The need to test anticancer drugs in multiple indications has been addressed by basket trials, which are Phase I or II clinical trials involving multiple tumor subtypes and a single master protocol. Basket trials typically involve few patients per type, making it challenging to rigorously compare responses across types. We describe the use of permutation testing to test for differences among subgroups using empirical null distributions and the Benjamini-Hochberg procedure to control for false discovery. We apply the approach retrospectively to tumor-volume changes and progression-free survival in published basket trials for neratinib, larotrectinib, pembrolizumab, and imatinib and uncover examples of therapeutic benefit missed by conventional binomial testing. For example, we identify an overlooked opportunity for use of neratinib in lung cancers carrying ERBB2 Exon 20 mutations. Permutation testing can be used to design basket trials but is more conservatively introduced alongside established approaches to enrollment such as Simon’s two-stage design.

篮子试验解决了在多种适应症中测试抗癌药物的需求，这些试验是涉及多种肿瘤亚型和单一主协议的 I 或 II 期临床试验。篮式试验通常每种类型的患者很少，因此很难严格比较不同类型的反应。我们描述了使用置换测试来测试使用经验零分布和 Benjamini-Hochberg 程序来控制错误发现的子组之间的差异。我们回顾性地将该方法应用于已发表的来那替尼、拉罗替尼、派姆单抗和伊马替尼的篮子试验中的肿瘤体积变化和无进展生存期，并发现了传统二项式测试遗漏的治疗益处的例子。例如，ERBB2外显子 20 突变。置换测试可用于设计篮子试验，但更保守地与已建立的注册方法一起引入，例如 Simon 的两阶段设计。