Objectives: The diagnosis of pancreatic ductal adenocarcinoma (PDAC) by endoscopic ultrasound-guided fine-needle aspiration (EUS-FNA) can be challenging to distinguish tumor cells from benign epithelium (BE). The aim of the present study was to set a minimal antibody panel to differentiate PDAC from contaminated BE in EUS-FNA specimens.
Methods: Immunohistochemistry using claudin 4, EZH2, Ki-67, maspin, p53, and S100P was performed on tissue microarray sections containing 53 PDACs and 33 BE as well as cell blocks of EUS-FNA including 53 PDACs and 22 BE. The positive rate was scored as 0 to 4+. The receiver operating characteristic curve was applied to determine a cutoff point, and the Classification And Regression Trees method was used to obtain a classification tree of the best panel.
Results: The cutoff point was 1+ for claudin 4, EZH2, Ki-67, p53, and S100P and 2+ for maspin. All BE scored 0 for p53. The classification tree revealed using p53, S100P, and claudin 4 was the most powerful. The sensitivity and specificity of the tree were 96.2% and 100% in tissue microarrays and 100% and 95.5% in EUS-FNA, respectively.
Conclusions: The classification tree using p53, S100P, and claudin 4 seems to successfully distinguish PDAC from the accompanying BE.