Integrative transcriptomic analysis identifies a novel gene signature to predict prognosis of pancreatic cancer in different subtypes

Background: Recent advances on pancreatic cancer molecular classifications have identified several subtypes with distinct characteristics, treatment response, and prognosis. We aim to identify the consensus gene signature that could predict the prognosis of pancreatic cancer. Methods: Transcriptomic data was acquired from TCGA database. Differentially expressed genes (DEGs) were identified by comparing the Basal-like, Quasi-mesenchymal and Squamous subtype to other subtypes. A new model was constructed by the least absolute shrinkage and selection operator to stratify patients into high and low-risk groups. The prognosis, transcriptomic profiles, and immune infiltration were examined between these groups. Results: We constructed a signature consisting of nine genes, and the GSEA analysis showed that the genomic profile of high-risk tumors is associated with the basal-like and squamous gene set enrichment. Patients with high-risk tumors had worse overall survival (P < 0.001) and progression free survival (P = 0.033), and are associated with a higher expression of KRAS downstream targets such as SDC1, ITGB4 and SLC2A1, which are involved in KRAS mediated macropinocytosis and tumor invasion. Meanwhile, several recurrence-associated genes increased in the high-risk tumors, including ITGA3 and TP63, which have been shown to mediate enhancer-dependent genomic reprogramming towards the squamous phenotype. The tumor immune infiltration profile analysis showed that high-risk tumors are characterized with an immune suppressive microenvironment. Conclusion: The integrative transcriptomic analysis identifies a consensus gene signature that can discriminate pancreatic cancer subtypes and determine patient prognosis by evaluating the genomic reprogramming and the level of immune infiltration profile in pancreatic cancer.