Cerebrovascular Reactivity and Subcortical Infarctions

Objectives  To investigate the association between different kinds of ischemic lesions and cerebrovascular reactivity (CR) and to evaluate their relationships with the major risk factors for stroke.

Subjects and Methods  We evaluated CR using the breath-holding index technique during bilateral transcranial Doppler monitoring of flow velocity in the middle cerebral arteries of 41 consecutive patients attending our clinic for a recent, first-ever, ischemic stroke and in 15 control subjects. Based on the location of the lesion determined by computed tomography, the following 3 types of infarctions were identified: cortical (or territorial), single subcortical, and subcortical with multiple silent subcortical infarctions. Patients with a condition of severe carotid artery stenosis or occlusion, which in itself could account for altered CR, were excluded from this study. All physiological and pathologic conditions that could possibly cause an impairment in CR were recorded.

Results  The breath-holding index was significantly lower in the multiple subcortical infarctions group than in the control subjects (P<.001), single subcortical infarctions group (P<.01), and cortical infarctions group (P<.01). In all of the groups male sex (P<.05) and a history of hypertension (P<.05), regardless of whether hypertension was treated, correlated with low CR. The multiple regression analysis indicated that the only significant factor able to influence the breath-holding index was the type of lesion.

Conclusions  Nonstenotic patients with first-ever stroke who had a recent symptomatic subcortical infarction associated with multiple silent infarctions seem to have an impaired cerebrovascular reserve capacity. The strong association of subcortical infarctions with multiple silent infarctions with low CR indicates the role of small vessel vasculopathy and hypoperfusion as possible pathogenetic mechanisms of subcortical infarctions with multiple silent infarctions.

THE CORRELATION between an impaired functional blood flow reserve capacity and the occurrence of brain infarction was previously reported in patients with severe carotid artery disease.^1-4 Patients with limited cerebrovascular reserve capacity have less adequate perfusion capacity than patients with normal reserves. An impaired cerebrovascular reserve capacity may increase the risk of cerebral ischemia in patients with major cerebral artery occlusion.^5,6 Patients with severe carotid artery stenosis or occlusion often have a borderzone distribution of brain infarction in the cerebral hemisphere ipsilateral to internal carotid artery disease.⁷ Border-zone distribution of infarction has traditionally been attributed to hypoperfusion related to reduced blood flow in zones between major hemispheric vascular territories. Moreover, cerebrovascular reactivity (CR) was found to be significantly reduced in low-flow infarctions compared with thromboembolic infarctions in patients with ipsilateral carotid stenosis.⁸ An association was found between CR and white matter lesions. This supports the hypothesis that these kinds of lesions may be associated with hemodynamic ischemic brain injury.⁹ In addition, patients with major cerebral arterial occlusive diseases and misery perfusion have a high risk of recurrent ischemic stroke.^6,10 These findings suggest that an impaired cerebrovascular reserve capacity is highly related to the occurrence of ischemic stroke. To our knowledge, the association between different types of ischemic lesions and CR has not been studied in patients who have had a stroke but who did not have severe carotid artery disease.

During the past decade, transcranial Doppler ultrasonography (TCD) has been widely used to assess blood flow velocities in the basal intracranial arteries and CR to various stimuli, including carbon dioxide (CO₂) reactivity. Many physiological and pathologic conditions such as age, sex, migraine, smoking, hypertension, and blood flow viscosity could account for changes in CR CO₂.^11-17 These conditions are considered the most common risk factors for stroke. Our study evaluated the association between different kinds of ischemic lesions and CR and their relationship to the above-mentioned risk factors for stroke. Thus, patients with a recognized potential source of CR failure, such as that observed with carotid artery stenosis or occlusion, were excluded from this study.

The study was prospective and consecutive and included all patients who have had an acute, first-ever stroke who were admitted to our neurology ward from January 1, 1998, to October 1, 1999. Patients were enrolled in this study if they fulfilled the following criteria: (1) their clinical symptoms correlated with a supratentorial ischemic lesion on computed tomography (CT), (2) Doppler ultrasonography excluded a hemodynamic stenotic disease of extracranial carotid and vertebral arteries, and (3) TCD revealed symmetrical middle cerebral artery (MCA) blood flow velocities and adequate temporal windows permitting acquisition of continuous bilateral blood flow velocities. Patients with a history of stroke were excluded from this study. Carotid artery evaluation was performed using a color-flow B-mode Doppler ultrasonography (model AU5; Harmonic Esaote Biomedica, Esaote S.p.A., Genoa, Italy), with a 7.5-MHz linear transducer. Plaque occurrence in the right and left carotid arteries and common carotid intima media thickness were evaluated. The type of ischemic lesion was determined by CT performed with a spiral CT scanner (Tomoscan SR 7000; Philips Medical Systems, Amsterdam, the Netherlands).

Of the 152 patients who had a first-ever ischemic stroke observed during the study period, 111 were excluded (21 owing to a subtentorial ischemic lesion, 56 owing to the presence of a hemodynamic stenotic disease of the extracranial carotid and vertebral arteries, and 34 owing to poor insonation of the temporal bone window or significant asymmetry of MCA blood flow velocities). Forty-one patients were included in this study. Thirty-two patients underwent brain magnetic resonance imaging (MRI) scanning (Gyroscan ACS-NT, 1.5 T; Philips Medical Systems). Twenty-one patients were also studied using MRI angiography. Basal TCD examination and MRI angiography did not reveal intracranial steno-occlusive lesions in the included patients. Based on the location of the lesion, 3 types of infarctions were classified. The first type, a cortical (or territorial) infarction, was defined as a case of first-attack infarction in which the CT scan showed a territorial infarction of a main intracerebral artery. The second type, a single subcortical infarction, was defined as a case of first-attack infarction in which the CT scan showed a single subcortical hemispheric infarction compatible with symptoms. According to the classification of Nakano et al,¹⁸ the infarctions were restricted to the basal ganglia and/or white matter on CT, and the overlying cerebral cortex appeared normal. The maximum diameter of the lesion exceeded 2.0 cm. The third type, a subcortical infarction with multiple silent subcortical infarctions, was defined as a case of first-ever stroke in which the CT scan showed multiple subcortical infarctions. Computed tomographic scans were examined by an expert reader (M.S.) blinded to the results of the TCD recordings. Evaluation of CR was performed by 2 operators (M.D. and F.P.) blinded to the CT findings.

The study was carried out in a quiet room with the patients lying in a comfortable supine position. Bilateral simultaneous flow velocity recording of MCAs was obtained using a transcranial Doppler instrument (Multi-DopX/TCD; DWL Elektronische Systeme GmBH, Sipplingen, Germany). Two dual 2-MHz transducers fitted on a headband and placed on the temporal bone window were used to obtain a bilateral continuous measurement of mean flow velocity (MFV) in the MCAs. Examination of vessels of the circle of Willis was performed as described by Aaslid et al.¹⁹ We obtained hypercapnia with breath holding²⁰ and evaluated CR using the breath-holding index (BHI) technique in the 41 patients and 15 healthy volunteers recruited from hospital personnel. The BHI is obtained by dividing the percent increase in MFV occurring during breath holding by the length of time (in seconds) the subjects hold their breath after a normal inspiration [({MFV at the end of breath holding − rest MFV}/rest MFV) × (100/s of breath holding)]. End-tidal expiratory CO₂ level was recorded using a capnometer (Normocap-oxy; Datex-Ohmeda S.p.A., Segrate, Italy). Mean blood pressure and heart rate were continuously monitored by means of a blood pressure monitor (2300 Finapress Ohmeda Medical, Laurel, Md). All subjects were normocapnic. The MFV at rest was obtained by the continuous recording of a 1-minute period of normal room air breathing. After a breath-holding period, the MFV, mean blood pressure, and heart rate were recorded over 4 seconds. Subjects were asked to hold their breath for 30 seconds. The end-tidal expiratory CO₂ level during the first exhalation after apnea was evaluated. The BHI was calculated when the rise in the level of end-tidal expiratory CO₂ from baseline to the first expiration after breath holding was more than 8 mm Hg. The efficacy of breath holding was checked with the respiratory activity monitor. All TCD data were stored on hard disk for off-line analysis.

Patients were examined twice, in the acute phase and in a follow-up visit 1 to 3 months after the acute onset of stroke. Data concerning this study refer to recordings performed during the follow-up visit since previous evidence indicates that cerebral hemodynamics can be impaired during the acute phase of stroke.

All the physiological and pathologic conditions that could account for the patients' altered CR were recorded. The following known or putative factors associated with risk were considered: age (≤50, >50, ≤65, or >65 years); sex; heavy alcohol consumption (≥300 g/wk), current daily smoking (≥10 cigarettes per day); hypertension (in treatment with antihypertensive drugs at the time of admission or hypertension diagnosed during the hospital stay); an elevated serum cholesterol level (total serum cholesterol level of ≥6.20 mmol/L [≥240 mg/dL] at the time of admission); an elevated hematocrit on admission; and the presence of diabetes mellitus, coagulopathies, and migraine.

As regards statistical analysis, at the first step several analyses of variance, with patients' characteristics (ie, age groups, sex, and others) as between-subjects factors and BHI as dependent variable, were used to assess the relationship between CR and the risk factors. In addition, analysis of variance with group (4 levels: the 3 groups of patients and the controls) as between-subjects factor and BHI as dependent variable was used to assess possible differences in CR among the different groups of patients and controls.

At the second step, to individuate which risk factors were more relevant on the BHIs, a multiple regression analysis was performed, entering the CR as a dependent variable and the type of lesion; sex; age; the presence of hypertension, diabetes mellitus, hypercholesterolemia, coagulopathies, and migraine; the use of antihypertensive treatment; the patient's tobacco use; an elevated hematocrit; and excessive alcohol consumption as independent variables. Each categorical variable was entered as a dummy variable, while age was a continuous variable. The forward stepwise method was chosen to individuate recursively the statistically significant factors. The statistical significance threshold was set at P<.05. All analyses were performed with StatSoft 5.0 for Windows statistical software (StatSoft Inc, Tulsa, Okla). The study was approved by the local ethics committee and all subjects gave their informed consent.

All patients included in the study performed the required task adequately. The period of apnea ranged from 29.1 to 30.6 seconds. Heart rate and mean blood pressure showed a slight increase after the end of the apnea period with respect to the baseline condition: 2% to 3% for heart rate and 3% to 4% for mean blood pressure. Forty-one patients (30 men and 11 women) were studied. Thirteen patients (group 1) had cortical (or territorial) infarctions (mean [±SD] age, 53.9 ± 11.8 years; age range, 34-83 years ), 14 patients (group 2) had single subcortical infarctions (mean age, 61.4 ± 9.2 years; age range, 41-76 years), and 14 patients (group 3) had subcortical infarction with multiple silent subcortical infarctions (mean age, 60.5 ± 10.5 years; age range, 44-76 years). All patients with subcortical infarction had an additional MRI scan that confirmed the solitary subcortical lesion revealed by CT.

Patients' characteristics and vascular risk factors are reported in Table 1. Mean (± SD) age and sex distribution for the controls were 57.66 ± 12.7 years (age range, 37-73 years) for 9 men and 6 women, respectively. No significant difference was noted in age and sex distribution among the groups. Regarding pharmacological treatment of vascular risk factors, no significant difference in the use of insulin, oral antidiabetes drugs, statins, and different classes of antihypertensive drugs was found among the patient groups.

Since the side of the stroke was not statistically found to influence CR in the 2 MCAs (ie, group 1, BHI [mean ± SD] of the symptomatic side: 1.24 ± 0.51, BHI of the asymptomatic side: 1.45 ± 0.51, P = .09; group 2, BHI of the symptomatic side: 1.33 ± 0.36, BHI of the asymptomatic side: 1.36 ± 0.39, P = 0.6; group 3, BHI of the symptomatic side: 0.97 ± 0.42, BHI of the asymptomatic side: 0.90 ± 0.36, P = .32), the mean of the right and left CR was used for further statistical analysis.

Smoking, diabetes mellitus, elevated serum cholesterol levels, hematocrit, coagulopathies, and use of alcohol were not found to affect CR significantly. Male sex (F_1,39 = 4.93, P = .03) and the presence of hypertension (whether treated or not) (F_1,39 = 4.1, P = .049] were found, regardless of group, to be significantly related to a low CR. However, a history of migraine was significantly related to a high CR (F_1,39 = 8.21, P = .007).

Regarding the comparison of the BHIs among the 3 groups of patients and controls, the group effect was significant (F_3,52 = 6.6, P<.001). In particular, the Tukey post hoc analysis showed that BHIs of the subcortical infarction with multiple silent infarctions group (Figure 1) was significantly lower than that of the controls (P<.001) and of both singular subcortical (P<.01) and cortical (or territorial) infarction groups (P<.01). No statistical difference of mean BHIs was observed for the controls and the territorial and singular subcortical infarction groups.

The multiple regression analysis indicated that the only significant factor able to influence the BHI was the type of lesion (r = 0.46, F_1,39 = 10.2, P = .003). No other variable could be entered for a better accounting of BHI variability. This result does not indicate that the type of lesion is an independent predictor of CR, since higher percentages of multiple infarction lesions were observed in patients with hypertension (P = .04, χ² test) and in patients who smoked (P = .04, χ² test). However, the result of multiple regression analysis suggests that, at the evaluation time, the strongest and unique factor able to explain BHI was the type of lesion. In particular, the patients in the subcortical infarction with multiple silent subcortical infarctions group were characterized by a significantly lower BHI than the other 2 groups (mean difference = 0.41), as confirmed by the above analysis of variance.

The primary purpose of this study was to assess the relationship between hemodynamic reserve capacity and distributions of ischemic lesions in patients who have had a stroke but who did not have carotid stenosis. Available data strongly link hypoperfusion with the occurrence of brain ischemia and infarction.²¹ Hypoperfusion is the proximate cause of all ischemic stroke; however, the extent of its role as the primary causative factor in stroke remains unclear.^5,22 The correlation between an impaired cerebrovascular reserve capacity and the occurrence of stroke in patients with severe internal carotid artery occlusive disease has been widely recognized.^1-4,6 However, the pathogenetic role of an impaired hemodynamic reserve capacity in stroke patients who did not have carotid stenosis has not been clarified.

The main finding of our study on patients without stenosis first-ever stroke is that subjects with lower CR were found to have subcortical infarction with multiple silent subcortical infarctions. We did not observe statistical differences in CR among the controls, the single subcortical infarction group, or the cortical (or territorial) infarction group. Previously, a reduction in CR was reported in low-flow infarction compared with that found in patients with cortical (or territorial) infarction.⁸ However, all of the patients included in that study had internal carotid artery occlusion and, as the authors^23-26 suggested, a restricted collateral blood supply contributed to their finding of low CR in low-flow infarction. Previous studies^23-26 also suggested that white matter infarctions in terminal distribution vessels may be a more common consequence of hypoperfusion. An association was previously shown between decreased CR and periventricular lesions using MRI in asymptomatic individuals²⁷ and hypertensive patients with leukoaraiosis.²⁸ An association was also reported between decreased CR and the size, location, and number of white matter lesions in elderly persons.⁹ However, all of these studies were conducted on both patients with and without stenosis, thus the effect of carotid stenosis on CR cannot be excluded.

It is generally accepted that among the pathogenetic causes of subcortical hemispheric infarctions are small vessel disease, thromboembolic occlusions of small arteries, and hemodynamic impairment in low-flow conditions.^8,9,18,23 Our finding that an impaired hemodynamic reserve capacity in patients without stenosis is associated with multiple subcortical ischemic lesions supports the hypothesis that some subcortical ischemic lesions may be associated with hemodynamic ischemic injury to the brain. Hypoperfusion could account for the finding of silent infarction in patients with first-ever, subcortical, symptomatic stroke. In our study, among the risk factors that could account for impaired CR, hypertension, and male sex were found to be significantly and independently from groups associated with low CR. The effect of hypertension and its treatment on CR were previously outlined.¹⁵ Hypertension is considered the most important single risk factor for ischemic stroke^29,30 and is considered one of the main risk factors for stroke recurrence.³¹ However, not all studies have shown that when blood pressure is controlled,³² the risk of stroke recurrence is reduced. We observed that despite whether they were being treated, the patients with hypertension had the lowest CR. This finding raises the critical issue of the efficacy of antihypertensive treatment for patients who have had a stroke.^29,33 Among risk factors for stroke, age, hypertension, and diabetes mellitus were not found to be significantly different in cortical and subcortical stroke.³⁴ However, hypertension was shown to be strongly and independently correlated with silent cerebral infarctions.^35-38 Silent cerebral infarctions are frequently shown by CT and MRI in the subcortical white matter or the basal ganglia in patients who have had a stroke and in elderly subjects.^35-37 Recently, it was suggested that silent cerebral infarctions appear first in the white matter in association with aging and hypertension and that the appearence of silent cerebral infarctions in the basal ganglia predicts a progression of generalized atherosclerosis.³⁹ Genetic risk factors for silent brain infarction have also been suggested.⁴⁰ The effect of hypertension on small vessels is well known. A vascular "remodeling" occurs in cerebral blood vessels during chronic hypertension.⁴¹ It has been suggested that this structural alteration impairs autoregulation, exposing the deep white matter to fluctuations in blood pressure.

We observed a significant increased CR in patients with a history of migraine. The study of interictal CR in migraineurs has provided contradictory results.¹³ Since in our study only 3 patients were migrainous, we cannot draw definitive conclusions concerning this issue.

Our study outlines the strong link between impaired CR and the finding of silent subcortical infarctions among patients who did not have carotid stenosis with a recent first-ever, symptomatic, subcortical stroke. The clinical relevance of hemodynamic factors in the pathogenesis of subcortical infarctions requires further investigation. Future studies on a larger number of patients are also needed to establish whether TCD evaluation of CR could be a convenient bedside test to differentiate subcortical stroke due to small vessel vasculopathy from that of other causes, especially emboli.

Accepted for publication July 13, 2000.

We thank P. Pasqualetti, PhD, for critically reading the manuscript and for his expert assistance with the statistical analysis.

Corresponding author: Letizia M. Cupini, MD, Clinica Neurologica, Universita' di Roma "Tor Vergata," Ospedale S Eugenio, P. le Umanesimo 10, 00144 Rome, Italy (e-mail lecupini@tin.it).