BACKGROUND: Barrett's esophagus is a well-known pre-malignant condition. Pathologic interpretation of biopsy specimens guides endoscopic surveillance as well as the therapeutic approach that will be carried out. However, the predictive value of histopathologic diagnosis can be questioned due to its poor intra- and interobserver reproducibility. AIMS: To assess intra- and interobserver variability in the diagnosis of Barrett's dysplasia. MATERIAL AND METHODS: Three-micrometer thick sections from biopsy specimens from 42 patients with Barrett's esophagus were stained with hematoxylin-eosin and PAS-alcian blue. The reading of the slides was carried out blindly in a light microscope. Intra and interobserver variability in the interpretation of the slides was determined by kappa statistics. RESULTS: The number of tissue specimens was 229, with average of 5.45 (1 to 18) fragments for patient. Low grade dysplasia was diagnosed by pathologists in 21.4% to 52.4% of the cases. The intra-observer agreement for the diagnosis of low grade dysplasia was slight (kappa = 0.30). The interobserver agreement for the diagnosis of low grade dysplasia was poor, with kappa scores between 0.05 and 0.16. The diagnosis of dysplasia, with agreement for all pathologists examining the same set of slides, was 14.3%. CONCLUSIONS: Pathologic interpretation of Barrett's dysplasia may be subject to marked intra- and interobserver variabiliaty. Interpretation of low grade dysplasia, as high grade dysplasia, should also be considered for review by two or more pathologists.
Barrett esophagus; Observer variation
