贝叶斯最优多阶段Ⅱa期自适应临床试验方法-BOP2设计

Bayesian optimal multi-phase adaptive method for Ⅱa clinical trials-BOP2 design

Ⅱ期临床试验的主要目的是评估新疗法的初步疗效,以早期无效终止和中期go/no-go决策为主要特征。此类设计的主要终点多集中在二元的疗效终点,例如客观缓解率。但随着新型分子靶向药物和免疫疗法的出现,主要终点变得更加复杂,可以是有序的或共同多(co-primary)终点的。为适应新形势下复杂终点的研究,本文介绍一种灵活的贝叶斯最优Ⅱ期(BOP2)设计方法,这是一种单臂Ⅱa设计方法,能够灵活处理简单(如二元)和复杂(如有序、嵌套和共同多)终点。并使用Dirichlet-多项式模型统一处理上述不同类型的终点。在每个中期阶段,通过评估事件的后验概率来做出go/no-go决策;并在原假设下,以决策概率为优化函数,可最大程度地提高检验效能或减少患者人数。模拟研究表明,较其他文献中的贝叶斯Ⅱ期设计相比,BOP2设计可精确控制Ⅰ类错误率,保证患者风险最小化,并具有更高的检验效能及更低的错误终止试验的风险,弥补了此前贝叶斯设计和频率设计之间的差距。此外,BOP2设计可以在试验设计阶段便枚举出终止边界值。这些特性使得BOP2设计可供众多用户和监管机构使用,且易于实现。本文同时在附件中给出了一个原文引理的证明,方便从业者更好理解和掌握此设计方法。

The primary objective of phase Ⅱ clinical trials is to evaluate a preliminary efficacy of a new treatment.It is characterized by early ineffective termination and interim go/no-go decision.The main endpoints of this type of design mostly focus on binary tumor response,such as objective response rate.However,with the advent of new molecularly targeted drugs and immunotherapies,the primary endpoint becomes more complex,often involving ordinal,nested,and co-primary endpoints.In order to adapt to the study of complex endpoints in the new situation,a flexible Bayesian Optimal Phase Ⅱ(BOP2)design is introduced,which is a Phase Ⅱa design and can simultaneously handle simple(e.g.,binary)and complicated(e.g.,ordinal,nested,and co-primary)endpoints under a unified framework.The Dirichlet-multinomial model is employed to accommodate different types of endpoints.At each interim stage,a go/no-go decision is made by evaluating a set of posterior probability of the event of interest.Under the null hypothesis,the posterior probability is optimized to maximize power or minimize the number of patients.Simulation studies show that compared to other existing Bayesian Phase Ⅱ designs,the BOP2 design explicitly controls the type Ⅰ error rate,ensures the minimization of the risk of patients,and has higher power and lowers the risk of incorrectly terminating the trial,which bridges the gap between the Bayesian designs and the frequentist designs.In addition,the stopping boundary of the BOP2 design can be enumerated during the experimental design stage.These features make the BOP2 design accessible to a wide range of users and regulatory agencies and particularly easy to implement in practice.A proof for the lemma in the original paper is given in the attachment such that practitioners can follow and grasp this design.