Abstract
Purpose :
Refractive error abnormalities are some of the most common ocular diseases in the world. However, the genetic factors of refractive error are still not well known and genome-wide association analyses (GWAS) have identified common variants of moderate to low effect. Our study analyzed exome-based array data from multiple European-ancestry cohorts using gene-based association tests to identify rare variants of potentially large effect associated with refractive error.
Methods :
We analyzed over 13,000 individuals of European ancestry, collected by the CREAM consortium, with quantitative spherical equivalent values of refractive error and exome-based array genotypes, using two gene-based tests for rare variants: the combined multivariate and collapsing (CMC) and variable threshold (VT) tests. Each test was run multiple times using different values of two parameters: the allele frequency definition of rare variants (<= 0.05, <=0.01) and the inclusion of either synonymous and nonsynonymous variants or nonsynonymous variants only. We also had three covariates (age, sex, and education) and all analyses were run with either no covariates, age and sex only, or age, sex, and education. Gene-based p-values were obtained for all analysis sets using a significance value of 1 x 10-5.
Results :
Across all analyses, we consistently found TLE3, GSTCD, ST6GALNAC5, S100A16 and GDF15 to be genome-wide significant, with ST6GALNAC5 (part of the sialic acid metabolism pathway) having the lowest overall p-value 1 x 10-10. These findings appear to be novel; these genes have no previous association to refractive error or other eye disorders. Overall, we found that the CMC tests yielded roughly 20 genome-wide significant genes and VT had approximately 60 genome-wide significant genes, with the CMC p-values slightly lower. The top genes tended to appear in both tests. Including all covariates tended to yield more significant gene associations. We also validated multiple genes found to be associated with refractive error in single variant GWAS, including CABP4.
Conclusions :
We have identified over 60 genome-wide significant genes for refractive error across our multiple analyses using the exome-based array and consistently identified the same top genes. This work is one part of a larger meta-analysis and will be combined with data from other European, Asian, and African ancestry datasets.
This is a 2020 ARVO Annual Meeting abstract.