Purpose: To evaluate the usefulness of magnetic resonance cholangiopancreatography (MRCP) in patients referred for assessment of the cause of elevated liver enzymes.
Methods: The MRCP scans of 170 patients between 2009 and 2011 with elevated liver enzyme levels were blindly and independently rereviewed by two experienced radiologists. Biochemical data were collected from the medical records. Receiver operating characteristics curve analyses were carried out. Cutoff levels of the enzymes to predict pathological MRCPs were determined by the area under the curve (AUC). Sensitivity, specificity, negative predictive value, and positive predictive value were calculated.
Results: MRCP scans of 134 patients were identically diagnosed by the two reviewers as nonpathological and 22 as pathological and 14 as uncertain. The agreement between the two reviewers was κ=0.62, indicating good agreement. The percentage of pathological MRCP performed because of elevated liver enzymes was only 14%. For patients with inflammatory bowel disease, the frequency was 36%. On receiver operating characteristics curve analysis, Az values were high for alkaline phosphatase 160 U/l (AUC=0.725) and γ-glutamyl transferase 270 U/l (AUC=0.617). Alkaline phosphatase and γ-glutamyl transferase had a high negative, but low positive predictive value for distinguishing pathological from normal scans.
Conclusion: MRCP does not contribute markedly toward the evaluation of the cause of elevated enzyme levels, except in patients with inflammatory bowel disease. These findings can provide practical guidelines for evaluation of patients with abnormal liver enzymes and for alleviation of the financial burden from health providers.