The diagnostic accuracy of endoscopic ultrasound in suspected biliary obstruction and its impact on endoscopic retrograde cholangiopancreatography burden in real clinical practice: A consecutive analysis

摘要

Background and Aims: Performing endoscopic ultrasound (EUS) before endoscopic retrograde cholangiopancreatography (ERCP) has been described to be useful in cases of suspected biliary obstruction where EUS can triage patients for ERCP. We aimed to determine the diagnostic accuracy of EUS and its impact on ERCP burden in real clinical practice. We also evaluated the safety and efficacy of EUS+ERCP in a single endoscopic session. Patients and Methods: Four hundred and eighteen consecutive patients with suspected but unexplained biliary obstruction referred for EUS before possible ERCP were evaluated. The diagnostic accuracy of EUS and its value in predicting the need for ERCP were determined. EUS established whether pancreaticobiliary disorder (PBD) was present and whether therapeutic ERCP was required. These decisions were matched with ERCP findings, histology, clinical course, and follow-up. Where ERCP was indicated, it was performed in the same endoscopic session. Results: EUS was performed in 412/418 patients (feasibility 98.5%), and ERCP was considered necessary in 64% (ERCP avoided in 36%). The single-session EUS and ERCP was safe and effective (264 patients). The diagnostic accuracy of EUS was as follows: choledocholithiasis 99%, malignant strictures 90%, and benign strictures 92%. EUS showed pathology in 42% of patients who had a nondilated biliary system at initial investigations. When EUS indicated a normal common bile duct (n=119), this had a 100% positive predictive value for non-necessity for ERCP. The median overall follow-up period was 12 months (range 6-34 months). Conclusion: EUS demonstrated high diagnostic accuracy in this mixed group of PBD. This accurately guided ERCP need and avoided unnecessary ERCP in 36%. EUS and ERCP in the same endoscopic session for the evaluation and management of PBD is technically feasible, with safety and efficacy profiles equivalent to that of each procedure performed independently in different sessions.