Background: A variety of different endoscopic resection techniques for early stage cancer of the upper Gl tract have been described that are more effective than strip blopsy. However, there is no report of a prospective randomized comparison of different techniques.

Methods: In a prospective randomized study, 100 consecutive endoscopic resections were performed in 72 patients with early stage esophageal cancer. Fifty endoscopic resections were performed with a “suck-and-ligate” device without prior submucosa injection and 50 with the cap technique with prior submucosa injection of a dilute saline solution of epinephrine. The main assessment criteria were maximum diameter of the resection specimen and of the resection area, and the complication rate.

Results: No significant differences were observed between the two groups with regard to the maximum diameters and calculated area of the resected specimens (ligation group: 16.4 [4.0]×11 [3.1] mm/185 [84] mm² vs. cap group: 15.5 [4.1]×10.7 [2.7] mm/168 [83] mm²), or the maxlmum diameters and calculated area of the endoscopic resection ulcers after 24 hours (ligation group: 20.6 [4.8]×14.3 [4.5] mm/314 [160] mm² vs. cap group: 18.9 [5.1]×12.9 [3.8] mm/260 [145] mm²). There was only a slight advantage (greater diameter of resection specimens) for the ligation group in patients who had prior endoscopic treatment. There was one minor eplsode of bleeding in each group; there was no severe complication. In 41 of 72 patients (57%), further endoscopic therapy after endoscopic resection was necessary because of residual neoplasia at the first follow-up endoscopy after resection (61 of 100 resection specimens [61%] had lateral margins that could not be evaluated because of coagulation artifact or contained malignancy but with the base of the lesion free of tumor).

Conclusions: The cap technique with submucosa injection and the ligation technique without submucosa injection are similar with respect to efficacy and safety for endoscopic resection of early stage esophageal cancers.