A novel risk model of cholesterol metabolism-related mRNAs for predicting overall survival and immune signature in glioma based on machine learning and multi-omics data.

Background: The intricate molecular pathways involved in gliomas, coupled with the shift toward more personalized treatment approaches, necessitate the identification of novel and dependable molecular targets. Cholesterol metabolism is a key player in modulating metabolic alterations in glioma cells and shaping the tumor immune microenvironment. Consequently, thoroughly examining messenger RNAs (mRNAs) within this pathway has emerged as a novel avenue for investigating the prognostic markers and immunotherapeutic targets for glioma, which also constitutes the objective of the present study. Methods: Transcriptomic profiles with matched clinical annotations were retrieved from The Cancer Genome Atlas (TCGA) and Chinese Glioma Genome Atlas (CGGA) cohorts, while additional single-cell expression data were accessed through the Gene Expression Omnibus (GEO) repository. Candidate cholesterol metabolism-related genes (CMRGs) were collected based on prior literature and curated databases. Prognostic genes were identified using stepwise Cox regression combined with least absolute shrinkage and selection operator (LASSO) regularization. On this basis, we formulated a multi-gene risk score to stratify patient survival. The score was incorporated with clinical covariates into a prognostic nomogram, and functional enrichment analyses were applied to characterize the associated molecular pathways. We further examined immune microenvironment differences, predicted potential response to immune checkpoint therapy, and estimated chemotherapeutic sensitivity. Finally, expression patterns of the selected genes were validated experimentally. Results: After stepwise selection, five mRNAs with prognostic value were incorporated into a survival risk score. In the independent validation cohort, this signature achieved favorable discrimination of overall survival (OS), with area under the curve (AUC) values of 0.893, 0.804, and 0.749 for 1-, 3-, and 5-year survival, respectively. Comparisons of immune infiltration, functional activity, ESTIMATE-derived scores, Tumor Immune Dysfunction and Exclusion (TIDE) prediction, and tumor mutation burden between risk groups confirmed the biological relevance of the model. Experimental assays further supported the predicted expression patterns and were consistent with our initial expectations. Conclusions: We established a cholesterol metabolism-associated mRNA signature to stratify glioma patients, and its predictive performance was confirmed in independent datasets and complementary analyses. This framework provides a practical tool for survival estimation and may contribute to optimizing immunotherapy decision-making in glioma management.