Abstract:
Background: Pancreatic cancer, particularly pancreatic ductal adenocarcinoma (PDAC), is known for its aggressiveness, early metastasis, and poor prognosis. Despite advancements in diagnostic techniques and treatments, the survival rate remains low. Understanding metabolic reprogramming in pancreatic cancer cells and its impact on the tumour microenvironment is crucial for developing new therapeutic strategies.
Methods: We utilized standardized RNA-Seq data from the TCGA-PAAD cohort, as well as microarray expression data from the GSE57495 and GSE21501 cohorts. Using gene set variation analysis (GSVA), we quantified the activity of metabolic pathways in tumour samples and performed unsupervised clustering to identify distinct metabolic subtypes. We analysed the associations between these subtypes and tumour immune cell infiltration, gene mutations, and drug sensitivity.
Results: This study integrated data from TCGA-PAAD, GSE57495, and GSE21501, forming a comprehensive dataset of 372 samples and 15,557 genes. Using gene set variation analysis (GSVA), we calculated enrichment scores for 72 metabolic pathways and identified 12 significantly enriched pathways. Unsupervised clustering revealed two metabolic subtypes (C1 and C2) with distinct overall survival (OS) rates. Further analysis revealed significant differences in pathway activity and immune cell infiltration between these subtypes. We identified 1,226 differentially expressed genes, and GO and KEGG enrichment analyses were performed to reveal key biological processes. A machine learning prognostic model developed using the random forest algorithm demonstrated robustness and predictive accuracy and was validated externally with additional datasets. The model risk score was significantly correlated with clinical features and served as an independent prognostic factor. Our findings revealed the mutational landscape and suggested potential therapeutic strategies on the basis of metabolic characteristics, enhancing the understanding of the roles of key gene in tumour biological features and the immune response. This study highlights the heterogeneity of metabolic pathways and the relationship of these pathways with the tumour immune microenvironment, providing a foundation for precision treatment strategies in pancreatic cancer.
Conclusions: Metabolic subtyping of pancreatic cancer revealed significant heterogeneity in metabolic pathways and the relationship of metabolic subtype with tumour immune microenvironment features. These findings provide a new theoretical basis for precision treatment strategies targeting metabolic pathways and immune evasion mechanisms in pancreatic cancer.
