MR imaging evaluation of the activity of Crohn's disease

OBJECTIVE. The purpose of this study was to evaluate the sensitivity and specificity of MR imaging in assessing the activity of Crohn's disease. SUBJECTS AND METHODS: Thirty symptomatic patients with Crohn's disease but uncertain disease activity were prospectively examined using MR imaging. Twenty-nine patients were scored using the Crohn's disease activity index. Six hundred milliliters of water orally and 1 mg of glucagon intramuscularly were given before imaging. Breath-hold images were obtained using T2-weighted turbo spin-echo, T1-weighted fast low-angle shot, and fat-suppressed gadolinium-enhanced T1-weighted fast low-angle shot sequences. Images were assessed by two radiologists who were unaware of the patient's symptoms, clinical scoring, and other imaging tests, and who reached a consensus about the imaging findings (bowel wall thickening, bowel wall enhancement, and perienteric changes) and determined the absence or presence of active disease in each patient. MR imaging findings were correlated with endoscopy and surgery. RESULTS. Twenty-three patients had active disease and seven patients had inactive disease. One hundred twenty-four of a total of 168 bowel segments were examined with both MR imaging and endoscopy or surgery. On a per patient basis, MR imaging had an overall sensitivity of 91% and a specificity of 71% for active disease. The Crohn's disease activity index had a sensitivity of 92% and a specificity of 28%. On a per segment basis, MR imaging had a sensitivity of 59% and a specificity of 93%. Bowel wall thickening of greater than 4 mm, bowel wall enhancement (ratio of signals intensity of abnormal to normal bowel > 1.3:1), and increased mesenteric vascularity were useful in identifying active disease. A layered enhancement pattern after the IV administration of gadolinium was highly specific for active inflammation. CONCLUSION. MR imaging is useful in assessing the activity of Crohn's disease and may be helpful when clinical scoring is equivocal.