Whole-exome sequencing in age-related macular degeneration

Abstract

Purpose: Genome-wide association meta analysis (GWAMA) has implicated common variations in 19 genes as AMD risk factors. Rare variants (RV) are not well captured by genome wide association studies and may represent some of the unexplained heritability in AMD risk. We used whole exome sequencing (WES) in phenotypically extreme individuals to identify RVs implicating novel AMD genes.

Methods: The GWAS 19 loci were used to calculate a genetic risk score (summed number of risk alleles weighted by effect sizes) for each individual. We created an ‘extreme’ case/control sample with the following characteristics: 20 individuals with bilateral neovascular AMD and the lowest calculated genetic risk score, and youngest ages at examination; 20 unaffected controls with no drusen, the highest calculated genetic risk score, and oldest ages at examination. Sequencing Capture was by SureSelect. Alignment and base calling used the Illumina CASAVA 1.6 pipeline, aligned to hg19 using BWA. Single nucleotide variants (SNV) and insertion-deletion variants (indels) were called by GATK Unified Genotyper with VQSR recalibration. Variants with VQSLOD<-3 and variant genotype likelihoods < 99 were excluded. All variants were annotated using SeattleSeq Annotation. Association of individual SNV with AMD was assessed by Fisher’s exact test. RVASSOC was used for the gene-based tests.

Results: No gene-based tests met Bonferroni-corrected significance (2.2 x 10-6). The most significant genes identified in the overall analysis are all novel. LRP1 is implicated both in the overall analysis and when considering only rare and damaging SNVs. The other most significant genes in the two analyses are different. Six of the 19 known loci have nominally significant gene-based tests. Associated SNVs in each gene are previously known variants.

Conclusions: Initial results of WES suggest additional genes with multiple rare SNVs may influence AMD. LRP1 is an APOE receptor and also involved in cholesterol transport. Restricting analysis to rare damaging variants reveals different results than considering the overall burden of SNVs, suggesting both analyses are needed to extract maximum information from sequence data. Novel coding variants in known AMD genes do not appear to explain the extreme phenotype studied here. WES of a replication dataset with an additional 20 cases/20 controls with extreme phenotype is in progress.

Keywords: 412 age-related macular degeneration • 539 genetics