Outcomes of endoscopic ultrasound and endoscopic resection of gastrointestinal subepithelial lesions: a single-center retrospective cohort study

Background Endoscopic resection (ER) is an emerging therapeutic alternative for subepithelial gastrointestinal lesions (SELs). We aimed to determine whether size, layer of origin, and histology based on endoscopic ultrasound (EUS) and EUS-guided sampling (EUS-GS) influenced the outcomes and selection of patients for ER. Methods We performed a retrospective review of patients who underwent EUS, EUS-GS and resection of SELs from 2012-2019. Two pathologists reviewed the histology and layer of origin of all resected specimens, serving as the criterion for EUS accuracy. Results Seventy-three patients were included, of whom 59 (81%) were gastric SELs. Per EUS, median lesion size was 21 mm (interquartile range 15-32), and 63 (86%) originated from the 4th layer. The overall accuracy of EUS and EUS-GS in predicting the layer of origin and histology was 88% (95% confidence interval [CI] 77-94%) and 96% (95%CI 87-98%), respectively. Based on EUS, 18 (25%) patients were referred for ER, 5 (7%) to laparoscopic-endoscopic cooperative surgery, and 50 (68%) to surgery. Size >20 mm was associated with the type of resection approach (P=0.005), while layer of origin and histology were not (P=0.06 and P=0.09, respectively). When SELs were inaccurately classified (n=4) there were no adverse events or revision of the resection approach. Conclusions EUS plays an important role in the outcome of resection approach for SELs, with size significantly influencing the selection for ER. In patients undergoing ER, no revised resections were needed when EUS was inaccurate.