Reproducibility of the histological diagnosis of celiac disease

Abstract

Objective. A small intestinal biopsy is considered to be the gold standard for the diagnosis of celiac disease (CD). However, the assessment of small intestinal histology may vary between pathologists. Our aim was, therefore, to determine the interobserver variability in the histological diagnosis of CD. Material and methods. Biopsy specimens of 297 pediatric patients suspected of having CD were revised by a single experienced pathologist and compared to the original reports. Mucosal changes were scored using the Marsh classification. In patients with a discrepancy in diagnosis, clinical and serological data were used to determine the most probable diagnosis. Results. Although the interobserver variability for the Marsh classification was found to be moderate with a Kappa value of 0.486, the Kappa value for the diagnosis reached an almost perfect agreement (0.850). Nevertheless, in 22 patients a different diagnosis was made by the second observer. Interestingly, in this subgroup relatively more biopsies were classified to be of suboptimal quality. Based on clinical presentation, serology and follow-up, 19 of those patients truly had CD. In 14 of them the diagnosis was originally missed by the first observer while five cases were under-diagnosed by the second pathologist. Conclusions. CD can be missed histologically due to assessment variation between pathologists. A final diagnosis of CD should be based on histology, serology as well as response to the diet.