Obstructive sleep apnea (OSA) is a chronic medical condition that is associated with intermittent hypoxemia during sleep (1). It occurs in at least 2 to 4% of the general population (2) and has been implicated as an independent risk factor for the development of stroke (3, 4). Although the pathophysiologic mechanism for the association between OSA and stroke is not known, it may occur through a reduction in cerebral vascular reactivity. The cerebral blood flow responses to acetazolamide and to CO2 are significant predictors of stroke in patients with carotid artery disease (5-8) and in those who have suffered cerebral infarction (9, 10). Although the predictive value of the cerebral blood flow response to hypoxia has not been studied in this context, it is an important physiologic mechanism that, if reduced, could predispose to cerebral infarction.
In healthy humans, the cerebral vasculature responds to hypoxia by vasodilation to increase blood flow and oxygen delivery to cerebral tissue. This response is endothelium dependent, involving the production of nitric oxide (NO) to relax smooth muscle in the vessel wall (11). Patients with OSA suffer oxidative stress, which may contribute to the development of endothelial dysfunction (12-15). Although the relationship between sleep apnea and endothelial dysfunction is somewhat controversial, previous studies have reported that oxidative stress reduces NO bioavailability (16, 17) and vasodilation in the peripheral vasculature (18-20). Furthermore, treatment of OSA by continuous positive airway pressure (CPAP) corrects oxidative stress (15), which improves NO bioavailability (16, 17) and peripheral vascular function (20). This phenomenon has not been studied in the cerebral circulation, where similar endothelial dysfunction is likely to occur in patients with sleep apnea.
We measured the cerebral blood flow response to hypoxia during wakefulness in patients with OSA before and after CPAP treatment and compared this with the cerebral blood flow response to hypoxia in healthy control subjects matched for gender and weight. We hypothesized that the cerebral blood flow response to hypoxia is reduced in patients with OSA and that this response is normalized by CPAP therapy. Some of the results from this study have been reported in abstract form (21).
METHODS
Study Subjects
Patients referred to the Sleep Center for evaluation of OSA had overnight cardiopulmonary monitoring at home. Patients who had evidence of significant OSA (i.e., respiratory disturbance index > 30 events/h) and met the remaining criteria for inclusion in the study were invited to participate. Patients were excluded from the study if they were obese (body mass index [BMI] > 35 kg/m^sup 2^); smoked; were taking medication; or had a history of cardiorespiratory disease, hypertension, or treatment with CPAP. These exclusion criteria were used to avoid all known factors that could alter the response of the cerebral vasculature to hypoxia. Healthy volunteers were recruited as control subjects. We recruited nonsmokers with a BMI less than 35 kg/m^sup 2^ who had no history of snoring or chronic medical disorders. All control subjects had overnight cardiopulmonary monitoring at home to rule out OSA.
The study protocol was reviewed and approved by the Conjoint Health Research Ethics Board at the University of Calgary, and all subjects gave written, informed consent to participate in the study.
Study Protocol
Measurement of the cerebral blood flow response to hypoxia was conducted in the Laboratory of Human Cerebrovascular Physiology at the University of Calgary. Subjects came to the laboratory on three occasions and were instructed not to eat or drink caffeinated beverages for 4 hours before their assessment. During the initial visit, resting end-tidal gases and cerebral blood flow velocity were determined, and subjects were familiarized with the experimental apparatus. The next day, subjects returned to the laboratory for baseline measurement of the cerebral blood flow response to hypoxia. The following night, patients with OSA had overnight attended polysomnography in the sleep laboratory at Foothills Medical Centre. During the first half of the sleep study, the severity of OSA was established. During the second half of the sleep study, patients were placed on CPAP, which was titrated to a level that controlled OSA. The following morning, patients were provided with a CPAP unit, which was set at the optimal pressure identified during polysomnography. Regular follow-up with a CPAP therapist was arranged to facilitate acclimatization to CPAP therapy. After 4 to 6 weeks of effective CPAP therapy, which was confirmed by objective monitoring of CPAP use and overnight cardiopulmonary monitoring while using CPAP at home, patients returned for follow-up measurement of the cerebral blood flow response to hypoxia. Control subjects also returned to the laboratory 4 to 6 weeks after their initial assessment for similar follow-up. Control subjects did not receive CPAP therapy.
Study Measurements
Home cardiopulmonary monitoring. Patients and control subjects were screened for OSA by continuous, overnight cardiopulmonary monitoring at home (Remmers Sleep Recorder Model 4.2; Saga Tech Electronic, Calgary, AB, Canada) (22). All participants were instructed how to set up the monitoring device at home to perform an unattended study. This device consists of an oximeter to record oxyhemoglobin saturation, a pressure transducer to record nasal airflow, a microphone to record snoring, an electrocardiogram to record heart rate, and a body position sensor. The oximeter has a high sampling frequency (1 Hz) and provides the data for an automated scoring algorithm, which calculates the respiratory disturbance index based on the number of episodes of oxyhemoglobin desaturation greater than 4% divided by the duration of the recording. The raw data were reviewed in each patient and control subject by an experienced sleep medicine physician (P.J.H.). The Remmers recorder, which was previously called "Snoresat," has been validated by comparison to attended polysomnography (22, 23).
Polysomnography. The diagnosis and severity of OSA was confirmed by overnight attended polysomnography. All polysomnograms were scored manually by registered polysomnographic technologists according to established criteria (24). Further details are outlined in the online supplement.
CPAP compliance. Once patients were fully acclimatized to CPAP therapy, their nightly use at the prescribed pressure was downloaded from the CPAP unit for 4 weeks before follow-up testing. CPAP compliance was considered acceptable if the device was used at least 4 hours per night (25).
Cerebral blood flow response to isocapnic hypoxia. The subject's resting end-tidal partial pressure of oxygen (PET^sub O2^) and CO2 (PET^sub CO2^) were measured over 10 minutes at the beginning of the experiment using dedicated software (Chamber v2.26; University of Oxford Laboratory of Physiology, Oxford, UK). The respired gas was sampled continuously (20 ml/min) through a fine catheter at the mouth and analyzed for PO^sub 2^ and PCO^sub 2^ by a mass spectrometer (AMIS 2000; Innovision, Odense, Denmark). The subject breathed normally through a mouthpiece with the nose occluded by a nose clip. Respiratory volumes were measured with a turbine and volume transducer, (VMM-400; Interface Associates, Laguna Niguel, CA), and respiratory flow direction and timing were obtained with a pneumotachograph and differential pressure transducer (RSS100-HR; Hans Rudolf, Inc., Kansas City, MO). Accurate control of the end-tidal gases was achieved by using specifically designed software (BreatheM v2.38; University of Oxford Laboratory of Physiology). This technique is known as dynamic end-tidal forcing (26). After a 10-minute lead-in period of isocapnic euoxia (PET^sub CO2^ = +1-0 mm Hg above resting values; PET^sub O2^ = 88.0 mm Hg), the inspired gas was rapidly changed (within two to three breaths) to a PET^sub O2^ of 50.0 mm Hg while PET^sub CO2^ was maintained at +1.0 mm Hg above resting values. Isocapnic hypoxia continued for 20 minutes, at which time the PET^sub O2^ was rapidly returned to 88.0 mm Hg and maintained constant for the final 10 minutes.
During the study, heart rate was recorded by three-lead electrocardiogram (Micromon 7142B monitor; Kontron Medical, Milton Keynes, UK), and mean peak blood flow velocity (VP) was measured in the middle cerebral artery by transcranial Doppler ultrasonography (TC22; SciMed, Bristol, UK) as previously described (26-28). Blood pressure was recorded continuously on the left hand by finger pulse photoplethysmography (Portapress; TPD Biomedical Instrumentation, Amsterdam, The Netherlands) and every 5 minutes from the right arm (Dinamap; Johnson and Johnson Medical, Inc., New Brunswick, NJ). Sa^sub O2^ was measured continuously by pulse oximetry (Model 3900; Datex-Ohmeda, Louisville, CO).
PO^sub 2^ and PCO^sub 2^ were sampled from the mouth by a computer program every 10 milliseconds, and PET^sub CO2^ and PET^sub O2^ were measured during each breath using dedicated software (BreatheM v2.38; University of Oxford Laboratory of Physiology). Cardiovascular data were acquired every 10 milliseconds and averaged over each heart beat by custom-designed data acquisition software (BreatheM v2.38; University of Oxford Laboratory of Physiology). VP was averaged over a 5-minute period during isocapnic euoxia and during the final 5 minutes of isocapnic hypoxia. The VP sensitivity to hypoxia was calculated as the percent change in VP from isocapnic euoxia to isocapnic hypoxia divided by the change in arterial oxyhemoglobin saturation. For the measurement of VP, sensitivity the arterial oxyhemoglobin saturation was calculated from PET^sub O2^ based on the Severinghaus transform as previously described (29). When the relationship between VP and the degree of hypoxia is expressed as a function of arterial oxyhemoglobin saturation, the relationship is linear, as demonstrated previously in our laboratory (30, 31).
Statistical Analysis
Data were compared using repeated-measures analysis of variance (SPSS, version 14.0; SPSS, Inc., Chicago, IL). When significant F ratios were detected, the simple effects test was applied post hoc to resolve differences. We confirmed that the data were normally distributed by plotting cumulative frequency against observed frequency and by performing tests for skewness and kurtosis and the Kolmogorov-Smirnov test. Pearson product-moment correlations were implemented to determine linear relationships between selected dependent variables. The level of significance was set at p < 0.05 for all statistical comparisons. All data are presented as mean ± SE.
RESULTS
We studied 18 men (8 with OSA and 10 healthy subjects). The two groups were of similar age (patients, 41 ± 2 yr; control subjects; 37 ± 3 yr) and BMI (patients, 30 ± 1 kg/m^sup 2^; control subjects, 28 ± 1 kg/m^sup 2^). By study design, patients had severe OSA associated with significant nocturnal hypoxemia, which was confirmed by home monitoring and polysomnography, whereas control subjects had no evidence of sleep apnea or nocturnal hypoxemia (Table 1). All patients with OSA were successfully treated with CPAP, which was reflected by reduction in the respiratory disturbance index to normal and correction of nocturnal hypoxemia. All subjects were compliant with CPAP therapy, using it 5.1 ± 0.4 hours per night during the 4 weeks before their final assessment.
There were no differences in isocapnic euoxic PET^sub O2^, PET^sub CO2^, mean arterial pressure, or VP between patients with OSA and control subjects at baseline or follow-up 4 to 6 weeks later (Table 2). However, before CPAP therapy, the cerebral blood flow sensitivity to hypoxia was significantly less in patients with OSA than in control subjects (p = 0.007) and returned to normal after treatment with CPAP (p = 0.40) (Figure 1). Six patients with OSA demonstrated an increase in the cerebral blood flow sensitivity to hypoxia, one showed no change, and one showed a decrease after CPAP therapy. The patient in whom the cerebral blood flow sensitivity decreased after CPAP therapy was incompletely treated with CPAP. He had persistent apnea (respiratory disturbance index = 12 events/h) during cardiopulmonary home monitoring despite using CPAP of 20 cm H2O. The cerebral blood flow response to hypoxia in one patient with OSA before and after 4 weeks of CPAP therapy is displayed in Figure 2. Before treatment with CPAP (Figure 2A), the response was blunted, and oscillations were excessive compared with a healthy subject. After CPAP therapy (Figure 2B), the response was of greater magnitude and was more sustained and similar to that of a healthy subject.
Before CPAP therapy, the cerebral blood flow sensitivity to hypoxia was significantly correlated with the apnea-hypopnea index (p = 0.04) and with mean nocturnal oxyhemoglobin saturation (p = 0.01) (Figure 3). The change in cerebral blood flow sensitivity to hypoxia after CPAP therapy correlated significantly with the baseline apnea-hypopnea index (r = 0.63; p = 0.005). These correlations imply that the severity of OSA and associated hypoxemia modulate the changes in cerebral blood flow sensitivity to hypoxia.
DISCUSSION
We found that the cerebral blood flow response to hypoxia is reduced in patients with OSA and that this is corrected by treatment with CPAP. Moderately strong relationships were found between the cerebral blood flow response to hypoxia and the apnea-hypopnea index and nocturnal oxyhemoglobin saturation, suggesting that patients with the most severe OSA and associated hypoxemia have the lowest cerebral blood flow response to hypoxia. The change in the cerebral blood flow response to hypoxia after CPAP therapy was greatest in those with the highest apnea-hypopnea index, suggesting that patients with severe OSA may have the most to benefit from treatment with CPAP.
This is the first report to demonstrate an altered cerebral blood flow response to hypoxia in patients with OSA. Specifically, the cerebral blood flow response to hypoxia was reduced by 42% compared with healthy subjects. Previous studies have demonstrated impaired autoregulation of the peripheral vasculature in patients with OSA, which has been attributed to endothelial dysfunction and diminished bioavailability of NO (18-20). The response of forearm blood flow to infusion of acetylcholine, which stimulates release of NO, was reduced by 39% (19), whereas the response to endothelium-independent stimuli, such as NO donors, L-arginine, and calcium-channel blockers, remained intact (18, 32). The impaired vascular response was corrected by treatment of sleep apnea with CPAP (20). This phenomenon has been attributed to intermittent hypoxia, which is supported by data from animal studies. Rats exposed to chronic intermittent hypoxia show a severely attenuated vasodilatory response of the middle cerebral artery to acetylcholine and hypoxia (33).
OSA is characteristically associated with repetitive oscillation in oxyhemoglobin saturation during sleep, which results in chronic exposure to intermittent hypoxia. Several pathophysiologic mechanisms may be considered that could link intermittent hypoxia to altered vascular function, including sympathetic nervous system activation (34-36), oxidative stress (37), and endothelial dysfunction (37). The impact of sympathetic nervous system activation on cerebrovascular function is poorly understood. Although the cerebral circulation has a rich sympathetic innervation, electrical stimulation of these nerves has little or no effect on cerebral blood flow in normotensive subjects (38, 39). Intermittent hypoxia also leads to oxidative stress through the formation of reactive oxygen species, which react with NO, an important vasodilator in the cerebral circulation (37, 40), thereby diminishing the bioavailability of NO and its vasodilator function. Plasma levels of NO derivatives are decreased in patients with OSA and increase after CPAP therapy, which supports this hypothesis (41,42). In addition, basal NO production is increased in patients with OSA after CPAP therapy, as indicated by a significantly greater reduction in forearm blood flow by the NO synthase inhibitor N^sup G^-monomethyl-L-arginine (32). We suspect that reduced NO bioavailability plays an important role in reducing hypoxic cerebral vasodilatation in patients with OSA, especially because hypoxia-induced cerebral vasodilatation is partly mediated by NO (11).
The cerebral blood flow response to hypoxia could have been independently altered by confounding factors such as age, atherosclerosis, hypertension, obesity, and smoking (43). To address this concern, we recruited younger patients who did not have overt cardiorespiratory disease, including hypertension, and who did not smoke. We excluded patients who were morbidly obese and matched our control group for age and weight. Although these exclusion criteria diminished the potential impact of confounding variables on our results, they limit the extension of our findings to the general stroke population, which includes patients with hypertension and those who develop stroke secondary to a cerebral embolus or hemorrhage (43). In contrast, patients with OSA typically develop stroke due to cerebral ischemia (44, 45) in which altered vascular reactivity is more likely to play a role.
We used Doppler ultrasound to measure blood flow velocity in the middle cerebral artery. Although this is not a direct measurement of cerebral blood flow, we believe that it is a reasonable estimate because the diameter of the middle cerebral artery varies by less than 4% during changes in arterial pressure and CO2 tension (46, 47) and because velocity and flow through the middle cerebral artery are highly correlated (27, 48).
We believe that our findings are clinically relevant to intracranial disease and lacunar stroke. There is increasing evidence that OSA increases the risk of stroke (3, 4), and this association is likely to increase as the population ages and the prevalence of obesity-related sleep apnea grows (49). Although many theories have been proposed, the basic mechanism responsible for the association between OSA and stroke is not known (50). The etiology is likely to be multifactorial and may include factors such as hypertension, carotid disease, and cardiac arrhythmias (50) in addition to altered vascular reactivity. Our findings provide the basis for future studies that will advance our understanding of the pathogenesis of stroke in patients with sleep apnea and thereby improve the management of this important vascular complication of a highly prevalent disease (1, 2).
Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript.
