Abstract
PURPOSE: To replicate previously reported genetic associations for the retinal aging clock-quantified as the retinal age gap-in an East Asian population and to assess cross-ancestry consistency of effect directions and sizes. DESIGN: Genetic association study with candidate-single nucleotide polymorphism (SNP) replication embedded in a population-based cohort. PARTICIPANTS: Individuals who participated in both first and second visits of the Nagahama Study with gradable fundus photographs and genome-wide genotype data. METHODS: A deep learning model fine-tuned on an independent subset (i.e., data from participants who attended only one of the 2 visits of the Nagahama Study) was used to estimate retinal age from color fundus photographs. We tested 13 previously reported lead variants; 7 were evaluable in our dataset (directly or as available leads). Linear regression models adjusted for standard covariates estimated SNP effects on the retinal age gap at the first visit. Statistical significance was set as Bonferroni-corrected P = 0.0071 (0.05/7). Secondary measures included directional concordance and effect-size correlation with discovery estimates. MAIN OUTCOME MEASURES: Association of previously reported retinal aging clock SNPs with the retinal age gap; number of loci meeting replication criteria; directional concordance; cross-study effect-size correlation. RESULTS: Age-prediction accuracy in the genetic cohort showed mean absolute error = 3.00 (first visit) and 3.42 years (second visit). Of 13 variants to be replicated, 7 were testable. The lead variant rs17713879 (SH3YL1) replicated after multiple-testing correction (β = -0.41 years per effect allele, standard error = 0.094, P = 9.8 × 10(-6)). After harmonizing effect directions, all 7 of 7 testable variants showed directional concordance with discovery findings (binomial P = 0.0078) and a strong cross-study effect-size correlation (r = 0.93). CONCLUSIONS: In an East Asian population, we replicate the SH3YL1 association with the retinal age gap and observe consistent direction and magnitude across all testable discovery variants, supporting a partially shared genetic architecture of retinal aging across ancestries. Larger transethnic meta-analyses and functional follow-up are warranted to refine causal biology. FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.