Background: The aim of this study was to evaluate the usefulness of pancreatic duct brushing for diagnosis of pancreatic carcinoma.
Methods: Brush cytology was attempted in 58 patients suspected of having pancreatic malignancy because of stricture of the main pancreatic duct, confirmed by endoscopic retrograde cholangiopancreatography. Thirty-eight patients were finally diagnosed by an operation or the clinical course as having pancreatic carcinoma, and the remaining 20 patients as having chronic pancreatitis. The usefulness of brush cytology for diagnosis of pancreatic carcinoma was estimated. We interpreted failures of pancreatic duct brushing to be false negatives when the lesion was malignant.
Results: In 48 of 58 patients (82.8%), brushing was successfully performed and satisfactory specimens were obtained. Brush cytology was positive in 25 of 38 patients with pancreatic carcinoma (sensitivity 65.8%) and negative in all patients without malignancy (specificity 100%). Overall accuracy was 76.4%. During 2001-2005, the number of back-and-forth motions of the brush was increased to more than 30 times. The sensitivity significantly improved from 43.8% in 1997-2000 to 81.8% in 2001-2005 (P < 0.05). The increased success rate of brushing by improvement of skill in manipulating the guidewire and increased number of cells smeared on glass slides by increased back-and-forth motion of the brush may account for this improvement over time. Moreover, the sensitivity in 2001-2005 was 85.7% if failures of brushing with pancreatic carcinoma are excluded. No major complications occurred, except for two patients with a moderate grade of acute pancreatitis.
Conclusions: Although further studies with a large number of patients are needed, our results suggest that with recent improvements of the brushing technique, pancreatic duct brushing is a useful and safe method for the differential diagnosis of malignancy from benign diseases of the pancreas.