Incomplete Retinal Pigment Epithelial and Outer Retinal Atrophy: Longitudinal Evaluation in Age-Related Macular Degeneration.

PURPOSE: To examine the association between incomplete retinal pigment epithelial and outer retinal atrophy (iRORA) on OCT imaging and the subsequent risk of developing geographic atrophy (GA) defined on conventional color fundus photography (CFP) and to compare this with the specific features that define nascent GA (nGA). DESIGN: Retrospective analysis of data from a longitudinal study. PARTICIPANTS: A total of 280 eyes from 140 participants with bilateral large drusen without specific nGA-defining features or late age-related macular degeneration (AMD) at baseline. METHODS: OCT imaging and CFP were performed at baseline and then at 6-month intervals for up to 36 months. Eyes that developed neovascular AMD were censored on the day it was detected. OCT volume scans were graded for the presence of iRORA and nGA separately, and CFP images were graded for the presence of GA. MAIN OUTCOME MEASURES: Association with and variance explained in time to GA development. RESULTS: A total of 58 eyes (21%) from 46 participants (33%) had iRORA at baseline, and a further 87 eyes (31%) developed iRORA over the follow-up period. Time-to-event analyses demonstrated that prevalent or incident iRORA was associated with an increased rate of GA development (adjusted hazard ratio [HR], 12.1; P = 0.021), as was incident nGA (adjusted HR, 78.6; P < 0.001). However, only the specific nGA features (adjusted P < 0.001), and not iRORA (adjusted P = 0.520), were associated with an increased rate of GA development when both features were included in the same multivariable model. The proportion of variance explained in the time to GA development by iRORA itself (R2 = 43%) was significantly lower than explained by nGA alone (R2 = 91%; P = 0.010). CONCLUSIONS: In this cohort, iRORA is a significant risk factor for GA development, but its association with GA development appears to be accounted for by the development of the specific features that define nGA. Although requiring replication, these findings provide useful guidance on the relative utility of nGA and iRORA as risk factors for GA and as potential surrogate end points for future interventional studies in the early stages of AMD.