Chronic elevation of plasma vascular endothelial growth factor-A (VEGF-A) is associated with a history of blast exposure

Abstract

Mounting evidence points to the significance of neurovascular-related dysfunction in veterans with blast-related mTBI, which is also associated with reduced [¹⁸F]-fluorodeoxyglucose (FDG) uptake.

The goal of this study was to determine whether plasma VEGF-A is altered in veterans with blast-related mTBI and address whether VEGF-A levels correlate with FDG uptake in the cerebellum, a brain region that is vulnerable to blast-related injury 72 veterans with blast-related mTBI (mTBI) and 24 deployed control (DC) veterans with no lifetime history of TBI were studied. Plasma VEGF-A was significantly elevated in mTBIs compared to DCs. Plasma VEGF-A levels in mTBIs were significantly negatively correlated with FDG uptake in cerebellum. In addition, performance on a Stroop color/word interference task was inversely correlated with plasma VEGF-A levels in blast mTBI veterans. Finally, we observed aberrant perivascular VEGF-A immunoreactivity in postmortem cerebellar tissue and not cortical or hippocampal tissues from blast mTBI veterans.

These findings add to the limited number of plasma proteins that are chronically elevated in veterans with a history of blast exposure associated with mTBI. It is likely the elevated VEGF-A levels are from peripheral sources. Nonetheless, increasing plasma VEGF-A concentrations correlated with chronically decreased cerebellar glucose metabolism and poorer performance on tasks involving cognitive inhibition and set shifting. These results strengthen an emerging view that cognitive complaints and functional brain deficits caused by blast exposure are associated with chronic blood-brain barrier injury and prolonged recovery in affected regions.

1. Introduction

Diagnostic and prognostic evaluation of individuals exposed to blast overpressures resulting in mild traumatic brain injury (TBI) relies on neurologic examination, cognitive testing, and brain imaging [1]. In spite of impressive advances [2], there remain significant gaps in our understanding about relationships between postconcussive cognitive status and central nervous system (CNS) dysfunction. Closing this knowledge gap is important for improving how latent injury progression is monitored, determining long term outcomes, and furthering mechanistic insights. Significant strides have been made in identifying molecular biomarkers found in blood, cerebrospinal fluid (CSF), and other biofluids from individuals with acute moderate to severe TBI [3]. However, less is known about candidate biofluid biomarkers capable of informing brain function in living individuals years after blast injury, although progress is accelerating rapidly [4–6]. Thus far, the many of the best characterized blood and CSF biomarkers normalize within days to weeks following a mTBI.

Blast exposure is increasingly recognized to be associated with chronic cerebrovascular pathology. A growing body of findings high-light the importance of blast-induced disturbances affecting the network of endothelial cells and perivascular astrocytes that comprise the neurovascular system of the brain [7–11]. Multiple preclinical findings indicate that blast exposure alters the level of vascular endothelial growth factor-A (VEGF-A), a key angiogenic factor regulating vascular endothelial survival, growth, and remodeling [12–15]. These findings, coupled with work showing that blast disturbs blood-brain barrier (BBB) integrity [9,10,16–18], which is also regulated by VEGF-A [19], and that a history of blast exposure disrupts regulatory networks inhibiting VEGF-A expression [18] prompted us to test the hypothesis that plasma VEGF-A is altered in veterans with chronic blast-related mTBI.

2. Materials and methods

2.1. Study participants

Studies were approved by the VA Puget Sound Human Subjects Committee and conformed to institutional regulatory guidelines and principles of human subjects protections in the Declaration of Helsinki. Written informed consent was obtained for all study participants. Lifetime history of both blast-related and impact-related mTBI was obtained using a semi-structured interview by two expert TBI clinicians described in detail elsewhere [20]. All participants were male. Females were eligible for study inclusion, but no females with blast-related mTBI enrolled. For inclusion in the mTBI group, participants must have had at least one blast exposure with acute symptoms meeting VA/DoD/American Congress of Rehabilitation Medicine criteria for mTBI. History of subconcussive blast exposure was not acquired. Study exclusion criteria included moderate-severe TBI, seizure disorder, insulin-dependent diabetes, current DSM-IV diagnosis of alcohol abuse or other substance abuse, schizophrenia or other psychotic disorders, bipolar disorder, dementia, and taking medications likely to affect cognition.

2.2. Positron emission tomography (PET)

Standard clinical PET acquisitions were performed 60 min after injection of [¹⁸F]-FDG (8–10 mCi) on either a GE Advance PET or Philips Gemini PET/CT scanner. Data from GE underwent 3D-filtered back projection reconstruction, while Philips data underwent OSEM reconstruction. Three 5 min frames were averaged following motion correction, smoothed using an 8×8×8mm Gaussian filter, then transformed directly into MNI standardized space using SPM12 [21]. Individual images were scaled for intensity using a VOI applied to parenchyma in PMOD (PMOD Technologies, Zurich), yielding unitless fractional uptake values in the images. The AAL VOI library [22] was applied to extract sub-region values. Multivariate regression analysis addressed possible confounds arising from use of two scanners with no significant effect of scan instrument confounding study interpretations.

2.3. Plasma protein measurements

Blood was collected from an intravenous catheter into sodium EDTA tubes. Blood cells were removed by centrifugation. Blood processing and freezing occurred within one hour of collection. CSF was obtained as described elsewhere [23]. Plasma proteins were measured using assays validated for human plasma with an electrochemiluminescense platform (Quickplex 120, MesoScale Discovery).

2.4. Cognitive testing

Scaled scores from the inhibition subscale of the Delis-Kaplan Executive Function Scale (DKEFS) Color Word Interference Test assessed response inhibition [24]. Time to complete the task was converted to age-adjusted scaled scores, with higher scores corresponding to better performance. Self-reported cognitive shifting abilities were assessed with the shift subscale of the Behavioral Rating Inventory for Executive Function-Adult Version (BRIEF-A) [25].

2.5. Human tissue samples

Autopsies were approved or exempt from human subjects regulations and IRB. Tissue samples were obtained from autopsies of blast-exposed veterans and age, sex-matched, non-exposed controls. Formalin-fixed, paraffin embedded sections of cerebellum with cerebellar cortex at the level (and including) dentate nucleus were received on charged microscope slides and submitted for immunostaining. Brain sections were stained using a commercially available kit (Opal Manual IHC Kit, Akoya Bioscience). Antibodies used were: anti-VEGF-A (Santa Cruz), anti-caveolin-1 (Cell Signaling) and anti-calbindin (EMD Millipore). Heat-mediated antigen retrieval was performed in AR6 buffer. Stained slides were mounted with ProLong Diamond antifade mountant (ThermoFisher). Confocal microscopy was performed using a Leica TCS SP5 II microscope. Images were acquired with the Leica Application Suite. All images are single z-plane scans. Post-acquisition image processing and figure preparation was accomplished using Leica Application Suite and Photoshop software (Adobe) that was limited to linear contrast and brightness adjustments and applied identically to mTBI and matching control images.

2.6. Statistics

As appropriate, Chi-square tests or between-subjects t-tests were used. P values denote two-tailed significance. Pearson and Spearman correlation p values correspond to two-tailed outcomes. Benjamini–Hochberg procedures corrected for multiple comparisons using a false discovery rate of 0.05 [26]. Statistical analyses were done using SPSS software (IBM).

3. Results

3.1. Study participant characteristics

72 Operation Iraqi Freedom/Operation Enduring Freedom (OIF/OEF) veterans with blast-related mTBI and 24 OIF/OEF Deployed Control (DC) veterans with no history of TBI were studied. These groups were matched for age (T [89]=0.606, n.s.), race (Chi-square = 0.157[df = 3], n.s.), and Apolipoprotein E genotype (Chi-square = 5.87[df = 4], n.s.). The mTBI group was evaluated an average of 5.5 (plasma collection) and 5.3 (imaging) years after their last reported mTBI (Table 1). As expected of this population [20,27], mTBI veterans had elevated comorbid posttraumatic stress disorder (PTSD) symptoms measured by the PTSD Checklist-Military version [28] (PCL-M:T [92]=7.13, p ≤ .0001); increased depressive symptoms measured by the Patient Health Questionaire-9 [29] (PHQ-9 total score:.T [93]=4.98, p ≤ .0001); poorer sleep quality measured by the Pittsburgh Sleep Quality Index [30] (PSQI: T [89]=6.00, p ≤ .0001); but not significantly increased alcohol use [31] (AUDIT-C:T [93]=1.37, n.s.) compared to DCs.

Table 1.

Subject characteristics of living Veterans.

Demographics	Blast-related mTBI	Deployed Controls (DC)	p value
	Mean (SD), range	Mean (SD), range
Age (years)	34.4 (9.7) 22–60	33.0 (7.0) 23–47	n.s.
Race %	4.2, 5.6, 76.1, 14.1	4.2, 4.2, 75.0, 16.6	n.s.
A, B/AA, W, O
Hispanic Ethnicity %	14.1	12.5	n.s.
Apolipoprotein E genotype % (2,3) (2,4) (3,3) (3,4) (4,4)	8.5, 1.4, 64.8, 22.5, 2.8	13, 0, 39.1, 39.1, 8.7	n.s.
Blast exposure/TBI history
Number of blast-related mTBIs during military service	29 (65) 1 > 100 Median = 11	N.A.
Number of lifetime mTBIs with LOC	2.4 (0.3) 1–12	N.A.
Time since last blast-related mTBI to blood draw (years)	5.5 (2.8) 0.6–12.3	N.A.
Time since last blast-related mTBI to PET scan (years)	5.3 (2.7) 0.5–12.1	N.A.
Behavioral and neurological measures
PCL-M total score	52.33 (16.2) 19–84	26.7 (11.8) 17–62	0.0001
PHQ-9 score	11.7 (7.1) 0–27	3.7 (5.7) 0–24	0.0001
PSQI score	11.3 (4.4) 1–20	5.4 (3.1) 0–11	0.0001
AUDIT-C score	3.5 (2.5) 0–10	2.7 (1.6) 0–5	n.s.
BRIEF-A Cognitive Shift T-scores	65.1 (12.3) 39–88	54.1 (11.3) 39–66	0.036
Stroop word color interference test	10.2 (3.6) 1–15	11.1 (2.5) 7–14	n.s.

For Race% A, B/AA, W, and O correspond to Asian, Black/African American, White, and Other, respectively. Percent self-identified Hispanic ethnicity indicated. All participants were male US military veterans. LOC denotes the lifetime number of reported losses of consciousness lasting less than 30 min. A LOC greater than 30 min met criterion for study exclusion. Posttraumatic stress disorder (PTSD) was evaluated by the PTSD Checklist-Military version (PCL-M). Depression was evaluated with the Patient Health Questionaire-9 (PHQ-9). Alcohol use was measured with the Alcohol Use Disorders Identification Test-Consumption (AUDIT-C). Sleep quality was evaluated using the Pittsburgh Sleep Quality Index (PSQI). Behavioral Rating Inventory for Executive Function-Adult version (BRIEF-A) cognitive shift subscale and Stroop word color interference test evaluated higher order cognitive flexibility. P-values for Apolipoprotein E genotype and Race denote results of Chi-Square analyses and Fischer’s Exact test for % Hispanic. No participants in this study had an apoE 2,2 genotype. P-values for Age, PCL-M, PHQ-9, PSQI, AUDIT-C, BRIEF-A.

3.2. VEGF-A is elevated in plasma from veterans with blast-related mTBI

Multiple preclinical reports have found that VEGF-A levels are altered by blast exposure [12,13,15,32–35]. However, in this regard little is known in humans. We tested the hypothesis that VEGF-A is elevated in plasma from veterans with blast-related mTBI using a validated electrochemiluminescent assay. Fig. 1 shows that plasma VEGF-A was significantly elevated in the mTBIs compared to DCs (T [94]= 2.60, p ≤ .011).

Fig. 1.

Plasma VEGF-A levels are chronically elevated in mTBI veterans with self-reported repetitive blast exposure.

VEGF-A levels measured on average 5.5 years after last reported mTBI were significantly elevated (p ≤ .011) in OIF/OEF veterans with self-reported repetitive blast exposures accompanied by acute mTBI symptoms (N = 72) compared to deployed control veterans (N = 24).

This platform also measured basic fibroblast growth factor-2 (bFGF), plasma placental growth factor (PLGF), KEK receptor tyrosine kinase (Tie2), VEGF-C, VEGF-D, and FMS-like tyrosine kinase 1 (FLK-1). In addition to VEGF-A, only bFGF levels were significantly increased in mTBIs versus DCs (bFGF:T [94]=2.95, p ≤ .004); and with no significant change in PLGF:T [94]=0.625, n.s.; Tie2:T [94]=0.297, n.s.; VEGF-C:T [86]=1.47, n.s.; VEGF-D:T [94]=0.83, n.s.; FLK-1:T [94] =0.37, n.s. Following correction for multiple comparisons (see Methods, 2.6), VEGF-A and bFGF differences remained statistically significant at the 0.05 level. In blast subjects, plasma VEGF-A and bFGF levels did not correlate with the log₁₀ transformed number of self-reported blast exposures (Pearson r = −1.36, n.s.; r = 0.058, n.s., respectively). In addition, none of these proteins differed significantly in the CSF of mTBIs compared to DCs (VEGF-A:T [66]=0.562, n.s., N = 47, 21 mTBI and DC, respectively; bFGF:T [52]=0.848, n.s., N = 36, 18; PGIF:T [66]=0.858, n.s., N = 47, 21; VEGF-C:T [16] =0.111, n.s., N = 13, 5; VEGF-D:T [65]=0.423, n.s., N = 46, 21; FLK 1:T [66]=0.025, n.s., N = 47, 21).

3.3. Plasma VEGF-A corresponds with cerebellar glucose uptake by FDG-PET

To investigate relationships between plasma VEGF-A and CNS function in blast mTBI veterans we quantified positron emission tomography (PET) fractional [¹⁸F]-fluorodeoxyglucose (FDG) uptake in cerebellum. We focused on the cerebellum because of its vulnerability to blast-related mTBI, which we and others have shown is associated with lower FDG uptake [9,36,37]. Based on prior findings that suggest the Crus I, Crus II and posterior aspects of the cerebellum display more robust chronic FDG-PET imaging abnormalities [9], five bilateral cerebellar volumes of interest (VOIs) were selected for study from standardized clinical PET/CT images in 68 veterans with blast-related mTBI. Fig. 2 shows that FDG uptake in cerebellar lobules VII, VIII, IX, Crus I, and Crus II were significantly negatively correlated with plasma VEGF-A levels (Pearson r = −0.301, p ≤ .013; r = −0.266, p ≤ .028; r = −0.262, p ≤ .031; r = −0.343, p ≤ .004; and r = −0.327, p ≤ .006, respectively, N = 68). All five regions remained statistically significant after correction for multiple comparisons. Collective analysis (lobules VII-IX + Crus I + Crus II) confirmed a significant correlation between increasing plasma VEGF-A levels and decreasing cerebellar FDG uptake (r = −0.325, p ≤ .007, N = 68). Linear regression analysis showed that this correlation remained significant after accounting for possible confounds arising from the use of two PET scanners (r = −0.294, p ≤ .013).

Fig. 2.

Higher plasma VEGF-A levels correspond with lower regional FDG-uptake in blast-mTBI veterans.

Resting state [¹⁸F]-fluorodeoxyglucose (FDG) PET imaging in veterans with blast-related mTBI (N = 68) was significantly negatively correlated with levels of plasma VEGF-A in cerebellar VOIs comprising left + right (A) Crus I; (B) Crus II; (C) Lobule 7; (D) Lobule 8; (E) Lobule 9; and (F) combined (Crus I, II, Lobules 7–9), indicating that increasing plasma VEGF-A levels corresponded with reduced metabolism in the cerebellum as measured by FDG-PET.

Further confirming the specificity of these findings to blast mTBI, FDG uptake in deployed controls was not correlated with plasma VEGF-A (Crus I: r = 0.134, n.s.; Crus II: r = 0.164, n.s.; VII: r = 0.257, n.s.; VIII: r = 0.158, n.s.; IX: r = 0.084, n.s.; combined Crus I, II, lobules VII-IX; r = 0.167, n.s.). Adding further to the specificity of the FDG/VEGF-A correlations in mTBI veterans, bFGF was not significantly correlated with FDG uptake after adjusting for multiple comparisons (Crus I: r = −0.251, p ≤ .039; Crus II: r = −0.194, n.s.; VII:r = −0.181, n.s.; VIII:r = −0.206, n.s.; IX r = −0.150, n.s.; combined: r = 0–0.212, n.s.).

3.4. Plasma VEGF-A corresponds with cognitive inhibition and set shifting impairment

We have previously reported correlations between blast exposure and impairments in sensorimotor integration [9]. However, among these veterans, motor-related symptoms are minor relative to complaints involving higher order cognitive and behavioral functions [38,39]. The cerebellum is part of a network that subserves a number of executive cognitive functions [40,41] that include supporting performance on the Stroop color interference test, a well-established measure of higher order cognitive inhibition of an over-learned perceptual response [24,42–44]. We identified a significant negative correlation between plasma VEGF-A levels in the blast mTBI group versus scores from the Inhibition subscale of the Delis-Kaplan Executive Function Scale (DKEFS) color/word interference test (Fig. 3A), an instrument used to assess response inhibition (Spearman r = −0.434, p ≤ .013, N = 32).

Fig. 3.

Objective and self-reported measures of cognitive executive functional impairment correspond with increasing plasma VEGF-A in blast-mTBI veterans.

(A) Scaled scores on the Stroop Color Word Interference Test, which is used to assess cognitive inhibition of over-learned word reading responses were significantly negatively correlated plasma VEGF-A levels in veterans with blast-related mTBI (N = 32). Lower scores correspond with worse performance. (B) Self-reported difficulties with shifting from one cognitive task or situation to another measured by the Behavioral Rating Inventory for Executive Function-Adult version (BRIEF-A) was significantly correlated with plasma VEGF-A levels in veterans with blast-related mTBI (N = 33). Higher BRIEF-A shift subscale T-scores correspond with greater difficulty with cognitive shifting.

These data indicate that lower performance on the Stroop task corresponded with higher plasma VEGF-A levels. To further test this idea we analyzed T-scores on the Behavioral Rating Inventory for Executive Function-Adult version (BRIEF-A), a self-report inventory that rates clinically significant difficulties in ability to cognitively set shift in different situations, including between tasks, from a way of thinking, or from an anticipated schedule [25]. Fig. 3B shows that plasma VEGF-A levels in the mTBI veterans were positively correlated with BRIEF-A shift subscale T-scores (Spearman’s r = 0.453, p ≤ .008), indicating that greater difficulty with cognitive shifting corresponded with higher plasma VEGF-A levels.

Executive cognitive functions engaged by the Stroop color/word test are commonly associated with prefrontal cortical brain regions [45]. Uniformly, plasma VEGF-A levels in the blast mTBIs did not correlate significantly with FDG uptake in superior frontal orbital, medial frontal, and inferior frontal orbital cortex (r = 0.097, n.s.; r = 0.152, n.s.; and r = −0.058, n.s, respectively) even though well-established cerebro-cerebellar connectivity networks functionally link a number of frontal cortical brain regions to the cerebellum [46,47].

3.5. Perivascular VEGF-A is increased in cerebellum of veterans with blast-related mTBI

Endothelial cells express luminal and abluminal VEGF receptors [48,49] that internalize VEGF-A [50]. In addition, astrocyte-associated extracellular matrix compartments can complex with and sequester VEGF-A in the brain [51]. Even though VEGF-A was not elevated in CSF of veterans with mTBI, it is still possible that VEGF-A could accumulate and become sequestered in the cerebellum.

To test this idea, we examined cerebellar VEGF-A immunoreactivity in a rare set of three postmortem specimens from veterans with blast-related mTBI and two comparable non-TBI veteran controls. TBI₁ was an active duty 46 year old male Navy SEAL who died from a self-inflicted gunshot to the head. Military blast exposure included multiple improvised explosive devices (IEDs) and other ordnance. Contact sports history consisted of high school, college, semi-professional football, and mixed martial arts. Medical/psychiatric history included diagnoses of PTSD, alcohol use disorder, hearing loss, and chronic pain. Neuropathologic findings were CTE (Stage 1) and interface astroglial scarring (IAS). TBI₂ was an active duty male Navy SEAL who died at age 35 by drowning. Military blast exposure included multiple IEDs and other ordnance. There was no history of participation in contact sports. Medical/psychiatric diagnoses included PTSD, alcohol/substance use disorder, chronic pain, paranoid ideation, and manic episodes. Neuropathological findings were IAS without tau pathology. TBI₃ was a male Army veteran who died at age 46 following complications of elective back surgery. This case had participated in Dr. Peskind’s study of blast mTBI and had extensive clinical characterization including lifetime history of mTBI as for the living Veterans reported here. Military blast exposure consisted of > 50 IEDs and other ordnance with acute symptoms consistent with VA/DoD criteria for mTBI and no history of contact sports or impact TBIs. Medical/psychiatric diagnoses included PTSD, alcohol use disorder, migraine headaches, chronic back pain, hearing loss, and tinnitus. Neuropathologic diagnoses included CTE (Stage 1) and meningitis. Control₁ was a male active duty Navy officer with no known TBIs who died at 43 due to cardiac arrest. Medical history included hypertension, hypercholesterolemia, and chronic obstructive pulmonary disease. There were no significant neuropathological abnormalities, including no evidence of tauopathy or neurodegeneration. Control₂ was a male Army veteran who died at age 57 due to cardiac arrest. He had no lifetime history of TBI. However, he reported falling down the stairs and bumping his head twice at the ages of 4 and 7 years without loss of consciousness or requiring any urgent care or emergency room evaluation. Medical/psychiatric diagnoses included headaches and remote history of a “mental breakdown.”

Fig. 4 shows VEGF-A immunoreactivity (green) was associated with microvascular profiles immunostained with the endothelial cell marker caveolin (blue), which were more readily detected in the blast-related mTBI cases than controls. Calbindin immunoreactivity (red) was used to label Purkinje cells. It is well established that cerebellar Purkinje cells express VEGF-A [52,53]. In keeping with this we observed a number of VEGF-A positive Purkinje cells, which supported the specificity of the immunostaining, but there was no apparent distinction in Purkinje cell VEGF-A expression in the mTBI versus control cases. We also examined VEGF-A expression in cortex and hippocampus from the same postmortem cases (Supplementary Fig. 1). Occasional VEGF-A positive peri-microvessel profiles were also observed in cortex and hippocampus, but unlike in the cerebellum there was no apparent difference between the control and TBI cases. As previously reported [54,55], neuronal VEGF-A immunoreactivity was evident in cortical and hippocampal neurons, but as with the cerebellum there was no notable distinction between the control and TBI cases.

Fig. 4.

VEGF-A accumulates in cerebellum of veterans with blast-related mTBI.

Cerebellar postmortem specimens from veterans and active duty servicemembers with blast-related (TBI_1–3) and two controls with no known TBI history (see Results, 3.5) were immunostained for VEGF-A (green), calbindin (red) that stains Purkinje cell bodies/processes, and caveolin (blue), which stains vascular endothelial cells. In distinction to controls (A-H), in the TBI cases (I-T), VEGF-A immunostaining was associated with microvascular profiles (caveolin-positive) (white arrow heads). In keeping with previous findings [52], prominent Purkinje cell body VEGF-A expression was observed, thereby supporting the specificity of the VEGF-A immunostaining, but did not reliably distinguish the mTBI and control cases. Scale bars = 50 μm.

4. Discussion

4.1. Elevated VEGF-A implicates chronic vascular dysfunction following blast-related mTBI

The intensive effort to identify chronic blood-borne biomarkers associated with repetitive mTBI in servicemembers and veterans has proven quite challenging. Recent advances indicate that altered protein expression of ubiquitination proteins [56], neurofilament light chain (NfL) [5], amyloid-β peptides [57,58] and tau species [58–60] are chronically elevated in blood and/or blood-derived exosomes from military service members with mTBI. Further, several epigenetic processes are altered in veterans with chronic mTBI [4,6,18]. Herein we report that VEGF-A is significantly elevated in plasma from blast mTBI veterans an average of 5.3 years after their last reported mTBI.

Whether aberrant VEGF-A levels contribute to sustained pathological processes or reflect ongoing compensatory mechanisms promoting vascular recovery cannot be addressed by these data. Increased VEGF-A levels could be part of the adaptive vascular responses to chronic injury, an idea consistent with recent findings of chronic neurovasculature thinning in blast-exposed animals [7]. Alternately, over time vascular remodeling may reach a stable homeostatic state, yet after multiple cycles of mTBI-induced activation, processes that promote VEGF-A secretion could become chronically hyperactive.

The initiating hypothesis of this study focused on VEGF-A, not bFGF, which we also found is elevated in mTBI veterans. Interestingly, bFGF interacts with VEGF-A signaling pathways [61,62]. How bFGF is involved in chronic blast-related mTBI cannot be addressed by these findings and is beyond the scope of the current report. Nonetheless, elevated plasma bFGF in blast mTBI veterans further supports the idea that vascular disturbances are an important feature of the chronic pathophysiology attending blast exposure. Importantly, the results herein are strongly supported by our recent report that microRNAs that directly suppress VEGF-A and bFGF expression are significantly reduced in plasma exosomes from veterans with blast-related mTBI [18].

4.2. Blast exposure is a unique method of inflicting trauma

VEGF-A is expressed throughout the body. In the CNS, VEGF-A expression is especially prominent in cerebellum and choroid plexus [63,64]. In searching for blood-borne TBI biomarkers, emphasis is often placed on molecules expressed selectively in brain. However, blast presents a more complex injury than blunt impact head trauma. The intense shock waves generated by high explosives propagate through the entire body, imparting simultaneous insults to multiple interacting organ systems, even when the victim is wearing protective body armor [65]. Blast-induced shock waves can induce biomechanical tissue injury throughout the body by multiple means that include shear stress, tensile stress, and formation of micro-cavitation bubbles that can damage cells [66] and may play important roles in mediating injury to small vessels [67,68] such as those in brain tissue. The fact that peripheral blast exposure of the torso in an animal model with the head shielded from blast forces still causes CNS pathology [69] lends importantly to the idea that blast overpressure injuries induce a form of polytrauma [34,69]. From this perspective, it is possible that VEGF-A disturbances may reflect blast-induced systemic insults with multiple functional consequences affecting both the brain and periphery together. The findings in this report necessarily leave open questions regarding the potential mechanisms and consequences of chronically elevated plasma VEGF-A. Nonetheless, it is possible that such elevations may reflect angiogenic regulatory mechanisms attempting to counteract chronically thinning cerebral vascular structures, which has been observed in blast-exposed rats [7]. Elevated plasma VEGF-A could also be indicative of dyshomeostasis of the dynamic interplay between endothelial cells and perivascular astrocytes of the BBB that is mediated, in part by VEGF-A [19] and which would be consistent with evidence of blast-induced BBB dysfunction [9,10,16–19].

4.3. Plasma VEGF-A association with cerebellar hypometabolism

Although there has long been evidence that the cerebellum is vulnerable to mild impact neurotrauma [70,71], this brain region has nonetheless received comparatively less attention than some other brain regions. We and others have reported that the cerebellum is vulnerable to blast-induced mTBI [9,10,20,37,65,72–85] [86]. Interestingly, a very recent report of tau PET imaging in veterans with blast TBI detected significant tau tracer uptake in the cerebellum and other brain regions that have not typically been associated with CTE tau pathology [87].

Using FDG-PET we found that glucose hypometabolism correlations with blast exposure were most prominent in the inferior/posterior cerebellum [9]. Therefore, we reasoned that if plasma VEGF-A levels could serve as a peripheral indicator of CNS function, such associations might more easily be ascertained by examining FDG uptake in cerebellum. Our findings support this idea. Plasma VEGF-A levels did not correlate with the frontal cortical areas examined. This does not mean these cortical regions are unaffected by blast mTBI nor does it suggest aberrant VEGF-A levels could not affect other brain regions. It is possible the cerebellum may be vulnerable to mTBI due to a number of factors that include: (i) very high neuronal density, in which approximately 69 billion of the 86 billion total neurons in adult human brain are in cerebellum [88]; (ii) high metabolic demand partially accounted for by high cell density; (iii) a cerebellar microvascular structure that is vulnerable to injury [89]; and (iv) the anatomical location of the cerebellum with respect to the posterior fossa.

4.4. Limitations

The correspondence between decreasing FDG uptake and increasing VEGF-A are well in keeping with multiple reports of blast-related neurovascular dysfunction [7–16]. Nonetheless, these data do not rule out other possibilities involving neuronal/synaptic loss or additional non-vascular functions of VEGF-A [90]. For example, VEGF-A complexed in extracellular matrices around Purkinje cells is a chemo-attractant that mediates granule cell/Purkinje cell connections [53]. It is not possible to determine if the VEGF-A immunoreactivity we observed in the blast mTBI cases is from peripheral or central sources. Importantly, because of the limited number of postmortem blast mTBI cerebellum cases available, these data must be interpreted cautiously to suggest only the possibility that elevated VEGF-A may localize in proximity to parenchymal perivascular and peri-astrocytic compartments. Similarly, our finding that VEGF-A accumulates in perivascular domains of the cerebellum, but not cortex or hippocampus, does not speak to the issue of where in the brain plasma VEGF-A levels may influence function.

It is difficult to assess the potential neurodegenerative specificity of elevated plasma VEGF-A to chronic blast-related mTBI. In distinction to blast-related mTBI, there are reports that VEGF-A levels are decreased in the serum of Alzheimer’s disease (AD) patients [91,92], as well as in peripheral mononuclear cells in Huntington’s disease patients [93], but is elevated in serum in multiple sclerosis patients [94]. Interestingly, serum VEGF levels are elevated in AD patients with cerebral microbleeds [95] and higher plasma VEGF levels correspond to lower Mini-Mental State Examination (MMSE) scores [96] [4,6,17,18].

Finally, it is important to recognize the limitations of history by retrospective recollection of blast mTBIs, as well as not acquiring history of subconcussive blast exposure. Both mTBI and DC veterans may have had at least some exposure to subconcussive blast, at the least in military training in preparation for deployment to the Iraq/Afghanistan combat theaters.

5. Conclusions

Plasma VEGF-A is chronically elevated in veterans with a history of repeated blast exposure causing mTBI. These findings add to the small number of blood-borne molecules thus far identified in this population during the chronic injury phase. This may reflect vascular dysregulation in the periphery and perhaps the CNS. Regardless of its source, increasing plasma VEGF-A levels correlate with decreasing glucose metabolism in the cerebellum, poorer performance on the Stoop color/word interference test, and increased difficulty in subjective cognitive set shifting. Overall, these findings significantly strengthen the idea that chronic focal vascular dysfunction may be an important pathophysiological consequence of repetitive blast exposure.

Abbreviations:

bFGF

basic fibroblast growth factor, FGF2

BBB

blood-brain barrier

BRIEF-A

Behavioral Rating Inventory for Executive Function-Adult Version

CNS

central nervous system

CSF

cerebrospinal fluid

CTE

chronic traumatic encephalopathy

DKEFS

Delis-Kaplan Executive Function Scale

DSM-IV

Diagnostic and Statistical Manual of Mental Disorders, 4th edition

FDG

[¹⁸F]-fluorodeoxyglucose

IAS

interface astroglial scarring

mTBI

mild traumatic brain injury

OEF/OIF/

Operation Enduring Freedom/Operation Iraqi Freedom

PET

positron emission tomography

PHQ-9

Patient Health Questionaire-9

PSQI

Pittsburgh Sleep Quality Index

PTSD

posttraumatic stress disorder

VA/DoD

Veterans Affairs/Department of Defense

VEGF-A

vascular endothelial growth factor-A)

VOI

volume of interest