Glioma region-specific metabolites associated with patient seizures

Seizures are often the first and only clinical symptom of glioma presence, especially in low-grade tumors. Anti-epilepsy drugs have shown mixed success due to tumor-induced biochemical changes that uniquely alter brain tissue functionality and activity. Recently, it was shown that glioma metabolism is differentially dysregulated between core (contrast enhancing) and edge (T2/FLAIR hyperintense) tissue. To provide insight into the biochemical alterations underlying seizures, this study hypothesized that glioma region-specific metabolites could be associated with seizure occurrence.

Log-transformed metabolite signal intensities were obtained by 2D liquid chromatography mass spectrometry/mass spectrometry of patient core and edge glioma samples (n = 39 patients). To identify metabolites associated with seizure occurrence, metabolites were used as features for classification of seizure via machine learning, including forward feature selection and rigorous model validation using repeated cross-validation and a test (holdout) set.

Pre-op seizure occurrence could be accurately classified with five core metabolites (AUROC_TEST=0.892, PRAUC_TEST=0.835, maximum F1_TEST=0.800). Edge metabolites provided a weaker signal, with seizure classification using two metabolites (AUROC_TEST=0.750, PRAUC_TEST=0.733, maximum F1_TEST=0.800). Seizures could also be accurately classified using one edge and four core metabolites (AUROC_TEST=0.871, PRAUC_TEST=0.889, maximum F1_TEST=0.800). Furthermore, seizure occurrence was associated with differential relative abundance of various core and edge metabolites, linked to dysregulation of metabolic pathways identified from KEGG database.

Tumor region-specific metabolites in glioma patients were associated with seizure occurrence, identifying metabolomic dysregulation potentially reflecting biochemical alterations. This study represents a first step towards identifying metabolites linked to seizure occurrence, with the ultimate goal of improving glioma patient outcomes.