A genetic risk score and number of myopic parents independently predict myopia

Abstract Purpose To investigate whether a genetic risk score ( GRS ) improved performance of predicting refractive error compared to knowing a child's number of myopic parents ( NMP ) alone. Methods This was a retrospective analysis of data from the Avon Longitudinal Study of Parents and Children ( ALSPAC ) birth cohort study. Refractive error was assessed longitudinally between age 7–15 using non‐cycloplegic autorefraction. Genetic variants ( n = 149) associated with refractive error from a Consortium for Refractive Error And Myopia (CREAM) genome‐wide association study were used to calculate a GRS for each child. Using refractive error at ages 7 and 15 years as the outcome variable, coefficient of determination ( R 2 ) values were calculated via linear regression models for the predictors: NMP , GRS and a combined model. Results Number of myopic parents was weakly predictive of refractive error in children aged 7 years, R 2 = 3.0% (95% CI 1.8–4.1%, p < 0.0001) and aged 15 years, R 2 = 4.8% (3.1–6.5%, p < 0.0001). The GRS was also weakly predictive; age 7 years, R 2 = 1.1% (0.4–1.9%, p < 0.0001) and 15 years R 2 = 2.6% (1.3–3.9%, p < 0.0001). Combining the 2 variables gave larger R 2 values at age 7, R 2 = 3.7% (2.5–5.0%, p < 0.0001) and 15, R 2 = 7.0% (5.0–9.0%, p < 0.0001). The combined model improved performance at both ages (both p < 0.0001). Conclusion A GRS improved the ability to detect children at risk of myopia independently of knowing the NMP . We speculate this may be because NMP captures information concerning environmental risk factors for myopia. Nevertheless, further gains are required to make such predictive tests worthwhile in the clinical environment.