Predictive value of neuron-specific enolase for prognosis in patients with moderate or severe traumatic brain injury: a systematic review and meta-analysis.

Abstract

Prognosis is difficult to establish early after moderate or severe traumatic brain injury despite representing an important concern for patients, families and medical teams. Biomarkers, such as neuron-specific enolase, have been proposed as potential early prognostic indicators. Our objective was to determine the association between neuron-specific enolase and clinical outcomes, and the prognostic value of neuron-specific enolase after a moderate or severe traumatic brain injury. We searched MEDLINE, Embase, The Cochrane Library and Biosis Previews, and reviewed reference lists of eligible articles to identify studies. We included cohort studies and randomized controlled trials that evaluated the prognostic value of neuron-specific enolase to predict mortality or Glasgow Outcome Scale score in patients with moderate or severe traumatic brain injury. Two reviewers independently collected data. The pooled mean differences were analyzed using random-effects models. We assessed risk of bias using a customized Quality Assessment of Diagnostic Accuracy Studies (QUADAS-2) tool. Subgroup and sensitivity analyses were performed based on a priori hypotheses. We screened 5026 citations from which 30 studies (involving 1321 participants) met our eligibility criteria. We found a significant positive association between neuron-specific enolase serum levels and mortality (10 studies, n = 474; mean difference [MD] 18.46 µg/L, 95% confidence interval [CI] 10.81 to 26.11 µg/L; I2 = 83%) and a Glasgow Outcome Scale ≤ 3 (14 studies, n = 603; MD 17.25 µg/L, 95% CI 11.42 to 23.07 µg/L; I2 = 82%). We were unable to determine a clinical threshold value using the available patient data. In patients with moderate or severe traumatic brain injury, increased neuron-specific enolase serum levels are associated with unfavourable outcomes. The optimal neuron-specific enolase threshold value to predict unfavourable prognosis remains unknown and clinical decision-making is currently not recommended until additional studies are made available.