Adaptive optics scanning laser ophthalmoscopy (AOSLO) measures in USH2A-related retinal degeneration

Abstract

Purpose : Although USH2A-related retinal degeneration is among the most common forms of Usher syndrome type 2 (USH2) and autosomal recessive retinitis pigmentosa (ARRP), prior studies have not described cone structure with high-resolution images. Adaptive Optics Scanning Laser Ophthalmoscopy (AOSLO) is a non-invasive method of visualizing cone photoreceptors with high resolution. We describe baseline cone spacing and explore inter-grader and inter-test variability using data collected in a multicenter, international natural history study of USH2A-related retinal degeneration.

Methods : Two baseline AOSLO montages of the cone mosaic were obtained from each of 14 participants (6 USH2, 8 ARRP; 4 female; 11 white) with USH2A-related retinal degeneration. Two graders masked to disease category measured cone spacing, cone density and an image quality score; poor quality images were confirmed not gradable by a masked clinician. Cone spacing measures were converted to Z-scores based on 37 normal subjects. Inter-test and inter-grader variability were evaluated using intraclass coefficient (ICC) and Bland-Altman plots. Association between cone spacing and clinical characteristics were assessed by linear mixed effects regression.

Results : Median age of participants at enrollment was 38 years (range: 20-59). Median disease duration was 14 years (interquartile range, IQR: 11-19) in USH2 and 6 years (IQR: 4-14) in ARRP participants. Cone spacing ranged from 0.78-2.44 arc minutes. Cone spacing was associated with eccentricity (P < 0.001), but no significant difference was observed by clinical diagnosis or disease duration. Bland-Altman plots showed high concordance between baseline 1 and baseline 2 (mean differences near 0, ICC = 0.71). Grader 2 reported slightly lower values than grader 1 (ICC=0.7). Quality score was higher in baseline 1 than baseline 2 (P = 0.02). Greater inter-grader variation was found in regions with poor image quality (P < 0.001).

Conclusions : The baseline AOSLO data revealed similar cone spacing between USH2 and ARRP participants. Image quality was significantly lower for the second baseline montage, suggesting one baseline image is sufficient. Adding a qualitative image quality score can be a useful approach to handle inter-grader inconsistencies while minimizing the loss of data.

This is a 2021 ARVO Annual Meeting abstract.