Usefulness of videoduodenoscopy and vital dye staining as indicators of mucosal atrophy of celiac disease:: assessment of interobserver agreement

Background: The present study was designed to determine the diagnostic usefulness of videoduodenoscopic inspection alone and the addition of vital dye staining in the detection of celiac disease. We additionally sought to evaluate interobserver agreement for specific duodenoscopic markers of mucosal atrophy. Methods: One hundred sixty-seven consecutive subjects who underwent duodenoscopy for intestinal biopsy were included in a prospective controlled study. Endoscopic examination was performed by experienced endoscopists according to a set protocol using methylene blue (1%) dye. All procedures were recorded on videotape, but only 20 (10 with atrophy and 10 normal) were used in a blinded, independent, randomized analysis by five reviewers to evaluate interobserver agreement. Endoscopic signs indicative of mucosal atrophy were as follows: reduction in the number or loss of Kerkring's folds, "scalloped" folds, "mosaic pattern," and visualization of the underlying blood vessels. Results: Eighty-seven patients had celiac disease (57 newly diagnosed, 30 when treated). Seven treated patients had nonatrophic mucosa. In 80 patients the final diagnosis excluded celiac disease. Videoendoscopic inspection alone correctly identified 75 of 80 patients with complete mucosal atrophy and 86 of 87 with normal mucosa. False-negative diagnoses occurred in treated celiac patients with mild atrophy. Mosaic pattern (89%) and scalloped folds (86%) were the most useful endoscopic signs. Vital dye staining, as assessed by experienced endoscopists, provided identical results to those obtained by inspection alone. Sensitivity, specificity, and positive and negative predictive values for the presence of one or more than one feature were 94%, 100%, 100%, and 96%, respectively. The agreement (kappa statistics) among observers was excellent for the mosaic pattern (kappa: 0.76 for both the videoendoscopic inspection alone and dye staining) and the scalloped folds (kappa: 0.83 and 0.76, respectively) and was fair (kappa: 0.41 and 0.59, respectively) for the reduction in the number or loss of duodenal folds. Conclusion: This study confirms that videoduodenoscopy is useful in the detection of intestinal atrophy. Dye staining produces a better delineation of scalloped folds and mosaic pattern in the atrophic mucosa, but did not provide additional information to the expert endoscopist. Finally, interobserver agreement was excellent for the most prevalent signs.