Clinicopathological features of narrow-band imaging endoscopy and immunohistochemistry in ultraminute esophageal squamous neoplasms.

To reveal clinicopathological features of narrow-band imaging (NBI) endoscopy and immunohistochemistry in ultraminute esophageal squamous neoplasms. If a lesion diameter was smaller or same compared with a width of closed biopsy forceps, a lesion was defined to be an ultraminute lesion. Twenty-five consecutive patients with 33 ultraminute esophageal lesions that were removed by endoscopic mucosal resection were included in the present study. We conducted two questionnaire surveys of six endoscopists by their retrospective review of endoscopic still images. The six endoscopists evaluated the endoscopic findings of the ultraminute lesions on still images taken by conventional white-light imaging endoscopy and non-magnified NBI endoscopy in the first questionnaire, and taken by magnified NBI endoscopy in the second questionnaire. An experienced pathologist who was unaware of any endoscopic findings made histological diagnosis and evaluated immunoexpression of p53 and Ki67. The 33 ultraminute lesions were all determined to be either 11 high-grade intraepithelial neoplasias (HGIENs) or 22 low-grade intraepithelial neoplasias (LGIENs). The tumor diameters were histologically confirmed to be <3 mm. All of the ultraminute tumors were visualized as unstained areas and brownish areas by real-time endoscopy with Lugol dye staining and non-magnified NBI endoscopy, respectively. All of the ultraminute IENs were visualized as brownish areas by real-time non-magnified NBI endoscopy. Three of the 25 patients with the ultraminute IENs (12%) had multiple brownish areas (more than several areas) in the esophagus on real-time non-magnified NBI endoscopy. All of the ultraminute IENs were visualized as unstained areas by real-time Lugol chromoendoscopy. Twenty of the 25 patients (80%) had multiple unstained areas (more than several areas) in the esophagus on real-time Lugol chromoendoscopy. The first questionnaire survey revealed that a significantly higher detection rate of the ultraminute IENs on non-magnified NBI endoscopy images compared with conventional white-light imaging endoscopy ones (100% vs. 72%, respectively: P < 0.0001). The second questionnaire survey revealed that presence rates of any magnified NBI endoscopy findings were not significantly different between HGIENs and LGIENs. Proliferation, dilation, and various shapes of intrapapillary capillary loops indicated remarkably high presence rates of more than 90% in both HGIENs and LGIENs. Six of 22 LGIENs (27%) and 3 of 11 HGIENs (27%) show a positive expression for p53. None of peri-IEN epithelia was positive for p53. A mean of Ki67 labeling index of LGIENs was 33% and that of HGIENs 36%. Ki67 labeling index was significantly greater in the LGIENs and HGIENs compared with that in the peri-IEN epithelia. There were no significant differences in p53 expression and Ki67 labeling index between the HGIENs and LGIENs. Non-magnified/magnified NBI endoscopy could facilitate visualization and characterization of ultraminute esophageal squamous IENs. The ultraminute HGIENs and LGIENs might have comparable features of magnified NBI endoscopy and immunohistochemistry.