Abstract
Purpose :
To determine if family history of age-related macular degeneration (AMD) and genetic variants identify eyes at higher risk for progression to advanced AMD (AAMD), after controlling for baseline demographics, behavioral factors, and macular status.
Methods :
Eyes with non-advanced AMD at baseline in the AREDS longitudinal cohort were classified using the AREDS severity score. Non-genetic and genetic predictors for progression to AAMD were evaluated. Cox proportional hazards models based on the eye as the unit of analysis were used to calculate hazard ratios (HR), accounting for correlated data. A composite risk score was calculated using beta estimates from demographic and behavioral variables (age, race, BMI, smoking status, AREDS treatment group, multivitamin intake), ocular factors (baseline AMD severity group and AMD status of fellow eye), AMD family history, and 12 statistically significant genetic variants (complement, angiogenesis, lipid, inflammatory, extra-cellular matrix, and DNA repair pathways) were selected from stepwise regression. Discrimination between progressing and non-progressing eyes in various models with different predictors was assessed using C-statistics and Net Reclassification Improvement (NRI).
Results :
Among 4910 eyes, 863 progressed to AAMD over 12 years. Baseline AMD severity group and status of the fellow eye were important predictors, and genetic factors provided additional discrimination. Family history of AMD also independently predicted progression after accounting for genetic and other covariates: 1 family member vs. none (HR = 1.21, 95% CI: 1.02, 1.43; P = 0.03); ≥ 2 family members vs. none (HR= 1.55, 95% CI: 1.26, 1.90; P < 0.001). The composite risk score predicted progression to AAMD (HR = 5.57, 90th vs. 10th percentile), providing superior fit versus other models with only ocular or non-genetic variables (NRI, P < 0.001). It also discriminated between progressing and non-progressing eyes within each baseline AMD severity group (see figure). An online risk calculator is available to implement these methods.
Conclusions :
Genetic variants and family history provided additional discrimination in AAMD prediction models, after accounting for ocular and other covariates.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.