Trial of 3-weekly Versus 5-weekly Schedule of S-1 Plus Cisplatin in Gastric Cancer: SOS

The primary endpoint of this study is Progression Free Survival (PFS). It is defined as the time from the date of randomization to the time of disease progression as assessed by the investigators, or death due to any cause. The primary goal of this study is to compare two different schedules of S-1 plus cisplatin combination treatments (3-weekly regimen vs. 5-weekly regimen) for advanced gastric cancer with respect to the PFS based on the hybrid design where we can test superiority and non-inferiority in the same trial (Reference: Journal of Clinical Oncology 25: 5019-5023, 2007, Boris Freidlin, et el). First, the non-inferiority hypothesis will be tested based on the non-inferiority margin 1.15. If the inferiority cannot be rejected(meaning non-inferiority is proven) then the superiority will be tested. If the superiority test is positive, then superiority is concluded; otherwise non-inferiority without superiority will be concluded.

The sample size was calculated from the following consideration:

For non-inferiority test: non-inferiority margin 1.15, 10 percent reduction of hazard ratio, power 80 percent, alpha 0.025, accrual period 36 months, follow-up period 12 months, and the expected median PFS of 6 months for 5 weekly regimen were assumed. Based on the above considerations, total of 560 patients will be need. With 10 percent follow-up loss, we need 622 patients.

For superiority test: the median PFS for 5-weekly regimen is expected to be 6 months and 7.5 months for 3-weekly regimen. With sample size of 560 patients calculated above, for detecting 1.5 months difference in the median PFS between the two groups , we will have 81 percent of power, one-sided 5 percent type I error. Using the log rank test assuming exponential underlying distribution, accrual period of 36 months, minimum of 12 months follow-up after the last enrolment, 516 events will be needed to show the superiority of 3 week cycle.