Longitudinal fundus imaging and its genome-wide association analysis provide evidence for a human retinal aging clock

Abstract

Biological age, distinct from an individual's chronological age, has been studied extensively through predictive aging clocks. However, these clocks have limited accuracy in short time-scales. Here we trained deep learning models on fundus images from the EyePACS dataset to predict individuals' chronological age. Our retinal aging clocking, 'eyeAge', predicted chronological age more accurately than other aging clocks (mean absolute error of 2.86 and 3.30 years on quality-filtered data from EyePACS and UK Biobank, respectively). Additionally, eyeAge was independent of blood marker-based measures of biological age, maintaining an all-cause mortality hazard ratio of 1.026 even when adjusted for phenotypic age. The individual-specific nature of eyeAge was reinforced via multiple GWAS hits in the UK Biobank cohort. The top GWAS locus was further validated via knockdown of the fly homolog, Alk, which slowed age-related decline in vision in flies. This study demonstrates the potential utility of a retinal aging clock for studying aging and age-related diseases and quantitatively measuring aging on very short time-scales, opening avenues for quick and actionable evaluation of gero-protective therapeutics.

Ageing Biomarkers Derived From Retinal Imaging: A Scoping Review

“…The ‘EyeAge’ model was trained on 217 289 images from 100 692 individuals in the EyePACS dataset, validated on 54 292 images from 25 238 individuals the same dataset and tested in both the UK Biobank and EyePACS dataset. The model achieved a MAE of 3.30, and a Pearson R of 0.87 for the UK Biobank test dataset, with corresponding figures of 2.86 and 0.95 for the EyePACS dataset (Ahadi et al, 2023). The ‘convolutional network-based model’ was trained on 98 400 photos from patients diagnosed with diabetes who were enrolled in the Retisalud programme of the Canary Islands Health Service.…”

“…Three models were designed to predict age from retinal images, namely: ‘ Retinal Age’ (Zhu, Shi, et al, 2023), ‘ EyeAge’ (Ahadi et al, 2023) and a ‘ convolutional network-based model’ (Abreu-Gonzalez et al, 2023). A fourth biological ageing model, ‘ RetiAGE’ (Nusinovici et al, 2022), was developed to estimate the probability of an individual being older than 65 years from a retinal image.…”

“…In the UK Biobank cohort, adjusted EyeAge achieved an all-cause mortality hazard ratio of 1.03, while RAG (referred to as EyeAge Acceleration) was associated with a higher risk of chronic obstructive pulmonary disease [p=0.0048], myocardial infarction [p=0.049], elevated systolic blood pressure [p=1.025e-7] and fluid intelligence scores [p=5.34e-27]. Additionally, a genome wide association study (GWAS) was performed as part of the study and found that candidate genes identified for EyeAge acceleration are associated with eye function and age-related disorders (Ahadi et al, 2023).…”

“…Three models were designed to predict age from retinal images, namely: 'Retinal Age' (Zhu, Shi, et al, 2023), 'EyeAge' (Ahadi et al, 2023) (Nusinovici et al, 2022).…”

Nurturing longevity through natural compounds: Where do we stand, and where do we go?

“…During this period, several other clocks emerged as significant contributors of assessing biological age. These include the mortality-predictive Grim-Age clock (A. T. Lu et al, 2019); the Ionome clock (W. that tracks the involvement of minerals and trace elements in aging; the microbiome clock (Galkin et al, 2020), which observes gut microflora aging; and the retinal clock (Ahadi et al, 2023) that utilizes fundus imaging. In regard to therapies, the first anti-aging clinical study, namely, the "Metformin in Longevity Study" (NCT02432287) trial, emerged in 2015 as part of the Targeting/Taming Aging with Metformin project.…”