Objective: HLA is the most strongly associated locus in rheumatoid arthritis (RA), accounting for up to one-third of the genetic contribution. Conditioning on the effect of true disease loci such as HLA can lead to increased power to detect effects at other loci and, in addition, allows investigation of the underlying disease models, including interactions. The aim of this study was to detect susceptibility loci for RA by conditioning on HLA in a large sample of affected sibling pairs (ASPs) and to test for evidence of interaction between novel loci and HLA.
Methods: Genotype data from 3 whole-genome linkage scans for RA in a US population and a UK population were pooled, resulting in a combined data set of 886 ASPs. This pooling of data increased the power to detect loci showing low levels of heterogeneity. Nonparametric linkage analysis was performed to identify regions of interest. Joint 2-locus analysis was then performed for HLA and each of the loci that demonstrated evidence of linkage in the 886 ASPs.
Results: Evidence for linkage was most significant at HLA (P = 4 x 10(-16)), with 7 non-HLA loci showing some evidence for linkage (P = 0.05-0.003). Joint modeling of these loci with HLA provided evidence for linkage at a genome-wide significance level for loci on 6q (P = 2.7 x 10(-6)) and 16p (P = 2 x 10(-4)).
Conclusion: These data provide the most convincing evidence to date that 6q and 16p harbor susceptibility genes. In addition, these loci may interact with HLA, facilitating the search for candidate genes within this region.