Abstract
Purpose :
To quantify photoreceptor loss during anti-VEGF therapy for neovascular age-related macular degeneration (nAMD) and correlate these findings with disease activity using precise artificial intelligence fluid quantifications.
Methods :
This study is a post-hoc analysis of data from the Fight Retinal Blindness! (FRB!) registry in Zürich. Spectral domain optical coherence tomography (SD-OCT) (Spectralis, Heidelberg Engineering, Germany) images of treatment-naïve patients with nAMD were processed at baseline and during follow-up of 3 years. A deep learning algorithm (Vienna Fluid Monitor, RetInSight, Austria) was used to automatically quantify intraretinal fluid (IRF), subretinal fluid (SRF) and pigment epithelial detachment (PED) volumes. Spatiotemporal correlation of fluid volumes with photoreceptor integrity was performed to identify early signs of atrophy progression in nAMD. The effect of fluid volumes on change of photoreceptor thickness at different timepoints was calculated using Wilcoxon rank-sum tests with bootstrapped confidence intervals.
Results :
Two hundred and eleven eyes from 158 patients were included which developed atrophy. The mean ± SD photoreceptor loss area in the central 6 mm was 1.81 ± 2.68 µm2 at baseline, increased to 4.21 ± 4.45 µm2 in the first year and reached 6.21 ± 6.15 µm2 at month 36. The mean photoreceptor thickness in the central 6 mm at baseline was 26.9 ± 4.7 µm and decreased to 21.4 ± 5.8 µm at month 36. Higher fluid volume (top 25%) of IRF and PED in the central 1mm and 6mm of the macula were significantly associated with more advanced photoreceptor thinning compared to the low fluid volume group (low 75%) during follow-up (figure 1). However, SRF volumes showed no impact on photoreceptor thickness or loss.
Conclusions :
The identification of early signs of atrophy in clinical practice is an important step towards a precise and personalized care, minimizing the risk of undertreatment. Detection of photoreceptor thinning and early loss of integrity dependent on retinal fluid behaviour has to be evaluated in a prospective manner. Artificial intelligence is best suited to model individual disease progression utilizing clinical and subclinical biomarker quantification in the real world.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.