7089 Endoscopic mucosal resection technique for upper gastrointestinal submucosal tumor.

Abstract

Background: Although endoscopic ultrasound (EUS) has been established as a best imaging technique in detection of gastrointestinal submucosal tumors (SMT), the accurate histologic diagnosis and differentiation between benign and malignant tumors are not satisfactory according to EUS criteria only. Endoscopic resection of gastrointestinal SMT, as an alternative to surgery is not widely used as a diagnostic or therapeutic choice because of the mostly benign nature and uncertain resection depth. We present our clinical experience of endoscopic mucosal resection (EMR) for various upper gastrointestinal SMTs and evaluate the safety and limitation of these techniques. Methods: Thirty-three patients (24 male, 9 female, mean age 45.6) with upper gastrointestinal SMTs underwent EMR between July 1997 and October 1999. Four lesions were located at esophagus, 23 at stomach, and 6 at duodenal bulb. These SMTs were confirmed by EUS using the Olympus GF-UM200/EU-M30 or ultrasonic probe (12/20 MHz). Lesions exceeding 3 cm, or arising from muscularis propria were excluded from this study. Two patients had endoscopic banding of the lesion prior to electrocautery resection. The other 31 patients received endoscopic mucosal resection by cap-fitted endoscope (EMRC) method. All patients had recieved H2 blocker therapy after overnight fasting for 4-6 weeks and endoscopy/ EUS follow-up thereafter. Results: The average resected specimen size was 8.6 mm (5-12 mm). Eradication is possible in 23 of these patients (70%). Inadequate samples for pathologic diagnosis 3 patients resulted from deep submucosal localization and incomplete detachment from muscularis propria after saline injection. The diagnosis of 33 SMTs were confirmed in 30 patients and included: 14 heterotopic pancreas, 6 Brunner's gland adenomas, 3 leiomyomas, 3 submucosal fibrosis, 1 lipoma, 1 gastritis cystica profunda, 1 carcinoid tumor, 1 vascular ectasia, 1 hemangioma, 1 inflammatory fibroid polyp, 1 granular cell tumor. No immediate bleeding or perforation resulted after EMR. Bleeding occurred on 3 patients (9.1%) within 48 hr, but all cases were controlled by epinephrine injection. All resection ulcers healed or became scar at followup. Conclusions: EMRC or band ligation could be a useful and easy method for histologic diagnosis and treatment of selected GI submucosal tumors. But artificial ulcer and delayed bleeding were the major complication. EUS and adequate submucosal saline injection prior to EMR were important in successful endoscopic resection.