Glial fibrillary acidic protein level on admission can predict severe traumatic brain injury in patients with severe multiple trauma: A single-center retrospective observational study

OBJECTIVE: This study aimed to clarify whether the glial fibrillary acidic protein (GFAP) and soluble protein-100β (S100β) can predict severe traumatic brain injury (TBI) in patients with severe multiple trauma. METHODS: This is a single-center retrospective observational study of 179 patients with severe multiple trauma. The GFAP and S100β were measured upon patient arrival at the hospital. We divided the patients into the severe TBI group (with a Traumatic Coma Data Bank classification of ≥III), the non-severe TBI group (non-TBI group [absence of abnormality on the computed tomography scan and extracranial injury], and the mild to moderate TBI group [TCDB classification I and II]). We compared biomarker levels between the two groups and then evaluated the accuracy of predicting severe TBI using a receiver operating characteristic curve. RESULTS: A total of 41 patients had severe TBI, and 138 had non-severe TBI. Mean GFAP levels were significantly higher in the severe TBI group (median, 6000 pg/mL; interquartile range [IQR], 651–15,548 pg/mL) than in the non-severe TBI group (median, 149 pg/mL; IQR, 0–695 pg/mL) (p < 0.0001). In contrast, there was no significant difference in S100β levels between the severe TBI group (median, 64 pg/mL; IQR, 0–536 pg/mL) and non-severe TBI group (median, 117 pg/mL; IQR, 0–403 pg/mL) (p = 0.637). The area under the receiver operating characteristic curve was 0.810 (p < 0.0001) for GFAP and 0.476 (p = 0.908) for S100β. For the GFAP, the optimal cutoff value for detecting severe TBI was 947 pg/mL (sensitivity, 75.6%; specificity, 78.3%). CONCLUSIONS: In patients with severe multiple trauma, the GFAP level at hospital arrival could predict severe TBI, whereas the S100β level was not a useful predictor.