Endoscopic and oncologic outcomes according to indication criteria of endoscopic resection for early gastric cancer: a systematic review and meta-analysis

Background: The criteria for endoscopic resection for early gastric cancer (EGC) have been expanded recently, and it has become acceptable to use techniques that are regarded as having equivalent technical and pathological outcomes to absolute indication (AI). However, the long-term oncological outcomes of expanded indication (EI) have yet to be clarified. This meta-analysis aimed to assess the long-term outcome of EI versus AI, to identify the endoscopic feasibility and safety according to the indication, and to provide the appropriate recommendations for each indication. Methods: Electronic databases including PubMed, EMBASE, Cochrane Central Register of Controlled Trials, and KoreaMed were searched for articles published between January 2000 and October 2014. After screening, the reviewers extracted the information from 12 retrospective cohort studies. A total of 9635 EGC lesions, 4150 lesions in the EI group and 5485 lesions in the AI group, were included in this study. Results: Meta-analyses showed that the local recurrence rate [risk ratio (RR) 1.34; 95 % CI 0.67–2.70] was not significantly higher in the EI group compared with the AI group, although the metachronous recurrence rate was higher in the EI group than in the AI group (RR 1.60; 95 % CI 1.22–2.10). The rates of en bloc resection [odds ratio (OR) 0.57; 95 % CI 0.41–0.78), complete resection (OR 0.37; 95 % CI 0.25–0.57), and curative resection (OR 0.34; 95 % CI 0.20–0.58) were significantly inferior in the EI group than in the AI group, whereas overall bleeding risk (RR 1.47; 95 % CI 1.19–1.82) and procedure-related perforation rate (OR 2.04; 95 % CI 1.56–2.68) were significantly higher in the EI group than in the AI group. Conclusions: This meta-analysis suggests that the EI group showed acceptable long-term outcomes with local recurrence rate that was not significantly inferior, although the metachronous recurrence rate was higher compared with that in the AI group.