Predictive Factors for the Outcome of Unsupervised Endoscopic Submucosal Dissection During the Initial Learning Curve with Prevalence-Based Indication

Background For an adequate educational strategy of ESD in non-Asian settings with prevalence-based indication it is essential to define adequate lesions, suitable for the beginner without on-site expert-supervision. Aims We analyzed possible predictors for outcome parameters of effectiveness and safety during the initial learning curve. Methods The first 120 ESDs of four operators ( n = 480), performed between 2007 and 2020 in four tertiary hospitals, were enrolled. Uni-/multivariable regression analysis was done with sex, age, pretreated lesion, lesion size, organ, and organ-based localization as possible independent predictors for en bloc resection (EBR), complication, and resection speed. Results Rates of EBR, complication, and resection speed were 84.5%, 14.2%, and 6.20 (± 4.45) cm 2 /h. Independent predictors for EBR were pretreated lesion (OR 0.27 [0.13–0.57], p < 0.001) and non-colonic ESD (OR 2.29 [1.26–4.17] (rectum)/5.72 [2.36–13.89] (stomach)/7.80 [2.60–23.42] (esophagus), p < 0.001), for complication pretreated lesion (OR 3.04 [1.46–6.34], p < 0.001) and lesion size (OR 1.02 [1.004–1.04], p = 0.012) and for resection speed pretreated lesion (RC – 3.10 [− 4.39 to – 1.81], p < 0.001), lesion size (RC 0.13 [0.11–0.16], p < 0.001) and male patient (RC − 1.11 [− 1.85 to − 0.37], p < 0.001). We found no significant difference in the incidence of technically unsuccessful resections in esophageal (1/84), gastric (3/113), rectal (7/181), and colonic (3/101) ESDs ( p = 0.76). Technical failure was mainly caused by complication and fibrosis/pretreatment. Conclusion During the initial learning curve of an unsupervised ESD program with prevalence-based indication, pretreated lesions and colonic ESDs should be avoided. In contrast, lesion size and organ-based localizations have less predictive value for the outcome.