Abstract
Purpose :
Microperimetry is a useful tool to assess associations between anatomical and functional data in geographic atrophy (GA). Due to the loss of retinal tissue in GA, microperimetry test points that fall within the lesion provide limited information as these areas represent scotomas (blind spots) with no sensitivity to light. This study compares various approaches to estimate the rate of retinal sensitivity loss and evaluates the precision of the estimates when using an optimized, patient-customized microperimetry grid in participants with GA.
Methods :
This prospective pilot study was conducted in participants with GA (n=11) at two sites in the United States. There were four on-study visits (Baseline, Day 7, Month 3 and 6) with each visit consisting of a 30-minute period of dark adaption followed by scotopic, then mesopic microperimetry assessments using a S-MAIA (CenterVue, Padova, Italy). The customized grids consisted of approximately 60 points across three zones: 9 points on the participants' preferred retinal locus (PRL), and 50 equidistant points in 2 concentric rings at 200 μm (Zone 1) and 450 μm (Zone 2) from the lesion border. Retinal sensitivity rate was estimated using mixed-effects models which were computed by using 1) the mean sensitivity per eye or 2) the sensitivities of each individual point. To assess the differences in precision of these different methodologies, relative errors were calculated and compared. Statistical analyses were conducted using R Software (v4.3.1) under RStudio.
Results :
Sensitivity Rates using overall mean sensitivity vs. all individual points, and per zone vs. all points per zone are as follows: (Uploaded Table)
Conclusions :
Relative errors using all points were generally lower than those of the overall or per zone mean sensitivities, indicating that more precise retinal sensitivity rates were observed when all individual points were used in the model. The variability of the estimated sensitivity rate was higher under scotopic conditions. This study suggests that estimating sensitivity rates using an all point, mixed effect modelling approach in an optimized, patient-customized microperimetry grid, provides more precision in the estimates than using mean sensitivity values.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.