Importance Existing outcome measures for tracking geographic atrophy (GA) progression have disadvantages. As a functional measure, best-corrected visual acuity (BCVA) is associated with quality of life and recognized by regulators but is subjective and noisy (ie, with a high level of test-retest variability). As a structural measure, GA area is objective and less noisy (ie, with a low level of test-retest variability) but assumes all macular locations are equally important. Objective To derive and validate a novel outcome measure, the Geographic Atrophy Weighted-by-Acuity Index (GAWAIN). This was designed as a structural measure to align with BCVA by empirical weighting of macular pixels arranged in concentric annuli. Exposures Derivation and validation of novel GA outcome measure. Design, Setting, and Participants This prognostic study included outcome measure development using in silico analyses. The data for the training/validation set were taken from participants in the Age-Related Eye Disease Study 2 (AREDS2) study (4313 visits, 1528 eyes), and for the external validation set, data were taken from participants in the GA Minocycline Trial (218 visits, 35 eyes). Study data were analyzed from April to October 2025. Main Outcomes and Measures Pearson correlation between the novel GA outcome measure (0-100) and BCVA (35-85 letter score); area under receiver operating characteristic curve for clinical BCVA thresholds. The macula was divided into 60 concentric annuli; GA occupancy of each annulus was calculated. Annulus weights were derived using ridge regression, with GA involvement of each annulus and age as predictors and BCVA deficit (100 − BCVA) as outcome. Results Training/validation was performed on 1120 participants (mean SD age, 77.1 7.0 years; 645 female 57.6%), and external validation was performed on 35 participants (mean SD age, 74.3 7.1 years; 21 female 60.0%). The annulus weights exhibited a decelerating decline with macular eccentricity. On internal validation, the novel GA outcome measure was correlated more strongly with BCVA deficit than GA area (Pearson r = 0.58 vs 0.32, difference = 0.27; 95% CI, 0.23-0.31; P lt; .001). This was also true on external validation for color fundus photograph–defined GA (Pearson r = 0.69 vs 0.58, difference = 0.12; 95% CI, 0.04-0.21; P = .005) and fundus autofluorescence–defined GA (Pearson r = 0.70 vs 0.56, difference = 0.13; 95% CI, 0.05-0.22; P = .002). On internal and external validation, the novel GA outcome measure predicted BCVA letter scores less than 36, 70, and 85 (20/200, 20/40, and 20/20, respectively) more accurately than GA area (each P lt; .001). Longitudinally, change in the novel GA outcome measure correlated more strongly with BCVA decline than change in GA area did (eg, Pearson r = 0.37 vs 0.28, difference = 0.09; 95% CI, 0.04-0.14; P lt; .001; internal validation). Conclusions and Relevance In this prognostic study, results suggest that the novel GA outcome measure aligned with BCVA, a key functional outcome measure, by empirical weighting of concentric annuli. This provided a surrogate measure to track predicted functional decline objectively and more meaningfully than GA area.
Hou et al. (Thu,) studied this question.