Contribution of low-frequency variants to variation in body mass index (BMI). V. Turcot1,2, Y. Lu3, J. Czajkowski4, H. M. Highland5, N. G. D. Masca6, A. Giri7, T. L. Edwards7, T. Esko8,9, M. Graff10, A. E. Justice10, C. Medina-Gomez11, C. Schurmann3, R. A. Scott12, K. Sin Lo1, S. S. Sivapalaratnam13,14, L. Southam15,16, K. Stirrups15, T. W. Winkler17, H. Yaghootkar18, K. L. Young10, A. L. Cupples19, T. M. Frayling18, J. N. Hirschhorn20,21,22, G. Lettre1,2, C. M. Lindgren23, K. E. North10,24, I. B. Borecki4, R. J. F. Loos3 For the BBMRI, the GOT2D, the CHARGE, and the GIANT Consortia 1) Montreal Heart Institute, Montréal, Québec, Canada; 2) Faculty of Medicine, Université de Montréal, Montréal, Québec, Canada; 3) The Genetics of Obesity and Related Metabolic Traits Program, The Charles Bronfman Institute for Personalized Medicine, Icahn School of Medicine at Mount Sinai, New York, NY, USA; 4) Department of Genetics Division of Statistical Genomics, Washington University School of Medicine, St. Louis, MO, USA; 5) Human Genetics Center, University of Texas Health Science Center, Houston, TX, USA; 6) NIHR Leicester Cardiovascular Biomedical Research Unit, University of Leicester, Leicester, UK; 7) Center for Human Genetics Research, Division of Epidemiology, Department of Medicine, Vanderbilt University, Nashville, TN, USA; 8) Estonian Genome Center, University of Tartu, Tartu, Estonia; 9) Children's Hospital Boston & Broad Institute, MA, USA; 10) Department of Epidemiology, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA; 11) Netherlands Consortium for Healthy Aging (NCHA), Departments of Epidemiology and Internal Medicine, Erasmus Medical Center, Rotterdam, The Netherlands; 12) MRC Epidemiology Unit, Institute of Metabolic Science, Addenbrooke's Hospital, University of Cambridge, Cambridge, UK; 13) Academic Medical Centre, Amsterdam, The Netherlands; 14) CGHR, Boston, MA, USA; 15) The Wellcome Trust Sanger Institute, Wellcome Trust Genome Campus, Hinxton, Cambridge, UK; 16) Wellcome Trust Centre for Human Genetics, University of Oxford, Oxford, UK; 17) Department of Genetic Epidemiology, Institute of Epidemiology and Preventive Medicine, University of Regensburg, Regensburg, Germany; 18) University of Exeter Medical School, Exeter, UK; 19) Boston University School of Public Health, Boston, MA, USA; 20) Divisions of Endocrinology and Genetics and Center for Basic and Translational Obesity Research, Boston Children's Hospital, Boston, MA, USA; 21) Broad Institute of the Massachusetts Institute of Technology and Harvard University, Cambridge, MA, USA; 22) Department of Genetics, Harvard Medical School, Boston, MA, USA; 23) Program in Medical and Population Genetics, Broad Institute of Harvard and MIT, Cambridge, MA, USA; 24) Carolina Center for Genome Sciences, University of North Carolina at Chapel Hill, Chapel Hill, NC, USA.
Genome-wide association studies (GWAS) have identified >90 loci for BMI. While low-frequency exonic variants are known to cause extreme and early-onset obesity, little is known about their role in obesity susceptibility in the general population. To estimate the contribution of low-frequency (MAF<5%) variants to variation in BMI, we performed a meta-analysis of exome array data in up to 249,395 individuals of predominantly European descent from 79 studies. Each study tested up to 246,127 autosomal and 5,072 X-chromosomal single nucleotide variants (SNVs) for association with inverse normally transformed residuals of BMI, adjusted for age and sex. Study-specific association results were combined using inverse variance-weighted meta-analysis and associations were considered significant if P<5x10-7. Three low-frequency autosomal SNVs were significantly associated with BMI. A rare SNV with a large effect on BMI was found in an unknown protein-coding gene (KIAA0754, MAF: 0.04%, P=4.8x10-7, effect sizeSE: 0.600.12 SD/minor allele [2.5 kg.m-2, or 7.2 kg for a 1.7m-tall person]). Two SNVs were located in genes (GPR61, ZBTB7B) also harboring common SNVs associated with BMI. The SNV in GPR61 (3.7%, P=2.8x10-21, 0.080.01SD/MA) is intronic, but flanked by a more common coding SNV (6.6%, P=1.3x10-10, 0.040.01SD/MA) in SYPL2, and an intronic SNV (4%) in GNAT2 that was previously GWAS-identified. While fine-mapping will be needed to identify the causal gene in this locus, of interest is that GRP61-deficient mice exhibit obesity and hyperphagia. The second low-frequency SNV is a missense variant in ZBTB7B (3.7%, P=3.8x10-8, 0.050.01SD/MA), located at the downstream-end of a long-range association peak (1Mb) of common SNVs (MAF>25%) that include coding variants in EFNA1 and UBQLN4. Conditional analyses to determine whether the low-frequency and common SNVs represent independent signals are ongoing. No associated X-chromosomal low-frequency SNVs were identified. Despite our large sample size, we identified only three low-frequency SNVs. We cannot exclude the possibility of other low-frequency SNVs with smaller effect sizes. Preliminary results indicate that associated low-frequency SNVs may be located near common SNVs and conditional analyses are needed to determine which SNVs are driving the associations. Ongoing analyses also include an expansion of the sample (>400,000), gene-based analyses and functional follow-up.
You may contact the first author (during and after the meeting) at