21 Associations Between Initial Injury Severity, Cerebral Metabolites, and the Local Connectome in Remote Mild-to-Moderate TBI

Abstract

Objective:

To determine the association between in-vivo spectroscopy metabolite data, the local connectome, and markers of initial injury severity (I.e., history of loss of consciousness; LoC) in traumatic brain injury (TBI), in a heterogenous sample of Veterans and non-Veterans with a history of remote mild-to-moderate TBI (I.e., >6 months).

Participants and Methods:

Participants with complete PRESS magnetic resonance spectroscopy (MRS) and diffusion weighted imaging (DWI) data (N = 41) were sampled from a larger multisite study of chronic mild-to-moderate TBI (Nmiid = 38; Nmoderate = 3; 54% with LoC; 46% with multiple TBI). The sample was predominantly male (76%) with ages ranging from 23-59 (M = 36.9, SD = 10.1), with 98% holding at least a high school degree (M = 14.5 years of education, SD = 2.4). Fully tissue-and-relaxation-corrected metabolite concentration estimates in the dorsal anterior cingulate (30x30x30mm voxel) were modeled using Osprey 2.4.0. Total creatine (tCr), total choline (tCho), total N-acetylaspartate (tNAA), glutamate/glutamine (Glx), and myo-inositol (mI) were analyzed. Logistic regression was used to measure the association between metabolites and history of TBI with LoC. Correlational connectometry using the normalized spin distribution function was performed for metabolites associated with LoC, to characterize the local connectome associated with metabolites of interest, controlling for age and sex, and correcting for multiple comparisons (FDR < .050 with 4000 permutations). A profile approach was used to interpret diffusion metrics, contrasting quantitative anisotropy (QA) with fractional anisotropy (FA). Local connectome tracks were then clustered to identify the larger white matter tract.

Results:

Glx (p = .008) and tCr (p = .032) were significantly associated with history of TBI with LoC. Increased Glx was associated with increased QA in 11,001 tracks, accounting for 1.4% of the total white matter tracks in the brain. 90% of tracks were identified in bilateral cingulum (33%), bilateral thalamic (13%), bilateral corticospinal (13%), corpus callosum (12%), left arcuate fasciculus (9%), left frontoparietal aslant tracts (6%), and bilateral inferior fronto-occipital fasciculus (4%) tracts. In contrast, FA was not associated with Glx. The same pattern emerged for tCr, with 10,542 tracks identified predominantly in bilateral cingulum (29%), corpus callosum (21%), bilateral corticospinal (15%), bilateral corticostriatal (7%), bilateral medial lemniscus (7%), left cortico-pontine (3%), left thalamic (2%), and bilateral superior longitudinal fasciculus (2%) tracts. Post-hoc exploratory analyses of mean QA across regions of cingulum found that increased QA was associated with self-report measures of headache intensity, fatigue, and perceived change in executive functioning.

Conclusions:

Results provided evidence that multimodal imaging can identify subtle markers of initial TBI severity years after injury. Neurometabolite concentrations were associated with diffuse changes in the local connectome; the pattern of discrepancy between FA and QA was suggestive of reduced potential for neuroplasticity. Exploratory analyses further indicated that variability in white matter density in the cingulum, an important connection for limbic regions, was associated with a range of problems commonly reported in clinical settings, which may be informative for diagnosis and treatment planning.