A Generalized Framework of Optimal Two-Stage Designs for Exploratory Basket Trials

The primary goal of an exploratory oncology clinical trial is to identify an effective drug for further development. To expedite the drug development process and increase the chance of finding active tumor indications where the treatment works, multiple tumor cohorts are often investigated simultaneously in a basket trial. In this article, we propose a generalized framework of an optimal basket trial design in the exploratory setting where tumor indications can be homogeneous (objective response rates of all indications are the same under both the null and alternative hypotheses) or heterogeneous (objective response rates are not the same under at least one of the null or alternative hypotheses). The proposed design prunes the inactive tumor indication in Stage 1 and pools the remaining tumor indications at the end of Stage 2 to evaluate the overall effectiveness of whether the treatment works in at least one tumor indication. The design parameters are optimized to minimize the expected sample size while explicitly controlling the global Type I and Type II error rates. In addition, we consider reallocating the planned Stage 2 sample size of pruned indications to achieve a higher power when the total planned sample size is fixed. Simulation studies are conducted to show the favorable operating characteristics of the proposed design under certain scenarios.

探索性肿瘤学临床试验的核心目标，是筛选出可推进后续开发的有效药物。为加快药物开发进程、提升找到治疗有效的活跃肿瘤适应证的概率，篮子试验（basket trial）通常会同步开展多个肿瘤队列的研究。本文针对探索性场景下的篮子试验提出一种通用优化设计框架，其中肿瘤适应证可分为同质性与异质性两类：同质性指所有适应证的客观缓解率（objective response rate）在原假设与备择假设下均保持一致；异质性则指至少在原假设或备择假设之一的场景下，各适应证的客观缓解率存在差异。所提设计在Ⅰ期阶段剔除无活性的肿瘤适应证，并在Ⅱ期结束时合并剩余肿瘤队列，以此评估该疗法是否在至少一种肿瘤适应证中展现整体疗效。该设计通过优化参数，在严格控制全局Ⅰ型错误率与Ⅱ型错误率的前提下，实现期望样本量的最小化。此外，在总计划样本量固定的情况下，本文还考虑将已剔除适应证的计划Ⅱ期样本量进行重新分配，以获得更高的检验效能。本研究通过模拟实验验证了所提设计在特定场景下具备优异的操作特性。