Global, multi-ethnic genome-wide association meta-analysis of body mass index: A.E. Locke 1; S. Vedantam 2,3; E. Marouli 4; S. Berndt 5; L. Yengo 6; A.R. Wood 7; T. Ferreira 8; S. Graham 9; on behalf of the Genetic Investigation of ANthropometric Traits (GIANT) Consortium

Obesity is a health crisis affecting more than 1/3 of the US population and contributing to increased risk for other serious health conditions such as heart disease, diabetes, and numerous cancers. Using body mass index (BMI) as a proxy for overall obesity (heritability estimated at ~40-60%), GIANT and other groups have committed substantial effort to understand the genetic basis and underlying biology of obesity. The nearly 200 loci that have been associated with BMI show strong signal enrichment in brain tissues and the central nervous system. However, the BMI-associated loci discovered to date account for <3% of phenotypic variance. Here, we describe our continuing efforts to identify genetic factors contributing to overall obesity through GWAS meta-analyses of BMI. We combined published results from previous GIANT GWAS meta-analyses of BMI (n~250,000) with a BOLT-LMM association analysis of a recent UK Biobank release (n=456,426) for a combined total of 681,275 European-descent participants. We limited fixed effects meta-analysis to the ~2.4M HapMap2 variants common to both studies. Through meta-analysis and approximate conditional and joint analysis using GCTA-COJO we identified 941 loci at a stringent threshold of P<1x10-8, including 656 primary associations and 285 secondary signals. These associations account for 6.0% of the BMI phenotypic variance. In tandem, we initiated a global GWAS meta-analysis of BMI based on 1000 Genomes and HRC reference panels covering >40M imputed variants. To date, we have received association summary statistics from >170 studies across six ethnic groups totaling >820,000 individuals, nearly half of whom are from non-European ancestry studies. The upcoming final meta-analysis of BMI including all these studies plus UK Biobank and several other large-scale GWAS efforts will comprise a sample size of >2 million individuals. Our analytical pipeline, built to deal with this sample size, will enable us to perform flexible and adaptable ad hoc meta-analyses, including full conditional meta-analysis and rare variant group-wise testing of both coding and non-coding regions. This large-scale multi-ethnic endeavor provides an unparalleled opportunity for fine-mapping associations to pinpoint candidate functional variants, which will speed the translation of genomic discoveries to the clinic by providing insight into functional biology of obesity.