Impact of Japanese Subpopulation Data and Overall Survival in Pivotal Trials on Oncology Drug Approvals in Japan

Globalization of oncology drug development has accelerated the adoption of multiregional clinical trials (MRCTs) as pivotal evidence for regulatory approvals in major markets, including Japan. While MRCTs enhance efficiency and generalizability, they introduce challenges related to inter-ethnic variability and data extrapolation. Japan's regulatory framework, guided by pharmaceuticals medical devices agency (PMDA) principles and the "Guideline for Clinical Evaluation Methods of Anticancer Drugs," emphasizes patient efficacy, safety and relevance, historically requiring local data to supplement global findings. This review analyzed the approvals of oncology drugs in Japan over 5 years, focusing on the role of Japanese subgroup data and overall survival (OS) outcomes in pivotal trials. Among 165 approvals for oncology drugs, OS in pivotal trials was set as a sole primary endpoint in 9.1% (15/165) and as co-primary/dual primary endpoints in 10.3% (17/165); alternative endpoints such as progression-free survival were accepted when OS was impractical. In 7.9% (13/165) of all approvals, Japanese subpopulation results diverged from trends of overall population (hazard ratio [HR] > 1 in Japanese subpopulation), and 4.2% (7/165) of the approvals involved trials in which the primary endpoint of OS (including a sole, co-primary or dual primary endpoint) was not achieved. The PMDA decisions reflected holistic evaluation, considering trial design integrity, secondary endpoints, and qualitative consistency rather than rigid HR thresholds. Differences in efficacy in Japanese subgroup rarely impacted labeling, underscoring reliance on global evidence with local scientific considerations. These findings highlight the PMDA's pragmatic approach to endpoint flexibility and extrapolation, reinforcing the importance of early regulatory engagement and strategic alignment in MRCT design for oncology drug development in Japan.