All Times EDT
Keywords: Bayesian optimal interval; Immunotherapy; Optimal biological dose; Toxicity efficacy probability interval; Utility-based interval; Cell therapy
Conventional dose finding designs in oncology drug development target on the identification of the maximum tolerated dose (MTD), with the assumption that the MTD has the most potential of clinical activity among those identified tolerable dose levels. However, immuno-oncology (I-O), especially in cell therapy area, may lack dose-efficacy monotonicity, posing significant challenges in the statistical designs for dose finding trials. A desirable design should empower the trial to identify the right dose level with tolerable toxicity and acceptable efficacy. Such dose is called as optimal biological dose (OBD), which is more appropriate to be considered as the primary objective of the first-in-human trial in I-O than MTD. We propose two model-assisted designs in this setting: the toxicity and efficacy probability interval-2 (TEPI-2) design and the utility-based interval (UBI) design that incorporate the toxicity and efficacy outcomes simultaneously and identify a dose that possesses high probability of acceptable efficacy with manageable toxicity. The proposed designs can generate decision tables before trial starts to facilitate practical and easy-to-implement applications. We present simulation results and case study to illustrate the performance and applicability of the proposed designs. Our proposed novel designs demonstrate superior performance in accuracy, efficiency and patient safety. Additionally, they can reduce the number of patients and shorten clinical development landscape significantly and enable to directly convert first-in-human study to pivotal phase I/II trial. We also illustrate the advantages of proposed methods to a CAR T-cell therapy phase I clinical trial for multiple myeloma and summarize our recommendations in the discussion section.