Confocal endomicroscopy for phenotypic diagnosis of gastric cancer.

BACKGROUND AND AIM: Relationships between mucin phenotype and malignant potential in gastric cancers have attracted attention. We attempted to assess the possibility of obtaining phenotypic diagnoses by confocal endomicroscopy.
METHODS: Confocal images of target lesions were obtained in 29 of 40 patients with gastric cancer. Appearances of the brush border, goblet cells, and gastric foveolar epithelium were investigated with immunohistochemical staining using CD10, MUC2, and human gastric mucin to evaluate phenotypic expression in gastric carcinomas. Confocal images were compared with immunohistochemical findings for goblet cells and brush borders.
RESULTS: Both the endoscopists and the pathologist obtained high accuracy rates for differential diagnosis. Sensitivity and specificity for goblet cells were 85.7% and 92.3% (Endoscopist A), and 85.7% and 88.5% (Endoscopist B). The kappa-value for correspondence between two endoscopists for the diagnosis of goblet cells in confocal images was 0.73. Sensitivity and specificity for the brush border were 93.8% and 91.7% (Endoscopist A), and 81.3% and 91.7% (Endoscopist B). The kappa-value for correspondence between two endoscopists for diagnosis of the brush border in confocal images was 0.79. Intestinal phenotypic gastric cancers show a brush border, goblet cells, or both. Sensitivity and specificity for the intestinal phenotype in confocal endomicroscopy were 90.9% and 77.8% (Endoscopist A), and 86.4% and 83.3% (Endoscopist B).
CONCLUSION: The confocal endomicroscopic diagnosis of the mucin phenotype in gastric cancers was limited to intestinal and mixed phenotypes, but may be useful for the diagnosis of mucin phenotype and differential diagnosis.