Performance characteristics of endoscopic ultrasound in the staging of pancreatic cancer: a meta-analysis

Abstract

Context: The optimal approach to pre-operative imaging assessment of pancreatic cancer is unknown.

Objective: The aim of this meta-analysis was to assess accuracy and performance characteristics of EUS in determining nodal staging, vascular invasion, and prediction of resectability of pancreatic cancer. A secondary aim was to perform head to head comparison of performance characteristics between EUS and CT for nodal staging, vascular invasion and resectability.

Design: Data from EUS studies were pooled according to bivariate generalized random effects model. Pooled estimates for CT were obtained from studies which performed head to head comparison between EUS and CT.

Patients: Patients with pancreatic cancer undergoing pre-operative imaging assessment.

Intervention: EUS.

Main outcome measure: Pooled sensitivity, specificity, positive and negative predictive values of EUS for nodal staging, vascular invasion and resectability.

Results: Forty-nine studies were considered of which 29 met inclusion criteria with a total of 1,330 patients. Pooled summary estimates for EUS-nodal staging were 69% for sensitivity and 81% for specificity. For vascular invasion, sensitivity was 85% and specificity was 91%. The sensitivity and specificity for resectability was 90% and 86%, respectively. CT scan showed lower sensitivity than EUS for nodal staging (24% vs. 58%) and vascular invasion (58% vs. 86%); however, the specificities for nodal staging (88% vs. 85%) and vascular invasion (95% vs. 93%) were comparable in studies where both imaging techniques were performed. The sensitivity and specificity of CT in determining resectability (90% and 69%) was similar to that of EUS (87% and 89%).

Conclusions: EUS is an accurate pre-operative tool in the assessment of nodal staging, vascular invasion and resectability in patients with pancreatic cancer.

Conflict of interest statement

Conflict of interest The authors do not have any conflict of interest to report

Figures

Figure 1

Flowchart of studies considered for inclusion in the meta-analysis.

Figure 2

Forest plot of sensitivity and specificity for endoscopic ultrasound (EUS). NS: nodal staging; VI: vascular invasion; R: resectability

Figure 3

Deek’s funnel plot to assess for small study effects. This is a method used for detecting small study effects based on a linear regression of log odds ratios on inverse root of effective sample sizes. When no small study effect exists, the test of the non-zero slope coeffcient has P value >0.1, as was shown in our study.

Figure 4

SROC in studies with both CT and EUS in assessing: a. nodal staging (NS); b. vascular invasion (VI); and c. resectability (R).