• We are sorry, but NCBI web applications do not support your browser and may not function properly. More information
Logo of ajhgLink to Publisher's site
Am J Hum Genet. Dec 2005; 77(6): 1044–1060.
Published online Nov 1, 2005. doi:  10.1086/498651
PMCID: PMC1285162

Replication of Putative Candidate-Gene Associations with Rheumatoid Arthritis in >4,000 Samples from North America and Sweden: Association of Susceptibility with PTPN22, CTLA4, and PADI4

Abstract

Candidate-gene association studies in rheumatoid arthritis (RA) have lead to encouraging yet apparently inconsistent results. One explanation for the inconsistency is insufficient power to detect modest effects in the context of a low prior probability of a true effect. To overcome this limitation, we selected alleles with an increased probability of a disease association, on the basis of a review of the literature on RA and other autoimmune diseases, and tested them for association with RA susceptibility in a sample collection powered to detect modest genetic effects. We tested 17 alleles from 14 genes in 2,370 RA cases and 1,757 controls from the North American Rheumatoid Arthritis Consortium (NARAC) and the Swedish Epidemiological Investigation of Rheumatoid Arthritis (EIRA) collections. We found strong evidence of an association of PTPN22 with the development of anti-citrulline antibody–positive RA (odds ratio [OR] 1.49; P=.00002), using previously untested EIRA samples. We provide support for an association of CTLA4 (CT60 allele, OR 1.23; P=.001) and PADI4 (PADI4_94, OR 1.24; P=.001) with the development of RA, but only in the NARAC cohort. The CTLA4 association is stronger in patients with RA from both cohorts who are seropositive for anti-citrulline antibodies (P=.0006). Exploration of our data set with clinically relevant subsets of RA reveals that PTPN22 is associated with an earlier age at disease onset (P=.004) and that PTPN22 has a stronger effect in males than in females (P=.03). A meta-analysis failed to demonstrate an association of the remaining alleles with RA susceptibility, suggesting that the previously published associations may represent false-positive results. Given the strong statistical power to replicate a true-positive association in this study, our results provide support for PTPN22, CTLA4, and PADI4 as RA susceptibility genes and demonstrate novel associations with clinically relevant subsets of RA.

Introduction

Rheumatoid arthritis (RA [MIM 180300]) is a common autoimmune disease characterized by chronic, destructive, and debilitating arthritis (Gabriel et al. 1999). The etiology of RA is unknown, but it is thought to have both a genetic and an environmental basis (Firestein 2003). Autoantibodies are detected in ~2/3 of patients with RA and predict more-severe disease (van Zeben et al. 1992; van Jaarsveld et al. 1999; Kroot et al. 2000). The two major types of autoantibodies used clinically to create RA subsets are rheumatoid factor (RF), which is an immunoglobulin specific to the Fc region of IgG, and anti–cyclic citrullinated peptide (CCP) antibodies, which are antibodies directed against peptides in which arginine has been posttranslationally modified to become citrulline (Schellekens et al. 2000). These autoantibodies are strongly correlated but may represent distinct clinical subsets of RA. For decades, it has been known that HLA-DR alleles within the major histocompatibility complex (MHC) are associated with RA (Stastny and Fink 1977) and are likely to contribute ~1/3 of the genetic risk in RA (Deighton et al. 1989; Rigby et al. 1991). Although the MHC associations with RA are complex (Jawaheer et al. 2002; Newton et al. 2004), the majority of the genetic signal from the MHC is explained by multiple alleles at the HLA-DRB1 locus (MIM 142857) (Hall et al. 1996; Jawaheer et al. 2002). These alleles are known collectively as “shared epitope” (SE) alleles because of their sequence similarity within the third hypervariable region (amino acids 70–74: Leu-x-x-Glu-Arg/Lys [Gregersen et al. 1987]). Non-MHC genes are likely to contribute to RA pathogenesis and disease heterogeneity but have been much more difficult to identify.

Genetic association studies of RA applied to non-MHC candidate genes have led to encouraging yet apparently inconsistent results (see Genetic Association Database Web site). One explanation for the inconsistency is insufficient power to detect modest effects in the context of a low prior probability of a true effect and the failure to assign a stringent P value to interpret significance (Hirschhorn and Daly 2005; Wang et al. 2005). Thousands of patients are required to detect an allele with an odds ratio (OR) less than that observed for HLA-DRB1, and the vast majority of association studies of RA have been performed with <1,000 samples (notable exceptions include studies by Barton et al. [2004b], Begovich et al. [2004], Prokunina et al. [2004], Kochi et al. [2005], and Swanberg et al. [2005]). The inconsistent results and lack of replication make interpreting the significance of these association studies a challenge, and it is difficult to determine which results are true positives and which are false positives.

In the current study, we sought to test alleles with a prior putative association with RA or other autoimmune disorders in a collection of 4,127 samples (2,370 RA cases and 1,757 controls) powered to detect modest genetic effects (Hirschhorn et al. 2002; Lohmueller et al. 2003). Our goal was to determine which alleles replicate in our collection from Sweden and North America, thus providing evidence of a true-positive association. We first performed a thorough review of the literature on RA candidate-gene association studies, to identify alleles with suggestive (but not definitive) evidence of association with the development of RA. We also reviewed the literature to identify alleles with convincing evidence of association with autoimmune diseases other than RA, since we hypothesized that common pathogenic mechanisms may be shared among RA and other autoimmune diseases (Yamada and Ymamoto 2005). Here, we describe our results of testing 17 alleles in 14 genes for association with RA susceptibility in two independent clinical collections consisting of Swedish or North American individuals of European descent. Our results validate the hypothesis that the R620W variant of PTPN22 (MIM 600716) is associated with the development of RF-positive (RF+) and anti-CCP–positive (CCP+) RA and provide support for an association of RA with variants in PADI4 (MIM 605347) and CTLA4 (MIM 123890).

Material and Methods

Clinical Samples

Two case-control clinical collections were used to test all 17 alleles for association between genetic variation and the development of RA. Clinical characteristics for both collections are provided in table 1. All individuals are of European descent. In total, 4,127 samples (2,370 RA cases and 1,757 controls) were available for analysis. This study was approved by the MIT Broad Institute Institutional Review Board.

Table 1
Clinical Characteristics of EIRA and NARAC Cohorts[Note]

The first cohort, the Epidemiological Investigation of Rheumatoid Arthritis (EIRA), is an inception cohort collected in Sweden between May 1996 and February 2001. This cohort has been described in more detail elsewhere (Stolt et al. 2003; Padyukov et al. 2004). Briefly, a case is defined as an individual who fulfills the American College of Rheumatology (ACR) 1987 criteria for the classification of RA (Arnett et al. 1988) and who has never before been diagnosed with RA. For each potential case, a control subject was randomly selected from the study base, with the subject’s age, sex, and residential area taken into consideration. In total, 1,530 confirmed RA cases and 881 controls were available for analysis. Low-resolution HLA typing was performed on ~94% of the samples by use of the sequence-specific primer-PCR method (HLA-DR low-resolution kit) as described elsewhere (Olerup and Zetterquist 1992).

The second collection is the North American Rheumatoid Arthritis Consortium (NARAC). This collection of affected sibling pairs has been described in more detail elsewhere (Jawaheer et al. 2001, 2004; North American Rheumatoid Arthritis Consortium Web site). Briefly, a case is defined as an individual who fulfills the ACR 1987 criteria for RA and who has at least one sibling with RA. At least one sibling must have documented erosions on hand radiographs, and at least one sibling must have disease onset between the ages of 18 and 60 years. In total, 840 affected cases from 463 families were used in our analysis. Unrelated matched controls (n=867) were selected from 20,000 healthy individuals who were enrolled in the New York Cancer Project (NYCP) (see New York Cancer Project Web site). Two controls were matched to a single randomly chosen affected sibling on the basis of sex, age (birth decade), and ethnicity (grandparental country/region of origin). Complete data for PTPN22 (Begovich et al. 2004; Lee et al. 2005) and partial data for CARD15 (Addo et al. 2005) have been published elsewhere for this cohort. PADI4 genotype data were generated at the National Institute of Arthritis and Musculoskeletal and Skin Diseases (NIAMS) by use of an overlapping set of NARAC and NYCP samples: 601 NARAC cases and NYCP controls were common to the current data set, and 1,115 individuals were unique.

In addition to these two cohorts, an additional collection (“arthritis cohort 3”) of 1,417 patients with RA (with no matched controls) was used to investigate the influence of PTPN22 alleles on RA phenotype in males and females and to study age-at-onset effects. These samples were used in a case-only analysis comparing allele frequencies in autoantibody-positive (RF+ and/or CCP+) disease with those in autoantibody-negative (RF− and/or CCP−) disease. Consequently, no matched controls were necessary. Patient samples were obtained from the Wichita Rheumatic Disease Data Bank (Wolfe et al. 2003); The Arthritis, Rheumatism and Aging Medical Information System; the National Inception Cohort of Rheumatoid Arthritis Patients (Fries et al. 2002) (130 RF+ male cases, 401 RF+ female cases, 42 RF− male cases, and 139 RF− female cases); and the Study Of New Onset Rheumatoid Arthritis (SONORA) (Bombardier et al. 2002; S.O.N.O.R.A. Web site) (128 RF+ male cases, 313 RF+ female cases, 70 RF− male cases, and 194 RF− female cases). These RA sample sets were combined, and, for comparison with the EIRA and NARAC collections, only patients of European descent were analyzed (n=1,417 total RA cases).

Literature Review

We reviewed the literature published in English before December 31, 2004, to identify non-MHC alleles that had results suggesting a putative association with RA. Inclusion criteria for a putative susceptibility allele were (a) location within a non-MHC gene; (b) association with RA susceptibility—that is, the presence or absence of RA (we did not include an allele when the reported association was limited to a secondary phenotype, such as disease severity, because our primary purpose was to replicate previous associations with RA susceptibility); (c) a SNP or insertion-deletion (indel) polymorphism (i.e., no microsatellite markers); (d) genotyping in at least 200 patients with RA, with samples other than EIRA or NARAC; and (e) significance at a level <.001 in one study or <.05 in two or more studies (nominal P values based on allele frequencies in cases vs. controls). We identified >200 published association studies of RA. From these studies, we identified 10 alleles within nine genes that met our inclusion criteria for testing in our combined RA sample collection (table 2). Specific alleles at PADI4 (Suzuki et al. 2003), CTLA4 (Rodriguez et al. 2002; Vaidya et al. 2002), IL4 (Cantagrel et al. 1999; Maksymowych et al. 2002), SLC22A4 (Tokuhiro et al. 2003), MIF (Barton et al. 2003), and TNFRSF1B (Barton et al. 2001; Dieude et al. 2002; Kyogoku et al. 2003) have been associated with the development of RA in some studies but not in others (Dahlqvist et al. 2002; Fabris et al. 2002; Huang et al. 2002; Yen et al. 2003; Barton et al. 2004b; Miterski et al. 2004; Orozco et al. 2004; van der Helm-van Mil et al. 2004; Newman et al. 2005). Other genes (e.g., IL3 [Yamada et al. 2001], RUNX1 [Tokuhiro et al. 2003], and HAVCR1 [Chae et al. 2004]) have demonstrated a positive association in a single study but have not yet been tested in an independent collection for replication. Only PTPN22 has been convincingly replicated in more than one independent study (Begovich et al. 2004; Hinks et al. 2005; Lee et al. 2005; Orozco et al. 2005; Steer et al. 2005; Viken et al. 2005; Zhernakova et al. 2005), including the NARAC samples.

Table 2
Results Testing Our Primary Hypothesis That an Allele with Prior Evidence of an Association with RA Susceptibility or Another Autoimmune Disease Replicates in Our Collection of 4,127 Clinical Samples from the EIRA and NARAC Clinical Collections[Note]

Our replication study was limited to those alleles with the strongest evidence of a prior association, for efficiency and to minimize multiple-hypothesis testing. We chose the CT60 variant (rs3087243) of CTLA4, rather than the serine→threonine missense SNP at amino acid 17 (rs231775), because the CT60 variant clearly influences the risk of Graves disease and type 1 diabetes (Ueda et al. 2003). We chose the slc2F1 (rs2073838) allele of SLC22A4 because this allele had the strongest association with RA in a Japanese population (Tokuhiro et al. 2003). We chose the PADI4_94 allele (rs2240340) of PADI4 because it had the strongest association in a Japanese population (Suzuki et al. 2003). IL4 did not strictly fit the criteria, but it was included in our study because it is located within the 5q31 cytokine cluster associated with inflammatory bowel disease (Rioux et al. 2001) and is in very close proximity to SLC22A4 and IL3, both of which have reported evidence of association with RA. Although alleles within FCGR3A satisfied the inclusion criteria, they were not studied further, because FCGR3A was not amenable to genotyping by use of our platform. PTPN22 has been tested in NARAC, but we included it in our analysis to replicate the result in EIRA and to explore clinically relevant RA subsets and epistatic interactions.

We also hypothesized that an allele associated with another autoimmune disease has an increased prior probability of being associated with RA—compared with a random allele chosen from the genome—on the basis of shared pathogenic mechanisms among autoimmune diseases. We selected seven alleles from five additional genes with convincing evidence of causality in either inflammatory bowel disease (CARD15 [Hugot et al. 2001; Ogura et al. 2001], DLG5 [Stoll et al. 2004], and IBD5 [Rioux et al. 2001; Daly and Rioux 2004]) or type 1 diabetes (SUMO4 [Guo et al. 2004] and CTLA4 [Ueda et al. 2003]), although previous studies have not shown an association between CARD15 and RA (Ferreiros-Vidal et al. 2003; Steer et al. 2003; Addo et al. 2005).

Genotyping

Genotyping was performed, as described elsewhere (Gabriel et al. 2002), by primer extension of multiplex products with detection by matrix-assisted laser desorption/ionization time-of-flight mass spectroscopy (Tang et al. 1999) with the use of a Sequenom platform. All genotyping was performed at the Broad Institute, except for the PADI4 variant in NARAC, which was genotyped at the NIAMS. The average completeness of genotypes was 97.8% at the Broad Institute and 98.7% at the NIAMS. Using genotypes performed in replicate (n=3,814), we documented a consensus error rate of 0.5% in the analyzed data. All SNPs were in Hardy-Weinberg equilibrium within individual sample collections (P>.01).

Statistical Analysis

To test our primary hypothesis (that an allele is associated with the development of RA) and secondary hypotheses (that particular alleles are associated with a related RA phenotype reported previously in the literature) in both EIRA and NARAC samples, one-tailed P values were calculated using 2-by-2 contingency tables of allele frequencies and a Fisher’s exact test, and ORs were reported relative to the published “susceptibility” allele. Results from the EIRA and NARAC cohorts were combined by Mantel-Haenszel meta-analysis of the ORs (Lohmueller et al. 2003), because we hypothesized that a true susceptibility allele would have an effect in both populations, despite possible genetic, ethnic, or clinical differences (Ioannidis et al. 2004). A P value of .05 was considered significant for replication of an allele with previous evidence of association (Wacholder et al. 2004). Prior to combining the EIRA and NARAC results, we tested for homogeneity in the ORs, using a Pearson χ2 goodness-of-fit test as described elsewhere (Lohmueller et al. 2003). The EIRA and NARAC cohorts demonstrated P values for homogeneity >.01 for all loci except PTPN22 (P=.0002). PADI4 (P=.03) and CTLA4 (P=.08) demonstrated a trend toward significant difference. We also tested our primary hypothesis by using logistic regression, controlling for familial relatedness in NARAC, age, PTPN22 and SE allele status, and autoantibody status. Relatedness among siblings was handled by use of robust (Huber/White/sandwich) variance estimates (Williams 2000). Genotype data from the NARAC cohort for two genes (PTPN22 [Begovich et al. 2004; Lee et al. 2005] and CARD15 [Addo et al. 2005]) have been reported elsewhere. The NARAC genotypes are included here for completeness, although tests of significance were based on EIRA genotype data. Genetic models were tested using a log-linear model of sample allele frequencies conditional on disease status (see Genetic Association Models Web site). Power calculations were performed under a multiplicative model (α = 0.01) by use of the program by Purcell et al. (2003).

Exploratory and stratified analyses for PTPN22, CTLA4, and PADI4 were tested using a Fisher’s exact test and logistic regression. Stratified analyses for the association of PTPN22, CTLA4, and PADI4 with RA case/control status were performed for three sets of subgroups: male versus female, RF+ versus RF−, and CCP+ versus CCP−. Subsequently, RA phenotype analyses for the association of PTPN22, CTLA4, and PADI4 within each nonoverlapping subgroup included comparison of male RA cases with female RA cases, comparison of RF+ cases with RF− cases, and comparison of CCP+ cases with CCP− cases. In addition, RF+ and CCP+ cases were compared with all controls. Analysis of sex-specific effects for PTPN22 was performed by pooling the allele frequencies across all three clinical collections and comparing male RF+ cases with female RF+ cases (by logistic regression). Linear regression was used to test whether a genotype influenced age at disease onset. Epistatic interactions between pairs of alleles were tested among PTPN22, CTLA4, PADI4, and HLA-DRB1 SE alleles, as recently advocated (Marchini et al. 2005), by use of logistic regression to determine whether genotype distributions were statistically different between cases and controls. High-resolution genotyping was available for NARAC, and the SE alleles included HLA-DRB1*0101, *0401, *0404, *0405, *0408, and *1001. In the EIRA cohort, HLA-DRB1*01, *04, and *10 were considered SE alleles (Fries et al. 2002), since only low-resolution genotyping was available. On the basis of NARAC data, we estimated that the fraction of HLA-DRB1*01, *04, and *10 individuals in the EIRA cohort who carry the SE allele is ~92%, 96%, and 81%, respectively. To estimate the false-positive rate, we implemented a Bonferroni correction of our PTPN22 results for three hypotheses (age, sex, and epistasis with HLA) and of our CTLA4 and PADI4 results for five hypotheses (age, sex, autoantibody status, and epistasis with HLA or PTPN22).

We performed a meta-analysis of all studies with published allele-frequency data for individuals of European ancestry, together with our genotype data, as described elsewhere (Lohmueller et al. 2003). We first tested for homogeneity in the ORs across studies, using a Pearson χ2 goodness-of-fit test. When studies across all ethnic populations were analyzed together, we observed heterogeneity for IL4, TNFRSF1B, and MIF. To minimize heterogeneity, we excluded studies performed in non-European populations. Among studies of individuals of European ancestry, we observed no evidence of significant heterogeneity (P>.01). We calculated the pooled OR under a fixed-effects model (Mantel-Haenszel meta-analysis). The original hypothesis-generating publication was excluded from the meta-analysis, when applicable, to avoid bias from the null hypothesis of no effect (Hirschhorn et al. 2002). Genotype data were available for IL4 (Cantagrel et al. 1999), MIF (Miterski et al. 2004), TNFRSF1B (Barton et al. 2001; Dahlqvist et al. 2002; Fabris et al. 2002; Kyogoku et al. 2003; Yen et al. 2003; van der Helm-van Mil et al. 2004), CARD15 (Ferreiros-Vidal et al. 2003; Steer et al. 2003; Addo et al. 2005), and CTLA4 (Barton et al. 2004b; Orozco et al. 2004) but were unavailable from some studies (Dieude et al. 2002; Maksymowych et al. 2002; Newman et al. 2005). We excluded MIF data from Barton et al. (2003), since 25% of the cases with inflammatory polyarthritis failed to develop RA. Only a single study (Martinez et al. 2005) has tested the PADI4_94 variant (rs2240340)—the variant was not tested by Barton et al. (2004a). Across all studies, there were insufficient data on clinically relevant subsets of RA (e.g., subsets based on autoantibody status and sex) available to perform a more refined meta-analysis.

Results

Description of EIRA and NARAC Clinical Cohorts and Estimation of Study Power

To maximize our power to detect a true causal RA allele, we included >4,000 clinical samples from two distinct cohorts: the EIRA (Stolt et al. 2003; Padyukov et al. 2004) and the NARAC (Jawaheer et al. 2001, 2004). The clinical features of the EIRA and NARAC cohorts are shown in table 1. The combined resource of >4,000 total case-control samples provides good power to detect even modest genetic effects; we have >80% power to detect the published OR for 14 of the 17 alleles selected for replication and >50% power to detect an OR [gt-or-equal, slanted]1.25 for all alleles (see the “Material and Methods” section). Thus, failure to replicate in our clinical collection substantially reduces the probability that an allele is a true causal allele.

Selection of Candidate Alleles and Testing of Primary Hypothesis

To maximize further our ability to detect a true positive result, we selected alleles from the published genetic association studies of RA and other autoimmune diseases. Our goal for the literature review was to identify alleles with suggestive evidence of association with RA (P<.001 in one study or P<.05 in two or more studies) or convincing association with a different autoimmune disease. Our primary hypothesis was that an allele that satisfies these criteria is associated with susceptibility to RA in our collection. Applying our inclusion criteria (see the “Material and Methods” section) to the RA and autoimmune literature, we identified 17 alleles within 14 genes with a putative association with RA or another autoimmune disease (table 2). Ten alleles within nine genes had suggestive statistical evidence in favor of an association with the development of RA. Only the R620W variant of PTPN22 had been convincingly replicated in more than one study (Hinks et al. 2005; Lee et al. 2005; Orozco et al. 2005; Steer et al. 2005; Viken et al. 2005; Zhernakova et al. 2005). Seven alleles within five genes had convincing evidence in favor of an association with a chronic autoimmune disease (either inflammatory bowel disease or type 1 diabetes), although previous studies have not shown an association between one of those five genes, CARD15, and RA (Ferreiros-Vidal et al. 2003; Steer et al. 2003; Addo et al. 2005).

To determine whether any of the 17 alleles identified through our literature review is associated with RA susceptibility (primary hypothesis), we genotyped each allele in our combined clinical collection of >4,000 case-control samples. For each allele, we first analyzed the cohorts individually; second, we performed a formal test for heterogeneity in the OR across the two cohorts; and third, we combined the samples in a meta-analysis to provide the best estimate of the OR (under the assumption of a homogeneous effect across cohorts). Of the 17 alleles tested, 3 replicated association, with a onetailed P value <.05 (table 2) in the combined samples: PTPN22, PADI4, and CTLA4. PTPN22 had not been tested previously in the EIRA cohort and thus represents an independent replication. The data for each of the three genes fit best under a multiplicative genetic model (compared with general, dominant, and recessive models). The CTLA4 result remained significant after we controlled for self-reported thyroiditis (with which CTLA4 demonstrates a genetic association [Ueda et al. 2003]) in the NARAC cohort (OR 1.19; 95% CI 1.03–1.38; P=.02). We did observe a P value <.05 for SUMO4 and HAVCR1, but it was for the opposite allele than the published susceptibility allele. (Because we report our results in table 2 with respect to the published putative susceptibility allele, SUMO4 and HAVCR1 have ORs that are significantly <1.0 but have P values close to 1.) None of the other alleles demonstrated a significant association in these two cohorts (table 2), even after we controlled for the effects of HLA-DRB1 and PTPN22 (data not shown).

Selection and Testing of Secondary Hypotheses

RA is a heterogeneous disease, and it is possible that an allele is associated with only a clinical subset of RA. Therefore, we used our literature review to define specific secondary hypotheses to test in our RA clinical collection, to ensure that we did not miss a positive association. Each secondary hypothesis was based on data from a previous publication that suggested a specific genetic association. The secondary hypotheses relate either to an RA subphenotype or to a specific genetic model. We defined five secondary hypotheses (with published ORs): (1) PTPN22 is associated with RF+ but not RF− disease (OR 2.0) (Begovich et al. 2004; Lee et al. 2005); (2) homozygotes for the PADI4 susceptibility allele are more likely to be CCP+ than are the other genotypes combined (OR 3.25) (Suzuki et al. 2003); (3) individuals homozygous for the putative susceptibility alleles at both SLC22A4 and RUNX1 have a marked increase in RA disease risk under a dominant (but not epistatic) model (OR 9.03) (Tokuhiro et al. 2003); (4) TNFRSF1B is associated with familial RA but not sporadic RA (OR 1.69) (Barton et al. 2001; Dieude et al. 2002); and (5) IL3 is associated with RA in females with earlyonset disease (OR 1.89) (Yamada et al. 2001). No other secondary hypotheses with published statistical evidence were identified for the remaining 11 alleles. Given the high ORs, we have >80% power to detect each hypothesis, with the exception of the one regarding SLC22A4 and RUNX1, for which there is limited power to detect an association. We modified the hypothesis to ask whether individuals homozygous at RUNX1 and either heterozygous or homozygous at SLC22A4 (“susceptible”) were at an increased risk of developing RA, compared with all other individuals (“nonsusceptible”). We have >80% power to detect this hypothesis (estimated OR 1.52, on the basis of genotype data from Tokuhiro et al. [2003]).

Each of our five secondary hypotheses was tested in the EIRA and NARAC clinical collections. As shown in table 3, we replicated the finding that PTPN22 is associated with RF+ disease but not RF− disease, using the previously untested EIRA cohort (P=.0001 in EIRA). We extend this observation in autoantibodypositive disease to include an association with CCP+ cases but not CCP− cases (table 3) in the EIRA cohort, which is consistent with previously published clinical observations that these autoantibodies are strongly correlated. Using data combined from EIRA and NARAC, we observe a trend toward an association with PADI4 and anti-CCP antibody status, as previously suggested, but this was not statistically significant (table 4). We found no significant evidence to support an additive model between SLC22A4 and RUNX1 (table 5), although we have limited power to detect the model proposed by Tokuriho et al. (2003), given the allele frequencies observed in our population; we found no significant evidence to support an association between IL3 and RA in females with early-onset disease (table 6); and we found no significant evidence for an association of TNFRSF1B with familial RA, represented by the NARAC affected sibling pairs (table 2).

Table 3
Results Testing Our Secondary Hypothesis That PTPN22 Is Associated with Autoantibody-Positive (AutoAb+) but Not Autoantibody-Negative (AutoAb−) Disease in the EIRA Cohort[Note]
Table 4
Results Testing Our Secondary Hypothesis That PADI4 Is Associated with CCP+ but Not CCP− Disease in the Combined EIRA and NARAC Collections[Note]
Table 5
Results Testing Our Secondary Hypothesis That SLC22A4 and RUNX1 Susceptible Genotypes Additively Influence the Risk of RA[Note]
Table 6
Results Testing Our Secondary Hypothesis That IL3 Is Associated with Early-Onset RA in Females[Note]

Exploring Novel Hypotheses for PTPN22

Having established clear evidence for an association of RA susceptibility with PTPN22 in our samples, we sought to explore whether the R620W variant is associated with a clinically relevant subset of RA. Each exploratory hypothesis was defined on the basis of phenotypes of clinical relevance to RA, as established through epidemiological studies, animal models, or convincing genetic studies of the HLA-DRB1 SE alleles. The hypotheses were considered exploratory because no suggestive genetic association for the individual genes had been reported previously (in contrast to our primary and secondary hypotheses). Our three exploratory hypotheses include: (1) the allele has a greater effect in one sex than in the other, (2) the allele influences age at disease onset, and (3) the allele is epistatic with HLA-DRB1 SE alleles.

In an analysis of EIRA and NARAC cases, together with a separate case-only collection of 1,417 RA cases (“arthritis cohort 3” [Lee et al. 2005]), we found modest support that the PTPN22 susceptibility allele has a greater effect in males than in females (P=.03) (table 7). This effect was most prominent in the EIRA cohort and the arthritis cohort 3. We also observed an age-at-onset effect for PTPN22: carriers of the susceptible T allele develop RA 2 years earlier than do those not carrying the allele (T/T or C/T mean ± SD, 48.2 ± 14.0; C/C, 50.1 ± 14.3; P=.004). We found no evidence to support an epistatic interaction between PTPN22 and SE alleles (data not shown) in the EIRA cohort, which is consistent with previously published reports on NARAC (Begovich et al. 2004). After application of a conservative Bonferroni correction for three hypotheses, the sex effect is no longer significant at a P value <.05, whereas the age-at-onset effect remains significant (P=.012).

Table 7
Results Testing Our Exploratory Hypothesis That PTNP22 Has a Greater Effect on RA Susceptibility in Males Than in Females[Note]

Stratified Analyses to Explain Heterogeneity for PTPN22, CTLA4, and PADI4

We observed a difference in the magnitude of the genetic effect, as measured by the ORs, between the EIRA and NARAC cohorts: PTPN22 is stronger and more significant in NARAC than in EIRA, and CTLA4 and PADI4 are only significant in the NARAC cohort. As part of our initial analysis, we formally tested for heterogeneity between the ORs of the two cohorts, and we observed a clear significant difference for PTPN22 (P=.0002) and a trend toward significance for PADI4 (P=.03) and CTLA4 (P=.08). Because the ascertainment criteria and clinical characteristics between the two cohorts are distinct, we sought to explore whether a clinically relevant RA subset or an epistatic interaction might explain the difference in observed ORs. Therefore, we stratified our analysis by available clinically relevant subsets of RA (sex, autoantibody status, and age at onset) and epistasis with HLA and/or PTPN22.

Stratified analyses demonstrated that CTLA4 has a stronger association in patients positive for RF or CCP autoantibodies (table 8). This effect is most apparent in a pooled analysis of EIRA and NARAC, when CCP+ cases are compared with all controls (OR 1.18; 95% CI 1.07–1.31; P=.0006). The result remains significant after a Bonferroni correction for five hypotheses (P=.003). No other clinically relevant subset demonstrated a significant association, and we found no evidence of epistasis between any two combinations of CTLA4 with HLA or PTPN22 alleles and PADI4 with HLA or PTPN22 alleles (data not shown). Overall, however, we were unable to find any variable(s) to explain the OR differences for the three genes between the EIRA and NARAC cohorts.

Table 8
Results Testing Our Exploratory Hypothesis That CTLA4 Is Associated with CCP+ but Not CCP− Disease in Both the EIRA and NARAC Cohorts[Note]

Meta-Analysis with Published Data

Alleles with modest impact on disease susceptibility may be missed by any single study. We therefore performed a meta-analysis, combining our results with previously published results for individuals of European descent, to determine whether subtle effects across all published studies could be detected, under the hypothesis that a true susceptibility allele has an effect in different populations despite possible ethnic or clinical differences. The results, shown in table 9, fail to demonstrate an association of RA with CARD15, IL4, MIF, or TNFRSF1B. In contrast, the meta-analysis of previously published data with our data for CTLA4 (table 10) and PADI4 (combined OR 1.12; 95% CI 1.03–1.23 [Martinez et al. 2005]) demonstrates continued evidence of association with RA.

Table 9
Meta-Analysis of IL4, MIF, TNFRSF1B, and CARD15 Allele-Frequency Data[Note]
Table 10
Meta-Analysis of CTLA4 CT60 Allele (rs3087243)[Note]

Discussion

Genetic association is a powerful statistical methodology to identify variants that influence the risk of developing a common disease such as RA. However, inconsistent results across multiple studies make it difficult to interpret which results represent true-positive association and, therefore, which genes truly influence the risk of RA. One explanation for the inconsistent results is insufficient power to detect modest effects in the context of a low prior probability of a true effect (Hirschhorn and Daly 2005; Wang et al. 2005). To overcome this limitation, we designed a study to test alleles with an increased prior probability of disease association in a sample collection powered to detect modest genetic effects. Our results confirm the association of PTPN22 with autoantibody-positive RA and provide support for an association with CTLA4 and PADI4. Stratified analysis of our data reveals a novel association with age at onset (PTPN22), sex (PTPN22), and autoantibodypositive RA (CTLA4), results that need to be replicated in an independent data set.

Our results in the EIRA cohort, together with other replication studies (Hinks et al. 2005; Lee et al. 2005; Orozco et al. 2005; Steer et al. 2005; Viken et al. 2005; Zhernakova et al. 2005), confirm the hypothesis originally described by Begovich et al. (2004) in the NARAC cohort that the PTPN22 R620W variant influences the risk of autoantibody-positive RA. We extend the finding that PTPN22 is associated with RF+ disease to include CCP+ disease. This result is not surprising, given the strong correlation between RF and CCP status. However, it is interesting to note that a more significant result was obtained with CCP status (table 3), which is consistent with the clinical observation that CCP is a more specific marker for RA than is RF (Schellekens et al. 2000). Together, these results provide the most convincing evidence to date of a non-MHC gene that influences the risk of developing RA.

Our stratified analyses of PTPN22 revealed novel associations with clinically relevant subsets of RA. We identified an association of PTPN22 with age at onset (P=.004), such that carriers of the susceptible allele develop RA ~2 years earlier than noncarriers do. We observed a similar result for the HLA-DRB1 SE alleles and age at onset (P=.0002; data not shown). We also observed that PTPN22 has a greater effect in RF+ males than in RF+ females (P=.03), a result which is no longer significant at the P<.05 level when we correct for our three exploratory hypotheses. Recently, Orozco et al. (2005) reported a similar trend in male versus female patients with RA. Limited clinical data for RA suggest that sex differences indeed exist (Weyand et al. 1998; Kuiper et al. 2001; Tengstrand et al. 2004), findings corroborated by animal models of arthritis (Remmers et al. 2002; Adarichev et al. 2003; Meng et al. 2004). These novel findings require replication in independent sample collections.

Interpreting the significance of CTLA4 and PADI4 is more challenging, but our data support the hypothesis that genetic variation within these genes influences the risk of developing RA. Our results are significant only in the NARAC cohort, although a trend toward an association with CTLA4 is noted in the EIRA cohort. Because the cohorts are clinically different, we stratified our analysis by sex, autoantibody status, and age at onset. In support of a true-positive association for the CT60 allele of CTLA4, our results are strengthened in the EIRA cohort when only CCP+ cases are compared with all controls (combined OR 1.18; P=.0006) or with CCP− cases (OR 1.18; P=.007). This result, if replicated, may indicate that CTLA4 influences the development of RA only in CCP+ patients. Our results remain significant after a formal meta-analysis with the published RA literature that includes individuals of European ancestry. Recently, Lei et al. (2005) demonstrated an association between the CT60 allele and RA susceptibility in Han Chinese (OR 1.41; 95% CI 1.10–1.82; P=.005), providing additional support for a true-positive association.

An alternate explanation for the modest association of CTLA4 and PADI4 with RA in our collection is that our result is a false-positive. Possible explanations for a false-positive result include technical artifact, population stratification, or an inappropriate statistical threshold of significance. Our genotyping efficiency and accuracy argue strongly against technical artifact (see the “Material and Methods” section). Both cohorts were matched on the basis of ethnicity, which reduces the likelihood of stratification. However, it is becoming increasingly clear that subtle stratification exists even in well-matched cohorts (Freedman et al. 2004; Marchini et al. 2004). Extensive genotyping of genomic controls (Pritchard and Donnelly 2001), which is beyond the scope of this project, is necessary to assess the extent of stratification in each cohort. Ultimately, additional studies in sufficiently powered cohorts of patients with RA, together with an updated meta-analysis of the published literature, are required to determine definitively whether genetic variation within PADI4 and/or CTLA4 influences the risk of developing RA.

If our CTLA4 and PADI4 results in >4,000 samples are true positives, and the actual allelic OR is ~1.10 (as indicated by the pooled analysis), then it is not surprising that smaller studies have failed to detect a statistically significant result. An allele at ~40% population frequency with an OR of ~1.10 under a multiplicative model requires ~2,700 cases and ~2,700 controls to have 80% power to detect a true-positive result (type I error rate 0.05) (see Genetic Power Calculator Web site), and previous studies were conducted with fewer samples than this (Barton et al. 2004a, 2004b; Orozco et al. 2004; Martinez et al. 2005). Consistent with this explanation, previous studies have shown that the putative susceptibility alleles were marginally increased in patients with RA, but not to a level that reached statistical significance (see table 10 for CTLA4 meta-analysis).

Our finding that the majority of previous association studies in RA may represent false-positive results is consistent with the published literature of other complex genetic diseases (Ioannidis et al. 2001; Lohmueller et al. 2003). Although it is possible that some of our results are false negatives, we have >80% power to detect the published OR in 14 of the 17 alleles selected for replication (and >50% for all alleles). We selected a nominal P value of .05 as significant, to minimize false-positive results and to maximize our confidence in a true-positive association. On the basis of empiric and theoretical data in complex traits, others have advocated similar replication P values (Lohmueller et al. 2003; Wacholder et al. 2004). Furthermore, a formal meta-analysis failed to detect a subtle effect that may have been missed by our replication in the NARAC and EIRA cohorts (table 9). Although we observed no significant evidence of genetic heterogeneity across studies, we excluded individuals of non-European descent. Of note, there are no formal tests of heterogeneity in patient selection criteria, and it may be that an alternate analytic method is necessary to combine effects across studies.

Other explanations for the lack of replication in our collection include genetic, ethnic, and clinical heterogeneity across study populations and the possibility that we did not test the actual causal allele in a given gene. It is possible that a true susceptibility allele in one ethnic population has no effect in a different ethnic population (“ethnic heterogeneity”). There is little evidence to support this hypothesis (Ioannidis et al. 2004), although a thorough evaluation is not yet possible, given the limited number of true susceptibility alleles tested across multiple ethnic populations. Differences in clinical features or ascertainment criteria across RA collections may account for the inability to replicate. For example, we observe a more significant association between RA and CTLA4 and PTPN22 alleles when CCP status is taken into consideration. In an attempt to control for clinical differences, we used our literature review to define secondary hypotheses that test for relevant subsets, but we found no evidence of association. Finally, we cannot exclude that other variants within these genes may be associated with RA. We selected alleles with the most-significant association from a previous study. If we assume that an allele is either the putative causal allele or in very strong LD with the causal allele, then our study design should replicate the original finding. Comprehensive testing of genetic variation within each gene may reveal alternate alleles that contribute to RA susceptibility, as recently suggested by a study of PTPN22 (Carlton et al. 2005).

We observed a difference in the ORs between the EIRA and NARAC cohorts for PTPN22, CTLA4, and PADI4. There are two possible explanations for the differences. First, the effect in NARAC and EIRA could actually be the same, with the different ORs simply reflecting sampling variance across the two populations. This is clearly not the case for PTPN22, which demonstrates a statistically significant difference between the two cohorts. Second, the OR in NARAC and EIRA could indeed be different as a result of ascertainment, clinical severity, “genetic loading” of affected sibling pairs, or an unmeasured variable. Our stratified analyses failed to identify any clear explanation for the difference, although we were unable to correct for disease severity, because of the lack of a suitable clinical variable in EIRA (e.g., disease activity score or radiographic erosions). The statistically significant differences in the ORs observed for CTLA4 and PADI4 are sufficiently modest such that we do not believe it is possible to determine these two association models at present. Future studies aimed at identification of covariates will be important for understanding the impact of these genes on RA clinical heterogeneity.

We did observe a P value <.05 for HACVR1 and SUMO4, but not with the published susceptibility allele. Therefore, these results may represent statistical fluctuation rather than a true-positive association. Recent data have called into question the validity of the SUMO4 association with type 1 diabetes (Park et al. 2005; Qu et al. 2005; Smyth et al. 2005). It should also be noted that the allele frequency of the SLC22A4 variant tested is significantly lower in our collection (minor-allele frequency [MAF]≈0.08) than in the Japanese population (MAF≈0.31), which limits our power to detect a true association (especially under a recessive genetic model). Studies in additional populations are necessary to test whether these or alternate alleles are associated with RA.

In conclusion, our results replicate the finding that PTPN22 is associated with RF+ RA and suggest the novel hypotheses that the allele influences age at RA onset and has a greater effect in males than in females. In addition, our study provides support for association of RA with CTLA4 and PADI4, a result that will need to be confirmed in additional studies. We find no evidence to support an association for the remaining alleles, which suggests that these alleles are not associated with RA with an OR [gt-or-equal, slanted]1.10. These results add to our understanding of the genetic basis of RA and emphasize the need for large collections of patients, updated meta-analysis, and stringent interpretation of statistical significance in future RA candidate-gene association studies.

Acknowledgments

R.M.P. is supported for this work by the National Institutes of Health (NIH) (grant K08 AI 55314). P.K.G. is supported for this work by the NIH (grants NO1-AR-2-2263 and R01-AR44222) and the National Arthritis Foundation. J.D.R. is supported by the National Institute of Diabetes and Digestive and Kidney Diseases and by the Crohn's and Colitis Foundation of America. This research was supported in part by the Intramural Research Program of the National Institute of Arthritis and Musculoskeletal and Skin Diseases of the NIH. E.W.K. is supported by NIH grants R01 AR49880-01, K24 AR0524-01, P60 AR47782-05. We are grateful to Abbott Laboratories, for the support of the SONORA study, and to Clair Bombardier, Elena Massarotti, and Michael Weisman, for their critical roles in the design and implementation of the SONORA cohort. We appreciate Julie Le’s technical contribution to genotyping done at the National Institute of Arthritis and Musculoskeletal and Skin Diseases and Noel Burtt and Christine Curley’s technical assistance at the Broad Institute.

Web Resources

The URLs for data presented herein are as follows:

Genetic Association Database, http://geneticassociationdb.nih.gov/
Genetic Association Models, http://statgen.iop.kcl.ac.uk/gpc/model/
Genetic Power Calculator, http://statgen.iop.kcl.ac.uk/gpc/
New York Cancer Project (NYPC), http://www.amdec.org/amdec_initiatives/nycp.html
North American Rheumatoid Arthritis Consortium (NARAC), http://www.naracdata.org/
Online Mendelian Inheritance in Man (OMIM), http://www.ncbi.nlm.nih.gov/Omim/ (for RA, HLA-DRB1, PTPN22, PADI4, and CTLA4)
S.O.N.O.R.A.—Study of New Onset Rheumatoid Arthritis, http://www.abbottimmunology.com/ri_sonora.asp

References

Adarichev VA, Nesterovitch AB, Bardos T, Biesczat D, Chandrasekaran R, Vermes C, Mikecz K, Finnegan A, Glant TT (2003) Sex effect on clinical and immunologic quantitative trait loci in a murine model of rheumatoid arthritis. Arthritis Rheum 48:1708–1720 [PubMed] [Cross Ref]10.1002/art.11016
Addo A, Le J, Li W, Aksentijevich I, Balow JJ, Lee A, Gregersen P, Kastner D, Remmers E (2005) Analysis of CARD15/NOD2 haplotypes fails to identify common variants associated with rheumatoid arthritis susceptibility. Scand J Rheumatol 34:198–203 [PubMed] [Cross Ref]10.1080/03009740510018561
Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, Healey LA, et al (1988) The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 31:315–324 [PubMed]
Barton A, Bowes J, Eyre S, Spreckley K, Hinks A, John S, Worthington J (2004a) A functional haplotype of the PADI4 gene associated with rheumatoid arthritis in a Japanese population is not associated in a United Kingdom population. Arthritis Rheum 50:1117–1121 [PubMed] [Cross Ref]10.1002/art.20169
Barton A, John S, Ollier WE, Silman A, Worthington J (2001) Association between rheumatoid arthritis and polymorphism of tumor necrosis factor receptor II, but not tumor necrosis factor receptor I, in Caucasians. Arthritis Rheum 44:61–65 [PubMed] [Cross Ref]10.1002/1529-0131(200101)44:1<61::AID-ANR9>3.0.CO;2-Q
Barton A, Jury F, Eyre S, Bowes J, Hinks A, Ward D, Worthington J (2004b) Haplotype analysis in simplex families and novel analytic approaches in a case-control cohort reveal no evidence of association of the CTLA-4 gene with rheumatoid arthritis. Arthritis Rheum 50:748–752 [PubMed] [Cross Ref]10.1002/art.20118
Barton A, Lamb R, Symmons D, Silman A, Thomson W, Worthington J, Donn R (2003) Macrophage migration inhibitory factor (MIF) gene polymorphism is associated with susceptibility to but not severity of inflammatory polyarthritis. Genes Immun 4:487–491 [PubMed] [Cross Ref]10.1038/sj.gene.6364014
Begovich AB, Carlton VE, Honigberg LA, Schrodi SJ, Chokkalingam AP, Alexander HC, Ardlie KG, et al (2004) A missense single-nucleotide polymorphism in a gene encoding a protein tyrosine phosphatase (PTPN22) is associated with rheumatoid arthritis. Am J Hum Genet 75:330–337 [PMC free article] [PubMed]
Bombardier C, Deaton R, Gregersen PK, Massarotti EM, Formica C, Weisman M (2002) Pattern of DMARD use in a North American cohort of patients with early rheumatoid arthritis (SONORA) [abstract]. Arthritis Rheum 46:S344
Cantagrel A, Navaux F, Loubet-Lescoulie P, Nourhashemi F, Enault G, Abbal M, Constantin A, Laroche M, Mazieres B (1999) Interleukin-1β, interleukin-1 receptor antagonist, interleukin-4, and interleukin-10 gene polymorphisms: relationship to occurrence and severity of rheumatoid arthritis. Arthritis Rheum 42:1093–1100 [PubMed] [Cross Ref]10.1002/1529-0131(199906)42:6<1093::AID-ANR5>3.0.CO;2-P
Carlton VE, Hu X, Chokkalingam AP, Schrodi SJ, Brandon R, Alexander HC, Chang M, Catanese JJ, Leong DU, Ardlie KG, Kastner DL, Seldin MF, Criswell LA, Gregersen PK, Beasley E, Thomson G, Amos CI, Begovich AB (2005) PTPN22 genetic variation: evidence for multiple variants associated with rheumatoid arthritis. Am J Hum Genet 77:567–581 [PMC free article] [PubMed]
Chae SC, Song JH, Shim SC, Yoon KS, Chung HT (2004) The exon 4 variations of Tim-1 gene are associated with rheumatoid arthritis in a Korean population. Biochem Biophys Res Commun 315:971–975 [PubMed] [Cross Ref]10.1016/j.bbrc.2004.01.154
Dahlqvist SR, Arlestig L, Sikstrom C, Linghult S (2002) Tumor necrosis factor receptor type II (exon 6) and interleukin-6 (-174) gene polymorphisms are not associated with family history but tumor necrosis factor receptor type II is associated with hypertension in patients with rheumatoid arthritis from northern Sweden. Arthritis Rheum 46:3096–3098 [PubMed] [Cross Ref]10.1002/art.10592
Daly MJ, Rioux JD (2004) New approaches to gene hunting in IBD. Inflamm Bowel Dis 10:312–317 [PubMed] [Cross Ref]10.1097/00054725-200405000-00020
Deighton CM, Walker DJ, Griffiths ID, Roberts DF (1989) The contribution of HLA to rheumatoid arthritis. Clin Genet 36:178–182 [PubMed]
Dieude P, Petit E, Cailleau-Moindrault S, Osorio J, Pierlot C, Martinez M, Faure S, Alibert O, Lasbleiz S, De Toma C, Bardin T, Prum B, Cornelis F (2002) Association between tumor necrosis factor receptor II and familial, but not sporadic, rheumatoid arthritis: evidence for genetic heterogeneity. Arthritis Rheum 46:2039–2044 [PubMed] [Cross Ref]10.1002/art.10101
Fabris M, Tolusso B, Di Poi E, Assaloni R, Sinigaglia L, Ferraccioli G (2002) Tumor necrosis factor-α receptor II polymorphism in patients from southern Europe with mild-moderate and severe rheumatoid arthritis. J Rheumatol 29:1847–1850 [PubMed]
Ferreiros-Vidal I, Barros F, Pablos JL, Carracedo A, Gomez-Reino JJ, Gonzalez A (2003) CARD15/NOD2 analysis in rheumatoid arthritis susceptibility. Rheumatology (Oxford) 42:1380–1382 [PubMed]
Firestein GS (2003) Evolving concepts of rheumatoid arthritis. Nature 423:356–361 [PubMed] [Cross Ref]10.1038/nature01661
Freedman ML, Reich D, Penney KL, McDonald GJ, Mignault AA, Patterson N, Gabriel SB, Topol EJ, Smoller JW, Pato CN, Pato MT, Petryshen TL, Kolonel LN, Lander ES, Sklar P, Henderson B, Hirschhorn JN, Altshuler D (2004) Assessing the impact of population stratification on genetic association studies. Nat Genet 36:388–393 [PubMed] [Cross Ref]10.1038/ng1333
Fries JF, Wolfe F, Apple R, Erlich H, Bugawan T, Holmes T, Bruce B (2002) HLA-DRB1 genotype associations in 793 white patients from a rheumatoid arthritis inception cohort: frequency, severity, and treatment bias. Arthritis Rheum 46:2320–2329 [PubMed] [Cross Ref]10.1002/art.10485
Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, Higgins J, DeFelice M, Lochner A, Faggart M, Liu-Cordero SN, Rotimi C, Adeyemo A, Cooper R, Ward R, Lander ES, Daly MJ, Altshuler D (2002) The structure of haplotype blocks in the human genome. Science 296:2225–2229 [PubMed] [Cross Ref]10.1126/science.1069424
Gabriel SE, Crowson CS, O’Fallon WM (1999) The epidemiology of rheumatoid arthritis in Rochester, Minnesota, 1955–1985. Arthritis Rheum 42:415–420 [PubMed] [Cross Ref]10.1002/1529-0131(199904)42:3<415::AID-ANR4>3.0.CO;2-Z
Gregersen PK, Silver J, Winchester RJ (1987) The shared epitope hypothesis: an approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum 30:1205–1213 [PubMed]
Guo D, Li M, Zhang Y, Yang P, Eckenrode S, Hopkins D, Zheng W, et al (2004) A functional variant of SUMO4, a new IκBα modifier, is associated with type 1 diabetes. Nat Genet 36:837–841 [PubMed] [Cross Ref]10.1038/ng1391
Hall FC, Weeks DE, Camilleri JP, Williams LA, Amos N, Darke C, Gibson K, Pile K, Wordsworth BP, Jessop JD (1996) Influence of the HLA-DRB1 locus on susceptibility and severity in rheumatoid arthritis. QJM 89:821–829 [PubMed]
Hinks A, Barton A, John S, Bruce I, Hawkins C, Griffiths CE, Donn R, Thomson W, Silman A, Worthington J (2005) Association between the PTPN22 gene and rheumatoid arthritis and juvenile idiopathic arthritis in a UK population: further support that PTPN22 is an autoimmunity gene. Arthritis Rheum 52:1694–1699 [PubMed] [Cross Ref]10.1002/art.21049
Hirschhorn JN, Daly MJ (2005) Genome-wide association studies for common diseases and complex traits. Nat Rev Genet 6:95–108 [PubMed] [Cross Ref]10.1038/nrg1521
Hirschhorn JN, Lohmueller K, Byrne E, Hirschhorn K (2002) A comprehensive review of genetic association studies. Genet Med 4:45–61 [PubMed]
Huang CM, Wu MC, Wu JY, Tsai FJ (2002) No association of interleukin-4 gene polymorphisms in Chinese patients with rheumatoid arthritis in Taiwan. Clin Exp Rheumatol 20:871–872 [PubMed]
Hugot JP, Chamaillard M, Zouali H, Lesage S, Cezard JP, Belaiche J, Almer S, Tysk C, O’Morain CA, Gassull M, Binder V, Finkel Y, Cortot A, Modigliani R, Laurent-Puig P, Gower-Rousseau C, Macry J, Colombel JF, Sahbatou M, Thomas G (2001) Association of NOD2 leucine-rich repeat variants with susceptibility to Crohn’s disease. Nature 411:599–603 [PubMed] [Cross Ref]10.1038/35079107
Ioannidis JP, Ntzani EE, Trikalinos TA (2004) “Racial” differences in genetic effects for complex diseases. Nat Genet 36:1312–1318 [PubMed] [Cross Ref]10.1038/ng1474
Ioannidis JP, Ntzani EE, Trikalinos TA, Contopoulos-Ioannidis DG (2001) Replication validity of genetic association studies. Nat Genet 29:306–309 [PubMed] [Cross Ref]10.1038/ng749
Jawaheer D, Li W, Graham RR, Chen W, Damle A, Xiao X, Monteiro J, Khalili H, Lee A, Lundsten R, Begovich A, Bugawan T, Erlich H, Elder JT, Criswell LA, Seldin MF, Amos CI, Behrens TW, Gregersen PK (2002) Dissecting the genetic complexity of the association between human leukocyte antigens and rheumatoid arthritis. Am J Hum Genet 71:585–594 [PMC free article] [PubMed]
Jawaheer D, Lum RF, Amos CI, Gregersen PK, Criswell LA (2004) Clustering of disease features within 512 multicase rheumatoid arthritis families. Arthritis Rheum 50:736–741 [PubMed] [Cross Ref]10.1002/art.20066
Jawaheer D, Seldin MF, Amos CI, Chen WV, Shigeta R, Monteiro J, Kern M, Criswell LA, Albani S, Nelson JL, Clegg DO, Pope R, Schroeder HW Jr, Bridges SL Jr, Pisetsky DS, Ward R, Kastner DL, Wilder RL, Pincus T, Callahan LF, Flemming D, Wener MH, Gregersen PK (2001) A genomewide screen in multiplex rheumatoid arthritis families suggests genetic overlap with other autoimmune diseases. Am J Hum Genet 68:927–936 [PMC free article] [PubMed]
Kochi Y, Yamada R, Suzuki A, Harley JB, Shirasawa S, Sawada T, Bae SC, Tokuhiro S, Chang X, Sekine A, Takahashi A, Tsunoda T, Ohnishi Y, Kaufman KM, Kang CP, Kang C, Otsubo S, Yumura W, Mimori A, Koike T, Nakamura Y, Sasazuki T, Yamamoto K (2005) A functional variant in FCRL3, encoding Fc receptor-like 3, is associated with rheumatoid arthritis and several autoimmunities. Nat Genet 37:478–485 [PMC free article] [PubMed] [Cross Ref]10.1038/ng1540
Kroot EJ, de Jong BA, van Leeuwen MA, Swinkels H, van den Hoogen FH, van’t Hof M, van de Putte LB, van Rijswijk MH, van Venrooij WJ, van Riel PL (2000) The prognostic value of anti-cyclic citrullinated peptide antibody in patients with recent-onset rheumatoid arthritis. Arthritis Rheum 43:1831–1835 [PubMed] [Cross Ref]10.1002/1529-0131(200008)43:8<1831::AID-ANR19>3.0.CO;2-6
Kuiper S, van Gestel AM, Swinkels HL, de Boo TM, da Silva JA, van Riel PL (2001) Influence of sex, age, and menopausal state on the course of early rheumatoid arthritis. J Rheumatol 28:1809–1816 [PubMed]
Kyogoku C, Tsuchiya N, Shibue T, Tokunaga K, Matsuta K (2003) TNFR2 position 196 polymorphism in Japanese patients with rheumatoid arthritis: comment on the article by Dieude et al. Arthritis Rheum 48:273–274 [PubMed] [Cross Ref]10.1002/art.10600
Lee AT, Li W, Liew A, Bombardier C, Weisman M, Massarotti EM, Kent J, Wolfe F, Begovich AB, Gregersen PK (2005) The PTPN22 R620W polymorphism associates with RF positive rheumatoid arthritis in a dose-dependent manner but not with HLA-SE status. Genes Immun 6:129–133 [PubMed] [Cross Ref]10.1038/sj.gene.6364159
Lei C, Dongqing Z, Yeqing S, Oaks MK, Lishan C, Jianzhong J, Jie Q, Fang D, Ningli L, Xinghai H, Daming R (2005) Association of the CTLA-4 gene with rheumatoid arthritis in Chinese Han population. Eur J Hum Genet 13:823–828 [PubMed] [Cross Ref]10.1038/sj.ejhg.5201423
Lohmueller KE, Pearce CL, Pike M, Lander ES, Hirschhorn JN (2003) Meta-analysis of genetic association studies supports a contribution of common variants to susceptibility to common disease. Nat Genet 33:177–182 [PubMed] [Cross Ref]10.1038/ng1071
Maksymowych WP, Suarez-Almazor ME, Buenviaje H, Cooper BL, Degeus C, Thompson M, Russell AS (2002) HLA and cytokine gene polymorphisms in relation to occurrence of palindromic rheumatism and its progression to rheumatoid arthritis. J Rheumatol 29:2319–2326 [PubMed]
Marchini J, Cardon LR, Phillips MS, Donnelly P (2004) The effects of human population structure on large genetic association studies. Nat Genet 36:512–517 [PubMed] [Cross Ref]10.1038/ng1337
Marchini J, Donnelly P, Cardon LR (2005) Genome-wide strategies for detecting multiple loci that influence complex diseases. Nat Genet 37:413–417 [PubMed] [Cross Ref]10.1038/ng1537
Martinez A, Valdivia A, Pascual-Salcedo D, Lamas JR, Fernandez-Arquero M, Balsa A, Fernandez-Gutierrez B, de la Concha EG, Urcelay E (2005) PADI4 polymorphisms are not associated with rheumatoid arthritis in the Spanish population. Rheumatology (Oxford) 44:1263–1266 [PubMed]
Meng HC, Griffiths MM, Remmers EF, Kawahito Y, Li W, Neisa R, Cannon GW, Wilder RL, Gulko PS (2004) Identification of two novel female-specific non-major histocompatibility complex loci regulating collagen-induced arthritis severity and chronicity, and evidence of epistasis. Arthritis Rheum 50:2695–2705 [PubMed] [Cross Ref]10.1002/art.20366
Miterski B, Drynda S, Boschow G, Klein W, Oppermann J, Kekow J, Epplen JT (2004) Complex genetic predisposition in adult and juvenile rheumatoid arthritis. BMC Genet 5:2 [PMC free article] [PubMed] [Cross Ref]10.1186/1471-2156-5-2
Newman B, Wintle RF, van Oene M, Yazdanpanah M, Owen J, Johnson B, Gu X, Amos CI, Keystone E, Rubin LA, Siminovitch KA (2005) SLC22A4 polymorphisms implicated in rheumatoid arthritis and Crohn’s disease are not associated with rheumatoid arthritis in a Canadian Caucasian population. Arthritis Rheum 52:425–429 [PubMed] [Cross Ref]10.1002/art.20854
Newton JL, Harney SM, Timms AE, Sims AM, Rockett K, Darke C, Wordsworth BP, Kwiatkowski D, Brown MA (2004) Dissection of class III major histocompatibility complex haplotypes associated with rheumatoid arthritis. Arthritis Rheum 50:2122–2129 [PubMed] [Cross Ref]10.1002/art.20358
Ogura Y, Bonen DK, Inohara N, Nicolae DL, Chen FF, Ramos R, Britton H, Moran T, Karaliuskas R, Duerr RH, Achkar JP, Brant SR, Bayless TM, Kirschner BS, Hanauer SB, Nunez G, Cho JH (2001) A frameshift mutation in NOD2 associated with susceptibility to Crohn’s disease. Nature 411:603–606 [PubMed] [Cross Ref]10.1038/35079114
Olerup O, Zetterquist H (1992) HLA-DR typing by PCR amplification with sequence-specific primers (PCR-SSP) in 2 hours: an alternative to serological DR typing in clinical practice including donor-recipient matching in cadaveric transplantation. Tissue Antigens 39:225–235 [PubMed]
Orozco G, Sanchez E, Gonzalez-Gay MA, Lopez-Nevot MA, Torres B, Caliz R, Ortego-Centeno N, Jimenez-Alonso J, Pascual-Salcedo D, Balsa A, de Pablo R, Nunez-Roldan A, Gonzalez-Escribano MF, Martin J (2005) Association of a functional single-nucleotide polymorphism of PTPN22, encoding lymphoid protein phosphatase, with rheumatoid arthritis and systemic lupus erythematosus. Arthritis Rheum 52:219–224 [PubMed] [Cross Ref]10.1002/art.20771
Orozco G, Torres B, Nunez-Roldan A, Gonzalez-Escribano MF, Martin J (2004) Cytotoxic T-lymphocyte antigen-4-CT60 polymorphism in rheumatoid arthritis. Tissue Antigens 64:667–670 [PubMed] [Cross Ref]10.1111/j.1399-0039.2004.00318.x
Padyukov L, Silva C, Stolt P, Alfredsson L, Klareskog L (2004) A gene-environment interaction between smoking and shared epitope genes in HLA-DR provides a high risk of seropositive rheumatoid arthritis. Arthritis Rheum 50:3085–3092 [PubMed] [Cross Ref]10.1002/art.20553
Park Y, Park S, Kang J, Yang S, Kim D (2005) Assessing the validity of the association between the SUMO4 M55V variant and risk of type 1 diabetes. Nat Genet 37:112 [PubMed] [Cross Ref]10.1038/ng0205-112a
Pritchard JK, Donnelly P (2001) Case-control studies of association in structured or admixed populations. Theor Popul Biol 60:227–237 [PubMed] [Cross Ref]10.1006/tpbi.2001.1543
Prokunina L, Padyukov L, Bennet A, de Faire U, Wiman B, Prince J, Alfredsson L, Klareskog L, Alarcon-Riquelme M (2004) Association of the PD-1.3A allele of the PDCD1 gene in patients with rheumatoid arthritis negative for rheumatoid factor and the shared epitope. Arthritis Rheum 50:1770–1773 [PubMed] [Cross Ref]10.1002/art.20280
Purcell S, Cherny SS, Sham PC (2003) Genetic Power Calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics 19:149–150 [PubMed] [Cross Ref]10.1093/bioinformatics/19.1.149
Qu H, Bharaj B, Liu XQ, Curtis JA, Newhook LA, Paterson AD, Hudson TJ, Polychronakos C (2005) Assessing the validity of the association between the SUMO4 M55V variant and risk of type 1 diabetes. Nat Genet 37:111–112 [PubMed] [Cross Ref]10.1038/ng0205-111
Remmers EF, Joe B, Griffiths MM, Dobbins DE, Dracheva SV, Hashiramoto A, Furuya T, Salstrom JL, Wang J, Gulko PS, Cannon GW, Wilder RL (2002) Modulation of multiple experimental arthritis models by collagen-induced arthritis quantitative trait loci isolated in congenic rat lines: different effects of non-major histocompatibility complex quantitative trait loci in males and females. Arthritis Rheum 46:2225–2234 [PubMed] [Cross Ref]10.1002/art.10439
Rigby AS, Silman AJ, Voelm L, Gregory JC, Ollier WE, Khan MA, Nepom GT, Thomson G (1991) Investigating the HLA component in rheumatoid arthritis: an additive (dominant) mode of inheritance is rejected, a recessive mode is preferred. Genet Epidemiol 8:153–175 [PubMed] [Cross Ref]10.1002/gepi.1370080303
Rioux JD, Daly MJ, Silverberg MS, Lindblad K, Steinhart H, Cohen Z, Delmonte T, et al (2001) Genetic variation in the 5q31 cytokine gene cluster confers susceptibility to Crohn disease. Nat Genet 29:223–228 [PubMed] [Cross Ref]10.1038/ng1001-223
Rodriguez MR, Nunez-Roldan A, Aguilar F, Valenzuela A, Garcia A, Gonzalez-Escribano MF (2002) Association of the CTLA4 3′ untranslated region polymorphism with the susceptibility to rheumatoid arthritis. Hum Immunol 63:76–81 [PubMed] [Cross Ref]10.1016/S0198-8859(01)00358-5
Schellekens GA, Visser H, de Jong BA, van den Hoogen FH, Hazes JM, Breedveld FC, van Venrooij WJ (2000) The diagnostic properties of rheumatoid arthritis antibodies recognizing a cyclic citrullinated peptide. Arthritis Rheum 43:155–163 [PubMed] [Cross Ref]10.1002/1529-0131(200001)43:1<155::AID-ANR20>3.0.CO;2-3
Smyth DJ, Howson JM, Lowe CE, Walker NM, Lam AC, Nutland S, Hutchings J, Tuomilehto-Wolf E, Tuomilehto J, Guja C, Ionescu-Tirgoviste C, Undlien DE, Ronningen KS, Savage D, Dunger DB, Twells RC, McArdle WL, Strachan DP, Todd JA (2005) Assessing the validity of the association between the SUMO4 M55V variant and risk of type 1 diabetes. Nat Genet 37:110–111 [PubMed] [Cross Ref]10.1038/ng0205-110
Stastny P, Fink CW (1977) HLA-Dw4 in adult and juvenile rheumatoid arthritis. Transplant Proc 9:1863–1866 [PubMed]
Steer S, Fisher SA, Fife M, Cuthbert A, Newton J, Wordsworth P, Lewis CM, Mathew CG, Lanchbury JS (2003) Development of rheumatoid arthritis is not associated with two polymorphisms in the Crohn’s disease gene CARD15. Rheumatology (Oxford) 42:304–307 [PubMed]
Steer S, Lad B, Grumley JA, Kingsley GH, Fisher SA (2005) Association of R602W in a protein tyrosine phosphatase gene with a high risk of rheumatoid arthritis in a British population: evidence for an early onset/disease severity effect. Arthritis Rheum 52:358–360 [PubMed] [Cross Ref]10.1002/art.20737
Stoll M, Corneliussen B, Costello CM, Waetzig GH, Mellgard B, Koch WA, Rosenstiel P, Albrecht M, Croucher PJ, Seegert D, Nikolaus S, Hampe J, Lengauer T, Pierrou S, Foelsch UR, Mathew CG, Lagerstrom-Fermer M, Schreiber S (2004) Genetic variation in DLG5 is associated with inflammatory bowel disease. Nat Genet 36:476–480 [PubMed] [Cross Ref]10.1038/ng1345
Stolt P, Bengtsson C, Nordmark B, Lindblad S, Lundberg I, Klareskog L, Alfredsson L (2003) Quantification of the influence of cigarette smoking on rheumatoid arthritis: results from a population based case-control study, using incident cases. Ann Rheum Dis 62:835–841 [PMC free article] [PubMed] [Cross Ref]10.1136/ard.62.9.835
Suzuki A, Yamada R, Chang X, Tokuhiro S, Sawada T, Suzuki M, Nagasaki M, Nakayama-Hamada M, Kawaida R, Ono M, Ohtsuki M, Furukawa H, Yoshino S, Yukioka M, Tohma S, Matsubara T, Wakitani S, Teshima R, Nishioka Y, Sekine A, Iida A, Takahashi A, Tsunoda T, Nakamura Y, Yamamoto K (2003) Functional haplotypes of PADI4, encoding citrullinating enzyme peptidylarginine deiminase 4, are associated with rheumatoid arthritis. Nat Genet 34:395–402 [PubMed] [Cross Ref]10.1038/ng1206
Swanberg M, Lidman O, Padyukov L, Eriksson P, Akesson E, Jagodic M, Lobell A, Khademi M, Borjesson O, Lindgren CM, Lundman P, Brookes AJ, Kere J, Luthman H, Alfredsson L, Hillert J, Klareskog L, Hamsten A, Piehl F, Olsson T (2005) MHC2TA is associated with differential MHC molecule expression and susceptibility to rheumatoid arthritis, multiple sclerosis and myocardial infarction. Nat Genet 37:486–494 [PubMed] [Cross Ref]10.1038/ng1544
Tang K, Fu DJ, Julien D, Braun A, Cantor CR, Koster H (1999) Chip-based genotyping by mass spectrometry. Proc Natl Acad Sci USA 96:10016–10020 [PMC free article] [PubMed] [Cross Ref]10.1073/pnas.96.18.10016
Tengstrand B, Ahlmen M, Hafstrom I (2004) The influence of sex on rheumatoid arthritis: a prospective study of onset and outcome after 2 years. J Rheumatol 31:214–222 [PubMed]
Tokuhiro S, Yamada R, Chang X, Suzuki A, Kochi Y, Sawada T, Suzuki M, Nagasaki M, Ohtsuki M, Ono M, Furukawa H, Nagashima M, Yoshino S, Mabuchi A, Sekine A, Saito S, Takahashi A, Tsunoda T, Nakamura Y, Yamamoto K (2003) An intronic SNP in a RUNX1 binding site of SLC22A4, encoding an organic cation transporter, is associated with rheumatoid arthritis. Nat Genet 35:341–348 [PubMed] [Cross Ref]10.1038/ng1267
Ueda H, Howson JM, Esposito L, Heward J, Snook H, Chamberlain G, Rainbow DB, et al (2003) Association of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune disease. Nature 423:506–511 [PubMed] [Cross Ref]10.1038/nature01621
Vaidya B, Pearce SH, Charlton S, Marshall N, Rowan AD, Griffiths ID, Kendall-Taylor P, Cawston TE, Young-Min S (2002) An association between the CTLA4 exon 1 polymorphism and early rheumatoid arthritis with autoimmune endocrinopathies. Rheumatology (Oxford) 41:180–183 [PubMed]
van der Helm-van Mil AHM, Dieudé P, Schonkeren JJM, Cornélis F, Huizinga TWJ (2004) No association between tumour necrosis factor receptor type 2 gene polymorphism and rheumatoid arthritis severity: a comparison of the extremes of phenotypes. Rheumatology (Oxford) 43:1232–1234 [PubMed]
van Jaarsveld CH, ter Borg EJ, Jacobs JW, Schellekens GA, Gmelig-Meyling FH, van Booma-Frankfort C, de Jong BA, van Venrooij WJ, Bijlsma JW (1999) The prognostic value of the antiperinuclear factor, anti-citrullinated peptide antibodies and rheumatoid factor in early rheumatoid arthritis. Clin Exp Rheumatol 17:689–697 [PubMed]
van Zeben D, Hazes JM, Zwinderman AH, Cats A, van der Voort EA, Breedveld FC (1992) Clinical significance of rheumatoid factors in early rheumatoid arthritis: results of a follow up study. Ann Rheum Dis 51:1029–1035 [PMC free article] [PubMed]
Viken MK, Amundsen SS, Kvien TK, Boberg KM, Gilboe IM, Lilleby V, Sollid LM, Forre OT, Thorsby E, Smerdel A, Lie BA (2005) Association analysis of the 1858C>T polymorphism in the PTPN22 gene in juvenile idiopathic arthritis and other autoimmune diseases. Genes Immun 6:271–273 [PubMed] [Cross Ref]10.1038/sj.gene.6364178
Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N (2004) Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst 96:434–442 [PubMed]
Wang WY, Barratt BJ, Clayton DG, Todd JA (2005) Genome-wide association studies: theoretical and practical concerns. Nat Rev Genet 6:109–118 [PubMed] [Cross Ref]10.1038/nrg1522
Weyand CM, Schmidt D, Wagner U, Goronzy JJ (1998) The influence of sex on the phenotype of rheumatoid arthritis. Arthritis Rheum 41:817–822 [PubMed] [Cross Ref]10.1002/1529-0131(199805)41:5<817::AID-ART7>3.0.CO;2-S
Williams RL (2000) A note on robust variance estimation for cluster-correlated data. Biometrics 56:645–646 [PubMed] [Cross Ref]10.1111/j.0006-341X.2000.00645.x
Wolfe F, Michaud K, Gefeller O, Choi HK (2003) Predicting mortality in patients with rheumatoid arthritis. Arthritis Rheum 48:1530–1542 [PubMed] [Cross Ref]10.1002/art.11024
Yamada R, Tanaka T, Unoki M, Nagai T, Sawada T, Ohnishi Y, Tsunoda T, Yukioka M, Maeda A, Suzuki K, Tateishi H, Ochi T, Nakamura Y, Yamamoto K (2001) Association between a single-nucleotide polymorphism in the promoter of the human interleukin-3 gene and rheumatoid arthritis in Japanese patients, and maximum-likelihood estimation of combinatorial effect that two genetic loci have on susceptibility to the disease. Am J Hum Genet 68:674–685 [PMC free article] [PubMed]
Yamada R, Ymamoto K (2005) Recent findings on genes associated with inflammatory disease. Mutat Res 573:136–151 [PubMed]
Yen JH, Tsai WC, Chen CJ, Ou TT, Lin CH, Hu CJ, Liu HW (2003) Tumor necrosis factor receptor 2 microsatellite and exon 6 polymorphisms in rheumatoid arthritis in Taiwan. J Rheumatol 30:438–442 [PubMed]
Zhernakova A, Eerligh P, Wijmenga C, Barrera P, Roep BO, Koeleman BP (2005) Differential association of the PTPN22 coding variant with autoimmune diseases in a Dutch population. Genes Immun 6:459–461 [PubMed] [Cross Ref]10.1038/sj.gene.6364220

Articles from American Journal of Human Genetics are provided here courtesy of American Society of Human Genetics
PubReader format: click here to try

Formats:

Related citations in PubMed

See reviews...See all...

Cited by other articles in PMC

See all...

Links

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...