Effectiveness of a new long cytology brush in the evaluation of malignant biliary obstruction: a prospective study.

Background: Cancer detection rates with biliary brush sampling remain disappointingly low. A low cellular yield is often the limiting factor in making a diagnosis of malignancy. The new Cytolong brush (Cook Endoscopy, Winston-Salem, NC) is 3 mm in diameter, 5 cm long, with stiffer bristles oriented at 45 degrees on a 7F sheath. We hypothesized that this new brush might improve cancer detection rates by increasing cellular yield.

Methods: Patients found to have a biliary stricture suspicious for neoplasia on ERCP were randomized to undergo brush sampling for cytology with a standard Geenen brush (GB; Cook Endoscopy, Winston-Salem, NC) [3 mm in diameter, 1.5 cm long, bristles oriented at 90 degrees on a 6F sheath] or the Cytolong brush (CB). Repeat sampling was then performed with the other brush. Stricture dilation was not performed prior to brushing. Specimen results were considered normal, atypical (considered benign), highly atypical (suspicious for cancer), or malignant. All specimens were assigned a cellularity score (0 to 3, insufficient to excellent). Final diagnosis was based on cytologic results plus surgery, EUS, autopsy, or clinical follow-up.

Results: From November 2001 to July 2003, 102 patients had specimens obtained from 94 malignancies (47% pancreatic cancer). The cancer detection rate was 25 of 94 (27%) using CB and 28 of 94 (30%) with GB (p = NS). No patient had positive cytology results with CB and negative cytology results with GB. The yield of the two brushes combined was 28 of 94 (30%). Cancer detection rates of 28% (18 of 64) and 31% (20 of 64) were found for CB and GB, respectively, in distal biliary strictures, and 23% (7 of 30) and 27% (8 of 30) in proximal strictures (p = NS). Insufficient or limited cellularity was seen less frequently with CB (11 of 98) than with GB (17 of 98), and the mean cellular yield was greater with CB than GB (2.6 vs 2.4, p = 0.006).

Summary: Despite improved cellularity, cancer detection rates were not improved by using the larger Cytolong brush in this study. There was no statistical difference between the brushes in both proximal and distal biliary strictures.

Conclusions: The yield of biliary brush cytology at ERCP remains low. Increasing brush size and bristle stiffness does not increase detection rates. Newer devices and processing techniques are required to allow detection rates to approach those attained in other GI tract malignancies.