Simulating the altered photoreceptor cell topography of progressive retinal dystrophies using OCT images and public data

Abstract

Purpose : The topography of rod and cone photoreceptor cells shapes the patterns of outer retinal loss observed in progressive cone-rod and rod-cone dystrophies. To better understand these diseases, we modeled the OCTs observed in progressive photoreceptor degenerations by simulating differential survival of rod and cone cells.

Methods : We assumed that the thickness of the outer retina is proportional to the topographic composition of living rod and cone photoreceptor cells. Consequently, the death of rods or cones will reduce the thickness of the outer retina proportionally to the baseline composition of these cells. By definition, 100% of rods and 100% of cones survive in eyes unaffected by disease. Using this framework, we imaged the maculas of an unaffected control using a Heidelberg Spectralis OCT, segmented the retinal layers using the Iowa Reference Algorithms, and located the fovea and the optic nerve within these volumes. We registered these volumes to published densities of photoreceptor cells (Curcio et al. J Comp Neurol 1990), interpolated the cell densities at every A-scan, and computed the expected proportion of rod and cone photoreceptor cells at every A-scan. To simulate the death of photoreceptor cells, we reduced the thicknesses of the outer retinal layers as a function of the contribution of rods and cones, reconstructed the segmentation, and resampled the voxels within each layer.

Results : When more cones are lost than rods, our model reproduces the basic pattern of a cone-rod dystrophy, such as autosomal recessive Stargardt disease. When more rods are lost than cones, our model reproduces the basic pattern of a rod-cone dystrophy, such as USH2A-associated retinitis pigmentosa. Users can explore the interactive model at https://observablehq.com/@barefootbiology/simulating-rod-and-cone-loss.

Conclusions : Clinicians can infer the approximate proportions of surviving rods and cones by comparing the simulated output to B-scans of actual patients. Computational biologists can use the model to communicate with clinicians by translating mathematical models of cell survival into the visual idiom of the clinic.

This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.