Mapping the shared and distinct HLA alleles for seropositive and seronegative rheumatoid arthritis. B. Han1,2,3, S. Eyre4,5, D. Diogo1,3,6, J. Bowes4,5, Y. Okada1,3,6, L. Padyukov7, R. Plenge1,3,6, L. Klareskog7, J. Worthington4,5, P. K. Gregersen8, P. I. W. de Bakker1,3,9,10, S. Raychaudhuri1,2,3,5,6 1) Division of Genetics, Brigham and Women's Hospital, Harvard Medical School, Boston, MA; 2) Partners HealthCare Center for Personalized Genetic Medicine, Boston, MA, USA; 3) Program in Medical and Population Genetics, Broad Institute of Harvard and MIT, Cambridge, MA, USA; 4) Arthritis Research UK Epidemiology Unit, Musculoskeletal Research Group, University of Manchester, Manchester Academic Health Sciences Centre, UK; 5) NIHR Manchester Musculoskeletal Biomedical Research Unit, Central Manchester NHS Foundation Trust, Manchester Academic Health Sciences Centre, UK; 6) Division of Rheumatology, Immunology, and Allergy and Division of Genetics, Brigham and Women's, Hospital, Harvard Medical School, Boston, Massachusetts, 02115, USA; 7) Rheumatology Unit, Department of Medicine, Karolinska Institutet and Karolinska University Hospital Solna, Stockholm, Sweden; 8) The Feinstein Institute for Medical Research, North Shore-Long Island Jewish Health System, Manhasset, New York, USA; 9) Department of Epidemiology, University Medical Center Utrecht, Utrecht, The Netherlands; 10) Department of Medical Genetics, University Medical Center Utrecht, Utrecht, The Netherlands.

   Motivation: Investigators have long speculated that the two subtypes of rheumatoid arthritis (RA), anti-citrullinated protein autoantibody positive (ACPA+) and negative (ACPA-), have distinct underlying genetic factors. The MHC region is the strongest genetic risk factor to ACPA+ RA, but plays a much more modest role in ACPA- RA. To understand the similarities and differences between these two disease subtypes, we fine-mapped and compared MHC associations. Results: Using densely genotyped SNP data consisting of 7,222 ACPA+ RA cases, 3,339 ACPA- RA cases, and 15,870 controls from six different cohorts (Eyre et al., Nat Gen, 2012), we imputed and tested HLA alleles in the two RA subtypes separately. We mapped associations to ACPA+ RA using forward search conditional analysis and confirmed previously published associations at amino acid sites at positions 13 (P < 10-705), 71, and 74 in HLA-DRB1, position 9 in HLA-B, and position 9 in HLA-DPB1. In addition, we identified a novel association at position 77 in HLA-A (P=1.710-8) located in the peptide binding groove implicating antigen presentation as the major mechanism by which MHC variation confers risk. Then in parallel we mapped associations to ACPA- RA. We recognized that ACPA- RA associations to the MHC might be confounded due to the inclusion of misclassified samples that are actually ACPA+ RA (false negative testing) or ankylosing spondylitis. We developed a novel statistical approach that estimates the proportion of misclassified samples and regresses out their effects. Using this approach we observed that each cohort consistently contained 3-9% of cases that likely had ankylosing spondylitis, and a variable number of cases that likely had ACPA+ RA. Controlling for misclassification effects, we identified the amino acid residues at position 13 in HLA-DRB1 as strongly associated with risk (Omnibus test P=1.210-16). Serine conferred the highest risk (OR=1.28, P=5.710-13); in stark contrast serine conferred protection to ACPA+ disease (OR=0.4). We also observed a shared association to the presence of an aspartate in position 9 in HLA-B (P=1.310-15, OR=1.38) with a more modest effect size than for ACPA+ disease (OR=2.1). Conclusions: Our analysis is the first to define specific amino acid sites for ACPA- RA, and demonstrates a distinct genetic basis for ACPA+ and ACPA- RA in the MHC region. Our analysis also underscores the importance of phenotypic classification for accurate fine-mapping.

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