P146. Small bowel Crohn's disease: magnetic resonance can predict with a high accuracy the presence of stenotic lesion. Results of a prospective study
S. Pous1, M. Frasson1, M. Aguas1, G. Sanchez1, G. Bastida1, C. Ramirez1, M. Llavador1, E. Garcia-Granero1, P. Nos1, 1La Fe Universitary Hospital, Valencia, Spain
Objective of this study is to assess the accuracy of the magnetic resonance (MRI) in the prediction of the number and features (inflammatory vs. fibrotic) of small-bowel lesions compared to intraoperative and pathological findings in patients with Crohn's disease.
This is a prospective study including all the patients with small bowel Crohn's disease operated on at our Specialized Colorectal Unit during the last 18 months. MRI was realized at maximum one month before surgery and the following MRI variables were considered for each lesion: wall thickness, postcontrast wall signal intensity (layered vs. homogeneous contrast enhancement), relative contrast enhancement, MaRIA score. During the surgery the small bowel was explored with calibration sphere to exclude the presence of hidden stenosis. MRI characteristics were compared to surgical exploration data (number and localization of the lesions) and pathological features (inflammation and fibrosis).
28 patients were included in the study and 66 small bowel lesions were identified. MRI accuracy for small bowel lesions identification was 87.1% with a 70% sensibility and 94% specificity (negative predictive value 88%, positive predictive value 82%). MRI accuracy for fistulas and abscess detection was 87.5% and 98.2%, respectively. MRI postcontrast wall signal intensity was not useful to determine the pathological features of the lesions (accuracy 63%), while the Relative Contrast Enhancement curve allowed to predict the presence of fibrosis or inflammation in the lesion with a 70% and 97% accuracy, respectively. MaRia score was a reliable tool to predict moderate (cut-off >16 and <20, area under the curve at ROC curve analysis 0.76, sensibility 70%, specificity 92%, accuracy 76%) and severe inflammation (cut-off >20, area under the curve at ROC curve analysis 0.74, sensibility 71%, specificity 80%, accuracy 76%).
MRI can predict the presence of stenotic small bowel lesions with a high accuracy in Crohn's disease. Pathological features can be predicted by MRI Relative Contrast Enhancement curve but not by MRI postcontrast wall signal intensity. MaRIA is a reliable tool to detect active inflammation in the lesions.