Thompson, Susan D.

Background/Purpose:Genome‐wide association studies have identified susceptibility loci for many autoimmune diseases, including rheumatoid arthritis (RA) and juvenile idiopathic arthritis (JIA). About 5% of children with JIA have rheumatoid factor positive arthritis that is phenotypically similar to adult seropositive RA, thus representing childhood onset RA (CORA). To understand the genetic architecture of CORA risk relative to other autoimmune diseases, we genotyped CORA cases and controls on the Immunochip, a custom array designed by the Immunochip Consortium to fine map autoimmune disease‐associated loci shared across 11 autoimmune phenotypes.Methods:Genotyping was completed on 340 CORA cases (mean onset age: 10.2 ± 4.2 yrs) and 11624 controls. Standard SNP and sample QC was performed (e.g., removing samples with call rate <98%, admixture outliers). To test for SNP association with CORA a logistic regression model was computed with Caucasian admixture proportions (ADMIXTURE) as covariates (SNPLash). False discovery rate (FDR) adjusted p‐values (PFDR) are reported to account for the actual number of tests computed.Results:SNP rs3129769, near HLA DRB1 was the most significantly associated (PFDR <3×10‐25), and is in linkage disequilibrium with the HLA DRB1 SNP reported in RA (rs660895, PFDR <1×10‐24). Outside the HLA region, 28 regions had ≥1 SNP meeting genome‐wide significance (PFDR<0.05). The best signal of association was on 22q13 (rs9610687, OR = 0.57, PFDR < 0.0002) near IL2RB, followed by the PTPN22 locus (rs6679677, OR = 1.83, PFDR < 0.0005), an intergenic SNP on chromosome 4 (rs970036, OR = 1.78, PFDR < 0.002) and an intronic SNP (19p13, rs3787016 PFDR < 0.002) in the POLR2E gene, which encodes a subunit of RNA polymerase II. Some loci which have previously been implicated in RA had evidence of association including MMEL (rs751358 PFDR < 0.003), IRF5 (rs4731531 PFDR < 0.006) and CCR6 (rs11575078 PFDR < 0.03). In addition using the FDR‐based threshold we identified several potential SNPs (rs10918214, 1q23; rs28491312, 14q31) not previously implicated in RA.Conclusion:Immunochip analysis of the largest cohort of CORA investigated to date has confirmed the HLA DRB1 association and identified several other loci associated with CORA. Comparisons among loci associated with CORA and those identified in RA and other forms of JIA are underway and will help delineate the unique genetic factors for CORA susceptibility.

ObjectiveChildren with childhood‐onset rheumatoid arthritis (RA) include those with rheumatoid factor or anti–citrullinated protein antibody–positive juvenile idiopathic arthritis. To test the hypothesis that adult‐onset RA–associated variants are also associated with childhood‐onset RA, we investigated RA‐associated variants at 5 loci in a cohort of patients with childhood‐onset RA. We also assessed the cumulative association of these variants in susceptibility to childhood‐onset RA using a weighted genetic risk score (wGRS).MethodsA total of 155 children with childhood‐onset RA and 684 healthy controls were genotyped for 5 variants in the PTPN22, TRAF1/C5, STAT4, and TNFAIP3 loci. High‐resolution HLA–DRB1 genotypes were available for 149 cases and 373 controls. We tested each locus for association with childhood‐onset RA via logistic regression. We also computed a wGRS for each subject, with weights based on the natural log of the published odds ratios (ORs) for the alleles investigated, and used logistic regression to test the wGRS for association with childhood‐onset RA.ResultsChildhood‐onset RA was associated with TNFAIP3 rs10499194 (OR 0.60 [95% confidence interval 0.44–0.83]), PTPN22 rs2476601 (OR 1.61 [95% confidence interval 1.11–2.31]), and STAT4 rs7574865 (OR 1.41 [95% confidence interval 1.06–1.87]) variants. The wGRS was significantly different between cases and controls (P < 2 × 10−16). Individuals in the third to fifth quintiles of wGRS had a significantly increased disease risk compared to baseline (individuals in the first quintile). Higher wGRS was associated with increased risk of childhood‐onset RA, especially among males.ConclusionThe magnitude and direction of the association between TNFAIP3, STAT4, and PTPN22 variants and childhood‐onset RA are similar to those observed in RA, suggesting that adult‐onset RA and childhood‐onset RA share common genetic risk factors. Using a wGRS, we have demonstrated the cumulative association of RA‐associated variants with susceptibility to childhood‐onset RA.