Advancing Platform Trials in Early Oncology by Using MATS and EXNEX in Randomized Controlled Trials.

The randomized controlled trial (RCT) is widely regarded as the gold standard in clinical development due to its ability to support robust statistical inference. In early-phase oncology development, however, trialists and statisticians may face obstacles such as small patient populations, which hinder the inclusion of a randomized control arm due to recruitment challenge and timeline urgency in drug development for these small populations. The issue is further exacerbated when drug development targets multiple similar but distinct patient populations, and meanwhile concurrent patients have prognoses that differ from those in previous trials within each patient population, upon which key design assumptions are based. Moreover, FDA's Project Optimus puts more emphasis to earlier dose optimization and the optimal dose may differ across treatment regimens and patient populations. To address this, we developed a platform trial design to support proof-of-concept (POC) decision-making in a randomized controlled trial (RCT) setting and dose optimization in alignment with FDA's Project Optimus. The method builds on the Multi-Arm Two-Stage (MATS) design (Jiang in Contemp Clin Trials 132:107278, 2023) by incorporating randomized control arms and inherits its feature of stopping futile drugs at the end of Stage 1, prior to further dose optimization in Stage 2. All are conducted within an RCT setting to enhance the robustness of decision making. To further utilize the patient resource in these small populations, we enable control borrowing across cohorts using the exchangeability-nonexchangeability (EXNEX) framework (Neuenschwander et al in Pharm Stat 15 (2):123-134, 2016), thereby improving statistical efficiency-an especially important consideration that the standard of care in these small populations may deliver similar efficacy outcome. We demonstrated how our method can promote efficient early oncology development in these small and similar patient populations through a case study.