Association Between Acute Plasma Glial Fibrillary Acidic Protein Levels and White Matter Integrity in Recent Trauma Survivors.

Traumatic brain injuries (TBIs) account for over 2.5 million emergency department (ED) visits each year in the United States. The bulk of TBI research in acute care settings has focused exclusively on individuals who receive computed tomography (CT) scanning. Recently, more sensitive TBI assessment tools have been identified, including blood biomarkers such as glial fibrillary acidic protein (GFAP) and magnetic resonance imaging (MRI) of white matter microstructural integrity. These methods can evaluate for evidence of TBI in CT-negative patients. However, limited prior work has investigated the relationship between blood GFAP levels and white matter microstructural integrity. The present study examined the association between acute (≤72 h) GFAP levels and white matter microstructural integrity at 2 weeks post-trauma among ED patients (n = 328) recruited after a qualifying traumatic experience. We hypothesized that acute GFAP concentrations would be associated with disrupted microstructural integrity, quantified with Tract-Based Spatial Statistics-derived diffusion tensor imaging (DTI) metrics (reduced fractional anisotropy and increased axial diffusivity [AD], radial diffusivity, and mean diffusivity), in tracts previously shown to be vulnerable to TBI. Our secondary hypothesis was that there would be a stronger association between GFAP levels and white matter integrity among female relative to male trauma survivors, in line with previous literature suggesting that females' axons are less resilient to stretch injury. To test the primary hypothesis, we fit four sets of linear regression models to assess the association between natural log-transformed acute GFAP concentration and DTI metrics for 15 white matter regions of interest (ROIs). The false discovery rate (FDR) was controlled at 0.05 for each set of models. Models were adjusted for age, sex, neuroimaging site, and a composite lifetime traumatic stress variable. To test the secondary hypotheses, we fit identical sets of linear regression models, with an additional interaction term between sex and ln(GFAP). We followed up all significant results surviving FDR correction with correlational tractography to examine the location and direction of tracts associated with the DTI metric passing through those ROIs. We observed significant and positive associations between acute GFAP and AD in the parahippocampal cingulum (B = 1.416 × 10^-5, standard error [SE] = 4.806 × 10^-6, p = 0.003, FDR-corrected p = 0.026), and the sagittal stratum (B = 1.012 × 10^-5, SE = 3.438 × 10^-6, p = 0.003, FDR-corrected p = 0.026), including fibers of the inferior longitudinal fasciculus and the inferior fronto-occipital fasciculus. Furthermore, the association between GFAP and AD of the parahippocampal cingulum was stronger in female trauma survivors compared with male trauma survivors (B = 2.898 × 10^-5, SE = 9.352 × 10^-6, p = 0.002, FDR-corrected p = 0.032). Our results suggest that among ED patients with low rates of intracranial injuries observed on CT and MRI scans, acute peripheral levels of GFAP may be able to predict later disruption in structural connectivity.