Large scale meta analysis of 250,000 individuals reveals novel biological pathways involved in adult human height. T. Esko1, AR. Wood2, S. Vedantam1, J. Yang3, TH. Pers1, SI. Berndt4, MN. Weedon2, G. Lettre5, J. O'Connell6, DI. Chasman7, G. Abecasis8, ME. Goddard3, RJF. Loos9,10, E. Ingelsson11, PM. Visscher3, JH. Hirschhorn1, TM. Frayling2, on. behalf of GIANT Consortium1 1) Divisions of Endocrinology, Boston Children's Hospital and Broad Institute, Cambridge, MA, USA; 2) Genetics of Complex Traits, Exeter Medical School, Exeter, UK; 3) University of Queensland Diamantina Institute, University of Queensland, Princess Alexandra Hospital, Brisbane, Queensland, Australia; 4) 4Division of Cancer Epidemiology and Genetics, National Cancer Institute, National Institutes of Health, Department of Health and Human Services, Bethesda, MD, USA; 5) Montreal Heart Institute, Montreal, Quebec, Canada; 6) Deptartment of Medicine, University of Maryland School of Medicine, Baltimore, MD, USA; 7) 7Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA; 8) Department of Biostatistics, Center for Statistical Genetics, University of Michigan, Ann Arbor, Michigan, USA; 9) MRC Epidemiology Unit, Institute of Metabolic Science, Cambridge, UK; 10) Mount Sinai School of Medicine, New York, NY, USA; 11) Department of Medical Sciences, Uppsala University Hospital, Uppsala, Sweden.
Adult height is a highly heritable polygenic trait that reflects the outcome of childhood growth, a fundamental developmental process. Studies of height have already employed large sample sizes (~130,000) and identified more loci (180) than for any other polygenic trait or disease, but have nonetheless only accounted for one tenth of the phenotypic variation. Although further expansion of GWAS sample size is predicted to substantially augment the explained heritability, it is not yet known if newly identified variants with very small effects will keep providing novel insight to human growth biology. We report results from nearly doubling the sample size of existing height GWAS data to ~250,000 individuals of European ancestry. By applying an approximate conditional analysis we identified 697 independent variants shown to cluster into 424 genomic loci (P<1e-4). Associated variants were not only non-randomly distributed with respect to functional and putatively functional regions of the genome (nsSNPs, eQTLs and epigenetic marks) but showed high allelic heterogeneity (262 secondary signals) in both established and novel loci. Variants in strictly novel loci were more prominently enriched for eQTLs and not for nsSNPs, which suggest that associations with smaller effect sizes will increasingly point to regulatory variants. By applying both existing and novel pathway enrichment and gene prioritizing bioinformatics tools, we provide evidence that identification of many 100s and even 1000s of associated variants will continue to provide biologically relevant information. The larger number of loci highlight many more significantly enriched growth-related pathways, including signalling by insulin-like growth factors, Hedgehog, WNT, BMPs and TGF-beta, mTOR, and MAPK, as well as chromatin remodelling. Our results prioritize 639 genes (FDR <5%), which reveal enriched expression in growth-related tissues (cartilage; P<1e-12) and include many known skeletal growth syndrome genes, This list provides new candidates for regulating human growth or for underlying diseases of abnormal growth. In conclusion, larger sample sizes have identified further height associated variants that as a group provide novel and more specific biological insights into human growth. By extension, larger GWA studies from other diseases and traits, for which there are fewer associated loci than for height, are likely to continue to provide additional insights into underlying biology.