Endoscopic ultrasound: a meta-analysis of test performance in suspected biliary obstruction

Endoscopic ultrasound (EUS) achieves high-resolution images of the bile duct and pancreas, while avoiding the risks of ERCP (endoscopic retrograde cholangiopancreatography). It appears comparable to MRCP (magnetic resonance cholangiopancreatography), although its use is less widely disseminated. We aimed to summarize EUS test performance in suspected biliary disease with meta-analysis.

MEDLINE search (January 1987-September 2006), selected reference lists, external experts, and manual search of abstracts were used. Studies permitting (re)construction of 2 x 2 tables for EUS versus a gold standard were used. Random-effects models were used to estimate pooled sensitivity and specificity after adjusting for a number of potential confounders. Summary receiver operating characteristic analysis, with the sensitivity corresponding to the point on the receiver operating characteristic curve where sensitivity equals specificity (Q*) and area under the curve, was performed. The effects of sample size, quality, disease prevalence and spectrum, pancreatitis, echoendoscope type, and EUS era on diagnostic performance were assessed. Performance regarding presence of obstruction, choledocholithiasis, and malignancy was analyzed.

Thirty-six eligible, non-overlapping studies met inclusion criteria (3532 subjects). EUS had a high overall pooled sensitivity (88%; 95% confidence interval, 85%-91%) and specificity (90%; 87%-93%) for biliary obstruction (area under the curve = 0.97; Q* = 0.92). EUS had higher sensitivity (89%; 87%-91%) and specificity (94%; 91%-96%) for choledocholithiasis than for malignancy (sensitivity, 78%; 69%-85%; specificity, 84%; 78%-91%). Smaller studies and ones mainly studying patients with suspected strictures were associated with lower test performance.

There is excellent overall accuracy for EUS in diagnosing choledocholithiasis, with less impressive results for malignancy (when fine-needle aspiration is not used).