Related Experiment Video
Updated: Sep 18, 2025
Author Spotlight: Overcoming Anti-VEGF Resistance Through Advanced Vascular Morphology Assessment in Choroidal Neovascularization
Published on: August 11, 2023
Real-world faricimab switch in France: artificial intelligence-based detection of changes in exudative signs in
Vincent Gualino1,2,3, Charles Sohier1,2, Maxime Sibert4
1Ophthalmology Department, Pierre-Paul Riquet Hospital, Toulouse University Hospital, Toulouse, France.
Purpose:
Some patients with neovascular age-related macular degeneration (nAMD) have persistent signs of exudation under treatment with intravitreal injections of anti-vascular endothelial growth factor (VEGF) agents. We examined the real-world anatomical responses among patients with suboptimal response and switched to faricimab using artificial intelligence (AI)-based retinal fluid quantification.
Methods:
A retrospective, multicentric, cohort study of patients in France with exudative signs and switched to faricimab without a new loading phase, maintaining the same prior injection interval. The RetInSight Fluid Monitor AI software quantified subretinal (SRF) and intraretinal (IRF) fluid on spectral domain optical coherence tomography. The primary outcome was change in SRF and IRF volumes in the central 1 and 6 mm retinal areas after one and two injections of faricimab.
Results:
74 patients (74 eyes) were included (mean age: 81.5±8.4 years, 49% male). Significant reductions were observed in mean 1 mm IRF (-2.9±18.3 nl; p=0.002), mean 6 mm IRF (-17.7±71.1 nl; p<0.001) and mean 6 mm SRF (-24.8±156.3 nl; p<0.001) volumes after one injection. The proportion of dry eyes (<5 nl for SRF and IRF in the 1 and 6 mm areas) increased from 0% at baseline to 32.4% after one injection and 48.4% after two injections. Lower baseline SRF volumes were predictive of dry response after one injection (adjOR 0.965; p=0.028) and lower baseline IRF volumes were predictive of dry response after two injections (adjOR 0.373; p=0.051).
Conclusion:
Nearly half of patients achieved a dry response after two injections. AI-assisted fluid quantification provided objective monitoring, identifying lower baseline SRF and IRF as predictive factors for good response. Limited patient inclusion means longer-term and larger prospective studies are now required using automated retinal fluid quantification to further refine the baseline characteristics of good switch responders to better adapt switch protocols.
