Cerebral hemodynamic impairment

Responses of the cerebrovasculature to reduced perfusion pressure.

Cerebral perfusion pressure in any arterial vascular territory is equal to the difference between the mean arterial pressure and the venous back pressure. Normally, the venous back pressure is negligible and the perfusion pressure is equal to the mean pressure within that arterial system. Arterial stenosis or occlusion can cause a reduction in the pressure in the distal arterial vessels. However, the degree of stenosis or the presence of arterial occlusion does not accurately predict the hemodynamic status of the distal circulation.1 Although stenoses of the extracranial carotid artery resulting in reductions in luminal diameter greater than 50 to 70% are known to reduce the distal pressure in some cases,4 collateral circulation can maintain normal cerebral perfusion pressure and normal flow in many of these patients. Up to 60% of patients with complete occlusion of the carotid artery may have no evidence of hemodynamic compromise in the distal cerebral circulation.5

When collaterals are not adequate to maintain normal perfusion pressure, reflex vasodilation occurs to maintain normal blood flow.6 This response, as well as the reflex vasoconstriction observed with increased perfusion pressure, is known as autoregulation. Autoregulatory vasodilation, or Stage 1 hemodynamic compromise, maintains normal flow by reducing the vascular resistance to arterial inflow. With further reductions in perfusion pressure, the capacity of autoregulatory vasodilation to maintain normal blood flow is overcome and blood flow begins to decrease. Although the delivery of oxygen falls, the brain can increase the amount of oxygen it extracts from the blood (oxygen extraction fraction [OEF]) to maintain normal cerebral oxygen metabolism (CMRO₂) and function (figure). 7,8 This phenomenon of reduced blood flow and increased oxygen extraction has been termed “misery perfusion” or Stage 2 hemodynamic failure.1,9 Once oxygen extraction becomes maximal, further decreases in perfusion pressure (and consequently blood flow) will lead to disruption in normal oxygen metabolism and ultimately to infarction.10

• Download figure
• Open in new tab
• Download powerpoint

Figure. Increased oxygen extraction despite normal blood volume and mean transit time. This patient is a 67-year-old man with a right internal carotid artery occlusion who presented with a right posterior frontal infarction 8 months before the PET examination. Note that the right hemisphere is on the right side of the PET images by laboratory convention. The top row of images demonstrate misery perfusion: The flow (cerebral blood flow [CBF] image on the left) is reduced (white arrows); the oxygen extraction fraction (OEF) is increased (white arrows on middle image); and cerebral oxygen metabolism (CMRO₂) remains normal and symmetric. The ratios of ipsilateral to contralateral mean hemispheric values of CBF and OEF measured in the middle cerebral artery territory in this patient fell outside of the range of ratios found in 18 normal control subjects, confirming the CBF reduction and OEF elevation distal to the carotid occlusion. The ipsilateral to contralateral CMRO₂, cerebral blood volume, and mean transit time (MTT) ratios were within the normal range, and the CBV and MTT images (bottom row) appear symmetric.

Methods of hemodynamic assessment.

Single measurements of cerebral blood flow (CBF) alone do not adequately assess cerebral hemodynamic status. First, normal values may be found when perfusion pressure is reduced, but CBF is maintained by autoregulatory vasodilation. Second, CBF may be low when perfusion pressure is normal. This can occur when the metabolic demands of the tissue are low. Reduced flow caused by reduced metabolic demand may not cause confusion when low regional CBF is measured in areas of frank tissue infarction. However, blood flow can also be reduced in normal, uninfarcted, tissue due to the destruction of normal afferent or efferent fibers by a remote lesion as well.5

Three basic strategies have been developed to assess regional cerebral hemodynamic status noninvasively. The normal compensatory responses of the brain and its vasculature to reduced perfusion pressure, as outlined previously, are assumed to be present. The first two strategies are used to indirectly identify the presence and degree of autoregulatory vasodilation (Stage 1). The third relies on direct measurements of oxygen extraction (Stage 2). It is important to note that these approaches are accurate only with uninfarcted brain tissue, studied at a time that is remote from a cerebral ischemic event.11

The first strategy relies on paired blood flow measurements with the initial measurement obtained at rest and the second measurement obtained following a cerebral vasodilatory stimulus. Hypercapnia, acetazolamide, and physiologic tasks such as hand movement have been used as vasodilatory stimuli. Normally, each will result in a robust increase in CBF. If the CBF response is muted or absent, preexisting autoregulatory cerebral vasodilation due to reduced cerebral perfusion pressure is inferred. Quantitative or qualitative (relative) measurements of CBF can be made using a variety of methods, including ¹³³xenon by inhalation or intravenous injection, SPECT, stable xenon CT (Xe-CT), PET, and MRI.5,12 Changes in the velocity of blood in the middle cerebral artery trunk or internal carotid artery can be measured with transcranial Doppler (TCD) and MRI.13,14 The blood flow or blood velocity responses to these vasodilatory stimuli have been categorized into several grades of hemodynamic impairment: 1) reduced augmentation (relative to the contralateral hemisphere or normal controls); 2) absent augmentation (same value as baseline); and 3) paradoxical reduction in regional blood flow compared with baseline measurement. This final category, also called the “steal” phenomenon, can only be identified with quantitative CBF techniques.15

The second Stage 1 strategy uses either the measurement of regional cerebral blood volume (CBV) alone or in combination with measurements of CBF in the resting brain to detect the presence of autoregulatory vasodilation. The CBV/CBF ratio (or, inversely, the CBF/CBV ratio), mathematically equivalent to the vascular mean transit time,5 may be more sensitive than CBV alone for the identification of Stage 1 hemodynamic compromise.16 However, it may be less specific. The CBV/CBF ratio may increase in low flow conditions with normal perfusion pressure, such as hypocapnia.17 Quantitative regional measurements of CBV and CBF can be made with PET or SPECT.5 MR techniques for the quantitative measurement of CBV have been developed.18 Patients are identified as abnormal with these techniques based on comparison of absolute quantitative values or hemispheric ratios of quantitative values to the range observed in normal control subjects. One issue that remains unresolved is to what extent autoregulatory vasodilation of arterioles gives rise to measurable increases in the CBV. Experimental data have produced conflicting results.19-21 Increases in CBV are found distal to stenotic or occluded carotid arteries in some patients. Thus, the sensitivity and specificity of CBV measurements for detecting reduced CPP is not known.

The third strategy relies on direct measurements of OEF to identify patients with increased oxygen extraction (Stage 2). At present, regional measurements of OEF can be made only with PET using O-15 labeled radiotracers.22 Both absolute values and side-to-side ratios of quantitative and relative OEF have been used for the determination of abnormal from normal. MRI measurements using pulse sequences sensitive to deoxy-hemoglobin, which is increased in regions with increased oxygen extraction, are being developed to provide similar information.23

Correlative data between methods.

The results of paired-flow studies with vasodilatory stimuli in any given patient may differ. The effects of acetazolamide on blood flow are different from hypercapnia or physiologic activation. For example, Kazumata et al. measured an increase in blood flow with hypercapnia in 10 of 11 patients with an absent increase or paradoxical decrease in CBF after acetazolamide.24 Inao et al. found a similar discordance between results with acetazolamide and neural activation.25 They found that CBF increased bilaterally with bimanual activation, but only contralaterally with acetazolamide. The exact mechanism by which acetazolamide causes an increase in flow is unknown. The evidence points to local, rather than systemic, effects on the cerebrovasculature.26 Possible local mechanisms include direct vasodilatory actions or secondary local metabolic changes due to carbonic anhydrase inhibition.26,27 The effect of acetazolamide can vary in response to the total dose, the timing of CBF measurement after the dose, and the age and sex of the patient.26,28,29 One possible explanation for the discordant results seen with acetazolamide and other vasodilatory stimuli was put forward by Inao et al.25 They suggested that regions with reduced baseline flow receive a lower dose of acetazolamide than regions with normal blood flow. The lower acetazolamide dose will result in a greater likelihood that a reduced or absent blood flow response will be seen with acetazolamide in these regions compared with hypercapnia alone, which produces a uniform stimulus throughout the brain.

An additional factor that may contribute to the discordance observed between acetazolamide and CO₂ or activation studies is the use of nonradioactive xenon as a blood flow tracer. The inhalation of xenon gas causes increases in blood flow.30 The steal phenomenon with acetazolamide is reported with much greater frequency with stable xenon blood flow methods than with O-15 labeled PET radiotracers.31,32 It is possible that the preexisting vasodilation caused by xenon augments the dose-dependent effects of acetazolamide. Finally, absolute blood flow, relative blood flow, and blood velocity, although related phenomena, are not equivalent. Dahl et al. found no relationship between the increase in velocity (by TCD) and the increase in blood flow (by Xe-CT) after acetazolamide.28

The correlation between the CBV/CBF ratio (or CBF/CBV ratio) and paired flow measurements has been good.33-35 These studies have compared measurements of CBV/CBF at rest and the quantitative CBF response to CO₂,33 the relative CBF response to acetazolamide,34 and blood velocity changes after CO₂ inhalation.35

The correlation between methods that evaluate Stage 1 and Stage 2 hemodynamic compromise in patients with cerebrovascular disease has been inconsistent. One issue that remains unclear is to what extent maximal autoregulatory vasodilation persists when oxygen extraction fraction is elevated, particularly in humans with chronic reductions in perfusion pressure. We have observed patients with increased OEF who did not have increased CBV or mean transit time (see figure). Whether this latter finding represents a variable vasodilatory capacity or some form of long-term compensatory response remains to be determined. Therefore, both the independence of the mechanisms of autoregulatory vasodilation and oxygen extraction and the changes occurring over time may be sources of difficulty in comparing the results of Stage 1 and Stage 2 methods.

Several studies have investigated the relationship between acetazolamide reactivity and OEF. The most consistent results were published by Hirano et al.,34 who reported that all seven patients with acetazolamide responses below the normal range also had OEF values above normal. However, some patients with increased OEF had normal acetazolamide responses. A sensitivity of 45% and a specificity of 98% of the steal phenomena after acetazolamide for increased OEF were reported by the same investigators.36,37 Similar data were reported by Nariai et al.38 Other investigators have reported no significant relationship between changes in blood flow after acetazolamide and quantitative values of OEF.31

Two comparative studies of CO₂ reactivity and OEF found linear relationships between the change in quantitative flow after CO₂ and the absolute value of quantitative OEF.33,39 However, in both studies, a threshold value CO₂ reactivity that included all patients with increased OEF would also include many patients with normal OEF (low specificity). Sugimori et al. found no relationship between blood velocity changes after CO₂ and OEF.35

The data for the correlation between the CBV/CBF or CBF/CBV ratios and OEF is variable. Two investigators found linear relationships.33,40 Two other studies found no correlation when all patients were included in the analysis.31,41

Association with stroke risk.

Several studies have evaluated the association of these different manifestations of hemodynamic impairment with subsequent ischemic stroke in patients with cerebrovascular disease. Proof that any of these abnormalities is an independent risk factor for subsequent stroke requires that several criteria be met (table 1).42 The current data regarding the association between hemodynamic compromise and stroke risk will be reviewed with these criteria in mind. To date, none of the hemodynamic measurements discussed in this review have been established as causal risk factors.2

First, the outcome must be defined in terms that allow accurate and reproducible measurement. The definition of outcome in all of the studies has been clinical stroke. Second, the population must be clearly defined and sampled in a valid manner. Problems with the definition have been 1) the inclusion, without distinction, of symptomatic and asymptomatic patients; and 2) the inclusion of mixed anatomic pathologies (extracranial carotid stenoses and complete occlusions, intracranial stenoses, and occlusions). In both instances, considerable differences in stroke risk may be present.43 Problems with sampling have included 1) failure to specify how the study cohort was derived from a larger sample, and 2) samples based on referral of patients for clinical evaluation of recalcitrant or unusual symptoms.

Third, the risk factor must be clearly defined and measured reproducibly. The risk factor in each of these studies is some manifestation of hemodynamic impairment. The distinction between abnormal and normal cerebral hemodynamic status necessarily differs between each method of hemodynamic assessment. However, each of the methods has been well validated in terms of quantitative accuracy and reproducibility.

Fourth, the association of the risk factor and the outcome must be performed in a valid manner. Ideally, studies associating risk factors with outcome should be as free from potential bias as possible. Problems encountered in many of the current studies include 1) retrospective assignment to low- and high-risk groups based on analysis of the relationship between risk factor and outcome (rather than prospective assignment based on predefined criteria for the presence or absence of a risk factor); 2) large numbers of patients censored owing to surgical revascularization or loss to follow-up; 3) failure to specify the number of patients lost to follow-up; 4) lack of blinding in ascertainment of study endpoint; and 5) proper statistical analysis to show a statistically significant relationship between the risk factor and the outcome. These problems introduce confounding biases that may invalidate the results of the study. For example, the reason many patients are lost to follow-up may be that they have reached the endpoint in question (ipsilateral stroke). Also, if investigators assessing outcome know that a risk factor is present, they may unconsciously search harder for evidence that an endpoint had been reached.

Fifth, the strength and independence of the association should be explored. Most studies did not address either issue adequately. Strength of association can be evaluated by a number of methods, including 95% confidence intervals for relative risk, odds ratios, or risk ratios. The independence of a risk factor must be tested against other known risk factors for the same outcome. These risk factors must be assessed and evaluated in a multivariate analysis. For example, patients with carotid occlusion and hemodynamic compromise may have a higher incidence of hypertension than patients with normal cerebral hemodynamics. This difference alone might account for a higher risk of stroke in the former group of patients.

Finally, a randomized, controlled trial that proves that modification of the risk factor changes outcome is necessary to prove a causal relationship between a risk factor and an outcome (the sixth requirement).

Several investigators have studied the association of Stage 1 hemodynamic compromise with stroke risk (table 2). Four found an association with stroke risk and three found none. However, all of these studies suffered from significant methodologic problems. Kleiser and Widder reported an association between abnormal blood velocity responses to hypercapnia (by TCD) and the risk of subsequent stroke in 85 patients with carotid occlusion.44 Both symptomatic and asymptomatic patients were included. The risk of contralateral stroke in the patients with a diminished or exhausted CO₂ reactivity was increased, which suggests that the groups were not matched for other stroke risk factors, which were not evaluated. A subsequent study by these same authors reported the outcome of 86 patients with carotid occlusion.45 A much lower risk of stroke was observed in this second study and the number of asymptomatic patients was not reported. ⇓

Yonas et al. reported an association of the steal phenomenon (reduced blood flow by Xe-CT) after acetazolamide and subsequent stroke.46 This study included patients with high grade carotid stenosis and patients with carotid occlusion. The hemodynamic data of patients with subsequent stroke were analyzed retrospectively to establish threshold values for the categorization of high- and low-risk groups. These authors subsequently repeated the analysis with an additional 27 patients.47 The hemodynamic criteria used to establish high- and low-risk groups were different from the prior analysis. Three of the five new strokes that occurred did so in patients who would not have met criteria for the first study and the definition of clinical outcome included contralateral stroke. Only two of these five new strokes were in the hemodynamically compromised territory of the occluded vessel.

Three studies have failed to find an association of Stage 1 hemodynamic impairment and stroke risk.48-50 The largest and most methodologically sound study was reported by Yokota et al.50 They prospectively evaluated 105 symptomatic patients with mixed lesions (unilateral occlusion or severe stenosis [>75% in diameter] of the ICA or proximal MCA) with a SPECT study of relative cerebral blood flow using ¹²³I IMP and measurement of cerebrovascular reactivity using acetazolamide. Other stroke risk factors were prospectively assessed. Thirteen strokes occurred during a median follow-up of 2.7 years: 7 strokes occurred in 39 patients with abnormal hemodynamics and 6 in the 39 patients with normal hemodynamics. The investigators were not blind to the results of the hemodynamic study. A relatively large number of patients (16) were censored from the study because of subsequent cerebrovascular surgery and a significant number of patients (11) were lost to follow-up.

Two studies have investigated the relationship between Stage 2 hemodynamic compromise and stroke risk (table 3). Both reported a positive association. The study by Yamauchi et al. was limited by the inclusion of patients with many different vascular lesions and a large number of censored patients.51 The strongest evidence for an association of hemodynamic compromise and stroke risk was provided by the St. Louis Carotid Occlusion Study.52 This was a blinded, prospective study of 81 patients with symptomatic carotid occlusion, which also specifically assessed the impact of other risk factors. The risk of all stroke and ipsilateral ischemic stroke in symptomatic Stage 2 subjects was significantly higher than in those with normal OEF (log rank p = 0.005 and p = 0.004, respectively). Univariate and multivariate analysis of 17 baseline stroke risk factors confirmed the independence of this relationship. The age-adjusted relative risk conferred by Stage 2 hemodynamic failure was 6.0 (95% CI 1.7 to 21.6) for all stroke and 7.3 (95% CI 1.6 to 33.4) for ipsilateral ischemic stroke.

Clinical applications.

The issue of hemodynamic risk is an important one for patients with carotid occlusion or intracranial atherosclerotic disease. Other conditions in which hemodynamic factors may play a role in the pathogenesis of ischemic stroke include vasospasm after subarachnoid hemorrhage and venous occlusive disease. This issue is moot for patients with extracranial carotid stenosis. Surgical endarterectomy has proved effective over medical management for these patients, regardless of mechanism.53,54 However, no therapy has proved effective for patients with carotid occlusion or intracranial atherosclerotic disease.

Approximately 15% of patients presenting with anterior circulation ischemic events are found to have complete occlusion of the ipsilateral carotid artery.5 These patients are at high risk for subsequent stroke (annual stroke rate of approximately 5 to 7%).55 The Extracranial to Intracranial (EC/IC) Bypass Trial was a large, international, randomized controlled trial that showed no benefit for EC/IC bypass over medical therapy in patients with symptomatic carotid occlusion.56 However, at the time of this trial, there was no proven method to identify patients at high risk due to hemodynamic factors. It is now established that many patients with complete carotid artery occlusion and normal cerebral hemodynamics have a low risk of subsequent stroke and therefore little to gain from EC/IC bypass. The failure of the EC/IC Bypass Trial to demonstrate efficacy may have been caused by the inclusion of many such patients.57

As many as 5 to 10% of patients presenting with strokes are found to have significant stenosis or occlusion of intracranial vessels.58 No benefit with revascularization was found for the 417 patients with intracranial stenosis or occlusion enrolled in the EC/IC Bypass Trial.56 A retrospective, unblinded, nonrandomized survey suggesting that anticoagulation may be superior to antiplatelet agents in these patients awaits confirmation from a prospective, randomized, blinded trial, now underway.59 Although hemodynamic impairment may be present in many patients with intracranial stenosis, the relationship between hemodynamic factors and stroke risk in this patient population remains to be defined. If hemodynamic factors are found to be important in the pathogenesis of ischemic stroke in patients with intracranial atherosclerotic disease, as in patients with extracranial carotid occlusion, then hemodynamic staging will be critically important for the design of therapeutic trials. Antithrombotic therapy may be more likely to be proved successful in patients with normal hemodynamics, whereas trials of revascularization via EC/IC bypass or angioplasty may be more likely to show a benefit if limited to patients with hemodynamic compromise.

At present, only one manifestation of hemodynamic compromise (increased OEF) has been established as an independent risk factor for subsequent stroke in patients with carotid occlusion.52 However, no causality has been established and no proven clinical utility for this information currently exists. Although the OEF abnormality is reversible with surgical revascularization,60 whether this procedure will reduce the risk of stroke remains to be determined. A randomized clinical trial of EC/IC bypass for patients with symptomatic carotid occlusion and increased OEF will be necessary.

The proper application of methods for assessing Stage 1 hemodynamic compromise at this time lies in further research. More data are needed for each modality to establish them as valid means for identifying patients at high risk for stroke due to hemodynamic factors. As with the measurement of OEF, there are no data to demonstrate that the use of Stage 1 hemodynamic measurements allows the selection of therapies that improve patient outcomes. If OEF measurements are proved to identify a group of patients who will benefit from EC/IC bypass, then large-scale comparative studies of methods of Stage 1 hemodynamic assessment should be pursued to determine their sensitivity and specificity for identifying patients with increased OEF. These Stage 1 methods may or may not prove to be appropriate substitutes for direct measurements of increased oxygen extraction.

Evidence of hemodynamic impairment by one method of hemodynamic assessment does not predict an abnormal result by another. Different techniques rely on different physiologic mechanisms from which the presence of reduced perfusion pressure is inferred. This may account for lack of correlation between different Stage 1 methods in the same patient. The degree of correlation between Stage 1 and Stage 2 techniques has also been variable. Different physiologic mechanisms behind the phenomena of autoregulatory vasodilation and increased oxygen extraction may be responsible.

Consequently, empirical proof linking each of these different measurements of hemodynamic compromise as risk factors to stroke risk is required. The proven association between the identification of increased OEF by PET and the risk of subsequent stroke in patients with symptomatic carotid occlusion does not indicate that hemodynamic impairment identified by other methods is also associated with stroke risk in this population. Furthermore, whether this association of increased OEF and stroke risk is true for asymptomatic patients or patients with stenosis rather than occlusion remains to be determined.

There is currently no sound evidence to support the use of these tools to guide the clinical management of patients with cerebrovascular disease. At present, the best application of methods for Stage 1 hemodynamic assessment lies in well-designed, prospective studies of hemodynamic impairment as independent risk factors for stroke in well-defined clinical populations. Conversely, Stage 2 hemodynamic compromise has been shown to be an independent predictor of stroke in patients with symptomatic carotid occlusion in a rigorous prospective study.52 EC/IC bypass can effectively treat this risk factor.9 A randomized trial of EC/IC bypass limited to patients with increased OEF will be required to prove a causal relationship between increased OEF and subsequent stroke. Until such a trial is completed, we do not believe that PET measurements of OEF or any other measurements of hemodynamic compromise should be used to select patients for surgical revascularization.

Note added in proof. Since the submission and acceptance of this manuscript, another study of hemodynamic compromise and stroke risk has been published.61 This was a well-designed and well-executed prospective study of 65 patients with both symptomatic and asymptomatic carotid occlusion. Hemodynamic compromise was assessed using a TCD measurement of blood velocity changes during breath holding. Multivariate analysis found only older age and impaired hemodynamics to be associated with subsequent TIA or stroke. A separate analysis limited to symptomatic patients was not reported, nor was the effect of different treatments the patients may have received.