Malignant strictures of the extrahepatic bile ducts are difficult to distinguish from benign strictures, particularly in patients with primary sclerosing cholangitis. Because attempts at diagnosing small cancers with fine‐needle aspiration biopsy are not possible in the absence of an associated mass lesion and because the sensitivity of exfoliative biliary cytology is controversial, brush cytology has been used as a potential means of establishing a specific diagnosis of bile duct carcinoma. Herein we report our experience with this technique when performed on 65 patients over a 5‐yr period. Each had at least one brushing. Thirty‐seven were found to have bile duct carcinoma and 28 were found to have benign strictures. Of these 37, the first brushing was positive for malignancy in 15 (40%), whereas four (11%) had cells suspected but not diagnostic of malignancy. Thirteen patients with bile duct carcinoma whose initial brushings were negative for malignancy had second brushings. Of these, five (38%) had malignant cells, whereas three (24%) yielded suspicious cells. Three of the eight whose first two brushings were negative for malignancy were found to have malignant cells on the third brushing. In contrast, of the 28 patients with benign strictures, malignant cells were never found. However, in two patients, suspicious cells were reported with the first but not the second brushing. A single negative or suspicious cytological finding decreased the probability of bile duct carcinoma to 43%. Two and three sequential negative tests reduced the probability to 32% and 0%, respectively. When suspicious cytological findings were excluded from the negative results, the probability of having bile duct carcinoma was further reduced to 41%, 20% and 0%, respectively. Using these results, we conclude that (a) a single cytological brushing of a bile duct stricture has a low yield, (b) the sensitivity of the test increases with repeated attempts, (c) the probability of having bile duct carcinoma after three sequential negative cytological brushings is very low (<6%) and (d) these data provide evidence for the usefulness of percutaneous transhepatic cholangiography and bile duct cytological findings in establishing a diagnosis of bile duct cancer. (HEPATOLOGY 1990;12:747–752). Copyright © 1990 American Association for the Study of Liver Diseases
