Background: Vascular invasion (VI) in a patient with pancreatic or periampullary cancers precludes surgery and indicates a poor prognosis. Published data on the accuracy of EUS in diagnosing VI is varied.
Objective: The aim of this meta-analysis was to evaluate the accuracy of EUS in diagnosing VI in patients with pancreatic and periampullary cancers.
Design: Data from EUS studies were pooled according to the Mantel-Haenszel and DerSimonian Laird methods.
Patients: EUS studies in which VI was confirmed by surgery or angiography were selected.
Interventions: EUS.
Main outcome measures: Pooled estimates of sensitivity, specificity, likelihood ratios, and diagnostic odds ratio of EUS.
Results: Data were extracted from 29 studies (N = 1308) that met the inclusion criteria. The pooled sensitivity of EUS in diagnosing VI was 73% (95% CI, 68.8-76.9) and the pooled specificity was 90.2% (95% CI, 87.9-92.2). The positive likelihood ratio for diagnosing VI by EUS was 9.1 (95% CI, 4.6-17.9) and the negative likelihood ratio was 0.3 (95% CI, 0.2-0.5). Diagnostic odds ratio, the odds of having VI in positive as compared with negative EUS studies, was 40.1 (95% CI, 16.1-99.9). The P value for chi(2) heterogeneity for all the pooled estimates was >.05.
Conclusions: Although EUS is the best noninvasive test to diagnose VI in pancreatic and periampullary cancers, this meta-analysis showed that the specificity (90%) is high but the sensitivity (73%) is not as high as suggested. Further refinements in EUS technologies and interpretation may improve the sensitivity for detecting VI.