A robust circulating microRNA signature for diagnosis and early detection in pancreatobiliary cancer

Optimized serum processing condition was evaluated with t-distributed stochastic neighbor embedding (t-SNE) visualization. The serum microRNA candidates were selected in disease association with Weighted Gene Coexpression Network Analysis (WGCNA). A microRNA signature, combination of multiple serum microRNAs, was examined in exploratory, validation and independent validation set. Biology of the diagnostic miRNAs was evaluated using human pancreatic cancer cells. Results: The 284 (healthy 150, PBca 134) of 827 serum samples were processed within 2hr of time-to serum collection, distributed at the same area of t-SNE map, assigned to exploratory set. The 193 optimized samples were assigned to validation (healthy 50, PBca 47) or independent validation set (healthy 50, PBca 46). Index-1 was a combination of the 5 serum microRNAs having disease-association on WGCNA, showed a sensitivity/specificity >80% and an AUC outperforming CA19-9 in exploratory, validation and independent validation set. The AUC of Index-1 for detecting T1 tumor was superior to CA19-9 (0.856 vs. 0.649, p = 0.038). Human pancreatic cancer cells had intra- and extracellular expression of miR-665, a component of Index-1. Transfection of miR-665 to a human pancreatic cancer cell inhibited cell growth. Conclusions: A serum microRNA signature, Index-1, was useful for detecting PBca, which would improve the early diagnosis of PBca.