Bile duct brushings in a pig model: examination of intraobserver variability and variability in specimen quality...

In patients with bile duct malignancy, bile duct brushing is plagued by a low yield diagnosing underlying malignancy. There are few data explaining why this is so. This porcine model was designed to examine three variables: 1) examination of inter-observer variability, 2) variability in specimen quality obtained in sequential animals, and 3) variability between two different brushes (one designed for colon with large bristles, one for duodenum with short bristles).

En bloc resection of liver, bile ducts, duodenum, and pancreas was performed on three 6-mo-old crossbred pigs at the time of commercial slaughter. In each pig, one common hepatic duct and one common bile duct brushing, all performed by the same investigator, were done. Ten identical vigorous passes were done with each brush (long bristle or short bristle) on virgin epithelium. Specimens were graded for cellularity by three cytopathologists who were blinded to the site or brush size. Interobserver variability as well as variability among sequential animals and between the two different brushes was compared.

Interobserver variability among the three cytologists was almost nil. Cellularity obtained using the short brush alone varied from unsatisfactory to high. Cellularity obtained using the long brush alone varied from unsatisfactory to moderate. Variability of cells obtained from one pig to the next ranged from unsatisfactory to high.

1) While interobserver variability was very low, variability in cellularity obtained from one pig to the next, and from one brush to the next, was very high. This sampling variability may partially explain the low yield in malignant cells in human malignant biliary brushing. Multiple brushings in one patient may alleviate part of this problem. 2) There was no advantage to either brush type (large bristle or small bristle).