INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Obstructive sleep apnea (OSA) is a common sleep disorder that is characterized by repetitive episodes of upper airway collapse during sleep. The resulting interruption of airflow, despite persisting respiratory efforts, leads to abrupt and marked falls in inspiratory intrathoracic pressure, which may have important effects on ventricular loading conditions. In addition, these episodes may be accompanied by hypoxemia, hypercapnea, and acidosis. Although the exact impact of obstructed inspirations on left ventricular (LV) performance remains controversial, most observers agree that such respiratory efforts cause transient, but substantial, increases in LV afterload (1). Recently, strong evidence of a direct causal relationship between chronic OSA and sustained, daytime hypertension was reported (2). Therefore, the syndrome of OSA can impact LV performance via two separate mechanisms. First, during the course of chronic OSA, multiple episodes of obstructed inspirations during sleep may have direct deleterious effects on ventricular function. Second, chronic OSA can lead to sustained hypertension, which has its own important effects on LV load.

A number of clinical observations support the notion that chronic OSA has important effects on LV function. For example, men with OSA have been shown to have increased LV mass even in the absence of hypertension (3). Furthermore, in patients with dilated cardiomyopathy and OSA, treatment of the sleep apnea with continuous positive airway pressure during sleep is associated with an increase in LV ejection fraction (4). Chronic OSA also appears to be an independent cardiovascular risk factor in that a number of epidemiologic studies have identified OSA as an independent risk for systemic hypertension, myocardial infarction, stroke, and sudden death (5, 6).

In the present investigation we used the impedance catheter technique to examine the acute effects of obstructive apneas on LV performance in a chronic, canine model of OSA. In addition to these acute measurements, we employed quantitative echocardiography to examine the impact of chronic OSA on LV systolic function and dimensions. The data presented are unique in two important ways. First, the animals were studied during a chronic phase of OSA that is associated with responses quite different from those observed during acute airway obstruction in previously healthy animals. Second, the model allowed beat-by-beat measurement of LV pressure and volume in nonsedated animals during spontaneous episodes of sleep.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The Canine Model of OSA

OSA was produced in the dog using a modification of the model described by our group (2). All surgical and experimental procedures were approved by the Animal Care Committee of the University of Toronto. Briefly, four dogs weighing 23 to 31 kg (three female, one male), trained to sleep in the laboratory, underwent two surgical procedures at least 2 mo before initiation of studies, the first to create a permanent side-hole tracheostomy, and the second to implant a three-channel telemetry unit (TLM11M3D70-CCP; Date Sciences, St. Paul, MN) for monitoring of arterial blood pressure, electroencephalogram (EEG), and cervical electromyogram (EMG). Surgery was performed under general anesthesia and aseptic conditions, as described previously in detail (2).

The radio-frequency signals of EEG, EMG, and blood pressure emitted by the telemetry unit were detected by water-resistant receivers (RL2000; Data Sciences) that were positioned around the pen in which the dog was housed. The EEG and EMG signals were processed and relayed continuously overnight to a microcomputer through an analog-to-digital converter. The computer sampled the signals and produced a judgment of sleep-wake state every 6 s based on the frequencies in the EEG signal and the amplitude of the averaged EMG signal (2). Once a period of sleep of predetermined length (generally 18 s) was identified by the computer, it generated a radio signal that was detected by a receiver-controller unit housed in a jacket worn by the dog. The signal activated a quiet, custom-designed occlusion valve attached to the endotracheal tube (Aire-Cuf, Bivona, IN; 9 mm internal diameter, 12.3 mm external diameter) through which the dog breathed, resulting in an obstructive apnea. When the dog awoke from sleep, the computer sent a signal to release the occlusion. Thus, the model simulated closely OSA in humans by producing repeated episodes of airway occlusion and arousal from sleep. Because the model used biotelemetry and computer technology, there were no lead attachments to the dogs, allowing them to move about freely; and the system required no human intervention, except for routine monitoring and maintenance.

Measurement of Responses to Acute Airway Occlusions

A total of three dogs underwent acute hemodynamic studies in which the impact of airway occlusion on left ventricular function was investigated. The dogs were studied after they had been subjected to chronic OSA for 25 to 95 d. These studies were performed in conscious animals during spontaneous episodes of sleep. One of the dogs was studied on two occasions, yielding a total of four hemodynamic studies. The fourth animal could not be studied because of problems with vascular access.

An 8-Fr introducer (Cordis Laboratories, Miami, FL), was inserted into the right carotid artery with the dog under light halothane anesthesia. Subsequently, a 7-Fr multielectrode impedance catheter was advanced to the left ventricle using continuous monitoring of the pressure waveform. The position of the impedance catheter in the ventricular apex was confirmed using transthoracic two-dimensional echocardiography (Hewlett-Packard Co., Medical Products Group, Andover, MA). Then, a 2-Fr micromanometer catheter (Model SPC-320; Millar Industries, Houston, TX) was placed within the lumen of the impedance catheter using a Y-connector. The volume catheter was connected to a stimulator-microprocessor system (Sigma V, Leycom; Cardiodynamics BV, Leiden, The Netherlands). The system uses a dual excitation algorithm and two pairs of stimulation electrodes with multiple intervening electrodes, spaced 1 cm apart to measure resistances. These resistance values are converted to "segmental" blood volume measures, which are added to yield "total" chamber volume.

The catheters and introducer were anchored with tape to the cervical skin, and the animal was allowed to recover from the effects of anesthesia for at least 1 h before initiation of the studies on the responses to acute airway occlusion. This length of time was sufficient for the physiologic responses to acute airway occlusion to return to normal (2). For these studies an electrocardiogram was recorded via platinum-coated subdermal needle electrodes (Type E2; Grass Instruments Co., Quincy, MA) placed in the chest wall. The EEG was monitored via two subdermal needle electrodes placed in the scalp on either side of the midline. An electrode was placed in the neck and served as common ground for both the electrocardiogram and the EEG. During these studies, the dogs breathed through a cuffed endotracheal tube (10 mm internal diameter) inserted through the chronic tracheostomy. The endotracheal tube was attached to the breathing circuit, which included a pneumotachograph (Fleish No. 2) and a three-way valve. Airflow was measured with the pneumotachograph and a differential pressure transducer (Validyne MP-45; Validyne Co., Northridge, CA). The airflow signal was integrated (Beckman 9873B resetting integrator coupler; Beckman Instruments, Fullerton, CA) to provide tidal volume. Airway pressure was measured with a transducer (P23Db transducer; Gould Statham, Cleveland, OH), which was calibrated against a water manometer.

For these studies of the responses to acute episodes of airway occlusion, the dogs were allowed to sleep in the laboratory. Airway occlusion during sleep was manually produced, at the end of expiration, by closure of the three-way valve attached to the breathing circuit. When arousal from sleep occurred, the occlusion was released. The presence of wakefulness and sleep were determined according to EEG and behavioral criteria.

All signals were recorded on a strip chart recorder (Beckman R711). The electrocardiogram, high fidelity LV pressure, LV volume, and tracheal pressure were digitally recorded at 300 Hz using a microcomputer equipped with a multichannel analogue-to-digital converter. Data files were acquired during episodes of sleep and obstructed inspiratory efforts and were recorded to disk for later analysis.

Effects of Chronic OSA

The chronic effects of OSA on resting LV function were examined using transthoracic echocardiography. For logistical reasons, in two dogs, echocardiograms were obtained prior to initiation of the OSA phase and approximately 3 months after initiation of OSA. In the other two dogs, the echocardiograms were obtained after 25 and 95 d, respectively, of chronic OSA, and after a 2-mo recovery period following the cessation of OSA. Echocardiographic studies were obtained in the dogs while they were awake, breathing normally, and lying in the left lateral decubitus position. Echocardiographic images were obtained using a Hewlett-Packard phased-array ultrasonoscope device with a 2.5 Mhz transducer (Hewlett-Packard Co., Medical Products Group). Images were obtained in the ventricular short axis and the apical four-chamber planes.

Data Aquisition

Hemodynamic data obtained during the acute studies were analyzed using a customized hemodynamic analysis package developed in Labview (Version 3.0; National Instruments Corporation, Austin, TX). Gain and offset corrections for LV volume were calculated by applying absolute LV end-systolic and end-diastolic volumes determined from quantitative echocardiography. Using this calibrated volume signal, end-systolic and end-diastolic LV volumes were calculated. The program also allowed for beat-by-beat determination of LV peak systolic and end-diastolic pressure along with the maximal first derivative of LV pressure (peak +dP/dt). The rate of isovolumic relaxation (Tau) was determined using two different techniques. The first method is a modification of the logarithmic method, such that Tau (T_L) = 1/slope of the regression line for the natural logarithm of LV pressure versus time for the period from peak -dP/dt to 5 mm Hg above LV end-diastolic pressure. The second method employed is the direct measurement of the pressure half-time (T_1/2). With this method, Tau is measured directly as the time required for LV pressure to fall to one half of its value at -dP/dt max.

In addition to measures of absolute LV pressure, the program calculated transmural LV pressure during airway occlusions, determined as the LV pressure minus the simultaneously measured tracheal pressure that was assumed to reflect intrathoracic pressure. From this derived pressure, transmural LV peak systolic and end-diastolic pressure were determined along with the maximal first derivative (peak transmural +dP/dt). Furthermore, the program also provided measures of both T_L and T_1/2 determined from this transmural pressure (TM T_L and TM T_1/2, respectively). During quiet respiration, prior to airway occlusion, intrathoracic pressure was assumed to be equivalent to atmospheric pressure. This assumption was considered reasonable since esophageal balloon catheter measures of intrathoracic pressure range from 5 to 8 cm H₂0 (approximately 4 to 6 mm Hg) during tidal breathing in healthy dogs (7). Tracheal pressure was also used as a surrogate for intrathoracic pressure during periods of airway occlusion when there was no airflow. However, this assumption could not be made during the period of exaggerated respiratory efforts after release of the airway occlusion when the large swings in intrathoracic pressure were not reflected in corresponding changes in tracheal pressure. Therefore, transmural LV pressures were not calculated during the period of recovery from airway occlusion.

Data Analysis

The effects of obstructed respiratory efforts on LV pressure and volume were approached in two ways. The first analysis examined the beat-by-beat effects of obstructed respiratory efforts on LV pressure and volume. In this case ventricular pressure and volume parameters were analyzed for the cardiac cycle immediately prior to an obstructed inspiration, during the first three cardiac cycles occurring within a single inspiratory effort against a closed airway, and, finally, for the two cardiac cycles immediately after cessation of the inspiratory effort. Several inspiratory efforts were made during each obstructive apnea, with the inspiratory efforts increasing progressively during each period. The beat-by-beat analysis was performed during the last third of the obstructive apnea, when inspiratory efforts were maximal. For each animal, data from two such obstructed inspiratory efforts were analyzed.

In the second analysis, a series of 15 to 30 cardiac cycles were analyzed during the period just prior to an episode of airway obstruction, during the period of obstructed respiratory efforts (which included both inspiratory and expiratory phases), and, finally, during the period of arousal immediately after release of the airway occlusion. For this analysis, all cardiac cycles from each period were analyzed and the results for each hemodynamic parameter were expressed as the mean. For each animal, two to three sequences of sleep, obstructive apnea, and recovery were analyzed.

Two-dimensional and M-mode echocardiographic data were analyzed using a custom designed quantitative echocardiographic work station. LV end-systolic and end-diastolic volumes were measured in the apical four-chamber view using a single plane, modified Simpson's rule. LV mass was estimated using the Penn convention (8).

Statistical Analysis

Data are presented as mean ± standard error of the mean. The effects of airway obstruction on heart rate, LV volume, and hemodynamic variables were analyzed by a one-way, repeated-measures analysis of variance. In the case of the beat-by-beat analysis, comparisons were made between the cardiac cycle immediately prior to the obstructed inspiratory effort and individual cardiac cycles occuring during and immediately after the obstructed respiratory effort. For the second analysis the mean value for each hemodynamic and LV volume variable during the preobstruction period was used as baseline. In cases where significant F values were found, specific preplanned comparisons between baseline and subsequent data points were performed with application of Scheffe's post hoc test for individual comparisons. Echocardiographic parameters were compared using paired t tests. Differences were considered significant if the null hypothesis could be rejected at the 0.05 probability level.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

During each hemodynamic study there were 10 ± 2 periods of sleep and obstructive apnea. The duration of obstructive apneas ranged from 15 to 40 s during which the dogs made an average of 7 ± 3 inspiratory efforts. Respiratory effort tended to increase during the period of airway occlusion, and the maximal inspiratory effort was associated with airway pressures of 40 to 60 mm Hg.

The effects of repeated respiratory efforts against a closed airway are demonstrated in Figures 1 and 2. It can be seen that such efforts have complex effects on LV pressure and volume (to be described below). The interaction is made more complex by marked changes in cardiac cycle length, which itself can have important effects on LV pressure and volume.

View larger version (50K): [in this window] [in a new window]   Figure 1.   Recordings of absolute ventricular pressure, tracheal pressure, and calculated transmural left ventricular pressure during a representative sequence of quiet respiration during sleep followed by a series of obstructed inspirations and subsequent arousal. Note the marked increase in left ventricular transmural systolic pressure during the occluded inspiratory efforts. Inspiratory efforts are easily identified as the periods during which tracheal pressure and diastolic left ventricular pressure become negative. (Tracings were obtained from Dog 1.)

View larger version (43K): [in this window] [in a new window]   Figure 2.   Recordings of tracheal pressure, absolute left ventricular pressure, transmural left ventricular pressure, and left ventricular volume during a series of inspiratory efforts against a closed airway. Note the marked increase in left ventricular transmural systolic pressure and volume during the occluded inspiratory efforts. Inspiratory efforts are easily identified as the periods during which tracheal pressure and diastolic left ventricular pressure become negative. (Tracings were obtained from Dog 2.)

Beat-by-Beat Effects of Obstructed Inspiration on LV Pressure and Volume

An example of the effects of obstructed inspiratory efforts on LV pressure and volume is shown in Figure 2. The effect of a single inspiratory effort on LV volume, pressure, and isovolumic performance indices is summarized in Table 1 and in Figure 3. The data presented are from the cardiac cycle immediately prior to an obstructed inspiratory effort, three consecutive cardiac cycles during an obstructed inspiration, and the two cardiac cycles immediately after the inspiratory effort. Absolute LV pressures, both peak systolic and end-diastolic, were significantly decreased during the period of airway occlusion. However, when the data were analyzed using transmural LV pressure, the opposite effect was observed because of the large subatmospheric intrathoracic pressure (Figure 1). In this case, there was a clear increase in transmural LV systolic pressure and a tendency for transmural LV end-diastolic pressure to rise. There was a significant decrease in LV end-diastolic volume during the first two cardiac cycles within an obstructed inspiration. In contrast, there was a small but highly significant increase in LV end-systolic volume during the period of obstructed inspiration. Therefore, during the period of obstructed inspiration there was a striking fall in stroke volume, which returned to normal immediately after the inspiratory effort ended.

View larger version (51K): [in this window] [in a new window]   View larger version (42K): [in this window] [in a new window]   Figure 3.   (Top four panels) Beat-by-beat effects of obstructed inspiration on absolute and transmural left ventricular peak systolic and end-diastolic pressures. (Bottom three panels) Beat-to-beat effects of obstructed inspiration on left ventricular end-systolic, end-diastolic, and stroke volumes. Preocclusion indicates cardiac cycle immediately preceding obstructed inspiration. During occlusion indicates the first three cardiac cycles during the period of obstructed inspiration. Postocclusion represents cardiac cycles immediately after the inspiratory effort. *p < 0.05 compared with preocclusion value.

In addition to the preceding observations, there was also a significant decrease in LV peak +dP/dt during the first two cardiac cycles within the period of obstructed inspiration (Table 1). Subsequently, LV peak +dP/dt returned to preocclusion values. A similar result was obtained when LV peak +dP/dt was derived from transmural LV pressure (data not shown). T_L and T_1/2 could not be evaluated during the period of obstructed inspiration because of the marked decrease in LV diastolic pressures, although values for both TM T_L and TM T_1/2 did not change during this period (Table 1). During the postocclusion period values of all measures of Tau returned to preocclusion values (Table 1).

LV Responses to Repeated Obstructed Respiratory Efforts

The effect of airway occlusion on heart rate, LV volumes, and pressures, integrated over time, is shown in Table 2. The data represent mean values for volume and pressure indices for 15 to 30 cardiac cycles immediately preceding a period of airway occlusion, during a series of obstructed respiratory efforts (including both the inspiratory and expiratory phases), and during the immediate recovery period after release of airway occlusion. When the data are looked at in this way, it can be seen that obstructed respiratory activity has important effects on LV volume and pressures. On average, LV end-diastolic volume did not change, but there was a significant increase in LV end-systolic volume (from 29 ± 8 preocclusion to 33 ± 9 ml during occlusion, p < 0.05). Therefore, during airway occlusion there was a significant decrease in left ejection fraction (from 48 ± 7 to 43 ± 7%, p < 0.05). Repeated obstructed inspirations caused a significant decrease in LV end-diastolic pressure (from 6 ± 1 to 3 ± 2 mm Hg, p < 0.05) and a small increase in LV peak systolic pressure, which was of borderline statistical significance (134 ± 11 versus 143 ± 14 mm Hg, p = 0.06). In contrast, the effect of repeated obstructed inspirations on LV transmural pressure was quite different, with a small increase in transmural LV end-diastolic pressure (7 ± 2 versus 10 ± 2 mm Hg, p = NS) and a highly significant increase in transmural LV peak systolic pressure (134 ± 12 versus 149 ± 15 mm Hg, p < 0.05). There was no significant change in LV peak +dP/dt or in the rate of isovolumic LV pressure relaxation (data not shown). LV volumes and pressures returned to preocclusion values during the postocclusion period (Table 2).

Effects of Chronic OSA on Echocardiographically Determined LV Volumes and Systolic Performance

As has been previously reported, chronic OSA caused sustained daytime hypertension in these animals (2). During the control phase (before initiation of OSA), the mean daytime arterial blood pressure was 97 ± 1 mm Hg. In contrast, the mean daytime arterial blood pressure at the time of echocardiography during the phase of OSA was 116 ± 7 mm Hg.

Two-dimensional echocardiograms obtained during the phase of OSA demonstrated an increase in LV end-systolic volume (from 20 ± 4 to 27 ± 6 ml, p < 0.05) (Table 3). There was also an increase in LV end-diastolic volume, although the increase was not statistically significant. Finally, during the phase of OSA, there was a significant decrease in LV ejection fraction as compared with control values (58 ± 3 versus 51 ± 3%, p < 0.05). There was no change in left ventricular wall thickness, and an increase in left ventricular mass that did not achieve statistical significance.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The relationship between sleep apnea syndromes and cardiovascular disease, identified in several epidemiologic studies, has led to a number of investigations concerning the effects of both apneas and changing airway pressure on LV function (1, 9). The present investigation is the first to report beat-by-beat measures of the effect of obstructed inspiratory efforts on LV pressure and volume in unanaesthetized animals during spontaneous sleep.

Analysis of the data confirms that the effect of obstructed inspiration on LV function is complex and varies on a beat-by-beat basis. Initially, LV end-diastolic volume falls, although this fall is transient, with ventricular volumes returning to normal immediately after the inspiratory effort. Although LV end-diastolic pressure decreases substantially, transmural LV end-diastolic pressure, a more accurate measure of LV distending pressure, tends to rise. The fact that LV end-diastolic volume falls in the absence of a change in transmural LV end-diastolic pressure is consistent with previous observations that LV filling is impaired by ventricular interaction and septal shift (11, 19, 20). Changes in cycle length could also have had an effect on LV end-diastolic volume. As can be seen in Figure 1, each obstructed inspiratory effort is associated with a transient increase in heart rate. This decrease in cycle length and diastolic filling could also have an impact on LV diastolic volumes. Obstructed inspiration also had important effects on LV end-systolic volume, with progressive increases in this variable during each cardiac cycle within the inspiratory effort. Once again, the change was transient, with end-systolic volumes returning to baseline values shortly after the end of the inspiratory effort. The mechanism of the increase in end-systolic volume appears to be multifactorial, although it may have occurred secondary to the marked increase in LV transmural systolic pressure. Changes in arterial oxygen saturation may also have played a role. Although not reported in the present experiment, previous experience with this model has demonstrated that during periods of airway occlusion arterial saturation fell by an average of 9% during non-rapid-eye-movement sleep and 25% during rapid-eye movement sleep (21). Whatever the mechanism, the combination of increased LV systolic transmural pressure and end-systolic volume represents an increase in LV systolic wall stress that may have important consequences for LV metabolism and function and may, in part, explain the reported association between OSA and myocardial ischemia (5). The increase in LV end-systolic volume observed during obstructed inspiratory efforts was associated with a significant decrease in stroke volume. This result is consistent with the recent report of Schneider and colleagues (22). In their chronically instrumented dog model, LV stroke volume was measured with an aortic flow probe, although LV pressure and dimensions were not assessed. During spontaneous episodes of sleep, obstructed inspiratory efforts were associated with signficant decreases in LV stroke volume.

When data from 15 to 30 cardiac cycles and a number of obstructed inspiratory-expiratory cycles were averaged, an integrated assessment of the overall effect of obstructive apnea on cardiac performance was obtained. Consistent with the beat-by-beat analysis, LV end-systolic volume increased during the period of airway obstruction, which was associated with an important decrease in ejection fraction. As was discussed above, the observed decrease in fractional shortening is likely a multifactorial process with contributions from changing afterload, preload, and inotropic state. Interestingly, there are some important differences when the data are examined in this way. First, LV end-diastolic volume did not change. Indeed, LV end-diastolic volume tended to increase in this analysis, likely reflecting the increase in cycle length and increase in diastolic filling time observed between inspiratory efforts. LV systolic pressure tended to increase during the time of airway obstruction, whereas there was a clear decrease in LV systolic pressure during individual obstructed inspirations. Once again, this difference results from the fact that LV systolic pressure was increased during the periods between inspiratory efforts.

The results of the present study are different from some previous reports of the effect of airway occlusion on LV performance in animals. Scharf and colleagues (23) have carried out detailed experiments examining the effects of airway occlusions on LV pressures and volumes in anaesthetized dogs. In this model they report that obstructed inspiration in animals breathing room air was associated with an increase in LV end-diastolic pressure but no change in transmural LV end- diastolic pressures. Transmural peak LV systolic pressure decreased and there was a significant fall in LV +dP/dt. Despite marked changes in load and peak +dP/dt, there was no change in LV segment lengths, suggesting no change in LV volume. It would appear that the discrepancies between their results and what we now report are based on methodologic differences. Most importantly, their studies involved an acute surgical preparation, with the animals under general anesthesia (23). In contrast, our experiments were performed in a long-term unanesthetized dog model during episodes of spontaneous sleep. In addition, in the report by Scharf and colleagues, the dogs were subjected to more profound periods of obstructive apnea, consisting of 1-min episodes of airway occlusion, separated by 2-min recovery periods, with the impact of airway occlusion being assessed after five to seven such cycles. In our preparation, the airway occlusions were shorter, being terminated physiologically by arousal from sleep, and the period of recovery between episodes of airway occlusion was longer.

Our observations clearly demonstrate that obstructed inspiratory efforts are associated with increases in LV afterload as reflected by marked increases in peak systolic LV transmural pressure and end-systolic volume. However, the concept that obstructed inspiratory efforts lead to increases in LV afterload has recently been challenged. Chen and Scharf (24) found that apnea induced by skeletal muscle paralysis in sedated animals caused greater increases in systemic blood pressure and more pronounced adverse effects on LV systolic function than did periods of obstructed inspiratory efforts. Subsequently, the same investigators reported that the hypertensive response to airway obstruction in sedated animals could be abolished with ganglionic blockade (25). Therefore, they concluded that increases in systemic blood pressure mediated by the autonomic nervous system have the greatest impact on LV function during periods of apnea, and that the contribution of changes in intrathoracic pressure is minimal. However, this conclusion can be challenged because periodic reductions in intrathoracic pressure (as demonstrated in the present study) can only serve to augment the increase in afterload induced by increases in systemic arterial pressure. Furthermore, it is not surprising that apnea induced by muscle paralysis in animals under light sedation can be associated with marked increases in systemic arterial pressure and decreases in LV systolic performance.

There have also been a number of human studies concerning the effects of obstructed inspiratory efforts on hemodynamics and LV performance. Most studies have examined the impact of the Müller maneuver (9, 16, 18), although there are some reports of the hemodynamic effects of obstructed apneas during sleep (14, 26). The Müller maneuver is physiologically somewhat different from obstructive inspirations during sleep. The former causes a sustained decrease in intrathoracic pressure, whereas the latter is associated with repeated, short inspiratory efforts against a closed airway. Despite these important differences, the hemodynamic and LV responses observed during the first few seconds of the Müller maneuver are similar to those observed with individual obstructed inspiratory efforts in the present study. It is well documented that the physiologic response to the Müller maneuver is dynamic, with clear differences between early and late events within the maneuver (9, 29). In this regard, most studies report a fall in absolute ventricular or systemic arterial pressures, but an increase in estimated transmural pressures (9, 16, 18, 29). The resulting increase in afterload is associated with a fall in stroke volume, and in most (11, 19, 20), but not all, studies (9, 12) a decrease in end-diastolic volume, a shift of the interventricular septum to the left, and an increase in end-systolic volume.

Only limited information is available concerning the effects of obstructed inspirations during sleep in humans with chronic OSA. Garpestad and colleagues (14), using radionuclide measures of LV volumes, examined changes in LV volumes during periods of OSA. They report a marked decrease in stoke volume at the time of arousal caused by an important increase in end-systolic volume. However, this change was observed in comparison to LV volumes during the period of obstructed apnea and no comparison was made with a true baseline prior to the period of airway obstruction. Although the mechanism of this effect remained unclear, arousal in their subjects was associated with a large increase in systolic arterial pressure. Therefore, it is not possible to determine from the report of Garpestad and colleagues (14) what effect obstructive apnea has on LV volumes and ejection fraction as compared with baseline values observed during sleep with normal respiration. Other reports of ventricular function responses to OSA have employed a transthoracic electrical impedance technique to measure LV stroke volume (26, 27). In these studies, episodes of OSA were associated with decreased stroke volume. Importantly, these studies did not report effects of OSA on LV volumes, and beat-to-beat data were not presented.

The use of an impedance catheter in the measurement of LV volumes has a number of potential limitations. First, the catheter provides an uncalibrated volume signal that can be transformed to absolute volumes using a number of approaches (30). In the present study we calibrated the volume signal using values for absolute LV volume derived from two-dimensional echocardiography. Because the purpose of the acute hemodynamic studies was to measure the change in LV volume in response to episodes of obstructed inspiration, the accuracy of absolute volume measurements is not of particular importance. A second potential problem with the impedance catheter relates to the impact of changing load during obstructed inspiration on the relationship between the impedance signal and LV volume. Some investigators have suggested that changes in parallel conductance induced by changes in LV volumes and loading conditions can lead to errors in the measurement of LV volume provided by the impedance technique (31, 32). Other investigators disagree with this view, reporting that the relationship between impedance catheter measurements and true LV volumes remains highly linear over a relatively wide range of ventricular load, allowing reliable predictions of changes in LV volume from changes in the impedance catheter signal (33, 34). Indeed, this controversy has been the subject of editorial comment where it was emphasized that no gold standard is available for the measurement of LV volume (35).

Because the impact of changing thoracic volume on this relationship has not been reported, it is possible that such changes might have some effect on the magnitude of the observed changes in LV volume. Nevertheless, this theoretical problem could not impact the direction of observed changes in LV volume. Indeed, when LV volume was examined on a beat-by-beat basis, obstructed inspiratory efforts were associated with an increase in end-systolic volume accompanied by a decrease in LV volume at end-diastole. Such opposite effects on LV volume cannot be explained by nonlinearity in the relationship between true LV volume and that estimated by the impedance catheter. Therefore we conclude that obstructed inspiratory efforts cause profound increases in ventricular afterload with subsequent impact on end-systolic volume.

The echocardiographic studies performed in these animals provide important information concerning the chronic effects of OSA and confirm previous suggestions that OSA may be associated with adverse effects on LV function. For example, Malone and colleagues (4) described LV dysfunction of unknown etiology in patients with chronic OSA. Abolition of OSA by continuous positive airway pressure was associated with marked improvements in LV ejection fraction. On the basis of these observations, these investigators hypothesized that OSA contributed to the development of LV dysfunction. The present findings provide strong support for their hypothesis. We observed a reduction in LV ejection fraction in association with an increase in end-systolic volume after a period of chronic OSA. These changes were observed in the absence of a significant change in end-diastolic volume. Whether this decrease in fractional shortening is secondary to episodic changes in loading conditions caused by obstructive apneas, or is related to the associated development of sustained daytime hypertension, or a combination of these factors, is a question that requires further investigation.

In summary, the present study provides important new information concerning the effects of chronic OSA on LV function. First, it provides the first direct measure of LV pressure and volume during episodes of obstructive apnea in animals who were not under the influence of general anesthesia. Indeed, it is the first report of the beat-by-beat analysis of the impact of obstructed inspiration on simultaneously acquired LV volume and pressure. Second, we have demonstrated that chronic OSA is associated with important decreases in LV ejection fraction. The fact that this decrease occurred in previously healthy animals with no evidence of cardiac disease is an important finding and raises the possibility that severe chronic OSA may lead to LV systolic dysfunction. Despite the potential importance of these findings it should be emphasized that these observations were made in an animal model of chronic OSA and that the impact of chronic OSA in humans could be different. For example, our observations concerning the acute impact of obstructed respiratory efforts on LV pressure and volume may have been influenced by the marked variation in heart rate observed in the dog. Further, the model created severe OSA in each animal studied, whereas in humans, the severity of the clinical syndrome is highly variable and often has important interactions with other disease states, particularly congestive heart failure. These differences merit some caution in interpretation of the results and emphasize the need to develop better methods of examining the impact of chronic OSA on LV function in humans.