Regulation of Cerebral Blood Flow after Spinal Cord Injury

CA

CA, which describes the ability of the brain to maintain CBF under a variety of perfusion pressures, can be divided into static and dynamic components. Static CA (sCA) is used to describe CBF during steady-state conditions ²³ and is traditionally considered to be held relatively constant over a wide range of perfusion pressures (i.e., mean arterial pressures (MAP) from 60 to 150 mm Hg). ^{24, 25} Lucas and associates, ²⁶ however, recently reported a relatively “pressure passive” CBF response to both increased and decreased BP induced pharmacologically. The assumption that diameter remained constant during the infusion of vasoactive substances was made based on the results of Giller and colleagues, ²⁷ who measured the outer diameter during craniotomy and found minimal change in diameter of 4%. Volumetric blood flow through the internal carotid artery during steady state phenylephrine infusions, however, showed increases in middle cerebral artery blood velocity, but apparently no measurable change in internal carotid artery flow, ²⁸ raising the possibility that elevations in flow velocity may reflect increase in arterial vascular tone, potentially of the insonated vessel itself. Such changes would invalidate the use of transcranial Doppler as an accurate surrogate of CBF. Discrepancies between these studies have not been resolved. Nevertheless, cerebral autoregulation has been shown to be reduced in acute ischemic stroke, ²⁹ traumatic brain injury, ³⁰ carotid artery stenosis, ³¹ and autonomic failure. ³²

Dynamic CA (dCA) on the other hand, describes the ability of the cerebrovascular system to oppose short-term changes in perfusion pressure over a period of less than 5 sec. ^23,33 Both sCA and dCA are thought to be regulated by a combination of myogenic, metabolic, neurogenic control mechanisms (for a review, see Paulson and coworkers, ²¹ ); however, sCA and dCA and may rely on relatively different influences from the various regulatory factors. ³⁴ For example, it has been suggested that neural control over cerebral vascular tissue more heavily influences dCA than sCA, ³⁴ while endothelial function has been shown to not relate to dCA. ³⁵ Myogenic control, on the other hand, is thought to play a role primarily in passive, compliance-based dCA (i.e., transfer function analysis phase) and less of a role in active, resistance-based dCA (i.e., gain and coherence). ^36,37 Further, although on a continuum, dysfunction in dCA is thought to occur before sCA in some pathological scenarios. ²³

Very recently, Pukayastha and colleagues ³⁸ showed, using prazosin (α_1a antagonist), that sympathetic control is important to dCA (namely, low-frequency [LF] gain) at rest and during exercise. ³⁸ Ogoh and associates ³⁹ also showed that the middle cerebral artery velocity (MCAv) responsiveness to thigh cuff deflation was reduced after generalized α-adrenergic blockade. ³⁹ Further, evidence showing increased transfer function analysis gain after cholinergic blockade through glycopyrrolate in the same frequencies as sympathetic control (i.e., >0.05 Hz) suggests that both the parasympathetic and sympathetic system are active in dCA that may predominantly oppose rising and falling perfusion pressures, respectively. ⁴⁰ Recent evidence in both humans ^16,41,42 and animals, ⁴³ however, supports the idea of hysteresis in CA. In other words, the brain can more effectively buffer hypertension compared with hypotension. Whether this is the case for SCI remains unknown, but if comparable to healthy humans would provide a means to defend against autonomic dysreflexia.

Cerebrovascular reactivity

The index of cerebrovascular reactivity (i.e., relative change in CBF in response to altered blood gas concentration) allows for quantification of the sensitivity of the intracranial vessels to a given stimulus. Most commonly, stimuli include either pharmacological or ventilatory driven alterations of arterial blood gas content (i.e., oxygen or carbon dioxide). A reduced CBF response to hypocapnia and hypercapnia has been shown to be related to obstructive and central sleep apnea, ^44,45 hypertension, ⁴⁶ carotid artery stenosis, ⁴⁷ congestive heart failure, ⁴⁸ and have independent predictive value for future ischemic stroke. ^{49 –51}

The overarching control of CBF is via PaCO₂. Specifically, CBF adjusts roughly 3.8%·mmHg⁻¹ change in PaCO₂ partial pressure, with a slightly more robust response to hypercapnia in comparison with hypocapnia. ^{52 –56} Briefly, CBF is thought to preferentially alter CO₂ and [H+] at the level of the brain stem to maintain pH within a narrow range. ⁵² Maintenance of cerebral pH helps stabilize respiratory and chemoreceptor centers, which are also sensitive to [H+]. The mechanisms through which this is achieved are only partially understood but thought to be the result of activated potassium channels in smooth muscle as well as rapid adjustment of several key vasoactive substances such as nitric oxide and prostaglandin. ⁵²

Neurovascular coupling

The close temporal and regional linkage between neural activity and CBF response is termed neurovascular coupling. CBF responses in the MCA, posterior cerebral artery and anterior cerebral artery to visual, verbal, and cognitive tasks have been widely measured. ^{57 –61} Previous work shows altered neurovascular coupling in those with hypotension, ⁶² type I diabetes, ⁶³ and Alzheimer disease ⁶⁴ ; however, to our knowledge, no study has directly related the neurovascular coupling response to cognitive function. The metabolic portion of CBF regulation is related directly to increased oxygen availability and reduced carbon dioxide during periods of increased CBF. ⁶⁵ Similarly to working muscle, a complex interplay occurs, whereas nitric oxide and other endothelial metabolites work to match blood flow provision to metabolic demand. ^66,67 This functional hyperemia uses intimate interactions between cerebral vascular cells, such as endothelium and smooth muscle, with neurons and glia (i.e., astrocytes, microglia, oligodendrocytes) to match neural activation with CBF. ⁶⁸

Cardiovascular consequences of SCI

The autonomic nervous system is composed of both sympathetic and parasympathetic divisions. Parasympathetic nervous outflow occurs from cranial nerves III, VII, IX, and X superiorly and S2–4 inferiorly. Only cranial nerves IX and X participate in cardiovascular control, and as they exit above the spinal cord level; therefore any parasympathetic influence on cerebrovascular after SCI tone would be intact. ⁸⁷ On the other hand, sympathetic nervous outflow occurs from the sympathetic preganglionic neurons (SPNs) localized within T1–L2 spinal segments of the spinal cord. These neurons synapse into the sympathetic paravertebral ganglia (ganglionic neurons in the sympathetic chain), ⁸⁸ and finally postganglionic fibers innervate target organs in the periphery. After SCI, tonic supraspinal control to the spinal SPNs is lost, and sympathetic outflow from the spinal cord below the level of injury is severely disrupted. ⁴ Drastic changes in systemic cardiovascular regulation occur in those with lesion levels above the sixth thoracic spinal cord segment. Lesions above T6 are associated with a loss of supraspinal control over the heart and splanchnic blood vessels, ⁸⁹ both of which are needed for effective long- and short-term BP regulation. ⁹⁰ Further to this, sympathetic control of the cerebral blood vessels is transmitted through the superior cervical ganglion, which originates from spinal nerves T1–T4. ⁹¹ As such, any SCI occurring above T4 would be expected to result in partial or complete (>T1) abolishment of sympathetic cerebrovascular control.

Also highly relevant to cerebrovascular regulation, endothelial function, estimated by flow mediated dilatation, has been shown to be systemically reduced after SCI, ⁹² in some cases in as little as 6 weeks post-injury. ⁹³ Nitric oxide is the primary substance causing vasodilatation in response to sheer-stress, ⁹⁴ and leads to smooth muscle relaxation and increased blood flow. ⁹⁵ The production of nitric oxide relevant to vascular control occurs in endothelial cells ⁹⁶ and mediates the relationship between cerebral vascular tone and the majority of signaling messengers (i.e, acetylcholine, adenosine diphosphate, adenosine triphosphate, bradykinin, endothelin, sodium fluoride, oxytocin, vasopressin, etc.). ⁹⁷ Reduced nitric oxide, however, does not appear to influence cerebral autoregulation but is associated with reduced cerebrovascular reactivity. ^{98 –101} After this, it could be suspected that systemic endothelial dysfunction influences the regulation of CBF (Fig. 2).

OPEN IN VIEWER

FIG. 2.

Illustration of theoretical mechanisms of cerebral blood flow control that would be impacted by high-level spinal cord injury (SCI). Briefly, decreased central sympathetic control of systemic blood vessels leads to greatly reduced cerebral perfusion pressure (CPP). Exacerbating the hypotension, those with SCI above T5 also have impaired central sympathetic control over the heart. Systemic endothelial (Endo) dysfunction may also lead to decreased CO₂ reactivity, neurovascular coupling, and potentially cerebral autoregulation. As mentioned in Figure 1, direct neurogenic control of cerebral vascular tone is controversial, but cerebral vessels are innervated by sympathetic fibres originating from the T1–4 level (superior cervical ganglia). As such, SCI above T5 would result in impaired or absent central sympathetic control of cerebrovasculature. denotes where cerebrovascular dysfunction may occur after SCI (1-endothelial dysfunction, 2- altered direct neural cerebrovascular influence, 3- sympathetic neural control over total peripheral resistance (TPR), 4- sympathetic neural control over cardiac output [Qc]). *Note that when a complete injury occurs, sympathetic vasomotor control is disrupted below the lesion level. ICP, intracranial pressure; CVP, central venous pressure.

sCA

Early studies from Nanda and coworkers ^72,73 that examined CBF using inhaled ¹³³Xenon clearance rate during supine-to-seated and supine-to-raised leg/lower body negative pressure (LBNP) maneuver reported that CBF was well maintained during an average 18 mm Hg drop in MAP (Fig. 3). ^72,73 Handrakis and colleagues ⁷⁸ provided more insight by measuring MCAv after angiotensin-converting enzyme-inhibitor infusion combined with 45 degrees tilt; results revealed that MCAv decreased to a similar rate in chronic HLSCI patients compared with able bodied (AB), although the MAP drop in HLSCI patients was markedly higher (p=0.02). More recent work has shown that sCA (albeit using the relationship between MAP and MCA [diastolic]) was similar between HLSCI and AB patients in response to head-up tilt, although the HLSCI group reported greater decreases in end-tidal CO₂. ⁸⁰ These authors showed an interesting correlation between supine plasma noradrenaline concentrations (but not low frequency oscillations in systolic BP) and sCA, suggesting a relationship between sCA and autonomic completeness. ⁸⁰

OPEN IN VIEWER

FIG. 3.

Illustration summarizing the primary studies examining static and dynamic cerebral autoregulation (dCA, sCA). 1 ⁷² ; 2 ⁷³ ; 3 ⁷⁸ ; 4 ⁷⁷ ; 5 ⁸⁰ (reduced phase, increased gain); 6 ⁷⁴ ; 7 ⁷⁹ .

Using a similar ¹³³Xenon technique in primarily patients with acute SCI (i.e., 92% <12 months since injury), Yamamoto and associates ⁷⁴ showed that the sCA (=ΔCBF/ΔMAP) response to 30-degree tilt was impaired in acute HLSCI (i.e., greater changes in CBF for a given change in MAP); however, those who had injuries occurring within 6 months of examination had significantly impaired sCA compared with those closer to the chronic phase. Because CBF is the final common pathway before the development of syncope, it is not surprising that reduced static CBF control is a major factor related to the increased risk of orthostatic intolerance after SCI. In support of this, Gonzalez and colleagues ⁷⁵ showed that orthostatically intolerant acute HLSCI had greater decreases in MCAv during steady-state 80 degrees tilt when compared with a tolerant chronic HLSCI group. ⁷⁵

A landmark study by Houtman and coworkers ⁷⁷ examined steady-state CBF velocity response to 5-min long bouts of progressive LBNP in a group of HLSCI patients. This study showed remarkably poor MAP control in HLSCI and large reductions in MCAv, but similar changes in cerebrovascular resistance (i.e., ΔMAP/ΔCBF) compared with AB indicating similar sCA. The large reductions in CBF velocity resulted in a trend for lower cerebral oxygenation in HLSCI. ⁷⁷ The finding that CBF is reduced more than perfusion pressure during LBNP has long been shown in AB, and although compelling evidence now exists to the contrary, was originally thought to result from systemic sympathetic activation of both muscular and cerebral vasculature during central hypovolemia. ^41,42,102 The majority (i.e., 8/10) of the HLSCI group in Houtman's the later work of Houtman and associates, ⁷⁷ however, had a decentralized superior cervical ganglion (as the result of complete injury above T1) and likely would not be capable of sympathetically mediated increases in cerebral vascular resistance. ⁷⁷ Because both HLSCI and AB persons reported similar changes in end-tidal CO₂ (i.e., a surrogate marker of PaCO₂), the findings of Houtman and colleagues ⁷⁷ support the contention that cerebrovascular changes during orthostatic challenge are primarily related to changes in PaCO₂. ^{103 –105}

Finally, two interesting studies examined the CBF velocity response to increasing BP in those with HLSCI after 2 min of cold water immersion of the hand ⁶⁹ or foot. ⁷⁰ Although not originally interpreted in this manner, it is likely that both these studies induced autonomic dysreflexia in HLSCI. ¹⁰⁶ As such, it is expected that in their HLSCI group (C4–C7), unopposed sympathetic outflow would have occurred not only in the systemic vascular bed, but also the cerebrovascular tissue. ⁸⁷ Both of these studies reported similar average cerebrovascular resistance during autonomic dysreflexia in HLSCI when compared with AB. Further, these studies also included a group of participants with injuries ranging T4–T6, which would be expected to have preserved central sympathetic control over the cerebrovasculature ⁹¹ and showed that when the nociceptive stimulus is applied below the injury level (i.e., the foot), increases in cerebrovascular resistance do not occur during autonomic dysreflexia, ⁷⁰ supporting the idea that sympathetic cerebrovascular control helps to mitigate flow increases during transient periods of increased perfusion pressure. ^42,43

Together, the available literature suggests that sCA is well maintained in those with chronic HLSCI, but may be impaired in the acute phase of injury. Although the early studies showed that absolute CBF response is similar in HLSCI compared with AB, more recent work shows that dysfunctional BP control leads to larger fluctuations in CBF, but similar changes in resistance. ⁷⁷ In one study, a small subset of three HLSCI patients suggests a strong systemic influence of sCA, because abdominal binders and compression stockings improved this metric. ⁷⁴ It is likely that cerebrovascular regulatory factors other than sympathetic control allow for preserved sCA in response to decreasing BP in HLSCI with sympathetic decentralization and endothelial dysfunction (i.e., the myogenic influence). In response to autonomic dysreflexia, those with injuries above the superior cervical ganglion are able to increase cerebrovascular resistance to match BP increases, while those with lower level lesions (with preserved central sympathetic control over cerebrovasculature) do not, likely because of baroreflex mediated sympathetic withdrawal above the lesion level. ⁸⁹

dCA

Only three studies have examined dCA in those with HLSCI, all of which used spontaneous transfer function analysis between MAP and MCAv and one with an additional analysis of MAP and posterior cerebral artery velocity (PCAv) (Fig. 3). ^79,80,82 Recently, Wilson and associates ⁷⁹ showed that supine coherence in the very low frequency (VLF) range was reduced in their group of six persons with HLSCI, while gain and phase were similar. ⁷⁹ We later showed that upright MAP-MCAv coherence was reduced in both the range. ⁸² Similarly, we reported reduced PCAv coherence in the LF range. ⁸² Sahota and colleagues ⁸⁰ showed that the supine LF gain was reduced and upright LF phase was increased, both indicative of reduced dCA. The latter study evaluated only the LF transfer function analysis for MAP-MCAv when coherence was >0.5. ⁸⁰ To our knowledge, no existing studies examine dCA using a model of perturbed BP/CBF SCI to improve the reliability of transfer function analysis. ¹⁰⁷

Considering the combined findings of these studies, spontaneous dCA is altered in those with HLSCI, and is possibly related to both sympathetic decentralization of the cerebrovasculature and increased parasympathetic tone after HLSCI. ¹⁰⁸ A fundamental inverse relationship has recently been shown between both spontaneous and non-spontaneous dCA and cardiac baroreflex sensitivity (BRS) in healthy AB patients. ¹⁶ According to this relationship, and our recent review highlighting that cardiac BRS is markedly reduced in HLSCI, it could be expected that coherence and MAP-MCAv gain would be reduced while phase would be increased. ⁹ It is possible that cardiovascular and autonomic pathology after HLSCI leads to a loss of this fundamental compensatory relationship. ¹⁶ More work should be completed, however, using larger sample sizes, incorporating VLF ranges as well as driving BP changes using either oscillatory LBNP or tilt to increase confidence in transfer function analysis estimates. ¹⁰⁹

Cerebrovascular reactivity

Early work using ¹³³Xenon as well as more current work using transcranial Doppler shows that the CBF response to hypercapnia is preserved in those with HLSCI. ^71,74,79 On the other hand, the response of CBF to hypocapnia does not appear as consistent. For example, Eidelman and colleagues ⁷¹ reported that the CBF decrease in response to hypocapnia was abolished after HLSCI. In contrast, other reports showed that the hypocapnic response of CBF is preserved. ^72,73,79 The preliminary study in this group by Eidelman et al., ⁷¹ however, may be confounded as four of the nine HLSCI participants to have BP recorded during hyperventilation showed BP changes meeting the criteria for autonomic dysreflexia, ⁸⁸ changes that would induce systemic, and more controversially, cerebral vasoconstriction. ⁸⁹ The vasoconstriction induced by autonomic dysreflexia would thereby augment hypocapnie induced vasoconstriction. ¹¹⁰ Differences between Eidelman and associates ⁷¹ and Nanda and colleagues ^72,73 in terms of hypocapnic CBF response in HLSCI are difficult to explain but may be because of different techniques for measuring CBF (i.e., modified vs. non-modified ¹³³Xenon inhalation).

Although hampered by a low sample size, Wilson and coworkers ⁷⁹ reported differences in subtle parameters of CBF between HLSCI and AB during both hypo- and hypercapnia. Pulsatile flow (pulsatility index) of MCAv decreased significantly more in response to hypercapnia in the HLSCI group, while the cerebrovascular resistance response to hypocapnia was reduced. These values suggest nuanced cerebrovascular changes occur after HLSCI, which influence the CBF response to PaCO₂. In support of the idea that the hypocapnic CBF response is blunted in HLSCI, Sahota and colleagues ⁸⁰ also showed that end-tidal CO₂ decreased significantly more in HLSCI compared with AB during upright tilt, but diastolic CBF velocity changed similarly. Although end-tidal CO₂ decreases may have been secondary to greater increases in dead space and/or decreases in cardiac output during head-up tilt in SCI compared with AB, ^111,112 this differential response between HLSCI and AB suggests a reduced CBF sensitivity to hypocapnia.

Pulsatile flow patterns represent an area of increasing interest within the field of CBF regulation, which has been suggested as a primary mechanism through which the brain maintains perfusion during low mean perfusion pressure, such as severe hemorrhage, and orthostatic hypotention, the latter of which is widespread in HLSCI. ^{89,113 –115} Hypercapnia preferentially alters the small distal resistance vessels, which regulates pulsatile flow, ¹¹⁶ and the adaptation of this vascular bed in HLSCI, as shown by Wilson and associates, ⁷⁹ may represent an adaptation to chronically altered hemodynamic and/or respiratory patterns in those with HLSCI as a means to maintain cerebral perfusion in the face of reduced perfusion pressure. ⁷⁹ Reduced cerebrovascular reactivity, especially to hypocapnia, has been previously related in patients and healthy persons to sleep apnea. ^48,117 Sleep apnea has been reported to occur in up to 40% of those with HLSCI, and cerebrovascular reactivity may be an important mediating factor. ¹¹⁸ Clearly, further examination of this relationship is warranted, especially considering the increased risk of stroke in the HLSCI population and the independent predictive value of cerebrovascular reactivity for stroke. ¹⁰⁰

Neurovascular coupling

Neurovascular coupling describes the hyperemic response to cognition. To our knowledge, only two studies have examined neurovascular coupling in those with HLSCI. Wecht and colleagues ⁷ showed that mean MCAv and cerebrovascular resistance did not change when completing a Stroop test in either AB or HLSCI. This technique is quite different from the established neurovascular coupling procedures, because Wecht and colleagues ⁷ reported mean values from three repeated 45-sec long Stroop tests, and it is not clear if an eyes-closed resting period occurred, or if the peak response (as opposed to the average) was different between groups. In many other studies, albeit using more established protocols, MCA, anterior cerebral artery, and posterior cerebral artery CBF velocity have been shown to increase in healthy controls during cognitive tasks and visual motor stimuli while cerebrovascular resistance decrease. ^57,59,119 Regardless, these authors reported a trend suggestive of a differing response in controls compared with combined HLSCI and low-level SCI (p=0.08), and a significant positive relationship between changes in cerebrovascular resistance during cognition and Stroop test performance in the AB group (indicating a relationship between neurovascular coupling and cognitive performance); this relationship did not occur in the HLSCI group. ⁷

Recently, we have revealed that a change in posterior cerebral artery CBF velocity during visual cortex stimulation is essentially absent in those with HLSCI (Phillips and associates, unpublished data). We also have shown that the hyperemic response to cognition improves after administration of an α_1a agonist (midodrine hydrochloride, which does not cross the blood brain barrier), albeit still reduced compared with matched controls (unpublished observations). Together, the scant available literature suggests that the neurovascular coupling response is heavily dependent on the underlying BP and systemic vascular tone, with HLSCI and decentralized systemic sympathetic vascular control resulting in an inability to match CBF to increases in neuronal activity. These preliminary data seem to link the plethora of evidence showing cognitive function is declined in those with hypotension with and without SCI. ^{62,120 –123} Other than hypotension, it is possible that nitric oxide availability, ¹²⁴ glucose intolerance, ^125,126 and dyslipidemia ¹²⁷ also mediate the relationship between altered neurovascular coupling and HLSCI. Interestingly, cessation of statin therapy, which acutely increases stroke risk, ¹²⁸ rapidly reduced neurovascular coupling through what is thought to be endothelium dysfunction. ⁶⁰