Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Ann Rheum Dis. Author manuscript; available in PMC 2010 September 24.
Published in final edited form as:
PMCID: PMC2945890

Associations between HLA, PTPN22, CTLA4 genotypes and RA phenotypes of autoantibody status, age at diagnosis, and erosions in a large cohort study



HLA-DRB1 shared epitope (HLA-SE), PTPN22 and CTLA4 alleles are associated with CCP+ RA.


We examined associations between HLA-SE, PTPN22, CTLA4 genotypes and RA phenotypes in a large cohort to (a) replicate prior associations with CCP status, and (b) determine associations with radiographic erosions and age of diagnosis.


689 RA patients from the Brigham RA Sequential Study (BRASS) were genotyped for HLA-SE, PTPN22 (rs2476601) and CTLA4 (rs3087243). Association between genotypes and CCP, RF erosive phenotypes and age at diagnosis were assessed with multivariable models adjusting for age, sex, and disease duration. Novel causal pathway analysis was used to test the hypothesis that genetic risk factors and CCP are in the causal pathway for predicting erosions.


In multivariable analysis, presence of any HLA-SE was strongly associated with CCP+ (OR 3.05 (2.18–4.25)), and RF+ (OR 2.53 (1.83–3.5)) phenotypes; presence of any PTPN22 T allele was associated with CCP+ (OR 1.81 (1.24–2.66)) and RF+ phenotypes (1.84 (1.27–2.66)). CTLA4 was not associated with CCP or RF phenotypes. While HLA-SE was associated with erosive RA phenotype (OR 1.52 (1.01–2.17)), this was no longer significant after conditioning on CCP. PTPN22 and CTLA4 were not associated with erosive phenotype. Presence of any HLA-SE was associated with on average 3.6 years earlier diagnosis compared with absence of HLA-SE (41.3 vs. 44.9 years, p=0.003) and PTPN22 was associated with 4.2 years earlier age of diagnosis (39.5 vs. 43.6 years, p=0.002). CTLA4 genotypes were not associated with age at diagnosis of RA.


In this large clinical cohort, we replicated the association between HLA-SE, and PTPN22 but not CTLA4 with CCP+ and RF+ phenotypes. We also found evidence for associations between HLA-SE, and PTPN22 and earlier age at diagnosis. Since HLA-SE is associated with erosive phenotype in unconditional analysis, but is not significant after conditioning on CCP, this suggests that CCP is in the causal pathway for predicting erosive phenotype.

Keywords: rheumatoid arthritis, age at diagnosis, PTPN22, HLA, CCP

Genetic factors are thought to be responsible for up to 50%−60% of RA risk.[13] Two genes have been unequivocally associated with RA susceptibility (HLA-DRB1 and PTPN22), while other genes demonstrate strong, but inconclusive risk (eg. CTLA, PADI4).[4] Although the HLA associations with RA are complex[5, 6], the majority of the genetic signal from HLA is explained by alleles at the HLA-DRB1 locus[7], and account for approximately 30% of the genetic risk of RA[1]. In individuals of European ancestry, the associated HLA-DRB1 alleles share a region of sequence similarity or “shared epitope” at amino acid positions 70–74 in the third hypervariable region of the HLA-DRB1 molecule[1] (HLA-SE). Outside the HLA, the only genetic polymorphism that has been associated with RA susceptibility in populations of European ancestry and replicated across multiple independent studies is PTPN22. [3, 814] A missense allele (C➔T) is associated with an increased risk of RA (rs2476601), with a summary odds ratio of 1.68 (1.53–1.84) from a meta-analysis.[15] Both the HLA-SE alleles, and PTPN22 allele are more strongly associated with the phenotype citric citrullinated peptide positive (CCP+) RA.[4, 1619] Cytotoxic T lymphocyte-associated antigen 4 (CTLA4) gene has an A➔G single nucleotide polymorphism (SNP) in the 3’ untranslated region (rs3087243), that is associated with increased risk of RA in several populations[2022], although the OR is much more modest (OR=1.20). In a large replication study CTLA4 was more strongly associated with CCP+ RA.[4]

Phenotyping of disease subgroups in rheumatoid arthritis (RA), such CCP status, is important in studies of RA genetic and environmental epidemiology.[17, 2325] Environmental risk factors for RA differ in CCP+ and CCP- subgroups.[25] HLA-SE demonstrates a significant gene-environment interaction with cigarette smoking for susceptibility to CCP+ RA[23, 26] and for CCP antibody status in RA[24] but not for CCP- RA.

Studies of genetic predictors of individual RA phenotypes have suggested associations between several genes and autoantibody status (CCP, RF), erosive disease and age at onset of RA.[4, 16, 17, 23, 2730] Several prior studies have suggested that the association between HLA and erosions is due solely to CCP status; PTPN22 and CTLA4 have not been extensively studied for these phenotypes. We sought to test whether these genetic factors are associated radiographic erosions and early age of diagnosis using data from a large observational RA cohort. Thus, our goal was to (a) replicate association of HLA, PTPN22, CTLA4 with CCP phenotype; (b) determine if genotypes predicted radiographic erosions independent of the CCP association; and (c) to study whether age of diagnosis is associated with genetic variation. We performed conventional association testing as well as novel causal pathway analysis as proposed by Cooper[31] and Li[32] to test the hypothesis that genetic risk factors and CCP are in the causal pathway in predicting erosions.


Study Population

BRASS (Brigham Rheumatoid Arthritis Sequential Study) is a prospective observational study of 968 RA patients receiving care at the Brigham and Women’s Hospital. Goals of the study are to: 1) determine and validate biomarkers that predict drug response and toxicity in RA; 2) determine and validate biomarkers that predict disease activity and prognosis in RA; and, 3) evaluate the natural history of treated RA by measuring clinical, functional and economic outcomes. Baseline evaluation includes demographic and clinical information, assessment of functional status, disease activity, comorbidity, laboratory testing and hand radiographs. A physical exam, including joint examination, assessment of pain and disease activity by MD and the patient are collected at baseline and yearly. Samples of blood for immunophenoptyping, including C reactive protein (CRP), cytokines, chemokines, and rheumatoid factor (RF), anti-cyclic citrullinated peptide (CCP) as well as blood specimens for DNA/RNA testing are collected and stored at baseline and yearly. During follow-up, patients are mailed a self-administered questionnaire every 6 months to collect information on disease severity, functional status, resource utilization, level of fatigue, employment status, medications, adverse events and intercurrent health events. Hand radiographs, PA views of the hands are performed at baseline and years 3 and 5 during the study. This analysis was limited to the baseline radiographs. The study was approved by the Partners Institutional Review Committee.

Laboratory Methods

Blood was collected at the baseline visit for genotyping. Samples were genotyped for HLA-SE alleles by low-resolution genotyping, and for PTPN22, and CTLA4 (CT60 allele) by Sequenom genotyping (San Diego, CA). HLA-SE alleles 01, 04, 10, and 14 were considered positive. Rheumatoid factor testing was performed by immunoturbidimetric technique on the Cobas Integra 700 analyzer (Roche Diagnostics -Indianapolis, IN), using reagents and calibrators from Roche. Anti-cyclic citrullinated peptide antibodies (CCP) were measured using a second generation ELISA assay (Inova Diagnostics, Inc. – San Diego, CA) with a titer of >20 considered as positive. Radiograph reports were reviewed by a study rheumatologist for evidence of erosions and coded as erosion present or absent.

Statistical analysis

Association between genotypes and dichotomous phenotypes for CCP, RF and erosive RA at baseline were assessed with logistic regression models adjusting for age, sex, and disease duration. Association between genotype and age at diagnosis of RA phenotype was assessed with general linear models adjusting for sex.

Since statistical correlation does not imply a causal relationship, we performed causal pathway analysis by the method described by Li, et al.[32] In brief, Cooper’s Local Causal Discovery algorithm (Cooper’s LCD) was used to explore the potential causal pathway between genotypes and phenotypes (Figure 1).[31] Each causal pathway analysis tested 3 variables: (x) genotype, (y) phenotype, and (z) phenotype. From prior knowledge we know the genotype (x), for example, HLA-SE genotype, cannot be caused by any intermediate phenotypes (y, z), for example, RF and CCP antibody status. If the pairwise unconditional correlations exists between the 3 variables, but the genotype (x) and one phenotype (z) is un-correlated conditional on the other phenotype (y), only one causal path can be derived from x→y→z (Figure 1). We conducted causal pathway analyses of the association between HLA-SE (x), CCP (y), and RF (z), and between HLA-SE (x), CCP (y), and erosion (z).

Figure 1
Causal Pathway Analysis


The BRASS research study began enrollment in March 2003, and has enrolled 968 patients to date. Genotype data and autoantibody status were available for 728 subjects at the time of this analysis. For this analysis, we included only Caucasian subjects, resulting in a sample of 689 Caucasian RA patients. Among these 689 patients, mean age is 58.0 (SD ±13.8) years, mean disease duration 15.4 (SD ±12.8) years, 110 (15.9%) are recent diagnosis RA, defined as < 2 years disease duration, 560 (81.2%) are female of whom 179 (32%) are in the premenopausal age range (age<51). Education level is 21.5% with high school or less and 78.5% with some college education. At baseline, 323 (47%) of patients report starting a new therapy for RA within the prior 12 months (Table 1). Baseline radiographs demonstrated presence of RA erosions in 374/627(59.7%) subjects with radiographs. There were 61.8% RF positive subjects, and 66.5% CCP positive subjects. Mean age at diagnosis of RA was 42.58 (±15.1). Genotype frequencies were similar to other RA cohorts (Table 2).[4]

Table 1
Characteristics of 689 Caucasian subjects in the BRASS cohort genetic analyses
Table 2
Genotypes for HLA-SE, PTPN22 and CTLA 4 among 689 Caucasian rheumatoid arthritis subjects in the BRASS cohort

We first attempted to replicate the finding that alleles within three genes, HLA-DRB1, PTPN22 and CTLA4, are associated with CCP+ RA. We created contingency tables for genotypes dichotomized by presence of any HLA-SE (single copy or double copy), the T allele of PTPN22 (single or double copy), the G allele of CTLA4 (single or double copy) and presence/absence of CCP and RF phenotypes. We tested for significance using logistic regression models assuming a multiplicative model for genotype, and adjusting for age, sex, and disease duration (Table 3). Presence of any HLA-SE was strongly associated with CCP+ phenotype (OR 3.05, 95% CI 2.19–4.25, p=2.9 × 10−9) and RF+ phenotype (OR 2.53, 95% CI 1.83–3.5, p=4.2 × 10−7); presence of any PTPN22 T allele was associated with CCP+ phenotype (OR 1.81, 95% CI 1.24–2.66, p=0.006) and RF+ phenotype (OR 1.84, 95% CI 1.27–2.66, p=0.002); and CTLA4 was not associated with CCP (OR 1.04, 95% CI 0.7–1.55) or RF phenotypes (OR=0.94, 95% CI 0.64–1.39).

Table 3
Genotype phenotype associations with autoantibody status in the BRASS study (n=689 Caucasian subjects)

We next sought to determine whether these genetic variants were associated with two markers of disease severity, radiographic erosions and age of diagnosis. Presence of any HLA-SE was associated with erosive RA phenotype in unadjusted logistic regression analysis (p=4 × 10−4), however this association was less strong in a logistic regression model that adjusted for age, sex and disease duration (OR 1.52, 95% CI 1.01–2.17, p=0.02) (Table 4). Presence of any PTPN22 T allele was not associated with erosive phenotype (OR = 1.14, 95% CI 0.77–1.71), nor was CTLA4 (OR=1.34, 95% CI 0.87–2.15) (Table 4).

Table 4
Genotype phenotype association with erosive RA phenotype in the BRASS study (n=689 Caucasian subjects)

Using general linear regression models for genotype as a predictor of age at diagnosis of RA, adjusted for sex, presence of any HLA-SE was on average associated with 3.6 years earlier age at diagnosis of RA compared with absence of HLA-SE (41.3 vs. 44.9 years, p=0.002) (Table 5). PTPN22 was on average associated with 4.2 years earlier age at diagnosis of RA (39.5 vs. 43.6 p=0.002) with the earliest age at diagnosis in those with the TT genotype (37.8 years). Adjusting for sex slightly attenuated these relationships. CTLA4 genotypes were not associated with age at diagnosis of RA in this dataset.

Table 5
RA genotypes as predictors of age at diagnosis of RA in the BRASS study (n=689)

We conducted two causal pathway analyses, adapted from Li, et al.[32], and illustrated in Figure 1. We asked whether a genetic variant (HLA-SE) contributed to RF phenotype, independent of CCP status (Causal Pathway 1) as a replication of the Li, et al analysis. We also asked whether HLA-SE contributed to erosion phenotype, independent of CCP status (Causal Pathway 2).

The first step in a causal pathway analysis[32] is to test unconditional associations between variables. We demonstrated strong relationships (p< 0.001) between all variables (Table 6).

Table 6
Unconditional association between variables

Conditional analysis of Causal Pathway 1, of the association between HLA-SE (x), CCP (y), and RF (z) demonstrated that HLA-SE (x) and RF (z) are not associated when conditioning on CCP (y) (Table 7). These results are similar to those shown in the causal pathway analysis by Li, et al.[32] Therefore the evidence supports a causal pathway from HLA-SE → CCP → RF, but not directly from HLA-SE → RF.

Table 7
Conditional analysis of causal pathways for HLA-SE, RF and CCP phenotypes in RA

Conditional analysis of Causal Pathway 2, of the association between HLA-SE (x), CCP (y), and erosion (z) demonstrated that HLA-SE (x) and erosion (z) are not associated when conditioning on CCP (y) (Table 8). Therefore the evidence supports a causal pathway from HLA-SE → CCP→ erosion, but not directly from HLA-SE → erosion. Since the analysis presented in Table 7 demonstrated that SE (x) and RF(y) are not associated when conditioning on CCP, we did not test for a causal pathway from SE → RF → erosion.

Table 8
Conditional analysis of causal pathways for HLA-SE, CCP and erosion phenotypes in RA


In this large observational RA cohort, we demonstrated strong associations between two RA genetic risk factors (HLA SE, PTPN22) and RA phenotypes. HLA-SE was strongly associated with CCP, RF, and erosive phenotypes, even after adjusting for age, sex, and disease duration. We demonstrated that PTPN22 was strongly associated with CCP and RF phenotypes, but not with erosive RA. We were unable to show any genotype-phenotype associations for CTLA-4, perhaps due to limited power as the published OR for susceptibility for CTLA-4 is 1.2 whereas for PTPN22 is 1.75 and for HLA-SE is 3.0. We found strong correlations between HLA-SE and both CCP and RF phenotypes, however, the HLA-SE association with the CCP phenotype was stronger than for RF phenotype. In this clinical cohort, using causal pathway analysis we replicated prior findings from the North American Rheumatoid Arthritis Consortium (NARAC) [18, 32], and the Leiden Early Arthritis Clinic (EAC)[17] that suggest that HLA-SE is causally associated with the CCP phenotype, and CCP is causally associated with the RF phenotype, but HLA-SE is not causally associated with the RF phenotype.

Our findings for HLA-SE and erosions in unconditional analyses are consistent with a meta-analysis of 30 studies published from prospective cohorts and cross-sectional studies involving 3,240 RA patients demonstrating an odds ratio of 2.0 (95% CI 1.8–2.2) for association of HLA-SE and erosions.[27] Our causal pathway analysis extends these observations to study the role of CCP antibodies. In our cross-sectional study, CCP is strongly associated with erosions, but HLA-SE is not associated with erosions, after conditioning on CCP status, suggesting that there is no causal pathway directly from HLA-SE to erosions. Our causal pathway findings are consistent with two studies from the Leiden EAC.[17, 30] Among RA patients in the EAC prospective cohort there were large differences in rates of erosion progression between CCP+RA and CCP-RA, with an additional effect of the HLA-SE on erosion progression only among the CCP+ group but no association for HLA-SE with erosions among the CCP- group.[17] Analysis of the undifferentiated arthritis patients in the EAC cohort followed prospectively for the development of RA demonstrated that HLA-SE alleles are primarily a risk factor for development of anti-CCP antibodies and are not an independent risk factor for progression to RA, after adjusting for CCP status.[30]

We found evidence of 3.6 years earlier age of diagnosis with HLA-SE in this large clinical cohort with a mean disease duration of 15 years. This work replicates other studies in which HLA-SE was associated with 6 years earlier age at onset in a seropositive RA cohort with < 15 months of disease duration in the US [28] as well as in a Korean population in which the specific allele, HLA DRB1*0405, was associated with 4 years earlier onset.[33] We found evidence for 4.2 years earlier diagnosis of RA for PTPN22, which is similar to the findings of 2 years earlier age of onset in samples from North America and Sweden.[4] In a population from the UK, PTPN22 was associated with 8.6 years earlier onset in homozygotes, and 4.7 years earlier onset in heterozygotes.[34]

The BRASS cohort is a well-educated, primarily Caucasian population, with long disease duration, treated at a tertiary referral center in the United States, all factors that may limit generalizability. Although disease duration is similar to that in the North American Rheumatoid Arthritis Consortium (NARAC), the rates of seropositivity and erosive disease are lower, since by design NARAC recruited more severe RA patients.[32] However age, disease duration, rates of erosive disease, and seropositivity are similar to those reported in the National Databank study of >14,000 patients enrolled from rheumatology practices across the US[35, 36], suggesting that BRASS subjects are more similar to RA patients seen in the community. The causal pathway approach is a statistical method that requires a number of assumptions, as discussed in detail by Cooper et al[31], and it is possible that our dataset does not meet all of the assumptions. The approach does allow for the presence of potential confounders, as long as the variable “x”, in this case genotype, is not caused by the confounder.

In conclusion, genotype-phenotype analysis of a large clinical cohort demonstrates the importance of considering phenotypes when studying genetic predictors. We replicated association of both HLA-SE, and PTPN22 with CCP+ RA compared with CCP- RA as well as associations with earlier diagnosis of RA, but were unable to demonstrate any associations for CTLA-4. Of the 3 genes studied, only HLA-SE was associated with radiographic erosions but this association was not independent of CCP status. The novel causal pathway analyses confirms prior studies that demonstrate the importance of antibodies to CCP in the pathogenesis of joint damage in RA and provides support to recent calls[17, 23] for considering CCP+ RA as a separate clinical entity within the overall RA phenotype.


We wish to thank the BRASS participants, the rheumatologists in the Robert B. Brigham Arthritis Center for their extensive time and efforts on behalf of the study, and our talented team of research assistants.


Dr. Karlson is supported by NIH grants R01 AR49880, and K24 AR0524-01. Robert Plenge is supported by NIH K08 A155314. The BRASS cohort and coauthors Jing Cui, Roberta Glass, Nancy Maher, Elena Izmailova, Michael Weinblatt, and Nancy Shadick are supported by Millenium Pharmaceuticals. Alex Parker was formerly supported by Millenium Pharmaceuticals and is currently supported by Amgen. Ronenn Roubenoff was formerly supported by Millenium Pharmaceuticals and is currently supported by Biogen/IDEC.

The Corresponding Author has the right to grant on behalf of all authors and does grant on behalf of all authors, an exclusive license (or non-exclusive for government employees) on a worldwide basis to the BMJ Publishing Group Ltd and its Licensees to permit this article to be published in Annals of the Rheumatic Diseases editions and any other BMJPGL products to exploit all subsidiary rights, as set out in our license


1. Gregersen PK, Silver J, Winchester RJ. The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis & Rheumatism. 1987;30(11):1205–1213. [PubMed]
2. MacGregor AJ, Snieder H, Rigby AS, Koskenvuo M, Kaprio J, Aho K, et al. Characterizing the quantitative genetic contribution to rheumatoid arthritis using data from twins. Arthritis Rheum. 2000;43(1):30–37. [PubMed]
3. Begovich AB, Carlton VE, Honigberg LA, Schrodi SJ, Chokkalingam AP, Alexander HC, et al. A missense single-nucleotide polymorphism in a gene encoding a protein tyrosine phosphatase (PTPN22) is associated with rheumatoid arthritis. Am J Hum Genet. 2004;75(2):330–337. [PubMed]
4. Plenge RM, Padyukov L, Remmers EF, Purcell S, Lee AT, Karlson EW, et al. 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. Am J Hum Genet. 2005;77(6):1044–1060. [PubMed]
5. Jawaheer D, Gregersen PK. Rheumatoid arthritis. The genetic components. Rheum Dis Clin North Am. 2002;28(1):1–15. v. [PubMed]
6. Newton JL, Harney SM, Wordsworth BP, Brown MA. A review of the MHC genetics of rheumatoid arthritis. Genes Immun. 2004;5(3):151–157. [PubMed]
7. Jawaheer D, Seldin MF, Amos CI, Chen WV, Shigeta R, Etzel C, et al. Screening the genome for rheumatoid arthritis susceptibility genes: a replication study and combined analysis of 512 multicase families. Arthritis Rheum. 2003;48(4):906–916. [PubMed]
8. Lee AT, Li W, Liew A, Bombardier C, Weisman M, Massarotti EM, et al. The PTPN22 R620W polymorphism associates with RF positive rheumatoid arthritis in a dose-dependent manner but not with HLA-SE status. Genes Immun. 2005;6(2):129–133. [PubMed]
9. Hinks A, Barton A, John S, Bruce I, Hawkins C, Griffiths CE, et al. 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. 2005;52(6):1694–1699. [PubMed]
10. Orozco G, Sanchez E, Gonzalez-Gay MA, Lopez-Nevot MA, Torres B, Caliz R, et al. Association of a functional single-nucleotide polymorphism of PTPN22, encoding lymphoid protein phosphatase, with rheumatoid arthritis and systemic lupus erythematosus. Arthritis Rheum. 2005;52(1):219–224. [PubMed]
11. Prescott NJ, Fisher SA, Onnie C, Pattni R, Steer S, Sanderson J, et al. A general autoimmunity gene (PTPN22) is not associated with inflammatory bowel disease in a British population. Tissue Antigens. 2005;66(4):318–320. [PubMed]
12. Viken MK, Amundsen SS, Kvien TK, Boberg KM, Gilboe IM, Lilleby V, et al. Association analysis of the 1858C>T polymorphism in the PTPN22 gene in juvenile idiopathic arthritis and other autoimmune diseases. Genes Immun. 2005 [PubMed]
13. Zhernakova A, Eerligh P, Wijmenga C, Barrera P, Roep BO, Koeleman BP. Differential association of the PTPN22 coding variant with autoimmune diseases in a Dutch population. Genes Immun. 2005;6(6):459–461. [PubMed]
14. Criswell LA, Pfeiffer KA, Lum RF, Gonzales B, Novitzke J, Kern M, et al. Analysis of Families in the Multiple Autoimmune Disease Genetics Consortium (MADGC) Collection: the PTPN22 620W Allele Associates with Multiple Autoimmune Phenotypes. Am J Hum Genet. 2005;76(4):561–571. [PubMed]
15. Lee YH, Rho YH, Choi SJ, Ji JD, Song GG, Nath SK, et al. The PTPN22 C1858T functional polymorphism and autoimmune diseases--a meta-analysis. Rheumatology (Oxford) 2006 [PubMed]
16. Padyukov L, Silva C, Stolt P, Alfredsson L, Klareskog L. A gene-environment interaction between smoking and shared epitope genes in HLA-DR provides a high risk of seropositive rheumatoid arthritis. Arthritis Rheum. 2004;50(10):3085–3092. [PubMed]
17. Huizinga TW, Amos CI, van der Helm-van Mil AH, Chen W, van Gaalen FA, Jawaheer D, et al. Refining the complex rheumatoid arthritis phenotype based on specificity of the HLA-DRB1 shared epitope for antibodies to citrullinated proteins. Arthritis Rheum. 2005;52(11):3433–3438. [PubMed]
18. Irigoyen P, Lee AT, Wener MH, Li W, Kern M, Batliwalla F, et al. Regulation of anti-cyclic citrullinated peptide antibodies in rheumatoid arthritis: contrasting effects of HLA-DR3 and the shared epitope alleles. Arthritis Rheum. 2005;52(12):3813–3818. [PubMed]
19. Johansson M, Arlestig L, Hallmans G, Rantapaa-Dahlqvist S. PTPN22 polymorphism and anti-cyclic citrullinated peptide antibodies in combination strongly predicts future onset of rheumatoid arthritis and has a specificity of 100% for the disease. Arthritis Res Ther. 2006;8(1):R19. [PMC free article] [PubMed]
20. Vaidya B, Pearce SH, Charlton S, Marshall N, Rowan AD, Griffiths ID, et al. An association between the CTLA4 exon 1 polymorphism and early rheumatoid arthritis with autoimmune endocrinopathies. Rheumatology (Oxford) 2002;41(2):180–183. [PubMed]
21. Rodriguez MR, Nunez-Roldan A, Aguilar F, Valenzuela A, Garcia A, Gonzalez-Escribano MF. Association of the CTLA4 3' untranslated region polymorphism with the susceptibility to rheumatoid arthritis. Hum Immunol. 2002;63(1):76–81. [PubMed]
22. Suppiah V, O'Doherty C, Heggarty S, Patterson CC, Rooney M, Vandenbroeck K. The CTLA4+49A/G and CT60 polymorphisms and chronic inflammatory arthropathies in Northern Ireland. Exp Mol Pathol. 2006;80(2):141–146. [PubMed]
23. Klareskog L, Stolt P, Lundberg K, Kallberg H, Bengtsson C, Grunewald J, et al. A new model for an etiology of rheumatoid arthritis: Smoking may trigger HLA-DR (shared epitope)-restricted immune reactions to autoantigens modified by citrullination. Arthritis Rheum. 2006;54(1):38–46. [PubMed]
24. Linn-Rasker SP, van der Helm-van Mil AH, van Gaalen FA, Kloppenburg M, de Vries RR, le Cessie S, et al. Smoking is a risk factor for anti-CCP antibodies only in rheumatoid arthritis patients who carry HLA-DRB1 shared epitope alleles. Ann Rheum Dis. 2006;65(3):366–371. [PMC free article] [PubMed]
25. Pedersen M, Jacobsen S, Klarlund M, Pedersen BV, Wiik A, Wohlfahrt J, et al. Environmental risk factors differ between rheumatoid arthritis with and without auto-antibodies against cyclic citrullinated peptides. Arthritis Res Ther. 2006;8(4):R133. [PMC free article] [PubMed]
26. Pedersen M, Jacobsen S, Garred P, Madsen HO, Klarlund M, Svejgaard A, et al. Strong combined gene-environment effects in anti-cyclic citrullinated peptide-positive rheumatoid arthritis: a nationwide case-control study in Denmark. Arthritis Rheum. 2007;56(5):1446–1453. [PubMed]
27. Gorman JD, Lum RF, Chen JJ, Suarez-Almazor ME, Thomson G, Criswell LA. Impact of shared epitope genotype and ethnicity on erosive disease: a meta-analysis of 3,240 rheumatoid arthritis patients. Arthritis Rheum. 2004;50(2):400–412. [PubMed]
28. Wu H, Khanna D, Park G, Gersuk V, Nepom GT, Wong WK, et al. Interaction between RANKL and HLA-DRB1 genotypes may contribute to younger age at onset of seropositive rheumatoid arthritis in an inception cohort. Arthritis Rheum. 2004;50(10):3093–3103. [PubMed]
29. van Gaalen FA, van Aken J, Huizinga TW, Schreuder GM, Breedveld FC, Zanelli E, et al. Association between HLA class II genes and autoantibodies to cyclic citrullinated peptides (CCPs) influences the severity of rheumatoid arthritis. Arthritis Rheum. 2004;50(7):2113–2121. [PubMed]
30. van der Helm-van Mil AH, Verpoort KN, Breedveld FC, Huizinga TW, Toes RE, de Vries RR. The HLA-DRB1 shared epitope alleles are primarily a risk factor for anti-cyclic citrullinated peptide antibodies and are not an independent risk factor for development of rheumatoid arthritis. Arthritis Rheum. 2006;54(4):1117–1121. [PubMed]
31. Cooper GF. A simple constraint-based algorithm for efficiently mining observational databases for causal relationships. Data Mining and Knowledge Discovery. 1997;1:203–224.
32. Li W, Wang M, Irigoyen P, Gregersen PK. Inferring causal relationships among intermediate phenotypes and biomarkers: a case study of rheumatoid arthritis. Bioinformatics. 2006;22(12):1503–1507. [PubMed]
33. Lee HS, Lee KW, Song GG, Kim HA, Kim SY, Bae SC. Increased susceptibility to rheumatoid arthritis in Koreans heterozygous for HLA-DRB1*0405 and *0901. Arthritis Rheum. 2004;50(11):3468–3475. [PubMed]
34. Steer S, Lad B, Grumley JA, Kingsley GH, Fisher SA. 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. 2005;52(1):358–360. [PubMed]
35. Choi HK, Hernan MA, Seeger JD, Robins JM, Wolfe F. Methotrexate and mortality in patients with rheumatoid arthritis: a prospective study. Lancet. 2002;359(9313):1173–1177. [PubMed]
36. Finckh A, Choi HK, Wolfe F. Progression of radiographic joint damage in different eras: trends towards milder disease in rheumatoid arthritis are attributable to improved treatment. Ann Rheum Dis. 2006;65(9):1192–1197. [PMC free article] [PubMed]