POAG genetic risk score performs worse in African-descent than European-descent samples, highlighting need for expanded genetic studies in diverse populations

Abstract

Purpose : Genetic risk scores (GRS), sums of additive genetic risk, have strong potential for application in clinical risk stratification and early disease diagnosis in primary open-angle glaucoma (POAG). Recent studies in primarily European-descent population samples suggest that GRS associate with POAG case status, diagnosis age, and disease-specific outcomes. Further study in diverse populations is necessary to ensure generalizability of GRS in POAG, particularly as POAG is generally worse in people of African descent (earlier onset, higher prevalence and likelihood of vision loss/blindness). We investigated a POAG-specific GRS of published risk variants and apply it to the diverse Million Veteran Program (MVP), an ongoing observational cohort study and mega-biobank with >825K Veterans.

Methods : GRS summing 83 POAG-associated risk alleles in MVP samples with POAG phenotype, imputed genetic data, and harmonized ancestry and race/ethnicity (HARE) data. Association analyses of unweighted and effect estimate-weighted GRS and binary POAG phenotype were performed via logistic regression in European-descent (3382 cases, 58,829 controls) and African-descent (2448 cases, 5665 controls) samples. Unadjusted models and models adjusting for age, sex, age*sex interaction, and 10 sample-specific principal components were evaluated. GRS score deciles and quartiles were tested for association with POAG phenotype via logistic regression. Receiver operating characteristic (ROC) curves were composed and statistically compared.

Results : GRS was significantly associated with POAG in unadjusted and adjusted models (P<5E-05). While odds ratio estimates generally increased across deciles, the trend was stronger in the European-descent sample than in the African-descent sample; this was consistently supported by ROC plots and statistical hypothesis testing. GRS identified European-descent cases more consistently than African-descent cases (top quartiles: 37%, 29%, respectively). Unweighted GRS outperformed the weighted GRS in all models.

Conclusions : The ability of known POAG-associated variants to differentiate POAG cases from controls was dampened in African-descent population sample compared to European-descent. These outcomes drive the necessity of further diverse POAG genomic studies.

This is a 2021 ARVO Annual Meeting abstract.