Background: Accurate nonoperative diagnosis of proximal biliary strictures (PBSs) is often difficult.
Objective: To report our experience with EUS-guided FNA (EUS-FNA) of PBSs following negative or unsuccessful results with brush cytology during ERCP.
Design: Retrospective cohort study.
Setting: Single, tertiary referral hospital in Indianapolis, Indiana.
Patients: Consecutive subjects from January 2001 to November 2004 who underwent EUS-FNA of a PBS documented by ERCP.
Interventions: EUS-FNA of PBS.
Main outcome measures: Performance of EUS-FNA, with the final diagnosis determined by surgical pathology study or the results of EUS-FNA and follow-up.
Results: A total of 291 biliary strictures undergoing EUS were identified. Of these, 26 (9%) had PBSs and 2 were excluded. EUS-FNA was not attempted in 1 because no mass was visualized. The second had a PBS seen on magnetic resonance cholangiopancreatography, but no ERCP was performed. Twenty-four patients (14 men; mean age, 68 years) underwent EUS-FNA of a PBS following ERCP brush cytology studies that were either negative/nondiagnostic (20) or unable to be performed (4). EUS visualized a mass in 23 (96%) patients, including 13 in whom previous imaging detected no lesion. EUS-FNA (median, 4 passes; range, 1-11) demonstrated malignancy in 17 of 24 (71%) patients with findings showing adenocarcinoma (15), lymphoma (2), atypical cytology (3), or benign cells (4). No complications were noted. Pathology results from 8 of 24 (33%) patients who underwent surgery showed hilar cholangiocarcinoma (6), gallbladder cancer (1), and a benign, inflammatory stricture (1). The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of EUS-FNA were 77% (95% confidence interval [CI], 54%-92%), 100% (95% CI, 15%-100%), 100% (95% CI, 83%-100%), 29% (95% CI, 4%-71%), and 79% (95% CI, 58%-93%), respectively.
Limitations: Histopathologic correlation of EUS-FNA findings was limited to 8 of 24 (33%) patients who underwent surgery.
Conclusions: EUS-FNA is a sensitive method for the diagnosis of PBSs following negative results or unsuccessful ERCP brush cytology. The low negative predictive value does not permit reliable exclusion of malignancy following a negative biopsy.