Statistical driver genes and the missing heritability of age-related macular degeneration

Abstract

Purpose : Age-related macular degeneration (AMD) is a progressive disease of the retina that is shaped by environmental and genetic risk factors. Previous studies have estimated the heritability (h²) of AMD based on variants identified in genome-wide association studies (GWAS); however, a substantial proportion of AMD h² cannot be explained by known loci. The International AMD Genomics Consortium (IAMDGC) performed one of the largest GWAS on advanced AMD (ADV) cases and controls and identified 34 loci anchored by 52 independent genomic variants contributing to the majority of AMD h². Using pathway analysis of the IAMDGC GWAS data, we identified 8 statistical driver genes (SDGs) including 2 novel SDGs not discovered by the IAMDGC GWAS: PPARA and PLCG2. We chose to further investigate these risk genes and estimate their contribution to the h² of ADV and its subtypes.

Methods : We performed genomic-relatedness-based restricted maximum-likelihood (GREML) analyses on ADV (n = 16,144 cases), geographic atrophy (GA) (n = 3,235 cases) and choroidal neovascularization (CNV) (n = 10,749 cases) subtypes to investigate the h² of the common genotyped variants in the 8 SDGs and separately in the 2 novel SDGs. These estimates were calculated in isolate and in combination with variants from the 34 AMD loci identified in the IAMDGC GWAS using various prevalence estimates. Covariates were included for downstream analyses. This research was conducted in compliance with institutional review boards as described in the IAMDGC GWAS.

Results : Variants from the 8 SDGs account for a small percentage of ADV, GA, and CNV h² (3.76±0.39%, 2.53±0.36%, and 3.71±0.41%, respectively). Heritability estimates for ADV, GA, and CNV based on the combination of SDGs and the 34 known loci are not significantly different from those calculated for the known loci alone, indicating that the 2 novel loci we identified via pathway analysis may contribute to AMD in a non-additive manner. Covariates did not influence the h² in any subtype tested.

Conclusions : Pathway analyses of GWAS data, which interrogate networks of genes in biological contexts, may be useful in identifying novel loci that contribute to the h² of complex disorders, especially those with potentially non-additive effects. Heritability analyses of these loci, especially for disease subtypes, may provide insights into the importance of specific genes in the genetic architecture of AMD.

This is a 2020 ARVO Annual Meeting abstract.