RESUMO
AIMS AND BACKGROUND: Up to now adjuvant chemotherapy after curative resection for gastric cancer (GC) has been considered an experimental approach. The results of existing phase III randomized trials comparing chemotherapy with control after surgery are controversial. Three meta-analyses have been published in recent years. It is likely that each of them presents a theoretical bias, mainly as regards the inclusion criteria of the trials. In this article we re-examine this potential bias, highlighting the differences between the present and past meta-analyses on adjuvant chemotherapy for GC. METHODS: Only randomized controlled clinical trials comparing systemic adjuvant chemotherapy with control after radical resection of GC were eligible. Total mortality was assessed as outcome measure of the treatment effect and a pooled odds ratio was calculated using the Peto-Mantel-Haenszel method. RESULTS: After the selection process 17 papers (18 comparisons) proved eligible for inclusion in the meta-analysis with a total of 3118 patients, of whom 1546 randomized to the treatment arms and 1572 to the control arms; 762 and 871 deaths occurred in the treatment and control arms, respectively. Statistical analysis suggests an absence of significant heterogeneity between the trials and a significant advantage in survival for adjuvant chemotherapy (pooled odds ratio, 0.72, 95% Cl, 0.62-0.84). CONCLUSIONS: Our meta-analysis would seem to indicate that adjuvant chemotherapy results in a significant survival advantage in patients with GC. However, this observation undoubtedly requires confirmation in large randomized controlled trials including cisplatin before adjuvant chemotherapy after curative resection for GC can be proposed for use in clinical practice.