Deciphering the Prognostic Implications of the Components and Signatures in the Immune Microenvironment of Pancreatic Ductal Adenocarcinoma

Background: The treatment modalities for pancreatic ductal adenocarcinoma (PDAC) are limited and unsatisfactory. Although many novel drugs targeting the tumor microenvironment, such as immune checkpoint inhibitors, have shown promising efficacy for some tumors, few of them significantly prolong the survival of patients with PDAC due to insufficient knowledge on the tumor microenvironment. Methods: A single-cell RNA sequencing (scRNA-seq) dataset and seven PDAC cohorts with complete clinical and bulk sequencing data were collected for bioinformatics analysis. The relative proportions of each cell type were estimated using the gene set variation analysis (GSVA) algorithm based on the signatures identified by scRNA-seq or previous literature. Results: A meta-analysis of 883 PDAC patients showed that neutrophils are associated with worse overall survival (OS) for PDAC, while CD8+ T cells, CD4+ T cells, and B cells are related to prolonged OS for PDAC, with marginal statistical significance. Seventeen cell categories were identified by clustering analysis based on single-cell sequencing. Among them, CD8+ T cells and NKT cells were universally exhausted by expressing exhaustion-associated molecular markers. Interestingly, signatures of CD8+ T cells and NKT cells predicted prolonged OS for PDAC only in the presence of "targets" for pyroptosis and ferroptosis induction. Moreover, a specific state of T cells with overexpression of ribosome-related proteins was associated with a good prognosis. In addition, the hematopoietic stem cell (HSC)-like signature predicted prolonged OS in PDAC. Weighted gene co-expression network analysis identified 5 hub genes whose downregulation may mediate the observed survival benefits of the HSC-like signature. Moreover, trajectory analysis revealed that myeloid cells evolutionarily consisted of 7 states, and antigen-presenting molecules and complement-associated genes were lost along the pseudotime flow. Consensus clustering based on the differentially expressed genes between two states harboring the longest pseudotime span identified two PDAC groups with prognostic differences, and more infiltrated immune cells and activated immune signatures may account for the survival benefits. Conclusion: This study systematically investigated the prognostic implications of the components of the PDAC tumor microenvironment by integrating single-cell sequencing and bulk sequencing, and future studies are expected to develop novel targeted agents for PDAC treatment.