Background and study aims: Intestinal Behçet's disease and Crohn's disease are chronic inflammatory bowel diseases that are difficult to distinguish from each other. We investigated their colonoscopic features and identified simple and valuable strategies for differential diagnosis. Patients and methods: Between 1995 and 2006, 250 consecutive patients with ulcers on colonoscopy (115 Behçet's, 135 Crohn's cases) were reviewed. All patients with Behćet's fulfilled the criteria of the International Study Group for Behćet's Disease or of the Behçet's Disease Research Committee of Japan, while Crohn's disease was confirmed by clinicopathological data. Patients were randomly allocated to a training set (70%) or a validation set (30%). Ulcer shapes, distributions, numbers, margins, and border contours, and the presence of aphthous, cobblestone, perianal, and strictured lesions were compared, in the training set. Univariate and multivariate analysis were performed, using the x2-test and logistic regression. In addition, a classification and regression tree (CART) was then used to generate simplified algorithms for differential diagnosis. Results: Round shape, five or fewer in number, focal distributions, and absence of aphthous and cobblestone lesions were significantly dominant features in Behçet's disease, according to multivariate analysis of the training set. The CART-generated algorithms proposed sequential use of shape (round, irregular/geographic, or longitudinal) and distribution (focal single/focal multiple, or segmental/diffuse). Diagnosis of Behçet's disease in the validation set produced sensitivity, specificity, and negative and positive predictive values of 94.3%, 90.0 %, 94.7%, and 89.2 %, respectively. Using the CART model, we made the correct diagnosis of intestinal Behçet's disease or Crohn's disease in 69 of 75 patients (92%). Conclusion: It was determined that round and longitudinal ulcers are suggestive of Behçet's disease and Crohn's disease, respectively. Irregular/ geographic-shaped ulcers and focal distributions are suggestive of Behçet's disease, while segmental/diffuse lesions suggest Crohn's.
All Science Journal Classification (ASJC) codes