Glioblastoma region-specific metabolic signatures reflect patient survival duration.
Spatial metabolic differences found in glioblastoma (GBM) tumor core (contrast enhancing) and peritumoral (T2/FLAIR hyperintense) edge tissue regions have recently enabled stratification of patient overall survival. However, the association between metabolic dysregulation and survival duration remains poorly understood. To gain further insight into the biological characteristics underlying longer vs. shorter survival, this study employed an interdisciplinary approach to analyze GBM region-specific metabolic signatures predictive of patient overall survival.
Patient survival data were paired with core and edge biopsy tumor metabolomic data (n = 37 pairs) obtained by 2D liquid chromatography-mass spectrometry/mass spectrometry. Metabolite expression was compared between patients with short (≤ median) and long (> median) overall survival using relative abundance analysis. Metabolic signatures predictive of patient survival were identified via a comprehensive machine learning (ML) workflow, including repeated nested cross validation and test (holdout) set evaluation. Pathways associated with key metabolites were identified from the KEGG database.
Core metabolite levels generally were increased and edge metabolite levels were decreased in patients with longer survival, Edge tissue metabolic signatures reflected survival duration better than signatures from core tissue. Metabolites differentiating short vs. long survival were associated with metabolic pathway dysfunction related to fatty acid and amino acid metabolism, glycolysis and gluconeogenesis, and ATP synthesis.
Interdisciplinary analysis of GBM region-specific metabolic signatures predictive of patient short vs. long survival can yield insight into the local and global metabolic dysfunction associated with survival duration.
Patient survival data were paired with core and edge biopsy tumor metabolomic data (n = 37 pairs) obtained by 2D liquid chromatography-mass spectrometry/mass spectrometry. Metabolite expression was compared between patients with short (≤ median) and long (> median) overall survival using relative abundance analysis. Metabolic signatures predictive of patient survival were identified via a comprehensive machine learning (ML) workflow, including repeated nested cross validation and test (holdout) set evaluation. Pathways associated with key metabolites were identified from the KEGG database.
Core metabolite levels generally were increased and edge metabolite levels were decreased in patients with longer survival, Edge tissue metabolic signatures reflected survival duration better than signatures from core tissue. Metabolites differentiating short vs. long survival were associated with metabolic pathway dysfunction related to fatty acid and amino acid metabolism, glycolysis and gluconeogenesis, and ATP synthesis.
Interdisciplinary analysis of GBM region-specific metabolic signatures predictive of patient short vs. long survival can yield insight into the local and global metabolic dysfunction associated with survival duration.