Risk Factors for 15-Letter Visual Acuity Loss from Geographic Atrophy Progression Over One Year in the Age-Related Eye Diseases Study 2.

Abstract

Purpose: A change of ≥ 15 letters in best-corrected visual acuity (BCVA) is typically defined as clinically significant by regulatory agencies, but risk factors for rapid 15-letter loss in geographic atrophy (GA) are poorly understood. The purpose was to identify independent risk factors for 15-letter loss within one year in eyes with GA.

Design: Post hoc analysis of the Age-Related Eye Disease Study 2.

Participants: 961 eyes (743 participants).

Methods: Annual fundus photographs were graded for GA presence/morphology. BCVA was measured using the Early Treatment Diabetic Retinopathy Study chart. Multivariable analyses comprised logistic regression for 15-letter loss within one year, based on (i) baseline variables (demographic, BCVA, and GA morphology variables, defined at first time-point with GA), (ii) genetic variables (CFH Y402H and ARMS2), and (iii) GA enlargement rate (from first time-point with GA).

Main outcome measures: 15-letter loss in BCVA within one year.

Results: During 1-year follow-up, BCVA declined by ≥ 15 letters in 53 eyes (5.5%). In a model with baseline variables, the risk factors were: age (adjusted odds ratio [aOR] 1.08, 95% confidence interval [CI] 1.04-1.14, p=0.0005), closer GA proximity to fovea (aOR 0.90 per 0.1 mm increase, 95% CI 0.84-0.97, p=0.005), current smoking (aOR 3.85, 1.49-9.97, p=0.005), and BCVA <20/40 (aOR 2.09, 95% CI 1.17-3.74; p=0.013). In a model with baseline variables and genotype, CFH was a risk factor (aOR 4.63, 1.27-16.9, p=0.020, 2 vs 0 risk alleles), while ARMS2 was not. In a model with baseline variables and GA enlargement rate, faster enlargement was a risk factor (aOR 1.12 per 0.1 mm/year increase, 95% CI 1.07-1.18, p<0.0001).

Conclusions: We identified multiple independent risk factors for rapid, clinically significant BCVA loss in GA. We also developed clinically relevant models for different scenarios. These can guide the design and interpretation of interventional trials aimed at decreasing vision loss in GA and provide prognostic information in clinical practice. The risk factors for BCVA loss and faster GA enlargement overlap only partially, so that trial inclusion criteria, power calculations, and covariate adjustment should differ according to the choice of a functional versus structural measure as the primary endpoint.