Relative efficacy of different MRI signs in diagnosing active Crohn's disease, compared against a histological gold standard.

INTRODUCTION: There has been growing interest in the use of MRI in assessing Crohn's disease because of its lack of ionising radiation. Many MRI signs have been described in the literature, but their relative utility is unknown. METHODS: The MRIs of the bowel performed at 1.5 Tesla were blindly reviewed on 26 patients with recent histology (surgery and/or colonoscopy and their associated reports) according to a dedicated pro forma. Each patient's bowel was divided into nine segments. Each segment was assessed as to the presence or absence of 15 MRI signs described in the literature: abnormal gadolinium enhancement (both subjective and objective), wall oedema, fat oedema, fat proliferation, nodal enlargement, free fluid, wall nodularity, serosal blurring, mural thickening >4 mm, stricture, multi-segmental disease, fistula, abscess and layered contrast enhancement. The results were compared against a histological gold standard with a six-point scale of disease severity. RESULTS AND CONCLUSIONS: MRI correctly identified all 15 patients with at least established mucosal disease, and three of eight with only mild mucosal disease. Combining these results, a positive MRI correlated highly with at least early mucosal disease (positive predictive value 95%), while the presence of established mucosal disease was unlikely if MRI was negative (negative predictive value 100%). The MRI signs found to be the most sensitive for detecting active Crohn's disease were those related to the bowel wall, namely, wall thickening, nodularity, contrast enhancement and oedema. The most specific signs were the presence of multi-segmental disease, layered contrast enhancement and complications (fistula and abscess).

Experience of endoscopic ultrasound in an Australian tertiary hospital.

BACKGROUND: There is minimal prospective data on endoscopic ultrasound (EUS) experience in Australia. This study aimed to review the recent EUS experience of a single Australian tertiary hospital. METHODS: Data of all EUS carried out at St Vincent's Hospital, Melbourne, Australia were prospectively collected from August 2002 to September 2005. Patient demographics, indications, anatomical sites of EUS and fine-needle aspiration (FNA) were extracted from the database. A subset of 231 patients were involved in a prospective study to assess the clinical effect of EUS: with accuracy and outcomes data derived from surgical histology and pre-EUS and post-EUS questionnaires. RESULTS: Five hundred and sixty-one procedures were carried out for 526 patients. Men comprised 57% (n = 300). Mean age was 60 years (range, 19-92 years). The indications were (i) oesophageal 30.7% (n = 172); (ii) pancreaticobiliary 29.6% (n = 166); (iii) mediastinal and lung 20.7% (n = 116); (iv) gastric 15.9% (n = 89); and (v) duodenal 3.2% (n = 18). Endoscopic ultrasound-guided FNA was carried out in 30.3% procedures (170/561). Overall EUS accuracy was 84% (55/67 patients) and EUS-guided FNA accuracy was 88% (28/32 patients). There were no major complications. CONCLUSIONS: This is the largest reported Australian experience of EUS to date. Overall EUS was accurate, safe with no major complications or mortality and had useful outcomes. The role, expertise and accessibility of EUS in the Australian clinical setting are anticipated to continue to increase.