Multi-Ancestry, Multitrait Polygenic Risk Scores for Myopia: Improved Accuracy and Clinical Potential

Abstract Myopia is an increasing global health concern and a leading cause of visual impairment. Genetic factors play a major role, and polygenic risk scores (PRSs) may help identify children at high risk of developing myopia. However, most PRSs are based on European populations, and accurately predicting risk across ancestries remains a challenge. We developed and evaluated PRSs for spherical equivalent refractive error (SER) and myopia using multitrait and multi‑ancestry genomewide association study data. A multitrait analysis of SER‑correlated traits identified 709 genomewide significant loci. PRSs were generated with SBayesRC for each ancestry group and for a combined multi‑ancestry model, and validated in the Australian Twins Eye Study and non‑European participants from the UK Biobank. The European PRSs explained approximately 20% of SER variance in Europeans and 18% in admixed Europeans and showed good transferability to South Asian (14%), East Asian (13%), and African (8%) groups. A multi‑ancestry PRS further improved prediction in Africans, explaining 9% of the variance. Predictive accuracy for high myopia was strong in the admixed group (AUC = 0.82, 95% CI [0.78, 0.87]), with all ancestry groups achieving AUCs of at least 0.70; European ancestry data were not available. PRS also predicted axial length in children, particularly those aged 5–8 years, where individuals in the lowest 10% of the PRS distribution had significantly longer axial lengths (β = 0.81 mm, p = 5.71 × 10 −3 ). These findings enhance genetic prediction of SER and myopia, showing the potential of multitrait, multi-ancestry PRS for early, equitable risk stratification.