A Systematic Review and Meta-Analysis of Tau Protein and Other Major Biomarkers that Predict Unfavorable Outcomes in Severe Traumatic Brain Injury

Abstract

Abstract Introduction: Initial traumatic brain injury (TBI) presentation and late clinical outcomes are usually evaluated by the Glasgow Coma Scale which lacks strong prognostic predictability. Several blood biomarkers have been linked to TBI, such as Tau, GFAP, UCH-L1, S-100B, and NSE. Clinical values of TBI biomarkers have yet to be evaluated in a focused multi-study meta-analysis. We reviewed relevant articles evaluating potential relationships between TBI biomarkers and both early and 6-month outcomes. Methods: All PubMed article publications from January 2000 to November 2023 with the search criteria: “Biomarker” AND “Traumatic Brain Injury” were included. Amongst all comparative studies, the sensitivity means and range values of biomarkers in predicting CT Rotterdam scores, early ICU admission in the early period, or predicting GOS-E < 4 at the 6-month period were calculated from confusion matrices. Sensitivity values were modeled for each biomarker across studies and compared statistically for heterogeneity and differences. Results: From the 65 articles that met our criteria, 13 were included in this study. Six articles involved early-period TBI outcomes and seven involved 6-month outcomes. In the early period TBI outcomes, GFAP had a superior sensitivity to UCH-L1 and S-100B, and similar sensitivity to the CT Rotterdam score. In the 6-month period TBI outcomes, total Tau and NSE both had significant interstudy heterogeneity, making them inferior to GFAP, phosphorylated Tau, UCH-L1, and S-100B, all four of which had similar sensitivities at 75%. This sensitivity range at 6-month outcomes was still relatively inferior to the CT Rotterdam score (85%). Conclusion: In this meta-analysis study, total Tau did not show any advantages in the severe TBI 6-month prognosis with GOS-E < 4, and phosphorylated Tau was similar in its sensitivity to other biomarkers such as GFAP and UCH-L1 and still inferior to the CT Rotterdam score. TBI biomarkers need stronger sensitivity evidence to support a more cost-efficient model.