An Accurate Prognostic Model and Treatment for Pancreatic Carcinoma Based on the Tumor Immune Microenvironment

BackgroundPancreatic adenocarcinoma (PAAD) is a highly malignant cancer with a poor prognosis. The tumor microenvironment (TME) is closely related to tumorigenesis, progression, and treatment. However, the relationship between TME immune cell genes and prognosis in PAAD is currently unclear.Methodsn this study, we identified three prognostic subtypes based on the TME by using data from The Cancer Genome Atlas (TCGA) database, The International Cancer Genome Consortium (ICGC) database and University of California Santa Cruz (UCSC) database. The Silhouette plot analysis was used to evaluate 758 immune genes expression in PAAD from each database, then to divide all samples into three subtypes (Clusters A, B, C) by Lasso’s binomial logistic regression. We analyzed the relationship between subtypes and prognosis by the survival R package. CIBERSORT was used for evaluating the expression changes of immune cells. We detect the copy number variation areas between two groups through GISTIC 2.0 algorithm. The TIDE network tool was used to predict the response of immune therapy.ResultsWe defined three clusters (Clusters A, B, and C) based on the analysis of immune gene expression. Cluster B got a worse prognosis than the other two clusters. The Cluster B group had the highest level of Macrophages M0 and Macrophage M2. NK cell resting was much higher in Cluster B than other groups in TME. Gene KRAS was mutated in 77% of all samples. Cluster C had a better immune therapy effect than others.ConclusionsWe found a news model to predicted patients’ prognosis who with pancreatic adenocarcinoma. Cluster B had the significant worse prognosis than other groups. Patients in Cluster C could get batter treatment effect by using immunotherapy.