P195 MRI is predictive of, and anti-TNF treatment changes, the clinical course of Crohn's disease strictures

Abstract Background Strictures are the most common Crohn’s disease (CD) complication but their natural history is unknown. There is a need to characterise inflammation and fibrosis, predict prognosis, and understand the impact of drug therapy. Methods Patients with a CD stricture diagnosed over a 5-year period with ≥12-month follow-up were reviewed for their clinical course, response to drug therapy, CRP, need for endoscopic dilatation, hospitalisation and surgery. Magnetic resonance enterography (MRE) scans at time of stricture diagnosis were reviewed blindly for disease extent and inflammation. Magnetic Resonance Index of Activity (MaRIA) score was calculated. Results Characteristics of stricture patients: 136 patients: 77 had 1 and 59 had ≥2 strictures. Median age at stricture diagnosis was 40. Thirty-four per cent had previous CD surgery. Fifty-seven per cent were de novo small bowel strictures, 33% anastomotic, and 10% colonic strictures. At stricture diagnosis, 28% of patients were already on anti-TNF therapy. Treatment: Median follow-up for those not requiring surgery was 41 months (IQR 26–56). Forty-six per cent of patients came to surgery for their stricture after a median of 6 months (IQR 2–11). Clinical and drug predictors of surgery: Hospitalisation due to obstruction predicted surgery (OR 2.7; p = 0.03) while anti-TNF therapy started at stricture diagnosis was associated with a reduced risk of surgery (p = 0.049). MRE predictors of outcome: On multiple logistic regression analysis MRE characteristics associated with increased risk for surgery were proximal bowel dilatation ≥30 mm diameter (OR 3.1; p = 0.005), bowel wall thickness at stricture (OR 2.5 for ≥10 mm; p = 0.01), and stricture length (OR 2.5 for >5 cm; p = 0.01). Eighty-one per cent of patients with all three adverse MRE features required surgery vs. 17% if none were present (p < 0.001; Figure 1). Accuracy for these three MRE variables combined for the prediction of future surgery was high (AUC 0.77). On univariate analysis mesenteric fat inflammation (p = 0.001), stricture bowel wall oedema (p = 0.002), MaRIA score (p < 0.001), and associated fistula (p = 0.02) were significant for surgical risk. Conclusions MRE findings are highly predictive of future surgery. Three simple findings (pre-stricture dilatation, bowel wall thickness, stricture length) are strongly predictive of subsequent surgery. These MRI findings predict future disease course and can identify patients who may benefi