Multimodal Imaging of Nonneovascular Age-Related Macular Degeneration

Genome-wide association meta-analysis for early age-related macular degeneration highlights novel loci and insights for advanced disease

“…Second, we contrasted advanced AMD effect sizes (IAMDGC data (9)) with early AMD effect sizes (our meta-analysis,) for the 34 known advanced AMD lead variants ( Figure 2, Table S13 ). We found two classes of variants: (1) 25 variants showed nominally significant effects on early AMD (P<0.05; “advanced-and-early-AMD loci”), all directionally consistent and all smaller for early vs. advanced AMD (OR early =1.04-1.47; OR adv =1.10-2.81); (2) nine variants had no nominally significant effect on early AMD (P≥0.05; “advanced-only AMD loci”). We did not find any variant with early AMD effects into the opposite direction as the advanced AMD effects.…”

“…We found several interesting aspects ( Table 3 ): (1) When prioritizing variants according to their statistical evidence for being the driver variant by computing 95% credible sets of variants (27), we found 23 and 294 credible set variants for the CD46 and TYR locus, respectively ( Table S3 ). (2) Using the Variant Effect Predictor (28), we assessed overlap of credible set variants with functional regulatory regions and found variants influencing the transcript and/or the protein for four genes ( Table S4 ): variants causing an alternative splice form for CD46 , a nonsense-mediated mRNA decay (NMD) for CR1L , a missense variant for TYR (rs1042602, r 2 =0.56 to the lead variant rs621313), and NMD variants for NOX4. (3) We investigated credible set variants for being an expression quantitative trait locus (eQTL) for any of the 17 genes in retina (Eye Genotype Expression database, EyeGEx (29)) or in 44 other tissues (Genotype-Tissue Expression database, GTEx (30)).…”

“…This would also suggest that mechanisms of complement system or lipid metabolism trigger both early and advanced disease. (2) For a mechanism that triggers advanced AMD no matter whether the person is “healthy” or has early AMD, we would anticipate a variant effect for advanced AMD, but not for early AMD ( Figure 3 , Model 2). This would be in line with our observed associations for the nine “advanced-AMD-only” loci.…”

“…The method assumes a single causal signal per locus. (2) Variant effect predictor (VEP) to explore whether any of the credible variants was located in a relevant regulatory gene region (28). (3) eQTL analysis: We downloaded expression summary statistics for the candidate genes in retina from the EyeGEx database (29) and for 44 other tissues from the GTEx database (30) (both available at www.gtexportal.org/home/datasets) and evaluated whether any of the credible variants showed significant effects on expression levels in the aggregated data.…”

“…We compared effect sizes in a scatter plot and clustered the lead variants by their nominal significant association on advanced and/or early AMD. We classify different types of loci in a similar fashion as done previously for adiposity trait genetics (41): (1) “advanced-and-early” AMD loci (P early <0.05, P adv <0.05), (2) “advanced-only” AMD loci (P early ≥0.05, P adv <0.05), (3) “early-only” AMD loci (P early <0.05, P adv ≥0.05).…”