[MRCP in primary sclerosing cholangitis]

Evaluation of MR-cholangiopancreaticography (MRCP) for the diagnosis of primary sclerosing cholangitis (PSC) in correlation with endoscopic retrograde cholangiopancreaticography (ERCP) and in comparison to the diagnostic accuracy of various T2-weighted sequences.

Fifty-five patients (34 males, 21 females; mean age 40 years, range 16 to 65 years) with suspected PSC were examined in a 1.5 T MR unit (Magnetom Vision, Siemens, Erlangen), using breath-hold transverse and coronal HASTE, paracoronal RARE and thin-sliced HASTE (TS-HASTE) sequences. Applying a five-point-scale, two blinded investigators assessed the image quality for ROC analysis. Morphologic criteria of PSC were documented and correlated with ERCP, which served as the gold standard, and sensitivity, specificity and diagnostic accuracy were calculated.

PSC was confirmed in 40 of 55 patients (ERCP 55 of 55, liver biopsy 37 of 55), with concomitant chronic ulcerative colitis in 27 and Crohn's disease in 6 of the 40 patients. Qualitative analysis of the image quality showed no significant difference between RARE, HASTE and thin-sliced HASTE sequences (3.4/3.5/3.2). The RARE sequence had the highest sensitivity (97 %), specificity (64 %) and accuracy (84 %) for the detection of PSC. The difference between HASTE and thin-sliced HASTE was statistically significant (p < 0.01). Of the 40 patients with confirmed PSC, 29 were followed by MRI and 3 underwent a liver transplantation within the follow-up period. A Klatskin tumor, which was misdiagnosed by MRCP, was diagnosed by brush biopsy in 1 of the 40 patients. Interobserver variability was adequate to good (kappa 0.4 to 0.7), depending on the chosen sequence.

MRCP seems to be a reliable non-invasive imaging method to diagnose and follow PSC. The RARE sequence showed the highest diagnostic accuracy of the T2-weighted sequences.