Objective: To evaluate the accuracy of digital image analysis (DIA) for distinguishing between benign and malignant strictures of the biliary tract.
Patients and methods: Our pathology databank was used to identify all biliary brush cytology specimens obtained during endoscopic retrograde cholangiopancreatography between June 1997 and June 1999. Corresponding medical records were reviewed to determine whether patients had benign or malignant strictures. Strictures were further classified into benign strictures with negative routine cytology, malignant strictures with negative routine cytology, and malignant strictures with positive routine cytology. Papanicolaou-stained smears of available brush cytology specimens were destained and then restained with Feulgen dye. Nuclear images were quantified for DNA content without knowledge of stricture type. DNA histograms were generated and ploidy results compared with the class of stricture.
Results: We analyzed 27 specimens from 69 confirmed benign or malignant strictures. Assuming that the presence of any aneuploid cells indicated malignancy, the sensitivity of DIA was 85%. Furthermore, aneuploid cells were detected by DIA in 13 of 16 specimens in which routine cytology was unrevealing.
Conclusion: Ploidy assessment by DIA has potential to enhance the sensitivity of diagnosing malignant strictures compared with routine cytology alone.