Context: Endoscopic brush cytology is a valuable technique for the diagnosis of pancreatobiliary malignancy. Despite its widespread use, the sensitivity of this method has been reported as approximately 50%. The specificity is usually higher than 95%. Few reports have systematically analyzed the reasons for this relatively low sensitivity.
Objectives: To determine the rate and reasons for false-negative diagnoses in endoscopic brushing cytology of biliary and pancreatic ducts based on the results of sensitivity, specificity, accuracy, and positive and negative predictive values.
Design: Retrospective analysis of laboratory data and slide review of false-negative cases.
Setting: Two tertiary care state university hospitals.
Patients: A total of 183 pancreatobiliary brushing specimens obtained from patients undergoing endoscopic retrograde cholangiopancreatography for biliary or pancreatic duct disease for a 4- to 5-year period.
Intervention: Endoscopic retrograde cholangiopancreatography brushings.
Main outcome measures: Determination of sensitivity, specificity, accuracy, and positive and negative predictive values. Analysis of false-negative results.
Results: The sensitivity, specificity, accuracy, and positive and negative predictive values, overall, were 48%, 98%, 79%, 92%, and 76%, respectively. Sampling error was a major cause of false-negative diagnoses (67%), followed by interpretive (17%) and technical errors (17%).
Conclusions: Improvements in sensitivity and diagnostic accuracy for cancer of the pancreatobiliary tract can be achieved by optimizing slide preparatory techniques. Also, enhancement of the cytologist's diagnostic skills enables the identification of the morphologic features of premalignant lesions. Repeat brushings are indicated for suspicious or negative results not consistent with the clinical or radiologic findings.