Strategies for improving diagnostic accuracy of biliary strictures.

Overview

abstract

BACKGROUND: Brush cytology is the initial intervention when evaluating biliary strictures. Biliary brush cytology is known for its low sensitivity (but high specificity) and may be accompanied by biopsies and/or fluorescent in situ hybridization (FISH) to improve diagnostic yield. This study aimed to identify features to enhance cytological sensitivity, and assess which sampling method(s) improve identification of pancreatobiliary adenocarcinomas (PBCa). METHODS: Seventy-three biliary stricture cases were retrieved (38 PBCa and 35 control benign strictures). Biliary brushings, FISH, and biopsies were reviewed. Cytology specimens were evaluated for cellularity and presence of drunken honeycomb (DH), loosely cohesive clusters of round cells (LCCRC), large atypical cells with foamy cytoplasm (LACF), and single vacuolated malignant cells (SCs). Biopsies were examined for the presence of stromal invasion (SI). RESULTS: Biliary brushings were scantly cellular in 47.4% of PBCa and 51.4% of controls, resulting in 69.6% nondiagnostic/false-negative cytology diagnoses. DH, LACF, and SCs were significantly associated with adenocarcinoma (P < .00001, .0033, and .00002, respectively). By univariate analysis, SCs and LACF were predictors of malignancy in brushings (P = .0002 and .05). By multivariate analysis, only SCs were predictive of malignancy (P = .002). SI facilitated the diagnosis in 9 biopsies. Sensitivity/specificity of brush cytology, FISH, and biopsy were 39.5%/94.3%, 63.9%/94.3%, and 84.2%/100%, respectively. CONCLUSIONS: The low sensitivity of biliary brushings results from limited cellularity. Identification of LACF, DH, and SCs improves sensitivity. Sampling of stromal tissue may facilitate PBCa diagnosis. Concurrent biopsies and FISH are helpful in enhancing the diagnostic yield of PBCa.