Background: Magnetic resonance cholangiopancreatography (MRCP) is a non-invasive and increasingly used procedure in cases involving biliary and pancreatic diseases. However, the accuracy of MRCP in differential diagnosis between pancreatic cancer and chronic pancreatitis has never been documented in a large prospective controlled study.
Methods: 124 patients were recruited for the study, selected from 141 consecutive patients with an average age of 55 years (range 19-80) who presented to our department between February, 1996, and January, 1998, with a strong clinical suspicion of pancreatic cancer. MRCP images were interpreted by a radiologist and a gastroenterologist who were unaware of the clinical diagnosis of patients. The exact diagnosis was based upon histological evidence from biopsy examination (surgical and fine needle biopsy) or a follow-up of at least 12 months.
Findings: Of the 124 patients, 37 (30%) had pancreatic carcinoma; 17 (14%) had other neoplastic pancreatic diseases; 57 (46%) had chronic pancreatitis; 13 (10%) pancreatic ducts were clear. The sensitivity of MRCP with respect to diagnosing pancreatic cancer was 84% and its specificity 97%. The corresponding values for endoscopic retrograde cholangiopancreatography (ERCP) were 70% and 94%, respectively.
Interpretation: MRCP is as sensitive as ERCP when detecting pancreatic carcinomas. Furthermore, it is feasible to presume that the use of MRCP may prevent inappropriate explorations of the pancreatic and common bileducts in cases of suspected pancreatic carcinomas, where interventional endoscopic therapy (ie, palliative common-bileduct drainage) is unlikely.