P310. Added value of anti CBir-1 antibodies to differentiate IBD unclassified from ulcerative colitis
O. Ben-Bassat1, W. Xu2, J.M. Stempak3, M.S. Silverberg1, G. Van Assche1, 1Mount Sinai Hospital, Gastroenterology, Toronto, Canada, 2University of Toronto, Dalla Lana School of Public Health, Toronto, Canada, 3Mount Sinai Hospital, Zane Cohen Centre for Digestive Diseases, Toronto, Canada
Serological markers have been applied with some success to reclassify patients with inflammatory bowel disease, type unclassified (IBDU) to Crohn's disease (CD) or ulcerative colitis (UC). Addition of anti-OmpC and anti-I2 to pANCA and ASCA antibodies in a serological panel has not been able to improve the discriminative model. Since a diagnosis of IBDU still represents a major challenge when surgical management is considered, we attempted to identify a characteristic profile of patients with IBDU using a panel of serological markers. In addition, we constructed a model to specifically identify a diagnosis of IBDU in patients with colonic IBD.
We performed a cross-sectional analysis of patients with IBDU (42 patients) ascertained through Mount Sinai Hospital IBD Centre in Toronto. Serological profiles (ASCA IgA/IgG, pANCA, anti-OmpC, anti-C-Bir1 and anti-I2) of these subjects were compared with the profiles of 617 subjects with CD and 420 with UC. IBD type and disease location were confirmed using the criteria of the NIDDK IBDGC phenotyping manual and the Montreal Classification. Univariate logistic regression models were applied to assess the association of single serological markers and IBD phenotypes. Multivariate analysis (MVA) was applied to assess the combined serological effects, adjusting for clinical factors such as age and gender. The discriminative value of the MVA model was expressed as area under the curve (AUC).
By univariate analysis, all of the serological markers tested showed significantly different positivity rates across the three groups (CD, UC, IBDU; p < 0.001). Logistic regression MVA models were performed to identify those with CD, UC or IBDU conditioned on disease location. A model containing ASCA, pANCA and anti-C-Bir1 (AUC=0.81) differentiated IBDU from CD. Anti-CBir-1 and ASCA discriminated IBDU from E2 UC (AUC=0.71) and E3 UC (AUC=0.66), respectively. Anti-OmpC or anti-I2 were not significantly useful in any of the MVA models.
The presence of anti-CBir-1 and ASCA and a negative pANCA favors CD over IBDU. Anti-CBir1 positivity has moderate predictive capacity to identify IBDU as compared to UC. The limited ability of anti-OmpC and anti-I2 to discriminate IBDU is confirmed. Validation of anti-CBir-1 as a marker for IBDU in larger cohorts is needed.