Evaluation of Additive Effects of Narrow Band Imaging (NBI) Magnification in Borderline Gastric Cancer Diagnosis: Randomized Comparative Study of Conventional/Dye Endoscopy With and Without NBI-Magnification Toshihisa Takeuchi*, Eiji Umegaki, Yuichi Kojima, Ken Narabayashi, Yukiko Yoda, Akira Imoto, Sadaharu Nouda, Kumi Ishida, Yosuke Abe, Daisuke Masuda, Takuya Inoue, Satoshi Tokioka, Kazuhide Higuchi 2nd Dep of Internal Medicine, Osaka Medical Collage, Takatsuki, Japan Introduction: Endoscopic submucosal dissection (ESD) is used globally for early gastric cancer, and accurate borderline lesion diagnosis is important. Current qualitative gastric cancer diagnosis involves magnification NBI. Distinguishing benign and malignant tumors by surface microstructural and capillary patterns, and diagnosing excavated gastric cancer histotypes, are reported to be possible, but there is no firm position on NBI-magnification borderline gastric cancer diagnosis. Objectives: To investigate additive effects of NBI-magnification for borderline diagnosis of ESD-target gastric cancer, and pathological characteristics of lesions not diagnosable even with NBI-magnification. Subjects and Methods: ESD-target gastric cancer cases were randomly allocated, 250 lesions each, to borderline diagnosis by conventional/dye endoscopy, with and without NBImagnification. The former used maximum magnification, structural reinforcement B8, and an Evis Lucera magnifying endoscope with a slit-tip transparent hood. Cancer and other tissues were distinguished by Yao’s VS classification system, and lesion border demarcation lines were determined. With each group, three endoscopists performed or attended pre-ESD borderline lesion diagnosis, marking approx. 5 mm outside lesions, and ESD. Preoperative borderline diagnosis and ESD specimens were compared pathologically, and the rate of failing borderline diagnosis (lateral margin positivity), and causative factors, were evaluated. Results: There were no significant inter-group differences in lesion size, shape, or position, or ESD duration. Conventional/dye endoscopy borderline diagnosis failure rate was significantly lower with than without NBImagnification, at 1.6% (4/250) and 5.6% (14/250; p 0.05). Lesions failing borderline diagnosis are (a) well-differentiated adenoma with mucosal surface layer replacement-type growth; (b) moderately differentiated adenoma dominated by mucosal middle layer invasion, with little surface layer atypia; or (c) undifferentiated cancer with mucosal middle layer invasion. Conclusions: In borderline gastric cancer diagnosis in ESD-target patients, addition of NBImagnification to conventional/dye endoscopy significantly increased diagnostic accuracy, showing the method’s additive effects. However, in histological evaluation of lesions (well-differentiated adenoma with little mucosal surface layer atypia, low-atypia cancer with mucosal middle layer invasion, etc) failing borderline diagnoses even with NBI-magnification, identification of the demarcation line was considered to be difficult.