Glaucoma Polygenic Risk Scores Demonstrate Heterogeneous Performance across 2 Large Multiethnic Cohorts

Glaucoma is a leading cause of irreversible blindness. While polygenic risk scores (PRSs) have shown promise for patient risk stratification, their consistency and reliability across diverse populations remain insufficiently characterized. This study aimed to determine whether published glaucoma PRSs demonstrate consistent population-level performance and individual-level risk assignment in 2 large, multiethnic cohorts.

A cross-sectional study.

A total of 243 300 individuals (7190 open-angle glaucoma [OAG] cases; 236 110 controls) from the All of Us (AoU) Research Program and 30 306 individuals (1899 cases; 28 407 controls) from the Mass General Brigham (MGB) biobank were included. Participants were aged ≥35 years with available genotype and electronic health record phenotype data.

We evaluated the performance of 11 published glaucoma PRSs in the Polygenic Score Catalog using logistic regression models. The agreement of PRSs in risk classification was evaluated using Pearson correlation and Jaccard index.

Associations between cataloged PRSs and OAG and agreement of PRSs in risk classification.

Higher PRSs were all significantly associated with higher odds of OAG; odds ratios (ORs) (per standard deviation) ranged from 1.17 to 1.41 (AoU) and 1.17 to 1.57 (MGB). In AoU, associations were strongest in individuals of European ancestry (ORs 1.22-1.55) and attenuated in African (1.08-1.21) and admixed American (1.09-1.33) ancestries. Despite consistent population-level performance, individual risk concordance across PRSs was low: median Pearson r = 0.42 (interquartile range, 0.35-0.50) and median Jaccard index = 0.16 (interquartile range, 0.13-0.19) for high-risk classification. This discordance persisted across cases, controls, ancestries, and both cohorts.

Our results demonstrate that while PRSs predict glaucoma risk at the cohort level, their instability at the individual level limits stand-alone clinical application. Polygenic risk scores should therefore be considered adjuncts to established glaucoma risk factors. Future work should prioritize developing tools to assess individual PRS reliability and standardizing analytic methods to improve comparability and clinical applicability.

Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.