Serum Metabolome Profiling Reveals Biomarkers to Predict Malignant Brain Edema Following Endovascular Stroke Therapy

Metabolomics reflects the body's metabolic state and holds substantial promise for identifying associated biomarkers and exploring pathological processes. This study used serum metabolomics to predict malignant brain edema (MBE) following endovascular therapy (EVT) for acute ischemic stroke (AIS) and investigate the underlying mechanisms. In this prospective observational study, we enrolled patients with anterior circulation large vessel occlusion who underwent successful recanalization post-EVT for AIS, including those with or without concomitant hemorrhagic transformation. Eligible patients were stratified into two groups according to the subsequent presence of MBE. Following propensity score matching to adjust for potential confounders, widely targeted metabolomic analysis was conducted on the serum samples. A prediction model, based on the identified differential metabolites, was then developed and validated in another independent cohort. Through widely targeted metabolomic analysis of serum samples from matched 46 patients, we identified 30 metabolites that were significantly altered between patients with and without MBE, including notable differences in acylcarnitine, lysophosphatidylcholine, riboflavin, and tyrosine. A prediction model incorporating 9 differential metabolites was constructed using 10-fold frame shift cross-validation, demonstrating prediction performance in the training set, test set, and external validation, with areas under the curve (AUC) of 0.842, 0.815, and 0.734, respectively. Dynamic change analysis based on multiple time points may suggest a potential role of diminished oxidative energy supply and sustained stress in MBE. The identified acylcarnitine and lysophosphatidylcholine might play influential roles in the pathogenesis of MBE. The prediction model derived from these differential metabolites holds promise as a noninvasive assay for the early detection of MBE and warrants additional refinement and validation.