Learning colorectal endoscopic submucosal dissection: a prospective learning curve study using a novel ex vivo simulator

Background Endoscopic submucosal dissection (ESD) is increasingly being used in Asia as a minimally invasive therapy to eradicate large laterally spreading superficial tumors in the colon. To date, the learning curve and effectiveness of ex vivo simulators in colonic ESD training remain unclear. The aim of the study is to determine the learning curve of colonic ESD in an ex vivo simulator. Methods We conducted a prospective study of colon ESD in ex vivo porcine colons in a prototype simulator. Three endoscopists with prior experience in gastric ESD but with no experience in colonic ESD each performed 30 ESD resections on standardized lesions in the rectosigmoid and left colon of the porcine simulator. Procedure time, en bloc resection status, and perforation were recorded. Results All 90 lesions were resected using the ESD technique. The mean time of procedure was 49.6 min (standard deviation 29.6 min). The aggregate rate of perforation was 14.4% and the aggregate rate of non-en bloc resection was 5.6%. Using a composite quality score integrating complications and procedural time, it was found that there was a significant difference between two local polynomial regression lines when using a cut-point at the 9th procedure ( p = 0.04), reflecting the point at which most of the learning curve is traversed. Conclusions In this study, there were significant improvements realized in colonic ESD performance after 9 colon ESD procedures in ex vivo specimens. Although training will depend on endoscopist skill and expertise, we suggest at least 9 ex vivo procedures prior to moving to live animal or proctored training in colonic ESD.