基于稳健估计模型的多区域临床试验（MRCT）的方法学研究

With the Globalization and multi-regionalization of Clinical Trials, multi-regional clinical trials(MRCT) have been an important part of global development of new drugs. However, multi-region commonly increases the difference and also adds complexity to the analysis. It has been an urgent and important realistic demand of great theoretical value to accurately and robustly estimate the regional effect of MRCT. But it is still an unsolved world problem, which hinders the development of new drug research and medical science. In this study, considering the design and analysis of MRCT, a new and robust semi-parametric model for binary, ranked (ordinal category) and survival endpoint will be established, based on the previous successful research for continuous end-point of MRCT . The model is divided into two parts, parameter parts which affecting the covariate of regional efficacy and nonparametric parts of other unknown factors. And the model estimation is based on contour likelihood method. Regional efficacy is tested by contour likelihood ratio statistics. Large sample theory is derived from mathematics. Through simulation and examples, the method deduced by theory is verified, and R software package is developed to facilitate application. The methods proposed in this project are original. The realization of this research goal not only has high theoretical value, but also has extremely important application value, especially when China becomes a member of ICH.

新药研发的临床试验日益全球化和多区域化，多区域临床试验（MRCT）在全球研发创新药物中发挥着重要作用。但多区域也增加了差异性及分析的复杂性。如何精确而稳健的估计MRCT的区域效应具有急迫和重要的现实需求及理论价值，但其目前仍是一个尚未解决的世界难题，因此阻碍了新药研发乃至医学科学的发展。建立在连续型终点MRCT的前期成功研究之上， 本研究针对MRCT的设计及分析，分别建立二分类型、等级（有序分类）及生存终点的新的、稳健的半参数模型。该模型包含影响区域疗效协变量的参数部分和其它未知因子的非参数部分。模型估计采用轮廓似然法，区域疗效应用轮廓似然比统计量检验，大样本理论由数学推导得出，通过模拟和实例来验证经理论推导出的方法，并开发R软件包以方便应用。本项目所提出方法均具有原创性，本研究目标的实现不仅具有较高的理论价值，还具有极为重要的应用价值，特别在中国成为ICH成员国之际。

本项目按计划完成了预期的研究目标，主要研究内容包括：① 建立了二分类终点半参数logistic回归，并将其用于MRCT研究，围绕MRCT设计提出了评估区域效应和区域治疗效应的稳健方法，并围绕该方法提出了相应的样本量估计算法；② 建立了等级（有序分类）终点半参数稳健估计模型，并将其用于MRCT研究，围绕MRCT设计提出了评估区域效应和区域治疗效应的稳健方法，并围绕该方法提出了相应的样本量估计算法；③ 建立了生存终点半参数稳健估计模型，并将其用于MRCT研究；④ 实现了考虑治愈的生存终点的半参数稳健估计模型。本项目成果主要体现在基础研究方面取得了扎实的进展，在主要研究成果已经发表的同时研究成果也已经成功地用于临床试验的分析，所以项目取得成果的总体情况符合项目预期。本项目所提出的方法均具有原创性，不仅具有较高的理论价值，还具重要应用价值。