Abstract
Purpose :
The large variability in response to anti-VEGF treatment of neovascular age-related macular degeneration nAMD adds complexity to the evaluation of novel therapies. To reduce uncertainty or potential bias in clinical trial design, and optimize patient recruitment and allocation, we have developed an empirical drug-disease progression model (EDDM) of ranibizumab treatment that can serve as a virtual treatment arm. The development, validation and potential applications of this novel approach are presented.
Methods :
An EDDM describing the time-course of best corrected visual acuity (BCVA) was established for 2421 patients with nAMD treated with intravitreal (IVT) ranibizumab or sham. The model, using pre-identified baseline characteristics, was used to predict the BCVA time-course in the phase 2 studies of faricimab (STAIRWAY) and the predictive performance of the model was evaluated using a receiver operating characteristic (ROC) curve.
Results :
The pre-identified baseline characteristics were BCVA, age, leakage size, central retinal leakage thickness, presence of cysts and CNV type. The model predictions for ranibizumab were in excellent agreement with the observed BCVA data. The estimated AUC of the ROC curve was 0.85, showing the high predictive performance of the model. The model was also used to reduce bias by comparing the observed results of the faricimab Q16w treatment with the simulated response to ranibizumab Q4w in a virtual population having the same baseline characteristics.
Conclusions :
We have developed and validated a virtual ranibizumab treatment arm that accurately predicts the trajectory of BCVA time course using 6 baseline characteristics. The potential to eliminate or reduce the size of an active comparator arm and adjust for bias using the virtual treatment approach warrants further considerations
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.