Factors that predict bleeding following endoscopic mucosal resection of large colonic lesions

Background and study aims: Endoscopic mucosal resection (EMR) for large colonic laterally spreading tumors (LSTs) is a safe, efficacious, and cost-effective treatment. The most common serious complication is delayed bleeding, which reduces these advantages, but consensus guidelines for large-polyp EMR do not exist. Patients and methods: Data from two large prospective intention-to-treat studies of EMR for colonic LSTs 20 mm or greater in size were analyzed. Data collection was comprehensive, and included patient and lesion characteristics. EMR technique and cessation of anticoagulant and antiplatelet therapy was standardized. Clinically significant delayed bleeding was defined as that requiring hospital admission. Results: EMR was performed on 302 lesions in 288 patients. There was clinically significant delayed bleeding in 21 cases (7%). Ten underwent colonoscopy. One required angiography. One required surgery after perforation following hemostatic clip placement. There were no deaths. Risk factors for bleeding on multivariate analysis were right colon location adjusted odds ratio (OR) 4.4, P = 0.01], use of aspirin (OR 6.3, P = 0.005), and age (OR per decade of age 1.70). All bleeds occurred before aspirin was restarted. Patient characteristics, including ASA grade and co-morbidity type, were not predictive. Despite requiring more complex EMR, larger lesion size (P = 0.2), multiple excisions rather than en bloc resection (P = 0.1), polyp morphology (P = 0.2), and previous attempts (P = 0.5), were not associated with increased risk. Conclusions: Proximal lesion location is a highly significant risk for clinically significant delayed bleeding following colonic EMR, and this knowledge could form the basis of a targeted therapeutic trial. Recent aspirin use also increases bleeding risk - specific consensus guidelines in this area are required for colonic EMR.