A comprehensive analysis of a machine learning-based risk assessment model for senescence-related gene characteristics in biliary tract cancer cells and its applications

Cholangiocarcinoma (CCA) is a highly aggressive malignancy characterized by significant heterogeneity and complex pathogenic mechanisms. Recent advancements in high-throughput sequencing and molecular biology have allowed for a deeper understanding of its underlying biological processes. In this study, we integrated transcriptomic data from the TCGA, GEO, and EMBL databases to explore the gene expression profiles associated with cellular senescence in CCA. Through advanced computational methods, including cox regression models, weighted gene co-expression network analysis (WGCNA), least absolute shrinkage and selection operator (least absolute shrinkageand selection operator, LASSO), and random forest algorithms, we identified four pivotal genes linked to cellular senescence: TFPI, MRPL43, FTSJ1, and MNAT1. Using these genes, we developed a cellular senescence gene score (CSGS) risk model. The CSGS exhibited strong prognostic capabilities, effectively stratifying patients based on their risk levels. Notably, high-risk patients had significantly poorer overall survival compared to those in the low-risk category. Functional enrichment analyses revealed that cellular senescence plays a critical role in CCA initiation and progression, particularly through modulating the immune microenvironment, triggering inflammatory responses, and driving metabolic reprogramming. Immune cell infiltration analysis highlighted an increase in immunosuppressive cells, such as M2 macrophages and regulatory T cells, within the tumor microenvironment of high-risk group, while a decrease in NK cells and CD8+ T cells was observed. These findings underscore the potential of immune evasion mechanisms in supporting tumor progression and their impact on the effectiveness of treatments. Furthermore, we examined the relationship between the core senescence-related genes and drug sensitivity. Our results demonstrated that the expression levels of TFPI, FTSJ1, and MNAT1 were associated with the sensitivity of CCA cells to several therapeutic agents, including AZD6482, selumetinib, rapamycin, and lapatinib. This suggests that these genes could serve as valuable molecular targets for precision medicine, potentially enhancing personalized treatment strategies for CCA. In conclusion, this study identifies cell senescence-related gene signatures and reveals their significant association with the prognosis of cholangiocarcinoma patients. These findings suggest that targeting the senescence microenvironment may serve as a potential strategy to improve clinical outcomes. The results provide novel biological insights for the molecular classification of cholangiocarcinoma and the exploration of therapeutic targets.