A candidate gene approach identifies the TRAF1/C5 region as a risk factor for rheumatoid arthritis

Abstract

Background: Rheumatoid arthritis (RA) is a chronic autoimmune disorder affecting approximately 1% of the population. The disease results from the interplay between an individual's genetic background and unknown environmental triggers. Although human leukocyte antigens (HLAs) account for approximately 30% of the heritable risk, the identities of non-HLA genes explaining the remainder of the genetic component are largely unknown. Based on functional data in mice, we hypothesized that the immune-related genes complement component 5 (C5) and/or TNF receptor-associated factor 1 (TRAF1), located on Chromosome 9q33-34, would represent relevant candidate genes for RA. We therefore aimed to investigate whether this locus would play a role in RA.

Methods and findings: We performed a multitiered case-control study using 40 single-nucleotide polymorphisms (SNPs) from the TRAF1 and C5 (TRAF1/C5) region in a set of 290 RA patients and 254 unaffected participants (controls) of Dutch origin. Stepwise replication of significant SNPs was performed in three independent sample sets from the Netherlands (ncases/controls = 454/270), Sweden (ncases/controls = 1,500/1,000) and US (ncases/controls = 475/475). We observed a significant association (p < 0.05) of SNPs located in a haplotype block that encompasses a 65 kb region including the 3' end of C5 as well as TRAF1. A sliding window analysis revealed an association peak at an intergenic region located approximately 10 kb from both C5 and TRAF1. This peak, defined by SNP14/rs10818488, was confirmed in a total of 2,719 RA patients and 1,999 controls (odds ratiocommon = 1.28, 95% confidence interval 1.17-1.39, pcombined = 1.40 x 10(-8)) with a population-attributable risk of 6.1%. The A (minor susceptibility) allele of this SNP also significantly correlates with increased disease progression as determined by radiographic damage over time in RA patients (p = 0.008).

Conclusions: Using a candidate-gene approach we have identified a novel genetic risk factor for RA. Our findings indicate that a polymorphism in the TRAF1/C5 region increases the susceptibility to and severity of RA, possibly by influencing the structure, function, and/or expression levels of TRAF1 and/or C5.

Figure 1. LD Structure and Haplotype Association across the TRAF1/C5 Region in Sample Set 1 (290 RA Patients and 254 Controls)

(A) Haplotype block structure was predicted on the basis of the strength of pairwise LD, which is presented as a 2×2 matrix; red represents very high LD (D'), white indicates absence of correlation between SNPs, and blue indicates high correlation with a low level of significance. SNPs that were chosen for haplotype analysis are indicated along the top by an asterisk. (B) Using htSNPs from each block, indicated by the asterisk in (A), haplotypes were inferred with a certainty of above 98% as represented by the R ² _h value. ^¥Comparisons were made between one haplotype versus all others in cases and controls. ^aThe protective haplotype that is significantly less frequent in cases. ^bThe susceptible haplotype that is significantly more frequent in patients. (C) Sliding window of the susceptible haplotype using consecutive two-SNP combinations of the htSNPs reveals that SNP14 and SNP15 account for most of the association observed. For each of the SNP pairs we show the –log₁₀ p-value. The dotted line indicates a nominal p-value of 0.005.

Figure 2. RA-Associated SNPs are Highly Linked to SNP14

Pairwise LD (r ²) between associated SNP14/rs10818488 and all other SNPs genotyped was calculated. For each of the SNPs listed along the x-axis we show the –log₁₀ (y-axis) of the p-values for RA patients versus controls. Dotted lines indicate a nominal p-value of 0.005 and a maximal r ² value of 1. In a logistics regression model, only SNP14/rs10818488 remained statistically significant (p = 6.16 × 10⁻⁴)

Figure 3. The A Allele of SNP14 Is Associated with Distinct Phenotypes of RA

(Top, table) The frequency of the A allele of patients from whom baseline autoantibody status (ACPA and RF) were available was calculated. ORs and p-values were calculated between each subgroup and controls, and indicated a predominantly higher frequency of the A allele in ACPA- and RF-positive patients. (Bottom, bar graph) Progression of joint damage in Sharp–van der Heijde units (“Sharp score”) is higher in the presence of the minor A allele of SNP14. Radiological data of 193 A carriers and 85 non-A carriers were available, and differences between the groups were calculated based on disease severity after 2 y corrected for baseline.