Statistical Methods for Multi-Drug Combinations

描述（由申请方提供）：本提案的目的是开发用于多药联合研究设计和分析的统计方法和算法。药物组合是治疗癌症、艾滋病毒和高血压等复杂疾病的标志。由于药物效应是剂量依赖性的，需要检查单个药物的多剂量，从而产生快速增加的组合数量和具有挑战性的高维统计问题。多药联合研究缺乏适当的设计和分析方法，导致许多治疗机会的错失。我们的初步研究已经确定了一个分析公式，用于确定两种抗癌药物的相对效力，这与该领域中恒定相对效力的常见假设相矛盾。此外，我们还开发了一种用于组合研究设计和分析的最大功效方法，以便最大限度地提高检测与相加偏离的统计功效，并且可以用中等样本量估计剂量-反应。目前多药联合研究多采用次优设计，如两药成对评价。我们提出了一种新的两阶段的程序，通过利用计算机模拟模型建立在单一药物的实验数据结合可用的网络或途径信息，并随后与有效的实验设计选定的多药物组合和数据的统计分析的初始选择。本研究将现代统计学方法、数学、药理学和计算机技术相结合，提出：（1）利用单药剂量反应数据和信号通路/网络信息，建立药物及其相互作用的最佳选择的统计模型和算法;（2）利用目标1中的计算机模拟结果，建立表征剂量反应的实验设计和统计分析;（3）开发用于多药物组合的相互作用（协同作用）分析的统计方法，（4）在已经由NCI和其他资助机构资助的研究中，在癌细胞系中测试这些方法;以及（5）丰富在目标1-3中开发的计算机程序。项目完成后，预计该方法将能够为更大的翻译研究社区服务，并且还将对统计研究产生影响， 高维数据

DESCRIPTION (provided by applicant): The goal of this proposal is to develop statistical methods and algorithms for the design and analysis of multi-drug combination studies. Drug combinations are the hallmark of therapies for complex diseases such as cancer, HIV and hypertension. Because drug-effect is dose-dependent, multiple doses of an individual drug need to be examined, yielding rapidly rising number of combinations and a challenging high dimensional statistical problem. The lack of proper design and analysis methods for multi-drug combination studies have resulted in many missed therapeutic opportunities. Our preliminary studies have identified an analytic formula for determining the relative potency of two anticancer drugs, which contradicts the common assumption of constant relative potency in the field. Furthermore, we have developed a maximal power approach for the design and analysis of combination studies so that the statistical power to detect departures from additively is maximized, the dose-response can be estimated with moderate sample size. Currently multi-drug combination studies have to resort to suboptimal design such as pair wise evaluation of two-drugs. We propose a novel two-stage procedure starting with an initial selection by utilizing an in silico model built upon experimental data of single drugs in conjunction with available network or pathway information and followed with efficient experimental designs on selected multi-drug combinations and statistical analysis of the data. Integrating modern statistical methods, mathematics, pharmacology and computing, we propose to (1) develop the statistical models and algorithms for the optimal selection of drugs and their interactions utilizing single-drug dose response data and signaling pathway/network information; (2) develop experimental designs and statistical analysis for characterizing dose-responses using the in silico results in Aim 1; (3) develop statistical methods for the interaction (synergy) analysis of multi-drug combinations, (4) test the methods in cancer cell lines in studies already funded by NCI, and other funding agencies; and (5) enrich computer programs developed in Aims 1-3. Upon completion of the project, it is anticipated that the method will be able to serve a much larger translational research community and it will also have bearing to statistical research dealing with high dimensional data.