A Novel MRI-Based Approach to Peripheral Refraction and Prediction of Myopia Progression

Abstract

Purpose: Optical solutions that create peripheral myopic defocus in the presence of a clear central image have shown to be effective as myopia treatment. This study investigates whether peripheral refraction measured via MRI and ray tracing can predict myopia progression in children. Methods: A total of 1635 children from the Generation R Study, a population-based birth cohort in Rotterdam, the Netherlands, underwent T2 weighted MRI scanning at age 9 years. At both ages 9 and 14 years, ocular biometry, and cycloplegic autorefraction were assessed. Retinal curvature radii were computed from MRI segmentations using semi-automated, customized image processing algorithms. Individual peripheral refraction profiles were modelled through ray tracing. Horizontal and vertical peripheral refraction was analysed at 50-degrees eccentricity. Relative peripheral refraction (RPR) was calculated by subtracting peripheral refraction from central cycloplegic refraction. Yearly myopia progression was calculated and stratified into quantiles (∆AL), and the effect of RPR on the quantile outcomes was examined using ordinal regression analyses. Predictive performance of RPR on development of myopia was evaluated using ROC-analysis (fast vs slow progressors) and a logistic regression (incident myopia). Results: At age 9 years, 207/1635 (13%) children had developed myopia. Myopic children had a significantly more hyperopic RPR compared to emmetropic children at all horizontal eccentricities (–1.8 ± 1.8D vs. 0.2 ± 2.1D) and vertical eccentricities (–1.0 ± 1.9D vs. 0.8 ± 2.2D). Higher vertical (OR: 1.08, CI: 1.02-1.14) and horizontal RPR (OR: 1.16, CI: 1.10-1.22) was associated with faster AL progression. Each diopter increase in vertical RPR (OR: 1.10, CI: 1.01-1.20) and horizontal RPR (OR: 1.23, CI: 1.13-1.35) was associated with an increased risk of incident myopia. ROC analysis indicated that RPR had a maximum predictive AUC of 0.77 for identifying fast progressors. Furthermore, MRI data revealed significant interindividual variations in retinal curvature (SD 1 mm), which resulted in clinically relevant peripheral refractive differences exceeding 8D among children with similar axial length and central SE, suggesting that standard defocus strategies may require individualization. Conclusions: Using this novel approach to calculate peripheral refraction, we provide evidence based on eye shape that peripheral hyperopic refractive error is more pronounced in myopic children and is strongly associated with myopia progression. The significant anatomical variability in retinal radii underscores the need for personalized treatment strategies, which may enhance the efficacy of optical interventions for myopia management.