Diagnosis of primary sclerosing cholangitis: a blinded comparative study using magnetic resonance cholangiography and endoscopic retrograde cholangiography

Background: We hypothesized that magnetic resonance cholangiography (MRC) may have less accuracy for the diagnosis and the assessment of the severity of primary sclerosing cholangitis (PSC) than endoscopic retrograde cholangiography (ERC). Objective: The aim of this study was to determine the diagnostic accuracy and interobserver agreement of both ERC and MRC in PSC. Design: A case-control study Setting: University Hospital. Patients: ERCs and MRCs of 36 patients with PSC and 51 controls (normal/other biliary tract diseases) were read in an independent, blinded, and random fashion by 2 magnetic resonance radiologists and 2 interventional endoscopists by using a previously validated classification system. Readers had no access to clinical history, laboratory results, or patient mix. Results: Extrahepatic ductal (EHD) and intrahepatic ductal (IHD) visualization was excellent in 64% of 66% of MRCs and 86% of 74% of ERCs. Sensitivity and specificity for diagnosis of PSC for readers 1 to 4 were 91% and 85%, 88% and 90%, 81% and 96%, and 83% and 96%. respectively. Receiver operating curve values were excellent for all readers (all > 0.9). Interobserver agreement (kappa statistics) for the diagnosis of PSC (MRC, 0.83; ERC, 0.73) and for identifying the presence of IHD strictures (MRC, 0.64; ERC, 0.86) was good for both modalities, but only ERC (ERC, 0.55; MRC 0.36) was good for the presence and the severity of EHD strictures. When assessment of disease severity was limited to the 36 patients with PSC, interobserver agreement was very poor for both MRC (0.23 and 0.07 for EHD and IHD, respectively) and ERC (0.24 and 0.34 for EHD and IHD, respectively). Limitations: The retrospective case-control study made it difficult to assess the impact of the diagnosis on patient management. Conclusions: ERC and MRC were comparable for diagnosing PSC, with very good interobserver agreement for the diagnosis of PSC and IHD strictures. Only ERC had good agreement for EHD strictures. Interobserver agreement was very poor for both MRC and ERC when disease severity of PSC was assessed.