EORTC Statistics Department, Av E Mounier 83/11, 1200 Brussels, Belgium.
Eur J Cancer. 2012 Jun;48(9):1386-91. doi: 10.1016/j.ejca.2011.12.024. Epub 2012 Jan 24.
Given the high failure rates and the increased costs of Phase III trials in oncology and the recent explosion of targeted agents, researchers are looking for better design strategies to try and optimise the use of available patients and financial resources. In this context, adaptive designs are seen as promising tools. We reviewed the different possible adaptations in the design of a clinical trial on the basis of the FDA guidance and summarized these. The pro and cons of adaptive designs are highlighted with a focus on one of the more 'controversial' adaptive designs, the sample size reassessment based on interim-effect size as proposed by Mehta and Pocock. While group sequential designs are preferable to such adaptive designs, both are difficult to implement in the case of rapid accrual and long time to event. Adaptive designs may have some potential in less favourable situations. However, the increase in overall power should be carefully weighted as well as the risk of a large negative trial. Adaptive designs need good, sometimes extensive, logistics. Some adaptive designs (e.g. group sequential designs) proved to be very useful and are already a part of the standard repertoire of trial designs used at European Organisation for Research and Treatment of Cancer (EORTC). Adaptive designs need strong measures to prevent bias that could otherwise become uncontrollable, particularly if interim results are leaked. This includes a prospective planning of adaptations. Finally, these studies currently have the potential to induce a heavy workload and cost linked to their regulatory management.
鉴于肿瘤学三期临床试验的高失败率和成本增加，以及最近靶向药物的大量涌现，研究人员正在寻找更好的设计策略，试图优化现有患者和财务资源的利用。在这种情况下，适应性设计被视为有前途的工具。我们根据 FDA 的指导意见，回顾了临床试验设计中可能的不同适应性，并对其进行了总结。重点介绍了一种更具“争议性”的适应性设计，即基于中期效应大小的样本量重新评估，同时强调了适应性设计的优缺点。虽然群组序贯设计优于此类适应性设计，但在快速入组和较长时间发生事件的情况下，这两种设计都难以实施。适应性设计在不太有利的情况下可能具有一定的潜力。然而，应该仔细权衡总体效能的增加，以及大的阴性试验的风险。适应性设计需要良好的、有时甚至是广泛的后勤支持。一些适应性设计（例如群组序贯设计）已被证明非常有用，并且已经成为欧洲癌症研究与治疗组织（EORTC）使用的标准试验设计组合的一部分。适应性设计需要采取强有力的措施来防止偏差，否则偏差可能变得无法控制，特别是如果中期结果被泄露。这包括对适应性进行前瞻性规划。最后，这些研究目前有可能因其监管管理而导致繁重的工作量和成本。
