OBJECTIVES: Clinically significant bleeding (CSPEB) is the most frequent adverse event following wide-field endoscopic mucosal resection (WF-EMR) of large sessile and laterally spreading colorectal lesions (LSL). There is limited knowledge regarding accurate prediction of CSPEB. We aimed to derive a score to predict the risk of CSPEB.METHODS: Data on patient and lesion characteristics and outcomes from WF-EMRs of LSL >= 20 mm at 8 referral hospitals were analyzed. The cohort was divided at random into equal sized training and test groups. Independent predictors of CSPEB in the training cohort were identified by multiple logistic regression analysis and used to develop a risk score. The performance of this score was assessed in the independent test cohort.RESULTS: Over 80 months to June 2015, 2,128 patients with 2,424 LSL were referred for WF-EMR. Two thousand and twelve patients were eligible for analysis. There were 135 cases of CSPEB (6.7%). In the training cohort of 1,006 patients, the independent predictors of CSPEB were lesion size >30 mm (odds ratio (OR) 2.5), proximal colonic location (OR 2.3), presence of a major comorbidity (OR 1.5), and epinephrine in injection solution (OR 0.57). The derived risk score comprised lesion size >30 mm (2 points), proximal colon (2 points), presence of major comorbidity (1 point), and absence of epinephrine use (1 point). The probabilities of CSPEB for scores of 0, 1, 2, 3, 4, and >= 5 in the training cohort were 1.5, 2.0, 5.6, 7.8, 9.1, and 17.5% and were 0.9, 6.7, 4.9, 6.2, 9.0, and 15.7% in the test cohort. The probabilities of CSPEB in those with low (score 0-1), medium (score 2-4), and elevated (score 5-6) risk levels were 1.7, 7.1, and 17.5% in the training cohort and 3.4, 6.2, and 15.7% in the test cohort.CONCLUSIONS: Patients at elevated risk of CSPEB can be identified using four readily available variables. This knowledge may improve the management of those undergoing WF-EMR and assist in designing studies evaluating CSPEB.

Systematic endoscopic assessment of advanced mucosal neoplasia (AMN - sessile or flat lesions ≥20mm in size) permits prediction of submucosal invasive adenocarcinoma (SMI) and guides the therapeutic strategy. Laterally spreading tumors (LSTs) with a non-granular (NG) surface or 0-Is or 0-IIc components are established risk factors for SMI, however lesion characteristics which influence the accuracy of SMI prediction are not well understood.