Cerebrovascular Neuroprotection after Acute Concussion in Adolescents

Abstract

Objective:

To assess acute cerebrovascular function in concussed adolescents (14–21 years of age), whether it is related to resting cerebral hemodynamics, and whether it recovers chronically.

Methods:

Cerebral vasoreactivity and autoregulation, based on middle cerebral artery blood flow velocity, was assessed in 28 concussed participants (≤14 days of injury) and 29 matched controls. The participants in the concussion group returned for an 8-week follow-up assessment. Over the course of those 8-weeks, participants recorded aerobic exercise frequency and duration.

Results:

Between groups, demographic, clinical, and hemodynamic variables were not significantly different. Vasoreactivity was significantly higher in the concussed group (p = 0.02). Within the concussed group, 60% of the variability in resting cerebral blood flow velocity was explained by vasoreactivity and two components of autoregulation – falling slope and effectiveness of autoregulation (adjusted R² = 0.60, p < 0.001). Moreover, lower mean arterial pressure, lower responses to increases in arterial pressure, and lower vasoreactivity were significantly associated with larger symptom burden (adjusted R² = 0.72, p < 0.01). By the 8-week timepoint, symptom burden, but not vasoreactivity, improved in all but four concussed participants (p < 0.01). 8-week change in vasoreactivity was positively associated with aerobic exercise volume (adjusted R² = 0.19, p = 0.02).

Interpretation:

Concussion resulted in changes in cerebrovascular regulatory mechanisms, which in turn explained the variability in resting cerebral blood flow velocity and acute symptom burden. Furthermore, these alterations persisted chronically despite symptom resolution, but was positively modified by aerobic exercise volume. These findings provide a mechanistic framework for further investigation into underlying cerebrovascular related symptomatology.

Of the nearly 200,000 concussions that occur annually, the highest rate is among those aged 10–19 years.¹ This high prevalence at a young age is a concern due to the potential for sequelae beyond primary symptom resolution.^2,3 Yet, the acute pathophysiology of concussion symptoms is poorly understood and derived primarily from animal models.⁴

In adolescents and young adults with sport-related concussion, standard neuroimaging is rarely able to identify signs of neural injury, and only 1.5–3% of CTs and MRIs demonstrate signs of neural injury within 30 days of injury.⁵ While advanced neuroimaging shows some alterations in the neural integrity and functional connectivity after a concussion, similar alterations are also reported in asymptomatic young adult athletes with a history of concussion,⁶ and correlate poorly with self-reported symptoms.⁷ Thus, acute and subacute symptom burden does not appear to result from a gross structural neural injury. Alternatively, physiologic disruption in mechanisms that support neural function may be the source of post-concussion symptoms.^8,9

After a concussion, optimal cerebral blood flow is necessary to meet the increased metabolic needs of the injured brain. However, data from both animals^4,10,11 and humans¹² suggest that cerebral blood flow may decrease after a concussion and can remain reduced for extended periods of time, resulting in an ‘energy mismatch,’ or neurometabolic crisis.⁴ While this seems reasonable, available data is conflicting: global and regional cerebral blood flow can be reduced,^13,14 unchanged,^15,16 or even increased¹⁷ in individuals who have suffered a concussion, and these changes may or may not relate to symptom burden.¹⁸ While part of this discrepancy may be due to methodological differences between the studies, it also highlights the possibility that concussion impacts the regulation of cerebral blood flow to meet metabolic demand, and not the resting blood flow per se.

In healthy individuals, the cerebral vasculature regulates cerebral blood flow by its ability to buffer against changes in arterial CO₂ (ie, by vasoreactivity) and in arterial pressure (ie, by autoregulation).^19,20 Data both from our group²¹ and others^22,23 have shown cerebrovascular dysfunction in adults with persistent post-concussion symptoms. However, it remains unknown if a similar alteration in cerebrovascular function exists acutely after a concussion, if these alterations are reflected in cerebral blood flow and/or relate to symptom burden, and whether these alterations last beyond the acute phase and overt symptom resolution. We hypothesize that in adolescents and young adults, components of cerebrovascular function (vasoreactivity and autoregulation) are disrupted acutely (within 14 days) after a sports-related concussion compared to control participants and the extent of their alterations were related to resting cerebral blood flow. We further sought to determine, longitudinally, whether any acute alteration in cerebrovascular function persisted at 8 weeks following injury.

Subjects and Methods

Participants

We recruited adolescents and young adults 14–21 years of age from a sport concussion clinic of a tertiary care regional children’s hospital between September 25, 2018 and March 5, 2020. Participants in the concussion group had sustained a physician-diagnosed concussion during sports or other recreational activities within 2 weeks of testing. We defined concussion consistent with the fifth Consensus Statement on Concussion in Sports guidelines, as a brain injury caused by a direct blow to the head, face, neck or elsewhere on the body, resulting in the rapid onset of impairment of neurological function.²⁴ Non-concussed control participants were recruited among those who presented to a sports injury prevention clinic with no complaints, and matched to concussion group characteristics (age and sex) at the time of recruitment.

Prior to study commencement, the study protocol was reviewed and approved by the local institutional review boards (protocol numbers 2017P002645 at SRH and IRB-P00024907 at BCH). All participants (older than 18 years of age) or their parents/guardians (if <18 years of age) gave informed consent, and all individuals younger than 18 years of age gave their assent to participate.

Participants were excluded if they had been diagnosed with a concussion in the year prior to initial assessment (other than the index concussion for the concussion group), a history of neurological surgery or seizure disorder, were using medication or a medical device that would alter heart rate, arterial pressure, or autonomic function, or had a current, serious, chronic medical condition that could interfere with study participation. After enrollment and screening, all participants completed an inperson assessment (described below) within 14 days of injury (or enrollment, for control subjects). The concussed participants visited the laboratory for an identical follow-up assessment 8 weeks after baseline. To quantify the average weekly aerobic exercise volume, participants completed standardized weekly logs where the amount of time spent exercising daily of at least moderate intensity was recorded. Moderate intensity was defined as a heart rate response of approximately 60–70% of age-predicted max HR. Aerobic exercise volume was calculated as the total minutes per week averaged across the 8 weeks. The mean exercise volume was determined from the mean exercise minutes per week based on the weeks with non-missing values.

Laboratory Testing

Prior to the cerebrovascular function assessments, all participants completed the Post-Concussion Symptom Inventory (PCSI),^25,26 which is similar to the Post-Concussion Symptom Score, and is validated for use within the age range of our study participants. Participants completed the 20 questions addressing the severity of different concussion-related symptoms, ranging from 0 (no symptoms) to 6 (maximum severity). Symptom severity was calculated as the sum of all responses, with scores ranging from 0 (no symptoms) to 120 (maximum severity on all symptoms). In addition to the total PCSI score, three different categories of PCSI—physical, cognitive, and fatigue-related – were calculated based on previously established criteria.²⁶

After the PCSI assessment, participants were instrumented for hemodynamic measurements. A standard 5-lead ECG (Dash 2000/5000, GE Medical Systems, Waukesha, WI, USA) was placed on the chest with lead II continuously recorded. A finger photoplethysmography system (Finometer, Finapres Medical System, Netherlands) was used to measure beat-by-beat arterial pressure for determination of changes in pressure, with a cuff placed on the non-dominant index or middle finger, depending on signal validity. Brachial oscillometric blood pressure (Dash 2000/5000, GE Medical Systems, Waukesha, WI, USA) was measured in the dominant arm and used as a standard measure of blood pressure as well as to calibrate the Finapres output signal. Throughout the protocols, end-tidal CO₂ (EtCO₂) was continuously measured by a nasal cannula connected to an infrared CO₂ analyzer (VacuMed, Model 17515A, Ventura, CA, USA).

A transcranial Doppler ultrasound (TCD, 2 MHz probe; MultiDop T2, DWL Elektronische System, Singen, Germany) was used to measure cerebral blood flow velocity of the right middle cerebral artery (MCAv) with a probe placed on the right temple using a signal depth between 45–65 mm. A Mueller-Moll probe fixation device held the ultrasound probe in place. The TCD signal was optimized by adjusting the probe angle and insonation depth settings. Probe location and settings were noted and used for the 8-week visit in the concussed participants.

Following instrumentation, 5 minutes of seated resting measures (heart rate, blood pressure, EtCO₂, and MCAv) were recorded continuously. Following the rest period, protocols to assess components of global cerebrovascular function (cerebral vasoreactivity and autoregulation) ensued. Assessments were performed with the individual in a seated position. All signals were digitized and stored at 1000 Hz (PowerLab, AD Instruments, Colorado Springs, CO, USA) and stored offline for analysis (LabChart, AD Instruments, Colorado Springs, CO, USA). Blood pressure and MCAv were decimated to 5 Hz and low pass filtered with a cutoff of 0.4 Hz to provide mean waveforms. Data were analyzed using custom software written in MATLAB (version R2019a, MathWorks, Natick, MA, USA).

Cerebral Vasoreactivity

Cerebral vasoreactivity was determined by assessing cerebrovascular conductance response to incremental hypercapnia elicited by rebreathing.^19,27 Subjects were fitted with a mouthpiece attached to a three-way respiratory valve unit, which, in turn, was attached to a 5-L rubber rebreathing bag. The valve allowed for an instantaneous switch from room air to the rebreathing bag, which had been prefilled with room air. EtCO₂ was measured continuously via a sample line connecting the mouthpiece to an infrared CO₂ analyzer. After a one-minute baseline collection of room air breathing, the valve was switched so the participant then inspired from and expired into the 5-L rebreathing bag for two-minutes; until an EtCO₂ concentration of 50–55 mmHg was reached. After 2 minutes, the valve was switched again allowing the participant to breathe room air. The exposure was deliberately limited to 2 minutes to minimize the impact of rebreathing on oxygen saturation while sufficiently increasing EtCO₂. Cerebral vasoreactivity was determined by plotting breath-by-breath cerebrovascular conductance (ie, MCAv/mean arterial pressure) against the corresponding EtCO₂ value. A piece-wise linear regression was used to determine the linear portion of the relationship between cerebrovascular conductance and EtCO₂, providing a slope of this relationship which was used to quantify vasoreactivity.

Cerebral Autoregulation

Cerebral autoregulation was determined using a resistance breathing technique which elicits oscillations in mean arterial pressure sufficiently large for characterization.²⁸ Participants breathed through a mouthpiece attached to a standard impedance threshold device made from latex-free PVC (PowerBreathe, England, UK), which was set to cause moderate breathing resistance (10–20 cm H₂O). This was to elicit fluctuations in blood – thus, perfusion – pressure sufficiently large enough to engage cerebral autoregulatory mechanisms.^19,29,30 Autoregulation was determined by deriving the relationship between mean arterial pressure (independent variable; input variable) and MCAv (dependent variable; output variable) fluctuations using a validated nonlinear and nonparametric approach based on projection pursuit regression (PPR).³¹ This approach has been validated and shown to have excellent reproducibility (Lin’s concordance coefficient >0.95 across repeat-assessments).³² It allows assessment of nonlinear relationships between blood pressure and cerebral blood flow velocity, and provides an assessment of four aspects of autoregulation: the range of blood pressures within which autoregulation was effective (ie, autoregulatory range), the effectiveness of autoregulation (ie, autoregulatory slope), and blood flow responses to decreases or increases in blood pressure outside the autoregulatory range (respectively, falling and rising slope).³² The linear slope of the pressure-flow relationship within each region provides a measure of the effectiveness of autoregulation. A lower gain indicates a smaller cerebral blood flow velocity response to changes in perfusion pressure within each range described above, whereas higher gains indicate larger flow responses to pressure changes.

Statistical Analysis

All statistical analyses were performed using R language for statistical computing (version 4.0.3). Continuous variables are reported as mean ± SD, and all categorical variables as count (%). Differences were considered statistically significant at p < 0.05. Grouplevel comparisons (concussed vs control) for continuous variables was determined using an ANOVA and for categorial variables a Chi-square test was used. A repeated measures ANOVA was used for comparisons of baseline and 8-week assessments within the concussed group.

Within the concussed group, to determine the relationship of post-concussion symptom burden to demographic, clinical, and hemodynamic variables, an initial ANCOVA model was used with age (continuous variable), sex (male or female), number of prior concussions, time since injury (days), baseline mean blood pressure, and measures of vasoreactivity and autoregulation (described above) as the full set of predictor variables. This was followed by a forward/backward stepwise model selection to identify variables with statistically meaningful contributions (based on Akaike Information Criteria, or AIC, as is standard for this methodology) while constraining model complexity to avoid superfluous relationships given the number of participants. We used the same approach to determine the relationship of resting MCAv to resting mean blood pressure and cerebrovascular function. A linear regression model was used to determine the relationship between exercise volume and cerebral vasoreactivity.

For all ANCOVA and regression models, model assumptions (multicollinearity, normality of the residuals and outliers, and homoscedasticity) were checked for conformity to statistical assumptions. For all models, explained variance (R²) is reported as adjusted values rather than observed ones; while the latter is a positively biased estimate of the population value, especially if the sample size is small, the former is not (adjusted by the degrees of freedom).

Results

Eighty adolescents and young adults were recruited and enrolled; 42 had sustained a concussion diagnosed by a sports medicine clinician and were symptomatic at the time of testing and 38 controls with no clinically diagnosed history of a concussion. Of these participants, we were able to obtain adequate hemodynamic and cerebral blood flow measurements in 57 participants (28 individuals who had a diagnosed concussion and 29 controls). Up to 20% failure of obtaining adequate signals are expected due to normal anatomic variations resulting in poor transtemporal insonation window, and it is consistent with our prior experience. We attribute the extra approximately 10% failure rate in our study to “impatience” and frequent head movement during the study/recording session.

Demographic, clinical, and hemodynamic variables were not significantly different between groups at baseline, including resting MCA flow velocity (Tables 1 and 2). At the group level, vasoreactivity was significantly higher among adolescents and young adults who sustained a concussion compared to control participants (Table 2 and Fig 1). Despite the statistically significant difference (p = 0.02), in the concussed group CVR was highly variable (coefficient of variation 42.1%).

TABLE 1.

Demographic and Clinical Variables at Baseline

	Control (n = 29)	Concussed (n = 28)	p value
Age	18.2 (2.3) [14.4–21.5]	17.4 (2.1) [14.3–21.9]	0.19
Female (%)	16 (55.2)	14 (50.0)	0.70
Height, cm	171.2 (10.5)	169.9 (9.4)	0.64
Weight, kg	67.4 (15.1)	65.6 (10.3)	0.61
Time since injury	-	11.4 (3.5) [4–14]	-
Number of prior concussions	-	1.2 (1.4) [0–5]	-
Total PCSI	-	41.6 (25.5) [4–114]	-
Resting RR interval, ms	882 (108)	864 (137)	0.58
RMSSD, ms	68.4 (43.3)	56.0 (28.0)	0.20

Continuous variables are mean (standard deviation). Categorical variables are count (percent). [ ], range. Significance, p < 0.05.

PCSI = Post-Concussion Symptom Inventory; RMSSD = root mean square of the successive differences in R-R intervals.

TABLE 2.

Hemodynamic and Cerebrovascular Variables at Baseline

	Control (n = 29)	Concussed (n = 28)	p value
Hemodynamic
Resting blood pressure (mmHg)
Systolic	108.3 (9.2)	109.5 (8.8)	0.61
Diastolic	59.9 (4.4)	62.3 (6.2)	0.10
Mean	76.1 (5.0)	78.0 (6.1)	0.19
Resting EtCO2 (mmHg)	33.7 (3.1)	33.5 (2.9)	0.84
Cerebrovascular
Resting MCAv (cm s−1)	57.4 (8.3)	58.7 (12.5)	0.65
Vasoreactivity (cm s−1 mmHg−1/mmHg CO2)	0.024 (0.009)	0.031 (0.013)	0.02
Autoregulation
Falling slope (cm s−1 mmHg−1)	0.0931 (0.431)	1.119 (0.525)	0.20
Autoregulatory slope (cm s−1 mmHg−1)	0.352 (0.246)	0.296 (0.199)	0.41
Autoregulatory range (mmHg)	6.1 (2.9)	5.4 (2.7)	0.39
Rising slope (cm s−1 mmHg−1)	0.944 (0.355)	0.898 (0.250)	0.62

Continuous variables are mean (standard deviation). Significance, p < 0.05. EtCO₂, end-tidal CO₂. Cerebrovascular autoregulation assessments are missing six participants in each group.

FIGURE 1:

Cerebral vasoreactivity at baseline in acutely concussed adolescents and young adults compared to age-matched control participants. Statistical significance remains when the two concussed participants with the highest vasoreactivity are removed from the cohort.

Within the concussion group, resting cerebral blood flow velocity was strongly and positively related to cerebrovascular function (vasoreactivity, falling slope, and autoregulatory slope), and these three components of cerebrovascular function explained almost two-thirds of resting MCAv (adjusted R² = 0.60, F _(3,18) = 11.7, p < 0.001). This was in contrast to the control group, for whom falling slope and autoregulatory slope, but not vasoreactivity, explained the variation in resting MCAv (adjusted R² = 0.46, F _(3,18) = 10.5, p < 0.001).

Within the concussed group with complete cerebrovascular function assessments (n = 22, both vasoreactivity and autoregulation), demographic (sex), clinical (number of prior concussions), hemodynamic (mean blood pressure), and cerebrovascular variables (vasoreactivity and autoregulation) explained over 75% of the variation for total PCSI score (Table 3). All other variables (ie, age and time since injury) were statistically removed from the model based on AIC during stepwise model selection (see Methods). Lower mean blood pressure, lower responses to increases in blood pressure, and lower vasoreactivity were significantly associated with higher total PCSI scores. This association was primarily explained by physical symptoms (adjusted R² = 0.72; adjusted R² = 0.34 when PCSI was confined to cognitive symptom scores only, and adjusted R² = 0.36 when PCSI was confined to fatigue symptoms scores).

TABLE 3.

Relationship of Symptom Burden to Demographic, Clinical, Hemodynamic, and Cerebrovascular Variables

	Total PCSI score Coefficients (SE)	Standardized coefficients (SE)
Sex (female)	50.81 (8.74) ***	50.81 (8.74)
Number of prior concussions	3.32 (2.33)	9.04 (6.35)
Resting mean blood pressure	−3.21 (0.56) ***	−39.18 (6.80)
Vasoreactivity	−627.12 (337.53) *	−16.45 (8.85)
Autoregulatory slope	−58.27 (18.58) ***	−23.15 (7.38)
Rising slope	23.58 (15.71)	11.77 (7.84)

R² = 0.81 (adjusted R² = 0.73), F_6,15 = 10.38, p < 0.01. Independent variables in the final model (shown in the table) were selected statistically (to minimize Akaike Information Criteria), via a forward/ backward model selection, from a bigger pool of variables (see text). Coefficients shown in bold denote variables that significantly contribute to total PCSI score (*** < 0.001, ** < 0.01, * < 0.05). Standardized coefficients reflect relative contributions of each variable to total PCSI score. Adjusted R² reflects explained variance adjusted for the degrees of freedom relative to the sample size.

Of the 28 individuals in the concussion group at baseline, 27 returned at the 8-week timepoint, with successful repeat assessments in 25 participants (Table 4). At the 8-week timepoint total PCSI score significantly improved (p < 0.01; 17 participants reported being symptom free), however, four individuals reported a PCSI score relatively similar to baseline (Fig 2). Twenty-two reported their average weekly exercise volume (259 ± 159 minutes per week, IQR 126–374 minutes). In the concussion group, there were no statistically significant differences in hemodynamic or cerebrovascular variables between baseline and 8-week timepoints; despite a majority of participants having symptom resolution, the magnitude of increase in vasoreactivity was still highly variable across individuals. At the individual level, an increase in vasoreactivity at the 8-week timepoint was positively associated with the 8-week average exercise volume (i.e., average minutes per week) (Fig 3; R² = 0.23, adjusted R² = 0.19, F _(1,20) = 6.0, p = 0.02).

TABLE 4.

Demographic, Clinical, Hemodynamic, and Cerebrovascular Variables within the Concussed Group at Baseline (<14 Days Post-Injury) and 8-Week Timepoints

	Baseline (n = 25)	8-week (n = 25)	p value
Total PCSI (SD) [Range]	44.4 (25.5) [4–114]	9.8 (17.0) [0–49]	<0.001
Resting RR interval, ms	866 (140)	861 (152)	0.84
RMSSD	54.2 (28.7)	53.1 (33.4)	0.85
Resting blood pressure (mmHg)
Systolic	109.1 (8.8)	109.4 (8.3)	0.87
Diastolic	62.8 (6.4)	61.8 (4.8)	0.48
Mean	78.2 (6.4)	77.7 (4.9)	0.69
Resting EtCO2 (mmHg)	33.4 (3.0)	33.2 (2.6)	0.80
Resting MCAv (cm s−1)	58.6 (12.5)	58.0 (10.8)	0.75
Vasoreactivity (cm s−1 mmHg−1/mmHg CO2)	0.031 (0.013)	0.028 (0.010)	0.22
Autoregulation
Falling slope (cm s−1 mmHg−1)	1.093 (0.524)	0.918 (0.433)	0.53
Autoregulatory slope (cm s−1 mmHg−1)	0.299 (0.208)	0.254 (0.203)	0.59
Autoregulatory range (mmHg)	5.3 (2.6)	5.3 (3.4)	0.85
Rising slope (cm s−1 mmHg−1)	0.900 (0.247)	0.991 (0.305)	0.40

Continuous variables are mean (standard deviation). Categorical variables are count (percent). [ ], range. Significance, p < 0.05. EtCO₂, end-tidal CO₂. Cerebrovascular autoregulation assessments are missing six participants in each group.

PCSI = Post-Concussion Symptom Inventory; RMSSD = root mean square of the successive difference in R-R intervals.

FIGURE 2:

Post-concussion symptom burden at baseline (acute phase, ≤14 days) and 8-week (chronic) assessment timepoint. PCSI, Post-Concussion Symptom Inventory.

FIGURE 3:

Linear relationship between aerobic exercise volume (average minutes per week) and change in cerebral vasoreactivity from baseline to 8-weeks in concussed adolescents and young adults.

Discussion

Our data suggest that a concussion may result in alterations in cerebrovascular regulatory mechanisms, which, in turn, helps explain the variability in resting cerebral blood flow velocity reported in prior work.^13–17 Our results also show that these changes (1) relate to acute (within 2 weeks of injury) post-concussion symptom burden, (2) persist chronically (8-week post-injury) despite symptom resolution, and (3) may be favorably modified by aerobic exercise. These results mechanistically support the current notion of the relationship between physiologic impact of concussion and concussion symptomatology.

Previous work in animals has demonstrated changes in cerebral blood flow after experimental mild traumatic brain injury.⁴ Partly based on this observation, both our group and others have suggested that there may be a metabolic mismatch between energetic requirements of neural activity and energy supply by global and regional perfusion, and this mismatch may underlie post-injury symptom burden.^4,20 However, in humans, demonstration of this mismatch has remained elusive. On one hand, prior research has shown that a concussion could indeed trigger pathophysiologic alterations in cerebral blood flow among pediatric patients.³³ On the other hand, these alterations may result in reduced,^13,14 similar,^15,16 or increased¹⁷ global and regional cerebral blood flow. Our observations provide a simple explanation for this discrepancy. We observed an acute change in key components of cerebrovascular function, vasoreactivity and autoregulation, which was highly variable yet related to cerebral blood flow and symptom burden among our sample of concussed participants. Thus, our results suggest an important role for change in cerebrovascular function, and not in resting cerebral blood flow per se, in pathophysiology of post-concussion symptom burden, although the high variability in this change may obscure its impact at the population level. This finding is supported by a recent study showing that cerebral vasoreactivity, and not cerebral blood flow itself, is different in concussed individuals compared to controls.²²

Our primary observation – that lower mean blood pressure, lower responses to increases in blood pressure, and lower vasoreactivity are associated with greater symptom burden – is counter-intuitive but not unexpected. Traditionally, higher systemic blood pressure, larger cerebrovascular responses to increases in arterial pressure (ie, autoregulation), or higher vasoreactivity are elements seen as an “impairment” in the context of brain injury. However, when interpreted in toto, these changes appear to point to a deliberate and collective physiologic response to increase cerebral perfusion to support neural metabolism: either by increasing systemic arterial (thus, cerebral perfusion) pressure or by increasing cerebrovascular responsiveness to it. This is consistent with the notion that appropriate physiological responses are needed for neurometabolic homeostasis and recovery.^4,20

These observations are consistent with a prior report that lower global and regional cerebral blood flow (presumably consequent to lower cerebrovascular responses) was more common in children and adolescents (8–18 years old) who were symptomatic, compared with controls.¹⁷ In fact, the same study demonstrated that cerebral blood flow was lower in children who recovered quickly, and that clinical recovery preceded recovery of cerebral blood flow.¹⁷ This is consistent with our observation that, longitudinally, cerebrovascular function was not different 8-weeks post-injury compared to the acute phase (<14 days) despite symptom resolution. In a prior study with adults with persistent post-concussion symptoms (162 ± 40 days postinjury) we found that higher vasoreactivity was associated with greater symptom burden.²¹ When considered together with this previous work,^17,21 our current results suggest that (1) cerebral perfusion and cerebrovascular responses are elevated after a concussion and can remain elevated even after overt symptom resolution, (2) the magnitude of this elevation is inversely proportional to acute symptom burden, and (3) cerebrovascular responses remain elevated chronically in those who do not recover²¹ but (4) return to normal in children who recover quickly.¹⁷

Neither the current study nor prior work cited above were designed to explore causal relationships between concussion, cerebrovascular function, and symptom burden. Conventional thinking appears to hold that alterations in cerebrovascular function after a concussion contributes to neurometabolic mismatch and subsequent post-concussion symptom burden.⁴ However, it is important to note that in our cohort, the amount of aerobic exercise over 8 weeks was associated with improved cerebral vasoreactivity (Fig 3), and engagement in exercise has been shown to be an essential component of post-concussion symptom recovery.^24,34,35 This supports the idea that alterations in cerebrovascular function—at least, the increase in vasoreactivity—after a concussion can be protective. However, in our study, most of the patients recovered before their 8-week visit, and thus, we can only provide circumstantial evidence.

As with any other study, our study has its limitations. First, as mentioned above, we could not provide direct causal evidence for a neuroprotective role of the reported changes in cerebrovascular function. This remains to be established in future studies properly designed to investigate causal relationships. Secondly, we could not assess the direct, longitudinal relationship between recovery of symptoms and cerebrovascular function, since two-thirds of our participants recovered by 8 weeks. While interpretation of our cross-sectional and longitudinal results are consistent with the prior data from our group and others, future work is needed to assess the relation between cerebrovascular function and rate of symptom recovery by assessing the two more frequently post-injury. Lastly, we assessed the middle cerebral artery blood flow velocity via TCD as a surrogate for cerebral blood flow. While TCD provides excellent temporal resolution to measure overall dynamic responses that represent downstream effects, it does not have enough spatial resolution to explore specific aspects of changes at the level of small vessels. In addition, TCD provides a measure of flow velocity, and not the absolute flow; the latter is sensitive to differences in vessel diameter. However, our results show a relationship between changes in cerebrovascular function and concussion, and whether this can be attributed to an injury-related change in vessel diameter in response to assessment procedures does not obviate our main conclusion.

Despite its limitations, our results provide evidence for physiologic changes in cerebrovascular function after a concussion in adolescents and young adults. In addition, a plausible mechanistic explanation of how a concussion may lead to a neurometabolic mismatch, which, in turn, may underlie post-concussion symptom burden has been provided. Importantly, our results provide evidence that alterations in cerebrovascular function may be neuroprotective, and that this may underlie the beneficial impact of aerobic exercise on post-concussion recovery.