Comparison of FAF and SD-OCT Geographic Atrophy Lesion Area and Growth Rate Measurements

Abstract

Purpose : While fundus autofluorescence (FAF) is the primary imaging modality to measure geographic atrophy (GA) growth in clinical trials, spectral domain optical coherence tomography (SD-OCT) imaging is routinely used in clinical care. This work aims to establish the relationship between FAF and SD-OCT GA-related measurements, to utilize SD-OCT for GA progression assessment.

Methods : In an IRB approved study, 12 patients (24 eyes) with both FAF and SD-OCT acquired within 60 days as part of routine clinical care at Cleveland Clinic were included. 76 image-pairs in total were analyzed with an average longitudinal follow up= 505±330 days. Trained, in-house experts manually segmented GA lesions in FAF following hypofluorescent areas demarcating GA. In SD-OCT, areas with complete retinal pigment epithelium and outer retinal atrophy (cRORA) were detected at the B-scan level using an automated algorithm with manual verification of the output. B-scan level cRORA segmentations were projected to create a 2D enface map depicting the total area of GA comparable to the segmented area in FAF. Total GA lesion area measurements were compared between modalities at each longitudinal time point. GA growth rates measured using longitudinal time points that spanned more than 180 days from baseline were also compared between modalities.

Results : GA area (mean±SD) was 6.1±4.3mm² (range: 0.0-15.3mm²) in FAF, 6.6±4.6mm² (range: 0.0-13.8mm²) in SD-OCT, and correlated with R²=0.98 and mean area difference= 0.6±0.5mm². Annual growth rate (mean±SD) was 0.8±0.6mm²/year in FAF, 1.1±0.9mm²/year in SD-OCT, and correlated with R²=0.60 and mean growth rate difference= 0.5±0.5mm²/year. Longitudinal changes in field of view (e.g., limited OCT field of view not capturing the entire lesion), areas with incomplete outer retinal atrophy appearing as hypofluorescent areas in FAF, inconsistent peripapillary atrophy annotation, and annotation variability proximal to fovea in FAF contributed to some measurement differences between the modalities.

Conclusions : FAF and SD-OCT total GA area measurements are highly correlated. While positive correlation was observed for growth rate measurements, smaller yearly growth in GA lesions resulted in decreased correlation coefficient. Accounting for factors contributing to differences across modalities could help minimize measurement variability when comparing incremental growth rates at short time intervals.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.