Background: Techniques to confirm suspected pancreaticobiliary (PB) malignancy when index sampling is non-diagnostic include cholangiopancreatoscopy (CP) and endoscopic ultrasound (EUS). However, comparative data are lacking.
Aim: The purpose of this study was to compare the yield of EUS and CP for the diagnosis of PB pathology.
Methods: Consecutive patients with indeterminate PB pathology who underwent both CP and EUS within 3 months of each other were retrospectively identified. For CP, tissue sampling included biopsy under direct inspection (cholangioscopy-directed biopsy), biopsy following CP with fluoroscopic guidance (cholangioscopy-assisted biopsy), or brush cytology. For EUS-FNA, lesions included ductal strictures or hypoechoic masses. A comparison of operating characteristics between CP and EUS utilizing tissue confirmation or 12-month clinical course consistent with either benign or malignant disease was performed.
Results: Between February 2000 and June 2007, 66 (33 males, 33 females, median age 64.5) patients with indeterminate PB pathology who had undergone both CP and EUS within 3 months of each other were included. Lesions amenable to sampling were noted in 59 CP and 50 EUS patients. On follow-up, 39 patients had neoplasia and 27 were benign. The sensitivity/specificity for the diagnosis of neoplasia for CP and EUS was 48.7/96.3 % and 33.3/96.3 %, respectively (comparison of sensitivities, P = 0.183). The combined (CP and EUS) sensitivity/specificity was 66.7/96.3 % (P = 0.0064 and P = 0.0001 comparing combined sensitivity vs. sensitivity of either CP alone or EUS alone, respectively).
Conclusions: In patients who undergo both EUS and CP for indeterminate PB pathology, the combined yield of EUS and CP to detect neoplasia appears to be higher than either examination alone.