P318 Accuracy of magnetic resonance enterography for extraluminal complications of Crohn's disease
Pecoraro G.M.*1, Orlando A.2, TesÈ L.3, Solina G.4, Di Mitri R.5, Mocciaro F.5, Renna S.2, Cottone M.2
1ARNAS Civico-Di Cristina-Benfratelli Hospital, Department of Gastroenterology, Palermo, Italy 2Riuniti Villa Sofia - Cervello Hospital, Department of Internal Medicine, Palermo, Italy 3Riuniti Villa Sofia - Cervello Hospital, Radiology, Palermo, Italy 4Riuniti Villa Sofia - Cervello Hospital, surgical department, Palermo, Italy 5ARNAS Civico-Di Cristina-Benfratelli Hospital, Department of Gastroenterology and Endoscopy, Palermo, Italy
Crohn's disease (CD) can be associated with intraluminal and extraluminal complications like abscess and fistula. More than half of patients with Crohn's disease need surgery to treat these complications at least once in their life. Aiming to make an appropriate choice of intervention, the magnetic resonance enterography (MRE) is largely used to assess the type and severity of these complications. In a previous prospective study pre-operative MRE has been able to predict surgical approach in the most of symptomatic patients, showing low sensitivity and high specificity for the diagnosis of fistula or abscess and high sensitivity and low specificity for stenosis.
We reviewed a sample from a referral center of 42 consecutive symptomatic CD patients addressed to surgery who underwent preoperative MRE. All data were recorded in an electronic database (Excel Microsoft). We analyzed the concordance between MRI and surgical findings for presence of fistulas or abscesses and other stenosis-related characteristics (location, number and extent). We calculated the proportions of VP, FN, FP and VN with 95% confidence limits and the accuracy of the index test (MRE) compared to the reference test (surgery) for the diagnosis of stenosis, fistula or abscess.
Fifty-two percent of patients were male, mean age was 40.5±13.6 years. Forthy-one (98%) were symptomatic. The median time between MRE performance and surgery was 2 months (range 1–24); time was 3 months in 29 pts (69%). A laparoscopic approach was performed in 69% of patients. The prevalence of fistula at surgery was 24% and of abscess 17%. MRE failed in detecting fistulas in 2 patients and falsely identified abscess in one patient. Sensitivity and specificity of MRE were 40% and 99% for fistula and 67% and 94% for abscess, respectively. Accuracy of MRE for fistula was 76% (95% CI 60–88%) and for abscess 88% (95% CI 69–93). For detection of single and multiple stenosis the sensitivity of MRE was 93% and 88% respectively.
The MRE was a fairly accurate imaging test for the characterization of stenosis and for the diagnosis of fistula or abscess. Taking into account limitations due to retrospective data, MRE shows high specificity and low sensitivity for the diagnosis of fistula and abscess according to data from literature. Moreover, it appears more useful to identify the site and extension of stenosis and the single stenosis with respect to the multiple. Aiming to optimize the decision about surgery, there is still a need for development of shared criteria among radiologists and surgeons to further assess the reproducibility of MRE and its accuracy for the diagnosis of complications in prospective studies of patients with CD