基于生物标志物的贝叶斯随机临床试验设计用于识别目标人群
本文讨论一种基于生物标志物的贝叶斯(BM-Bay)随机临床试验设计,该设计纳入连续尺度测量并预设切点或分级尺度的预测性生物标志物以定义多个患者亚群。通过设置期中分析及决策标准,基于时间至事件结局的疗效评估,实现正确高效识别目标患者人群。大量模拟研究评估了所提方法在不同临床场景下的操作特性,并应用于一项随机II期免疫肿瘤学临床试验设计。
要点
- BM-Bay设计利用连续或分级生物标志物定义多个亚群,通过期中分析决策标准识别敏感亚群并排除不敏感亚群。
- 决策标准基于时间至事件结局的疗效评估,模拟研究评估正确识别概率和预期患者数等操作特性。
- 该方法已应用于一项随机II期免疫肿瘤学临床试验的设计示例。
核心结论
- 提出的BM-Bay设计通过期中分析决策标准,可基于时间至事件结局疗效评估,同时纳入敏感亚群和排除不敏感亚群,以正确高效识别目标患者人群。
研究设计与样本
- 本文为方法学文章,讨论基于生物标志物的贝叶斯(BM-Bay)随机临床试验设计,并进行大量模拟研究评估其操作特性,另应用于一项随机II期免疫肿瘤学临床试验设计。
主要发现
- 模拟研究在广泛临床场景下评估了所提方法的操作特性,包括正确识别目标亚群的概率和预期患者数。
局限性
- 原文未报告局限性
以上解读依据论文摘要(全文未开放获取或许可不允许翻译转载)。
原文
The challenges and potential benefits of incorporating biomarkers into clinical trial designs have been increasingly discussed, in particular to develop new agents for immune-oncology or targeted cancer therapies. To more accurately identify a sensitive subpopulation of patients, in many cases, a larger sample size-and consequently higher development costs and a longer study period-might be required. This article discusses a biomarker-based Bayesian (BM-Bay) randomized clinical trial design that incorporates a predictive biomarker measured on a continuous scale with pre-determined cutoff points or a graded scale to define multiple patient subpopulations. We consider designing interim analyses with suitable decision criteria to achieve correct and efficient identification of a target patient population for developing a new treatment. The proposed decision criteria allow not only the take-in of sensitive subpopulations but also the ruling-out of insensitive ones on the basis of the efficacy evaluation of a time-to-event outcome. Extensive simulation studies are conducted to evaluate the operating characteristics of the proposed method, including the probability of correct identification of the desired subpopulation and the expected number of patients, under a wide range of clinical scenarios. For illustration purposes, we apply the proposed method to design a randomized phase II immune-oncology clinical trial.
引用
- Sugitani Y, Morita S, Nakakura A, et al. Biomarker-based Bayesian randomized clinical trial design for identifying a target population. Stat Med. 2023. PMID: 37078122. doi:10.1002/sim.9749
供执业医师工作参考,不替代临床判断;数值与分级以指南原文为准。