Impact of baseline imbalances on the efficacy of pegcetacoplan for the treatment of geographic atrophy (GA): A post hoc analysis of OAKS, DERBY, and FILLY

Abstract

Purpose : To examine the potential impact of baseline imbalances on the 12-month results of OAKS, DERBY, and FILLY.

Methods : This is a post hoc analysis of the OAKS (NCT03525613), DERBY (NCT03525600) and FILLY (NCT02503332) studies. OAKS and DERBY are 24-month, phase 3, randomized studies comparing the efficacy and safety of monthly or every-other-month (EOM) intravitreal pegcetacoplan with sham in patients with GA secondary to age-related macular degeneration. FILLY was a phase 2 study with a similar design. The primary endpoint was change in absolute GA lesion size from baseline to Month 12 in OAKS and DERBY and change in square root transformed lesion size in FILLY. OAKS and FILLY met the primary endpoint while DERBY narrowly missed. In the original, pre-specified analysis, fellow eye CNV (OAKS and DERBY only), time, and baseline lesion size were covariates. To investigate differences in the treatment effect estimate, a systematic covariate analysis was conducted. Eight variables, including 7 baseline characteristics related to GA progression (study eye focality, lesion location, lesion size, pseudodrusen, low-luminance deficit (LLD), intermediate/large drusen number, GA laterality) and region were selected to investigate imbalances between treatment arms within each study. Variables with imbalances meeting a p<0.2 threshold in any of the studies were included in the common adjusted model and efficacy analyses were repeated.

Results : Four variables with imbalances met the p<0.2 threshold for inclusion in the adjusted model: study eye lesion location (OAKS), study eye intermediate/large drusen number (DERBY, FILLY), study eye LLD (FILLY), and study eye lesion focality (DERBY). Before adjustment, absolute change in GA lesion growth reduction was 22% and 16% (OAKS), 12% and 11% (DERBY), and 31% and 21% (FILLY) for the monthly and EOM arms, respectively, at Month 12. Accounting for imbalanced variables, GA lesion growth reduction was 26% and 18% (OAKS), 16% and 15% (DERBY), and 25% and 18% (FILLY) for the monthly and EOM arms, respectively.

Conclusions : After adjusting for imbalances, results are more consistent across the studies, but this analysis does not fully explain the disparities nor replace the primary analysis. Future studies could consider incorporating additional variables as covariates or pre-specifying a plan for covariate adjustment.

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