Assessment of Ellipsoid Zone Integrity and Other Quantitative OCT Biomarkers for Intermediate AMD Progression to Atrophy.

PURPOSE: To evaluate the association between baseline optical coherence tomography (OCT) biomarkers and 2-year disease progression from intermediate age-related macular degeneration (iAMD) to advanced atrophic AMD, defined by OCT, with emphasis on ellipsoid zone (EZ) attenuation/loss and other quantitative structural features. DESIGN: Retrospective cohort study. PARTICIPANTS: 502 eyes with iAMD and no evidence of atrophy (i.e., RPE loss with associated outer retinal atrophy) or exudation at baseline, each with 2-year follow-up OCT imaging. METHODS: Baseline and 2-year spectral-domain OCT scans were analyzed using a validated, machine learning-enhanced multilayer segmentation platform with subsequent certified reader review and correction. Quantitative OCT parameters included EZ-RPE thickness, partial and total EZ attenuation, drusen volume, hyperreflective foci (HRF) count, total RPE loss, and outer nuclear layer-RPE (ONL-RPE) thickness. Fully automated deep-learning models quantified hypertransmission and EZ at-risk (i.e., a model developed to identify areas of abnormal EZ-RPE thinning in the absence of RPE loss). A random forest classifier was trained using baseline features, with performance assessed using 5-fold stratified cross-validation. MAIN OUTCOME MEASURES: Development of OCT-defined advanced atrophic AMD defined by total RPE loss with associated outer retinal atrophy with an area ≥0.05 mm2 (i.e., cRORA equivalent) at 2 years and baseline OCT biomarkers associated with progression. RESULTS: Eyes that developed advanced atrophic AMD showed significantly greater partial and total EZ attenuation, reduced EZ-RPE and ONL-RPE thickness, higher drusen volume, and greater HRF counts than non-converters (all p<0.05). Deep-learning-derived EZ at-risk and hypertransmission metrics were also significantly higher in converters. The cross-validated predictive model incorporating all baseline features achieved a mean area-under-the-ROC curve (AUC) of 0.85 ± 0.02. EZ integrity metrics and HRF count ranked as the most influential predictors of progression. CONCLUSIONS: Quantitative OCT biomarkers, particularly EZ integrity measures and HRF count, are strongly associated with 2-year progression to advanced atrophic AMD in iAMD. These findings support the role of quantitative EZ integrity metrics and hyperreflective foci features for early disease risk stratification and may inform the design of prevention-focused or early-intervention clinical trials aimed at delaying the onset of atrophic changes.