A Large Genome-Wide Association Study of Age-Related Hearing Impairment Using Electronic Health Records

Age-related hearing impairment (ARHI), one of the most common sensory disorders, can be mitigated, but not cured or eliminated. To identify genetic influences underlying ARHI, we conducted a genome-wide association study of ARHI in 6,527 cases and 45,882 controls among the non-Hispanic whites from the Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort. We identified two novel genome-wide significant SNPs: rs4932196 (odds ratio = 1.185, p = 4.0x10-11), 52Kb 3’ of ISG20, which replicated in a meta-analysis of the other GERA race/ethnicity groups (1,025 cases, 12,388 controls, p = 0.00094) and in a UK Biobank case-control analysis (30,802 self-reported cases, 78,586 controls, p = 0.015); and rs58389158 (odds ratio = 1.132, p = 1.8x10-9), which replicated in the UK Biobank (p = 0.00021). The latter SNP lies just outside exon 8 and is highly correlated (r2 = 0.96) with the missense SNP rs5756795 in exon 7 of TRIOBP, a gene previously associated with prelingual nonsyndromic hearing loss. We further tested these SNPs in phenotypes from audiologist notes available on a subset of GERA (4,903 individuals), stratified by case/control status, to construct an independent replication test, and found a significant effect of rs58389158 on speech reception threshold (SRT; overall GERA meta-analysis p = 1.9x10-6). We also tested variants within exons of 132 other previously-identified hearing loss genes, and identified two common additional significant SNPs: rs2877561 (synonymous change in ILDR1, p = 6.2x10-5), which replicated in the UK Biobank (p = 0.00057), and had a significant GERA SRT (p = 0.00019) and speech discrimination score (SDS; p = 0.0019); and rs9493627 (missense change in EYA4, p = 0.00011) which replicated in the UK Biobank (p = 0.0095), other GERA groups (p = 0.0080), and had a consistent significant result for SRT (p = 0.041) and suggestive result for SDS (p = 0.081). Large cohorts with GWAS data and electronic health records may be a useful method to characterize the genetic architecture of ARHI.Author Summary: Age-related hearing impairment (ARHI) is one of the most common sensory disorders. While ARHI effects can be mitigated with current technologies, it cannot be cured or eliminated. It is thus hoped that identification of genetic influences on ARHI may one day lead to curative therapies. Towards this goal, the current study utilized electronic health record data from non-Hispanic whites in the Genetic Epidemiology Research on Adult Health and Aging (GERA) cohort to conduct a genome-wide association study of ARHI, and tested the significant variants for replication in other GERA race/ethnicity groups, independent GERA phenotypes, and self-reported ARHI from the UK Biobank. We discovered two genome-wide significant SNPs. The first was novel and near ISG20. The second was in TRIOBP, a gene previously associated with prelingual nonsyndromic hearing loss. Motivated by our TRIOBP results, we also looked at exons in known hearing loss genes, and identified two additional SNPs, rs2877561 in ILDR1 and rs9493672 in EYA4 (at a significance threshold adjusted for number of SNPs in those regions). These results suggest that large cohorts with GWAS data and electronic health records may be a useful method to characterize the genetic architecture of ARHI.