Pancreatic adenocarcinoma: regression analysis to identify improved cytologic criteria

Abstract

The incidence of pancreatic adenocarcinoma is increasing and it is usually unresectable at the time of diagnosis. Consequently, fine-needle aspiration biopsy (FNAB) is being used more frequently for diagnosis. The reported sensitivity of diagnosing pancreatic adenocarcinoma by FNAB has varied between 50% and 100%. In an attempt to increase the diagnostic sensitivity, we retrospectively reviewed a series of pancreatic FNABs. Fifteen cytologic criteria were evaluated in 78 patients who had pancreatic FNABs. Of these patients, 49 had primary adenocarcinomas and 29 had benign, non-neoplastic lesions. Using a stepwise logistic regression analysis we identified three key cytologic criteria for this diagnosis. Our study identified anisonucleosis (P = 0.001), large nuclei (P = .007), and nuclear molding (P = .03) as the significant cytologic features for diagnosing pancreatic adenocarcinoma. In combination, these three criteria had a sensitivity of 98% and a specificity of 100%.