Diagnostic interobserver variability in Crohn's disease- and ulcerative colitis-associated dysplasia: a multicenter digital survey from the IG-IBD Pathologists Group

BACKGROUND: Crohn's disease (CD) and ulcerative colitis, two forms of inflammatory bowel disease (IBD), are chronic and relapsing conditions of the gastrointestinal tract both characterized by long lasting chronic inflammation and increased risk of dysplasia and colorectal cancer (CRC). The aim of our study was to evaluate the interobserver agreement about IBD-associated dysplasia among pathologists belonging to the Italian Group for Inflammatory Bowel Diseases (IG-IBD P). METHODS: The present multicenter survey was performed using telepathology, supported by an open source E-learning platform. Biopsy specimens from 30 colonoscopies and from 20 patients were included. The glass slides of any case, including clinical and endoscopic data, were digitalized and uploaded on the E-learning platform. All the digital slides were grouped in 54 diagnostic "blocks". Blinded histopathological evaluation on all the digital slides was performed by 20 gastrointestinal pathologists. Closed-ended questions about (1) the occurrence of IBD; (2) the classification of IBD (as UC or CD); (3) the presence of active versus quiescent disease; (4) the presence of dysplasia; (5) the possible association of dysplasia with the sites of disease (dysplasia-associated lesion or mass-DALM vs adenoma-like mass-ALM); (6) the grading of dysplasia according to the ECCO guidelines (negative, indefinite, low grade, high grade categories) and (7) the presence of associated serrated features, were proposed in each case. Inter-observer agreement was evaluated by mean agreement percentage and kappa statistic, when suitable. RESULTS: The diagnosis of IBD was confirmed in 19 of 20 patients, 17 of 19 being classified as UC, 2 as CD. The mean interobserver agreement percentages about (1) the evidence of IBD, (2) the presence of either UC or CD and (3) the activity grading resulted to be 80%, 69% and 86%, respectively. Dysplasia was detected in 8/20 patients, with moderate agreement between pathologists (mean 72%, k 0.48). Particularly, low grade dysplasia was found in 13 biopsies (combined k 0.38), whereas high grade dysplasia in 8 (combined k 0.47). When the endoscopic and histopathological data were combined, features consistent with DALM were found in 6 of 20 patients with low grade dysplasia and those consistent with ALM in 2 patients with low grade dysplasia in a single biopsy (mean agreement: 86%). An associated serrated pattern was discovered in 4 patients (7 biopsies). CONCLUSIONS: Our study showed moderate interobserver agreement about the histopathological detection and classification of IBD-associated dysplasia. Further efforts should be undertaken to integrate the histopathological data with both the ancillary tests and molecular investigations.