[Restricted mean duration of response: a novel endpoint for efficacy evaluation in oncology clinical trials]

In oncology clinical trials, the duration of response (DOR) is limited by the inclusion of only responders, leading to selection bias and violating the principle of randomization. To address this limitation, this study introduces the restricted mean duration of response (RMDOR) as a novel composite efficacy endpoint. The definition, parameter estimation, and hypothesis testing procedures of RMDOR were systematically described, and an example analysis was conducted using data from a phase Ⅲ non-small cell lung cancer clinical trial with the package "PBIR". RMDOR integrates information from the objective response rate, time to response, and DOR at the intention-to-treat population level, effectively compensating for the limitations of single endpoints. The example analysis demonstrated that RMDOR provides a stable and rational quantification of treatment differences between therapeutic regimens. RMDOR offers a new methodological tool for oncology trials that preserves the advantages of randomization under the intention-to-treat principle while comprehensively reflecting treatment benefits. Its combined use with traditional endpoints is recommended in future clinical studies.