INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Loss of sleep may decrease waking hypercapnic ventilatory response (1) but less well understood are the effects of sleep loss on the cerebrovascular response to hypercapnia. Hypercapnia is typically a powerful stimulus for vasodilation with subsequent increase in cerebral blood flow. However, nonrapid eye movement (NREM) sleep, although characterized as a mildly hypercapnic state, is accompanied by an approximate reduction of 25% in both cerebral blood flow and cerebral O₂ consumption, particularly in stages 3 and 4 (2). During REM sleep, regional changes in cerebral blood flow have been noted, with some areas showing an appreciable increase and others showing a decrease (3).

Responsiveness of the cerebral vasculature to hypercapnic challenge during sleep has not been studied directly. However, indirect evidence from studies of the cerebral blood flow in patients with sleep apnea suggests altered vascular responses during transient episodes of hypoxia and hypercapnia. Several groups have reported data on cerebral blood flow velocities, using transcranial Doppler (TCD) ultrasound during apneic events. Fischer and coworkers (4) reported that patients with sleep apnea have lower middle cerebral artery (MCA) mean flow velocities compared with controls during REM and NREM stages of sleep and after sleep. Netzer and colleagues (5) reported a significant reduction in MCA flow velocities with apneic episodes. They observed that reduction in MCA flow velocities was much more prominent in obstructive apneas compared with central apneas. Hajak and colleagues (6) reported an increase in MCA flow velocities with each apneic episode during all stages of sleep, with the maximum increase observed during REM sleep. Siebler and Nachtmann (7) reported a rapid increase in blood flow velocities associated with rapid decrease in velocities at the termination of apneic episodes. Balfors and Franklin (8) similarly reported an increase in flow velocity with each apneic episode, followed by a transient decrease in blood flow velocity. However, prolonged reductions of mean flow velocities were observed with repetitive episodes of apnea. Although the reasons for these conflicting results are unclear, they may be attributed to differences in the nature of apnea (central versus obstructive), temporal discrimination between velocities at onset and termination of apnea, and the frequency of preceding apneic episodes. Potentially relevant to these studies is an older study by Loeppky and coworkers (9), who found that sleep apnea patients had reduced daytime carotid artery blood flow in response to CO₂ rebreathing relative to controls, although their responses to hypoxia were similar.

More recently, a waking, morning reduction of vasodilatory capacity of the cerebral vasculature in response to hypercapnia has been reported in awake, normal subjects who did not undergo polysomnography (10); however, morning change in other vascular beds was not studied. Speculations on the underlying mechanism of the reported findings have focused on circadian rhythms in nitric oxide synthesis or metabolism (11); however, sleep-related factors have not been assessed. In the current study we attempted to replicate this morning reduction in hypercapnic reactivity and determine to what extent specific features of sleep might impact on this morning change in reactivity to CO₂. We included individuals both with and without disordered breathing in sleep to examine whether sleep apnea and any of its concomitants (stages, fragmentation, hypoxemia, hypercapnia) might be related to such a phenomenon.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Patients undergoing diagnostic nocturnal polysomnography at the Emory Sleep Disorders Center at Wesley Woods Hospital (Atlanta, GA) between March and August 1996 were invited to participate in the study. Patients with prior history of cerebrovascular disease were not included. The protocol was approved by the Emory University Human Investigations Committee and all subjects gave informed consent. We recorded age, sex, height and weight (for calculations of body mass index), presence of hypertension, and smoking history for each patient. Fifty patients were invited to participate in the study. In 20 patients either the TCD ultrasound or overnight capnography were considered inadequate for further analysis. There were no differences between the patients in the final sample and the excluded subjects in terms of age, sex, body mass index, total sleep time, respiratory disturbance index, or movements with arousal index. The final sample consisted of 30 patients (mean age, 50.4 yr; SD, 12.8 yr), of which 18 were men and 12 were women.

Polysomnography

Overnight in-laboratory polysomnography was performed on Grass (West Warwick, RI) model 78 polysomnographs using conventional techniques for recording electroencephalography (EEG), electrooculography (EOG), and surface mentalis electromyography (EMG) and standardized criteria for staging sleep (12). Additional channels recorded electrocardiography (modified lead II), bilateral surface anterior tibialis EMG, respiratory effort (Protech [Woodinville, WA] piezoelectric transducers) and air flow (nasal/oral thermistors). Oxygen saturation was recorded with pulse oximetry (Ohmeda Biox [Louisville, CO] model 3700) and end-tidal CO₂ was monitored with an infrared CO₂ analyzer (Novametrix Capnogard [Wallingford, CT] model 1265) via nasal cannula.

Transcranial Doppler Ultrasound Measurements

Transcranial Doppler (TCD) ultrasound provides a relatively new method for investigating the human cerebral circulation. TCD ultrasound records the blood flow velocity at certain points in the large arteries within the skull. In contrast, the other major methods for estimating cerebral blood flow are based on injection or inhalation of a radioactive tracer and visualizing the distribution of tracer either through single photon emission computed tomography (SPECT) or positron emission tomography (PET). Conventionally used tracers in SPECT scans are -emitting compounds that can penetrate the blood- brain barrier and include ¹³³Xe (gas), ^99mTc-HMPAO (d,l-hexamethylpropyleneamine oxime), and ^99mTc-ECD (ethyl cysteinate dimer). PET scans use H₂¹⁵O to measure cerebral blood flow. Both SPECT and PET scans estimate brain tissue perfusion, i.e., the blood flow at the level of cerebral microcirculation. Although brain tissue perfusion represents the physiologically important variable, both SPECT and PET scans have limited use in bedside evaluation owing to technical difficulties and the cost involved. TCD ultrasound records the velocity of flow and is used as an indirect measure of cerebral blood flow. Previous investigations have demonstrated a good correlation between concomitant measurements of cerebral blood flow velocity by TCD ultrasound and cerebral blood flow by SPECT (13). Insonation of the middle cerebral artery has been recommended as the artery of choice both for measurement of hemispheric cerebral blood flow and evaluation of CO₂ reactivity (13, 14). The CO₂ reactivity has been used as an indirect measure of the autoregulatory capacity of the brain. TCD ultrasound provides excellent temporal resolution and therefore is good for assessing the rapid responses of cerebral vessels to various stimuli (14).

Each patient underwent two TCD studies, one at night and the other in the morning, immediately before and after the polysomnographic recordings. All TCD measurements for all patients were made by A.I.Q. or W.C.W. Final morning awakenings ("lights on") were made by the overnight polysomnographic technologist only when one of these two investigators arrived in the laboratory, thus allowing morning measurements to be made as soon as possible after awakening, while the patient lay in bed. Both of the individuals making the measurements were experienced in TCD ultrasound and were blinded to nocturnal polysomnographic findings when making morning measurements. Patients were awakened per routine laboratory schedule; no attempt was made to control stage of sleep from which final awakening was made.

Cerebral blood flow velocity (CBFV) in the right middle cerebral artery was measured by TCD ultrasound by the transtemporal approach, using a 2-MHz probe at a depth of 50-55 mm at rest. Diastolic blood flow velocity measurements were used for all analyses. After measurements made with room air, the patient was given a mixture of 5% CO₂ to inhale via a nonrebreathing face mask to serve as a vasodilatory stimulus (15, 16). End-tidal CO₂ was recorded continuously (Novametrix Capnogard) during TCD measurements to document any increases in expired CO₂. After end-tidal CO₂ reached a steady state, TCD measurements were repeated. For each set of TCD readings (room air, 5% CO₂), three consecutive readings were recorded and averaged.

Data Analyses

The following variables were generated from the overnight polysomnogram: total sleep time (TST) (in minutes), sleep efficiency (SE) (%); proportion of TST (%) spent in stage 1, stage 2, stages 3/4, and REM; respiratory disturbance index (RDI) (apneas and hypopneas per hour of sleep); lowest Sa_O2 (%); desaturation  4% index (D4I) (number of desaturations greater than or equal to 4% per hour of sleep); desaturation < 90% index (D90I); nocturnal CO₂ retention (%) (percentage of all breaths with expired CO₂ greater than 45 mm Hg); and movements with arousal index (MAI) (number of periodic leg movements with EEG arousals per hour of sleep). Simple evening-to-morning differences in CBFV were tested with paired t tests. To examine motor vasomotor reactivity as a function of time, we computed a two-factor repeated-measures analysis of variance with main effects of time (evening versus morning) and condition (room air versus inhaled CO₂) as well as their interaction. Normality of all continuous distributions was checked and data were transformed with log or inverse transforms as necessary (17).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Mean polysomnographic data for the 30 patients are shown in Table 1. A total of 9 had RDIs greater than 15 events per hour and 10 had D4Is greater than 10 desaturations per hour. These two measures were highly intercorrelated (r = 0.90, p < 0.0001). Some evidence of overnight CO₂ retention (at least 10% of expired breaths with end-tidal CO₂ values in excess of 45 mm Hg) was present in 15 of these patients, and the percentage of expired breaths associated with high CO₂ values was uncorrelated with RDI, D90I, and D4I.

Presleep mean CBFVs were 44.5 (SD = 17.8) and 50.9 (SD = 19.7) cm/s on room air and CO₂, respectively. Morning mean CBFVs were 42.7 (SD = 16.9) and 47.3 (SD = 18.8) cm/s on room air and CO₂, respectively. ANOVAs indicated a highly significant effect of time (Wilks'  = 0.69, F = 13.12, df 1, 29; p < 0.001) and condition (Wilks'  = 0.41, F = 42.40, df 1, 29; p < 0.0001) with evidence of an interaction effect as well (Wilks'  = 0.84, F = 5.37, df 1, 29; p < 0.03). Pairwise comparisons indicated that CBFVs during CO₂ inhalation were significantly higher than on room air (t = 7.28, p < 0.0001 for evening; t = 4.69, p < 0.0001 for morning) and significantly lower in the morning than on the preceding evening on both room air and on CO₂ (t = 2.35, p < 0.03 and t = 3.93, p < 0.0005, respectively). Difference scores (morning minus evening) in CBFV on room air and CO₂ were unrelated to total sleep, sleep efficiency, or sleep stages. However, overnight CO₂ retention and greater sleep fragmentation (higher movement with arousal index, MAI) were related to a smaller morning decline in CBFV during CO₂ inhalation (reduced CO₂ vasoreactivity) (r = 0.44, p < 0.02; r = 0.49, p < 0.006, respectively). Subsequent stepwise multiple regression indicated that both variables predicted morning hypercapnic reactivity (F = 7.89, p < 0.003; cumulative r² = 0.369); however, MAI entered the model first and predicted approximately twice the variance of overnight CO₂ retention (partial r² = 0.243 versus 0.126). This model, although highly significant, still left approximately 63% of the variance in morning hypercapnic reactivity unaccounted for. The evening-to-morning difference in CBFV was significantly greater (i.e., a larger morning decline) during 5% CO₂ inhalation compared with the difference observed between CBFVs on room air (evening versus morning) (t = 2.32, p < 0.03).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We confirmed the previously noted, presumably normal finding that vasodilation of the middle cerebral artery during CO₂ inhalation is decreased in the morning compared with similarly measured response from the preceding evening (10). Our data also suggest that uninterrupted sleep (cf. sleep fragmentation) could play an important role in the extent of such morning vasomotor reactivity. Sleep fragmentation has been associated with considerable neurobehavioral (18) and, possibly, cardiovascular (19) morbidity. In our data, sleep fragmentation, and to a lesser extent, overnight measures of CO₂ retention predicted a reduced vasodilatory response to a mild hypercapnic challenge on awakening.

One limitation of this study is that TCD ultrasounds attempted in a number of patients were considered inadequate for analysis. They were mainly attributed to difficulty in insonating the middle cerebral artery, intolerance to the face mask, or incomplete recording of nocturnal capnography. This situation is not atypical for attempted sonography of the middle cerebral artery. About 10% of patients have an inadequate bone window, precluding any insonation of the MCA (20). The frequency of inconclusive recordings further vary depending on the study protocol and degree of specificity maintained. A high degree of specificity was maintained for the present study to ensure that true changes in vasomotor reactivity were being analyzed.

These results may have relevance for epidemiologic data suggesting that the period proximal to morning awakening is vulnerable for stroke (21). Explanations of this temporal pattern have varied and have included hematological factors such as increased platelet aggregability on assuming upright stance (22) and increased morning fibrinolytic activity (23) or sleep-specific factors such as REM sleep propensity and sleep apnea (24). Our results suggest that particular features of sleep, such as the number of arousals, may also be important variables to consider. To this extent, individuals who periodically arouse during sleep from any causes, such as periodic leg movements, might also be vulnerable to derangements of normal cerebrovascular function. It is of note that prolonged sleep fragmentation in an animal model has been shown to lead to hemodynamic changes similar to those seen in sleep apnea (25).

The role of sleep apnea in cardiovascular disease in general and cerebrovascular disease, specifically, remains highly controversial (e.g., see Reference 26). However, a diverse set of studies has linked sleep apnea and cerebral ischemic events. These data include retrospective, epidemiologic case control studies (27); prospective epidemiologic studies of both snoring (28) and sleep duration and napping (29); and polysomnographic studies of stroke patients subsequent to their stroke (30). Arousals during sleep have been implicated in such pathophysiological events observed in patients with sleep apnea. The current findings support previous literature suggesting that morbidities associated with sleep apnea are most likely to be associated with sympathetic activation accompanying arousals from sleep (31, 32).

Apart from issues related to sleep fragmentation, the current study also has demonstrated that decreased brain blood flow velocity during sleep, a finding demonstrated in both normal subjects and sleep apnea patients (6), persists into waking. This has been noted previously in brief episodes of wakefulness within the sleep period as well as on awakening in the morning (6). Such postsleep decreases in blood flow velocity have been confirmed in normal volunteers by Kuboyama and co-workers (33) using TCD ultrasound and with H₂¹⁵O positron emission tomography by Braun and colleagues (34). Taken together, these findings provide physiologic operationalization of a neurobehavioral phenomenon sometimes referred to as "sleep inertia" (35), or the carryover of sleep-related processes into waking. Such uncoupling of electrophysiologically defined state from metabolism (6) may be characteristic of the postsleep period. Whether this effect might represent a circadian rhythm independent of sleep is unclear. Similarly, whether this reduction might be associated with cerebrovascular morbidity apart from brief arousals is unknown, but conceivably it might represent yet another possible mechanism underlying the predisposition for stroke near the time of final morning awakening.