P301 Agreement in histologic diagnosis of preneoplastic lesions in inflammatory bowel disease: Results from a national survey in Italy
G. Canavese1, L. Saragoni2, T. Salviato3, L. Reggiani Bonetti4, G. Leoncini5, L. Baron6, M. Facchetti7, C. Vignale1, V. Villanacci7
1Department of Pathology, S. Giovanni Battista Hospital, Torino, Italy, 2Pathology Unit, Forlì Hospital, Forlì, Italy, 3Pathology Unit, ASUITS Trieste, Trieste, Italy, 4Policlinico di Modena, Pathology Unit, Modena e Reggio Emilia University, Modena, Italy, 5ASST del Garda, Desenzano del Garda, Pathology Unit, Brescia, Italy, 6S. Leonardo Hospital, Castellammare di Stabia, Pathology Unit, Castellammare di Stabia, Italy, 7Spedali Civili di Brescia, Pathology Unit, Brescia, Italy
Background
Inflammatory bowel diseases (IBD) are chronic inflammatory conditions of gastrointestinal tract: the inflammatory damage increases the risk of developing preneoplastic and neoplastic conditions. Therefore, a good agreement between pathologists in the detection of preneoplastic lesion is essential in the management of IBD patients, in order to decrease the risk of progression towards neoplastic lesions. An agreement study on 4 pathologists and 38 cases of dysplasia demonstrated a fair agreement (k=0.4).1 A similar study demonstrated that the lowest level of agreement in the category indefinite for dysplasia (κ = 0.251).
Methods
The study consisted of a survey about diagnostic agreement in a series of preneoplastic lesions of IBD-affected patients, based on digital images. The occurrence of the disease and the occurrence of dysplasia were considered in the study. The study enclosed biopsy specimens from 30 colonoscopies and 1 surgical specimen, related to 20 patients with a clinical pattern of IBD from 4 reference centres in Italy. Digital slides were uploaded in an open-source learning platform. For each endoscopy, sampling sites with similar morphology were aggregated in 54 ‘blocks’, and a series of close-ended questions about (A) the occurrence of IBD (active, in remission, absent, not evaluable) and (B) the evidence of dysplasia (LG, HG, absent, undefined) have been submitted for every block. For each case, a final comprehensive evaluation about (1) the occurrence of IBD (present, absent, unsuitable for assessment), (2) the disease classification (ulcerative colitis, Crohn’s disease, unfit for differential diagnosis, not possible for lack of clinical data), (3) the occurrence of IBD-related dysplasia (IBD with dysplasia, dysplasia not IBD-related, dysplasia untestable, absent) and (4) the classification of dysplasia (sporadic adenoma, ALM, DALM, unclassifiable) were provided. Twenty gastrointestinal pathologists from as many centres in Italy were enrolled.
Results
325 of the 400 tests were successfully concluded.
Agreement values
| overall agreement | mean k value 0.59 |
| evidence of IBD in the cases | k value not assessable for excess of concordant answers |
| differential diagnosis SPORADIC ADENOMA / DALM / ALM | k value not assessable for excess of concordant answers |
| evidence of dysplasia in the cases | mean k value 0,37 |
| overall evidence of dysplasia in blocks | mean k value 0.42 |
Conclusion
A preliminary data analysis demonstrated a good agreement about the occurrence of IBD, and a lower agreement about the occurrence of dysplasia and its classification in IBD. Analysis of the correlation between agreement and clinical-histologic parameters could provide interesting spotlights on diagnostic algorithms in this field.