Clinical determinants of retinal age gap estimated from fundus photographs in glaucoma patients

Retinal age gap (RAG), defined as the difference between artificial intelligence-predicted retinal age and chronological age derived from fundus photographs, has been proposed as a potential biomarker of biological aging; however, the influence of ocular factors on RAG remains unclear. This retrospective observational study investigated clinical determinants of RAG in 283 eyes of adults who underwent 45-degree color fundus photography at the University of Yamanashi Hospital between February 2010 and April 2025. Retinal age was estimated using a deep learning-based algorithm developed from the Japan Ocular Imaging Registry. Associations between RAG and ocular and clinical variables were evaluated using linear mixed-effects models to account for within-subject correlations. The mean chronological age was 65.5 ± 11.7 years, the mean predicted retinal age was 59.9 ± 17.0 years, and the mean RAG was - 5.7 ± 19.1 years. Lens status was strongly associated with RAG, with phakic eyes showing greater RAG values than pseudophakic eyes. Worse visual acuity and more hyperopic refractive error were also associated with greater RAG values. Overall glaucoma severity was not significantly associated with RAG, and mean deviation was not associated with RAG when analyzed as a continuous variable. In phakic eyes, glaucoma eyes showed greater RAG values than control eyes. These findings indicate that fundus-based retinal age estimation is influenced by ocular structural and optical factors, particularly lens status, and should be interpreted with consideration of ocular conditions.