INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The clinical syndrome of obstructive sleep apnea (OSA) is characterized by repeated episodes of upper airway occlusion during sleep, resulting in recurrent asphyxia and disruption of sleep. Epidemiological studies have identified OSA as a risk factor for hypertension, myocardial infarction, stroke, and sudden death, but the physiological mechanisms underlying these associations have not been defined (1). The strongest association demonstrated to date in clinical and epidemiological studies is between OSA and hypertension (2). Recently, we demonstrated experimentally a direct etiologic link between OSA and sustained daytime hypertension in an animal model (4).

In the human syndrome of OSA, it has been postulated that the paroxysmal sympathetic discharges and surges of blood pressure (BP) associated with obstructive apneas during sleep may be the mechanism leading to daytime hypertension (5). In addition, there is evidence that patients with OSA have a higher level of sympathetic activity during wakefulness and sleep than normal subjects, as indicated by microneurography, and that treatment of OSA with the application of continuous positive airway pressure (CPAP) decreases sympathetic activity (6, 7).

The changes in sympathetic activity in patients with OSA could be related to changes in baroreflex set point for the maintenance of BP. The arterial baroreceptors play a major role in the reflex regulation of BP and normally function over a relatively brief time frame to maintain BP within certain limits (8). For example, an increase in BP results in an increase in baroreceptor firing, resulting in a decrease in sympathetic activity and an increase in vagal firing, causing a decrease in peripheral vascular resistance, stroke volume, and heart rate (HR) (9). Thus, a rise in BP initiates compensatory responses that restore pressure to lower levels (9). However, recurrent exposure to elevated BP, as encountered in OSA, may lead to resetting of the baroreceptors to a higher pressure. Resetting can be defined as "decreased pressure-sensitivity" of the baroreceptors, i.e., decreased baroreceptor discharge for a given level of pressure (10). Conceivably, resetting of the baroreceptors could contribute to the increased sympathetic nerve activity noted in patients with OSA.

The literature on the effect of OSA on baroreceptor function is limited and conflicting. Some investigators have reported no difference in baroreflex slope between apneic and nonapneic patients (11), whereas others have reported depressed sensitivity (12). However, the method of baroreflex testing was substantially different in these two studies. In addition, comparison of baroreflex function between control subjects and patients with OSA is difficult owing to confounding variables that independently affect baroreceptor function, including age and BP. Nevertheless, in the study by Carlson and colleagues (12), differences in baroreflex sensitivity persisted despite statistical correction for age and BP.

We have recently induced sustained increases in BP in a canine model of OSA (4). In the present study, we used this model to determine whether OSA leads to changes in the arterial baroreceptor control of HR. Because each animal served as its control, we were able to examine the specific effects of OSA on the arterial baroreflex control of HR in the absence of other confounding variables.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Induction of OSA

All surgical and experimental procedures were approved by the Animal Care Committee of the University of Toronto. Studies were performed in the three adult dogs (2 female, 1 male; weight, 23 to 31 kg) that participated in the study on the effect of OSA on daytime BP (4). As described previously, each dog had a permanent side-hole tracheostomy and an implanted three-channel telemetry unit (TLM11M3D70-CCP; Data Sciences, St. Paul, MN) for monitoring of arterial BP, electroencephalogram (EEG), and nuchal electromyogram (EMG) (4, 13, 14). The radiofrequency signals of EEG, EMG, and BP emitted by the telemetry unit were detected by three water-resistant receivers (RL2000; Data Sciences) that were positioned around the pen in which the dog was housed. A multiplexer (RMX10; Data Sciences) was used to process and select the strongest signal from the receiver in closest proximity to the dog. The EEG and EMG signals were then converted from digital to analog (DL10; Data Sciences) where they were amplified and filtered (1 to 50 Hz for EEG, 10 to 100 Hz for EMG). The signals were then sampled at 300 Hz by a second digital-to-analog board attached to a personal computer (Lab Master DMA) that produced a judgment of sleep-wake state for every 6-s epoch using the custom-designed program previously validated and described (13, 14). Once a period of sleep of predetermined length (at least 18 s) was identified, a radio signal was produced by the computer and detected by a receiver-controller unit housed in a jacket worn by the dog. This signal resulted in closure of a quiet custom- designed valve attached to an endotracheal tube (internal diameter, 9 mm; Aire-Cuff; Bivona, Gary, IN) through which the dog breathed, thereby producing an obstructive apnea. When the dog awoke from sleep, the computer generated a signal to release the occlusion. Thus, the model simulated closely the clinical syndrome of OSA by producing repeated episodes of airway occlusion and arousal from sleep (4, 15).

OSA was produced in each dog over a period of 1 to 3 mo. The severity of the disorder, defined by the number of apneas per hour of sleep (i.e., apnea index), was allowed to increase by changing the number of consecutive epochs of sleep required to generate the signal to close the occlusion valve. As a result, the apnea index was allowed to increase from 10 to 30 events per hour of sleep on Nights 1 to 7 to a plateau of 50 to 60 events per hour of sleep after 14 nights.

Arterial Baroreflex Control of HR

Each animal was assimilated into the laboratory environment during a control period of 1 to 2 mo before induction of OSA, during 1 to 3 mo of OSA, and during a recovery period of 1 mo after cessation of OSA. Control measurements of baroreflex regulation of heart rate were made 2 to 3 wk before initiation of OSA. Baroreflex measurements were made during the last week of the OSA phase. During the recovery phase, baroreflex measurements were made at least 4 wk after cessation of OSA, when daytime BP had returned to control levels.

The arterial baroreflex control of HR was examined in the daytime, with the dogs lying recumbent. Studies were performed during wakefulness as determined by EEG and behavioral criteria (16). During the studies, the dogs breathed through a cuffed endotracheal tube (10 mm internal diameter) inserted through the chronic tracheostomy. The endotracheal tube was attached to a pneumotachograph (Fleisch No. 2) and a differential pressure transducer (Validyne MP-45) for measurement of airflow. The airflow signal was integrated (Beckman 9873B resetting integrator coupler) to provide tidal volume. Airway pressure was measured with a transducer (Gould Statham P23Db transducer), which was calibrated against a water manometer. In addition to the raw ECG signal, the instantaneous HR was also derived (Cardiotachometer Coupler, Beckman, type 9857).

Arterial BP was measured with the implanted catheter and telemetry system that was validated previously (13). Briefly, the radiofrequency signals from the implanted catheter were detected by a receiver mounted beside the animal, sent to a computer (IBM compatible 486/33 MHz, 16 MB RAM), and captured by a data acquisition system (Dataquest IV; Data Sciences). Because the pressure implant sensed absolute pressure (i.e., relative to vacuum), an electronic barometer (C11PR; Data Sciences) was incorporated into the system to correct for changes in barometric pressure. The BP signal was also converted from digital to analog form, and was recorded on chart paper (at 5 mm · s¹) and stored on magnetic tape (Hewlett Packard 3968A, DC-156 Hz bandwidth). The implanted catheter was calibrated periodically by comparing the BP values from the telemetry system to those measured with a manometer-tipped catheter inserted acutely into the contralateral femoral artery (13).

For purposes of these studies, the dogs were mechanically ventilated using a volume-cycled ventilator (Model No. 613; Harvard Medical Apparatus Inc., South Natick, MA) to produce a constant respiratory rate and tidal volume throughout the experiment in order to avoid the confounding influence of changes in spontaneous breathing pattern and blood gases on blood pressure. The dogs were ventilated with room air to a steady-state level of PCO₂, approximately 2 mm Hg below the level measured during at least 5 min of spontaneous quiet breathing on the day of the experiment. This degree of hyperventilation was sufficient to abolish spontaneous breathing movements, as judged by the smooth and reproducible flow and pressure traces produced by the ventilator, and as confirmed previously by silencing of respiratory muscle discharges (17, 18). The volume of the ventilator was set at 100 ml above the tidal volume observed during spontaneous breathing and the ratio of inspiratory time to expiratory time was 2:3, similar to normal values. After the desired PCO₂ had been reached and the dog was comfortable, the level of mechanical ventilation was held constant.

Once a stable ventilatory pattern was established, the arterial baroreflex control of HR was examined by measuring HR responses to graded changes in BP that were produced with infusions of phenylephrine (20 to 120 mg, intravenously) and of sodium nitroprusside (25 to 125 mg, intravenously), administered over 30 to 60 s. To minimize the effects of respiratory sinus arrhythmia on HR, which are prominent in dogs, steady-state systolic BP and R-R intervals were measured over 45 to 60 s. Different dosages of the vasoactive agents were used to obtain a range of BP changes (± 20 mm Hg) from control level. The different dosages and vasoactive agents (i.e., phenylephrine and nitroprusside) were administered in a random order. Infusions were separated by sufficient time to allow the return of BP to the control value (at least 20 min).

Data Analysis

For each drug infusion trial, steady-state systolic BP and R-R interval were measured over 45 to 60 s. Multiple drug infusions (n = 10 to 14) were performed in each dog to obtain an assessment of the arterial baroreflex control of HR over a range of pressures. The relationship between R-R interval and systolic BP was graphed in each dog for the three phases of the study.

In general, two statistical questions were asked: (1) Did OSA result in a shift of the systolic BP versus R-R interval curve to higher pressures? (2) Did OSA result in a change in gain (i.e., slope) of the systolic BP versus R-R interval curve? The changes in baseline daytime BP and HR during the OSA phase were compared with control values. To determine if there was a change in slope during the three phases, the relationship between systolic BP and R-R interval was assumed to be linear. Therefore, linear regression analysis was performed using commercially available software (Sigmastat; Jandel Scientific, San Rafael, CA) to determine the slope and correlation values. The slope values for each dog in the control, OSA, and recovery phase were then compared by repeated-measures analysis of variance (ANOVA). Differences were considered statistically significant if the null hypothesis was rejected at a level of p < 0.05 using a two-tailed test. Data are presented as mean ± standard error (SEM).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

After the induction of OSA, all three dogs developed sustained daytime hypertension compared with the control period (increase in mean daytime BP 6.0, 13.1, and 26.8 mm Hg, respectively, in dogs 1, 2, and 3 [4]). Resting daytime HR decreased by 3.9 beats/min in dog 1, and increased by 2.9 and 2.5 beats/min in dogs 2 and 3, respectively, but none of these changes was statistically significant (p  0.2). The change in resting BP without an associated change in HR indicates that the set point of the baroreflex had been altered.

Figure 1 shows the relationship between systolic BP and R-R interval in each of the three dogs, before, during, and following OSA. To simplify the statistical analysis, the relationship between systolic BP and R-R interval was assumed to be linear. Table 1 shows the slopes and correlation coefficients for these linear regression relationships. R values for all correlations were greater than 0.8, indicating that the linear regression model was appropriate for the analysis of the data. There was no statistically significant difference between the slope values in the control, OSA, and post-OSA phases (ANOVA; all p values > 0.2). The lack of change in baroreflex sensitivity (i.e., slope) is also demonstrated in Figure 2, where the change in R-R interval is plotted against the change in systolic BP.

View larger version (13K): [in this window] [in a new window]   Figure 1.   Relationship between systolic BP and R-R interval pre-OSA (filled circles), during OSA (open squares), and post-OSA (open triangles). Lines represent calculated linear regressions. Note that the OSA data points are shifted to the right compared with pre- and post-OSA points.

View larger version (13K): [in this window] [in a new window]   Figure 2.   Relationship between changes in systolic BP and changes in R-R interval pre-OSA (filled circles), during OSA (open squares), and post-OSA (open triangles). The curves represent the best fit mathematically for the control (pre-OSA) data. Note that the OSA data points fall on the same curve as the control points.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

We have utilized a canine model to examine the effects of experimentally induced OSA on baroreflex control of HR. We have demonstrated that although OSA produces sustained daytime hypertension, there is no associated change in daytime HR or change in slope of the baroreflex. The findings indicate that OSA is associated with a resetting of the baroreflex to higher pressures, without a change in sensitivity.

There have been few studies of the independent effects of OSA on baroreflex function. In a study by O'Donnell and coworkers (19), sleep-deprived dogs subjected to 12 h of intermittent airway obstruction developed a significant increase in mean arterial BP in the morning, but no decrease in HR. These findings led the authors to suggest that the night of repetitive airway obstruction increased the set point of the cardiac baroreflex, which may have contributed to maintaining an elevated BP after cessation of airway obstruction. However, their conclusions were based on only a single measurement on the baroreflex curve. In humans, Ziegler and coworkers (11) demonstrated no statistically significant difference in baroreflex slope (using phenylephrine injections) between apneic and nonapneic subjects. However, the patients were not matched for age and gender.

In contrast, a recent study by Carlson and colleagues (12) compared baroreceptor function in patients with OSA with that of control subjects who were matched for gender and age, but not for body mass index and BP. The changes in R-R interval and mean sympathetic nerve activity evoked by nitroprusside-induced reductions in BP revealed a depressed sensitivity in both muscle sympathetic activity and heart rate in the patients with OSA, with the magnitude of depression being dependent upon both age and systolic BP. The discrepancy between their findings and those of our study may be related to the method of examining the baroreflex, in that Carlson and colleagues (12) used only one vasoactive agent (nitroprusside), and therefore did not examine the full baroreflex pressure-response slope. Furthermore, their non-OSA control group was not completely matched to the patient group, and the patients were not studied after recovery from OSA. Alternatively, the discrepancy may relate to the differences in chronicity of OSA between our dogs (1 to 3 mo) and their patients (presumably years). If so, our findings would suggest that depression of baroreflex sensitivity is a secondary event in OSA, and not the underlying cause of sustained daytime hypertension.

Despite the differences in species, our findings are likely relevant to human OSA, given the similarities between this canine model of OSA and the human condition (15). However, it is theoretically possible that bypassing of the upper airway in our dogs somehow altered the BP response to OSA. Although there was variability among the dogs in the degree of daytime hypertension they developed, each animal served as its own control; hence the variability among the dogs had little impact on the findings.

Human hypertension is characterized both by a rightward shift in the operating set point of the blood pressure-heart rate relationship (i.e., resetting), and by a blunting of the magnitude of the HR response to increases or decreases in arterial pressure (i.e., impaired baroreflex sensitivity or gain) (20). The latter change has been attributed to a decrease in baroreceptor afferent nerve firing in response to increases or decreases in arterial distending pressure (10, 20). However, in the present experiments, there was a rightward resetting without any concomitant reduction in the slope, or gain, of the blood pressure-heart rate relationship, and the baroreflex control of HR reverted to its baseline characteristics once the OSA and hypertension reversed. These differences between the present and previous models of experimental and, indeed, human hypertension suggest that this resetting developed as a result of central neural adaptations to repetitive periods of apnea, hypoxia, hypercapnia, arousal, or surges in BP, rather than as a consequence of damage to or desensitization of the baroreceptor afferent nerve endings. It is conceivable, however, that had OSA been sustained for a longer period of time, desensitization of or damage to the baroreceptors would have resulted in a decrease in baroreflex gain. Regardless of the mechanisms involved, the parallel shift of the baroreflex curves during OSA indicates that the same systolic BP was associated with a lower R-R interval (or higher HR), and that the baroreflex was no longer capable of counteracting the higher BP (10), and may have been acting indirectly to maintain it (21).

Methodological Considerations

We chose to use vasoactive drugs to test the arterial baroreflex control of HR because this method holds a number of practical advantages over other approaches. Unlike the neck chamber method, the approach we used is unobtrusive and the subject is not aware that the pressure is being perturbed (20). In addition, the method does not require the cooperation of the subject. We performed the testing under standardized conditions of constant mechanical hyperventilation to avoid the important effects of random changes in breathing pattern on BP and HR (20). There are, however, limitations to the method we used to test the arterial baroreflex. First, this method allowed us to examine the effects of OSA on the baroreflex control of HR, but not of peripheral vascular resistance. Second, the infusion of vasoactive agents may elicit a response from nonarterial baroreceptors located in the heart and in the pulmonary circulation (22).

We chose to administer the vasoactive drugs by infusion rather than by bolus injection for a number of reasons. First, the prominent sinus arrhythmia normally present in dogs tends to obscure the HR response to bolus injections of drugs. Second, the infusion method, titrated to specific BP levels, allows careful definition of the entire systolic blood pressure-R-R interval relationship (20). Third, the use of steady-state BP and R-R intervals for the analysis overcomes the need for precise synchronization of arterial BP and R-R interval measurements (20). Finally, the steady-state techniques reflect the integrated contributions to HR of both the cardiac sympathetic and vagus nerves, whereas the ramp bolus method is thought to be influenced predominantly by vagal inputs (22).

A potential confounding effect of the infusion technique is that a given infusion could have long-lasting effects that might influence the baseline BP and therefore modify the response to a subsequent infusion. To minimize this effect, we allowed at least 20 min between drug infusions and ensured that BP had returned to control levels (i.e., preinfusion) before initiating the next infusion. The use of both pressor and depressor agents allowed us to define a greater range of the relationship between BP and R-R interval, but the use of both phenylephrine and nitroprusside ignores potential hysteresis of the baroreflex with rising versus falling pressures (20). However, we did not observe evidence of hysteresis in these studies.

We focused the analysis on the relationship between systolic BP and R-R interval because the highest pressure may be the best index of the stimulus acting on the baroreceptor. However, we also analyzed the relationship between mean arterial BP and R-R interval and despite greater variability, the conclusion was the same: specifically, that there was no change in the slope of the baroreflex during OSA.

A potentially important limitation of this study is that the conclusions are based on data from only three dogs. As a result, there is a possibility that the negative finding (i.e., no change in sensitivity of the baroreflex control of HR during OSA) is a false negative, related to the power of the statistical analysis. However, unlike many other studies of baroreflex sensitivity, each dog in our study served as its own control, a clear advantage, given the inherent variability in baroreflex sensitivity from one animal to another. Furthermore, the slopes of the relationships between BP and R-R interval in each of the control, OSA, and recovery phases were based on a large number of data points (10 to 14 in each phase). As a result, the calculated variance of each of the slopes was relatively small (Table 1), which would favor the finding of a statistical difference, if such differences existed. Finally, studies of other models of hypertension in dogs have demonstrated decreases in baroreflex sensitivity in the range of 44 to 51% (23, 24). This magnitude of decrease is well beyond the 95% confidence limits of the baroreflex sensitivity curves of the current study. Hence, had the development of systemic hypertension in our canine model of OSA involved a physiologically important decrease in baroreflex sensitivity, the decrease would have been readily identified statistically despite the small number of animals in the study. Furthermore, the fact that baroreflex sensitivity decreased slightly in one dog during the OSA phase, increased slightly in another, and was unchanged in the third, argues against a change in sensitivity being responsible for the increase in BP that was consistently observed in each dog during OSA.

In conclusion, we have demonstrated that the sustained daytime hypertension that accompanies experimental OSA in the dog is not associated with changes in HR or in sensitivity of the arterial baroreflex control of HR.