Endoscopic Ultrasound: A Meta-Analysis of Test Performance in Suspected Biliary Disease

Purpose: EUS achieves high-resolution images of the bile duct (CBD) and pancreas. It avoids the risks of ERCP, and may be comparable to MRCP, although its use is less widely disseminated for biliary disease. We aimed to summarize EUS test performance in suspected biliary disease with a meta-analysis.

Methods: Studies were identified via Medline search (Jan 88-Dec 04) and citation indexing. Included studies permitted (re)construction of 2×2 tables for EUS vs. a gold standard (direct CBD opacification and/or adequate clinical follow-up). Two independent observers graded study quality (blinding, consecutive enrollment, single gold standard, and use of direct cholangiography). The effects of disease prevalence and sample size, clinical context (obstructive jaundice, pancreatitis, etc), radial vs. linear echoendoscope, and EUS “era“ (publication year) on diagnostic performance were assessed. Random-effects models were used to estimate, Q*, after adjusting for sample size, prevalence, EUS era, and study quality, as well as pooled sensitivity (SN) and specificity (SP). Summary receiver-operating characteristic analysis was performed, and area under the curve (AUC) calculated. Performance for detecting presence of obstruction, CBD stones and malignant obstruction were analyzed. Influence of FNA has not been assessed.

Results: 34 eligible, non-overlapping studies met inclusion criteria (3,447 patients). EUS had high overall pooled SN (89%; [95% CI: 86–91%]) and SP (91% [87–94%]) for biliary obstruction (weighted AUC=0.97). EUS had higher SN (89% [86–91%]) and SP (94% [91–96%]) for CBD stones than for malignancy (SN 81% [75–86%], SP 85% [80–91%]). Confounding factors were not significant.

Conclusion: Like MRCP, there is excellent overall accuracy for EUS in diagnosing CBD stones, with less-impressive results for malignancy when FNA is not used; the latter will improve when FNA is considered.