Contribution of coding variation to type 2 diabetes-related quantitative traits in 13,000 exomes from multiple ancestries. X. Sim1, H. M. Highland2, M. A. Rivas3, H. K. Im4, A. K. Manning5, A. Mahajan3, A. E. Locke1, N. Grarup6, P. Fontanillas5, A. P. Morris3, T. M. Teslovich1, J. Flannick5, C. Fuchsberger1, K. Gaulton3, H. M. Kang1, J. B. Meigs7, C. M. Lindgren3, T2D-GENES and GoT2D Consortia 1) Department of Biostatistics and Center for Statistical Genetics, University of Michigan, Ann Arbor, MI, USA; 2) Human Genetics Center, University of Texas Health Science Center at Houston, Houston, TX, USA; 3) Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK; 4) Department of Health Studies, University of Chicago, Chicago, USA; 5) Program in Medical and Population Genetics, Broad Institute, Cambridge, MA, USA; 6) Novo Nordisk Foundation Center for Basic Metabolic Research, Faculty of Health and Medical Sciences, University of Copenhagen, Copenhagen, DK; 7) Massachusetts General Hospital, Boston, MA, USA.

   Within the T2D-GENES and GoT2D consortia, we performed deep exome sequencing of 18,162 genes in ~13,000 type 2 diabetes case-control individuals from five major ancestry groups (African-American, East Asian, European, Hispanic, and South Asian). We are carrying out quantitative trait (QT) association analysis to assess the role of coding variation in type 2 diabetes-related lipid, anthropometric, and glycemic traits (the latter only analyzed in non-diabetic controls). Current analysis of exome sequence data from ~10,000 multi-ethnic individuals (2,000 from each ancestry group) identified ~2.5 million variants, 40% of which are non-synonymous. Within the non-synonymous variants, only 4% are polymorphic across all ancestry groups while 76% are ancestry-specific. Similarly, 5% of synonymous variants are polymorphic across all ancestry groups and 78% are ancestry-specific. Applying a combination of single variant and gene-level association tests, some interesting results from analyses to date identified (1) coding variants associated with diabetes-related traits specific to different ancestries, and (2) multiple non-synonymous variants exhibiting allelic heterogeneity across different ancestries. Gene-based association identified four loss-of-function variants significantly associated with LDL-cholesterol at APOB (P = 4x10-7), present in four individuals (1 African-American, 1 Hispanic, and 2 South Asians), who exhibited extremely low LDL-cholesterol levels (< 60mg/dl). We observed exome-wide significant gene-based association signals for body mass index (BMI) at EIF3G (P = 6.9x10-7), specific to African-Americans. The gene-based association was driven primarily by a single missense variant (P = 1.2x10-7) observed in five individuals, all of whom had BMI < 25kg/m2. Finally, we detected evidence of association with decreased glycated hemoglobin (HbA1c) at the G6PD locus in African American (minor allele frequency [MAF] = 0.12), East Asian (MAF = 0.02), and South Asian (MAF = 0.02) ancestries, at different index missense variants, suggesting allelic heterogeneity at this locus. This diverse catalog of coding variation across wide allelic spectrum will further facilitate characterization of coding variants, and larger sample sizes may be needed to elucidate the role of low frequency and rare exonic variant associations in diabetes-related quantitative traits.

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