Fang Hong-Bin, Tian Guo-Liang, Li Wei, Tan Ming
Division of Biostatistics, University of Maryland Greenebaum Cancer Center, Baltimore, Maryland 21201, USA.
J Biopharm Stat. 2009 Jul;19(4):625-40. doi: 10.1080/10543400902964019.
The study of drug combinations has become important in drug development due to its potential for efficacy at lower, less toxic doses and the need to move new therapies rapidly into clinical trials. The goal is to identify which combinations are additive, synergistic, or antagonistic. Although there exists statistical framework for finding doses and sample sizes needed to detect departure from additivity, e.g., the power maximized F-test, different classes of drugs of different does-response shapes require different derivation for calculating sample size and finding doses. Motivated by two anticancer combination studies that we are involved with, this article proposes dose-finding and sample size method for detecting departures from additivity of two drugs with linear and log-linear single dose-response curves. The first study involves combination of two drugs, where one single drug dose-response curve is linear and the other is log-linear. The second study involves combinations of drugs whose single drug dose-response curves are linear. The experiment had been planned with the common fixed ratio design before we were consulted, but the resulting data missed the synergistic combinations. However, the experiment based on the proposed design was able to identify the synergistic combinations as anticipated. Thus we shall summarize the analysis of the data collected according to the proposed design and discuss why the commonly used fixed ratio method failed and the implications of the proposed method for other combination studies.
由于药物组合在较低毒性剂量下具有疗效潜力，且需要迅速将新疗法推进到临床试验中，因此药物组合的研究在药物开发中变得至关重要。目标是确定哪些组合是相加、协同或拮抗的。尽管存在用于确定检测偏离相加性所需剂量和样本量的统计框架，例如功效最大化F检验，但不同剂量反应形状的不同类药物在计算样本量和确定剂量时需要不同的推导方法。受我们参与的两项抗癌联合研究的启发，本文提出了用于检测具有线性和对数线性单剂量反应曲线的两种药物偏离相加性的剂量确定和样本量方法。第一项研究涉及两种药物的组合，其中一种单一药物剂量反应曲线是线性的，另一种是对数线性的。第二项研究涉及单一药物剂量反应曲线为线性的药物组合。在咨询我们之前，该实验采用了常见的固定比例设计进行规划，但所得数据遗漏了协同组合。然而，基于所提出设计的实验能够如预期那样识别出协同组合。因此，我们将总结根据所提出设计收集的数据的分析，并讨论常用固定比例方法为何失败以及所提出方法对其他联合研究的影响。
