Development of Fibrosis and Atrophy in nAMD in regard to fluid volume (IRF, SRF, PED) and fluctuation localization (1, 3, 6mm)

Abstract

Purpose : Age-related macular degeneration (AMD) in its advanced form as neovascular AMD (nAMD) can lead to subretinal fibrosis and scarring. Additionally, macular atrophy (MA) is regularly seen in patients with long-term nAMD. The clinical parameters influencing the development of fibrosis and atrophy are still not clarified. Apart from the influence of fluid distribution on the functional outcomes, its effect on the development of fibrotic and atrophic alterations could give important insight for better understanding the pathogenesis of end-stages. Hence, we investigated the influence of fluid distribution fluctuation localization on fibrosis and atrophy using AI-supported fluid quantification and localization.

Methods : 569 treatment naïve nAMD patients were included in the analysis. Sub- and intra-retinal fluid (SRF and IRF) and pigment epithelium detachment (PED) were quantified on optical coherence tomography (OCT) volumetric scans using the Vienna Fluid Monitor, Version 2 (RetInSight, Vienna, Austria) at all timepoints until appearance of atrophy or fibrosis evaluated by the Vienna Reading Center (VRC). A disease activity-dependent treat&extend-based anti VEGF treatment regimen was used.

Results : A logistic regression model for fibrosis/atrophy was created. ROC-curves for fibrosis showed a specifity of 0.793 and a sensitivity of 0,583, AUC: 0.700 (0,532/0.833, AUC: 0.735 after stepwise model selection) and for atrophy 0,644/0,723 AUC: 0.718 and 0,659/0.734, AUC: 0.716, accordingly. Importantly, fluid distribution and dynamic showed a significant influence on the development of fibrosis/atrophy. If the standard deviation of IRF increased by 1-unit (100nl) (p=0.006) odds to develop fibrosis increased by 4.4. Hence, patients with little IRF at baseline that increased rapidly during the following visits were likely to develop fibrosis. However, atrophy did not seem to be influenced by IRF significantly, whereas PED and SRF showed a significant influence on the development of MA.

Conclusions : Automated assessment of fluid type and volume using artificial intelligence allows an objective and precise assessment of the activity of IRF, SRF and PED and therefore its influence on the development of fibrosis or atrophy. AI-based fluid monitoring offers guidance to improve anatomical and visual outcomes in the long-term management of nAMD in clinical routine.

This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.