Retinal Melanin Measured From Routine Fundus Photos is Associated With Age-Related Macular Degeneration Risk

Abstract

Purpose : Melanin in the retinal pigmented epithelium is hypothesized to be protective in age-related macular degeneration (AMD). Prior investigations have been limited by small cohorts and require specialized equipment. We used the retinal pigment score (RPS), a technique that measures the degree of retinal pigment from color fundus photographs, in a cohort of self-reported white patients to investigate if retinal pigment is associated with AMD risk (Figure 1).

Methods : Color fundus photographs from the UK Biobank were analyzed in this study. AMD diagnosis and laterality of disease were determined by patient questionnaire. The RPS is a single continuous variable generated from the retinal background using machine learning to extract the chromaticity (Figure 1). A higher score corresponds to more retinal pigment. Univariate tests were performed for RPS, age (per decade), Townsend Deprivation Index (TDI), and sex between patients with AMD and without. Continuous predictors were evaluated with independent t-tests and categorical predictors with chi-squared tests. From the univariate tests, the statistically significant predictors of age and RPS were incorporated as linear predictors in a generalized additive model. This model also included an additional term for random effects at the patient level.

Results : A total of 44,336 patients from the UK Biobank were included, of which there were 449 eyes with AMD from 387 patients. Univariate analysis found RPS and age to be significantly associated with AMD. In multivariate analysis, after controlling for age, higher RPS was associated with a lower risk of AMD diagnosis (OR: 0.84 [0.74 - 0.94]) (Table 1).

Conclusions : More retinal pigment quantified by the RPS was associated with a lower risk of AMD among white patients. Our results suggest that retinal pigment, thus melanin, may play a protective role in AMD even among patients of the same self-reported ethnicity.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.

 

Fig 1. a) schematic of the pipeline to extract the retinal pigment score (RPS), b) example fundus photos used in the cohort with varying RPS. A continuous scale of color associated with RPS is shown below the fundus photos.

 

Table 1. a) significance of univariate tests for predictors and AMD diagnosis. b) results from a multivariate generalized additive model with linear predictors of RPS and age with additional patient level random effects.