Genome-wide association studies for corneal and refractive astigmatism in UK Biobank demonstrate a shared role for myopia susceptibility loci

Abstract

Previous studies have suggested that naturally occurring genetic variation contributes to the risk of astigmatism. The purpose of this investigation was to identify genetic markers associated with corneal and refractive astigmatism in a large-scale European ancestry cohort (UK Biobank) who underwent keratometry and autorefraction at an assessment centre. Genome-wide association studies for corneal and refractive astigmatism were performed in individuals of European ancestry (N = 86,335 and 88,005 respectively), with the mean corneal astigmatism or refractive astigmatism in fellow eyes analysed as a quantitative trait (dependent variable). Genetic correlation between the two traits was calculated using LD Score regression. Gene-based and gene-set tests were carried out using MAGMA. Single marker-based association tests for corneal astigmatism identified four genome-wide significant loci (P < 5 × 10^-8) near the genes ZC3H11B (1q41), LINC00340 (6p22.3), HERC2/OCA2 (15q13.1) and NPLOC4/TSPAN10 (17q25.3). Three of these loci also demonstrated genome-wide significant association with refractive astigmatism: LINC00340, HERC2/OCA2 and NPLOC4/TSPAN10. The genetic correlation between corneal and refractive astigmatism was 0.85 (standard error = 0.068, P = 1.37 × 10^-35). Here, we have undertaken the largest genome-wide association studies for corneal and refractive astigmatism to date and identified four novel loci for corneal astigmatism, two of which were also novel loci for refractive astigmatism. These loci have previously demonstrated association with axial length (ZC3H11B), myopia (NPLOC4), spherical equivalent refractive error (LINC00340) and eye colour (HERC2). The shared role of these novel candidate genes for astigmatism lends further support to the shared genetic susceptibility of myopia and astigmatism.

Fig. 1

Estimates of SNP-heritability (h²_SNP) using GCTA. Error bars represent the standard error of the h²_SNP estimate

Fig. 2

Manhattan and Quantile–Quantile plots for GWAS of corneal astigmatism and refractive astigmatism using BOLT-LMM. a, b Corneal astigmatism; c and d refractive astigmatism. Manhattan plots (a, c): upper horizontal line indicates the genome-wide significance threshold at P = 5 × 10⁻⁸; lower horizontal line indicates P = 1 × 10⁻⁵. Quantile–Quantile plots (b, d): Y-axis shows observed negative log₁₀ p values and X-axis shows expected negative log₁₀ p values according to the null hypothesis of no genetic association. Diagonal line = line of unity (observed = expected)

Fig. 3

Regional association plots for loci demonstrating genome-wide significant association (P < 5 × 10⁻⁸) in GWAS for Corneal Astigmatism and Refractive Astigmatism using BOLT-LMM. a–d Corneal astigmatism; e–g refractive astigmatism. In order of chromosome: a rs12032649, b rs196052, c rs1129038, d rs62075722, e rs12196123, f rs1129038, and g rs34635363. Symbol shading denotes linkage disequilibrium (r²) values of variants with respect to the lead marker (named and highlighted). NB: rs14879552 is a synonym for rs12032649