Arterial Spin Labeling Reveals Elevated Cerebral Blood Flow with Distinct Clusters of Hypo- and Hyperperfusion after Traumatic Brain Injury

Abstract

Imaging detection of brain perfusion alterations after traumatic brain injury (TBI) may provide prognostic insights. In this study, we used arterial spin labeling (ASL) to quantify cross-sectional and longitudinal changes in cerebral blood flow (CBF) after TBI and correlated changes with clinical outcome. We analyzed magnetic resonance imaging scans from adult participants with TBI requiring hospitalization in the acute (2 weeks post-injury, n = 33) and chronic (6 months post-injury, n = 16) phases, with 13 participants scanned longitudinally at both time points. We also analyzed 18 age- and sex-matched healthy controls. Whole-brain CBF maps were derived using a three-dimensional pseudo-continuous arterial spin label technique. Mean CBF across tissue-based regions (whole brain, gray matter, and white matter) was compared cross-sectionally and longitudinally. In addition, individual-level clusters of abnormal perfusion were identified using voxel-based z-score analysis of relative CBF maps, and number and volume of abnormally hypo- and hyperperfused clusters were assessed cross-sectionally and longitudinally. Finally, all CBF measures were correlated with clinical outcome measures. Mean global and gray matter CBF were significantly elevated in acute and chronic TBI participants compared to controls. Participants with better outcome at 6 months post-injury tended to have higher CBF in the acute phase compared to those with poorer outcome. Acute TBI participants had a significantly greater volume of hypo- and hyperperfused brain tissue compared to controls, with these regions partially normalizing by the chronic phase. Our findings demonstrate global elevation of CBF with focal hypo- and hyperperfusion in the early post-injury period and suggest a reparative role for acute elevation in CBF post-TBI.

Introduction

Development of therapeutics for traumatic brain injury (TBI) has been hindered by a lack of validated diagnostic and prognostic biomarkers that can guide therapy.^1
–3 Although magnetic resonance imaging (MRI) has revolutionized the diagnosis and management of numerous neurological disorders, its application for prognostication in TBI remains limited. Patients who experience ongoing disability after TBI often exhibit no abnormalities on conventional MRI sequences, and, even when present, post-traumatic lesions correlate poorly with TBI-related symptoms.⁴ The development of neuroimaging biomarkers for more-precise stratification and prognostication is critical to enable targeted selection of patients for clinical trials and therapeutic interventions.⁵

Cerebral perfusion alterations after TBI are of particular interest, given that changes in vascular function have been linked to functional deficits after TBI⁶ and may be amenable to therapeutic intervention.⁷ One measure of vascular function is cerebral blood flow (CBF), which can be quantified non-invasively with MRI using arterial spin labeling (ASL).⁸ Changes in CBF have been detected using ASL in the chronic phase of mild TBI^9
–11 and the acute¹² and chronic¹³ phases of moderate-to-severe TBI. However, the direction and time course of CBF changes have varied significantly, with some studies reporting global/regional reduction of CBF after TBI^9,14 and others reporting regional elevation.^11,15
–17 The relationship between CBF and clinical outcome also remains unclear, with both hypoperfusion^11,15
–17 and hyperperfusion^18,19 linked to poor outcome. Significant intersubject variation in CBF,^20,21 heterogeneity in the pattern and spatial distribution of injury,^12,21
–23 and dynamic changes in CBF over time may all contribute to these inconsistencies. Longitudinal follow-up and individual subject analysis is necessary to better characterize these changes.

In this study, we aimed to clarify the utility of CBF as a biomarker for TBI by taking these factors into consideration. We focused on a population of TBI participants who met the Glasgow Coma Scale (GCS) criteria for mild injury but required hospitalization for their TBI. For each participant, we quantified tissue-based measures of CBF, as well as individual-level voxel-based measures. We performed cross-sectional comparisons (TBI vs. controls), as well as longitudinal comparisons (from 2 weeks to 6 months post-injury). Finally, we examined the relationship between tissue- and voxel-based measures of CBF and clinical outcome measures at 6 months post-injury.

Methods

Study population

We studied 33 hospitalized adults with TBI who presented to Penn Presbyterian Medical Center, the level 1 trauma center of the University of Pennsylvania, within 24 h of injury and were able to undergo 3 Tesla (T) MRI scanning at ∼2 weeks post-injury. TBI participants were included if they sustained a non-penetrating traumatic impact to the head, presented with symptoms consistent with TBI as defined by the Department of Defense,²⁴ and required hospitalization. All participants had GCS scores between 13 and 15, but required hospitalization because of their TBI. Accordingly, >80% of TBI participants had findings on head computed tomography (CT). Participants were excluded if they had a history of pre-existing serious neurological or psychiatric disease, comorbid disabling condition limiting outcome assessment, current pregnancy, or were incarcerated. For participants unable to provide informed consent, consent was obtained from a legally authorized representative. Healthy controls (HCs) were enrolled if they had no history of TBI, pre-existing disabling neurological or psychiatric disorders, or current pregnancy. The protocol was approved by the University of Pennsylvania Institutional Review Board.

Of 33 acute TBI participants, 19 had clinical outcome data collected in the chronic phase (6 months post-injury), including 13 who underwent a second scan in the chronic phase. Additionally, 3 participants were scanned only in the chronic phase, of which 2 were identified to meet study inclusion criteria during clinical visits and 1 was unable to complete their acute scan. We included 18 age- and sex-matched healthy volunteers as an HC group.

Clinical data collection

Demographic information, medical history, admission injury characteristics, and other clinical information were collected from hospital chart review and personal interview, when needed, for all participants. Outcome measures, including Glasgow Outcome Score-Extended (GOSE)²⁵ and Rivermead Post-Concussion Symptom Questionnaire (RPQ)²⁶ score, were collected 6 months post-injury using in-person, structured interviews. GOSE was measured based on disability attributable to TBI.

Image acquisition

Participants underwent brain MRI on a 3T scanner (Siemens Prisma; Siemens, Elangen, Germany) in the acute (2 weeks post-injury) and/or chronic (6 months post-injury) phase of injury. MRI included the following structural sequences: high-resolution three-dimensional (3D) T1-weighted magnetization-prepared rapid acquisition with gradient echo, fluid-attenuated inversion recovery (FLAIR), and susceptibility-weighted imaging. Background-suppressed ASL was acquired using a pseudo-continuous labeling scheme along with a 3D spiral readout with the following sequence parameters: repetition time (TR) = 4.2 sec, echo time = 10.03 ms, flip angle = 90 degrees, resolution = 3.75 mm³ isotropic, field of view = 240 × 240 mm, matrix = 64 × 64, bandwidth = 400 Hz/px, labeling time (τ) 1.8 sec, post-label delay 1.8 sec, and 18 label-control pairs. The labeling plane was placed perpendicular to the straight segments of the cervical carotid and vertebral arteries based on time-of-flight angiography. Two unlabeled M0 images with TR = 6 sec and without background suppression were acquired and averaged for use in CBF quantification. All MRI examinations received clinical interpretation by a board-certified subspecialist neuroradiologist, who identified the presence and location of any focal anatomical TBI-related abnormalities.

Image processing

ASL data were post-processed using the BASIL²⁷ toolbox in FSL,²⁸ which included motion correction of the raw label-control images, pair-wise label-control subtraction, division of each subtraction image by the mean M0 image, and application of the recommended model²⁹ (Eq. 1) to create a CBF time series using the following parameters: labeling efficiency (α) 85%; T1_blood 1.65 sec; and brain/blood partition coefficient (λ) 0.9 mL/g. A final denoised absolute CBF map was obtained by excluding degraded CBF time points using the structural correlation outlier rejection method and averaging across retained CBF time points.³¹ The final CBF maps were visually reviewed by a neuroradiologist to ensure data quality and assess for transit time artifact. Five subjects (2 acute, 1 chronic, and 2 controls) had scans with poor data quality, were excluded from analysis, and are not included in data reported here.

Structural images were used to create a population-specific anatomical template using the Advanced Normalization Tools (ANTs),³⁰ which was downsampled to a resolution of 2 mm³ to better match the resolution of ASL data, thereby reducing multiple comparisons in subsequent voxel-based CBF analysis and reducing the magnitude of CBF data interpolation during coregistration. The FSL boundary-based registration tool was used to compute a linear registration between each participant's structural image and the mean of the ASL time series, and structural images were normalized to the template using the ANTs symmetrical diffeomorphic registration tool (SyN) with B-spline interpolation.³¹ For tissue-level analysis, global (whole brain), gray matter, and white matter CBF were extracted in structural image space using masks created from the ANTs structural image segmentation, and tissue-based masks were visually reviewed for adequate gray matter/white matter classification. For voxel-based analysis, relative CBF (rCBF) maps were created for each participant by normalizing to the mean gray matter CBF. Each subject's rCBF map was registered to the 2-mm³ template using a single subject-specific transformation obtained from concatenating the ASL-to-T1 registration matrix with the T1-to-template warp field.

z-score maps for each TBI participant were created in template space relative to the voxel-based mean and standard deviation of the pooled rCBF data from all controls. z-score maps for each control participant were created using the leave-one-out approach,³² excluding that participant's data from the pooled rCBF data. Voxels with z-score ≥2.5 were identified as abnormally hyperperfused, whereas voxels with z-score ≤–2.5 were identified as abnormally hypoperfused. The z-score threshold of ±2.5 was chosen based on maximization of the observed effect size (Cohen's d) in discriminating TBI participants from controls³³ (d = 3.0 for hypoperfusion, d = −2.9 for hyperperfusion). A cluster size threshold of five contiguous voxels (40 mm³) was also imposed to reduce spurious extrema.

Total number and volume of abnormally hypo- and hyperperfused clusters were extracted using FSL and used in subsequent analysis. In TBI participants scanned longitudinally, mean rCBF within hypo- and hyperperfused clusters, based on the acute scan, was calculated. The same regions were identified in the chronic scan in order to assess change in mean rCBF within these clusters from the acute to chronic phase of injury.

Statistical analysis

Global, gray matter, and white matter mean CBF were compared between acute and chronic TBI participants and HCs. Number and volume of hypo- and hyperperfused clusters were also compared between acute and chronic TBI participants and HCs. In TBI participants scanned longitudinally, tissue- and voxel-based measures of CBF were compared between the 2-week and 6-month time points. Group-level comparisons were performed using Mann-Whitney U tests, and longitudinal comparisons were performed using mixed-effects linear modeling, covaried for age and years of education. In the models, time was treated as a continuous variable, and all models included a random intercept. Model fit was assessed using Q-Q and residual plots. Associations between CBF measures and clinical outcomes were assessed using unpaired t-tests, comparing participants with complete (GOSE-TBI = 8) and incomplete (GOSE-TBI <8) recovery and comparing participants with total RPQ score above and below the median. Clinical outcomes data were dichotomized because of small sample size and variability in measurements. Statistical analysis was performed using GraphPad (Mann-Whitney U tests; GraphPad Software Inc., La Jolla, CA) and R software (the lme4 package was used for mixed-effects linear modeling; R Foundation for Statistical Computing, Vienna, Austria).

Results

We analyzed 54 unique participants, including 33 TBI participants scanned in the acute phase (median 17 [interquartile range {IQR}, 14–22] days post-injury); 16 TBI participants scanned in the chronic phase (median 195 [IQR, 187.0–201.5] days post-injury), of which 3 persons were scanned only in the chronic phase; and 18 age- and sex-matched controls. Of 33 acute TBI participants, 13 completed a second MRI 6 months after injury (195 [179.0–201.5] days post-injury), and 19 (57.6%) had 6-month clinical outcome data. The majority of participants were Caucasian and male. HCs had significantly more years of education compared to acute TBI participants (mean 16.4 vs. 14.2 years; p = 0.02). Most participants (81.8%) had positive findings on head CT, including extra-axial blood, subarachnoid hemorrhage, and parenchymal contusion. Overall, the cohort exhibited good recovery at 6 months post-injury (GOSE-TBI 7 [7–8]) and reported low-to-medium post-concussive symptoms (RPQ 10 [1–13]). Additional demographic and clinical data are presented in Table 1.

Table 1.

Demographic and Clinical Information

	TBI-Acute	TBI-Chronic	TBI-Longitudinal^a	HC
n	33	16	13	18
Demographic information
GCS, n (%)
13–15	33 (100)	16 (100)	13 (100)
Injury cause, n (%)
Road traffic accident	15 (45.5)	8 (50)	6 (46.2)
Accidental fall	10 (30.3)	5 (31.3)	4 (30.8)
Violence/assault	5 (15.2)	2 (12.5)	2 (15.4)
Other	3 (9.1)	1 (6.3)	1 (7.7)
Race, n (%)
White	20 (60.6)	11 (68.8)	9 (69.2)	11 (61.1)
Black	11 (33.3)	4 (25)	3 (23.1)	3 (1.7)
Asian	2 (6.1)	1 (6.3)	1 (7.7)	3 (1.7)
Other	0 (0)	0 (0)	0 (0)	1 (0.6)
% male	81.8	81.3	76.9	83.3
Age at injury (years, mean ± SD)	41.0 ± 15.9	36.8 ± 16.5	37.1 ± 14.7	36.6 ± 6.5
Time to scan (days, median [IQR])	17 [14–22]	195 [187.0–201.5]	195 [179.0–201.5]
Education (years, mean ± SD)	14.2 ± 2.5	13.8 ± 2.9	14.3 ± 2.1	16.4 ± 2.8
CT abnormalities, n (%)	27 (81.8)	12 (75)	10 (76.9)
Extra-axial blood	26 (78.8)	8 (50)	8 (61.5)
Subarachnoid hemorrhage	17 (51.5)	8 (50)	7 (53.8)
Parenchymal contusion	10 (30.3)	6 (37.5)	5 (38.5)
Presence of LOC, n (%)	24 (72.7)	12 (75)	10 (76.9)
Length of hospital stay (days, mean ± SD)	2.5 ± 1.8	2.1 ± 1.5	2.0 ± 1.4
Clinical outcomes (median [IQR])
n	22	11	10
2-week GOSE-TBI	6 [5–7]	6 [5–7]	5.5 [5–7]
2-week RPQ	14 [8–27]	14 [10–29]	18.5 [13–29]
n	19	12	11
6-month GOSE-TBI	7 [7–8]	7 [7–8]	7 [7–8]
6-month RPQ	10 [1–13]	7 [1–12.5]	9 [1–13]

Participants in the TBI-Longitudinal arm were scanned in both the acute and chronic phases of injury and are also included in the count for the TBI-Acute and TBI-Chronic arms.

TBI, traumatic brain injury; HC, healthy control group; GCS, Glasgow Coma Scale; SD, standard deviation; IQR, interquartile range; CT, computed tomography; LOC, loss of consciousness; GOSE-TBI, Glasgow Outcome Scale-Extended, measured based on disability attributable to TBI; RPQ, Rivermead Post-Concussion Symptom Questionnaire.

Cross-sectional analysis

Mean global and gray matter CBF were significantly increased in the acute and chronic TBI groups compared to controls (global CBF [median]: 30.1, 35.3, and 33.5 mL/min/100 g for HC, TBI-Acute [p = 0.003], and TBI-Chronic [p = 0.03], respectively; gray matter CBF: 37.0, 42.3, and 40.6 mL/min/100 g for HC, TBI-Acute [p = 0.01], and TBI-Chronic [p = 0.04], respectively; Fig. 1). Mean white matter CBF was not significantly different between groups (21.6, 22.9, and 23.6 mL/min/100 g for HC, TBI-Acute [p = 0.2], and TBI-Chronic [p = 0.2], respectively). Similar findings were observed after excluding participants with intraparenchymal lesions on head CT (Supplementary Table S1). There was no significant difference in mean global, gray matter, or white matter CBF in acute TBI participants with and without intraparenchymal lesions on head CT (Supplementary Table S2). There was no significant difference in mean global, gray matter, or white matter CBF based on level of education (>12 vs. ≤12 years; Supplementary Table S3). There was no significant correlation between years of education and CBF when assessed within each arm.

FIG. 1.

Cerebral blood flow is significantly increased in TBI compared to HCs. (A) Mean global absolute CBF was significantly increased in acute and chronic TBI participants compared to controls (HC; HC [median]: 30.1; TBI-Acute: 35.3 [p = 0.003]; TBI-Chronic: 33.5 mL/min/100 g [p = 0.03]). (B) Similar findings were observed in gray matter CBF (HC: 37.0;; TBI-Acute: 42.3 [p = 0.01]; TBI-Chronic: 40.6 mL/min/100 g [p = 0.04]). (C) White matter CBF was not significantly different between HC and TBI groups (HC: 21.6; TBI-Acute: 22.9 [p = 0.2]; TBI-Chronic: 23.6 mL/min/100 g [p = 0.2]). Median, interquartile range, and minimum/maximum are indicated by the box-and-whiskers plot. *p < 0.05. CBF, cerebral blood flow; HCs, healthy controls; TBI, traumatic brain injury. Color image is available online.

Using a z-score cutoff of ±2.5 and cluster size threshold of 40 mm³, total volume of abnormally hypoperfused (z ≤ −2.5) and hyperperfused (z ≥ 2.5) tissue was significantly higher in acute TBI participants compared to controls (volume of hypoperfused tissue [median]: 126.6, 345.1, and 154.4 mm³ for HCs, TBI-Acute [p = 0.004], and TBI-Chronic [p = 0.3], respectively; volume of hyperperfused tissue: 184.1, 421.1, and 252.5 mm³ for HC, TBI-Acute [p = 0.03], and TBI-Chronic [p = 0.5], respectively; Fig. 2). Number of hypoperfused clusters was also higher in acute TBI participants, although not statistically significant (number of hypoperfused clusters [median]: 34.5, 49.0, and 39.0 for HCs, TBI-Acute [p = 0.08], and TBI-Chronic [p = 0.4], respectively; number of hyperperfused clusters: 46.0, 56.0, and 44.0 for HCs, TBI-Acute [p = 0.14], and TBI-Chronic [p = 0.5], respectively). Based on visual inspection of all acute TBI scans, a mean of 1.4 ± 2.0 clusters was associated with visible T1 or FLAIR lesions. Clusters of abnormal perfusion were observed in both gray and white matter and exhibited focal and multi-focal patterns (Fig. 3). No significant difference in voxel-based measures was noted in chronic TBI participants compared to controls.

FIG. 2.

Voxel-based analysis reveals increased volume of abnormal perfusion in acute TBI compared to healthy controls. Voxel-based analysis was used to identify clusters of abnormal perfusion, using a z-score cutoff of ±2.5 and a size cutoff of five contiguous voxels (40 mm³). Acute TBI participants exhibited (A) significantly greater volume of relative hypoperfusion compared to controls (HC; HC [median {IQR}]: 126.6 [50.2–240.3]; TBI-Acute: 345.1 [131.4–767.3; p = 0.004]; TBI-Chronic: 154.4 [89.7–528.6] mm³ [p = 0.3]), as well as (B) significantly greater volume of relative hyperperfusion compared to controls (HC: 184.1 [128.0–551.1]; TBI-Acute: 421.1 [192.7–1057.0; p = 0.03]; TBI-Chronic: 252.5 [133.6–754.6] mm³ [p = 0.5]). No significant differences were noted in chronic TBI participants compared to controls. Log-transformed volume is shown in log mm³. Median, IQR, and minimum/maximum are indicated by the box-and-whiskers plot. *p < 0.05. HCs, healthy controls; IQR, interquartile range; TBI, traumatic brain injury. Color image is available online.

FIG. 3.

Examples of rCBF maps in the acute and chronic phases. Acute CT and longitudinal FLAIR images, rCBF maps, z-score maps, and thresholded clusters are shown (blue clusters indicate relative hypoperfusion, or z < −2.5; red clusters indicate relative hyperperfusion, or z > 2.5). (A) Example of multi-focal reduction in rCBF with no TBI-related abnormalities on FLAIR imaging: Participant 1 is a 34-year-old male who was in a motor vehicle accident, with positive loss of consciousness and GCS of 15 on arrival. Initial CT showed evidence of incidental cavernomas, more clearly observed on MRI, but no TBI-related lesions. GOSE-TBI at 6 months was 7. Regions of multi-focal hypo- and hyperperfusion are seen in the acute rCBF map, unrelated to incidental cavernomas, and partially resolve by the chronic phase. (B) Example of focal reduction in rCBF: Participant 2 is a 28-year-old male who fell 20 feet, with positive loss of consciousness and GCS of 13 on arrival. Initial CT showed left frontal subarachnoid hemorrhage (white arrow on CT), left subdural hematoma, and 5-mm rightward midline shift. GOSE-TBI at 6 months was 7. Acute reduction in rCBF is seen in the left frontal lobe (more clearly seen in the z-score map, black arrow), associated with and surrounding a small parenchymal contusion (white arrow on FLAIR). Partial normalization of this focal reduction is observed by 6 months post-injury. (C) Example of focal reduction in rCBF: Participant 3 is a 57-year-old male who was assaulted, with no loss of consciousness and GCS of 15 on arrival. Initial CT demonstrated a hemorrhagic parenchymal contusion in the left temporal lobe (white arrow on CT). GOSE-TBI at 6 months was 7. Acute reduction in rCBF is seen at the left temporal contusion (more clearly seen in the z-score map, black arrow) and surrounding brain tissue, corresponding to the FLAIR lesion (white arrow on FLAIR). Partial normalization of this focal reduction is observed by 6 months post-injury, although volume of hyperperfusion is increased in the chronic phase in this participant. CT, computed tomography; FLAIR, fluid-attenuated inversion recovery; GCS, Glasgow Coma Scale; GOSE, Glasgow Outcome Sclae-Extended; MRI, magnetic resonance imaging; rCBF, relative cerebral blood flow; TBI, traumatic brain injury. Color image is available online.

Longitudinal analysis

In TBI participants scanned longitudinally, at 2 weeks and 6 months post-injury (n = 13), global, gray matter, and white matter CBF did not change significantly over time (Supplementary Fig. S1). There was a weak correlation between 2-week and 6-month mean gray matter CBF (Supplementary Fig. S2).

Total volume of hypo- and hyperperfused tissue decreased from 2 weeks to 6 months post-injury, reaching statistical significance for hypoperfused tissue only (Fig. 4). Mean rCBF within clusters of hypo- and hyperperfusion, defined in the acute scan, partially normalized by 6 months post-injury, increasing within regions of acute hypoperfusion and decreasing within regions of acute hyperperfusion (Fig. 5). rCBF normalization in hypo- and hyperperfused clusters was observed in association with both anatomical lesions and structurally normal brain tissue on T1 and FLAIR imaging (Fig. 3). There was a strong positive correlation between 2-week and 6-month mean rCBF within acute clusters for each participant (Pearson, r = 0.72 for hypoperfused clusters; r = 0.72 for hyperperfused clusters), as well as in the pool of clusters from all participants (r = 0.89 for hypoperfused clusters; r = 0.83 for hyperperfused clusters; Supplementary Fig. S3).

FIG. 4.

Voxel-based analysis reveals normalization of regions of abnormal hypoperfusion from 2 weeks to 6 months post-injury. Voxel-based analysis was used to identify clusters of abnormal perfusion in patients who were scanned longitudinally, using a z-score cutoff of ±2.5 and a size cutoff of five contiguous voxels (40 mm³). (A) After adjusting for age and years of education, volume of hypoperfusion was found to decrease significantly over time (estimated change of −0.14 [log mm³]/month; p = 0.04). (B) Volume of hyperperfusion was also found to decrease over time, though the difference was not statistically significant (estimated change of −0.075 [log mm³]/month, p = 0.31). Each line represents an individual participant. ß is the coefficient for the “time since injury” variable, representing estimated change in log-transformed volume of hypo-/hyperperfusion (mm³) per month. Color image is available online.

FIG. 5.

Mean rCBF within abnormal clusters normalizes from 2 weeks to 6 months post-injury. Voxel-based analysis was used to identify clusters of abnormal perfusion in participants who were scanned longitudinally, using a z-score cutoff of ±2.5 and a size cutoff of five contiguous voxels (40 mm³). (A) After adjusting for age and years of education, the mean rCBF was found to increase significantly within hypoperfused clusters (estimated change of 0.033/month; p < 0.0001). (B) Mean rCBF was found to decrease significantly within hyperperfused clusters (estimated change of −0.037/month; p < 0.0001). Each line represents an individual participant. ß is the coefficient for the “time since injury” variable, representing estimated change in rCBF per month. rCBF, relative cerebral blood flow; TBI, traumatic brain injury. Color image is available online.

Correlation with clinical outcome

Of 19 participants with available clinical outcome data, all had 6-month GOSE-TBI of ≥6 (6, upper moderate recovery; 7, lower good recovery; 8, complete recovery), as expected given the relative insensitivity of GOSE for subtle disabilities in patients with relatively mild brain injuries. Participants with complete recovery had higher gray matter CBF at 2 weeks post-injury, which trended toward significance (GOSE = 8 [median]: 45.1 mL/min/100 g, GOSE <8: 38.8 mL/min/100 g; p = 0.08; Fig. 6A). Participants with lower symptom burden also had higher gray matter CBF at 2 weeks post-injury, which trended toward significance (RPQ ≤ median: 45.1 mL/min/100 g, RPQ > median: 38.8 mL/min/100 g; p = 0.09; Fig. 6B). Participants with better outcome had significantly higher mean gray matter CBF compared to controls (GOSE = 8 vs. controls: p = 0.01; RPQ > median vs. controls: p = 0.009), whereas participants with worse outcome were not significantly different from controls. Similar findings were observed with global CBF. As expected, clinical recovery was inversely correlated with symptom burden (Pearson r = −0.41, p = 0.06), and participants with complete clinical recovery at 6 months had significantly lower symptom score compared to those with incomplete recovery (complete recovery: RPQ [median] = 1.54.4; incomplete recovery: RPQ = 11.5; p = 0.009). No significant relationship was observed between voxel-based measures of CBF and clinical outcome measures.

FIG. 6.

Clinical recovery at 6 months post-injury is associated with greater gray matter CBF in the acute phase. (A) Participants who reached complete recovery at 6 months post-injury (GOSE-TBI = 8) had higher mean gray matter CBF at 2 weeks post-injury compared to those who had incomplete recovery (GOSE-TBI <8), which trended toward statistical significance (GOSE = 8 [median]: 45.1 mL/min/100 g, GOSE <8: 38.8 mL/min/100 g; p = 0.08). (B) Similarly, participants with lower symptom burden at 6 months post-injury (RPQ ≤ median [10]) also had higher mean gray matter CBF at 2 weeks post-injury compared to those with greater symptom burden, which trended toward statistical significance (RPQ ≤ median: 45.1 mL/min/100 g; RPQ > median: 38.8 mL/min/100 g; p = 0.09). Median, interquartile range, and minimum/maximum are indicated by the box-and-whiskers plot. Dotted line indicates the median 2-week gray matter CBF of HCs (37.0 mL/min/100 g). *p < 0.05, relative to HCs. CBF, cerebral blood flow; GOSE-TBI, Glasgow Outcome Scale-Extended, measured based on disability attributable to TBI; HCs, healthy controls; RPQ, Rivermead Post-Concussion Symptom Questionnaire. Color image is available online.

Discussion

We found that absolute CBF, as determined using ASL, is globally increased in the acute and chronic phases of injury in a relatively mildly injured cohort of TBI participants. Participants with better outcome at 6 months post-injury tended to have higher global CBF in the acute phase. In TBI participants scanned longitudinally, we observed coexisting regions of abnormal hypo- and hyperperfusion in the acute phase of injury, which partially normalized by 6 months post-injury. There was a strong correlation between 2-week and 6-month mean CBF within these abnormal clusters, and only a weak correlation between 2-week and 6-month global CBF, suggesting that the observed normalization was not attributable to regression-to-the-mean.

Changes in CBF after TBI are commonly reported, but the direction of change has not been consistent, in part because of variability in methodological factors. In our relatively mildly injured cohort, we found that global CBF was increased at the 2-week and 6-month time points after injury. In addition, we found that better outcome was associated with global elevation of CBF in the acute phase. Notably, the mean global CBF of participants with worse outcome was comparable to that of controls, whereas the mean global CBF of participants with better outcome was significantly increased relative to controls. Based on these findings, we hypothesize that this observed elevation in perfusion in our relatively mildly injured study population may represent an adaptive reparative mechanism, achieved only in a subset of participants. We also find significantly greater volume of abnormal perfusion in TBI participants in the acute post-injury phase compared to controls, including clusters of both hypo- and hyperperfusion, which trended toward normalization by 6 months post-injury. Importantly, although some of these regions are associated with structural injury observed on T1/FLAIR sequences, most are distinct, reflecting more diffuse injury effects than those detected using standard MRI protocols.

Several studies have reported reduced CBF after TBI, and hypoperfusion and cerebral ischemia are strongly linked to negative outcomes after severe neurological insult,³⁴ potentially occuring as a secondary downstream sequelae of the initial structural injury and autoregulatory failure in the face of increased metabolic demand.²³ However, findings of post-injury hypoperfusion are more commonly demonstrated in the hyperacute phase of injury,^14,17,35 as well as in more-severe injuries, whereas hyperperfusion is often observed after milder injuries.^18,19,36,37 Although hyperemia has also been linked to detrimental effects, such as vasospasm and increased intracranial pressure, particularly in more-severe injuries,²³ we hypothesize that the observed post-injury hyperperfusion in our relatively mildly injured cohort does not primarily reflect vascular damage or autoregulatory failure, but rather represents an appropriate compensatory metabolic or inflammatory response as a result of injury. Ultimately, additional research assessing metabolic demand at the time of CBF rise will be necessary to distinguish blood flow alterations because of metabolic derangements from those attributable to primary vascular dysfunction. Cerebrovascular reactivity (CVR) mapping, which provides a more-specific assessment of microvascular function, will be helpful in making this distinction and is an area of active investigation.

This study was limited by the small number of participants scanned longitudinally. Additionally, we do not account for participant characteristics, including substance use, smoking history, or body mass index, given that this exploratory study was not powered to examine all potential confounders. This will be an important consideration for future work. Analyzing regional distribution of abnormally perfused clusters within a larger study population will also be important in identifying brain regions particularly susceptible to changes in perfusion after injury. Additionally, evaluating the relationship between CBF and associated volumetric changes will be useful in assessing long-term sequelae of TBI. We did not mask focal lesions when quantifying tissue-based CBF measures and recognize that large lesions may disproportionately decrease CBF measures. However, in 3 participants with large FLAIR lesions on imaging, we observed an increase, rather than decrease, in global CBF relative to HCs. Further, there was no significant difference in global CBF in participants with and without intraparenchymal blood on CT, suggesting that focal lesions did not significantly contribute to our findings in the present study.

Finally, numerous vascular-directed therapies exist in medicine and are beginning to be applied to TBI in both pre-clinical and clinical studies.³⁸ Our group previously demonstrated that phosphodiesterase-5 inhibitors, commonly used in erectile dysfunction and primary pulmonary hypertension, can transiently rescue CVR in chronic TBI patients, with a trend toward clinical improvement.³⁹ Future work may aim to determine whether similar changes in ASL-derived CBF can be detected in response to vascular-directed therapies.

Using ASL to quantify and characterize changes in cerebral perfusion after TBI provides diagnostic and prognostic insights that may not be accessible through conventional neuroimaging techniques. Identifying patients with alterations in vascular function after injury will not only allow physicians to target clinical follow-up to patients more likely to have poor outcome, but also enable more-precise patient stratification in clinical trial enrollment in order to hasten the development of new therapeutic interventions for TBI.

Footnotes

Authors' Contributions

L.X., J.B.W., R.D.A., and D.K.S. conceived and designed the study, analyzed and interpreted the data, and prepared the manuscript. J.J.K., C.E.L., D.P., H.I.C., and J.S. contributed to the revision of the manuscript and approved its submission for publication. E.S., B.M., C.D., L.W., M.D.L., J.M., and H.Z. contributed to data collection and data integrity. P.S. contributed to data analysis.

Funding Information

Work in the authors' laboratory was supported by the Pennsylvania Department of Health. This work was also funded by NINDS U01 NS086090, DoD W81XWH-14-2-0176, W81XWH-19-2-0002, and NINDS K23-NS104239-03.

Author Disclosure Statement

No competing financial interests exist.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Table S3

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

References

Sandsmark

D.K.

(2016). Clinical outcomes after traumatic brain injury. Curr. Neurol. Neurosci. Rep. 16, 52.

Diaz-Arrastia

, Kochanek

P.M.

, Bergold

, Kenney

, Marx

C.E.

, Grimes

J.B.

, Loh

, Adam

G.E.

, Oskvig

, Curley

K.C.

, and Salzer

(2014). Pharmacotherapy of traumatic brain injury: state of the science and the road forward: report of the Department of Defense Neurotrauma Pharmacology Workgroup. J. Neurotrauma, 31, 135–158.

Saatman

K.E.

, Duhaime

, Bullock

, Maas

A.I.

, Valadka

, and Manley

G.T.

(2008). Classification of traumatic brain injury for targeted therapies. J. Neurotrauma, 25, 719–738.

Lee

, Wintermark

, Gean

A.D.

, Ghajar

, Manley

G.T.

, and Mukherjee

(2008). Focal lesions in acute mild traumatic brain injury and neurocognitive outcome: CT versus 3T MRI. J. Neurotrauma, 25, 1049–1056.

Haacke

E.M.

, Duhaime

A.C.

, Gean

A.D.

, Riedy

, Wintermark

, Mukherjee

, Brody

D.L.

, DeGraba

, Duncan

T.D.

, Elovic

, Hurley

, Latour

, Smirniotopoulos

J.G.

, and Smith

D.H.

(2010). Common data elements in radiologic imaging of traumatic brain injury. J. Magn. Reson. Imaging, 32, 516–543.

Kenney

, Amyot

, Haber

, Pronger

, Bogoslovsky

, Moore

, and Diaz-Arrastia

(2016). Cerebral vascular injury in traumatic brain injury. Exp. Neurol. 275, 353–366.

Sandsmark

D.K.

, Bashir

, Wellington

C.L.

, and Diaz-Arrastia

(2019). Cerebral microvascular injury: a potentially treatable endophenotype of traumatic brain injury-induced neurodegeneration. Neuron, 103, 367–379.

Haller

, Zaharchuk

, Thomas

D.L.

, Lovblad

, Barkhof

, and Golay

(2016). Arterial spin labeling perfusion of the brain: emerging clinical applications. Radiology, 281, 337–356.

Wang

, West

J.D.

, Bailey

J.N.

, Westfall

D.R.

, Xiao

, Arnold

T.W.

, Kersey

P.A.

, Saykin

A.J.

, and McDonald

B.C.

(2015). Decreased cerebral blood flow in chronic pediatric mild TBI: an MRI perfusion study. Dev. Neuropsychol. 40, 40–44.

10.

Clark

A.L.

, Weigand

A.J.

, Bangen

K.J.

, Merritt

V.C.

, Bondi

M.W.

, and Delano-Wood

(2020). Repetitive mTBI is associated with age-related reductions in cerebral blood flow but not cortical thickness. J. Cereb. Blood Flow Metab. 41, 431–444.

11.

Bonne

, Gilboa

, Louzoun

, Kempf-Sherf

, Katz

, Fishman

, Ben-Nahum

, Krausz

, Bocher

, Lester

, Chisin

, and Lerer

(2003). Cerebral blood flow in chronic symptomatic mild traumatic brain injury. Psychiatry Res. 124, 141–152.

12.

Launey

, Fryer

T.D.

, Hong

Y.T.

, Steiner

L.A.

, Nortje

, Veenith

T.V.

, Hutchinson

P.J.

, Ercole

, Gupta

A.K.

, Aigbirhio

F.I.

, Pickhard

J.D.

, Coles

J.P.

, and Menon

D.K.

(2020). Spatial and temporal pattern of ischemia and abnormal vascular function following traumatic brain injury. JAMA Neurol. 77, 339–349.

13.

Kim

, Whyte

, Patel

, Avants

, Europa

, Wang

, Slattery

, Gee

J.C.

, Coslett

H.B.

, and Detre

J.A.

(2010). Resting cerebral blood flow alterations in chronic traumatic brain injury: an arterial spin labeling perfusion FMRI study. J. Neurotrauma, 27, 1399–1411.

14.

Wang

, Nelson

L.D.

, LaRoche

A.A.

, Pfaller

A.Y.

, Nencka

A.S.

, Koch

K.M.

, and McCrea

M.A.

(2016). Cerebral blood flow alterations in acute sport-related concussion. J. Neurotrauma, 33, 1227–1236.

15.

Ware

J.B.

, Dolui

, Duda

, Gaggi

, Choi

, Detre

, Whyte

, Diaz-Arrastia

, and Kim

J.J.

(2020). Relationship of cerebral blood flow to cognitive function and recovery in early chronic traumatic brain injury. J. Neurotrauma, 37, 2180–2187.

16.

Soustiel

J.F.

, Glenn

T.C.

, Shik

, Boscardin

, Mahamid

, and Zaaroor

(2005). Monitoring of cerebral blood flow and metabolism in traumatic brain injury. J. Neurotrauma, 22, 955–965.

17.

Wang

, Nencka

A.S.

, Meier

T.B.

, Guskiewicz

, Mihalik

J.P.

, Brooks

M.A.

, Saykin

A.J.

, Koch

K.M.

, Wu

Y.-C.

, Nelson

L.D.

, McAllister

T.W.

, Broglio

S.P.

, and McCrea

M.A.

(2019). Cerebral blood flow in acute concussion: preliminary ASL findings from the NCAA-DoD CARE consortium. Brain Imaging Behav. 13, 1375–1385.

18.

Churchill

N.W.

, Hutchison

M.G.

, Graham

S.J.

, and Schweizer

T.A.

(2017). Symptom correlates of cerebral blood flow following acute concussion. Neuroimage Clin. 16, 234–239.

19.

Stephens

J.A.

, Liu

, Lu

, and Suskauer

S.J.

(2018). Cerebral blood flow after mild traumatic brain injury: associations between symptoms and post-injury perfusion. J. Neurotrauma, 35, 241–248.

20.

Henriksen

O.M.

, Larsson

H.B.

, Hansen

A.E.

, Grüner

J.M.

, Law

, and Rostrup

(2012). Estimation of intersubject variability of cerebral blood flow measurements using MRI and positron emission tomography. J. Magn. Reson. Imaging, 35, 1290–1299.

21.

Liu

A.A.

, Voss

H.U.

, Dyke

J.P.

, Heier

L.A.

, and Schiff

N.D.

(2011). Arterial spin labeling and altered cerebral blood flow patterns in the minimally conscious state. Neurology, 77, 1518–1523.

22.

Andriessen

T.M.

, Jacobs

, and Vos

P.E.

(2010). Clinical characteristics and pathophysiological mechanisms of focal and diffuse traumatic brain injury. J. Cell. Mol. Med. 14, 2381–2392.

23.

Werner

, and Engelhard

(2007). Pathophysiology of traumatic brain injury. Br. J. Anaesth. 99, 4–9.

24.

Management of Concussion/mTBI Working Group. (2009). VA/DoD clinical practice guideline for management of concussion/mild traumatic brain injury. J. Rehabil. Res. Dev. 46, CP1–CP68.

25.

Jennett

, Snoek

, Bond

M.R.

, and Brooks

(1981). Disability after severe head injury: observations on the use of the Glasgow Outcome Scale. J. Neurol. Neurosurg. Psychiatry, 44, 285–293.

26.

King

N.S.

, Crawford

, Wenden

F.J.

, Moss

, and Wade

D.T.

(1995). The Rivermead Post Concussion Symptoms Questionnaire: a measure of symptoms commonly experienced after head injury and its reliability. J. Neurol. 242, 587–592.

27.

Chappell

M.A.

, Groves

A.R.

, Whitcher

, and Woolrich

M.W.

(2008). Variational bayesian inference for a nonlinear forward model. IEEE Trans. Sig. Proc. 57, 223–236.

28.

Jenkinson

, Beckmann

C.F.

, Behrens

T.E.

, Woolrich

M.W.

, and Smith

S.M.

(2012). Fsl. Neuroimage. 62, 782–790.

29.

Alsop

D.C.

, Detre

J.A.

, Golay

, Günther

, Hendrikse

, Hernandez-Garcia

, Lu

, MacIntosh

B.J.

, Parkes

L.M.

, Smits

, van Osch

M.J.P.

, Wang

D.J.J.

, Wong

E.C.

, and Zaharchuk

(2015). Recommended implementation of arterial spin-labeled perfusion MRI for clinical applications: a consensus of the ISMRM Perfusion Study Group and the European Consortium for ASL in Dementia. Magn. Reson. Med. 73, 102–116.

30.

Avants

B.B.

, Tustison

N.J.

, Song

, Cook

P.A.

, Klein

, and Gee

J.C.

(2011). A reproducible evaluation of ANTs similarity metric performance in brain image registration. Neuroimage, 54, 2033–2044.

31.

Avants

B.B.

, Epstein

C.L.

, Grossman

, and Gee

J.C.

(2008). Symmetric diffeomorphic image registration with cross-correlation: evaluating automated labeling of elderly and neurodegenerative brain. Med. Image Anal. 12, 26–41.

32.

Friedl

, and Stampfer

(2014). Jackknife resampling. Wiley StatsRef: Statistics Reference. Online.

33.

Amyot

, Kenney

, Moore

, Haber

, Turtzo

L.C.

, Shenouda

, Silverman

, Gong

, Qu

B.-X.

, Harburg

, Lu

H.Y.

, Wassermann

E.M.

, and Diaz-Arrastia

(2018). Imaging of cerebrovascular function in chronic traumatic brain injury. J. Neurotrauma, 35, 1116–1123.

34.

Jeremitsky

, Omert

, Dunham

C.M.

, Protetch

, and Rodriguez

(2003). Harbingers of poor outcome the day after severe brain injury: hypothermia, hypoxia, and hypoperfusion. J. Trauma Acute Care Surg. 54, 312–319.

35.

Meier

T.B.

, Bellgowan

P.S.

, Singh

, Kuplicki

, Polanski

D.W.

, and Mayer

A.R.

(2015). Recovery of cerebral blood flow following sports-related concussion. JAMA Neurol. 72, 530–538.

36.

Doshi

, Wiseman

, Liu

, Wang

, Welch

R.D.

, O'Neil

B.J.

, Zuk

, Wang

, Mika

, Szaflarski

J.P.

, Haacke

E.M.

, and Kou

(2015). Cerebral hemodynamic changes of mild traumatic brain injury at the acute stage. PloS One, 10, e0118061.

37.

Sours

, Zhuo

, Roys

, Shanmuganathan

, and Gullapalli

R.P.

(2015). Disruptions in resting state functional connectivity and cerebral blood flow in mild traumatic brain injury patients. PLoS One, 10, e0134019.

38.

Sandsmark

D.K.

, Bashir

, Wellington

C.L.

, and Diaz-Arrastia

(2019). Cerebral microvascular injury: a potentially treatable endophenotype of traumatic brain injury-induced neurodegeneration. Neuron, 103, 367–379.

39.

Kenney

, Amyot

, Moore

, Haber

, Turtzo

L.C.

, Shenouda

, Silverman

, Gong

, Qu

B.-X.

, Harburg

, Wassermann

E.M.

, Lu

, and Diaz-Arrastia

(2018). Phosphodiesterase-5 inhibition potentiates cerebrovascular reactivity in chronic traumatic brain injury. Ann. Clin. Transl. Neurol. 5, 418–428.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.01 MB

0.39 MB

0.50 MB

0.91 MB