Metabolomic and lipidomic profiling reveals distinct subtypes of glioblastoma IDH-wildtype and shared metabolic features with astrocytoma IDH-mutant grade 4

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Abstract Background Glioblastoma IDH-wildtype (GBM) is a metabolically diverse and aggressive brain tumor. While transcriptomic subtypes are well-defined, they lack a direct understanding of the tumor’s metabolism. We used integrated, untargeted metabolomics and lipidomics to define the functional metabolic landscape of GBM and astrocytoma, IDH-mutant, Grade 4, and to identify intrinsic metabolic subtypes with specific vulnerabilities. Methods Brain tissue from GBM patients (n = 38), astrocytoma IDH-mutant, Grade 4 (n = 5), and non-neoplastic controls (n = 20) underwent untargeted metabolomic and lipidomic profiling. Data were analyzed using Partial Least Squares Discriminant Analysis (PLS-DA), Principal Component Analysis (PCA), and K-means clustering. Quantitative pathway enrichment analyses were performed based on established databases and Z-scores are reported for effect sizes. A multinomial logistic regression classifier was trained to assess diagnostic utility. Results GBM and IDH-mutant astrocytoma, Grade 4 exhibited distinct metabolomic profiles, driven by 2-hydroxyglutarate accumulation in IDH-mutants. Machine learning classification trained on the metabolomic profiles achieved 95% diagnostic accuracy. Within GBM, K-means clustering of the metabolomics dataset revealed three functionally distinct subtypes. Cluster 1 was defined by fatty acid turnover. Cluster 2 was characterized by the robust accumulation of triglycerides, alongside a significantly reduced hexosylceramide-to-ceramide ratio. Cluster 3 was distinguished by upregulated one-carbon and folate metabolism, a significant depletion of acylcarnitines, and susceptibility to ferroptotic cell death. Conclusions Metabolomic and lipidomic profiling holds promise as an accurate diagnostic tool, while also enabling a deeper metabolic understanding of high-grade gliomas. This approach resulted in the identification of three distinct metabolic subtypes of GBM, each possessing unique therapeutic vulnerabilities.