Project 1: Innovative Biomarker-Integrated Clinical Trial Design and Analysis

项目1：创新的生物标志物整合临床试验设计和分析 项目摘要 随着生物医学技术的进步，生物标志物在疾病进展中发挥着越来越重要的作用， 癌症患者的诊断和治疗选择。纳入生物标志物的创新临床试验设计 信息可以提高研究效率，需要更少的患者，并降低成本，从而加速癌症 药物开发该研究项目的广泛，长期目标是开发创新的统计 解决生物标志物整合临床试验设计和分析问题的方法。有 该项目的四个具体目标。第一个目标是开发设计和分析方法，以确定最佳的 基于生物标志物的亚组或具有至事件时间终点的临床试验的最佳生物标志物临界点。 将为非随机和随机试验开发统计方法。第二个目标是投资-- 门控适应于随机化前纵向生物标志物的个性化治疗策略。统计 基于选择模型的方法将用于联合建模纵向生物标志物和存活结果。 用于测试纵向生物标志物和治疗之间相互作用的样本量和功效计算 将被开发。第三个目标是开发生物标志物富集临床应用的设计和分析方法， 使用替代物依赖性和标记物依赖性取样的试验。当标志物阳性的流行率超过- 然而，由于剂量低，传统的设计需要较大的样本量，这限制了许多研究的可行性。 将研究一类新的基于生物标记的成本效益设计，并研究与实施有关的问题。 的设计，效率增益的条件，和新的估计的渐近性质将被调查。 用于量化利用生物标志物指导治疗所产生的益处的新的统计测量将 也被调查。第四个目标提出了两阶段临床试验，以发现个性化的预测 生物标志物。将探索两阶段设计，并确定该两阶段设计的样本量和功效公式 将开发，特别是对于新的第二阶段，其中辅助信息从第一阶段将是 incorporated.为了达到上述目的，我们将（1）研究所提出的方法的理论特性 基于现代经验过程理论和其他先进的统计理论;（2）检查绩效 透过广泛的模拟研究，了解建议的方法在实际情况下的应用情况;及（3）发展方便使用者的 用于所开发方法以及样本量和功效计算工具的软件，并予以传播 免费向公众开放。

Project 1: Innovative Biomarker-Integrated Clinical Trial Design and Analysis PROJECT SUMMARY With advances in biomedical technology, biomarkers are playing an increasingly important role in disease prog- nosis and treatment selection for cancer patients. Innovative clinical trial designs that incorporate biomarker information can improve study efficiency, require fewer patients, and reduce costs, thereby accelerating cancer drug development. The broad, long-term objective of this research project is to develop innovative statistical methodology to address issues in the design and analysis of biomarker-integrated clinical trials. There are four specific aims in this project. The first aim develops design and analysis methods to identify the optimal biomarker-based subgroups or the optimal biomarker cut-point for clinical trials with time-to-event endpoints. Statistical methods will be developed for both non-randomized and randomized trials. The second aim investi- gates a personalized treatment strategy that is adaptive to pre-randomization longitudinal biomarkers. Statistical methods based on selection models will be used to model longitudinal biomarkers and survival outcomes jointly. Sample size and power calculations for testing the interactions between longitudinal biomarkers and treatments will be developed. The third aim is to develop design and analysis methods for biomarker-enrichment clinical trials using surrogate-dependent and marker-dependent sampling. When the prevalence of marker positive pa- tients is low, the traditional design requires a large sample size, which can limit the feasibility of many studies. A new class of biomarker-based cost-effective designs will be studied and issues related to the implementation of the design, conditions for efficiency gain, and asymptotic properties of the new estimates will be investigated. Novel statistical measures for quantifying the benefit resulting from utilizing biomarkers to direct treatments will also be investigated. The fourth aim proposes two-phase clinical trials for discovering personalized predictive biomarkers. A two-phase design will be explored and sample size and power formulae for this two-phase design will be developed, especially for the novel second phase where auxiliary information from the first phase will be incorporated. For all of the aims, we will (1) investigate the theoretical properties of the proposed methodology based on modern empirical process theory and other advanced statistical theory; (2) examine the performance of the proposed methods in practical settings through extensive simulation studies; and (3) develop user-friendly software for the developed methods and for the sample size and power calculation tools and disseminate them freely to the general public.