Lymph node metastasis in pancreatic ductal adenocarcinoma (PDAC) is shown to be related with poor prognosis. To construct an immune-related gene prognostic risk model for PDAC and clarify the molecular and immune characteristics and the benefit of immune checkpoint inhibitor (ICI) therapy in prognostic risk model-defined subgroups of PDAC, we analyze the association between the density of immune cell infiltration and lymph node metastatic status and further study the potential role of immune cells, immune cell–related genes, and immunotherapy outcomes in PDAC patients using bioinformatics models and machine learning methods. Based on The Cancer Genome Atlas (TCGA), PACA-AU and PACA-CA data sets, 62 immune-related hub genes were identified by weighted gene co-expression network analysis (WGCNA). Four genes were selected to construct a molecular subtype identification based on the type 1 T helper cell–related hub genes by using the Cox regression method. We found that lower type 1 T helper cell abundance was correlated with prolonged survival in PDAC patients. Further, prognostic risk score model constructed with the type 1 T helper cell-related signature showed high accuracy in predicting overall survival and response to immunotherapy. This study might improve the understanding of the role of type 1 T helper cells in PDAC patients and aid in the development of immunotherapy and personalized treatments for these patients.

Pancreatic cancer is a highly fatal disease and an increasing common cause of cancer mortality. Mounting evidence now indicates that molecular heterogeneity in pancreatic cancer significantly impacts its clinical features. However, the dynamic nature of gene expression pattern makes it difficult to rely solely on gene expression alterations to estimate disease status. By contrast, biological networks tend to be more stable over time under different situations. In this study, we used a gene interaction network from a new point of view to explore the subtypes of pancreatic cancer based on individual-specific edge perturbations calculated by relative gene expression value. Our study shows that pancreatic cancer patients from the TCGA database could be separated into four subtypes based on gene interaction perturbations at the individual level. The new network-based subtypes of pancreatic cancer exhibited substantial heterogeneity in many aspects, including prognosis, phenotypic traits, genetic mutations, the abundance of infiltrating immune cell, and predictive therapeutic efficacy (chemosensitivity and immunotherapy efficacy). The new network-based subtypes were closely related to previous reported molecular subtypes of pancreatic cancer. This work helps us to better understand the heterogeneity and mechanisms of pancreatic cancer from a network perspective.