Adenoma recurrence after piecemeal colonic EMR is predictable: the Sydney EMR recurrence tool

Background and Aims: EMR is the primary treatment of large laterally spreading lesions (LSLs) in the colon. Residual or recurrent adenoma (RRA) is a major limitation. We aimed to identify a robust method to stratify the risk of RRA. Methods: Prospective multicenter data on consecutive LSLs >= 20 mm removed by piecemeal EMR from 8 Australian tertiary-care centers were included (September 2008 until May 2016). A logistic regression model for endoscopically determined recurrence (EDR) was created on a randomly selected half of the cohort to yield the Sydney EMR recurrence tool (SERT), a 4-point score to stratify the incidence of RRA based on characteristics of the index EMR. SERT was validated on the remainder of the cohort. Results: Analysis was performed on 1178 lesions that underwent first surveillance colonoscopy (SC1) (median 4.9 months, interquartile range [IQR] 4.9-6.2). EDR was detected in 228 of 1178 (19.4%) patients. LSL size >= 40 mm (odds ratio [OR] 2.47; P < .001), bleeding during the procedure (OR 1.78; P = .024), and high-grade dysplasia (OR 1.72; P = .029) were identified as independent predictors of EDR and allocated scores of 2, 1, and 1, respectively to create SERT. Lesions with SERT scores of 0 (SERT = 0) had a negative predictive value of 91.3% for RRA at SC1, and SERT was shown to stratify RRA to specific follow-up intervals by using Kaplan Meier curves (log-rank P < .001). Conclusions: Guidelines recommend SC1 within 6 months of EMR. SERT accurately stratifies the incidence of RRA after EMR. SERT = 0 lesions could safely undergo first surveillance at 18 months, whereas lesions with SERT scores between 1 and 4 (SERT 1-4) require surveillance at 6 and 18 months.