Magnetic Resonance Cholangiopancreatography

A Meta-Analysis of Test Performance in Suspected Biliary Disease

7 October 2003 | Volume 139 Issue 7 | Pages 547-557

Background: Magnetic resonance cholangiopancreatography (MRCP) is one of many newer noninvasive tests that can image the biliary tree.

Purpose: To precisely estimate the overall sensitivity and specificity of MRCP in suspected biliary obstruction and to evaluate clinically important subgroups.

Data Sources: MEDLINE search (January 1987 to March 2003) for studies in English or French, bibliographies, and subject matter experts.

Study Selection: Studies were included if they allowed construction of 2x2 contingency tables of MRCP compared with a reasonable gold standard for at least 1 of the following: the presence, level, or cause of biliary obstruction.

Data Extraction: Two independent observers graded study quality, which included consecutive enrollment, blinding, use of a single (versus composite) gold standard, and nonselective use of the gold standard. Logistic regression was used to examine the influence of publication year, quality score, proportion of patients having a "direct" gold standard, and clinical context on diagnostic performance.

Data Synthesis: Of 498 studies identified, 67 were included (4711 patients). Mixed-effect models were used to estimate the sensitivity and specificity, and quantitative receiver-operating characteristic analysis was performed. Magnetic resonance cholangiopancreatography had a high overall pooled sensitivity (95% [±1.96 SD: spread of SD, 75% to 99%]) and specificity (97% [spread of SD, 86% to 99%]). The procedure was less sensitive for stones (92%; odds ratio, 0.51 [CI, 0.35 to 0.75]) and malignant conditions (88%; odds ratio, 0.28 [CI, 0.18 to 0.44]) than for the presence of obstruction. In addition, diagnostic performance was higher in studies that were larger, did not use consecutive enrollment, and did not use gold standard assessment for some patients.

Conclusions: Magentic resonance cholangiopancreatography is a noninvasive imaging test with excellent overall sensitivity and specificity for demonstrating the level and presence of biliary obstruction; however, it seems less sensitive for detecting stones or differentiating malignant from benign obstruction.

Editors' Notes Context Noninvasive imaging of the biliary tree with magnetic resonance cholangiopancreatography (MRCP) may detect certain abnormalities better than others. Contribution This meta-analysis summarized 67 studies that compared MRCP with other gold standard tests for diagnosing biliary obstruction (for example, endoscopic retrograde cholangiopancreatography, endoscopic ultrasonography, and intraoperative or intravenous cholangiography). Magnetic resonance cholangiopancreatography compared well with most other tests, but it seemed slightly less sensitive for diagnosing stones (92%) and for differentiating benign from malignant obstruction (85%) than it was for identifying the presence (99%) and level (96%) of biliary obstruction. Cautions Study quality varied, and some sensitivity estimates had wide confidence intervals. –The Editors

 

Author and Article Information

From the University of Calgary, Calgary, Alberta, Canada; and McGill University, Montreal, Quebec, Canada.

Acknowledgments: The authors thank Dr. Patrice Bret, Radiologist-in-Chief at the Mount Sinai Hospital and the University Health Network and Professor of Radiology at the University of Toronto, and Dr. Matthew Barish, Director of 3D & Image Processing Center and Assistant Professor of Radiology at Harvard Medical School, for their thoughtful review of our bibliographic search and help in identifying additional published or unpublished studies.

Grant Support: Dr. Romagnuolo is a population health investigator funded by the Alberta Heritage Foundation for Medical Research. Dr. Bardou was funded by the Tomlinson (McGill University) Fellowship. Dr. Rahme was funded by the Arthritis Society (Toronto, Ontario, Canada). Dr. Joseph was funded by Canadian Institutes for Health Research. Dr. Reinhold is the Medical Director, Oncology Division, Synarc Inc., San Francisco, California. Dr. Barkun is a research scholar funded by the Fonds de la Recherche en Santé du Québec.

Potential Financial Conflicts of Interest: None disclosed.

Requests for Single Reprints: Joseph Romagnuolo, MD, FRCPC, MSc (Epid), University of Calgary Medical Clinic, 3350 Hospital Drive NW, G-179, Calgary, Alberta T2N 4N1, Canada; e-mail, j.romagnuolo{at}ucalgary.ca.

Current Author Addresses: Dr. Romagnuolo: University of Calgary Medical Clinic, 3350 Hospital Drive NW, G-179, Calgary, Alberta T2N 4N1, Canada.

Dr. Bardou: LPPCE, Faculty of Medicine, 7, bd Jeanne d'Arc, BP87900, 21079 Dijon Cedex, France.

Dr. Rahme: Research Institute, McGill University Health Centre, Division of Clinical Epidemiology, Montreal General Hospital, 1650 Cedar Avenue, Room L10-408, Montreal, Quebec H3G 1A4, Canada.

Dr. Joseph: Department of Epidemiology and Biostatistics, McGill University and Division of Clinical Epidemiology, Montreal General Hospital, 1650 Cedar Avenue, Room L10 509, Montreal, Quebec H3G 1A4, Canada.

Dr. Reinhold: McGill University Health Centre, Montreal General Hospital Site, Department of Radiology, 1650 Cedar Avenue, Montreal, Quebec H3G 1A4, Canada.

Dr. Barkun: Montreal General Hospital, 1650 Cedar Avenue, D7-148, Montreal, Quebec H3G 1A4, Canada.

Author Contributions: Conception and design: J. Romagnuolo, M. Bardou, E. Rahme, A. Barkun.

Analysis and interpretation of the data: J. Romagnuolo, M. Bardou, E. Rahme, C. Reinhold, A. Barkun.

Drafting of the article: J. Romagnuolo, M. Bardou, A. Barkun.

Critical revision of the article for important intellectual content: J. Romagnuolo, M. Bardou, E. Rahme, L. Joseph, C. Reinhold, A. Barkun.

Final approval of the article: J. Romagnuolo, M. Bardou, E. Rahme, L. Joseph, C. Reinhold, A. Barkun.

Provision of study materials or patients: J. Romagnuolo, M. Bardou, C. Reinhold.

Statistical expertise: J. Romagnuolo, E. Rahme, L. Joseph.

Obtaining of funding: A. Barkun.

Administrative, technical, or logistic support: J. Romagnuolo, M. Bardou, A. Barkun.

Collection and assembly of data: J. Romagnuolo, M. Bardou, C. Reinhold.