靶向临床试验不同设计方法的统计学性能比较

目的: 本文拟对三种常用的靶向临床试验设计方法进行统计学性能模拟研究,为进行靶向临床试验提供设计选择依据。方法: 在全随机设计、富集设计、策略设计三种常用的靶向临床试验设计方法下,设定不同的参数条件,借助模拟方法探讨其在同等参数条件下样本量与把握度之间的关系,并依此对比各设计方法的统计学性能。结果: 在随机分配两组样本含量相同的条件下,富集设计总能获得最大的把握度。除了在靶向治疗药物对无靶点病人产生负影响的情况下,全随机设计都可以获得比策略设计更大的把握度。当试验药物除了对靶点阳性病人有效,如果对靶点阴性病人也有一定疗效时,在相同把握度情况下,富集设计为筛选合格病人所需的样本量会大于全随机设计。结论: 如果靶向治疗药物对无靶点病人也有疗效,或者目前尚没有很好的靶点诊断方法并且病人中靶点阳性率较高,我们建议首选可进行亚组分析的全随机设计。

Statistical performance comparison of different design methods of targeted clinical trials

AIM: In recent years, the targeted clinical trials increase rapidly with various design methods. The choice of these design methods, however, has not been widely recognized. In this paper, we conducted a simulation study to assess the statistical performance of three commonly used targeted clinical trial design methods to provide evidence for design choice.METHODS: For three commonly used targeted clinical trial design methods (full randomized design, enrichment design and strategy design); we investigated the relationship between sample size and statistical power under different parameters to compare the statistical performance of the design methods by simulation method.RESULTS: If two groups were assigned with the same sample size, the enrichment design always obtained the largest power. Except the condition that targeting therapeutic drugs had adverse effect on target negative patients, the full randomized design can obtain larger power than the strategy design. When the targeted drugs also had a certain beneficial effect for the target negative patients, the enrichment design required larger sample size for screening than the full randomized design to maintain the same power.CONCLUSION: If the targeted drugs have a certain effect for non-target patients, or there is no accurate diagnosis of target and the target positive rate in patients is high, we recommend the full randomized design with subgroup analysis.