Objectives: To determine the efficacy of endoscopic bile duct brush cytology for diagnosis of pancreaticobiliary malignancies and to provide guidelines for interpretation of dysplastic cytology.
Methods: Consecutive endoscopic bile duct brush cytology specimens were classified by an independent cytopathologist as benign, low- or high-grade dysplasia, or cancer. A final diagnosis was established in a blinded fashion by histopathology, radiographic evidence of metastatic disease, or independent clinical follow-up. Sensitivity and specificity were adjusted for dysplastic cytology, and likelihood ratios were determined for each diagnosis and used for calculation of posttest probability of malignancy.
Results: Dysplasia was found in 23% of 168 consecutive bile duct brushings performed in 149 patients. Sensitivity of brush cytology was 37% and specificity 100%; its likelihood ratio for malignancy ranged from 3.4 for high-grade dysplasia, to 1.1 for low-grade dysplasia, to 0.6 for benign. For a patient with a 50% pretest probability of malignancy, finding of high-grade dysplasia changed the posttest probability to 77%, low-grade dysplasia to 52%, and benign to 38%.
Conclusion: Cytological dysplasia occurs frequently, with high-grade dysplasia being strongly suggestive of malignancy. Presented likelihood ratios can be used to calculate the posttest probability of malignancy for any diagnosis.