Can an endocytoscope system (ECS) predict histology in neoplastic lesions?

Abstract

Background and study aims: An endocytoscope system (ECS) has recently been developed with the possibility of super-high magnification of gastrointestinal mucosa, thus allowing in vivo imaging of living cells. The aim of the present study was to assess the potential of ECS in the prediction of histology in both normal gastrointestinal mucosa and neoplastic lesions.

Patients and methods: In total, 76 patients (57 men, 19 women; age range 37-86 years) with neoplastic lesions in the esophagus, stomach, or colon were enrolled into the study and underwent esophagogastroduodenoscopy or colonoscopy. After staining with 1% methylene blue, the mucosa was examined with the ECS probe (x 450 and x 1100 magnification), and video sequences were recorded on video disk. Biopsies from the examined areas were taken for histology and served as the gold standard. The endocytoscope video sequences were evaluated by two blinded pathologists. Finally the results were compared with those resulting from the evaluation of an experienced endoscopist who was aware of the macroscopic endoscopic pictures and the endocytoscope image results.

Results: A total of 25 patients with esophageal lesions, 28 patients with colonic lesions, and 23 patients with gastric lesions were examined. The sensitivity and specificity for the evaluation of the blinded pathologists was 81% and 100%, respectively, in the esophagus, 56% and 89% in the stomach, and 79% and 90% in the colon. If an endoscopist evaluated the endocytoscopic pictures in combination with the macroscopic endoscopic images sensitivity and specificity increased significantly.

Conclusions: First experiences with ECS show good sensitivity rates even by blinded assessment for esophageal and colonic lesions. Sensitivity for neoplastic lesions in the stomach is lower because of gastric mucous secretion. Combining the endoscopic and cytoscopic appearance of the lesion may further enhance the diagnostic value of the method.