Structure/Function Correlations in USH2A- associated Usher Syndrome

Abstract

Purpose : Usher Syndrome is a leading cause of deaf-blindness, with variants in USH2A being the most prevalent cause. With growing interest in this condition, and potential therapeutics going to trial, further understanding of how structure and function change through the disease course is required. OCT and microperimetry may provide more sensitive markers of progression and need to be further investigated.

Methods : Patients with molecularly confirmed biallelic USH2A variants were recruited at Moorfields Eye Hospital. Subjects were monitored at two time points (12.8 +/- 2 months apart). Heidelberg Spectralis OCT images were acquired. Novel OCT segmentation was developed using UNET with EfficientNet Backbone, based on 2770 double graded images. Micropermetry was performed using MAIA Expert Exam. Custom analysis was developed in Matlab, to evaluate visual sensitivity volume across the macular and by ring (fixation, 2°, 6°, 10°). Comparison between visits used Wilcox Signed Rank test, between devices used Mann-Whitney.

Results : Thirty-five patients were recruited (44 eyes of 23 patients included). Macular sensitivity loss was not significant between visits (p=0.08). Analysis of the individual ring sensitivity was also not significant between exams. Analysis of EZ area showed 5.02% (range:0.02-0.12%) loss across timepoints. EZ thickness compared to MP volume were significantly correlated (fixation- p=0.03; 2°- p=0.04, 6°- p=0.01, 10°- p=0.003). ONL thickness compared to MP volume were significantly correlated 2°- p=0.03, 6°- p=0.01, 10°- p=0.01).

Conclusions : Novel OCT analysis showed better sensitivity for progression than manual measures, or functional analysis. Therefore, more robust analysis should be used, included deep learning trained segmentation models, for future analysis. There was a significant relationship in ONL and EZ thickness with MP sensitivities when compared over specific regions.

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