Abstract
Purpose :
Individuals receiving treatment for neovascular age-related macular degeneration (nAMD) can often develop the same pathology in the fellow eye. We describe the changes in volumes of OCT biomarkers for fellow eyes that do not convert to nAMD during treatment by segmenting OCT volume scans using an artificial intelligence (AI) system previously published by De Fauw et al. (2018).
Methods :
This study included data from the Moorfields Eye Hospital AMD database for all patients that began treatment for nAMD between June 2012 and June 2017. This included AI-derived segmentation data from Topcon 3D OCT-2000 scans. Only eyes that did not convert by 24 months were used in this analysis. The biomarkers analysed include neurosensory retina (NSR), hyperreflective foci (HRF), retinal pigment epithelium (RPE) and drusen. The median relative change in volume from baseline (month 0) to specified time points of 4, 12, and 24 months was calculated as a percentage.
Results :
At baseline our data included a total of 1347 non-converting fellow eyes. The NSR volume remained consistent over the follow-up period, reducing by 0.22% at 4 months and 1.45% by 24 months, relative to baseline. Similarly, RPE volumes reduced by 0.18%, 0.41% and 1.19% from baseline at 4, 12 and 24 months respectively. The volume of HRF changed more considerably, initially increasing by 10.10% and 15.35% at 4 and 12 months respectively, with an overall increase of 51.29% by 24 months. The drusen volume increased by 3.46%, 9.57% and 19.07% at 4, 12 and 24 months, respectively.
Conclusions :
The volumes of NSR and RPE tissues in non-converting fellow eyes remained stable over the 24 month follow-up period. In contrast, HRF volume increased considerably, particularly in the second year of follow-up. The drusen volume demonstrated a steady increase over the follow-up period. Longitudinal quantitative analysis of retinal tissue volumes in fellow eyes of patients with nAMD, enabled by AI segmentation, may provide insights into OCT biomarkers that could predict disease progression.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.