BACKGROUND AND AIMS: Discrimination of gastric adenomas from adenocarcinomas by conventional endoscopy is difficult. Therefore, we evaluated the usefulness of magnifying endoscopy combined with narrow-band imaging for this differential diagnosis. METHODS: Forty-nine consecutive gastric lesions were diagnosed as adenomas by conventional endoscopy with forceps biopsy and finally resected by endoscopic submucosal dissection. The findings from magnifying endoscopy with narrow-band imaging were retrospectively classified into five types according to the marginal crypt epithelium and microvascular pattern: Types I and II (clear marginal crypt epithelium combined with regular or unclear microvascular pattern) and Types III, IV, and V (unclear marginal crypt epithelium combined with regular, irregular, or unclear microvascular pattern). RESULTS: Conventional endoscopy showed 39 flat elevated-type lesions (0-IIa) and 10 flat elevated-type lesions with depression (0-IIa+IIc). The patterns on magnifying endoscopy with narrow-band imaging were Type I (n = 8), Type II (n = 8), Type III (n = 2), Type IV (n = 30), and Type V (n = 1). The final histological diagnoses after endoscopic submucosal dissection were adenoma (n = 20), adenocarcinoma in adenoma (n = 22), and adenocarcinoma (n = 7). The cancer-bearing rates were Type I (0%), Type II (0%), Type III (100%), Type IV (89.7%), and Type V (100%). Among the expert endoscopists, intra- and interobserver κ values for each type were 0.85 each, with 92.0% and 88.0% consensus of diagnoses, respectively. CONCLUSIONS: Magnifying endoscopy with narrow-band imaging is a powerful tool for diagnosing gastric borderline lesions.
BACKGROUND AND AIMS: Discrimination of gastric adenomas from adenocarcinomas by conventional endoscopy is difficult. Therefore, we evaluated the usefulness of magnifying endoscopy combined with narrow-band imaging for this differential diagnosis. METHODS: Forty-nine consecutive gastric lesions were diagnosed as adenomas by conventional endoscopy with forceps biopsy and finally resected by endoscopic submucosal dissection. The findings from magnifying endoscopy with narrow-band imaging were retrospectively classified into five types according to the marginal crypt epithelium and microvascular pattern: Types I and II (clear marginal crypt epithelium combined with regular or unclear microvascular pattern) and Types III, IV, and V (unclear marginal crypt epithelium combined with regular, irregular, or unclear microvascular pattern). RESULTS: Conventional endoscopy showed 39 flat elevated-type lesions (0-IIa) and 10 flat elevated-type lesions with depression (0-IIa+IIc). The patterns on magnifying endoscopy with narrow-band imaging were Type I (n = 8), Type II (n = 8), Type III (n = 2), Type IV (n = 30), and Type V (n = 1). The final histological diagnoses after endoscopic submucosal dissection were adenoma (n = 20), adenocarcinoma in adenoma (n = 22), and adenocarcinoma (n = 7). The cancer-bearing rates were Type I (0%), Type II (0%), Type III (100%), Type IV (89.7%), and Type V (100%). Among the expert endoscopists, intra- and interobserver κ values for each type were 0.85 each, with 92.0% and 88.0% consensus of diagnoses, respectively. CONCLUSIONS: Magnifying endoscopy with narrow-band imaging is a powerful tool for diagnosing gastric borderline lesions.