Learning Curve for Endoscopic Submucosal Dissection With an Untutored, Prevalence-Based Approach in the United States

Background & Aims: Endoscopic submucosal dissection (ESD) is widely used in Asia to resect early-stage gastrointestinal neoplasms, but use of ESD in Western countries is limited. We collected data on the learning curve for ESD at a high-volume referral center in the United States to guide development of training programs in the Americas and Europe. Methods: We performed a retrospective analysis of consecutive ESDs performed by a single operator at a high-volume referral center in the United States from 2009 through 2017. ESD was performed in 540 lesions: 449 mucosal (10% esophageal, 13% gastric, 5% duodenal, 62% colonic, and 10% rectal) and 91 submucosal. We estimated case volumes required to achieve accepted proficiency benchmarks (>90% for en bloc resection and >80% for histologic margin-negative (R0) resection) and resection speeds >9cm²/hr. Results: Pathology analysis of mucosal lesions identified 95 carcinomas, 346 premalignant lesions, and 8 others; the rate of en bloc resection increased from 76% in block 1 (50 cases) to a plateau of 98% after block 5 (250 cases). The rate of R0 resection improved from 45% in block 1 to >80% after block 5 (250 cases) and ∼95% after block 8 (400 cases). Based on cumulative sum analysis, approximately 170, 150, and 280 ESDs are required to consistently achieve a resection speed >9cm²/hr in esophagus, stomach, and colon, respectively. Conclusions: In an analysis of ESDs performed at a large referral center in the United States, we found that an untutored, prevalence-based approach allowed operators to achieve all proficiency benchmarks after ∼250 cases. Compared with Asia, ESD requires more time to learn in the West, where the untutored, prevalence-based approach requires resection of challenging lesions, such as colon lesions and previously manipulated lesions, in early stages of training.