Prognostic Value of Blood-Based Protein Biomarkers in Traumatic Brain Injury: A Living Systematic Review and Meta-Analysis.

Abstract

Circulating biomarkers might improve the prediction of outcomes in patients with traumatic brain injury (TBI) beyond current approaches. Robust and up-to-date evidence is required to support their clinical utility and integration into medical practice to guide decision-making. Our objective was to critically appraise the existing evidence for six core blood-based TBI biomarkers (S100 calcium-binding protein B, glial fibrillary acidic protein [GFAP], neuron-specific enolase, ubiquitin C-terminal hydrolase-L1 [UCH-L1], tau and neurofilament proteins), in predicting outcome after TBI. Electronic databases, including Medline and Embase, were searched for articles published from their inception to October 2023. Studies were included if they evaluated the accuracy of blood biomarker concentrations at hospital presentation for outcome prediction in adult patients with TBI. Outcomes assessed were mortality, Glasgow Outcome Scale (GOS)/GOS extended (GOS-E), or the Rivermead Post-Concussion Symptoms Questionnaire (RPQ). Study selection, data extraction, and quality assessment using the modified Quality Assessment of Prognostic Accuracy Studies tool were performed by two authors independently, with disagreements being resolved through discussion or arbitration. If appropriate, a meta-analysis was conducted by calculating the weighted summary area under the curve (AUC) and using a bivariate regression model. Of 12,792 retrieved records, 32 articles, including 7481 patients with TBI, were selected as relevant. Two biomarkers showed strong associations with in-hospital and 6-month mortality: GFAP (unadjusted pooled AUC 0.81 [95% confidence interval [CI] 0.75-0.87] and 0.82 [0.80-0.85], respectively) and UCH-L1 (0.80 [0.74-0.85] and 0.83 [0.77-0.88]). Their addition to models that included established risk factors consistently improved the predictive value, though models and performance varied substantially across studies. In four studies measuring both markers, UCH-L1 outperformed GFAP in improving risk stratification when added to established prediction models. At ∼1.5 ng/mL (five studies), the summary sensitivity of GFAP for predicting mortality was 78% (95% CI 67-85%), and the summary specificity was 79% (95% CI 64-89%). The other assessed biomarkers had fair to good performance in mortality prediction with unclear added benefits. Neurofilament light (NfL) (three studies) demonstrated the strongest association in predicting a 6-month poor outcome (GOS-E ≤4; GOS ≤3) (unadjusted pooled AUC 0.81 [95% CI 0.75-0.87]), whereas the other assessed biomarkers had a fair performance with unclear or irrelevant added value. All core biomarkers had only marginal or no association with incomplete recovery and post-concussion symptoms/syndrome, as assessed by RPQ. Serious problems were found in the design and analysis of many of the studies. We conclude that admission measurements of core blood TBI biomarkers, in particular GFAP and UCH-L1, are strongly associated with mortality. There remains little evidence that any of these markers are ready for clinical implementation for prognostic purposes. Future work focused on the intended use and applying unbiased rigorous analysis methods is necessary to demonstrate that the biomarker test results are "prognostically actionable."