Abstract
Purpose :
To use statistical modeling to identify baseline characteristics that best correlate with the ability to extend faricimab injection intervals in DME patients during the first year (Y1) of the phase 3 YOSEMITE & RHINE studies and to understand how these characteristics may impact treatment intervals.
Methods :
YOSEMITE(NCT03622580) & RHINE(NCT03622593) were 2 identical, double-masked, randomised, comparator controlled, Phase 3 trials of intravitreal faricimab in patients with center-involving DME. Patients were randomized 1:1:1 to faricimab 6.0mg Q8W after 6 initial Q4W doses; faricimab 6.0mg per personalized treatment interval (PTI) after 4 initial Q4W doses; or aflibercept 2.0mg Q8W after 5 initial Q4W doses. PTI patients followed a treat-and-extend–based regimen (up to Q16W interval dosing) using prespecified BCVA and CST criteria to adjust dosing. This post-hoc analysis evaluated clinically relevant ocular baseline characteristics associated with Y1 treatment intervals. Patients were categorised into 3 subgroups based on their treatment interval at Y1: Q16, Q12 or Q4/Q8. The characteristics were inputted into a R package ‘rpart’ (R version 4.0.3; package version 4.1.15) to create a classification tree model that chose the most impactful variables and their associated thresholds that differentiated the 3 subgroups.
Results :
The ocular baseline characteristics inputted into the model were: BCVA, CST, presence of macular ischemic non-perfusion, previous anti-VEGF treatment, presence of SRF and IRF. Baseline CST was the most impactful with faricimab treatment interval extensions in Y1, with CST subgroups of <407μm (n=213), 407-576μm (n=269), >=576μm (n=108) identified for the 3 treatment intervals (Q16, Q12, Q4/8). After CST, the model chose BCVA as the next most impactful baseline characteristic (Figure 1).
Conclusions :
Among the selected baseline characteristics from YOSEMITE & RHINE trial patients, baseline CST and BCVA had the strongest impact on the ability to extend treatment intervals at Y1. Other factors, including presence of IRF, SRF and macular ischemic non-perfusion were not as strongly associated. This analysis may provide further insight in predicting treatment intervals when initiating treatment with faricimab for DME.
This abstract was presented at the 2023 ARVO Annual Meeting, held in New Orleans, LA, April 23-27, 2023.