OBJECTIVES

Regulatory agencies such as the FDA and EMA have issued guidelines on how to evaluate non-inferiority. However, there is no clear guidance on how to select the non-inferiority margins (NIFm's) in oncology clinical trials. This study is to identify previously used NIFms in oncology clinical trials for key endpoints.

METHODS

A systematic Medline literature review of NIFms in oncology clinical trials was conducted. Non-inferiority randomized clinical trials published in the English between Jan 2000 and Dec 2017 were included. Only studies reporting margins for the response rate (RR), progression-free survival (PFS), overall survival (OS), and safety endpoints were included. The following data items were extracted: indication, treatment setting, sample size, primary endpoint, treatment effect measure, and defined margin. Study selection and data extraction were performed by two independent investigators.

RESULTS

Out of 635 screened search hits, 99 reports were included in this analysis. For both RR and safety, the rate difference was used as the measure of treatment effect. For RR, among 18 reports (median sample size: 293; range: 98 to 828), the median and mean of NIFms were 15% and 13%, respectively. For safety endpoint, we identified two studies with a NIFm of 10% and 15%. For both PFS and OS, the hazard ratio was used as the measure of treatment effect. For PFS, among 29 reports (median sample size: 402; range: 85 to 2,098), the median and mean of NIFms were 1.300 and 1.314, respectively. For OS, among 50 reports (median sample size: 586; range:22 to 1,725), the median and mean of NIFms were 1.250 and 1.272, respectively. On average, retrieved studies with a total sample size of 1000 or 500 had larger NIFms.

CONCLUSIONS

There is no consensus on the appropriate NIFms for key endpoints in oncology trials. More guidance from regulatory agencies is needed.