Oral Presentation Annual Meetings of the Endocrine Society of Australia and Society for Reproductive Biology and Australia and New Zealand Bone and Mineral Society 2016

Genome-wide association study of bone mineral density in the UK Biobank Study identifies over 376 loci associated with osteoporosis (#142)

JOHN P KEMP 1 , JOHN A MORRIS 2 , CAROLINA MEDINA-GOMEZ 3 , CELIA GREGSON 4 , VINCENZO FORGETTA 2 , KATERINA TRAJANOSKA 3 , NICOLE M WARRINGTON 1 , JIE ZHENG 5 , STEPHEN KAPTOGE 6 , FERNANDO RIVADENEIRA 3 , JONATHAN H TOBIAS 4 , CHERYL L ACKERT-BICKNELL 7 , BRENT RICHARDS 2 , David M Evans 1
  1. University of Queensland Diamantina Institute, Woolloongabba, QLD, Australia
  2. Departments of Medicine, Human Genetics, Epidemiology and Biostatistics, Lady Davis Institute, McGill University, Montreal
  3. Department of Epidemiology, Erasmus Medical Center, Rotterdam
  4. School of Clinical Sciences, University of Bristol, Bristol
  5. MRC Integrative Epidemiology Unit, University of Bristol, Bristol
  6. Department of Public Health and Primary Care, University of Cambridge, Cambridge
  7. Center for Musculoskeletal Research, University of Rochester, New York

Quantitative ultrasound (QUS) is a non-invasive technique that measures the speed and attenuation of ultrasound through bone, and estimates bone mineral density (eBMD). To identify genetic loci associated with BMD variation, we used BOLT-LMM to perform a GWAS of eBMD of the heel on 140,623 individuals of European descent (75,275 females and 65,349 males) from the UK Biobank Study. Our GWAS identified 403 independent SNPs from 376 loci (defined as r2 < 0.05 and +/-500 MB) attaining genome-wide significance (P < 5x10-8), which jointly explained 13% of the variation in heel eBMD. These included the majority of SNPs previously associated with DXA derived BMD, as well as 308 novel loci, many of which contained genes that have not been previously implicated in bone physiology. LD score regression and bivariate G-REML analyses revealed moderate to high positive genetic correlations between heel eBMD and DXA-derived BMD measures, and a negative genetic correlation with fracture further confirming the validity of the eBMD measure. We also found significant genetic correlations with BMI, celiac disease, educational attainment, age at menarche and LDL cholesterol, but not cigarettes per day, despite the observational association between smoking and risk of osteoporosis. We implemented in-silico fine-mapping by constructing credible sets with a Bayesian method, followed by coding and non-coding SNP annotation. Results from fine-mapping strongly implicated novel and known genes and predicted causal SNPs. Preliminary functional studies showed increased expression of a novel gene Zhx3 in maturing calvarial osteoblasts and that Zhx3 knock-out mice had increased whole body BMD compared to wild type mice. In summary, using UK Biobank, we increased the number of BMD loci 5-fold, identified traits and diseases sharing etiologic pathways with osteoporosis, and implicated novel proteins which will serve as potential drug targets to improve the care of patients suffering from this common costly disease.