Discovery of three novel risk loci for age-related macular degeneration by trans-ethnic genome-wide association analysis in the VA Million Veteran Program

Abstract

Purpose : The Million Veteran Program (MVP) hosts a biobank consisting of DNA specimens from >600K US Veterans, with genotyping available for >480K subjects of varying ethnic backgrounds. Using individual survey data and electronic health records from this repository, we conducted genome-wide association (GWAS) analyses for age-related macular degeneration (AMD).

Methods : We evaluated 8 algorithms for AMD phenotyping by assessing the strength of association between each phenotype and DNA variants at the known AMD loci CFH and ARMS2/HTRA1 loci to choose the most favorable phenotypic contrasts. Defining AMD cases as having at least two ICD9/ICD10 codes for the presence of AMD in Veterans >50 years of age, and controls as having either eye exam visits with no AMD codes or no eye exam visits in participants >65 years of age led to effect estimates that best approximated those reported in the literature. We then tested 312,204 common DNA variants (minor allele frequency >1%) for an association with this AMD phenotype using logistic regression and adjusting for sex and principal components of ancestry, separately in European-American (EA; 17,818 cases/142,522 controls), African-American (AA; 1,247/26,720) and Hispanic-American (HA; 544/7,425) Veterans. A multi-ethnic fixed effects meta-analysis was conducted (total N = 196,276).

Results : In the EA sample, 12 of 34 known AMD risk loci (Fritsche et al., Nat Genet 2016;48:134) were detected at genome-wide significance (GWS) and 8 more had a suggestive association (p<0.0001). New GWS loci in OCA2/HERC2 (15q; rs12913832) and near CLUL1 (18p; rs485741), and suggestive loci at CAND2 (3p; rs2305397) and COL10A1 (6q; rs1064583) were identified. Compared with other groups, the AA genetic risk profile was the most dissimilar, with only the HLA region (rs2858870) achieving a GWS signal. The HA sample attained GWS at CFH and ARMS2/HTRA1, and had suggestive association at C2 and HLA. The trans-ethnic meta-analysis was dominated by EA signals, with COL10A1 and OCA2/HERC2 loci attaining GWS alongside 9 previously reported loci (Fritsche et al., 2016).

Conclusions : We conducted the largest multi-ethnic GWAS for AMD. We confirmed prior risk loci for EA and identified three novel loci. The genetic risk profile of AA differs widely from EA and HA Veterans.

This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.