Potential of integrating human genetics and electronic medical records for drug discovery: the example of TYK2 and rheumatoid arthritis. D. Diogo1,2,3, K. P. Liao2, R. S. Fulton4, R. R. Graham5, J. Cui2, J. C. Denny8, T. Behrens5, M. F. Seldin6, P. K. Gregersen7, E. Mardis4, R. M. Plenge1,2,3, The RACI, i2b2-Rheumatoid Arthritis, CORRONA 1) Division of Rheumatology, Immunology & Allergy, Brigham and Women's, Hospital, Boston, Massachusetts; 2) Division of of Genetics, Brigham and Women's, Hospital, Harvard Medical School, Boston, Massachusetts; 3) Medical and Population Genetics Program, Broad Institute, Cambridge, Massachusetts; 4) The Genome Institute, Washington University School of Medicine, St. Louis, Missouri; 5) ITGR Human Genetics Group, Genentech Inc, San Francisco, California; 6) Department of Biochemistry and Molecular Medicine, University of California, Davis, California; 7) The Feinstein Institute for Medical Research, North Shore-Long Island Jewish Health System, Manhasset, New York; 8) Department of Biomedical Informatics and department of Medicine, Vanderbilt University, Nashville, Tennessee.

   Human genetics has the potential to uncover disease-associated alleles that result in gain-of-function (GOF) or loss-of-function (LOF), thereby linking target perturbation with relevant human physiology. In particular, genes with LOF alleles that protect from disease represent promising targets for pharmacological manipulation. Here, we integrated deep sequencing with large-scale genotyping to search for protein-coding variants that influence risk of rheumatoid arthritis (RA). We selected 845 genes, based on findings from genome-wide association studies (GWAS) in RA and other autoimmune diseases, as well as candidate genes based on the pathophysiology of RA, and targeted the protein-coding exons for sequencing in 1,118 RA cases and 1,118 matched controls of European ancestry. We observed an excess of rare alleles (MAF<1%) predicted to be damaging in controls in TYK2, a gene coding for a member of the Janus kinases (JAK) (P=0.003). The signal of association was driven by 2 variants with MAF>0.5%. In addition, we observed an excess of true rare variants in controls in the kinase 1 domain-coding region of TYK2 (P=0.006). We further analyzed the role of low-frequency and common TYK2 protein-coding variants in two large collections of case-control samples genotyped on the Exomechip or the Immunochip. We demonstrate that three protein-coding variants predicted to be damaging independently protect against RA (P=1.6x10-27). Importantly, recent studies have shown that two of these variants are LOF mutations that affect TYK2 kinase activity. To assess for pleiotropic effects of the three alleles - and especially phenotypes that may be considered adverse events following pharmacological inhibition of TYK2 - we linked TYK2 genetic data with clinical data from electronic medical records (EMR) for 3,102 individuals (European ancestry). In a permutation-based analysis, we observed an enrichment of association at clinical diagnoses related to infection, suggesting that individuals carrying TYK2 RA-protective haplotypes are at higher risk of infection. Together, our results highlight the role of TYK2 in RA pathogenesis and suggest that pharmacological manipulation of TYK2 would be effective at treating the symptoms of RA with an adverse drug event profile that is predictable from human clinical data. These findings highlight the potential of integrating information from electronic medical records with human genetics for drug discovery in complex traits such as RA.

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