Endoscopic resection (ER) has become the preferred curative-intent treatment for early upper gastrointestinal cancer, given its superior safety profile compared to surgery. Over the past decade, technological and procedural innovation has substantially expanded the scope, safety, and precision of endoscopic submucosal dissection (ESD) and related techniques. This review synthesises current evidence on key advances relevant to upper gastrointestinal ESD practice. Enhanced imaging modalities have improved lesion detection and characterisation, as well as recognition of intraoperative anatomical structures during third-space endoscopy. A new generation of therapeutic endoscopes combines high-definition optics with substantially improved tip-down angulation and channel size, addressing a longstanding gap between diagnostic-class image quality and procedural capability. Resection strategies—including mechanical traction systems, saline immersion therapeutic endoscopy (SITE), and luminal drainage techniques—have reduced procedural complexity and improved dissection conditions. Dedicated closure technologies have improved management of large resection defects, potentially reducing resection-related morbidity. Deep resection techniques, including submucosal tunnelling endoscopic resection (STER), device-assisted endoscopic full-thickness resection (FTRD), knife-assisted full-thickness resection (kFTR), and endoscopic intermuscular dissection (EID), are extending organ-preserving resection to deeply invasive cancers and subepithelial lesions. Management of non-curative ESD resections is being refined through multicentre risk stratification studies. Advances in simulation, competency-based training, and artificial intelligence hold promise for standardising technique acquisition and real-time procedural support. Together, these innovations are reshaping upper gastrointestinal oncology by positioning minimally invasive, organ-preserving digestive endoscopy as a central therapeutic strategy.
Sorge et al. (Thu,) studied this question.