Genome-wide association study meta-analysis brings monogenic hearing loss genes into the polygenic realm

BackgroundDisabling age-related and noise-induced sensorineural hearing loss (SNHL) affects 5% of the global population and is associated with isolation, depression, cognitive decline, and dementia. SNHL is strongly heritable with a polygenic signature of common genetic variants conferring small risks individually, thus requiring very large studies to achieve statistical power in genome-wide association studies (GWAS). MethodsWe present the first report of a clinical SNHL GWAS in the Million Veteran Program (MVP) cohort. GWAS findings are examined separately within MVP (210,240 cases and 265,275 controls), including multi-ancestry analysis, then combined and contrasted with the United Kingdom Biobank (UKB) self-reported hearing loss study (87,056 cases and 163,333 controls). We perform functional mapping and annotation, gene prioritization, gene-based and gene-set analysis, and cochlear cell type enrichment, including human single-cell results, then compare our results to known hereditary hearing loss genes. Summary results are leveraged to characterize the genetic architecture of SNHL. ResultsSubstantial genetic overlap is seen between MVP and UKB despite differences in phenotypes, demographics, and environmental exposures. Individual GWAS and meta-analyses identify 108 loci, including 54 loci containing novel prioritized genes and/or protein-coding genes. Significant gene ontology pathways include "sensory perception of mechanical stimulus," "ear development," "actin-binding," and "cytoskeletal protein binding," indicative of mechanosensory structure and function in the cochlea. GWAS and gene-based analyses implicate 17 known familial hearing loss genes, expressing proteins in regions of hair cells, Type 1 neurons, stria vascularis, and subcellular regions of stereociliae, rootlets, and notably, mechanoelectrical transduction tip-links. Although risk for polygenic SNHL is predominantly captured by SNPs outside of congenital HHL genes (HHL), SNP-based partitioned heritability estimates show a 3.26-fold enrichment of HHL relative to other genes. ConclusionIn the largest GWAS to date, combining clinical diagnosis of hearing loss and self-report data from two cohorts, we identify 108 loci with 54 novel genes. Despite the increased enrichment of HHL genes, 97% of the risk for adult SNHL is captured by SNPs outside of HHL genes. Although SNHL in the UKB and MVP were assessed using different phenotypes, genetic signals between the two cohorts are predominantly shared, and locus discovery is boosted through increased sample size in meta-analysis.