LncRNAs associated with vascular mimicry establish a novel molecular subtype and prognostic model for pancreatic cancer

Background Vascular mimicry (VM) epitomizes an innovative tumor angiogenesis pathway, potentially serving as an alternate conduit under the assumption of traditional tumor angiogenesis pathway inhibition. The role of VM in pancreatic cancer (PC), however, remains unexplored. Methods Using differential analysis and Spearman correlation, we identified key long non-coding RNAs (lncRNAs) signatures in PC from the collected set of VM-associated genes in the literature. We identified optimal clusters using the non-negative matrix decomposition (NMF) algorithm, and then compared clinicopathological features and prognostic differences between clusters. We also assessed tumor microenvironmental (TME) differences between clusters using multiple algorithms. Using univariate Cox regression analyses as well as lasso regression, we constructed and validated new lncRNA prognostic risk models for PC. We used Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) to analyze model-enriched functions and pathways. Nomograms were then developed to predict patient survival in association with clinicopathological factors. In addition, single-cell RNA-sequencing (scRNA-seq) analysis was used to analyze the expression patterns of VM-related genes and lncRNAs in the PC of TME. Finally, we used the Connectivity Map (cMap) database to predict local anaesthetics that could modify the VM of PC. Results In this study, we developed a novel three-cluster molecular subtype using the identified VM-associated lncRNA signatures of PC. The different subtypes have significantly different clinical characteristics and prognostic value, and also show differential treatment response and TME. Following an in-depth analysis, we constructed and validated a novel prognostic risk model for PC based on the VM-associated lncRNA signatures. Enrichment analysis suggested that high riskscores were significantly associated with functions and pathways, including extracellular matrix remodeling, et al. In addition, we predicted eight local anaesthetics that could modulate VM in PC. Finally, we discovered differential expression of VM-related genes and lncRNAs across various cell types within pancreatic cancer. Conclusion VM has a critical role in PC. This study pioneers the development of a VM-based molecular subtype that demonstrates substantial differentiation in PC populations. Furthermore, we highlighted the significance of VM within the immune microenvironment of PC. Moreover, VM might contribute to PC tumorigenesis through its mediation of mesenchymal remodeling and endothelial transdifferentiation-related pathways, which offers a new perspective on its role in PC.