Human serum metabolites as outcome predictors in moderate and severe traumatic brain injury

Abstract

Traumatic brain injury (TBI) is major global health problem. Outcome of TBI varies between death and good neurological recovery. Several clinical characteristics are known to affect outcome, still prognostication after TBI has proven difficult. Novel biomarkers have been studied, but none have performed well enough to be useful in clinical work. Metabolic profiling has shown promise in many disease entities including TBI. In this study metabolic profile is compared between unfavorable outcome patients and favorable outcome patients during first week after the injury. Also, between fatalities and survivors. Prognostic model using logistic regression was generated and its performance was compared to existing IMPACT and CRASH models. Metabolic profile was found to differ between unfavorable and favorable outcome patients. It was observed that metabolic difference is likely to change in unfavorable and favorable outcome patients during the first week after injury. Change in metabolic profile is even greater between fatalities and survivors than between unfavorable and favorable outcome patients in first days after injury. Metabolic modeling shows promise as outcome predictor in moderate and severe TBI and performs with good accuracy in single center setup but validates poorly. Existing clinical models perform with poor accuracy in this cohort. Larger studies are needed in future to validate these findings. The changing difference in metabolic profiles between unfavorable and favorable outcome patients during the first week after injury is interesting finding, which have not been reported before.