Fluid/function correlation using AI-based quantification versus central subfield thickness in treatment-naïve and pre-treated patients with neovascular AMD in a real-world setting.

Abstract

Purpose: To investigate the association between best-corrected visual acuity (BCVA) and quantitative macular fluid volumes, compared to central subfield thickness (CST) in treatment-naïve and previously treated patients with active neovascular age-related macular degeneration (nAMD).

Methods and analysis: Baseline data were collected from 290 eyes of 290 participants consecutively enrolled in a prospective, randomized phase III clinical trial. Intraretinal fluid (IRF), subretinal fluid (SRF) and pigment epithelial detachment (PED) volumes were quantified and localized using an MDR-certified AI algorithm (Fluid Monitor, RetInSight). Fluid volumes and CST were included in linear regression models for comparison.

Results: Significantly greater IRF volumes within each macular region were observed in treatment-naïve patients, whereas larger PED volumes contributed to higher CST values in pretreated patients. In both subgroups, the largest proportion of BCVA variance could be explained by measuring IRF and SRF volumes within the entire 6-mm area (adjusted R² = 0.140 and 0.225, respectively). In pre-treated eyes, CST explained only half as much BCVA variance as the 6-mm fluid model, and the model's fit was even poorer when compared to the CST model in the treatment-naïve subgroup (adjusted R² = 0.078 vs. 0.198).

Conclusion: The examination of IRF and SRF volumes significantly impacts BCVA in nAMD. The weaker association of CST highlights its limitations as a parameter of disease activity. These findings emphasize the necessity of distinct fluid volume quantification as a relevant surrogate for visual function loss or benefit in nAMD, with particular emphasis on treatment duration and fluid in regions beyond the central 1-mm.