Reliability of Quantitative Autofluorescence Imaging in a Multicenter Study Involving Patients with Stargardt Disease

Abstract

Purpose : The aim of this study is to establish the reliability of quantitative autofluorescence (qAF) in a multicenter setting in order to validate qAF as an endpoint for interventional clinical trials in Stargardt disease (STGD1).

Methods : A total of 104 STGD1 patients (208 eyes) underwent qAF imaging at two visits (screening and baseline, mean interval 20 ± 13 days), with two qAF recordings per visit, as part of the Stargardt Remofuscin Treatment Trial (EudraCT No. 2018-001496-20) and according to standardized image acquisition protocol including operator training and certification. qAF₈ values were determined by calculating the mean qAF values within the 8 middle segments of the Delori pattern. Eyes were independently graded by the concerned study site and the central reading center. Intra- and inter-visit as well as interobserver (study site versus reading center) reproducibility of qAF₈ levels were obtained using intraclass correlation (ICC) and coefficient of repeatability (CR) based on the method of Bland-Altman.

Results : At the first visit, there were significant differences in qAF₈ values between the six study sites (p<0.001). Overall, the mean qAF₈ value at baseline, 463.55±107.81 units, was not significantly different from the value at screening in the same patients, 467.63±108.53 units (p=0.753). qAF₈ reproducibility was ± 26.1% for intra-visit, ± 40.3% for inter-visit, and ± 21.6% for the interobserver reproducibility. Intra-visit and interobserver reliability of qAF₈ was good to excellent for all sites (ICC of 0.84-0.99). However, we observed a significant difference of 8.80 ± 56.18 qAF₈ units between both movies acquired on the same day and 17.49 ± 48.60 between observers (p=0.007 and p<0.001). Variability between visits was higher with ICC of 0.43-0.83, but there was no significant difference in the qAF₈ units measured at screening and baseline (p=0.210).

Conclusions : Real-life test-retest variability of qAF is higher in the multicenter setting than previously reported in single center studies. Possible explanations include the demanding imaging protocol, operator variability and differences in patient characteristics. qAF is a successful tool to diagnose STGD1. However, when using qAF as clinical trial endpoint, increased variability must be considered, and larger sample sizes may be required to demonstrate treatment effects.

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