图示法在多重性问题中的应用及KeyNote-048案例分析

Application of graphical approach in multiple testing and KeyNote-048 case analysis

控制总I类错误是监管机构审查药物临床试验的基本标准,而实际临床研究中普遍存在多重性问题导致I类错误的膨胀。图示法是一种使用图形方式表现的多重性检验程序,其可以直观展示不同假设检验间的相关性和Ⅰ类错误的重新分配策略,并可以融合多种多重性调整方法以应对复杂临床试验。KeyNote-048是一项帕博利珠单抗治疗晚期头颈癌患者的随机对照研究,其包含多个临床终点、多组间比较、亚组分析和期中分析共4种多重性问题。通过图示法,KeyNote-048展示了14个假设检验的相关性和预设Ⅰ类错误,并根据α重新分配策略设定不同情况下假设检验的检验水平。最后本文讨论了实际应用中图示法应注意的问题。

Control type I error is a basic criterion for regulatory agency to review the clinical trial.However,multiplicity in the real trials often leads the inflation of type I error.Graphical approach is a multiplicity test procedure in graphical presentation.The graphical approach can intuitively display the correlation between multiple testing and the redistribution of type I error,and it can also merge a variety of multiplicity adjustment methods in complex clinical trials.KeyNote-048 is a randomized controlled study of pembrolizumab in the treatment of advanced head and neck cancer.It includes 4 types of multiple issues including multiple clinical endpoints,multiple arms,subgroup analysis,and interim analysis.Through graphical methods,KeyNote-048 shows the correlation of 14 hypothesis tests and pre-defined type I errors and sets the test level of hypothesis tests in different situations according to theαredistribution strategy.Finally,this paper discusses the problems that should be paid attention to in the actual application of the graphical approach.