ObjectiveNuclear factor erythroid 2-related factor 2 (Nrf2) may harbor endogenous neuroprotective role. We strived to ascertain the prognostic significance of serum Nrf2 in severe traumatic brain injury (sTBI).MethodsThis prospective cohort study included 105 controls and 105 sTBI patients, whose serum Nrf2 levels were quantified. Its relations to traumatic severity and 180-day overall survival, mortality, and poor prognosis (extended Glasgow Outcome Scale score 1–4) were discerned using multivariate analysis.ResultsThere was a substantial enhancement of serum Nrf1 levels of patients (median, 10.9 vs. 3.3 ng/ml; P < 0.001), as compared to controls. Serum Nrf2 levels were independently correlative to Rotterdam computed tomography (CT) scores (ρ = 0.549, P < 0.001; t = 2.671, P = 0.009) and Glasgow Coma Scale (GCS) scores (ρ = −0.625, P < 0.001; t = −3.821, P < 0.001). Serum Nrf2 levels were significantly higher in non-survivors than in survivors (median, 12.9 vs. 10.3 ng/ml; P < 0.001) and in poor prognosis patients than in good prognosis patients (median, 12.5 vs. 9.4 ng/ml; P < 0.001). Patients with serum Nrf2 levels > median value (10.9 ng/ml) had markedly shorter 180-day overall survival time than the other remainders (mean, 129.3 vs. 161.3 days; P = 0.002). Serum Nrf2 levels were independently predictive of 180-day mortality (odds ratio, 1.361; P = 0.024), overall survival (hazard ratio, 1.214; P = 0.013), and poor prognosis (odds ratio, 1.329; P = 0.023). Serum Nrf2 levels distinguished the risks of 180-day mortality and poor prognosis with areas under receiver operating characteristic curve (AUCs) at 0.768 and 0.793, respectively. Serum Nrf2 levels > 10.3 ng/ml and 10.8 ng/ml discriminated patients at risk of 180-day mortality and poor prognosis with the maximum Youden indices of 0.404 and 0.455, respectively. Serum Nrf2 levels combined with GCS scores and Rotterdam CT scores for death prediction (AUC, 0.897; 95% CI, 0.837–0.957) had significantly higher AUC than GCS scores (P = 0.028), Rotterdam CT scores (P = 0.007), or serum Nrf2 levels (P = 0.006) alone, and the combination for poor outcome prediction (AUC, 0.889; 95% CI, 0.831–0.948) displayed significantly higher AUC than GCS scores (P = 0.035), Rotterdam CT scores (P = 0.006), or serum Nrf2 levels (P = 0.008) alone.ConclusionIncreased serum Nrf2 levels are tightly associated with traumatic severity and prognosis, supporting the considerable prognostic role of serum Nrf2 in sTBI.