Fine mapping of the MHC in >60,000 samples by the International IBD Genetics Consortium: identification of multiple predisposing and protective variants that are mostly distinct between Crohn's disease and ulcerative colitis. P. Goyette1,2 on behalf of the International inflammatory bowel disease genetics consortium (IIBDGC) 1) Research Center, Montreal Heart Institute, Montreal, Quebec, Canada; 2) Department of Medecine, Université de Montréal, Montreal, Quebec, Canada.

   The major histocompatibility complex (MHC) is a well-established risk locus for both Crohn's disease (CD) and ulcerative colitis (UC). Previous studies of the MHC region have been hampered by insufficient density of variants tested in order to capture the extensive genetic variation of the region, in particular within the highly polymorphic classical HLA loci, as well as insufficient power to fine-map signals in this region with high linkage disequilibrium. In order to meet these challenges the IIBDGC has performed a large-scale high density genotyping project using the Immunochip custom genotyping array in over 67,000 samples. Following stringent QC, 7041 SNPs in the extended MHC (Chr6:25Mb-34Mb) were tested for association, using a logistic regression model controlling for ancestry, in 18,446 CD and 14,366 UC cases, as well as 33,933 healthy controls. In addition, alleles at the classical human leukocyte antigen (HLA) loci, and amino acids variation relevant to their antigen binding properties, were imputed from the SNP data. The primary association analyses in CD and UC confirm association to variants rs9264942 (1.16, P=1.52x10-25) and rs6927022 (1.56, P=2.7x10-169) for CD and UC, respectively, as previously reported by a recent IIBDGC study (Jostins et al., Nature, 2012). We used forward conditional logistic regression analysis to identify independent genome-wide significant associations. In CD we identified up to 10 independent signals, including 3 in the class II region and 4 near class I genes, mostly centered near HLA-B. In UC we identified up to 12 independent signals, including 6 in the class II region and 3 near classical class I genes. Interestingly, only three of the independent signals are shared between CD and UC, including a non-synonymous coding risk variant in CFB and a combination of rare protective alleles at HLA-B and HLA-C (HLA-B*5201 and HLA-C*1202). Both diseases show associations at HLA-A, but to SNPs tagging different HLA-A alleles; with CD associated to HLA-A*0301 while UC is associated the HLA-A*0201. While CD and UC share the majority of known risk alleles outside the MHC (110 of 163 known IBD risk loci), they only share a limited number of the independent alleles identified within the MHC. The current study demonstrates a strong but not exclusive role for classical class I and II loci in CD and UC, with only few MHC signals localizing outside the classical HLA loci.

You may contact the first author (during and after the meeting) at