Department of Biostatistics, The University of Texas MD Anderson Cancer Center, 1400 Pressler Street, 4-5062 Pickens Academic Tower, Houston, TX 77030-1402, USA.
Nat Rev Clin Oncol. 2011 Nov 8;9(4):199-207. doi: 10.1038/nrclinonc.2011.165.
Modern oncology drug development faces challenges very different from those of the past and it must adapt accordingly. The size and expense of phase III clinical trials continue to increase, but the success rate remains unacceptably low. Adaptive trial designs can make development more informative, addressing whether a drug is safe and effective while showing how it should be delivered and to whom. An adaptive design is one in which the accumulating data are used to modify the trial's course. Adaptive designs are ideal for addressing many questions at once. For example, a single trial might identify the appropriate patient population, dose and regimen, and therapeutic combinations, and then switch seamlessly into a phase III confirmatory trial. Adaptive designs rely on information, including from patients who have not achieved the trial's primary end point. Longitudinal models of biomarkers (including tumor burden assessed via imaging) enable predictions of primary end points. Taking a Bayesian perspective facilitates building an efficient and accurate trial, including using longitudinal information. A wholly new paradigm for drug development exemplifying personalized medicine is evinced by an adaptive trial called I-SPY2, in which drugs from many companies are evaluated in the same trial--a phase II screening process.
现代肿瘤药物开发面临的挑战与过去大不相同，必须相应地进行调整。三期临床试验的规模和费用继续增加，但成功率仍然低得令人无法接受。适应性试验设计可以使开发更具信息量，解决药物是否安全有效，并展示如何进行药物的投放以及针对哪些人群。适应性设计是指累积数据被用于修改试验进程的设计。适应性设计非常适合同时解决许多问题。例如，一项试验可以确定合适的患者人群、剂量和方案以及治疗组合，然后无缝切换到三期确证试验。适应性设计依赖于信息，包括未达到试验主要终点的患者的信息。生物标志物的纵向模型（包括通过影像学评估肿瘤负担）可用于预测主要终点。采用贝叶斯观点有助于构建高效准确的试验，包括使用纵向信息。一个名为 I-SPY2 的适应性试验体现了药物开发的全新范例，即个性化医学，该试验评估了许多公司的药物——这是一个二期筛选过程。
