| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Clinical and Molecular Epidemiology Unit (F.K.K., J.P.A.I.), Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina 45110, Greece; Translational Research Center (T.Ak., H.H.), Kyoto University Hospital, Kyoto University School of Medicine, Kyoto, Japan; Division of Endocrinology and Diabetes (T.Aw., S.Ka., S.Ku.), Department of Medicine, Saitama Medical University, Saitama, Japan; Third Department of Internal Medicine (Yoshiy.B., Yoshio.B.), Showa University School of Medicine, Tokyo, Japan; Department of Pathology (D.A.C.), University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania; Department of Hematology (I.F.), Bone Marrow Transplantation and Blood Neoplastic Diseases, Medical Academy, Wroclaw, Poland; Institute of Immunology and Experimental Therapy (I.F., L.K., E.P.), Polish Academy of Sciences, Wroclaw, Poland; Shiraz Institute for Cancer Research (A.G.), Medical School, Shiraz University of Medical Sciences, Shiraz, Iran; Institute of Biomedical Research (S.C.G., J.M.H.), Division of Medical Sciences, The Medical School, University of Birmingham, Birmingham, United Kingdom; Department of Endocrinology (Y.H., I.M.), Kurume University School of Medicine, Kurume, Fukuoka, Japan; Human Molecular Genetics Laboratory of the Department of Forensic Medicine (R.P., M.C.), Departments of Diabetology, Newborn Pathology and Birth Defects, of Medical Genetics, and of Endocrinology (T.B.), Medical University of Warsaw, Warsaw, Poland; Department of Internal Medicine (P.-W.W., R.-T.L.), Chang Gung Memorial Hospital-Kaohsiung Medical Center, Chang Gung University College of Medicine, Kaohsiung, Taiwan; Department of Endocrinology (T.B.), Medical Research Center, Polish Academy of Science, Warsaw, Poland; Department of Molecular Diagnostics (E.I.C.), National Research Center GosNIIgenetika, Moscow, Russia; Graduate Institute of Medical Genetics and Department of Clinical Research (S.-H.H.J.), Kaohsiung Medical University Hospital, Kaohsiung, Taiwan; Endocrine and Metabolism Research Center (G.-H.R.O.), Namazee Hospital, Shiraz, Iran; Division of Endocrinology and Metabolism (M.T.), Showa University Fujigaoka Hospital, Yokohama, Kanagawa-ken, Japan; Department of Medical Information (T.T.), Showa University School of Pharmaceutical Science, Tokyo, Japan; Biomedical Research Institute (J.P.A.I.), Foundation for Research and Technology-Hellas, Ioannina, Greece; and Institute for Clinical Research and Health Policy Studies (J.P.A.I.), Department of Medicine, Tufts-New England Medical Center, Tufts University School of Medicine, Boston, Massachusetts
Address all correspondence and requests for reprints to: John P. A. Ioannidis, M.D., Professor and Chairman, Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina 45110, Greece. E-mail: jioannid{at}cc.uoi.gr.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Objective: The aim was to generate large-scale evidence on whether the CTLA-4 polymorphisms (A49G and CT60) and haplotypes thereof increase the susceptibility to GD and/or HT.
Design, Setting, and Participants: Meta-analyses of group-level data were reviewed from 32 (11,019 subjects) and 12 (4,479) published and unpublished studies for the association of the A49G polymorphism with GD and HT, respectively (PubMed and HuGeNet search until July 2006). There were 15 (n = 7246) and six (n = 3086) studies available for the CT60 polymorphism, respectively. Meta-analyses of individual-level data from 10 (4906 subjects) and five (2386) collaborating teams for GD and HT, respectively, were also reviewed.
Main Outcome Measures: Association of gene variants and haplotypes with GD and HT was measured.
Results: Group-level data suggested significant associations with GD and HT for both A49G [odds ratios 1.49 (P = 6 x 10–14) and 1.29 (P = 0.001) per G allele, respectively] and CT60 [1.45 (P = 2 x 10–9) and 1.64 (P = 0.003) per G allele, respectively]. Results were consistent between Asian and Caucasian descent subjects. Individual-level data showed that compared with the AA haplotype, the risk conferred by the GG haplotype was 1.49 (95% confidence interval 1.31,1.70) and 1.36 (95% confidence interval 1.16,1.59) for GD and HT, respectively. Data were consistent with a dose-response effect for the G allele of CT60.
Conclusion: The CT60 polymorphism of CTLA-4 maps an important genetic determinant for the risk of both GD and HT across diverse populations.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Human CTLA-4 gene consists of four exons and three introns (8). Many studies conducted over the last decade have claimed associations of AITDs with an adenine to guanine transition at position 49 of exon 1 (A49G) (9), but several other polymorphisms of the same gene have also been evaluated. The results have not always been consistent. A detailed genomic analysis of CTLA-4 in GD, HT, and type 1 diabetes mellitus involving 108 single nucleotide polymorphisms (SNPs) was published in Nature in 2003 (3). The G allele of the +6230G>A (CT60) polymorphism showed very strong association to GD. Other polymorphisms in addition to CT60, such as JO31, JO30, and JO27_1, were also highly associated and found to be in strong linkage disequilibrium (LD) (i.e. they tend to exist together), making it very difficult to map disease susceptibility to a single SNP. In an attempt to gain a greater understanding of the effects of individual CTLA-4 variants and address some of the published inconsistencies, we performed a collaborative international meta-analysis. We have included both published and unpublished data from a network of investigators working in the field. Investigators provided individual-level information from their databases, which allowed more detailed haplotype analyses.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We first identified all published studies that examined the association of any CTLA-4 gene polymorphism with AITDs. Sources included MEDLINE, EMBASE, and HuGeNet (last search update performed in June 2006). The search strategy was based on combinations of "CTLA4," "CTLA-4," "cytotoxic T-cell lymphocyte associated antigen 4," "CD152," "CD28," "thyroid,*" "Graves,*" and "Hashimoto,*" limited to humans without language restriction. References of retrieved articles were also screened.
Studies were eligible if they had determined the distribution of alleles and/or genotypes for any of these polymorphisms in unrelated cases with one or more types of AITD and in unrelated controls without AITDs. We did not consider family-based studies of pedigrees with several affected cases per family. We also excluded studies that did not discriminate between the various types of AITD because all analyses were to be performed separately for GD and HT.
When the published literature was accumulated, the meta-analysis coordinators (F.K.K. and J.P.A.I.) communicated with the corresponding investigators of all eligible studies, seeking their participation in a collaborative meta-analysis of individual-level data. The prerequisite for participation was that investigators should be able to supply individual-level genotype data on GD and/or HT cases and controls for both A49G and CT60 polymorphisms on their study populations. Participating investigators were also encouraged to provide individual-level data on additional CTLA-4 SNPs, whenever available.
Databases
For published articles, two investigators (F.K.K. and J.P.A.I.) independently extracted data and reached consensus on potential disagreements. The following information was sought from each report: authors; journal and year of publication; country of origin; selection and characteristics of cases and controls; demographics; "racial" descent of the study population (Asian, Caucasian, African-American, and other); eligible and genotyped cases and controls; and genotype distributions. Whenever a study team included two or more "racial" descent subgroups, these were treated as separate studies in all analyses.
Furthermore, we examined whether matching of cases and controls had been performed, there was specific mention of blinding of the personnel that performed the genotyping to the clinical status of the subjects, the genotyping method had been validated, and genotype frequencies in control groups were in Hardy-Weinberg equilibrium (HWE) according to an exact test.
Individual-level data were sent to the coordinators, and checked for logical errors, inconsistencies, and potential deviations from HWE among controls. Queries were sent back to the primary investigators for clarification and resolution.
Statistical analyses
Group-level data. Analyses of group-level data were performed using all published information, as well as any additional unpublished data retrieved from the investigators. Whenever investigators provided updated information besides the published data, we kept only the updated data to avoid double counting. Primary analyses compared allele frequencies for each polymorphism. We also addressed recessive and dominant models.
The odds ratio (OR) was used as the metric of choice. Heterogeneity across all eligible comparisons was tested using the 
2-based Cochran’s Q statistic (significant for P < 0.10) (10) and the I2 statistics (considered very large for I2 
75% and large for values of 50–74%) (11). Data were combined using both fixed effects [Mantel-Haenszel (12)] and random effects [DerSimonian and Laird (13)] models. Unless stated otherwise, random effects estimates are reported. Subgroup analyses estimated ORs per "racial" descent subgroup.
We also performed recursive cumulative meta-analysis to evaluate whether the summary OR for the allele contrast changed as more data accumulate (14, 15). We used the nonparametric 
correlation coefficient (16) to evaluate whether the magnitude of the observed association is related to the variance of each study ("small-study effects").
Individual-level data. Analyses of individual-level data focused on the A49G and CT60 polymorphisms for which meaningful amounts of data were available to be examined as haplotypes across a large number of studies. First, we examined whether analyses using group-level data from the studies with individual-level information gave similar results for each polymorphism as those obtained from the group-level analyses, including all studies. Then, the main analyses of individual-level data used haplotypes.
Haplotype reconstruction of the A49G and CT60 polymorphisms was performed using the population genotypic data separately for each case group (GD or HT) and controls of each participating team. The possible haplotypes are GG, GA, AG, and AA (the first allele corresponds to the A49G and the second one to the CT60 polymorphism, respectively). Haplotypes were inferred performing 100 iterations and using 100 individuals (randomly chosen) in each input file; 96.2% of the haplotypes were inferred with a probability exceeding 90%. Presented analyses used the most likely inferred haplotype for each subject. Analyses weighting each haplotype by its probability of inference yielded very similar results (data not shown).
Primary analyses used logistic regression to calculate in each study the OR per haplotype copy using the AA haplotype as reference. We then combined the natural logarithms of the ORs for each haplotype using an inverse variance random effects model. Between-study heterogeneity was measured with the Q and I2 statistics. Secondary analyses considered for each haplotype two variables instead of one (having one or two copies).
Individual-level data were used to calculate separately for each "racial" descent the attributable fraction (AF) (the complete list of data can be found in supplemental Appendix 1, which is published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org).
Analyses were conducted in Intercooled Stata 8.2 (Stata Corp., College Park, TX) using the meta and the metan module. Haplotype reconstruction was performed in PHASE 2.1 (17, 18) using the -T option. P values were two-tailed.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The electronic search yielded 157 articles. Of those, 114 were excluded (Fig. A1, published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). A total of 43 articles examining the relation of AITD and the CTLA-4 polymorphisms were eligible (3, 9, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59) (the complete list of articles can be found in supplemental Table A1, which is published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). There were 27 and nine studies that contained data on the A49G polymorphism and GD (3, 9, 21, 23, 25, 26, 27, 29, 30, 31, 32, 33, 36, 37, 38, 39, 42, 43, 44, 46, 47, 49, 50, 51, 52, 53, 57) and HT (3, 26, 30, 31, 35, 41, 48, 58, 59), respectively. One of the eligible articles included subjects from two different racial descent groups (37). Therefore, a total of 28 comparisons of published studies were considered for the A49G polymorphism and GD. Seven and three studies reported data for the CT60 polymorphisms and GD (3, 43, 48, 50, 52, 53, 54) and HT (3, 48, 54), respectively. There was considerable diversity of ethnic groups and eligibility criteria (the complete list of results can be found in supplemental Table A1, which is published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). For controls, varying details were presented regarding the extent of testing that had been performed to exclude controls with disturbed thyroid function. There were 19 studies (3, 9, 19, 20, 21, 22, 25, 29, 31, 32, 33, 35, 36, 38, 41, 46, 47, 48, 50, 51, 52, 53, 55, 56, 57, 59, 60) that excluded subjects with a family history of AITD and/or other autoimmune disorders from controls.
Seven studies matched for age (9, 38, 39, 43, 44, 48, 50), six for gender (9, 38, 43, 44, 48, 50), four for geographic region (32, 35, 42, 52), one for age, gender, and geographic region (50), one for ethnical descent, age, and gender (9), and five for ethnical descent (3, 19, 42, 52, 59). PCR methods were used for genotyping. No articles mentioned explicit blinding of the personnel that performed the genotyping. In three studies (two for A49G polymorphism and one for CT60), the distribution of genotypes in the control group deviated significantly from HWE (36, 38, 41).
There were 10 teams of investigators that provided individual-level genotyping data for both A49G and CT60 polymorphisms, all teams provided data for GD and five of them provided data for HT, as well. Four teams were from Europe and six from Asia. Nine teams had already published data on A49G polymorphism (3, 26, 28, 29, 37, 38, 43, 44, 47), but only two on CT60 polymorphism (3, 48). Three teams provided genotyping data from different cohorts than those previously published (29, 38, 48). Seven teams clarified that they had used blinding of personnel in genotyping.
Data on other CTLA-4 polymorphisms were more limited (the complete list of data can be found in supplemental Tables A17–A23, which are published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org).
Group-level data
A49G polymorphism.
The analyses included a total of 4848 cases with GD, 866 with HT, and 7314 controls (Table 1
) (the complete list of cases can be found in supplemental Table A2, which is published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). The frequency of the G allele was 48.9% among control subjects (62.0% and 36.9% among Asian and Caucasian descent, respectively). The overall prevalence was 26.5% (40.5% and 14.0%, respectively) for G/G homozygosity and 42.0% (40.2% and 42.7%, respectively) for G/A heterozygosity.
|
and Fig. 1A). Effect sizes were consistent across subgroups of different racial descent (Table 1)
|
The G allele conferred an almost 1.3-fold increase in the susceptibility of HT that was significant (P = 0.001), and there was substantial between-study heterogeneity (I2 = 54%; Table 1 and Fig. 1B). Similar effects were found for different ethnic groups (Table 1), with heterogeneity only in the Asian studies. When the analysis was limited to studies in which published or unpublished data were available for both A49G and CT60 polymorphisms, the summary OR for the G allele was still 1.31 (I2 = 0%). Sensitivity analysis excluding studies with unpublished data yielded similar results (OR 1.28; I2 = 65%). No studies had significant deviations from HWE in their control groups.
The magnitude of the overall OR diminished modestly over time for GD (from 1.64 in 1996, to 1.49 in the final analysis) and HT (from 1.57 in 1997, to 1.29 in the final analysis). There was no evidence that more precise studies showed more conservative results for the association of the G allele with GD or HT either, than less precise studies (P = 0.24 and P = 0.68 for GD and HT, respectively).
CT60 polymorphism.
Group-level data included 3047 GD cases, 839 HT cases, and 3741 healthy controls (Table 1) (the complete list of cases can be found in supplemental Table A5, which is published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). The G allele frequency was 64.3% in controls (76.0% in subjects of Asian descent and 52.7% in those of Caucasian descent), G/G homozygosity had a frequency of 44% (57.6 and 29.3%, respectively), while G/A heterozygotes were 41.9% (36.8 and 47.3%, respectively) of the controls.
The G allele increased 1.45-fold the odds of GD (P = 2 x 10–9); there was large heterogeneity (I2 = 56%) (Table 1 and Fig. 1C). The results were similar in subjects of Asian and Caucasian descent (Table 1), with between-study heterogeneity only in the Asian descent studies. Analyses limited to studies in which both A49G and CT60 had been genotyped yielded an OR of 1.45 (I2 = 56%) per allele. A sensitivity analysis including only previously published studies yielded similar results (OR 1.40; I2 = 57%). After excluding one study (52) that significantly deviated from HWE, neither the effect size (OR 1.43) nor the heterogeneity (I2 = 58%) changed.
For HT, the G allele of the CT60 polymorphism increased the odds 1.64-fold (P = 0.003) with very large heterogeneity (I2 = 83%) (Table 1 and Fig. 1D). The data came from Asian descent studies, with the exception of a single study on Caucasian descent subjects, in which a very strong effect was seen (OR 2.83). A sensitivity analysis including only previously published studies yielded similar results (OR 1.68 per G allele; I2 = 91%). All studies conformed to HWE.
Compared with the first study, the magnitude of the OR diminished slightly for GD (from 1.53 in 2003, to 1.45 in the last analysis) and more prominently for HT (from 2.83 in 2003, to 1.64 in the last analysis). No evidence was found that more precise studies showed more conservative results than less precise studies (P = 0.40 and P=0.19 for GD and HT, respectively).
Individual-level database
Individual-level data were available for 2306 GD cases, 657 HT cases, and 2530 controls for the A49G, and 2276 GD cases, 662 HT cases, and 2469 controls for the CT60 polymorphism. The allele and genotype frequency for each polymorphism were similar to those of the larger group-level database (the complete list of results can be found in supplemental Tables A8–A11, which are published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). Controls did not deviate significantly from HWE in any study teams. Moreover, the summary effect using group-level information from these studies on each polymorphism was similar to the aforementioned analyses, including all studies (the complete list of results can be found in supplemental Tables A12 and A13, which are published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org).
The database included a total of 5586 subjects: GD, 2334; HT, 680; and controls, 2572. The haplotype frequencies among controls for the: GG haplotype was 47.9% (63.1% in Asian descent controls, 38.4% in Caucasian descent, and 29.5% in Iranian descent, respectively); AG was 14.3% (11.7, 16.0, and 16.0%, respectively); GA was 1.6% (1.4, 1.6, and 2.0%, respectively); and AA was 36.2% (23.8, 44.0, and 52.5%, respectively) (the complete list of frequencies can be found in supplemental Tables A14 and A15, which are published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org).
For GD (Table 2 and Fig. 2A), the GG haplotype increased the odds by 1.49-fold per copy [95% confidence interval (CI) 1.31–1.70; P = 2 x 10–9; I2 = 48%)] compared with the AA haplotype. The result was consistent in Asian and Caucasian descent subjects (1.57 and 1.52, respectively). The AG haplotype also increased the odds of GD by 1.35-fold (Table 2 and Fig. 2B). No differences were found for Asian and Caucasian descent subgroups (1.30 and 1.42, respectively). The GA haplotype did not differ significantly in the GD risk overall (OR 0.78; Table 2 and Fig. 2C). After stratifying for CT60, the OR per copy of G allele of A49G was 1.13 (95% CI 0.98–1.31).
|
|
and Fig. 3A). No effect was shown for the AG haplotype, but the CIs cannot exclude a modest effect (OR 1.02; Table 2 and Fig. 3B). For both GG and AG, the effects were stronger in the subjects of Caucasian descent than in subjects of Asian descent (1.59 vs. 1.26 and 1.62 vs. 0.86, respectively). The GA haplotype had a nonsignificant effect, but data were very sparse (only 14 HT cases and 33 controls had the GA haplotype).
|
The AF according to the primary analysis was 32% for Asian descent and 29% for Caucasian descent subjects for GD, and 19 and 28%, respectively, for HT. Results were similar in the secondary analysis (AF = 24–39%).
Other polymorphisms
No formally significant results were seen overall for the 104-bp allele of the (AT)n microsatellite, the C(-318)T polymorphism, or the JO27_1 polymorphism, either for GD or HT (the complete list of results can be found in supplemental Appendix Tables A15–A17, A20, and A21, which are published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). The G allele of the JO31 polymorphism [four studies (5412 alleles)] and the G allele of the JO30 polymorphism [three studies [2796 alleles)] showed a possible association to GD, but the effects were modest (OR 1.40, 95% CI 1.15–1.72, P = 0.01, I2 = 57%; and OR 1.25, 95% CI 0.99–1.59, P = 0.06, I2 = 68%, respectively) (the complete list of results can be found in supplemental Tables A20, A21, and A23, which are published as supplemental data on The Endocrine Society’s Journals Online web site at http://jcem.endojournals.org). Moreover, JO31 and JO30 were in LD with the CT60 polymorphism (r2 = 0.65 and 0.79, respectively) and among themselves (r2 = 0.55).
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Although environmental agents are undoubtedly important for the development of AITD in susceptible individuals, it has been estimated in twin studies that almost 80% of the predisposition to GD is due to genetic factors (60). The genetics do not represent a simple Mendelian model (61). Several genes may be associated with AITD. The human leukocyte antigen region, CTLA-4 gene, and PTPN22 gene have shown the strongest results to date (62, 63, 64, 65). Given the observed OR and high allele frequencies in the examined populations, the AF for CT60 is in the range of 20–30%. We should acknowledge that genetic effects may occasionally be overestimated due to biases, however, this is one of the largest AFs ever shown in the genetics of complex diseases (66).
Functional evidence supports the role of CTLA-4 in autoimmunity. CTLA-4 was recently described as a gatekeeper of conjugation timing (67). Reduced conjugation might protect against prolonged contact periods of cytotoxic T lymphocytes with autoantigen-defined targets. CT60 G haplotypes produce less soluble CT60 transcript than A haplotypes (3). However, the disease-implicated haplotypes may extend over the long costimulatory receptor region of chromosome 2 consisting of CD28, CTLA-4, and ICOS (68). The CTLA-4 CT60 A protective allele haplotype goes with the most common extended haplotype (15-2-4) in Caucasians. The relative role of other polymorphisms and extended haplotypes may be further clarified in additional large-scale studies.
Some caveats should be discussed. Most of the group-level data showed large between-study heterogeneity. This could be due to bias, chance, or genuine diversity of genetic effects. We found no evidence of differences according to racial descent, even though the allele frequencies differed across racial subgroups. The smaller published studies did not show different results compared with the larger ones. However, there was a suggestion that the first studies may have yielded somewhat stronger effects (15). This is consistent with a "winner’s curse phenomenon" in which early data show exaggerated effects. Thus, the group-level derived estimates may be modestly inflated, as suggested also by the trend for relatively smaller effects sizes for the G allele of CT60 in the individual-level data. The latter may provide more reliable estimators and more options for deciphering the relative contribution of each polymorphism, but they are also not necessarily devoid from potential biases. Practical considerations did not allow shipping of specimens for regenotyping at a central facility. Nevertheless, genotyping error for SNPs should be low at experienced facilities. All control genotype frequencies were consistent with HWE in the individual-level database, but this was not so in the group-level data. However, some analyses have significant between-study heterogeneity, even in the individual-level data. Besides biases, this could be attributed to differences in terms of disease phenotype (e.g. presence of ophthalmopathy and/or of other autoimmune diseases) among AITD cases. AITD is rare in men to allow evaluation of gender differences. Moreover, data on age of onset were not sufficiently standardized across studies to allow meaningful investigation of age-related effect modification.
In conclusion, despite these caveats, our collaborative analysis shows consistent associations between GD and HT with CT60. This association crosses ethnic barriers, and we can make a reasonable estimate of the implicated OR. Although we still cannot identify a single etiological polymorphism, our study confirms the important role of the CTLA-4 locus in determining the risk of AITD.
Glossary of statistical terms
Effect size: The magnitude of the association (e.g. odds ratio).
"Fixed effects" model: Considers that the variability of the results between studies of the meta-analysis is exclusively due to random variation (chance). Therefore, if all the studies were infinitely large they would give identical results.
"Random effects" model: Assumes a different underlying effect for each study of the meta-analysis and takes this into consideration as an additional source of variation, leading to wider confidence intervals than the fixed effects model. The model tries to measure the mean and dispersion of these study-specific effects.
Heterogeneity: Denotes the diversity in a meta-analysis due to clinical differences (participants, interventions, outcomes) or methodological differences (study design, quality, analysis) among the included studies.
I2 (I-square): Describes the percentage of total variation across studies included in a meta-analysis that is due to heterogeneity rather than chance. I2 lies between 0% and 100%. A value of 0% indicates no observed heterogeneity, and larger values show increasing heterogeneity.
Attributable fraction: Denotes the proportion of a disease (or other outcome of interest) in the community that can be explained by the presence of a risk factor. It is mainly influenced by the prevalence of the risk factor, thus common risk factors can have high attributable fractions even when the effect size is not large.
|
Footnotes |
|---|
Authors’ contributions: F.K.K. and J.P.A.I. organized and coordinated the meta-analysis. All other authors were responsible for the design and conduct of individual-level studies and contributed their databases in the meta-analysis. F.K.K. and J.P.A.I. performed the statistical analyses, and F.K.K. wrote the first draft of the manuscript. All authors interpreted the results and commented on the manuscript.
Disclosure Statement: The authors have nothing to disclose.
Conflict of interest and source of funding: All authors declare that they have no conflict of interest.
Role of the funding source: The funding sources had no role in the study design, data collection, data analysis, results interpretation, and preparation of the manuscript or decision to publish.
First Published Online May 15, 2007
Abbreviations: AF, Attributable fraction; AITD, autoimmune thyroid disease; CI, confidence interval; CTLA-4, cytotoxic T-lymphocyte associated antigen 4; GD, Graves’ disease; HT, Hashimoto thyroiditis; HWE, Hardy-Weinberg equilibrium; LD, linkage disequilibrium; OR, odds ratio; SNP, single nucleotide polymorphism.
Received January 22, 2007.
Accepted May 9, 2007.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  J. Daroszewski, E. Pawlak, L. Karabon, I. Frydecka, A. Jonkisz, M. Slowik, and M. Bolanowski Soluble CTLA-4 receptor an immunological marker of Graves' disease and severity of ophthalmopathy is associated with CTLA-4 Jo31 and CT60 gene polymorphisms Eur. J. Endocrinol., November 1, 2009; 161(5): 787 - 793. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Kamesh, J. M. Heward, J. M. Williams, S. C. L. Gough, K.-M. Chavele, A. Salama, C. Pusey, C. O. S. Savage, and L. Harper CT60 and +49 polymorphisms of CTLA 4 are associated with ANCA-positive small vessel vasculitis Rheumatology, October 8, 2009; (2009) kep280v1. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. Sun, Z. Hu, H. Shen, and D. Lin Genetic Polymorphisms in Cytotoxic T-Lymphocyte Antigen 4 and Cancer: The Dialectical Nature of Subtle Human Immune Dysregulation Cancer Res., August 1, 2009; 69(15): 6011 - 6014. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. J Kahaly Polyglandular autoimmune syndromes Eur. J. Endocrinol., July 1, 2009; 161(1): 11 - 20. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Lantz, M. Abraham-Nordling, J. Svensson, G. Wallin, and B. Hallengren Immigration and the incidence of Graves' thyrotoxicosis, thyrotoxic multinodular goiter and solitary toxic adenoma Eur. J. Endocrinol., February 1, 2009; 160(2): 201 - 206. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Huber, F. Menconi, S. Corathers, E. M. Jacobson, and Y. Tomer Joint Genetic Susceptibility to Type 1 Diabetes and Autoimmune Thyroiditis: from Epidemiology to Mechanisms Endocr. Rev., October 1, 2008; 29(6): 697 - 725. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Ban, D. A. Greenberg, T. F. Davies, E. Jacobson, E. Concepcion, and Y. Tomer 'Linkage Analysis of Thyroid Antibody Production: Evidence for Shared Susceptibility to Clinical Autoimmune Thyroid Disease J. Clin. Endocrinol. Metab., September 1, 2008; 93(9): 3589 - 3596. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Hayashi, T. Kouki, N. Takasu, S. Sunagawa, and I. Komiya Association of an A/C single nucleotide polymorphism in programmed cell death-ligand 1 gene with Graves' disease in Japanese patients. Eur. J. Endocrinol., June 1, 2008; 158(6): 817 - 822. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Dideberg, G. Kristjansdottir, L. Milani, C. Libioulle, S. Sigurdsson, E. Louis, A.-C. Wiman, S. Vermeire, P. Rutgeerts, J. Belaiche, et al. An insertion deletion polymorphism in the Interferon Regulatory Factor 5 (IRF5) gene confers risk of inflammatory bowel diseases Hum. Mol. Genet., December 15, 2007; 16(24): 3008 - 3016. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
