preferentially impacted by different pathological mechanisms. The differential expression of sHLA concentrations observed in this study requires further investigation to determine if this is directly related to immune responsiveness or is an epiphenomenon of the pathogenetic process. In a recent study, an increase in serum sHLA-I in MS patients treated with interferon beta 1b was reported, and the elevation correlated with response to therapy \[39\]. However, whether sHLA from other body fluids follows a similar pattern remains to be determined as evidenced by the reciprocal relationship between CSF and serum sHLA-I levels in MS reported by these same investigators \[[@B15]\]. It appears from this preliminary study that sHLA-II is the predominant class of sHLA molecules present in the CSF and saliva of MS patients. This is in contrast to CSF and saliva sHLA-I, which we have found to be in non-detectable quantities. The reduced sHLA-I and augmented sHLA-II observed in these body fluids may reflect the active stage of RRMS, triggered by the stimulation of immune system in the absence of immunosuppressive therapy. Our results indicate that measurement of saliva sHLA-II may be a potential noninvasive biological marker of disease activity in a primary CNS disease such as MS. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' contributions ======================= Dr. Adamashvili has contributed to this manuscript by providing expertise for laboratory measurement of soluble HLA-I and soluble HAL-II. Drs. Minagar and Kelley have contributed to this manuscript by recruiting and examining multiple sclerosis patients, interpretation of data and preparing the manuscript. Dr. Gonzalez-Toledo has contributed to this manuscript by interpreting the neuro-radiology studies and generating neuro-radiology data. Dr. Featherston has contributed to this manuscript by doing statistical analysis and generating the figures. Acknowledgements ================ This study was support by a grant from Serono, Inc., Rockland, MA (U.S.A.). We would like to thank Dr. Stephen Jaffe for critical review of this manuscript and for patient referral.	PubMed Central	2024-06-05T03:55:59.698991	2005-6-2	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180848/", "journal": "J Neuroinflammation. 2005 Jun 2; 2:13", "authors": [ { "first": "Irena", "last": "Adamashvili" }, { "first": "Alireza", "last": "Minagar" }, { "first": "Eduardo", "last": "Gonzalez-Toledo" }, { "first": "Liubov", "last": "Featherston" }, { "first": "Roger E", "last": "Kelley" } ] }
PMC1180849	Introduction ============ Balance and postural control in humans is often studied by measuring the sway and/or muscle EMG response to a controlled mechanical perturbation, mainly taking the form of forward and backward or side-to-side platform translations, and foot dorsi- and plantar-flexing rotations \[[@B1]-[@B7]\]. Perturbations have also taken the form of a sudden push or pull to the upper body or waist while subjects stand or walk \[[@B8]-[@B13]\]. While these studies provide a better understanding of postural reflexes to mechanical perturbations, the conditions for the responses often do not correspond to the natural conditions in which individuals with balance impairments fall. Falls in individuals with balance impairments mainly occur during common, everyday activities \[[@B14]-[@B16]\]. Individuals with balance impairments are also susceptible to self-initiated perturbations (cognitively or externally cued but without external forces) such as sudden stops \[[@B17],[@B18]\], turns \[[@B19]\], or stepping corrections to avoid obstacles \[[@B20],[@B21]\]. Numerous studies on balance and postural control from the perspective of non-linear dynamics have been published in the last decade \[[@B22]-[@B28]\]. Collins et al. \[[@B22]\] applied the analysis of Brownian motion (stabilogram-diffusion analysis) to undisturbed standing and concluded that, compared to young healthy subjects, elderly subjects utilized open-loop control schemes for longer periods of time before closed-loop feedback mechanisms were initiated, but that their closed-loop postural control mechanisms were more stable. Mitchell et al. \[[@B25]\] used stabilogram-diffusion analysis to show that people with Parkinson\'s disease (PD) compensate for less stable open-loop control in the anteroposterior direction with increased closed-loop control in mediolateral direction. Van Emmerik et al. \[[@B23]\] applied dimensionality analysis to quiet standing of healthy people and people with tardive dyskinesia, and reported that loss of variability, rather than high sway amplitude, may cause postural instability. Studying the relative phase dynamics between the movements of upper and lower extremities as a function of walking velocity in healthy persons and people with PD, van Emmerik and Wagenaar \[[@B29]\] reported that in PD persons the ability to switch between coordination patterns (flexibility) was reduced whereas the within-pattern variability was decreased (hyperstability) compared to healthy participants. This finding was consistent with the neurological symptom \'rigidity\' assessed by means of the Columbia rating scale. Results were also corroborated by van Emmerik et al. \[[@B30]\], who reported smaller changes in mean relative phase between transversal pelvic and thoracic rotations and a lower variability in relative phase in a PD group compared to a group of healthy individuals. The locomotor stability of people with other neurologic deficits, such as vestibular hypofunction and cerebellar pathology, has received less attention \[[@B31]-[@B34]\], and has not been assessed during perturbed locomotor tasks. The objective of the present study was to investigate the stability of stepping in people with peripheral and central vestibular dysfunction by means of an easily controlled sensory (auditory) perturbation test that is functional and self-initiated (via external cue). We have previously reported a cadence controlled, repeated bench stepping task for studying people with vestibular \[[@B33],[@B34]\] and cerebellar pathology \[[@B32]\]; our results show this activity challenges participants\' locomotor and balance systems. In this report, we applied an auditory perturbation by suddenly changing the cadence of the metronome (100 beat/min to 80 beat/min) at a predetermined time during the trial. The effects of the perturbation on the stability of the movement patterns were studied by applying tools derived from non-linear dynamics. We hypothesized that, when compared to healthy participants, 1) balance impaired participants (vestibular hypofunction and cerebellar pathology) would demonstrate more variability when the perturbation is applied, and 2) recover more slowly from the perturbation. This study should be useful in the development of new approaches for assessing treatment efficacy. Methods ======= Participants and Procedures --------------------------- Participants consisted of five healthy adults (HE: mean age = 43.4 ± 15.5 years), six adults with vestibular hypofunction (VH: mean age = 45.3 ± 10.2 years), and three adults with cerebellar pathology (CB: mean age = 55.6 ± 12.0 years). Sample characteristics are summarized in Table [1](#T1){ref-type="table"}. HE participants were free of orthopaedic, neurologic or other conditions affecting physical performance or balance. Participants with CB were diagnosed by a neurologist\'s examination of the patients\' signs and symptoms and from Magnetic Resonance or Computed Tomography brain scans \[[@B35]\]. Participants with VH were diagnosed using a vestibular test battery and by an otoneurologist\'s examination as either bilaterally (BV) or unilaterally (UV) deficient \[[@B36],[@B37]\]. BV was diagnosed as abnormal vestibulo-ocular reflex gains (at least 2.5 standard deviations below normal) on computerized sinusoidal vertical axis rotation testing, and bilaterally absent caloric responses as determined by cold and warm water stimulation. UV was diagnosed by demonstration of at least one of the following: 30% unilaterally reduced caloric response, positional nystagmus while lying with the damaged ear down, and confirmatory abnormalities on rotational testing. Beyond their respective primary diagnoses, persons with VH and CB had no evidence of other conditions that could affect balance control. All participants signed informed consent forms prior to testing according to institutional guidelines on human research. Specific diagnoses are listed for each participant in Table [2](#T2){ref-type="table"}. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Subject characteristics ::: Age (yrs) Height (m) Weight (kg)\* ------------------------------------------------------------------------ ---------------- -------------- ----------------- [Healthy Participants (5 females)]{.underline} Mean 43.4 1.58 53.6 St. Dev. 15.5 .18 5.0 Range 24.2 -- 59.58 1.22 -- 1.73 45.0 -- 59.1 [Vestibular Hypofunction Participants (5 females, 1 male)]{.underline} Mean 45.3 1.67 92.6 St. Dev. 10.2 .09 28.7 Range 29.92 -- 61.60 1.55 -- 1.83 54.55 -- 145.45 [Cerebellar Pathology Participants (2 females, 1 male)]{.underline} Mean 55.61 1.63 73.87 St. Dev. 11.99 .08 15.09 Range 39.58 -- 68.42 1.55 -- 1.73 56.36 -- 93.18 \* Significant between-groups difference for healthy vs. vestibular hypofunction participants (p= .05) ::: ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Individual subject diagnoses and perturbation error responses. ::: Participant Diagnosis \Emax ^†^Emin ^‡^Edif ------------- ----------------------------------------------- -------- --------- --------- 1 HE -- Healthy .26 .10 62.0 2 HE -- Healthy .31 .15 52.6 3 HE -- Healthy .39 .15 60.5 4 HE -- Healthy .42 .13 68.6 5 HE -- Healthy .46 .17 63.7 6 VH -- Unilateral vestibular hypofunction .46 .11 77.0 7 VH -- Unilateral vestibular hypofunction .56 .23 59.2 8 VH -- Bilateral vestibular hypofunction .61 .34 44.9 9 VH -- Unilateral vestibular hypofunction .61 .16 73.6 10 VH -- Unilateral vestibular hypofunction .91 .23 74.3 11 VH -- Unilateral vestibular hypofunction .94 .27 71.1 12 CB -- Idiopathic spinocerebellar degeneration .52 .26 49.3 13 CB -- Cerebellar dysfunction .64 .22 65.3 14 CB -- Idiopathic spinocerebellar degeneration .94 .39 59.0 \Emax = Maximum fractional error at initiation of perturbation; ^†^Emin = Fractional error recovery at 2--3 cycles after perturbation; ^‡^Edif = Percent difference in fractional error response from initiation to recovery. ::: Participants performed 30 second repeated bench stepping trials using a step up forward/step down backward paradigm: participants were instructed to step forward onto the platform and then step backward off the platform (Figure [1](#F1){ref-type="fig"}), leading with their dominant leg, and synchronizing their foot strikes with the beats of an electronic metronome. The dominant leg was determined by asking participants to pantomime kicking a ball. The platform consisted of two side-by-side 7.6 × 57.6 × 23.0 cm (height × width × depth) blocks placed on the front halves of two 60 cm long Kistler force plates (Kistler Instruments, Inc. Winterthur, Switzerland). Bilateral, three-dimensional body segment kinematics were collected at 152 Hz with four SELSPOT (Selective Electronics, Inc. Partille, Sweden) optoelectric cameras. The cameras were used to track arrays of infrared light emitting diodes embedded in rigid plastic disks, securely strapped to eleven body segments (both feet, shanks, thighs and upper arms, and pelvis, thorax and head). Whole body center of gravity (CG) was computed as previously described by Riley et al. \[[@B38]\] Briefly, center of mass in the global reference frame of each of the eleven body segments during a trial were multiplied by their corresponding segment masses, summed, and divided by the total body mass, to arrive at the whole body CG position as a function of time. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### Three-dimensional android reconstruction of a representative healthy subject performing the stepping task. (a-b-c) The subject steps forward onto the platform with their dominant leg; (c-d-e) steps backward off the platform with their dominant leg. The task is performed repeatedly over a 30 second period (approximately 12 cycles). ::: ![](1743-0003-2-9-1) ::: Participants performed one-to-two unperturbed stepping trials (constant cadence), followed by one cadence perturbation stepping trial. Perturbation trials were performed by changing (within one beat) the metronome frequency during the stepping trial from 100 to 80 beats per min (bpm) at 10 seconds into the trial, and then from 80 to 100 bpm at 20 seconds into the trial. There were two exceptions: one healthy subject continued at 80 bpm instead of returning to 100 bpm at 20 seconds, and one cerebellar pathology patient, who was unable to reach 100 bpm cadence, performed the trial at 80-60-80 bpm. Participants were aware that the cadence would change during the perturbation trial, but not when it would change. Data Analysis ------------- A two-dimensional phase plot was constructed from the anterior/posterior (A/P) velocity component of the whole body CG, X(t) versus X(t+T), where Xwas the order parameter (in this case A/P velocity of the CG), twas time, and Tthe lag time. The appropriate lag time was determined from the first inflection point (zero crossing) of the autocorrelation function of X(t). To simplify the analysis description, we use x(t) = X(t) and y(t+T) = X(t+T). To represent the perturbation response, the attractor trajectory x(t), y(t+T) was compared at each time frame to a reference trajectory x~p~(τ\'), y~p~(τ\') derived from the attractor trajectory prior to cadence perturbation for each subject. The reference trajectory was generated by first estimating the geometric center x~o~, y~o~of the entire attractor time history t~t~, where t~t~= 30-T. ![](1743-0003-2-9-i1.gif) ![](1743-0003-2-9-i2.gif) A phase angle φ(t) was then computed from t= 0 to t~p~seconds (at time step 1 / f= 1 / 152 Hz = 0.0067 seconds) between x(t), y(t+T) and x~o~, y~o~from the expression ![](1743-0003-2-9-i3.gif) and forced to range between 0 and 2πradians (instead of -πto π) and converted to degrees. Time t~p~was 10-T seconds, just prior to onset of the perturbation. The φ(t) array was then sorted into φ\'(τ), where τwas an index array corresponding to ascending values of φ(t) (from 0 to 360). Attractor dimensions were then sorted into x\'(τ) and y\'(τ) and an nth order Fourier series fit was conducted for x\'(τ) and y\'(τ) variables separately, using φ\'(τ) as the independent variable. A 10^th^order fit was found to minimize the residuals. A new independent variable φ~p~(τ\') = 0, 1, 2, \..., 360 was then prescribed and used to compute the reference trajectory coordinates x~p~(τ\') and y~p~(τ\'), where ![](1743-0003-2-9-i4.gif) The Fourier coefficients were computed from ![](1743-0003-2-9-i5.gif) ![](1743-0003-2-9-i6.gif) ![](1743-0003-2-9-i7.gif) where n= f·t~p~, fis the sampling frequency and t~p~the time duration, dis a degree to radian conversion (π/180), and kis the harmonic index. Computation of y~p~(τ\') proceeded in a similar manner. The perturbation magnitude was estimated by computing the Euclidian distance, expressed as the squared fractional error, ε, between the length, r, of a line between x(t), y(t+T) and x~o~, y~o~and length, r~p~, of a line between x~p~(τ\'), y~p~(τ\') and x~o~, y~o~. The latter dimension was determined by first calculating the angle of r(ie. using equation 3), φ~r~, rounding it to the nearest degree, and using it as an index, τ\' = φ~r~to find the corresponding x~p~(τ\'), y~p~(τ\') coordinates. The error was then calculated from ![](1743-0003-2-9-i8.gif) where t= 0 to 30-Tseconds (see also Figure [2](#F2){ref-type="fig"}). ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Schematic computation of the attractor trajectory error. All attractor points from time = 0 to 30-Tseconds are compared to the reference trajectory established for the first 10-Tseconds based on the squared fractional difference, ε, in their radial dimensions from the geometric center of the attractor trajectory orbit. ::: ![](1743-0003-2-9-2) ::: To compare groups of participants, the error data for each subject was first binned into 2 second intervals (a total of 5 intervals) between the 10 second and 20 second marks. The peak error was then documented for each bin. The maximum value of the five peaks (Emax, occurring in the first or second bin) and minimum value of the five peaks (Emin, occurring in the last bin) were then recorded for each subject. The magnitudes of Emax and Emin both represent the stability of the participants following the auditory perturbation. The magnitude of Emin also indicates participants\' ability to recover. We also analyzed the difference between Emax and Emin (Edif), as a measure of participants\' recovery response, relative to their initial perturbation response. Analysis of variance (ANOVA) was used to compare dependent variables (Emax, Emin and Edif) among groups of participants at an alpha level of .05. All statistical comparisons were conducted using SPSS (v10, SPSS, Chicago, IL). Results ======= There were no significant differences in age (p= .50) and height (p= .59) between groups, but weight was significantly greater for the VH participants compared to the HE participants only (p= .05). Cadence Perturbation Analysis ----------------------------- Figure [3](#F3){ref-type="fig"} illustrates for a representative HE participant the two dimensional attractor and reference trajectory for A/P velocity of the CG during a repeated stepping test with no cadence perturbation (Figure [3a](#F3){ref-type="fig"} and [3b](#F3){ref-type="fig"}), and with a cadence perturbation (Figure [3c](#F3){ref-type="fig"} and [3d](#F3){ref-type="fig"}). The calculated error for the attractors (left panel) are shown in error plots (right panel). The sharp transition in the error at 11--12 seconds (Figure [3d](#F3){ref-type="fig"}) corresponds to the cadence transition from 100 steps/minute to 80 steps/minute. Figure [3d](#F3){ref-type="fig"} indicates that the HE participant was able to return to a stable trajectory within 2 to 3 cycles, though the error remained slightly higher than prior to the perturbation. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Attractor trajectory error for a representative healthy subject during the stepping task. The top panels are: (a) The A/P CG velocity attractor during an unperturbed cadence trial; (b) The attractor error for the unperturbed trial; (c) The A/P CG velocity attractor during an perturbed cadence trial; (b) The attractor error for the perturbed trial. Note the delay response in the attractor relative to the cadence change (it will require at minimum one step to realize the beat has changed). ::: ![](1743-0003-2-9-3) ::: Representative stepping perturbation data for a VH and a CB participant are shown in Figure [4](#F4){ref-type="fig"}. The left panels of Figure [4](#F4){ref-type="fig"} demonstrate erratic attractor behavior in these individuals, and the right panels of Figure [4](#F4){ref-type="fig"} shows the resulting error calculations for these participants. Compared to the HE subject in Figure [3](#F3){ref-type="fig"}, data in Figure [4](#F4){ref-type="fig"} shows that a return to a stable trajectory does not occur within 2 to 3 cycles for those with balance disorders. As with the healthy subject (see Figure [3d](#F3){ref-type="fig"}), there is a time delay between perturbation onset and response of the attractor. Error measures (Emax, Emin and Edif) for all participants are summarized in Table [2](#T2){ref-type="table"}. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Attractor trajectories for two representative balance impaired patients during the stepping task. The top panels are for a patient with cerebellar dysfunction: (a) The A/P CG velocity attractor during a perturbed cadence trial; (b) The attractor error for the perturbed trial. The bottom panels are for a patient with vestibular hypofunction: (c) The A/P CG velocity attractor during a perturbed cadence trial; (b) The attractor error for the perturbed trial. As with the healthy subject (see Figure 3), there is a time delay between perturbation onset and response of the attractor, however, this particular cerebellar subject was suddenly confused by the change and momentarily lost the pace. ::: ![](1743-0003-2-9-4) ::: Our hypothesis that balance impaired participants would demonstrate a greater perturbation response than healthy participants, as measured by the fractional error variables, was supported. One-way ANOVA revealed significant between-groups differences for Emax (p= .019), and Emin (p= .028). Both balance impaired groups had significantly higher Emax than HE participants (CB: p= .049; VH: p= .026), but were not different from each other (p= .985). Using age and weight as covariates did not change the significant outcomes; both Emax and Edif were significantly different between groups (p= .027 and p= .023, respectively) when controlling for these potentially confounding variables. Mean errors for the CB group, VH group and the HE group are shown in Figure [5](#F5){ref-type="fig"}. It should be noted that the highest error observed (.94) was for both a CB and a VH participant (Table [2](#T2){ref-type="table"}). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Maximum (Emax) and minimum (Emin) peak squared fractional errors from 2 second interval bins during 10 seconds following the perturbation. Peak error Emax at perturbation (p= .019) and peak error Emin after 10 seconds (p= .028) were greater for balance impaired patients compared to healthy subjects. ::: ![](1743-0003-2-9-5) ::: Our hypothesis that balance impaired participants demonstrate a slower recovery to the perturbation response than healthy participants was also supported. Although Emin was significantly different between groups (p= .028), it was only significantly higher for CB participants compared to HE participants (p= .026); there was no significant difference between VH and HE participants (p= .147). Interestingly, the between-groups differences in Edif approached the level of significance (p= .051). The reason became clear when Edif was expressed as percent decease: all three groups decreased their error by approximately 60% in the 10 second interval following onset of the cadence perturbation: the recovery time for balance impaired participants was longer than for healthy participants because their error response was so much higher. Discussion ========== While measures of standing stability are commonplace, measures of locomotor stability in balance impaired individuals are few \[[@B29],[@B31]-[@B34],[@B39]\]. In this report we describe a locomotor perturbation test and analytical procedure for quantifying postural control during a dynamic functional motor task. The findings of the present study indicate that both balance impaired groups (vestibular hypofunction and cerebellar pathology) revealed a more variable stepping pattern and a slower recovery as a result of the cadence perturbation compared to the healthy participants, suggesting the balance impaired individuals experienced difficulty maintaining fluid movement during the trial, with a diminished ability to predict future position of the whole body CG. However, as shown by Table [2](#T2){ref-type="table"} and Figure [5](#F5){ref-type="fig"}, our data do not discriminate between peripheral and central vestibulopathy, or within a diagnostic group (bilateral vs. unilateral vestibular hypofunction); indeed, a larger study would be needed to test the power of the protocol and analytical method for this purpose. While the error means for both balance impaired groups were not statistically different, and the highest error response (.94) was observed in both the CB and VH participants, the most interesting responses were observed in the CB group. Although qualitative, observation of computer animated stepping trials suggested that two of the three CB participants were unable to smoothly adjust their stepping cadence when the cadence perturbation was applied, and appeared to have difficulty regaining the inter-limb coordination required to match the new metronome beat. This supports our previous finding that people with CB have poor inter-limb coordination during a repeated stepping task compared to their healthy counterparts \[[@B32]\]. Furthermore, Timman and Horak \[[@B40]\] found that participants with cerebellar pathology are less able to scale anticipatory postural adjustments when stepping was cued with a backward translation of the support surface. Our data suggests that cerebellar pathology also affects the ability to scale postural adjustments during unanticipated cadence perturbation. VH participants had a slightly, though not significantly, lower error response than CB participants, and had significantly higher error response compared to HE participants. This latter finding also supports our previous reports that people with VH are less stable \[[@B34]\] and less smooth \[[@B33]\] during a stepping task than are their healthy counterparts. The perturbation response for the VH group was probably not due to difficulty controlling interlimb coordination, but rather, due to cadence corrective action (after the perturbation) coming too late to slow down the center of gravity after the perturbation is cognitively realized. The late corrective action, allowing the attractor trajectory to deviate further from its orbit, was perhaps due to additional time required of visual and proprioceptive mechanisms to re-assert control over head and gaze stability. Van Emmerik and Wagenaar \[[@B29]\] studied the relative phase and frequency dynamics of interlimb coordination and trunk rotation during walking in people with Parkinson\'s disease (PD) and healthy participants when systematically varying walking speed. Their findings revealed that people with PD often have a reduced ability to switch between walking patterns and relatively more stable coordination patterns compared to young healthy participants. They hypothesized that the hyper-stable coordination patterns in PD cause a reduced flexibility (that is, ability to switch between coordination patterns). The results of the present study indicate that the balance impaired individuals have a larger variability in stepping behavior and a slower recovery (longer relaxation time) as a result of the perturbation. It suggests that a hypo-stable stepping pattern results in a slower recovery from a perturbation, which makes, for example, balance impaired individuals more at risk for falls. Van Wegen et al. \[[@B28]\] reported that healthy elderly and people with PD show a decreased time-to-contact variability in body sway during quiet standing in the medio-lateral direction; older adults and people with PD remained a larger distance from their stability boundary than young participants. In addition, it was found that during walking, in the higher frequency ranges (3--12 Hz), younger participants had higher power than the older participants, while in the lower frequency ranges (0--3 Hz), the older participants had higher power than the younger participants (see also van Emmerik et al. ..\[[@B30]\]). In their approach to coordination, fluctuations i.e., variability, can play a functional role in the stabilization and adaptation of coordination patterns. From this perspective, a reduction in variability (hyper-stability) also has a negative impact on movement coordination. The findings of the present study strongly suggest that in people with peripheral and central vestibulopathy the flexibility of movement coordination is reduced (increased variability) as a result of hypo-stable stepping patterns. On the basis of the above-mentioned findings we hypothesize that a similar problem in stability and flexibility during stepping or walking may exist in healthy elderly at risk for falls. The shape of the attractor in Figure [3](#F3){ref-type="fig"} for a HE subject resembles a diamond and has closely packed orbital trajectories. When a cadence perturbation is applied, the predictive quality of the attractor breaks down during the transition from a 100 bpm orbital trajectory to an 80 bpm orbital trajectory. Even the HE subject shown in Figure [4](#F4){ref-type="fig"} required two to three steps to restabilize the new trajectory. Participants with peripheral (VH) and central (CB) vestibulopathy disorders did not transition as smoothly as HE participants when moving between 100 bpm and 80 bpm, however, they appeared to adapt at a similar rate over a 10 second interval. Testing for a longer interval at 80 bpm following the perturbation onset, however, would be required to determine if indeed there are differences in recovery rate; the need for longer testing was exemplified by the fact that error magnitudes did not return to pre-perturbation levels for any participants. It is important to note that the recovery time following perturbation depends on when the perturbation occurs within the stepping cycle, and the feedforward nature of volitional stepping. These factors probably contribute to the variability in Emax and Emin times, and hence influence the recovery time response, more so than system time constants (such as the 6 msec VOR response or 100 msec \"long loop\" response to the brain and back to muscle \[[@B41]\]). The cadence, and cadence transition, applied for participants may also be a factor influencing the results. To assess the sensitivity of the attractor geometry to the stepping rate, and thereby provide a rationale for the cadence perturbation rates chosen to conduct on participants, we examined the attractor geometry for several participants (not included in this study) who performed the stepping trials at different cadences (60--152 bpm). When plotting the attractor radius against stepping rate (data not presented here), we found a curvilinear relationship suggesting the attractor radius peaks in dimension at about 120 bpm, but that the difference between 100 bpm and 80 bpm was sufficiently large and linear. We concluded that our choice of a cadence perturbation was appropriate for the participants studied. It is important to note that we did not quantify hearing ability of the study participants, although no participant indicated hearing impairment on their entry medical screening. Quantifying hearing ability would be important for a larger study because the perturbation requires one to detect the metronome transition. Because the sample was small, we also chose to ignore the gender of the participants. Indeed, a larger study would not ignore such influences. Furthermore, there were differences in age (though not statistically significant) and weight (significant at p= .05 between VH and HE) among groups that cannot be ignored, as response latency in concurrent cognitive tasks may be influenced by age-related and other impairment \[[@B42]\]. However, because we found that the between-groups differences persisted when age and weight were used as covariates, we are confident in our conclusion that balance-impairment explained the majority of the differences observed between groups. We also analyzed only the velocity perturbations in the anterior-posterior direction. It is reasonable to expect that a similar analysis of the medio-lateral velocities may yield interesting results. In cases where the body CG cannot be estimated accurately (for example, when using systems that only track a few body segments), other measures, such as pelvis or trunk marker velocities should be compared to the results in this analysis which uses the whole body CG velocity derived from an 11 segment inertial body model \[[@B38],[@B43],[@B44]\]. We conclude that the cadence perturbation test is useful for quantifying locomotor stability control in people with peripheral or central vestibulopathy. People with damaged vestibular systems or those with cerebellar damage performed significantly worse on the cadence perturbation tests compared to healthy participants. Clearly our results are not necessarily generalizable due to the small pilot study sample used and the above identified limitations; however, the data presented do suggest that the tool used to quantify stepping stability, derived from non-linear dynamics, is useful and sensitive enough to detect the effects of stepping cadence changes when controlled by external auditory cues. Competing interests =================== The author(s) declare that they have no competing interests. Authors\' Contributions ======================= All authors participated in the overall study design, contributed to the interpretation of data and writing/editing of the manuscript, and have read and approved the final manuscript. CAM conceived the hypotheses, developed and programmed the non-linear dynamic analysis methods, carried out the data analysis, and prepared the manuscript; DEK was the principal investigator of the project; RW was the project consultant. Acknowledgements ================ Supported by the National Institutes of Health (R01-AG11255, R21-AT000553). The authors thank Dov Goldvasser, MScE, for assistance with data processing.	PubMed Central	2024-06-05T03:55:59.700790	2005-5-27	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180849/", "journal": "J Neuroengineering Rehabil. 2005 May 27; 2:9", "authors": [ { "first": "Chris A", "last": "McGibbon" }, { "first": "David E", "last": "Krebs" }, { "first": "Robert", "last": "Wagenaar" } ] }
PMC1180850	The National Institutes of Health has issued a new policy for open access to all NIH-funded research, whereby NIH grantees are strongly encouraged to deposit their complete manuscripts and supplementary material to an Internet repository, thereby enabling the rapid dissemination of research results and the long-term archiving of scientific literature \[[@B1],[@B2]\]. Entitled \"Policy on enhancing public access to archived publications resulting from NIH-funded research,\" the Public Access Policy is a major step towards the fulfillment of a universal right to access scientific information without barriers and free of charge (i.e., Open Access). The new policy derives historically from a request made by the US House Committee on Appropriations to the National Library of Medicine in 2004 to identify potential remedies to \"ensure that taxpayer-funded research remains in the public domain\" and to \"alleviate the restrictive trend in information technology\" caused by a \"dramatic rise in medical research data subscription costs\" \[[@B3]\]. Publish or Perish ----------------- Publications are perhaps the sole currency of scientific research\--for it is publications which beget funding, and in turn, funding which begets more publications\--and as such they are vitally important to the career of any research scientist. How, when, and where the research is published can be as significant as the research results themselves since the influence of a research article may only be as potent as its ability to attract an audience of readers and thereby disseminate through the field. Indeed, the root of the word publication implies its dissemination to a publicreadership generally, and in this way the progress of science is archived in the historical record. True to this spirit, the NIH has initiated the Public Access Policy \[[@B1],[@B2]\] and has created a single repository (PubMed Central, or PMC) to archive the corpus of biomedical research\--past, present, and future. This Public Access Policy follows on the heels of a similar initiative in the United Kingdom last year when the House of Commons Science and Technology Committee recommended the promotion of Open Access in the UK to all publicly funded scientific research \[[@B4]\]. Open Access, or How to Archive Yourself for the Ages ---------------------------------------------------- The NIH proposal mandates Open Access, but only to those research articles deriving in part or whole from direct costs provided by NIH grants. Nonetheless, this policy will likely apply to a major fraction of all research publications. By its own estimation, the NIH currently funds at least ten percent (65,000 articles) of all biomedical literature annually \[[@B5]\]. Moreover, PubMed Central will further expand due to the continuing submission (since its inception in 2000) of all final articles published in Open Access journals. To date, PubMed Central \[[@B6]\] archives approximately 100,000 articles from over 130 biomedical journals. Such a single repository, covering the full spectrum of research literature and freely available to the world, is revolutionary. The NIH directive, although strongly encouraged, was issued merely as a \"request,\" and so the successful implementation of the Public Access Policy will rest on the shoulders of researchers, and traditional publishing houses, and their collective wherewithal to build the archive by willfully submitting manuscripts to the PMC. The journal The Proceedings of the National Academy of Sciences USA (PNAS)has fully embraced the new NIH open-access policy. PNAShas in the past and will continue in the future to deposit all of its final, publisher-formatted articles into the PMC repository after a six-month embargo, regardless of the funding source. Furthermore, PNAShas adopted a democratic compromise by allowing authors to choose an \"Open Access option\" whereby the publisher-formatted final edition of their paper can be made freely available at PMC and PNASimmediately upon its online publication \[[@B7]\]. For this, PNASimposes a modest surcharge (\$750) that is in line with the \"author pays\" model used presently by most of the approximately 1500 Open Access journals \[[@B8]\]. Open Access already exists \[[@B9]\] and other traditional, subscription-based journals are invited to follow the bellwether example of PNAS, and at least explore the possibility of making the transition to an Open Access model of publishing. Indeed it seems that an increasing number of subscription-based journals are exploring hybrid approaches, such as that used by PNAS, that enable authors to offer their content online by paying publication costs for open access. But the publishing world is uniquely devoid of market forces that might expedite this transition. One need only look to the skyrocketing subscription costs (print or online) for scientific and medical journals, and the inability of institutional libraries to afford them. A study by the NIH concluded that journal prices increased during the last decade at a rate that was over 6 times inflation \[[@B10]\]. A recent editorial published in the New England Journal of Medicineacknowledges that some journal editors and publishers will perceive some aspects of the Public Access Policy as \"potential threats\" to their \"revenue sources\" \[[@B11]\]. In addition, some traditional publishers have assaulted Open Access journals and the NIH policy initiative with dubious arguments; the response of the Open Access community speaks for itself \[[@B1],[@B2],[@B12],[@B13]\]. Open Access, or How to Improve your Impact Score ------------------------------------------------ A study by Thomson ISI--the company that assigns impact factors to journals--indicates that, among the 200 Open Access journals it monitors, these journals are competitive with their traditional counterparts: Open Access journals \"adhere to high publishing standards, are peer reviewed comparably to other journals in their respective fields, and are cited at a level that indicates they compete favorably with similar journals in their field\" \[[@B14]\]. One telling example is the high impact factor of the relatively new BMC Bioinformatics\[[@B15]\]. Since Open Access journals are freely available to all researchers with a connection to the Internet, the journals\' impact factors are enhanced de factodue to the immediacy and ease with which their research articles may be retrieved, disseminated, and cited by scientific colleagues. It is this immediacy that, in part, fuels the movement towards Open Access, for such immediacy is required now more than ever given the breakneck speed of scientific progress. This is reflected in the changing landscape of scientific publishing as it transitions from print-based to electronic journals for the communication of biomedical research. BioMed Central (BMC) and the Public Library of Science (PLoS) are two current, and successful, Open Access publishing models that provide immediate access to biomedical research articles. With high visibility and high quality articles monitored by strict peer-review standards, these publishing \"houses\" provide an alternative to traditional print journals and embody the ideals recently articulated in the NIH Public Access Policy. Thus, we encourage compliance with the new NIH policy and, furthermore, would suggest that authors submit their research articles to an Open Access journal, all of which archive their content in PubMed Central. Alternatives to traditional subscription-based journals continue to gain momentum as policies in support of Open Access are made by some of the world\'s largest science funding agencies, namely, the NIH, the Howard Hughes Medical Institute, and the Wellcome Trust. All these agencies now provide generous budget supplements to cover the costs of publication in Open Access journals \[[@B10]\]. It is important to emphasize that, just as indirect costs included in NIH grant funds have long been used for financing journal subscriptions, NIH grant policies permit the use of grant funds to pay for publication costs. In fact, the NIH has made the conservative estimate that \$30 million already is paid annually in direct costs to traditional print journals to defray page charges and other publication costs \[[@B5]\]. This sum would easily cover the author expense of 30,000 Open Access publications, if one assumes, conservatively, an average cost of \$1,000 per article. Publication models evolve over time. We have entered a very exciting period in which the scientific community can choose between two models, the traditional, subscription-based model or the Open Access model. As the number of Open Access journals and publishing houses grows, the scientific community is now in the position to \`vote\' for one or the other model. We firmly believe that immediate and unrestricted access to scientific information will be the gold standard for scientific publication and urge every researcher to submit their manuscripts to Open Access journals. The growing number of freely available articles, which are archived in PubMed Central, marks the trend towards Open Access. Competing interests =================== GCI, CD and PWT declare that there are no competing interests. CS is Deputy Editor of Molecular Cancerand receives no remuneration for his efforts. CS is exempted from Molecular Cancer\'sArticle Processing Fee. Authors\' contributions ======================= GCI wrote and finalized this manuscript. CS, CD and PWT provided critique and suggestions. All authors read and approved the final manuscript.	PubMed Central	2024-06-05T03:55:59.703363	2005-6-2	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180850/", "journal": "Mol Cancer. 2005 Jun 2; 4:20", "authors": [ { "first": "Gregory C", "last": "Ippolito" }, { "first": "Christian", "last": "Schmidt" }, { "first": "Chhaya", "last": "Das" }, { "first": "Philip W", "last": "Tucker" } ] }
PMC1180851	Background ========== The dissimilatory ferric reductase, which are found in bacteria are an essential part of the iron cycles \[[@B1]\] and are essentially intracellular, but one extracellular one was isolated from Mycobacterium paratuberculosis\[[@B2]\]. Another possible mechanism could be active in this process since it was discovered that some bacteria reduce Fe^3+^oxides by producing and secreting small, diffusible redox compounds that can serve as electron shuttle between the microbe and the insoluble iron substrate \[[@B3]\]. The role of excreted compounds in extracellular electron transfer was recently reviewed \[[@B4]\]. The presence of hydrogenase in fungus as Fusarium oxysporumwas demonstrated with washed cell suspensions that had been grown aerobically and anaerobically in a medium with glucose and salts amended with nitrate \[[@B5]\]. The nitrate reductase was apparently essential for ferric iron reduction \[[@B6]\]. Many fungi that exhibit these characteristic properties, in general, are capable of reducing Au (III) or Ag (I) \[[@B7]\]. Besides these extracellular enzymes, several naphthoquinones \[[@B8]-[@B10]\] and anthraquinones \[[@B11]\] with excellent redox properties, were reported in F. oxysporumthat could be act as electron shuttle in metal reductions \[[@B3]\]. Although it is known that microorganisms such as bacteria, yeast and now fungi play an important role in remediation of toxic metals through reduction of the metal ions, this was considered interesting as nanofactories very recently \[[@B12]\]. Using these dissimilatory properties of fungi, the biosynthesis of inorganic nanomaterials using eukaryotic organisms such as fungi may be used to grow nanoparticles of gold \[[@B13]\] and silver \[[@B14]\] intracellularly in Verticilliumfungal cells \[[@B15]\]. Recently, it was found that aqueous chloroaurate ions may be reduced extracellularly using the fungus F. oxysporum, to generate extremely stable gold \[[@B16]\] or silver nanoparticles in water \[[@B17]\]. Other process, which was described in the literature, was related to produce silver nanoparticles through oligopeptides catalysis, precipitating the particles with several forms (hexagonal, spherical and triangular) \[[@B18]\]. However, in the fungal reduction of Ag ions led colloidal suspension, differently that in the oligopeptides case. Then the mechanistic aspects are still an open question, however this process occur in the fungal case probably either by reductase action or by electron shuttle quinones or both. Our aims in this research are to compare different strains of F. oxysporumin order to understand if the efficiency of the reduction of silver ions is related to a reductase or quinone action. Results and Discussion ====================== The Erlenmeyer flasks with the F. oxysporumbiomass were a pale yellow color before the addition of Ag^+^ions and this changed to a brownish color on completion of the reaction with Ag^+^ions for 28 h. The appearance of a yellowish-brown color in solution containing the biomass suggested the formation of silver nanoparticles \[[@B21]\]. The UV-Vis spectra recorded from the F. oxysporum07SD strain reaction vessels (Method A) at different times of reaction is presented in Figure [1](#F1){ref-type="fig"}. The strong surface plasmon resonance centered at ca. 415--420 nm clearly increases in intensity with time. The solution was extremely stable, with no evidence of flocculation of the particles even several weeks after reaction. The inset of Figure [1](#F1){ref-type="fig"} shows UV-Vis spectra in low wavelength region recorded from the reaction medium exhibited an absorption band at ca. 265 nm and it was attributed to aromatic amino acids of proteins. It is well known that the absorption band at ca. 265 nm arises due to electronic excitations in tryptophan and tyrosine residues in the proteins. This observation indicates the release of proteins into solution by F. oxysporumand suggests a possible mechanism for the reduction of the metal ions present in the solution \[[@B17]\]. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### UV-Vis spectra recorded as a function of time of reaction of an aqueous solution of 10^-3^M AgNO~3~with the fungal biomass (07SD). The inset shows the UV-Vis absorption in the low wavelength region. ::: ![](1477-3155-3-8-1) ::: Figure [2](#F2){ref-type="fig"} shows the fluorescence emission spectra of fungal filtrate of one of the strain (07SD). An emission band centered at 340 nm was observed. The nature of the emission band indicates that the proteins bound to the nanoparticle surface and those present in the solution exist in the native form \[[@B22]\]. The similar results were observed for all the studied strains as shown in Table [1](#T1){ref-type="table"}. In Table [1](#T1){ref-type="table"}, the 07SD strain appeared as the most efficient one in the silver nanoparticles production. Apparently, the different efficiencies are related to the reductase and/or to the quinone generation and will be discussed later. A destabilization of the nanoparticles is evident in the case of F. oxysporum534, 9114 and 91248 strains at 28 hrs, as indicated by a decrease in the 420 nm absorption. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Fluorescence emission spectrum recorded from the silver nanoparticles-fungus reaction mixture. The excitation wavelength was 260 nm. ::: ![](1477-3155-3-8-2) ::: ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Relative absorbance at 420 nm of several F. oxysporumstrains ::: Strains Relative Absorbance (420 nm)^a^ --------- --------------------------------- --------- ---------- 0 hours 6 hours 28 hours O6 SD 1 2.041 4.854 07 SD 1 4.007 7.922 534 1 1.682 0.712 9114 1 1.686 0.572 91248 1 3.549 1.372 ^a^Normalized absorbance. ::: Similarly, when the biomass was immersed in water and only the fungal filtrate (Method B) was added to a 10^-3^M AgNO~3~solution, the initially colorless aqueous solution changed to a pale yellowish-brown within 28 h of reaction (data not shown), clearly indicating that the reduction of the ions occurs extracellularly through reducing agents released into the solution by F. oxysporumas it shows the UV-Vis spectra for the 07SD strain (Fig. [3](#F3){ref-type="fig"}). ::: {#F3 .fig} Figure 3 ::: {.caption} ###### UV-Vis spectra recorded as a function of time of reaction of an aqueous solution of 10^-3^M AgNO~3~with the fungal filtrate (07SD). The inset shows the UV-Vis absorption in the low wavelength region. ::: ![](1477-3155-3-8-3) ::: Figures [4](#F4){ref-type="fig"} and [5](#F5){ref-type="fig"} shows the SEM micrograph recorded from the silver nanoparticle (Method A). This picture shows silver nanoparticles aggregates. In this micrograph, spherical nanoparticles in the size range 20--50 nm were observed. The nanoparticles were not in direct contact even within the aggregates, indicating stabilization of the nanoparticles by a capping agent. This corroborates with the previous observation by Ahmad et al. \[[@B17]\] in their study on F. oxysporum. The same micrograph in the Method B was observed (not showed). In the analysis by energy dispersive spectroscopy (EDS) of the silver nanoparticles was confirmed the presence of elemental silver signal (Figure [6](#F6){ref-type="fig"}). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### SEM micrograph from F. oxysporum07 SD strain at ×11000 magnification. ::: ![](1477-3155-3-8-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### SEM micrograph from F. oxysporum07 SD strain at ×40000 magnification. ::: ![](1477-3155-3-8-5) ::: ::: {#F6 .fig} Figure 6 ::: {.caption} ###### EDS spectra of silver nanoparticles. ::: ![](1477-3155-3-8-6) ::: The TLC (Cromatography of Thin Layer) analysis on silica gel 60 plates using chloroform-methanol-acetic acid (195:5:1) showed a spot with Rf value of 0.65, and using benzene-nitromethane-acetic acid (75:25:2) showed a spot with Rf value of 0.85, corresponding to 2-acetyl-3,8-dihydroxy-6-methoxy anthraquinone or its isomers at 2-acetyl-2,8-dihydroxy-6-methoxy anthraquinone (Scheme 1). This was corroborated by the fluorescence spectrum of the filtrate (Method A), which indicates an anthraquinone fluorescence moiety \[[@B11]\]. The excitation spectra at the maximum emission (550 nm) fit quite well with the absorption spectrum of the anthraquinone in Figure [7](#F7){ref-type="fig"}. ::: {#F7 .fig} Figure 7 ::: {.caption} ###### Fluorescence emission spectrum from the aqueous solution of 10^-3^M AgNO~3~with the fungal biomass (07SD). The excitation wavelength was 465 nm. The inset shows the fluorescence excitation spectrum (λ emission at 550 nm). ::: ![](1477-3155-3-8-7) ::: The Figure [8](#F8){ref-type="fig"} shows the nitrate reductase through the reaction of nitrite with 2,3-diaminophthalene. The emission spectrum exhibits two major peaks of fluorescence intensity at 405 and 490 nm corresponding to the emission maximum of the and 2,3-diaminonapthotriazole, DAN (excess) respectively. The intensity of these two bands increased with the addition of a 0.1% KNO~3~solution, confirming the presence of nitrate reductase. ::: {#F8 .fig} Figure 8 ::: {.caption} ###### Fluorescence emission spectra for the reaction of nitrite with 2,3-diaminophthalene. In the emission spectra the curves A and B were, respectively: fungal filtrate and fungal filtrate and 0.1% KNO~3~solution. The maximum excitation wavelength was at 375 nm. ::: ![](1477-3155-3-8-8) ::: It appears that the reductase is responsible for the reduction of Ag^+^ions and the subsequent formation of silver nanoparticles. The same observation was reported with another strain of F. oxysporumand it was pointed out that this reductase was specific to F. oxysporum. However, Fusarium moniliforme, did not result in the formation of silver nanoparticles, neither intracellularly nor extracellularlybut contained intra and extra cellular reductases in a similar fashion as F. oxysporum\[[@B17],[@B23]\]. This is an indication that probably the reductases in this kind of Fusariumare important for Fe (III) to Fe (II) but not to Ag (I) to Ag (0). Moreover, in F. moniliformeanthraquinones derivatives were not detected unlike the case of F. oxysporum. Both fusariumwere alike in the production of naphthaquinones \[[@B8]\] but differed in the production of anthraquinones. Probably, in our case, Ag (0) reduction was mainly due to a conjugation between the electron shuttle with the reductase participation as shown in Figure [9](#F9){ref-type="fig"}. ::: {#F9 .fig} Figure 9 ::: {.caption} ###### Hypothetical mechanisms of silver nanoparticles biosynthesis. ::: ![](1477-3155-3-8-9) ::: Conclusion ========== Even though gold/silver nanoparticles have been synthesized using prokaryotes such as bacteria \[[@B24],[@B25]\] and eukaryotes such as fungi \[[@B13],[@B14]\], the nanoparticles grow intracellularly, except in the case of a recent report in which F. oxysporumwas used. In that case the nanoparticles grew extracellularly \[[@B17]\]. In our case, all the F. oxysporumstrains studied exhibited silver nanoparticle production capacity, however, depending on the reductase/electron shuttle relationships under these conditions. Biologically synthesized silver nanoparticles could have many applications, in areas such as non-linear optics, spectrally selective coating for solar energy absorption and intercalation materials for electrical batteries, as optical receptors, catalysis in chemical reactions, biolabelling \[[@B26]\], and as antibacterials capacity \[[@B27]\]. Methods ======= The F. oxysporumstrains used were the following: O6 SD, 07 SD, 534, 9114 and 91248 from ESALQ-USP Genetic and Molecular Biology Laboratory-Piracicaba, S.P., Brazil. The fungal inoculates were prepared in a malt extract 2% and yeast extract 0.5% at 28°C in Petri plates. The liquid fungal growth was carried out in the presence of yeast extract 0.5% at 28°C for 6 days. The biomass was filtrated and resuspended in sterile water. Silver reduction and its characterization ----------------------------------------- Method A: In the silver reduction, the methodology described previously was followed \[[@B17]\]. Briefly, approximately 10 g of F. oxysporumbiomass was taken in a conical flask containing 100 mL of distilled water. AgNO~3~solution (10^-3^M) was added to the erlenmeyer flask and the reaction was carried out in the dark. Periodically, aliquots of the reaction solution were removed and the absorptions were measured using a UV-Vis spectrophotometer (Agilent 8453 -- diode array). Method B: Another test was also carried out as following: approximately 10 g of F. oxysporumbiomass was taken in a conical flask containing 100 mL of distilled water, kept for 72 h at 28°C and then the aqueous solution components were separated by filtration. To this solution, AgNO~3~(10^-3^M) was added and kept for several hours at 28°C. The silver nanoparticles were characterized by scanning electron microscopy (SEM) and energy-dispersive spectroscopy (EDS) at a voltage of 20 kV (Jeol -- JSM-6360LV) and previously coated with gold under vacuum. Determination of the electron-shuttling compounds ------------------------------------------------- Release of electron-shuttling compounds was followed the methodology described previously \[[@B11]\]: In order to determine the water-soluble quinones that might function as an electron shuttle, cultures were filtered 4--6 weeks, and the filtrate adjusted to pH 3 with HCl 1 M. The acidified solution was then passed through a column with ion exchange resin (Amberlite^®^) for absorption of the pigments. Compounds were removed from the column by elution with acetone, the acetone removed using a Buchi rotary evaporation and the aqueous phase extracted 3 times with ethyl acetate. All ethyl acetate extractions were combined and reduced using the rotary evaporator. After that, 2 μL samples were repeatedly spotted on a Silica gel 60 plate until a spot was visible under UV light at 254 nm. Samples were resolved using a chloroform-methanol-acetic acid (195:5:1) and benzene-nitromethane-acetic acid (75:25:2) system designed to mobilize polar pigments. Plates were air dried, and spots visualized under UV light \[[@B19]\]. Nitrate reductase assay ----------------------- Nitrate reduction was demonstrated in the same medium (Method A and B) of the same growth broth of F. oxysporumwith the addition of 0.1% of KNO~3~\[[@B6]\]. The nitrate reductase test was made after 2 days by fluorometric method \[[@B20]\]. Briefly, 100 μL fungal filtrate and 200 μL of dionized water. To this, 10 μL of freshly prepared 2,3-diaminonaphtalene (DAN) (0.05 mg/mL in 1 M HCl) is added and mixed immediately. After 10 min incubation at 20°C, the reaction was stopped with 5 μL of 0.1 M NaOH. The intensity of the fluorescent signal produced by the product was maximized by the addition of base. The 2,3-diaminonapthotriazole formation was measured using a Perkin-Elmer (LS-55) luminescence spectrophotometer with and excitation wavelength at 375 nm and the emission band measured at 550 nm \[[@B20]\]. Determination of the tryptophan/tyrosine residues ------------------------------------------------- Presence of tryptophan/tyrosine residues in proteins release in the fungal filtrated was analyzed by fluorescence \[[@B17]\]. The fluorescence measurements were carried out on a Perkin-Elmer (LS-55) luminescence spectrophotometer. The exitation wavelength was 260 nm, close to maximal optical transitions of the tryptophan and tyrosine. Authors\' contributions ======================= ND conceived the study, together with OLA and EE and participated in its design and coordination and collected all the data and wrote the paper. PDM obtained all the SEM views, performed the enzymatic assays, the electron shuttling aspects and discussed the three related parts in the manuscript. GIHS performed all the fungal tests and measured all the spectroscopic variations of the plasmon resonance of the silver nanoparticles supervised by EE. OLA also supervised all the nanoparticles aspects in this work. All authors read and approved the final manuscript. Acknowledgements ================ Supports from Brazilian Network of Nanobiotechnology, CNPq/MCT and FAPESP are acknowledged. We acknowledge Dr. Fernando de Oliveira from NCA-UMC for the UV-Vis analyses support.	PubMed Central	2024-06-05T03:55:59.704615	2005-7-13	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180851/", "journal": "J Nanobiotechnology. 2005 Jul 13; 3:8", "authors": [ { "first": "Nelson", "last": "Durán" }, { "first": "Priscyla D", "last": "Marcato" }, { "first": "Oswaldo L", "last": "Alves" }, { "first": "Gabriel IH", "last": "De Souza" }, { "first": "Elisa", "last": "Esposito" } ] }
PMC1180852	Background ========== Integration of viral genomes into host cell DNA is a key element of the life cycle and pathogenesis of many viruses. DNA viruses integrate by relying solely on cell machinery. In contrast, retroviruses possess a specialized endonuclease, also designated integrase (IN), which is essential for their replication (for a review, see \[[@B1]\]). After entering a target cell, reverse transcriptase (RT) converts genomic RNA into linear double-stranded cDNA with a copy of the viral long terminal repeat (LTR) at each end. Such linear genomic cDNA included in a preintegration complex (PIC) \[[@B2]-[@B9]\] can be used as a template for integration in vivo. Consequently, circular viral genomes that are detected in infected cells were considered until now as «dead-end» molecules, without essential function in the integration process and the viral cycle in general \[[@B8]\]. Integration mediated by the retrovirus IN occurs in two catalytic steps, referred to as 3\'-processing and strand transfer (or joining), respectively. Interestingly, the two steps appeared on distinct reactions catalyzed by virus IN in two different compartments in the infected cells. The strand transfer reaction joins viral DNA to cellular DNA in the cell nucleus. The viral cDNA ends are used to cut the target DNA in a staggered manner, which covalently links the viral 3\' ends to the 5\' phosphates of the cut (for reviews see \[[@B10],[@B11]\]. The 3\' hydroxyl groups at the LTR termini are the nucleophiles that promote DNA strand transfer \[[@B12]\]. Efficient strand transfer requires previous endonucleolysis of DNA that produces recessed 3\'hydroxyl ends \[[@B3],[@B5]\]. This occurs in the cytoplasm very soon after reverse transcription is completed \[[@B13]-[@B16]\], as viral genomes with blunt ends are extremely rare in the infected cytoplasm. Following these reactions, host cell enzymes likely repair the gap remaining between host and provirus DNA \[[@B17],[@B18]\]. IN recognizes and acts on short sequences (12 to 20 bp) called attachment (att) sites that are located at the LTRs \[[@B19]\]. Attsite includes the invariant CA dinucleotides, which are conserved in all retroviruses whereas the other nucleotides of the attsite, while not conserved in sequence, form an (imperfect) inverted repeat (IR) in all retroviruses, that has to be maintained intact for viral replication. Attmutagenesis experiments showed that mutation in one LTR precludes the processing of the other, demonstrating that activity of IN is concerted onto the two viral LTRs that are simultaneously cleaved in vivo\[[@B20]\]. The structural basis of such concerted processing of both extremities is unknown. More surprisingly, in the case of spumaretroviruses, a subfamily of retroviruses that share some features of DNA viruses \[[@B21]-[@B23]\], the IN may process only one of the two LTRs, although the attsites are present at the two LTRs. Based on the sequences of both 2-LTR DNA and integrated proviruses, an asymmetric processing of attsites has been proposed, in which IN may cleave the right, U5 end and may leave the left, U3 end intact \[[@B24],[@B25]\]. As the human spumaretrovirus (PFV) IN presents the usual features of other IN and carries out in vitroan endonucleolytic activity, as well as strand transfer and disintegrase activities \[[@B26],[@B27]\], the reason for this unusual mechanics is not understood at present. The attrecognition site of IN is present at least one time on all forms of viral DNA. In addition to linear and integrated forms, viral DNA is found in the infected cells as covalently closed DNA circles containing either one or two copies of the LTR, referred to as 1-LTR and 2-LTR circles, respectively \[[@B2]\]. Interestingly in the 2-LTR circles, the attsites are in a closed configuration due to the juxtaposition of the two LTRs and are included within a palindromic motif formed by the inverted repeat sequences in all retroviruses \[[@B28]-[@B31]\]. These 2-LTR circles are believed to result from a direct covalent joining of LTR ends at the so-called circle junction \[[@B32],[@B33]\]. Circularization is thought to occur by blunt-end ligation of the ends of linear proviral DNA, even no direct evidence has been provided until now to support this hypothesis. 2-LTR could be formed in part by the non-homologous end-joining (NHEJ) pathway of DNA recombination \[[@B34]\]. The two-LTR circle forms could, theoretically, serve as a potential precursor for the integrated provirus \[[@B4]\]. In spleen necrosis virus (SNV), Rous sarcoma virus (RSV), avian sarcoma virus (ASV) and avian leukosis virus (ALV), closed circular forms were initially proposed to act as substrates templates for integration \[[@B31],[@B32],[@B35]\], although these reports have not been substantiated. Although they are currently described in a productive infection as \"dead end\" molecules, precisely because of their incapacity to be directly integrated \[[@B8]\], intriguing observations invite some to reconsider their place. First, 2-LTR molecules were shown to be used as functional templates for the transcription machinery in HIV infected cells \[[@B36]-[@B39]\]. Second, 2-LTR viral DNA were detected in the cytoplasm of MLV and PFV infected cells at a very early time post infection, suggesting that they are not formed in the nucleus by an alternative fate to the integration way \[[@B40],[@B41]\]. In this context, we asked whether 2-LTR circles, rather than being substrate for integration nor \"dead end\" molecules, would be used as substrates for a preintegrative endonucleolytic activity of PFV IN. Such interrogation comes within the scope of the more global questioning concerning the pleiotropic actions of IN. Indeed, the mechanisms underlying the essential requirement for integration are still unclear in the retrovirus cycle. Why is integration critical for viral production when unintegrated DNA is abundant and competent for transcription \[[@B36]-[@B39],[@B42]-[@B45]\]? Is it possible that preintegrative function of IN explain its essential requirement rather than integration per se? Indeed, in addition to its roles in the establishment of the proviral integrated state, IN participates to other critical steps, such as reverse transcription \[[@B23],[@B46]-[@B52]\], nuclear import of HIV-1 preintegration complex (PICs) (for a review, see \[[@B53]\]), and the postintegration step of viral particle assembly (reviewed in \[[@B54]\]). Among the PIC constituents, IN is a logical and probable candidate for facilitating the efficient nuclear import of cDNA, since it has karyophilic properties \[[@B55]-[@B61]\]. Reflecting the pleiotropic activities of IN, non-replicative IN mutants of HIV were divided in two phenotypic classes depending on their defects \[[@B54]\]. The properties of IN mutants of PFV are less extensively described, and we suspected that PFV IN could play a key role in early preintegrative steps. In an attempt to better characterize the properties and substrates of the original IN of PFV, we analyzed both its in vivoproperties and in vitroactivity. We observed that the 2-LTR circles could serve as templates for the 3\' processing reaction of the IN. This allows spumaretrovirus to follow a symmetrical mechanism of integration and leads to reexamine the role of 2-LTR molecules and the importance of preintegrative function of IN. Results and discussion ====================== The mutations inPFV IN do not alter its karyophilic property ------------------------------------------------------------ Retroviral INs from oncoviruses \[[@B62],[@B63]\], lentiviruses \[[@B55],[@B59],[@B64],[@B65]\] and spumavirus \[[@B66]\] are karyophilic proteins, since they localize to cell nuclei in the absence of any other viral protein. Nuclear accumulation of INs may be a general feature of retroviruses. The intrinsic karyophilic property of retrovirus INs could be of high importance for the import of preintegration complex containing viral genomes in the nucleus (for a review, see \[[@B53]\]), where the transcription step occurs. The 39-kDa PFV virus IN \[[@B67]\] shares significant homologies with other retroviral INs including an amino-terminal HHCC zinc finger, a D, D~35~, E typical active site, and a DNA binding domain (Figure [1A](#F1){ref-type="fig"}) \[[@B68]-[@B70]\]. Three PFV-1 constructs with point mutations at conserved residues of IN were generated: (1) a His^42^Leu mutation within the HH-CC zinc finger domain that has been suggested to be involved in DNA binding (mutant M5, Figure [1A](#F1){ref-type="fig"}). ([2](#F2){ref-type="fig"}) an Ile^106^Thr mutation which had been described to abolish the in vitrointegration activity of the protein due essentially to a strong defect in strand transfer, the 3\'processing reaction being carried out with an efficacy of 35% compared to the WT IN (mutant M9) \[[@B24]\] and; (3) an Asp^160^Gly mutation (mutant M8) in the invariant catalytic triad which has been shown to impair PFV replication \[[@B24]\], likely due to a defective catalytic activity of the protein, as reported for HIV \[[@B69]\]. As expected, by using a vector encoding PFV-1 IN fused to the Flag epitope, we confirmed that PFV-1 WT IN shares the karyophilic properties as other retroviral IN. PFV-1 IN expressed in Hela-transfected cells was indeed confined to the cell nucleus as detected by immununofluorescence staining (figure [1B](#F1){ref-type="fig"}). We then evaluated the effects of the IN mutations onto the ability of IN to spontaneously localize into cell nucleus. None of the mutations we introduced did affect the nuclear accumulation of the protein (figure [1B](#F1){ref-type="fig"}) indicating that these mutations do not affect the ability of IN to be retained in the nucleus by tethering the chromosomes and/or the karyophilic character of IN. We conclude that the IN mutant phenotypes did not result from altered IN cellular localization. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### **The mutations in PFV-1 IN do not alter its karyophilic property.(A)*Schematic representation of foamy virus IN showing conserved motifs and residues between retroviral INs (IN-WT). Critical amino acid residues were mutated as indicated: M5 was mutated within the HH-CC zinc finger domain. In the M8 virus, Asp^160^in the invariant conserved catalytic triad, was changed to a glycine residue. Such a mutation has been shown to impair PFV-1 replication \[24\], likely due to a defective catalytic activity of the protein, as reported in HIV \[50\]. Another mutation was introduced at Ile^106^in the M9 mutant, since this mutation had been described to abolish the in vitrointegration activity of the protein \[24, 27\]. (B)Confocal microscopy analysis of WT PFV-1 IN and of mutants M5, M8, M9 IN. HeLa cells were transfected with plasmids expressing the WT or mutant IN, fused to the Flag epitope. After 36 hours, cells were fixed, permeabilized, and stained with anti-Flag-antibodies. Series of optical sections at 0.7-μm intervals were recorded. One representative medial section of the immunofluorescence staining is shown. ::: ![](1742-4690-2-31-1) ::: PFV harboring mutant IN genes are impaired in their replication at an early step -------------------------------------------------------------------------------- In order to study the impact of IN mutations in the viral context, the three mutations were introduced in the viral molecular clone PFV-1. We first analyzed overall infectivities in situations allowing the dissociation between early and late stages of viral replication. After transfection in FAB cells, transient viral production was found to be similar for both wild type parental and mutant viruses, as measured by reverse transcriptase activity in culture supernatants (Figure [2A](#F2){ref-type="fig"}). In these cells, only the late phase of virus replication is required to produce virions as transfection allows processes related to the synthesis of viral DNA to be bypassed. Certain point mutations in MLV or HIV IN were indeed described to impair the late replication steps such as virion assembly, production or maturation (viruses classified as class II IN mutant) \[[@B38],[@B52],[@B71]-[@B74]\]. This suggested that none of the mutations affected any of the late viral replicative steps, from viral transcription to the release of viral particles (Figure [2A](#F2){ref-type="fig"}). The impact of IN mutations on viral infectivity was further evaluated in a one-round infection assay based on indicator FAB cells \[[@B75]\]. This assay requires de novosynthesis of the viral Tas protein that trans-activates an integrated β-galactosidase reporter gene under the control of PFV LTR in the indicator cells. All mutations were found to affect viral replication in this assay, as well as in multiple-cycle assays in human glioblastoma U373-MG or Baby Hamster Kidney (BHK-21) cells (not shown). Since the DNA transfection experiments demonstrated that viral transcription itself was not affected by the IN mutations, the inability of these mutants to induce expression of the virus trans-activation dependent reporter gene (Figure [2B](#F2){ref-type="fig"}) indicates that their replication is impaired at an early step, between virus entry and transcription. Of importance, the M9 virus retained nearly 50% of the replication ability of its wild-type counterpart, which was striking in view of the reported inability of IN mutated at this site to integrate DNA mimicking PFV-1 LTR ends in vitro\[[@B27]\]. These data confirm that IN integrity is required for PFV replication. As for other retroviruses, it participates at an early pre-transcriptional stage of the replication cycle. Interestingly, it appeared that PFV can still replicate with an IN that has lost its in vitrostrand transfer activity. Similar paradoxical observations have already been reported for HIV \[[@B39],[@B51],[@B76]\]. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### *Impact of the IN mutations on viral replication.(A)The late replicative steps -- from viral transcription until the release of new virions in the cell supernatant- were studied by determining the reverse transcriptase (RT) activity in the culture supernatant of FAB cells transfected with equal quantities of the various proviral molecular clones. (B)*To study the early replicative steps, viral infectivity was determined in a single-cycle replication assay using FAB-indicator cells \[75\]. Cells were exposed to equal amounts of wild-type or IN-mutated viruses for 24 hours, as determined by RT-activity measurements in viral supernatants. Infections were assessed by measuring β-galactosidase activity in cell extracts. Data represent the mean of triplicate infections (+/- SD). ::: ![](1742-4690-2-31-2) ::: PFV-1 replication defective IN mutants display an abnormal pattern of viral DNA synthesis with an accumulation of 2-LTR circles ------------------------------------------------------------------------------------------------------------------------------- To further document the early steps at which the replication of defective mutant IN viruses is impaired, detailed kinetic analyzes of the different viral DNA forms were conducted in infected cells. The importance of IN in the virus replication might be very early since it participates to reverse transcription \[[@B23],[@B46]-[@B52]\], and may be even in close contact with the viral DNA all along its synthesis since it was shown to directly interact with the RT \[[@B46],[@B47]\]. U373-MG cells were exposed to equal amounts of viral particles. At various time-points after infection, DNA was extracted from infected cells and analysed for total viral DNA content by real-time PCR amplifying a gagregion. This PCR reaction amplifies all complete reverse transcription products. As shown in Figure [3A](#F3){ref-type="fig"}, all IN-defective viruses produced viral DNAs containing gagsequences indicating that their reverse transcription proceeded through both strand transfers. This DNA represented newly synthesized molecules since the RT-inhibitor AZT abolished DNA production (Figure [3A](#F3){ref-type="fig"}). However, the amount of viral DNA accumulating in cells infected with M5 and M8 mutant viruses was reduced, as compared to the DNA contents in wild-type virus-infected cells. After 24 hours of infection, viral DNA production increases in cells infected with wild-type or M9 virus (data not shown), likely reflecting new viral cycles which only take place under conditions of productive infection. These data indicate that M5 and M8 IN mutations affect reverse transcription, an IN mutant phenotype also observed in other retroviruses \[[@B38],[@B50],[@B51],[@B61]\]. ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Decreased viral DNA production by IN-defective viruses is concomitant with an abnormal accumulation of LTR-LTR junctions. Quantification of viral DNA synthesis was carried out by real-time PCR amplification of total DNA extracts from U373-MG infected cells (equal virion levels as measured by reverse transcriptase activity), collected 3, 6, 10, and 24 hours post-infection. An m.o.i. of 1 for the WT infection as determined by the FAB assay was used. Data are presented for 10^6^cells as measured by quantification of the nuclear β-globin gene and standard deviations representing variations between two quantifications of the same sample are given. To ensure that only freshly synthesized DNA, and not contaminating DNA contained in the viral particles input, was analyzed, all infections were performed in parallel control experiments under AZT treatment that inhibits viral neosynthesis. Representative kinetics from 4 independent experiments is presented. (A)Total viral DNA was detected using primers allowing amplification of the region of the PFV cDNA at the 5\' end of the gaggene \[40\]. (B)Viral DNA with 2-LTR junctions was measured using primers that cross the junction between the two LTRs as previously described \[40\]. (C)The abundance of 2-LTR molecules is expressed as the percentage of 2-LTR copies relative to the total viral DNA (gag) at each infection time-point. ::: ![](1742-4690-2-31-3) ::: Various DNA extracts were then analyzed for their content in molecules carrying 2-LTR junctions. As previously shown \[[@B40]\], viral DNA containing a LTR-LTR junction could be detected as early as 3 hours post-infection, and it continuously increased during viral replication (Figure [3B](#F3){ref-type="fig"}). The kinetics of production of 2-LTR species for IN mutant viruses paralleled that of the wild-type virus, indicating that their reverse transcription products were quite compatible with the formation of viral DNA containing LTR-LTR junctions. Using these quantitative data, we calculated the ratio of 2-LTR versus gagcontaining DNA in the same extracts. As for other retroviruses \[[@B77],[@B78]\], viral DNA species with an LTR-LTR junction represented a minority of the total viral DNA, from 0.6% early in the replicative cycle to a maximum of 9% 24-hour post-infection, in the case of wild-type virus (Figure [3C](#F3){ref-type="fig"}). Interestingly, for all IN-mutant viruses, we noticed a marked increase in the proportion of 2-LTR species as compared to the wild-type virus. The over-representation of 2-LTR molecules increased all along infection, reaching a remarkable 35% of total viral DNA in the case of the M8 mutant (Figure [3C](#F3){ref-type="fig"}). 2-LTR PCR does not allow to distinguish between 2-LTR circles and other molecules containing a LTR-LTR junction such as concatemeric linear or circular genomes. As the later molecules were not described, we assume that the 2-LTR junctions we quantified are indeed carried by circular genomes as in other retroviruses. However, such circles were difficult to detect during spumavirus infection by Southern blot \[[@B79]\], and further studies will be required to precisely answer this question. Our kinetic analyses revealed that the impaired global production of viral DNA due to inactivation of IN was associated with an abnormal accumulation of 2-LTR DNA species. Importantly, this overaccumulation of 2-LTR species has also been associated with IN-defective HIV viruses \[[@B50],[@B80]-[@B82]\]. To explain this observation, it is currently assumed that linear HIV DNA, representing the precursor of integration \[[@B3],[@B5]\], accumulates because it cannot be integrated and is rerouted into the circularization pathway producing 2-LTR molecules in the nucleus \[[@B29],[@B83]-[@B85]\]. However, 2-LTR circles are also detected in WT infected cells. In this case, 2-LTR formation was suggested to result from aberrant attsequences preventing their recognition by IN \[[@B83]\]. Moreover, since 2-LTR molecules have been detected both in the cytoplasm and the nucleus of PFV WT infected cells \[[@B40]\], as well as at very early time-points in cytoplasm of MLV infected cells \[[@B41]\], overproduction of 2-LTR DNA cannot simply be explained by such a rerouting of non-integrated viral DNA. Alternatively, PFV-1 IN might be directly involved in the processing and/or turnover of viral DNA containing LTR-LTR junctions explaining their accumulation when IN is defective. To address this hypothesis, we tested whether PFV-1 IN might use LTR-LTR circle as a substrate in vitro. PFV IN can specifically cleave the conserved palindromic sequence found at LTR-LTR junctions to generate 3\'-end processed LTRs ------------------------------------------------------------------------------------------------------------------------------- Sequences located at each end of linear proviral DNA, that are essential for recognition by IN, define the viral attachment (att) site. We analyzed sequences connecting the LTRs in the 2-LTR viral DNAs produced in infected cells. We found that these sequences bear a long palindrome composed of a central 8-base motif, flanked on each side by another 12-base palindrome separated from the central one by a 2-nucleotide insertion (Figure [4A](#F4){ref-type="fig"}). This 20 nucleotide-long bipartite palindrome was highly conserved in 36/40 of the sequenced clones as well as in U373-MG-infected cells, and corresponded to the juxtaposition of blunted 5\'-LTR and 3\'-LTR ends \[[@B24]\]. Palindromic sequences at the LTR-LTR junctions of the 2-LTR circles were also described in ASV and HIV-1 infected cells, each of them having its unique and specific palindrome (Figure [4D](#F4){ref-type="fig"}) \[[@B29],[@B31]\]. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### *PFV-1 IN specifically cleaves the conserved palindromic sequence found at LTR-LTR junctions.(A)The LTR-LTR junction in infected cells forms a 20 nucleotide-long bipartite palindrome. The LTR-LTR viral DNAs were PCR-amplified, cloned and sequenced following 5-days infection of BHK-21 cells with wild type virus. The vast majority of sequences (90%) were similar whereas approximately 10% had some divergence of the U3 junction. (B)The LTR-LTR junction is cleaved by recombinant PFV IN. This purified IN was shown to be functional by its 3\' processing activity on the blunt-ends of PFV LTR (see lanes 3 and 7, panel C) and its strand transfer activity (not shown). The U5 strand of an oligonucleotide spanning over the WT LTR-LTR palindromic junction was labelled at its 5\' extremity, annealed to its U3 complementary strand and incubated in the presence of PFV-1 IN. Products were resolved on a 15% denaturing polyacrylamide gel. A G+A chemical sequencing reaction was run alongside to identify the cleavage site. A specific cleavage immediately downstream of the conserved 5\'CA was obtained. The complementary strand was used for the U3 LTR-LTR junction. (C)*The cleavage of the LTR-LTR junction by IN is operating on the two strands of the palindrome leading to cohesive digestion fragments (lanes 2 and 6) indistinguishable from the products generated by the classical 3\' processing in vitroreaction on the blunt-ended LTRs (lanes 3 and 7). Cleavage products were obtained as for panel B. 3\' processing of either U5 or U3 blunt double-stranded LTRs was carried out under similar conditions and products were run alongside to confirm the structure of the palindrome cleavage products. Lanes 2, 3, 6, 7 and 10: 150 nM PFV-1 IN; Lanes 1, 4, 5, 8 and 9: 150 nM IN + 20 mM EDTA. EDTA was used to impair the cation-dependant activity of IN. This digestion is highly specific of the viral palindromic sequence since a mutated palindrome (which sequence is indicated panel D) was not cleaved by IN (lane 10). (D)A palindrome motif is required for cleavage by PFV-1 IN. Cleavage of oligonucleotides with mutations that disrupt the palindrome motif (mutated nucleotides different from the PFV wild-type sequence are marked with an asterisk), and with a scrambled sequence was assessed. Oligonucleotides carrying different palindromes chosen because they correspond to LTR-LTR junctions of other retroviruses such as HIV-1 and MLV were also tested as putative substrates of the PFV-1 IN. Assays were performed under the same conditions as in Fig. 3C. The ability of the IN to cleave the oligonucleotides onto their two strands is indicated in the right column. The vertical arrow indicates the cleavage site of the wild-type PFV LTR-LTR junction. These experiments were found reproducible in four independent assays. ::: ![](1742-4690-2-31-4) ::: Since inactivation of PFV IN led to the accumulation of 2-LTR viral DNA containing a palindrome reminiscent of enzymatic restriction sites, we tested whether this palindrome was a possible substrate for the endonuclease activity of IN, as proposed for avian retroviruses \[[@B86]\]. Recombinant PFV IN was produced in E. coli and purified on nickel column. The purified IN, able to catalyze integration in vitro, was incubated with a double stranded ^32^P-labeled oligonucleotide containing the palindrome. Reaction products were analyzed by electrophoresis in a polyacrylamide sequencing gel. A cleavage product appeared in the presence of IN confirming that IN harbors endonuclease activity. Moreover, the digestion fragment was found to be unique (Figure [4B](#F4){ref-type="fig"} and [4C](#F4){ref-type="fig"}, lanes 2 and 6) and corresponded to a cut between the two consecutive adenines in the middle of the palindrome, as determined by comigration of the sequencing reaction (Figure [4B](#F4){ref-type="fig"}, lane (G+A)). This digestion was dependent on IN activity as only the initial oligonucleotide was detected when IN was inactivated by EDTA treatment (Figure [4B](#F4){ref-type="fig"} and [4C](#F4){ref-type="fig"}, lanes 1 and 5). Moreover, this activity of PFV-1 IN was highly dependent on the target sequence since oligonucleotides carrying mutations that disrupt the palindromic character of the LTR-LTR junction (Figure [4C](#F4){ref-type="fig"} lane 10 and Figure [4D](#F4){ref-type="fig"}), and an irrelevant scrambled oligonucleotide (Figure [4D](#F4){ref-type="fig"}) did not undergo specific cleavage. Finally, PFV-1 IN did not cleave palindromes that are found at HIV-1 and MLV retroviral LTR-LTR junctions (Figure [4D](#F4){ref-type="fig"}). These data demonstrated that IN double-stranded DNA cleavage activity is restricted to the palindrome at the LTR-LTR junction found in corresponding infected cells and thus carries the same sequence specificity as already documented for the 3\'processing of LTR extremities \[[@B26]\]. Detailed analysis indicated that the digestion had operated on the two strands (U5- and U3-end labeling) of the oligonucleotide substrate generating cohesive ends with a 5\'-protuding AT (compare lanes 2 and 3, or 6 and 7, Figure [4C](#F4){ref-type="fig"}). Altogether, these data reveal a new substrate for IN endonuclease activity. This endonucleolytic activity is able to cleave specifically the palindromic sequence generated at the LTR-LTR junctions of viral DNA. The cleavage of 2-LTR circles into linear genomes justifies revisiting them as functional intermediates in the retroviral cycle. This is reinforced by recent observations showing their stability and contribution to the viral transcription \[[@B36],[@B37],[@B77],[@B78]\]. Interestingly, many DNA viruses replicate by using circular intermediates resembling the retroviral 2-LTR circles, and require the activity of a virally encoded endonuclease reminiscent of the IN. Identification of new IN activity should improve our understanding of the early steps of the retroviral replication cycle, allow screening of anti-retroviral drugs as well as design of new non-integrating retroviral vectors. That IN operates on 2-LTR molecules to produce linear DNA with each LTR end 3\'-processed avoids the need for asymmetrical integration in spumavirus ---------------------------------------------------------------------------------------------------------------------------------------------------- PFV IN was suggested to be unrelated to other retrovirus INs because of its apparent inactivity on the U3 LTR end of linear molecules, and the integration process of spumavirus was proposed to be asymmetrical \[[@B24],[@B25]\]. The asymmetric integration has been deduced from the sequences of both integrated and 2-LTR viral molecules (Figure [5A](#F5){ref-type="fig"}). The usual replication model supposes that the reverse transcription stage leads to linear DNA with blunt-ends. However, these ends are difficult to detect and sequence. Their structure had been previously deduced from the sequence at the LTR-LTR junctions. Indeed, the latter are themselves supposed to be formed by the intramolecular ligation between the two blunt-ends of linear DNA by an unidentified mechanism. As only two nucleotides are lost during integration, the PFV integration process was proposed to be unusual (figure [5A](#F5){ref-type="fig"}). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### *Asymmetric integration is not required to understand the sequences of integrated and 2-LTR molecules observed in PFV-1 infected cells.(A)The asymmetric integration in PFV-1 virus was proposed to account for the sequences of both integrated and 2-LTR viral molecules as observed in the infected cells \[24, 25\]. This unusual proposed integration was able to solve the problematic lost of only 2 nucleotides between U5 extremity of the integrated molecules and the putative U5 free end, whereas the U3 end remains unchanged. This assertion was based on the following model: the linear substrate for integration is produced by two 3\'-processing reactions at each end of a blunt molecule. Of note, such blunt linear molecules have never been detected in infected cells and their structure was deduced from the observed 2-LTR circles sequences. Such deduction is based on the idea that 2-LTR circles result from the ligation of blunt linear DNA. However the actors of this reaction are still unknown. (B)*We propose a revised version where the PFV-1 integration remains classical. A single reaction of PFV-1 IN onto the palindrome at the LTR-LTR circle junction can generate a linear DNA with its two 3\' ends processed. The subsequent integration then eliminates the two nucleotides that are lost between the observed sequences of the LTR-LTR junction and the integrated provirus. ::: ![](1742-4690-2-31-5) ::: In light of our observation that 2-LTR molecules are possible substrates for PFV-1 IN (Figure [4](#F4){ref-type="fig"}), the 3\'-processing of both ends of the linear DNA might be generated in a single reaction that produces the two 3\'-processed ends simultaneously (Figure [5B](#F5){ref-type="fig"}). Such concerted processing might explain the influence of one LTR on the processing of the other, as observed for HIV-1 \[[@B20]\]. The subsequent integration of such processed extremities would eliminate the two nucleotides that are lost between the LTR-LTR junction and the integrated provirus. No asymmetric integration is required to account for the previous observations \[[@B24],[@B25]\]. This mechanic, when generalized to other retroviruses carrying a different palindrome at the LTR-LTR junction, would result during integration in the loss of the number of nucleotides comprised between the conserved CA. In support of our symmetrical integration model, Pahl and Flügel \[[@B26]\] previously reported an efficient 3\'-processing activity of PFV IN on LTR containing the two additional nucleotides AT. The substrate of concerted processing corresponds to the extended substrate they tested. We confirmed the 3\'-processing cleavage of the extended U3 LTR carrying an additional AT (Figure [4C](#F4){ref-type="fig"}), as well as the fact that the 3\'-processing does not occur onto the shorter U3 LTR lacking these nucleotides (not shown). Integration depends on preintegrative IN activity ------------------------------------------------- Integration was reported to be a very rare event in spumaviruses \[[@B87],[@B88]\], except in chronically infected cell situations \[[@B89]\]. To document this point in our conditions, we quantified the integration events for PFV-1 WT and IN mutants. To this end, we designed a highly sensitive quantitative real-time RACE-PCR reaction, amplifying Alu-LTR junctions between the cell genome and integrated proviruses (detecting 25 integrated proviruses per 50 000 cells, Figure [6A](#F6){ref-type="fig"}). U373-MG cells were infected with equivalent amounts of viral particles as measured by RT activity and the quantity of integrated viral molecules was analyzed 24 hours later, a time-point at which the first round of infection is achieved. As shown in Figure [6A](#F6){ref-type="fig"}, and as expected \[[@B87],[@B88]\], only a small fraction of total wild-type PFV DNA was integrated (range of 0.9--2.1%). The M8 and M9 mutant INs used in our study failed to integrate oligonucleotides mimicking the PFV LTR DNA ends into a target plasmid in vitro\[[@B26]\]. We therefore assessed the ability of viruses carrying the same IN mutations to integrate in vivo. We could detect integrated DNA after infection with viruses carrying inactive INs (Figure [6B](#F6){ref-type="fig"} upper panel). However, with the exception of the semi-replicative M9 virus, IN mutants yielded significantly fewer integrated proviruses than the wild-type (Figure [6B](#F6){ref-type="fig"}). Similar observations have been reported in cells infected with IN-defective HIV and the presence of integrated proviruses was attributed to integrase-independent integration events depending on cell enzymes \[[@B81]\]. Another explanation could rely on the fact that IN mutants produced less linear DNA as a substrate for integration. The altered viral DNA production is likely reflected by the reduced amounts of total viral DNA quantified in the same extracts (Figure [6B](#F6){ref-type="fig"} lower panel). We compared integration ratios with and without functional IN by normalizing integrated proviruses values with the total number of viral DNA copies present in infected cells. Strikingly, the percentage of integrated DNA was not modified by the presence of a defective IN (Figure [6C](#F6){ref-type="fig"}). Thus, the level of integrated provirus depends on the global viral DNA pool available in the infected cells. And such global viral DNA content itself depends on the early activity of the viral IN as shown above. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### *Integration of IN-defective viruses.(A)A quantitative assay based on a real-time RACE-PCR reaction was designed, amplifying Alu-LTR junctions between the cell genome and integrated proviruses twenty-four hours post-infection. PCR amplifications of existing Alu-PFV-1 LTR junctions were subjected to a second quantitative round of real time PCR with PFV-1 LTR-specific primers. Fluorogenic hybridization probes were used to quantify the amplification products. Infected cells with known copy numbers of integrated proviruses were used as quantification standards. The assay is highly sensitive since it allows detecting 25 proviruses copies in 50,000 human cells. Control reactions are detailed in the Material and methods section. (B)*Detection of integrated viral DNA following infection of IN-mutated viruses. Quantitation of viral DNA accumulated in PFV-1 infected cells was carried out by real-time PCR of total DNA extracts from U373-MG infected cells (m.o.i. of 1) collected at the completion of the first viral replication cycle, 24 hours post-infection. Total viral DNA (gagquantifications) and integrated proviruses were quantified in duplicate using real-time PCRs. Data obtained in one representative infection from four independent experiments are expressed as integrated DNA copies per million cells (logarithmic scale) as determined by a human β-globin quantification in cell extracts (\"Integrated provirus\" panel). Total DNA copies per million cells (logarithmic scale) present in the same extracts are presented in the lower panel. Standard deviations representing variations between two quantifications of the same sample are given. (C)Integration efficiency in PFV-1 infected cells. Integration efficiency was determined by normalizing the number of integrated proviruses (mean of duplicates) with the total number of viral DNA molecules (mean of duplicates) present in the same extract. Raw LightCycler data from four independent experiments are presented in the upper table. Mean of integration efficiencies from these four experiments are figured in the lower histogram. ::: ![](1742-4690-2-31-6) ::: Role of IN in PFV retrovirus replication cycle ---------------------------------------------- We conclude from these experiments that PFV IN displays a specific activity on the 2-LTR circles, which may constitute a substrate for the 3\'processing reaction in vivo. This action of IN generates linear DNA that might be then integrated in the cell genome following a classical symmetrical integration process. The fact that early actions of IN may influence later steps of replication, including integration, certainly participates in the pleiotropic effects of IN mutations. Finally, IN seems to be essential not because of its participation to the integration per sebut for its upstream activities able to influence integration efficacy. Our findings that a loss of endonuclease IN activity results in both LTR-LTR accumulation and an associated reduction in viral DNA production leads us to propose a direct role for retroviral integrase in the production of viral DNA. Thus, a modified replication model is presented in Fig. [7B](#F7){ref-type="fig"}. It is accepted that the encounter between viral DNA and IN occurs very shortly after viral DNA synthesis, since cytoplasmic viral DNA is mostly found as linear molecules with 3\' processed ends resulting from IN endonucleolytic action in the cytoplasm \[[@B13]-[@B15]\]. In our model, DNA molecules containing LTR-LTR junction would be generated during the reverse transcription process and cleaved rapidly by the IN, leading to the production of linear DNA harboring 3\'-processed ends. This would account for the rarity of linear DNA with blunt ends in the cytoplasm of infected cell, as well as for the presence of 2-LTR circles in the cytoplasm of retrovirus infected cells at early times post infection \[[@B40],[@B41]\]. Additionally, it would explain the data from attsite mutagenesis experiments showing that mutation of one LTR precludes the processing of the other LTR \[[@B20]\]. These results were initially interpreted to represent a concerted activity of IN on the two viral LTRs ends that must be simultaneously cleaved in infected cells. In view of our results, these data might be understood as resulting from the endonucleolytic activity of IN on palindromic LTR-LTR junctions. Such processed DNA could then undergo integration. In this interpretation, a unique endonucleolytic action of IN at an early step would explain many of the phenotypes associated with IN mutations, including the increasing abundance of 2-LTR molecules at the expense of linear and integrated DNA in IN-defective viruses. It underlines that in vivointegration is performed in two steps that are uncoupled both in time and in space, ie3\' processing in the cytoplasm and integration per sein the nucleus. It also illustrates why and how certain in vitrointegration-defective viruses such as our M9 mutant or HIV mutants \[[@B39],[@B51],[@B76]\] are still replicative. The IN activity demonstrated in this report allows processing the circles -- currently considered as dead-end molecules- into the replication pathway. Additional support to this conclusion is present in the HIV literature where episomal circular DNA were shown to turn over by degradation rather than through death or tissue redistribution of the infected cell itself in HIV-1 infected individuals \[[@B42]\]. Finally, our data imply that circular retroviral genomes are fully functional replication intermediates, first as substrates for transcription and second as precursors of linear unintegrated DNA. ::: {#F7 .fig} Figure 7 ::: {.caption} ###### *Role of IN in retrovirus replication cycle.(A)*Classical model of early steps in retrovirus replication. IN plays a role in the 3\' processing as well as in the integration itself, these two steps being separated both in time and in space. Following synthesis of linear blunt-ended DNA in the cytoplasm (step 1 in Fig. 7A), IN cleaves their 3\' termini, thus eliminating the terminal two bases from each 3\'end (step 2). The resulting recessed 3\'OH groups provide the attachment sites of the provirus to host DNA, an attachment which is performed only after import of 3\'processed DNA into the nucleus where the final step of the integration process occurs (step 3). Circular DNA carrying LTR-LTR junctions are reportedly formed from linear DNA viathe action of cellular ligases (step 4). The circularization is considered to be an alternate fate of linear DNA that has not integrated, and may indirectly explain why DNA bearing LTR-LTR junctions accumulates to high levels in cells harboring integration-defective viruses. This classical model considers that functions of IN in processes other than integration are secondary. (B)Alternate retrovirus replication model. IN cleaves the LTR-LTR junction generated at the reverse transcription step (step 1) to produce 3\'end-processed linear DNA (step 2). This specific activity of the IN explains the pleiotropic effects of this protein and the phenotypes associated with its mutagenesis. First, since linear DNA is the direct product of a reaction that is catalyzed by IN, its levels would decrease under IN-defective conditions. Moreover if LTR-LTR junction molecules indeed constitute the substrate for IN, their amount would increase as a direct consequence of defective IN. Second, decreased levels of integrated proviruses would be an indirect result of the decreased pool of 3\'processed IN-catalyzed linear DNA molecules that are available for integration (step 3). In this model, 2-LTR molecules are a replication-intermediate. Low levels of these molecules would be due to their rapid processing by IN in the wild-type infections. Rapid processing might also explain the presence of linear molecules with 3\' processed ends in the cell cytoplasm during diverse retroviral infections, even though no blunt-ended linear molecules can be recovered from infected cells. Thus, apart from participating in retroviral DNA integration per se, IN would act upstream by controlling linear DNA production. This function of IN, as included in the modified replication model presented here, provides a parsimonious interpretation of the pleiotropic effects observed in cells infected with IN mutants. ::: ![](1742-4690-2-31-7) ::: Although the consensus sequences in the C ter region of IN may differ between the lentiviruses and the nonlentiviruses, the carboxyterminal region of IN is well conserved in all retroviruses \[[@B80]\], and further studies are now required to evaluate whether the revised replication model we propose here, applies to all retroviruses. The fact that the typical phenotype associated with a defective IN, either due to mutations or inhibitors, resulting in reduced DNA synthesis but a persistence of integration and an accumulation of 2-LTR molecules, is commonly observed among retroviruses \[[@B73],[@B82],[@B90]\], argues in favour of a conserved IN function. Such an early participation of IN sheds new light on reports showing both that viral transcription occurs from nonintegrated HIV DNA \[[@B38],[@B44],[@B45],[@B91]\], and that the most prevalent form of HIV DNA during the asymptomatic phase of infection is full-length unintegrated DNA \[[@B42],[@B92]\]. Whereas IN activity is clearly required, formation of integrated provirus as an obligate step of retroviral replication now needs to be reconsidered. On the other hand, early preintegrative activities of IN are of capital importance. This provides new answers to the puzzling question of why is integration essential to retrovirus replication, when many authors have shown that unintegrated genomes are abundant and expressed \[[@B36]-[@B39],[@B42]-[@B45],[@B93]\]. Our proposal is simply: integrase is essential, integration is not; and IN is required given its critical preintegrative influence on genomic DNA production in vivo, as we precisely measured here. Given the above, retroviruses better fit the classical schemes of distinct lytic and lysogenic phases exemplified by the lambda phage: integration (lysogeny) contributes to viral persistence and pathogenesis, but it is not essential for acute viral production (lytic cycle). Finally, a fascinating evolutionary conservation appears between retroviruses and DNA viruses (such as poxviruses). All use circular DNA intermediates and a specialized endonuclease activity for genome production. Methods ======= Cells, virus infections and reagents ------------------------------------ BHK-21, FAB, HeLa and U373-MG cells were cultivated in DMEM with 10% foetal calf serum, 1 μg per ml of streptomycine-streptavidine. For FAB indicator cells, 1 μg per ml of G418 (Sigma) was added. PFV-1 virus stocks were prepared by transfecting BHK-21 cells with the PFV-1 molecular WT and mutant clones using the calcium phosphate method. Cells were infected by WT and mutant viruses with same amounts of viral particles, as evaluated by a reverse transcription assay. The culture medium was changed two hours post-infection with fresh medium. Cell free virus stocks were titrated on FAB cells \[[@B75]\]. In some experiments, infected cells were treated with 3\'-azido-3\'-deoxythymidine (AZT, Sigma) at 100 μM. DNA quantifications by real time PCR ------------------------------------ Total DNAs were extracted from 10^6^cells using the DNA Blood Mini kit (Qiagen) in a final volume of 200 μl and analysed by real time PCR as described previously \[[@B40]\]. Integrated viral DNA was also quantified by two rounds of PCR \[[@B94]\]. The first one amplifies integrated DNA using primers ALU1 (5\'-CCT CAG CCT CCC GAG TAG CTG GGA-3\'), ALU2 (5\'-CTG TAA TCC CAG CAC TTT GGG AGG C-3\'), and λ TSPA (5\'-ATG CCA CGT AAG CGA AAC TTA GTA TAA TCA TTT CCG CTT TCG-3\'). Sequence in bold represents a sequence in the lambda phage, which is unknown in all mammals\' databanks. The other part of the sequence of λ TSPA primer can hybridize in PFV LTR. Amplification was performed in a 20 μl reaction volume containing 1X Light Cycler Fast Start DNA Hybridation probes, 3.5 mM MgCL~2~, 300 nM of primer ALU1, ALU2 and 10 nM of primer λ TSPA. The same mix, containing only primer λ TSPA, was prepared. DNA from U373-MG chronically infected cells was used as a standard for integrated copies. All reactions were further diluted in a final volume of 200 μl of water. 2 μl over 200 μl was used for the second PCR. This amplification was performed with 300 nM of each primers Nested R (5\'-GAA ACT AGG GAA AAC TAG G-3\'), lambdaT (5\'-ATG CCA CGT AAG CGA AAC T-3\') and 100 nM of each hybridation probes SpuFL (5\'-CAC TCT CGA CGC AGC GAG TAG TGA A X-3\') and SpuLC (5\'-GCC TCC CGT ACA ATC TAG AAA CTA TCC T p-3\'). This assay is quite specific of integrated provirus only, as attested by performing the following control reactions: -- a carry-over control in which all primers were omitted in the first PCR, data obtained indicated always that the second-round amplification of nonpreamplified viral DNA is efficiently prevented; -a parallel reaction with the Aluprimers in the first-round PCR, in order to calculate the linear amplifications resulting from all the viral DNA species. The copy number due to the linear amplification was systematically subtracted from the signal obtained in the presence of Aluprimer. We evaluated that this interfering amplification never exceeded 6.7 % of the global amplification. Quantifications were performed with the LightCycler software Version 3.5 according to manufacturer\'s instructions. Virion-associated RT assays --------------------------- 48 hours post transfection viral supernatants were collected. 10 μl of viral supernatant was incubated with 20 μl of reaction buffer (Tris pH 8 50 mM -- KCl 75 mM -- Dithiotreitol 2 mM -- rA/dT 25 μg/ml -- NP40 0,05% -- MnCl~2~5 mM -- dTTP α-^32^P 20 μCi/ml). The reaction mixtures were incubated at 37°C for 90 min. 10 μl of the reaction was spotted onto DE81 filter and allowed to dry. The filters were washed four times with 2xSSC (1xSSC is 0.15 M NaCl plus 0.015 M sodium citrate) for 5 min each, followed by two washes with 95% ethanol. The filters were then dried and counting by scintillation fluid. Construction of Flag-PFV IN mutants and their cell localisation by immunofluorescence staining ---------------------------------------------------------------------------------------------- To express the INs in the absence of other viral products, we used the pFlag expression vector \[[@B95]\]; in which we inserted the PFV-1 IN sequence under the control of the simian virus 40 promoter. The IN fragment was amplified by PCR with the following primers, which created a BamH1and an XhoIrestriction site at the 5\' and 3\' ends, respectively, of the IN sequence: 5\'-GGA TCC TAC ATA TTT TTT AGA AGA TGG C-3\'; and 5\'-CTC GAG TTA TTC ATT TTT TTC CAA TGA TCC-3\'. The resulting PCR fragment was digested with BamHIand XhoIand ligated into the corresponding cloning sites of pSG-Flag \[[@B95]\], in the plasmid called pSG-FlagIN PFV. The pSG-FlagIN PFV expression vector was used for the mutagenesis, with the Quick Change mutagenesis kit (Stratagene), and the primers: 5\'-CAA TTT GGC TCT CAC AGG ACG TGA AGC C-3\' and 5\'-GGC TTC ACG TCC TGT GAG AGC CAA ATT G-3\' for the M5 mutant; 5\'-ATT CAC TCT GGT CAA GGT GCA GC-3\' and 5\'-GCT GCA CCT TGA CCA GAG TGA AT-3\' for the M8 mutant; and 5\'-GGC AAA GGG CCA GTA TAG TCA AT-3\' and 5\'-ATT GAC TAT ACT GGC CCT TTG CC-3\' for the M9 mutant. HeLa cells (2 × 10^5^) were spread on glass coverslips in 24-well plates, transfected with 1 μg of the corresponding plasmids, and stained for immunofluorescence 36 hours later. Cells were fixed in 3.7% formaldehyde-PBS for 20 min, washed three times in PBS, and incubated for 10 min in 50 mM NH~4~Cl to quench free aldehydes. Cells were washed three times in PBS and incubated in a permeabilization buffer (0.05% saponin, 0.01% Triton X-100, 2% bovine serum albumin, PBS) for 15 min and incubated 1 h with the first MAb (M2 anti-Flag MAb at 7.5 μg/ml) in permeabilization buffer. Cells were washed three times in permeabilization buffer and incubated with Cy3-conjugated anti-mouse MAbs (Amersham) at a final dilution of 1:200. Cells were washed three times in permeabilization buffer and once in PBS and mounted in 133 mg of Mowiol (Hoechst) per ml-33% glycerol-133 mM Tris HCl (pH 8.5). Confocal microscopy was performed and optical sections were recorded. One representative medial section was mounted by using Adobe Photoshop software. Construction of PFV proviruses ------------------------------ We inserted a DNA fragment containing the PFV-1 IN sequence into a Litmus 38 plasmid, in which a PacIsite had been added. The viral fragment was amplified by PCR with the following primers: 5\'-GGA TCC TAC ATA TTT TTT AGA AGA TGG C-3\' and 5\'-CTC GAG TTA TTC ATT TTT TTC CAA TGA TCC-3\', and cloned after a BspEI-PacIdigestion into the modified Litmus. This plasmid containing the WT IN was used for the mutagenesis, with the Quick Change mutagenesis kit and the primers used above for the expression IN vector mutagenesis. After the mutagenesis, the PacI-BspEIdigestion fragments from the mutated Litmus vectors were substituted for the corresponding sequence of the PFV-1 full-length clone. All constructions were confirmed by DNA sequencing of the entire PCR-amplified fragment. 2 LTR junction sequence analysis -------------------------------- Total DNA from acutely BHK-21 infected cells of two independent infections were extracted and analyzed by a PCR amplification specific for the LTR-LTR junction from the 2-LTR circles, using the following primers: R, 5\'-TAC GAG ACT CTC CAG GTT TG-3\'; and U3, 5\'-CGA CGC AGC GAG TAG TGA AG-3\' and the Pfu polymerase (Stratagene) \[[@B40]\]. PCR products were cloned in a pSK+ plasmid (PCR-Script cloning kit, Stratagene). 50 independent cloned were sequenced. Construction and purification of PFV recombinant IN --------------------------------------------------- Histidine-tagged PFV-1 IN, corresponding to aminoacids 752-1143 of the Pol polyprotein, was expressed and purified by nickel affinity. The preparation and purification of recombinant PFV-1 IN protein were performed as described for HIV IN \[[@B96]\]. To obtain wild type IN protein, plasmid pET15b (Novagen) was digested with NdeIand BamHI. The DNA fragment containing the PFV IN was obtained from pHSRV clone C55 by PCR using the PfuDNA polymerase (Stratagene). The sequence of the primers used to amplify the fragment were 5\'-ACA TAT GTG TAA TAC CAA AAA ACC AAA CCT GG-3\' and 5\'-AGG ATC CTT ACT CGA GTT CAT TTT TTT C-3\'. PCR amplifications were done at 92°C for 1 min, 55°C for 45 s, and at 72°C for 90 s; the cycle was repeated 28 times. The resulting PCR fragment were digested with NdeIand BamHIand ligated into the corresponding cloning sites of pET15b. Plasmid pET15bIN was used to express the His-tagged IN in E. coliBL21 (DE3) cells. 500 ml of BL21 (DE3) pET15bIN cells was grown at 37°C in LB medium (supplemented with 50 mg/ml ampicilin) to an A^600^of 0.6--0.8. To induce IN protein expression, isopropyl-1-thio-β-D-galactopyranoside was added to a final concentration of 1 mM; bacteria were grown for another 4 hours and harvested by low speed centrifugation. The pellet was resuspended in 24 ml of 50 mM Tris-HCl, pH8, 1 M NaCl, 4 mM β-mercaptoethanol (buffer A). Cells were lysed with French Press and centrifugated at 14,000 rpm and 4°C for 30 min to remove cells debris The supernatant was filtered (0.45 μm) and incubated over night with Ni-NTA agarose beads (Qiagen). The beads were washed with 10 volumes of buffer A. Then, IN was purified under native conditions according to manufacturer\'s instructions using batch procedure. His-tagged IN was eluted with buffer A supplemented with 50 μM ZnSO~4~and 1 M imidazole. The IN concentration was adjusted to 0.1 mg/ml in buffer A and dialysed over night against 20 mM Tris-HCl, pH 8, 1 M NaCl, and 4 mM β-mercaptoethanol. Fractions were aliquoted and rapidly frozen at -80°C. Nucleic acid substrates ----------------------- All oligonuleotides U5B (5\'-CCT TAG GAT AAT CAA TAT ACA AAA TTC CAT GAC AAT-3\'), (U5A 5\'-ATT GTC ATG GAA TTT TGT ATA TTG ATT ATC CTA AGG-3\'), U3 B (5\'-ATT GTG GTG GAA TGC CAC TAG AAA T-3\'), U3A (5\'-ATT TCT AGT GGC ATT CCA CCA CAA T-3\'), LTR-LTRB (5\'-CCT TAG GAT AAT CAA TAT ACA AAA TTC CAT GAC AAT TGT GGT GGA ATG CCA CTA GAA AT-3\') and LTR-LTRA (5\'-ATT TCT AGT GGC ATT CCA CCA CAA TTG TCA TGG AAT TTT GTA TAT TGA TTA TCC TAA GG-3\') were purchased from Eurogentec and further purified on an 15% denaturing acrylamide/urea gel. 100 pmol of U5 B, U3 B and LTR-LTR B were radiolabeled using T4 polynucleotide kinase and 50 μCi of \[γ-^32^P\]ATP (3000 Ci/mmol) during 2 hours at 37°C. The T4 kinase was heat inactivated, and unincorporated nucleotides were removed using a Sephadex G-10 column (Pharmacia). NaCl was added to a final concentration of 100 mM and complementary unlabeled strand was added to either U5 B, U3 B or LTR-LTR B. The mixture was heated to 90°C for 3 min, and the DNA was annealed by slow cooling. LTR processing, LTR-LTR junction cleavage ----------------------------------------- Processing and LTR-LTR cleavage were performed in buffer containing 50 mM Hepes, 5 mM DTT and 10 mM MgCl~2~. 150 nM of PFV-1 IN was used for reaction. The reaction was initiated by addition of substrate DNA, and the mixture was incubated 2 hours at 37°C and stopped by phenol/chloroform extraction. DNA products were precipitated with ethanol, dissolved in TE containing 7 M urea and electrophoresed on a 15% denaturing acrylamide/urea gel. Gels were analysed using a STORM Molecular Dynamics phosphorimager. List of abbreviations ===================== Att, attachment site HIV, human immunodeficiency virus IN, integrase LTR, long terminal repeat PFV, primate foamy virus PIC, preintegration complex RT, reverse transcriptase WT, wild-type Authors\' contributions ======================= OD carried out all the experiments concerning the phenotype analysis of the viruses in the cell context including constructions, viral kinetics and real-time PCR, and participated to the analysis of the data. CP contributed to the design and coordination of the study, supervised the experimental work, participated in the analysis and interpretation of the data, and drafted figures and the manuscript. HL participated in the acquisition of the biochemical datas and in their interpretation. GM contributed to the acquisition of biochemical datas. JFM contributed and supervised biochemical analysis of integrase in vitro*. PS conceived the original ideas, designed and coordinated the study, and took part in writing the manuscript. All authors read and approved the final manuscript. Acknowledgements ================ We warmly acknowledge Olivier Neyrolles, Sebastien Petit and the OCU for stimulating remarks and daily help. We are grateful to William Jacques Speare and Alexandre Matet for their corrections and for continued enthusiastic discussion regarding this research. We also thank Marc Alizon, Olivier Danos and Olivier Schwartz for stimulating and thoughtful comments, and constructive criticisms on the manuscript. We finally thank Naomi Taylor and Marc Sitbon for insightful discussions concerning the retrovirus replication models, as well as for their meticulous reading of our original manuscript.	PubMed Central	2024-06-05T03:55:59.706415	2005-5-18	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180852/", "journal": "Retrovirology. 2005 May 18; 2:31", "authors": [ { "first": "Olivier", "last": "Delelis" }, { "first": "Caroline", "last": "Petit" }, { "first": "Herve", "last": "Leh" }, { "first": "Gladys", "last": "Mbemba" }, { "first": "Jean-François", "last": "Mouscadet" }, { "first": "Pierre", "last": "Sonigo" } ] }
PMC1180853	Background ========== From a genome of only 9000 nt in length, HIV-1 directs the synthesis of 15 proteins essential for its replication and dissemination (for review see ref. \[[@B1]\]). In order to generate mRNAs required for the synthesis of these proteins, HIV-1 uses the cellular splicing machinery. Through alternative splicing of its primary RNA transcript containing 4 donor sites (D1, D2, D3 and D4) and 8 acceptor sites (A1, A2, A3, A4a, A4b, A4c, A5 and A7), more than 30 different mRNAs are generated and divided into three classes of 2 kb, 4 kb and 9 kb in length (Figure [1](#F1){ref-type="fig"}) \[[@B2]\]. The 2 kb mRNAs are fully spliced and principally encode the regulatory proteins Tat and Rev and accessory proteins Nef and Vpr. The single-spliced 4 kb RNAs are bicistronic and code for the Env glycoproteins and viral factor Vpu, and the unspliced 9 kb RNA serves both as mRNAs for the Gag and Gag-Pol polyproteins as well as pre-genomic RNA for Gag assembly. Rev is crucial because it directs the export of the unspliced and single-spliced mRNAs from the nucleus to the cytoplasm that permits their translation \[[@B3],[@B4]\]. A fine tuning of splicing is then critical to ensure the balance between spliced versus unspliced viral RNAs. ::: {#F1 .fig} Figure 1 ::: {.caption} ###### HIV-1 splicing pattern. Schematic representation of HIV-1 proviral DNA. Open boxes represent the open reading frames encoding the viral proteins. Black boxes represent exons generated by combination of donor sites (D1 to D4) and acceptor sites (A1 to A7). The viral translation initiator codons are indicated by AUG. ::: ![](1742-4690-2-33-1) ::: HIV-1 splicing regulation relies on the presence of (i) suboptimal splice sites \[[@B5],[@B6]\], (ii) exonic and intronic cis-acting elements \[[@B7]-[@B15]\] and (iii) trans-acting factors (generally hnRNPs and SR proteins) that mediate their effects by binding these elements \[[@B16]-[@B19]\]. SR proteins belong to a conserved family of structurally and functionally related phosphoproteins (for review, ref. \[[@B20]\]). These proteins participate in constitutive splicing by causing stabilizing interactions with components of the splicing machinery and are able to influence the choice of splicing sites in alternative splicing (for review see ref. \[[@B20]\]). The high level of conservation of the splicing pattern in different HIV expressing cells suggests that splicing regulation is critical for efficient virus replication \[[@B2],[@B21],[@B22]\]. Because SR proteins ASF/SF2, SC35, 9G8 and SRp40 have been shown to cause an imbalance in the HIV-1 splicing pattern in vitroand ex vivo\[[@B19],[@B23]-[@B26]\], we investigated the impact of SR protein over-expression on virus production and infectivity in a human cell line expressing infectious HIV-1. In the present study we show that overexpression of one of the three SR proteins ASF/SF2, SC35 and 9G8 together with HIV-1 strongly affected the full length viral RNA splicing pattern, notably resulting in a strong reduction of the genomic RNA and Env mRNA levels. As a consequence, only small amounts of viral particles were produced which, however, retained part of their infectivity. Results ======= SR proteins alter the splicing pattern of HIV-1 ----------------------------------------------- Human cells (293T) were co-transfected by the calcium phosphate precipitation method with 10 μg of HIV-1 pNL4-3 \[[@B27]\] and 10 μg of irrelevant plasmid pCLacZ (control) or 5--10 μg of one of the SR protein-expression vectors, pXJ41-ASF, pXJ42-PR264 and pXJ42-9G8, encoding respectively ASF/SF2, SC35 and 9G8 proteins \[[@B26],[@B28]\]. Expression of HIV-1 and SR proteins in co-transfected cells was verified by immunoblotting assays (data not shown). We first performed RT-PCR in conditions previously described \[[@B2],[@B29]\] to verify that SR proteins modified HIV-1 splicing pattern as reported elsewhere \[[@B26]\]. Multiple-spliced 2 kb mRNAs isolated from ASF/SF2 over-expressing cells showed that Vpr1, Tat2 and Tat3 were strongly increased as compared with the control (Figures [1](#F1){ref-type="fig"}, [2A](#F2){ref-type="fig"}). These observations were confirmed by the analysis of the 4 kb mRNAs where Tat6 and Vpr3 mRNAs became the most represented in these conditions probably at the expense of the Env mRNA which proved to accumulate at a low level (Figure [2B](#F2){ref-type="fig"}). SC35 and 9G8 overexpression led to similar splicing patterns where Tat1 and Tat5 mRNAs were the most abundant spliced isoforms (Figures [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). In the case of SC35, splicing was almost completely driven towards Tat1 production. Because Tat2 and Tat6 required splicing at site A1 and Vpr1, Vpr3 and Tat3 mRNAs at site A2, we concluded that ASF/SF2 participated in a positive regulation of splicing at sites A1 and A2, while SC35 and 9G8 preferentially enhanced splicing at site A3 necessary for Tat mRNA synthesis (Figure [1](#F1){ref-type="fig"}). These results are in agreement with those obtained in HeLa cells using a truncated non-infectious HIV-1 DNA construct \[[@B26]\] and showed that SR proteins profoundly changed the HIV-1 splicing pattern. However the effects observed in the present experimental conditions were stronger than with the incomplete HIV-1 DNA construct \[[@B26]\]. ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Regulation of HIV-1 alternative splicing by SR proteins. Analysis of 2 kb (A) and 4 kb (B) mRNAs was performed by RT-PCR using 10 μg of total cellular RNA extracted from 293T cells transfected by HIV-1 pNL4.3 only (lane 1) or together with one SR plasmid (lanes 2--4). Viral mRNAs were identified according to the nomenclature of Purcell and Martin \[2\]. ::: ![](1742-4690-2-33-2) ::: To further study the SR-mediated commitment of the full length viral RNA to splicing, that is increasing the ratio of viral spliced versusunspliced RNAs, we purified total RNAs from cells expressing HIV-1 and either one of the SR proteins and subjected 10 μg total RNA to Northern blot analysis with an HIV-1 env-specific probe. In control HIV-1 cells, 8 % of HIV-1 RNA remained unspliced while this amount was lowered to 0.5% by ASF/SF2 and SC35, and to 1.5% by 9G8. This also caused a decrease of total intracellular viral RNAs by two to five fold (Table [1A](#T1){ref-type="table"}). We concluded that SR proteins are general activators of HIV-1 splicing, negatively regulating the steady state level of full length viral RNA. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Relative levels of the three viral mRNA classes. The amounts of radioactivy in mRNA signals identified by Northern blotting or by slot blotting experiments (see methods) were measured using a Storm scanner. (A) Relative levels of total intracellular viral RNA were determined as the sum of the radioactivity in the 3 signals corresponding to the 2, 4 and 9 kb mRNAs from the same experiments. Levels are expressed as the percentages of total viral RNA in cells transfected with HIV pNL4.3 only used as a reference (100 %) or with HIV-1 pNL4.3 and an SR plasmid. For the same degree of DNA transfection, the percentages of the unspliced and spliced mRNAs were calculated relative to the total viral RNA considered as 100 %. (B) Values of genomic RNA packaged into a standardized amounts of virions (CAp24 ELISA) are reported relative to the virions produced in the absence of SR protein overexpression (100%). ::: \(A) CELLS \(B) VIRIONS ---------------- ------------ -------------- ------ ----- HIV-1(%) 100 8 92 100 \+ ASF/SF2 (%) 25 0.5 24.5 24 \+ SC35 (%) 20 0.5 19.5 25 \+ 9G8 (%) 51 1.5 49.5 37 ::: Alterations of HIV-1 splicing pattern by SR proteins modify viral protein synthesis ----------------------------------------------------------------------------------- The profound modifications of the HIV-1 splicing pattern by overexpression of one of the SR proteins were expected to strongly influence viral protein synthesis. Since the unspliced viral RNA serves both as the mRNA for Gag and Gag-Pol synthesis and as the pregenome, we expected the levels of Gag and newly made virions to be strongly reduced by the SR proteins. To this end, levels of intracellular HIV Gag were assessed by CAp24 ELISA on cell lysates 48 h after DNA transfection (see methods). To measure the levels of virion production, culture supernatants were harvested every day for two days, pooled, clarified by filtration and ultracentrifuged through a 20 % sucrose cushion. Pelleted viral particles were resuspended in TNE buffer (see methods) and virus production was monitored by CAp24 ELISA. Series of measurements indicated that ASF/SF2 and SC35 caused about a 10--12 fold reduction of total Gag synthesized while 9G8 reduced it by roughly 4 fold. These results are in agreement with the relative levels of the unspliced viral RNA in HIV-1 producer cells (Table [1A](#T1){ref-type="table"}). Next we evaluated the relative amounts of cell-associated versusvirion-associated Gag. Despite the low levels of total Gag synthesized as measured by CAp24 ELISA, cell-associated Gag was found at unexpected high levels when either one of the SR proteins was overexpressed. Indeed, cell-associated Gag levels were found to be about 40%, 80% and even 250% upon overexpressing ASF, SC35 and 9G8, respectively, as compared with control HIV-1 cells (Figure [3A](#F3){ref-type="fig"}). Pol was expressed as evidenced by Gag processing and the presence of reverse transcriptase (RT) in the newly formed infectious virions (see below). The pattern of Gag processing by the viral protease was only slightly influenced by overexpressing one of the SR proteins (Figure [3B](#F3){ref-type="fig"}, compare lanes 2--4 to 1, upper panel). ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Influence of SR proteins on HIV-1 protein synthesis. 293T cells (2 × 10^5^per well) were transfected with 1 μg of HIV-1 pNL4.3 in the presence of increasing amounts of plasmid encoding either ASF/SF2, SC35 or 9G8. DNA concentrations were maintained constant by supplementation with the pCLacZ control plasmid which also served to monitor transfection efficiency. Values reported here correspond to assays carried out with a HIV to SR DNA molar ratio of 1:05. Cells were recovered two days after DNA transfection. A: Levels of Gag production were assessed by CAp24 antigen ELISA and expressed as pg of CA per μg of total cellular proteins. Note that ASF had a clear negative impact on Gag accumulation in cells whereas 9G8 had an opposite effect. B: Equivalent amounts of CAp24 antigen as measured by ELISA were subjected to western blotting. The same membrane was alternatively probed with the respective antibodies as indicated on the right: anti-CAp24 for Gag, anti-Vpr for p15, anti-NEF for p27 and anti-TMgp41 for Env. The viral Gag, Vpr, NEF and Env proteins are indicated according to their molecular weights in kDaltons. Note that SR proteins did not change the Gag processing pattern (compare lanes 2--4 and 1). ASF caused an indirect increase of Vpr cellular accumulation (lane 2) in agreement with its positive effect on Vpr mRNA level (Figure 1). On the other hand SC35 and 9G8 had an opposite effect (lanes 3--4). All Env levels were low (lanes 2--4) except in the control (lane 1). ::: ![](1742-4690-2-33-3) ::: A large fraction of the 4 kb mRNAs codes for Env. The very low level of Env glycoproteins present in cells is consistent with the fact that SR proteins strongly reduced the encoding viral mRNA (Figure [2B](#F2){ref-type="fig"}; Figure [3B](#F3){ref-type="fig"} bottom panel). Last we analysed viral protein synthesis directed by the multiple spliced 2 kb mRNAs, coding for the regulatory proteins Nef and Vpr and the trans-acting factors Tat and Rev. Only the expression of Vpr was found to be markedly enhanced by ASF/SF2 in agreement with the increased level of Vpr mRNAs (Figure [2](#F2){ref-type="fig"}; Figure [3B](#F3){ref-type="fig"}, compare lanes 1 and 2; and data not shown). Thus we can conclude that the SR proteins have a strong indirect impact on viral protein synthesis due to their alterations of the HIV-1 splicing pattern. Only the rather high level of cell-associated Gag appears to contradict this view (see discussion). Influence of the SR proteins on Gag and Env expression analysed by immuno-confocal microscopy --------------------------------------------------------------------------------------------- To better understand the influence of the SR proteins on Gag and Env synthesis, we examined by immunofluorescence staining and confocal laser microscopy (CLSM), co-expression of the two major viral structural proteins in individual cells. HIV-1 expressing cells were subjected to immuno-staining using anti-MA for Gag (green staining) and anti-gp120 for Env (red staining) antibodies, and all stainings were viewed by confocal microscopy (Figure [4A](#F4){ref-type="fig"}) (see methods). It is noteworthy that most, if not all, cells co-expressed Gag and Env which accumulated at the plasma membrane and in intracellular vesicles (merge picture in Figure [4A](#F4){ref-type="fig"}). Co-expression of HIV-1 Gag and Env was confirmed by examining 100 cells where Gag only cells were hardly found, as expected with complete HIV-1 (Figure [5](#F5){ref-type="fig"}). ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Confocal microscopy of cells co-expressing HIV-1 Gag, Env and SR-protein. Panel A: 293T cells expressing HIV-1 pNL4.3 were subjected to immuno-staining using anti-Map17 (green staining) and anti-Env gp120 (red staining) antibodies and staining was viewed by confocal microscopy as described in methods. Most if not all cells expressed Gag and Env but only partial colocalization was seen (merge picture). Right panel corresponds to the same cells viewed by phase contrast microscopy. Panel B: same as in A except that His tagged-ASF/SF2 SR protein was overexpressed by DNA transfection with about 75% transfection efficiency (see methods). ASF/SF2 protein is localized in the nucleus (blue staining) and its overexpression caused a drastic reduction of Env level while Gag remained well expressed in agreement with the western blot data (Figure 3) but with an heterogenous pattern (first panel). Panel C: same as in A except that His tagged-SC35 SR protein was overexpressed by DNA transfection with about 75% transfection efficiency (see methods). SC35 protein (nuclear blue staining) overexpression caused a reduction of Env level while Gag was still highly expressed in agreement with the western blot data (Figure 2). Note that in all cases examined here (anti-Map17; green staining in panel A to C) Gag was found to accumulate at the plasma membrane and in intracellular compartments corresponding to vesicles \[42\] (Muriaux et al., unpublished data). ::: ![](1742-4690-2-33-4) ::: ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Influence of SR protein on cellular levels of HIV-1 Gag and Env. 293T cells expressing HIV-1 and one SR protein (either ASF, SC35 or 9G8) were immuno-stained, examined and counted using Confocal Laser Scanning Microscopy (see figure 4). Numbers are representative of more than 100 SR positive cells. For all experiments we evaluated the expression of Gag and Env, and SR protein when applicable. The numbers are expressed as the percentage of all SR positive cells given a DNA transfection efficiency of 70--75% (not shown). When HIV-1 pNL4.3 was transfected alone, 100% of the cells were found to co-express Gag and Env (first bar). Upon co-transfection with the ASF coding DNA, a majority of the cells only expressed ASF and about half of them expressed Gag and the SR protein (see ASF bars). Upon co-transfection of pNL4.3 and either the SC35 or 9G8 coding plasmid, a majority of cells expressed Gag and the SR protein (see SC35 and 9G8 bars). ::: ![](1742-4690-2-33-5) ::: Overexpressing ASF/SF2 as evidenced by a blue nuclear staining in most cells (Figure [4B](#F4){ref-type="fig"}) caused a drastic reduction of Env but only moderately affected Gag (Figure [4B](#F4){ref-type="fig"}, green and red stainings) in agreement with the western blot data (Figure [3B](#F3){ref-type="fig"}, lane 2). As above, Gag was seen to accumulate in intracellular vesicles and at the plasma membrane while Env was expressed in a heterogeneous manner and mainly located in the cell interior (Figure [4B](#F4){ref-type="fig"}, HIV Env panel), probably in the Golgi area and in intracellular vesicles (Figure [4B](#F4){ref-type="fig"}, merge picture). Quantitative values on 100 cells, taking into account that 70--75% of the cells were positively transfected, showed that co-expression of Gag and ASF was observed in 25% of the cells while Gag, Env and ASF was seen in only 10% of the cells. At the same time 65% of the cells expressed ASF only (Figure [5](#F5){ref-type="fig"}, bars labelled ASF). These results further showed that the ASF/SF2 SR protein can have a drastic negative impact on HIV-1 since its overexpression caused a nearly complete suppression of Gag and Env expression in a large fraction of the cells (Figures [4B](#F4){ref-type="fig"} &[5](#F5){ref-type="fig"}). SC35 (Figure [4C](#F4){ref-type="fig"}) and 9G8 (not shown) SR proteins had less pronounced effects since a majority of the cells coexpressed Gag and one SR protein (Figure [5](#F5){ref-type="fig"}; 45 to 55 % see bars labelled gag+SC35 and gag+9G8, respectively) or evenly in the case of Gag, Env and 9G8 (Figure [5](#F5){ref-type="fig"}; see bar labelled gag+env+9G8). These observations suggest that the SR proteins can have differential effects on HIV-1 structural protein expression. The influence of the SR proteins on Gag and Env synthesis was further evaluated with respect to virion production and infectivity. Influence of SR proteins on virion production and infectivity ------------------------------------------------------------- This was examined by monitoring the levels of HIV-1 virion production under conditions of increasing expression of the SR proteins. As shown in Figure [6A](#F6){ref-type="fig"}, SR proteins overexpression induced a dose-dependent inhibition of virion production as compared with control cells co-transfected with HIV-1 pNL4.3 and an irrelevant expression vector. A high dose of SR DNA, notably ASF/SF2, caused a nearly complete inhibition of virion production. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Expression of viral proteins results from alterations of splicing pattern. 293T cells (2 × 10^5^per well) were transfected with 1 μg of HIV-1 pNL4.3 in the presence of increasing amounts of plasmid encoding either ASF/SF2, SC35 or 9G8 (ratios indicate molar amounts of HIV-1 DNA vsSR-expressing vector). DNA concentrations were maintained constant by supplementation with the pCLacZ control plasmid which also served to monitor transfection efficiency. A: Viral production was monitored by CAp24 antigen ELISA and expressed as ng of p24 per ml of medium (see methods). Results are representative of 3 independent experiments. Note that the effect of ASF/SF2 on virion production was already drastic at a HIV/SR molar ratio of 1:0.5. B: The pelleted viral particles were tested for their content in Gag, Pol, Env and Vpr proteins. Equivalent amounts of CAp24 antigen measured by ELISA were subjected to Western blotting. The same membrane was alternatively probed with the respective antibodies as indicated on the right: anti-CAp24 for Gag, anti-RT for p66 and p51, anti-Vpr for p15 and anti-TMgp41 for Env. The viral Gag, RT, Vpr and Env proteins are indicated according to their molecular weights in kDaltons. Note that fully mature CAp24 and RTp66/p51 were abundant in all virion preparations. ASF caused an indirect increase of Vpr incorporation in virions (lane 2) whereas SC35 and 9G8 had an opposite effect (lanes 3--4). All Env levels were low (lanes 2--4) except in the control (lane 1). ::: ![](1742-4690-2-33-6) ::: Protein composition of the virions generated by cells overexpressing one of the SR factors, at a HIV-1/SR DNA ratio of 1:0.5, was investigated by western blotting using antibodies against the major core component, CAp24, the RT enzyme, viral factor VPR and the envelope glycoprotein TMgp41. As shown in Figure [6B](#F6){ref-type="fig"}, CAp24 and RT were found as processed Gag protein and Pol enzyme, respectively, in proportions similar or close to wt HIV-1 particles (see panels labelled αCAp24 and αRT). On the contrary, VPR was more abundant in virions upon overexpression of ASF/SF2 in agreement with higher levels of the corresponding viral mRNA and protein in cells (Figures [2](#F2){ref-type="fig"} and [3B](#F3){ref-type="fig"} lane 2). With SC35 and 9G8 Vpr was hardly detected in virions in agreement with the very low level of Vpr mRNA and protein in cells (Figure [3B](#F3){ref-type="fig"} lanes 3--4). All SR proteins examined negatively impacted on the incorporation of Env TMgp41 in virions (Figure [6B](#F6){ref-type="fig"}, lanes 2--4), again in agreement with the fact that Env mRNA and protein levels were drastically reduced in cells (Figures [2](#F2){ref-type="fig"} and [3B](#F3){ref-type="fig"}). To test whether the decreased level of cellular unspliced viral RNA also caused an attenuation of genome packaging into newly made virions, viral particles corresponding to the same amounts of CAp24 were used to purify the genomic RNA which was analyzed by slot-blotting using a gag-specific probe. For all overexpressed SR proteins, genomic RNA packaging was reduced from 3 to 4 fold compared with control virions (Table [1B](#T1){ref-type="table"}). To determine the infectivity of virions produced by cells overexpressing one of the SR proteins, the same amount of virus-associated genomic RNA was used to infect Hela P4 cells, a HeLa subtype that constitutively expresses the CD4 receptor and contains the lacZgene under the control of the HIV-1 LTR. One day later, blue cells were counted allowing us to assess virus infectivity (see methods). Upon overexpression of each one of the SR proteins, virus infectivity, based on the same amount of genomic RNA, was found to be 30 to 60% of the control virus, or 6 to 12 fold less based on identical amounts of CAp24-associated particles. It can be concluded that overexpression of each one of the SR proteins caused a strong reduction of the unspliced viral RNA in cells, and this had a more pronounced effet on virion production than on Gag synthesis (Figures [2](#F2){ref-type="fig"}, [3](#F3){ref-type="fig"}, [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"}, [6](#F6){ref-type="fig"}). At the same time levels of genomic RNA packaged into progeny virions remained high (Table [1](#T1){ref-type="table"}). These findings are in full agreement with the fact that the genomic RNA is considered to be an indispensable partner of Gag in the course of virus assembly. Discussion ========== In the present study, we show that the overexpression of either one of three different SR proteins, namely ASF/SF2, SC35 and 9G8, profoundly affected the HIV-1 splicing pattern (Figure [1](#F1){ref-type="fig"}) \[[@B26]\], resulting in a drastic decrease of virus production. However, the progeny virions still made retained part of their infectivity. SR protein overexpression caused an oversplicing of the HIV-1 full length transcript and confirm that the targets of activation depend on the SR protein overexpressed. Indeed, ASF/SF2 stimulates splicing at sites A1 and A2, while SC35 and 9G8 preferentially enhance splicing at site A3 (Figures [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}). In addition to being general activators of constitutive splicing, results reported here confirm that each one of the three SR proteins exerts specific effects on the alternative splicing of HIV-1 primary RNA transcript (Figure [2](#F2){ref-type="fig"}) \[[@B26]\]. Little is known about the HIV-1 A1 site. Here, we show that ASF/SF2 participates in the utilization of A1 by a mechanism that requires further investigations. Many elements act in concert to repress splicing at site A2 such as its intrinsic weakness \[[@B5],[@B6]\] and the existence of the hnRNP A/B dependent ESSV located in the noncoding exon flanking sites A2 and D3 \[[@B8],[@B9]\]. Then how does ASF/SF2 exert this control ? The exon bridging hypothesis proposes that U1 snRNP binding to the downstream donor site acts to increase splicing efficiency at the upstream flanking acceptor site (for review see \[[@B20]\]). Also, SR proteins are known to stabilize U1 snRNP binding on suboptimal donor sites. Consequently, one can imagine that ASF/SF2 reinforces splicing at site A2 by stabilizing spliceosomal interactions at the suboptimal site D3. Accordingly, other SR proteins like SC35 or 9G8 would be expected to have the same effect as ASF/SF2 on site A2 and thus on Vpr mRNA synthesis. This prediction is inconsistent with our data (Figure [3](#F3){ref-type="fig"}) since overexpression of SC35 and 9G8 did not increase Vpr RNA level. Another possibility is that ASF/SF2 positively regulates splicing at site A2 by counteracting the effect of ESSV. ESSV represses splicing at site A2 by binding cellular hnRNP A/B proteins \[[@B8],[@B9]\]. This binding prevents the assembly of U2AF on the polypyrimidine tract (PPT) and subsequently the formation of a functional spliceosome between sites D1 and A2 \[[@B9]\]. SR proteins are thought to also activate weak acceptor sites by facilitating the recruitment of U2AF on the PPT \[[@B20]\]. It is tempting to speculate that the ratio between hnRNP A/B and ASF/SF2 bound close to site A2 modulates the binding of U2AF at this site. This effect of SR proteins is generally mediated by a splicing enhancer, but whether an ASF/SF2-dependent splicing element is required here remains to be determined. Site A3 is used to generate Tat mRNAs. Like A2, A3 is intrinsikly weak and repressed by hnRNP A/B-dependent ESS2 and hnRNP H-dependent ESS2p \[[@B17],[@B29]\]. Our results show that a strong positive control is exerted by SR proteins SC35 and 9G8 at this level. Findings on SC35 are consistent with the recent data of Zahler et al. \[[@B19]\] reporting a novel ESE downstream of A3 that reinforces A3 utilization in the presence of a high level of SC35 in vitro. In addition, hnRNPs act in a trans-dominant manner to counteract that of SC35 in vitro\[[@B19]\]. Taken together these results strongly suggest that changing the hnRNP/SC35 ratio probably leads to activation or repression of splicing at site A3. Little is known on the implications of SR protein 9G8 in HIV-1 splicing. The present data show that 9G8 appears to function in a way similar to SC35 (Figure [2](#F2){ref-type="fig"}). As for ASF/SF2, the exon bridging hypothesis can be mentioned but 9G8 acts mainly by binding specific cis-acting RNA elements \[[@B30]\]. Even if overexpression of SC35 and 9G8 caused a large accumulation of Tat mRNAs, it is likely that these two SR proteins act by distinct mechanisms. Indeed, firstly SELEX experiments showed that 9G8 and SC35 recognize different consensus RNA sequences \[[@B30]\]. Secondly, ESS2 mutations that in vitrostrongly reinforce the binding of SC35 on an ESS2-containing transcript have no effect on 9G8 binding to the same substrate \[[@B19],[@B30]\]. Sequences important for 9G8 splicing activation remain to be determined. In conclusion, the exact molecular mechanisms by which high levels of SR proteins cause a strong enhancement of genomic RNA splicing and consequently a severe inhibition of HIV-1 virion production remain to be determined. This is presently under investigation. As expected, the profound changes of the HIV-1 splicing pattern caused by the overexpression of one of the SR proteins ASF/SF2, SC35 or 9G8 inhibited viral protein synthesis, notably that of the structural proteins Gag and Env (Figures [3](#F3){ref-type="fig"}, [4](#F4){ref-type="fig"}, [5](#F5){ref-type="fig"}) and consequently, virion production (Figure [6A](#F6){ref-type="fig"}). Still in agreement with such alterations of the HIV-1 splicing pattern (Figures [1](#F1){ref-type="fig"}, [2](#F2){ref-type="fig"}), Vpr synthesis was upregulated by ASF whereas SC35 and 9G8 had an opposite effect (Figure [3B](#F3){ref-type="fig"}). But it was surprising to find a rather high heterogeneous level of cell-associated Gag (Figure [3A](#F3){ref-type="fig"}). To explain this apparent discrepancy, one should remember that the unspliced viral RNA performs two essential functions, firstly as the mRNA for Gag and Gag-Pol synthesis and secondly as the pregenome for Gag assembly (reviewed in ref. \[[@B31]\]). In fact, the Gag assembly process requires two platforms that are the genomic RNA through specific NC-genomic RNA interactions \[[@B31],[@B32]\] and a cellular membrane in which Gag is anchored viaMA-membrane interactions (reviewed in ref. \[[@B33]\]). Membranes are not limiting whereas the full length RNA is probably limiting due to its mobilization by the translating ribosomes (reviewed in ref. \[[@B34]\]). Actually, the fate of the full length viral RNA appears to result from a subtle balance between Gag translation on ribosomes and core assembly governed by Gag-genomic RNA interactions (reviewed in refs \[[@B31],[@B35]\]). In the presence of high amounts of the SR proteins, the unspliced viral RNA is even more limiting and thus probably rarely available for assembly. Hence, the cell-associated Gag corresponds to newly made free Gag molecules as well as Gag in newly assembled core nucleocomplexes which accumulate in the cell (Figure [4](#F4){ref-type="fig"}) before being released. According to this scheme of virus assembly, the low levels of the pregenomic RNA (Table [1A](#T1){ref-type="table"}) and Env (Figure [3B](#F3){ref-type="fig"}) upon SR overexpression may very well explain why Gag assembly and virion release are most probably slowed down (Figure [6A](#F6){ref-type="fig"}). In agreement with this interpretation, the level of packaged genomic RNA into newly formed viral particles was decreased by 65 to 75% (Table [1B](#T1){ref-type="table"}). Also in agreement with the above interpretation is the observation that production of high titer lentivectors necessitates expression of the recombinant viral RNA at high levels in vector producing cells \[[@B36]\]. Despite the low level of incorporated Env (Figure [6](#F6){ref-type="fig"}), virions produced by cells overexpressing one SR protein retained part of their infectivity on Hela P4 cells. This was not unexpected since only a minimal amount of Env appears to be required to drive infectivity in certain model cell systems \[[@B37]\]. Conclusion ========== In summary, the data presented here show that elevated concentrations of SR proteins in HIV-1 expressing cells differentially affected viral RNA and protein expression, resulting in a strong decrease of viral progeny made. Thus one can speculate that the coordinated regulation of HIV-1 splice site utilization by SR proteins is of critical importance to maintain high levels and balanced ratios of the viral RNAs and hence of the viral proteins made in order to direct optimal virus assembly and production. Thus, HIV-1 probably needs to interact with the splicing machinery. In accordance with this view, SC35 is up-regulated and 9G8 down-regulated in HIV-1 infected cells \[[@B38],[@B39]\]. On a more general basis it has been found that SR proteins influence expression and replication of other viruses such as human papilloma virus type 16 \[[@B40]\] and Adenovirus \[[@B41]\]. Materials and methods ===================== Plasmids -------- Plasmids pXJ41-ASF, pXJ42-PR264 and pXJ42-9G8 \[[@B28]\] that respectively encode ASF/SF2, SC35 and 9G8 in eukaryotic cells were provided by J. Stevenin and R. Gattoni (IGBMC Strasbourg, France). The HIV-1 molecular clone is pNL4.3 (GenBank \#M19921) \[[@B27]\]. Cell cultures, transfections and infections ------------------------------------------- HeLa P4 (provided by P. Charneau) and 293T cells (provided by Genethon) were maintained in Dulbecco\'s modified Eagle\'s medium supplemented with 10 % fetal calf serum, 2 mM glutamine and antibiotics (penicillin-streptomycin; Invitrogen). One day before transfection, 3 × 10^6^293T cells were inoculated in 10-cm Petri dish (except for experiments of Figure [1](#F1){ref-type="fig"}, see legend). One day later, cells were transfected with 10 μg of proviral pNL4.3 and 4--10 μg of SR-expressing vector or of control plasmid pCLacZ by the calcium phosphate precipitation technique according to manufacturer instructions (Gibco). As observed by immuno-staining and confocal microscopy, more than 70% of the cells were positively transfected. After 12 h, cell culture supernatants were substituted by fresh culture medium. Forty-eight hours after transfection, supernatants were harvested, clarified by filtration through 0.8 μm-pore size filters and ultracentrifuged through a 20 % sucrose cushion. Pelleted viruses were resuspended in TNE buffer (25 mM Tris HCl pH 7.5; 150 mM NaCl; 1 mM EDTA). Virus production was monitored in cell culture supernatants and in virus pellets with a CAp24 ELISA capture assay (kindly provided by Valérie Cheynet and Bernard Mandrand, BioMérieux). Viral titrations were performed by infection of HeLa P4 cells (1.5 × 10^5^cells per well of a 24-well plate) with purified viruses containing 1 ng of genomic RNA. After 24 h, cells were fixed and incubated in the presence of X Gal substrate at 37°C until blue color development was complete. Viral titers were determined by counting the number of blue cells in threee different wells. RNA isolation, Northern blotting and RT-PCR ------------------------------------------- Transfected cells were harvested 48 h after transfection and washed in phosphate-buffered saline (PBS). Two-third of cells were resuspended in PBS and total cellular RNA was extracted with TRIzol reagent as recommended by the manufacturer (Invitrogen). Culture supernatants were treated as indicated above and the level of CAp24 antigen measured by ELISA. Northern blotting of intracellular viral RNAs was performed with 10 μg of total RNA and slot-blot of virion-associated genomic RNA was performed with 10 ng of CAp24 antigen. After transfer onto a Hybond-N+ membrane (Amersham Pharmacia Biotech), viral RNAs were probed with radiolabeled fragments from Env for Northern blot and from Gag region for slot-blot. All the mRNAs species were quantified by Storm (Amersham). The splicing products were analyzed by RT-PCR as previously described \[[@B10]\], except that the forward PCR primer was Odp045 \[[@B2]\]. The PCR products were fractionated on a 6 % acrylamide-7M urea electrophoresis gel and autoradiographed. Individual HIV-1 mRNA species were named according to the nomenclature of Purcell and Martin \[[@B2]\]. Immunoblotting -------------- One-third of transfected cells were washed in PBS and lysed in PBS containing 0.5 % Triton. After CAp24 ELISA measurement in cell lysates and in virus pellets, samples were added to 3X gel loading buffer (0.5 M Tris-HCl pH 6.8; 0.8 % SDS; 40 % glycerol; 5 % β-mercaptoethanol; 0.03 % bromophenol blue). For immunoblotting, samples containing equal amounts of CAp24 antigen were loaded on a 10 % SDS-PAGE and fractionated proteins were transferred onto a Hybond P membrane (Amersham Pharmacia Biotech). Viral proteins were probed with monoclonal anti-CAp24 (BioMérieux), polyclonal anti-Vpr (\#3951, NIH USA), polyclonal anti-Nef (\#331, NIH USA) or monoclonal anti-TMgp41 (Ab 41A9; Hybridolab, Pasteur) antibodies. The bound antibodies were detected with peroxidase-conjugated anti-mouse IgG antibodies and visualized by the SuperSignal West Pico Chemiluminescent Substrate (Pierce). Immunofluorescence staining and confocal microscopy imaging ----------------------------------------------------------- Transfected 293 T cells were grown on poly-lysine coated coverglass dishes and fixed 24 h post transfection in 3% paraformaldehyde (diluted in Phosphate Buffer Saline -- PBS) for 20 min. The fixative was then removed and the free aldehydes were quenched with 50 mM NH~4~Cl. Cells were then permeabilized using 0.2 % Triton X-100 for 5 min and blocked in 1% BSA. The fixed cells were incubated for one hour at room temperature with primary antibodies: rabbit anti-MAp17 (NIH, USA), human anti-HIV-1 gp120 Mab(b12) (NIH, USA), rabbit anti-His for His-tagged 9G8 and ASF proteins, and mouse anti-HA1 for HA-tagged SC35 protein (Sigma). The corresponding fluorescent Alexa^®^488, 546 and 633-conjugated secondary antibodies were used at 0.5 μg/ml (Molecular probes). Coverslips were washed three times with PBS and mounted on microscope slides with Mowiol (Sigma). Images were acquired on Axioplan 2 Zeiss CLSM 510 confocal microscope with Argon 488/458, HeNe 543, HeNe 633 lasers and plan apochromat 63 × 1.4 oil Ph3 objective, supplied with LSM 510 3.4 software. Abbreviations used ================== HIV-1, Human Immunodeficiency Virus type 1. CAp24, viral capsid protein p24. MAp17, viral matrix protein p17. RT, reverse transcriptase. SR, splicing regulatory proteins. hnRNP proteins, heterogenous ribonucleoparticle proteins. Site A, splicing acceptor site. Site D, splicing donor site. Kb, kilobases. WT, wild type. PBS, Phosphate Buffer Saline. BSA, bovine serum albumin. Mab, monoclonal antibody. Authors\' contributions ======================= SJ carried out analyses on SR protein-mediated effect on HIV-1 RNA splicing. DD was in charge of cell culture and transfection assays. DM performed immuno-confocal microscopy experiments and analyses. JLD is the lab head and arranged the manuscript. Competing interests =================== The author(s) declare that they have no competing interests. Acknowledgements ================ We are grateful to J. Stevenin and R. Gattoni for their gifts of anti-ASF/SF2, anti-SC35 and anti-9G8 antibodies and for the SR protein expression plasmids, B. Mandrand (CNRS BioMérieux) for anti-CAp24 ELISA and the NIH (USA) for reagents. Work supported by ANRS, Sidaction and the European TRIoH Consortium. SJ was the recipient of an ANRS fellowship.	PubMed Central	2024-06-05T03:55:59.711508	2005-5-22	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180853/", "journal": "Retrovirology. 2005 May 22; 2:33", "authors": [ { "first": "Sandrine", "last": "Jacquenet" }, { "first": "Didier", "last": "Decimo" }, { "first": "Delphine", "last": "Muriaux" }, { "first": "Jean-Luc", "last": "Darlix" } ] }
PMC1180854	Background ========== The use of cytology brushes for the purpose of obtaining respiratory cells from adults for clinical and research purposes is well established. This technique is generally reported to be safe, both in adults with pulmonary disease and in healthy volunteers \[[@B1]\]. Samples are usually obtained under direct vision using a bronchoscope. However, we and others have recently used non-bronchoscopic brushing to sample airway epithelial cells from children \[[@B2],[@B3]\]. This method has several advantages over bronchoscopic brushing as it is simple and quick to perform, and there is no need for a bronchoscope. Although we have now successfully used this method in combination with a variety of cellular and molecular techniques to investigate the role of the epithelium in childhood asthma, initially, very little had been published on the topic. In addition, the difficulty in obtaining target organ tissue from children has meant that most information regarding common childhood diseases such as asthma has been derived from studies performed in adults. We currently understand little about the molecular mechanisms involved in the pathogenesis of asthma. The airway epithelium is an especially attractive target in which to identify new molecular mechanisms and therapeutic targets because it is critically involved in the development of asthma as the first cell of contact with the environment. This is particularly pertinent since it is likely that dysregulated epithelial repair in childhood asthma is a critical determinant of disease progression in adults. To this end, the available evidence suggests that epithelial fragility and dysfunction as well as accumulation of sub-epithelial fibroblasts and remodelling of the airway wall in asthma can occur early in childhood \[[@B4]\]. The purpose of this report is to summarise our experiences with non-bronchoscopic epithelial brushing techniques and subsequent sample processing. We have detailed the use of cells for gene and protein expression, and for cell culture. The use of primary culture systems has important advantages over the use of immortalized cell lines, and allows functional experiments to be conducted. We aim to encourage the use of these techniques to study normal developmental processes in the lung and the early pathophysiology of respiratory diseases. Methods ======= Children admitted to Princess Margaret Hospital for the purposes of elective surgery for non-respiratory complaints were recruited for this study. Children were usually having minor gastrointestinal or ear, nose and throat surgery. The study was approved by the Princess Margaret Hospital for Children Ethics Committee, and written informed consent was obtained from the parents of the children prior to sampling (see appendix for parent information sheet). The asthmatic and atopic backgrounds of the children were determined by allergen-specific IgE testing and a validated asthma and allergy questionnaire was administered to the parent or legal guardian \[[@B5]\]. Prior to surgery, each child was anaesthetised and intubated. A nylon cytology brush (BC 25105, Olympus, Australia) was used to sample cells from the airway. The plastic sheath protecting the brush was removed and discarded, as retracting the brush into the tightly fitting sheath would dislodge cells. The unprotected brush was inserted directly through the endotracheal tube, advanced until resistance was felt, and rubbed against the epithelial surface to sample cells. The brush was then withdrawn and the tip cut off into 5 ml of culture media (RPMI-1640 containing 10% (v/v) heat inactivated foetal calf serum). This process was repeated at least once. A sub-group of twenty five of these children was studied to assess how well the technique was tolerated. Respiratory variables were monitored before, during, and after the brushing procedure. Symptoms following the brushing were recorded by contacting the parents within one week of the procedure. Children who underwent non-bronchoscopic brushing (12 male, mean age 10.2, SE 0.71) were compared to 24 control children who were anaesthetised and intubated only for similar procedures (15 male, mean age 10.9, SE 0.76). Sample Processing ----------------- Epithelial cells were collected into RPMI-1640 containing 10% (v/v) heat inactivated foetal calf serum. Samples were processed immediately. An aliquot of the cell suspension was diluted 1:2 with 0.4% trypan blue, and applied to a haemocytometer. The total number of cells, and the percentage of viable cells, was determined under a light microscope, within 15 minutes of collection. Each sample of cells obtained by non-bronchoscopic brushing was processed to allow for use in multiple investigative techniques; 1 × 10^6^cells were used for cell culture or protein extraction, 0.5 × 10^6^cells to produce cytospin slides for immunocytochemical studies, and RNA was extracted from the remaining 1 × 10^6^cells. For culture, cells were washed once in RPMI-1640 media and the cell pellet resuspended in bronchial epithelial basal media (BEBM, Clonetics, CA) supplemented with bovine pituitary extract (50 mM), insulin (5 mM), hydrocortisone (0.5 mM), gentamicin (0.001%, v/v), amphotericin B (0.0005%, v/v), retinoic acid (0.1 μM), transferrin (10 mM), triiodothyronine (6.5 μM), epinephrine (6.5 μM) and human recombinant epidermal growth factor (EGF: 0.5 μM). The cells were then seeded into a culture vessel (25 cm^2^growth surface area) pre-coated with a mixture of fibronectin, collagen and bovine serum albumin, and maintained at 37°C in a humidified incubator. Twenty-four hours post-isolation, unattached cells were collected. These cells were reseeded into the same culture vessel, with fresh media containing Ultroser G (2% v/v; BioSepra, CA), a serum substitute. The collection and reseeding of viable unattached cells was repeated at both 48 and 72 hours post isolation. Subsequent cultures were fed every second day and were usually passaged every 13--16 days. Before the remaining cell suspension was used for protein and RNA extraction, and to produce cytospin slides, the macrophages were removed by positive selection: the cell suspension was added to a culture dish that had been previously coated with CD-68 antibody (Dako, Australia). The plate was incubated for 20 minutes (37°C, 5% CO~2~) to allow the macrophages to adhere. The suspended epithelial cells were aspirated from the plate, and the macrophages removed using trypsin (0.25%) for subsequent analysis. The macrophage depleted cell suspension was used to produce cytospins, extract protein, and extract RNA. Cytospin slides were prepared by centrifuging epithelial cells onto a glass slide in a cytocentrifuge (Hettich). Slides were air dried, fixed in 4% paraformaldehyde for 10 minutes, and then stored at -20°C until required. Immunocytochemical staining of the cytospins was used to confirm the purity of the epithelial sample. Antibodies against cytokeratin (a marker for tissue of epithelial origin), α -smooth muscle actin (a marker of myofibroblasts, myoepithelial cells and smooth muscle cells), smooth muscle myosin (a marker of smooth muscle cells), and vimentin (a marker of mesenchymal cells) were used to confirm epithelial cell phenotype. Immunocytochemical techniques provide only semi-quantitative data about protein expression, and are not sensitive enough to determine levels of protein expression accurately; therefore protein was extracted for analysis with Western blotting. Protein was extracted from the pelleted epithelial cells by lysing the cells in 200 μl of an SDS extraction buffer (20 mM Tris, 1 mM SDS, 1 mM DTT) in the presence of protease inhibitors (Sigma). A commercial assay (Micro BCA Protein Assay, Pierce Biotechnology) was used according to the manufacturers instructions to determine the concentration of total protein in the cell lysate. For each sample, 50 μg of protein was subjected to 12% SDS-PAGE, and immunoblotted with anti-β-actin antibody. Antibody binding was detected with ECL Plus Western Blotting Detection Reagents (Amersham Biosciences). Total RNA was extracted from epithelial cells using the QIAGEN RNeasy kit (Vic, Australia). RNA quality and quantity was assessed using the Agilent Bioanalyser (Vic, Australia). RNA was prepared from 18 subjects according to a modified version of the protocol of Baugh et al \[[@B6]\] and hybridised to microarrays. For our first study, gene expression profiles from 9 mild, asymptomatic asthmatics were compared to 9 healthy children for a total of 18 arrays. Expression of genes in cells from asthmatic and healthy children was compared using Affymetrix Human Genome U133 Arrays (HG-U95Av2), which examine the expression of approximately 23,000 genes. Real-time PCR was used to validate the array data for specific genes. Real-time PCR was conducted as previously described \[[@B3]\]. Data was reported as mean (SE) and analysed by independent samples t-test. Significance was taken as p \< 0.05. The proportion of basal cells was analysed using analysis of variance (ANOVA) to compare the difference between the three phenotype groups. Results ======= The procedure is well tolerated ------------------------------- The only significant adverse event reported after the procedure was cough (Table [1](#T1){ref-type="table"}). The nursing staff recorded cough in 20% of patients undergoing brushing. Parentally reported cough was higher, with 40% of parents reporting cough in those children undergoing the procedure. Cough persisting longer than 4 hours was recorded in 32% of patients, and 28% had cough persisting to the following day. Despite this, 100% of parents said that if asked by a friend \"Should my child participate in this study?\" they would answer \"yes\". Children who underwent non-bronchoscopic brushing were, on average, ventilated for a longer period than the control group (27 vs. 22.3 min), but this was not statistically significant (p = 0.066, table [1](#T1){ref-type="table"}). There was no difference in any other clinical variables recorded, including the lowest level of oxygen saturation, and the time spent in recovery. ::: {#T1 .table-wrap} Table 1 ::: {.caption} ###### Symptoms following non-bronchoscopic brushing. Symptoms following non-bronchoscopic brushing, and observations of respiratory parameters before, during, and after brushing. Cough was the only reported symptom, and all parents said they would recommend the study to a friend. ::: Controls mean (SE) Sampled mean (SE) p-value ------------------------------ -------------------- ------------------- --------- Number of participants 24 25 \- Cough recorded 0% 20% (8%) 0.022 Parental reported cough \- 40% (10%) \- Cough \> 4 h \- 32% (9%) \- Cough following day \- 28% (46%) \- Recommend to friend? \- 100% (0%) \- Length of Ventilation (mins) 22.3 (1.85) 27 (1.68) 0.066 Lowest oxygen saturation 98.6 (0.12) 96.8 (1.02) 0.083 Supplemental oxygen 79% (8%) 80% (8%) 0.944 Highest respiratory rate 22.7 (0.77) 22.2 (0.58) 0.607 Time in recovery (mins) 21.5 (2.22) 18.4 (1.40) 0.246 ::: Symptoms and respiratory variables were compared between asthmatic and non-asthmatic children undergoing non-bronchoscopic brushing (Table [2](#T2){ref-type="table"}). The nursing staff did not record cough in any of the asthmatic patients. Parentally reported cough was higher in the asthmatic subjects (43% vs. 39%), but this was not statistically significant (p = 0.863). Parents of asthmatic children also reported a higher level of cough persisting to the following day (43% vs. 22%), however this was not statistically significant (p = 0.322). ::: {#T2 .table-wrap} Table 2 ::: {.caption} ###### Symptoms following non-bronchoscopic brushing. Comparison of symptoms following non-bronchoscopic brushing, and observations of respiratory parameters before, during, and after brushing in asthmatic and non-asthmatic patients. ::: Non-asthma mean (SE) Asthma mean (SE) p-value ------------------------------ ---------------------- ------------------ --------- Number of participants 18 7 \- Cough recorded 28% (11%) 0% (0%) 0.129 Parental reported cough 39% (12%) 43% (20%) 0.863 Cough \> 4 h 28% (11%) 43% (20%) 0.489 Cough following day 22% (10%) 43% (20%) 0.322 Recommend to friend? 100% (0%) 100% (0%) \- Length of Ventilation (mins) 28.9 (1.92) 22.1 (2.86) 0.071 Lowest oxygen saturation 96.9 (1.33) 96.4 (1.46) 0.845 Supplemental oxygen 83% (9%) 71% (18%) 0.524 Highest respiratory rate 22.6 (0.86) 21.0 (0.72) 0.282 Time in recovery (mins) 18.8 (1.73) 17.3 (2.50) 0.644 ::: Epithelial cell retrieval and processing ---------------------------------------- The mean cell retrieval from 151 brushings was 2.67 × 10^6^cells (SE = 0.16 × 10^6^, Figure [1A](#F1){ref-type="fig"}) and on average, 17.3% of cells were viable (SE = 2.15%, Figure [1B](#F1){ref-type="fig"}). The number or viability of cells retrieved was not related to asthmatic or atopic status. The epithelial phenotype of the cells was confirmed by immunostaining (Figure [2](#F2){ref-type="fig"}). To confirm that the population of epithelial cells sampled did not vary between phenotypes, the number of cells expressing the basal cell marker isolectin B4 was determined. The proportion of basal cells was independent of phenotype (Figure [3](#F3){ref-type="fig"}). ::: {#F1 .fig} Figure 1 ::: {.caption} ###### A: Histogram of number of cells sampled in 151 non-bronchoscopic brushings of the paediatric airway. The number of cells retrieved ranged between 0.1 million and 11.8 million (mean 2.7 million). B: Histogram of the percentage of viable cells in 43 non-bronchoscopic brushings of the airway. The mean viability was 17.3%. ::: ![](1465-9921-6-53-1) ::: ::: {#F2 .fig} Figure 2 ::: {.caption} ###### Immunocytochemical confirmation of epithelial cell phenotype. A: Pan-cytokeratin, B: α-smooth muscle actin, C: Smooth muscle myosin, D: Vimentin. ::: ![](1465-9921-6-53-2) ::: ::: {#F3 .fig} Figure 3 ::: {.caption} ###### Percentage of basal cells in brushings obtained non-bronchoscopically from the paediatric airway. A: Histogram of the percentage of basal cells in each brushing. On average, 10.04% of cells were basal cells. B: Percentage of basal cells retrieved in brushings from each phenotype. The percentage of basal cells retrieved was not different depending on phenotype. ::: ![](1465-9921-6-53-3) ::: In our hands, the cell culture success rate is such that 93% of samples exhibit some growth, 80% grow to confluence, and 80% survive a second passage. Initial cultures consisted of a heterogeneous cell population composed of both terminally differentiated epithelial cells as indicated by the presence of cilia, but also non-ciliated epithelial cells. Subsequently passaged cultures were found to solely exhibit the non-ciliated epithelial morphology. No differences in cellular morphology have been observed between cultures established from healthy and asthmatic phenotypes. Established cultures have been successfully grown up to passage 7. When the cells reach 80% confluence in their second passage (Figure [4](#F4){ref-type="fig"}), they are used for functional studies. ::: {#F4 .fig} Figure 4 ::: {.caption} ###### Phase-contrast micrograph of cultured epithelial cells ::: ![](1465-9921-6-53-4) ::: Protein was extracted from the epithelial cells; the average protein yield was 74 μg (SE = 8.79 μg, N = 18) per 1 × 10^6^cells. This was sufficient for detection of β-actin using Western blotting, demonstrating the feasibility of determining protein expression from cells obtained by the non-bronchoscopic brushing technique. RNA extraction yielded on average 2.9 μg of RNA (SE = 0.21, N = 58, Figure [5A](#F5){ref-type="fig"}), as assayed by Agilent Bioanalyser (Vic, Australia). The ribosomal RNA ratio was used as a measure of RNA quality; the average ratio was 1.52 (SE = 0.12, N = 58, Figure [5B](#F5){ref-type="fig"}). This was sufficient RNA to use both for microarray analysis (1 μg) and real-time PCR analysis (0.5 μg). ::: {#F5 .fig} Figure 5 ::: {.caption} ###### Quantity and quality of RNA extracted from brushings obtained non-bronchoscopically from the paediatric airway. A: Histogram of the amount of RNA extracted. On average, 2.9 μg of RNA was extracted from 2 ml of cell suspension. B: Histogram of the quality of RNA. The ribosomal RNA ratio was used as a measure of RNA quality; on average the RNA quality was 1.52. ::: ![](1465-9921-6-53-5) ::: Preliminary data from the microarrays showed that gene expression in epithelial cells from asthmatic subjects was significantly different to that in healthy subjects (Figure [6](#F6){ref-type="fig"}). Expression of genes in the asthmatic cohort ranged from 6-fold up-regulated to 5-fold down-regulated compared to expression in the healthy controls. ::: {#F6 .fig} Figure 6 ::: {.caption} ###### Scatter plot of average gene expression (x axis) vs. log of fold change in gene expression. Those genes which are differentially regulated more than 1.5 fold are highlighted. ::: ![](1465-9921-6-53-6) ::: Discussion ========== We have demonstrated that non-bronchoscopic brushing is safe and well tolerated in children. The main symptom reported following brushing was cough, which was reported more frequently by parents than the nursing staff. The procedure was acceptable to participants, as none of the patients experienced symptoms that would cause the parents not to recommend the study to a friend. Respiratory variables and symptoms in children undergoing brushing were no different in asthmatics compared to non-asthmatics (table [2](#T2){ref-type="table"}). The level of cough reported by parents was increased in asthmatic children, however this difference was not significant. The difference was most likely due to increased observation of their child\'s symptoms, as there was no difference in the level of cough reported by nursing staff in asthmatic and non-asthmatic children. The non-bronchoscopic brushing technique was able to harvest useful quantities of epithelial cells, with a mean viability of 17.3%. Studies in adult asthmatics, harvesting cells under direct vision, report average viability ranging from 25--30%, with a significantly lower viability in the asthmatic epithelial cells \[[@B1],[@B7],[@B8]\]. In this study, the viability of the epithelial cells was not significantly different in the asthmatics. This may reflect the mild nature of asthma experienced by these children, as previous studies have reported the level of viability is correlated with the severity of the disease \[[@B8]\]. This report outlines a broad range of techniques that can be used to study a single sample of airway epithelium, thus maximising the information obtained from one brushing of the airway. Previous studies conducted in children have detailed only limited use of the sample \[[@B2]\]. We report that it is possible to obtain cells of sufficient quality and quantity to allow investigation using cell culture, immunohistochemistry, western blotting, real-time PCR, and microarray. We have used immunocytochemical staining to examine expression of protein, and sufficient protein can be extracted to perform Western blot analysis. We have cultured the epithelial cells, and can maintain and propagate the cultures successfully over seven passages. Whilst non-bronchoscopic brushings of the paediatric airway are easily and safely obtained, it is important to maximise the amount of information obtained from each brushing. To this end, we have outlined a program of non-bronchoscopic brushing that utilises a wide range of techniques, allowing a broad approach to the study of the airway epithelium in children. Conclusion ========== The purpose of this study was to demonstrate how a single sample of airway epithelial cells obtained by non-bronchoscopic brushing can be used to study gene and protein expression and provides sufficiently viable cells to allow cell cultures to be established for functional analysis. The technique is particularly useful for studying the paediatric airway because it is simple and minimally invasive, and can be used to overcome a major obstacle to our understanding of paediatric respiratory disease namely a paucity of target organ tissue. We have presented data that can be used by other groups to establish similar programs, bench-mark for quality assurance and respond to common questions by human research ethics committees. Authors\' contributions ======================= CL carried out the protein and RNA analysis, was involved in the cell culture work, and drafted the manuscript. SB designed and carried out the paediatric complications study, and participated in sample processing. AK refined the method for culture of airway epithelial cells. SS conceived of the non-bronchoscopic brushing program. DK established the cell culture program and with SS, participated in the design and coordination of this study, and contributed to the manuscript. Appendix ======== Parent Information Sheet ------------------------ ### Does raised exhaled nitric oxide reflect unrecognised airway inflammation in healthy children? You are being invited to take part in a research study. Before you decide, it is important for you to understand why the research is being done and what it will involve. Please take time to read the following information carefully. Ask us if there is anything that is not clear or if you would like more information. Take time to decide whether or not you wish to take part. GENERAL INFORMATION: Nitric oxide (NO) is a molecule that is involved in many physiological processes in the body including inflammation. NO is detectable in exhaled breath (eNO) and is raised in asthmatics. Many studies have demonstrated that exhaled NO might be a useful marker of airway inflammation in asthma thus aiding diagnosis and monitoring of disease activity. However, children who respond to skin allergy tests (atopic) and who do not have any respiratory symptoms also have raised eNO. We do not know why eNO is raised in healthy atopic children but it may also be due to inflammation in the airways that is not presently causing respiratory problems. If we can determine what causes raised eNO in healthy atopic children we will better understand how this test will help us monitor airway disease. \"What will happen if I agree for my child to take part?\" The present study will use standard diagnostic techniques to investigate whether inflammation that might not otherwise be recognised is the cause of raised NO levels in the breath of children with atopy. In order to study this issue we will recruit children, WITH and WITHOUT ATOPY, who are at Princess Margaret Hospital for day surgery. We will use the following strategy: • You will be asked to complete a questionnaire about your child\'s symptoms of allergy and asthma and medications that may effect eNO levels. • If your child is old enough they will be asked to blow into a machine to determine their eNO levels. This may occur either before the operation or at the time of their next appointment. • During surgery we will obtain a brushing of the surface of the windpipe. • A blood sample will be taken for allergy testing to common allergens. It will also be used for gene testing for NO genes and asthma genes. [Measurements we will make:]{.underline} 1\. Prior to surgery Exhaled nitric oxide-- To measure eNO we will ask your child to take a deep breath in and blow into a machine while trying to maintain a constant expiratory flow. 2\. During surgery Brushings-- To collect cells from the wall of the airways we will pass a fine brush through the larynx (voice box) into the windpipe and rub along the wall of the windpipe a few centimetres below the vocal cords. If you wish, we will demonstrate the techniques and equipment used before you decide to go ahead. Blood sample-- Blood will be collected once your child is asleep. [Risks:]{.underline} The brushing is a simple test and takes less than 5 minutes to complete. In adults it is performed without an anaesthetic. The test will not be carried out if your child\'s anaesthetist or surgeon believes the test will interfere with your child\'s treatment. We have routinely performed hundreds of these tests without incident. A dry cough is the only adverse symptom reported, it seems to occur in approximately half of the children involved in our study. [Benefits:]{.underline} We will be able to provide you with information regarding the allergic status of your child. All information that is collected about your child during the course of the research will be kept strictly confidential. Any information about your child that leaves the hospital will have your/his/her name and address removed so that you/he/she cannot be recognised from it. It is up to you to decide whether or not to take part. If you decide to take part you will be given this information sheet to keep and be asked to sign a consent form. If you decide to take part you are still free to withdraw at any time and without giving a reason. This will not affect the standard of care your child receives. If you have any complaints about any aspect of the study you can contact the Executive Director Medical Services of PMH (Dr Geoff Masters) on 9340 1550 Thank you for reading this information sheet. If you have any further questions with regard to this study they can be discussed with Dr Stephen Stick (Telephone 9340 8830) Dr Scott Burgess (Hospital switch board -- Telephone 9340 8222 page 2025 at any time) Acknowledgements ================ This work was supported by a grant from the NH&MRC Australia (303145). We gratefully acknowledge the assistance of Professor Charles Mackay and Trina So (Garvan Institute of Medical Research, Sydney, Australia) in the hybridisation of the microarrays, and Professor David Erle (University of California, San Francisco) in the preliminary analysis of the microarray data.	PubMed Central	2024-06-05T03:55:59.714890	2005-6-8	{ "license": "Creative Commons - Attribution - https://creativecommons.org/licenses/by/4.0/", "url": "https://www.ncbi.nlm.nih.gov/pmc/articles/PMC1180854/", "journal": "Respir Res. 2005 Jun 8; 6(1):53", "authors": [ { "first": "Catherine", "last": "Lane" }, { "first": "Scott", "last": "Burgess" }, { "first": "Anthony", "last": "Kicic" }, { "first": "Darryl", "last": "Knight" }, { "first": "Stephen", "last": "Stick" } ] }

Subsets and Splits

No community queries yet

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