Artificial intelligence-quantified functional and structural outcomes after 12 months of faricimab in neovascular age-related macular degeneration: A real-world longitudinal cohort at Fundación Oftalmológica Los Andes

Abstract Introduction: Faricimab is the first bispecific antibody approved for the treatment of neovascular age-related macular degeneration (nAMD), combining dual inhibition of VEGF-A and angiopoietin-2. Although it has demonstrated efficacy in the TENAYA/LUCERNE trials, evidence in Latin American populations and in real-life settings is limited. Methods: Retrospective observational study in 28 eyes of 24 patients treated with faricimab for nAMD in a tertiary center in Chile. Visual and structural changes at month 12 were evaluated by automated quantitative analysis with artificial intelligence (RetinAI Discovery™), including biomarkers such as subretinal fluid (SRF), intraretinal fluid (IRF), pigment epithelial detachment, retinal thickness, retinal pigment epithelium (RPE), and EZ. Linear mixed models, logistic regression, and negative binomial regression were used. Results: Best-corrected visual acuity (BCVA) improved significantly at 12 months (+0.08 logarithm of the minimum angle of resolution; P = 0.02), with no differences between naïve patients and switchers. SRF and IRF volumes were reduced, mainly in the first 3 months. No significant correlations were found between biomarker change and visual improvement. BCVA was the only significant predictor of visual response (odds ratio = 680; P = 0.042). The mean number of injections was 6.4 ± 2.9, with no significant differences between groups. Conclusions: In this Latin American cohort, faricimab showed sustained functional and anatomical effectiveness in real life. However, visual outcomes were more modest than in pivotal trials. The structural response was greater in eyes with higher baseline fluid load, without necessarily translating into greater visual gain. Prospective studies are required to validate these findings.