Test-retest variability of static perimetry and microperimetry results from the RUSH2A study

Abstract

Purpose : Static perimetry (SP) and microperimetry (MP) are key outcome measures for the Rate of Progression in USH2A-related Retinal Degeneration (RUSH2A) study. Each test is performed 3 times at baseline and results are averaged in analysis. However, each test requires a high level of concentration and substantial time (≈ 9 to 64 minutes for SP and ≈ 9 to 14 minutes for MP). Testing may require more than one day due to patient fatigue. We evaluated the variability of mean retinal sensitivity of SP and MP and total hill of vision (V_TOT, decibel-steradians) of SP exams to help future study design and possibly reduce patients’ burden in future clinical research.

Methods : Participants completed SP and MP (optional at sites with MAIA) testing 3 times during the baseline period (within 14 days). Examination results were scored by reading centers. Reproducibility of SP and MP scores was evaluated using intra-class coefficients (ICCs) and Bland-Altman analysis for participants with all three gradable tests.

Results : Among 105 participants targeted for repeat testing, 102 (97%) completed 3 SP tests. V_TOT of SP was available for 101 (96%) participants and was similar across the 3 tests with mean (SD) equal to 32.7 (24.1), 31.2 (23.4), and 31.7 (23.9) dB steradians. The ICC among the sets of three SP tests was 0.96. The 95% limits of agreement between V_TOT from tests 1 and 2 were ± 9.2 dB steradians. Mean sensitivity of SP across the 3 tests was 11.7 (5.9), 11.4 (5.8), and 11.5 (5.9) dB. The ICC among the 3 sets of SP test results was 0.96. The 95% limits of agreement between SP tests 1 and 2 were ± 2.3dB. 84 (90%) of targeted participants had 3 gradable MP tests. Mean sensitivity of MP was similar across the 3 tests with mean (SD) equal to 5.7 (4.9), 5.6 (5.0), and 5.5 (5.0) dB. The ICC among the 3 sets of MP test results was 0.95. The 95% limits of agreement between SP tests 1 and 2 were ± 3.2 dB. With ICCs near 1, there is little gain in precision of the baseline measurement by averaging results from more than 1 test.

Conclusions : The baseline SP and MP measurements in RUSH2A suggesting minimal learning effect. Repeated testing for SP and MP may not be necessary. Future clinical trials may be able to eliminate repeat testing and substantially reduce testing burden.

This is a 2020 ARVO Annual Meeting abstract.

 

Figure 1. Bland-Altman Plot of Difference vs Mean for 1^st and 2^nd V_TOT of SP tests