Clinical Prognostic Modeling and Paired Blood-CSF Metabolomic Profiling for Outcome Prediction in Isolated...

Objectives: This study aimed to develop a prognostic model for isolated moderate-to-severe traumatic brain injury (TBI) (Glasgow Coma Scale [GCS] ≤ 12) using readily available variables and to explore paired blood-cerebrospinal fluid (CSF) metabolomic signatures. Methods: Consecutive TBI patients admitted between January 2019 and June 2025 were retrospectively analyzed. Multivariate logistic regression with bootstrap internal validation identified predictors of 6-month unfavorable outcome and in-hospital mortality. Untargeted metabolomics was performed on paired blood and CSF samples from 30 matched male patients. Results: Among 405 patients, 266 (65.7%) had unfavorable outcomes and 54 (13.3%) died in hospital. Rotterdam CT Score (odds ratio [OR] 10.59, 95% confidence interval [CI] 6.19-18.14), initial lactate (OR 1.81, 95% CI 1.38-2.36), and blood glucose (OR 1.40, 95% CI 1.21-1.64) predicted unfavorable outcome (area under the receiver operating characteristic curve [AUC] 0.97). GCS motor score (OR 0.50, 95% CI 0.37-0.66), initial lactate (OR 1.57, 95% CI 1.31-1.91), and follow-up lactate (OR 1.57, 95% CI 1.34-1.88) predicted mortality (AUC 0.96). Blood metabolomics revealed enrichment in energy and lipid metabolism pathways. CSF metabolomics highlighted neurotransmitter pathway dysregulation and neuroinflammatory markers, with depleted kynurenic acid in both biofluids. Conclusions: Readily available admission variables enable early bedside risk stratification in TBI. Metabolomic profiling links unfavorable outcomes to systemic energy-lipid dysregulation and central neuroinflammatory-neurotransmitter disturbances, with the tryptophan-kynurenine axis as a potential therapeutic target for neuroprotective strategies.