Prospective randomized controlled study comparing cell block method and conventional smear method for pancreatic juice cytology.

To elucidate the diagnostic efficacy of the cell block (CB) method by comparing it with that of conventional smear cytology for pancreatic juice obtained by endoscopic retrograde cholangiopancreatography (ERCP) in a randomized controlled trial fashion. A total of 170 patients with pancreatic lesions suspicious of being malignant who underwent pancreatic juice collection without giving secretin under ERCP were enrolled in this study. After sampling, the pancreatic juice was randomized to the CB method (n = 85) or to smear cytology (n = 85). CB sections were subjected to hematoxylin-eosin, periodic acid Schiff-Alcian blue, and immunohistochemical stains. Both Papanicolaou stain and Giemsa stain were used for smear cytology. The final diagnosis was malignancy in 54 patients: pancreatic cancer, 45; intraductal papillary-mucinous carcinoma, six; and endocrine tumor, three. The number of patients with a cytological borderline malignancy in the CB group (3.5%) was significantly smaller than that in the smear group (27.1%) (P < 0.001). The diagnostic accuracy of the CB method and that of smear cytology were 76.5% (65/85) and 74.1% (63/85), respectively (P = 0.72), and their respective sensitivities were 50% (14/28) and 38.5% (10/26) (P = 0.39). The sensitivity of the CB method (88.9%) was better than that of smear cytology (42.9%) for invasive ductal carcinoma in the pancreas head (P = 0.048). The CB method using immunostaining for pancreatic juice cytology showed a much lower rate of equivocal borderline malignancy and a tendency for a higher diagnostic yield compared with smear cytology. Its diagnostic sensitivity, however, was not satisfactory except for pancreatic-head cancer.