Long-term outcomes and interval cancers after endoscopic mucosal resection of large colorectal polyps

Background Data on the short- and long-term outcomes of endoscopic mucosal resection (EMR) for large colorectal polyps are essential for guiding treatment algorithms. Objective To evaluate the short- and long-term outcomes of EMR for large rectal and colonic polyps. Design We conducted a retrospective cohort study of consecutive patients who underwent EMR for colorectal polyps ≥20 mm at a single tertiary centre between 2000 and 2020. Procedures were performed by 15 colonoscopists. Data were retrieved from hospital records and the National Cancer Registry. The primary outcome was post-colonoscopy colorectal cancer (PCCRC; median follow-up 9.1 years). The main secondary outcome was recurrence risk, assessed in patients with available follow-up data (71.2% follow-up rate; median follow-up, 15.5 months). Cox proportional hazards models were used to analyse these long-term outcomes. Candidate confounders included age, sex, lesion size, morphology, location (rectum vs colon), prior treatment, technical failure, resection technique (en bloc vs piecemeal) and grade of dysplasia or invasive cancer. Results A total of 2413 lesions in 2205 patients were analysed; 26.3% were rectal and 73.7% colonic. Overall, 27.3% of lesions were pedunculated, 27.5% sessile and 45.2% flat. Rectal lesions more often harboured invasive cancer (OR 1.66, 95% CI 1.17 to 2.37; p=0.005), an association driven predominantly by laterally spreading lesions (OR 5.20, 95% CI 2.68 to 10.09). Rectal location was independently associated with recurrence (16.6% vs 6.9% for colonic lesions; HR 1.57, 95% CI 1.18 to 2.08; p=0.002), and recurrence was a strong predictor of PCCRC (2.4% vs 0.9% for colonic lesions; HR 10.32, 95% CI 4.59 to 23.20; p<0.001). In contrast, rectal location was not an independent predictor of PCCRC. Conclusion Large rectal lesions were more likely to harbour invasive cancer and carried a higher risk of recurrence. Although recurrence independently predicted PCCRC, rectal location alone did not.