Delayed treatment effect predicting (DTEP) model for guiding immuno-oncology trial designs

Abstract

Listen

Translate article

BackgroundThe presence of delayed treatment effects (DTE) is common in immuno-oncology trials. However, conventional trial designs often overlook the potential presence of DTE, which can result in an underestimation of the required sample size and loss of statistical power. Conversely, when there is actually no apparent delay in treatment effects, alternative trial designs for addressing DTE may lead to an overestimation of sample size and unnecessary prolongation of the trial duration. To mitigate this challenge, we propose the use of a DTE predicting (DTEP) model to better guide immuno-oncology trial designs. MethodsThe DTEP model was developed and validated using data from 147 published randomized immuno-oncology trials. The eligible trials were divided into a training set (approximately 75% of the trials) and a test set (approximately 25%). We employed linear discriminant analysis (LDA) to develop the DTEP model for predicting the DTE status using baseline characteristics available at the trial design stage. The receiver operating characteristic (ROC) curve was utilized to assess the ability of the model to distinguish between trials with and without DTE. We further re-conducted the JUPITER-02 trial in a simulation setting, employing three design approaches to assess the potential benefits of utilizing the DTEP model. ResultsBaseline characteristics available during the trial design stage, including cancer type, line of treatment, and experimental and control arm regimens were incorporated, and high accuracy in predicting the DTE status in both the training set (area under the operating characteristic curve (AUC), 0.79; 95% confidence interval (CI), 0.71–0.88) and test set (AUC, 0.78; 95% CI, 0.66–0.90) was achieved. Notably, the model successfully predicted the DTE status in two randomized trials among the test sets that were conducted by our team (ESCORT-1st (absence of DTE) and JUPITER-02 (presence of DTE)). In silico re-conduct of the JUPITER-02 trial further showed that the statistical power would be markedly improved when trial designs were guided by the DTEP model. ConclusionsThe DTEP model can significantly enhance the precision and effectiveness of immuno-oncology trial designs, thereby facilitating the discovery of effective immunotherapeutics in a more streamlined and expedited manner.