Genome-wide association study imputed to 1000 Genomes reveals 18 novel associations with type 2 diabetes. R. A. Scott1, R. Magi2, A. P. Morris3, L. Marullo4, K. Gaulton5, M. Boehnke6, J. Dupuis7, M. I. McCarthy5, L. J. Scott6, I. Prokopenko8, DIAGRAM+ consortium 1) MRC Epidemiology Unit, University of Cambridge School of Clinical Medicine, Institute of Metabolic Science, Cambridge Biomedical Campus, Cambridge, UK; 2) Estonian Genome Center, University of Tartu, Tartu, Estonia; 3) Department of Biostatistics, University of Liverpool, Liverpool, UK; 4) Department of Life Sciences and Biotechnology, University of Ferrara, Ferrara, Italy; 5) University of Oxford, Wellcome Trust Centre for Human Genetics, Oxford OX3 7BN, UK; 6) Center for Statistical Genetics and Dept of Biostatistics, University of Michigan, Ann Arbor, Michigan 48109, USA; 7) Department of Biostatistics, Boston University School of Public Health, Boston, Massachusetts, 02118, USA; 8) Genomics of Common Disease, Imperial College London, London, UK.
Genome-wide association studies (GWAS) imputed to Hapmap reference panels have identified over 70 loci showing associations with type 2 diabetes (T2D). Associated variants identified by GWAS to date are exclusively common (MAF>5%). The 1000 Genomes (1000G) reference panel is a comprehensive catalogue of variation down to 1% MAF and allows imputation of a larger number of variants, including many at lower frequency. Here, we aimed to identify new low frequency and common variant associations with T2D by 1000G imputation. We performed an inverse-variance weighted meta-analysis of genome-wide data from up to 26,676 individuals with T2D and 132,532 controls of European ancestry from 18 studies. Genetic data were imputed using the March 2012 1000G multi-ethnic reference panel. Following discovery analyses, we sought follow-up for single nucleotide variants (SNVs) reaching 5x10-5 in up to 14,545 cases and 38,994 controls of European ancestry genotyped on the Metabochip. We also defined credible sets of SNVs in a 1Mb window that were 99% likely to contain the causal variant. Analysis of ~12M SNPs revealed 18 loci showing new associations with T2D (p<5x10-8). All lead SNVs were common (MAF>10%). Seven loci contained genome-wide significant associations in discovery analyses, including variants in or near CENPW (chr6:126792095; OR [95% CI]=1.10[1.06,1.13]), HSF1 (chr8:145536056; 1.08[1.05,1.11]), PLEKHA1 (chr10:124186714; 1.09[1.06,1.11]), HSD17B12 (chr11:43877934; 1.08[1.05,1.11]), CMIP (chr16:81534790; 1.08[1.05,1.10]), APOE (chr19:45411941; 1.13[1.09,1.17]) and HORMAD2 (chr22:30599562; 1.13[1.09,1.18]). Eleven further signals were seen after follow-up in or near GLP2R, ATP5G1, NHEG1, MAP3K11, ACSL1, ABO, TCF19, HLA-DQ1, UBE3C, NRXN3, and ZZEF1. In credible-set mapping of 48 known loci, 17 contained fewer than 20 SNVs in the 99% credible set. We identified seven loci for which the previous Hapmap lead SNV was not included in the 99% credible set, including BCAR1, where the 1000G-lead SNV (OR=1.16, p=3.7x10-11) had a stronger association than the previous HapMap lead (OR=1.1, p=1.9x10-6). Imputation using 1000G reference panel has allowed identification of 18 novel associations with T2D and identified new lead SNVs at known T2D loci, including some for which the previous Hapmap-based lead SNV is no longer in the credible set. Further annotation analyses in these regions will inform the extent to which 1000G imputation can aid in fine-mapping T2D associations.
You may contact the first author (during and after the meeting) at