Meta-analysis supporting noninferiority of oxaliplatin plus S-1 to cisplatin plus S-1 in first-line treatment of advanced gastric cancer (G-SOX study): indirect comparison with S-1 alone.

ABSTRACT

BACKGROUND: The Randomized Phase III Study Comparing Oxaliplatin plus S-1 with Cisplatin plus S-1 in Chemotherapy-naïve Patients with Advanced Gastric Cancer (G-SOX) showed the noninferiority of S-1 (an oral fluoropyrimidine-derivative dihydropyrimidine dehydrogenase inhibitor) plus oxaliplatin combination therapy (SOX) to S-1 plus cisplatin therapy (CS) in overall survival [hazard ratio (HR) from proportional hazard model 0.958, 95 % confidence interval (CI) 0.803-1.142; noninferiority margin 1.15]. To further clarify the clinical position of SOX in advanced gastric cancer (AGC), a meta-analysis including information from other reported studies was conducted. METHODS: In addition to G-SOX, Japanese phase III clinical trials including S-1 monotherapy were included in the analyses. Individual patient data for SOX (318 patients) and CS (324 patients) from G-SOX, as well as those for S-1 (160 patients) from the Randomized Phase III Study Comparing the Efficacy and Safety of Irinotecan plus S-1 with S-1 Alone as First-line Treatment for Advanced Gastric Cancer (GC0301/TOP-002), were available. Published clinical information for S-1 from other studies (total 705 patients) was also collected. A Weibull distribution was assumed for overall survival time, and parameters for SOX, CS, and S-1 were estimated parametrically. Posterior HR distributions were obtained with a Bayesian approach. RESULTS: The HR of SOX to S-1 was 0.817 (95 % credible interval 0.704-0.939), and the probability of the HR <1.00 was 99.8 %. The HR of CS to S-1 was 0.871 (95 % credible interval; 0.754-0.998), and the probability of the HR <1.00 was 97.6 %. The HR of SOX to CS in G-SOX was 0.942 (95 % credible interval; 0.789-1.117), and the probability of HR <1.15 was 98.9 %. CONCLUSION: This meta-analysis indicates that SOX was superior to S-1 and noninferior to CS in AGC.