INTRODUCTION

Top Abstract Introduction Methods Results Discussion References

SLEEP-RELATED BREATHING DISORDERS are present in ~50% of patients with stable, symptomatic congestive heart failure (CHF) (12, 17). Several lines of evidence indicate that both obstructive and central sleep apnea have an adverse impact on survival and disease progression, even when these patients receive optimal medical treatment for CHF (10, 13). This effect could occur through both hemodynamic and nonhemodynamic mechanisms (3). Our objective in this study was to focus on potential hemodynamic mechanisms by which such breathing disorders could further compromise the already failing heart and circulation.

One mechanism that could disturb circulatory homeostasis and is unique to obstructive sleep apneas is the generation of exaggerated negative inspiratory intrathoracic pressure against the occluded upper airway. Negative inspiratory intrathoracic pressure swings increase left ventricular afterload by increasing systolic left ventricular transmural pressure, reduce left ventricular filling by mechanisms arising from ventricular interdependence, and impair left ventricular relaxation (4, 35). As a result, obstructive apneas can reduce stroke volume in subjects with normal ventricular function (29, 31). Responses to negative intrathoracic pressure during apnea have been studied in subjects with normal ventricular function (4, 22, 27-29, 31), but the effect of obstructive apneas on cardiac and systemic hemodynamics have not been specifically examined in detail in patients with impaired ventricular systolic function. Moreover, these published reports do not describe the beat-by-beat time course of these hemodynamic responses, and the potentially confounding influence of apnea, itself, was not controlled for in those experiments. Thus our understanding of the hemodynamic consequences of upper airway occlusion is incomplete.

It is important to characterize the potential adverse effects of obstructive apnea in CHF for several reasons. Obstructive apnea is present in at least 10% of CHF patients studied (12, 17). Because the failing heart is much more sensitive to changes in afterload than is the normal heart (21), any adverse effect of increasing left ventricular systolic transmural pressure on stroke volume and cardiac output is likely to be exaggerated in such patients. In a substantial number of CHF patients, distension of the right ventricle during the generation of negative intrathoracic pressure could impair left ventricular diastolic filling through mechanisms such as pericardial constraint and leftward shift of the interventricular septum (1). Because CHF patients often suffer from inadequate tissue perfusion, any further reduction in systemic or regional blood flow during obstructive apneas could have greater functional importance and clinical impact than in patients with normal ventricular function. Observations from our own laboratory indicate that obstructive sleep apnea can play an important role in the progression of CHF, because ventricular systolic function improves when upper airway obstruction is abolished by nasal continuous positive airway pressure (13). It is even possible that unexplained nocturnal death in some patients with CHF (16, 19) represents an extreme manifestation of the nocturnal pulmonary edema or angina that has been reported to arise as a consequence of obstructive apneas (5, 7, 13).

Our objective in these experiments was to characterize the nature and time course of hemodynamic responses to negative intrathoracic pressure generated during apnea in patients with CHF due to systolic dysfunction. Because it is not possible to compare the effects of obstructive apneas of exactly the same length, occurring within the same sleep state, within and between different patients, we simulated the immediate hemodynamic effects of obstructive apneas without inducing confounding variables and interactions arising from hypoxia, hypercapnia, and arousals from sleep, by having patients perform Mueller maneuvers to two prespecified target intrathoracic pressures. Breath holds of equal length were introduced as time controls for apnea. Our hypotheses were, first, that generation of negative intrathoracic pressure and apnea together, during Mueller maneuvers, would cause greater immediate reductions in stroke volume and systolic blood pressure in these patients than would apnea alone and, second, that the magnitude of these changes would be related to the intensity of the stimulus, namely, the negative intrathoracic pressure generated during this maneuver. Because blood pressure tends to decrease over the course of obstructive apneas in patients with CHF (30), we anticipated that the immediate hemodynamic responses to Mueller maneuvers would change over time in association with alterations in left ventricular afterload and preload. Our study design allowed us to address this question on a beat-by-beat basis.

	METHODS

Top Abstract Introduction Methods Results Discussion References

Subjects. We studied eight patients <75 yr of age with chronic (>6 mo in duration) CHF due to either ischemic or idiopathic dilated cardiomyopathy. They were recruited by advertisement from our institutional Heart Failure Program. All patients were in sinus rhythm, suffered from exertional dyspnea despite medical therapy, and had left ventricular ejection fractions of 45% measured at rest by ⁹⁹Tc equilibrium radionuclide angiography. Ischemic cardiomyopathy was diagnosed either by demonstration of coronary occlusion or flow-limiting stenosis (>75% stenosis) on coronary angiography or by a history of documented myocardial infarction. Idiopathic dilated cardiomyopathy was diagnosed by the presence of global left ventricular hypokinesis, a left ventricular end-diastolic dimension 60 mm, normal coronary arteries or non-flow-limiting epicardial coronary narrowing (<50% stenosis) on coronary angiography, and by the absence of histological evidence of myocarditis on endomyocardial biopsy (14). Exclusion criteria included: primary mitral or aortic valvular heart disease and cardiac pacing. The protocol was approved by the Human Subjects Review Committee of the University of Toronto, and all patients gave written informed consent before their participation.

Arterial and esophageal pressure. Finger blood pressure was measured beat by beat by using the volume-clamp method (Finapres, Ohmeda 2300, Englewood, CO), with the arm and hand maintained in the horizontal position throughout the study. This method has been validated against acute changes in intra-arterial pressure during Mueller maneuvers (33). Esophageal pressure (Pes) was measured from an esophageal balloon catheter system attached to a pressure transducer (Validyne MP, 45 ± 50 cmH₂O, Northridge, CA) to quantify pleural pressure. The balloon was placed in the esophagus according to the method of Baydur et al. (2), such that a given change in mouth pressure was accompanied by an equal change in Pes during occluded breaths. Oxyhemoglobin saturation was monitored continuously with an ear pulse oximeter (Oxyshuttle; Sensormedics, Anaheim, CA). R wave-to-R wave (R-R) interval was determined from a precordial electrocardiogram lead. Signals were recorded continuously onto a strip-chart recorder (Gould model 2800S, Cleveland, OH).

Stroke volume and cardiac output. All measurements were performed by using an echocardiographic Doppler technique (Ultramark 8, Advanced Technology Laboratories, Bothell, WA) previously described for our laboratory (18). With patients in the supine position, maximum instantaneous aortic flow velocity was measured in the ascending aorta by using continuous-wave Doppler (2.25 MHz) directed through the suprasternal window. Stroke volume was calculated as the product of the mean time-velocity integral and the cross-sectional area of the aortic annulus orifice (A) calculated as A = (D/2)², where D is the diameter of the aortic annulus obtained from a prior parasternal long-axis view at baseline. Echocardiographic Doppler estimates of stroke volume have been validated under experimental conditions similar to those described here (8). Although such measurements tend to systematically underestimate the absolute stroke volume, they accurately reflect changes in stroke volume (11). Cardiac output was calculated from the product of heart rate and stroke volume. Stroke volume index and cardiac index were then calculated. In addition, to take into account possible alterations in thoracic configuration that might affect measures of time-velocity integrals from the suprasternal window during Mueller maneuvers, we performed initial validation experiments in three of the patients. Time-velocity integrals were acquired from the suprasternal window and from the right carotid artery, an extrathoracic site that would not be affected by alterations in thoracic configuration. Separate measurements were made from each site during baseline tidal breathing and two Mueller maneuvers at a target Pes of 40 cmH₂O (see below). There were no significant differences in the change in time-velocity integrals from baseline between the suprasternal window and the carotid artery averaged over the first five cardiac cycles [24 ± 10 (SD)% vs. 18 ± 24%, P = 0.7] or the last five cardiac cycles of the Mueller maneuvers (33 ± 33% vs. 21 ± 7%, P = 0.34). Thus the direction and magnitude of changes in time-velocity integrals measured from the suprasternal notch parallel those measured from the right carotid artery.

Protocol. Diuretics were withheld the morning of each study. Patients were studied while in a supine position. To test our two hypotheses, responses to interventions were compared with baseline hemodynamic values recorded during quiet breathing before these breath holds and Mueller maneuvers. To provide a time control with which to compare the independent hemodynamic responses to negative intrathoracic pressure generated during apnea, patients performed breath holds of 15-s duration. To determine the immediate and subsequent hemodynamic effects of negative intrathoracic pressure generated during simulated obstructive apnea, patients performed sustained Mueller maneuvers for 15 s. All respiratory maneuvers were performed at end expiration. With a nose clip in place, inspiratory effort was generated against a mouthpiece with a small air leak through a 21-gauge needle to prevent closure of the glottis. Mouth pressure was monitored by visual feedback from the pressure gauge by the patient to maintain the target intrathoracic pressures of 20 cmH₂O (MM 20) and 40 cmH₂O (MM 40). During preliminary studies, we determined that 15 s was the maximum duration CHF patients could maintain a Mueller maneuver without undue discomfort or oxyhemoglobin desaturation. All patients performed several practice Mueller maneuvers before actual data collection. Data were obtained during one breath hold and two Mueller maneuvers at the first target pressure, selected at random, followed by another breath hold and two Mueller maneuvers at the other target pressure. Breath holds and Mueller maneuvers were separated by 3-min rest periods.

Data analysis. Systolic Pes was determined by measuring Pes synchronously with the peak systolic blood pressure. Systolic Pes before respiratory maneuvers was used as the baseline value for subsequent interventions. As a measure of left ventricular afterload, systolic left ventricular transmural pressure was calculated as systolic blood pressure  systolic Pes (18). Beat-by-beat measurements were obtained during each breath hold, MM 20, and MM 40, and average values for each beat were determined for each individual. To determine the immediate response to these interventions, group mean values were calculated for each of the first five beats during the baseline control period before breath holds, MM 20, and MM 40. Each of the first five beats during the breath holds, MM 20, and MM 40 was then compared with the mean of the five baseline control beats by a one-way analysis of variance for repeated measures with a Dunnett's correction. To determine whether averaged values of variables for the first five and last five beats differed from the baseline control values and from each other within a given maneuver and whether responses during MM 40 and MM 20 differed from the apneic time control period of breath holds and from each other at the same time intervals, mean changes from baseline over the first and final five beats were compared among the breath holds, MM 20, and MM 40, using a two-way analysis of variance with Student-Newman-Keuls post hoc tests. A P value <0.05 was considered statistically significant. All data are means ± SD.

	RESULTS

Top Abstract Introduction Methods Results Discussion References

Characteristics of the patients. All eight CHF patients studied were men. Their characteristics and medications are shown in Table 1. They had severe left ventricular systolic dysfunction, as indicated by their markedly depressed left ventricular ejection fractions. All were on appropriate and optimal medical therapy for CHF.

Immediate responses to breath holds and Mueller maneuvers. Figure 1 shows beat-by-beat data over the first five beats under each condition. During the breath hold (Fig. 1, left), there was a small increase in systolic Pes, relative to baseline, reflecting the absence of inspiratory reductions in Pes. However, there were no significant changes in systolic left ventricular transmural pressure, systolic blood pressure, stroke volume index, or R-R interval. In contrast, both MM 20 (Fig. 1, middle) and MM 40 (Fig. 1, right) caused immediate reductions in systolic Pes and blood pressure. Because the drop in systolic Pes was greater than the fall in systolic blood pressure, systolic left ventricular transmural pressure increased immediately. Stroke volume index also decreased instantaneously at the onset of MM 20 and MM 40, but R-R interval remained constant.

View larger version (27K): [in this window] [in a new window]   Fig. 1.   Beat-by-beat data in patients with congestive heart failure during breath hold and Mueller maneuvers at target esophageal pressure (Pes) of 20 and 40 cmH2O (MM 20 and MM 40, respectively). Systolic Pes (Pessys) increased slightly during breath holds because of absence of negative inspiratory Pes swings. However, there were no changes in systolic left ventricular transmural pressure (LVPtmsys), systolic blood pressure (BPsys), stroke volume index (SVI), and R-R interval during the breath hold. In contrast, generation of negative intrathoracic pressure during MM 20 and MM 40 was associated with increases in LVPtmsys and reductions in BPsys and SVI on 1st and subsequent cardiac cycles. No change in R-R interval was observed. * P < 0.01 vs. mean of 5 baseline beats by Dunnett's test.

Time course of hemodynamic responses. Changes in systolic Pes, systolic and diastolic blood pressures, and systolic left ventricular transmural pressure over time are shown in Table 2. Absolute data for R-R interval, stroke volume, and cardiac indexes appear in Table 3. During the breath holds, none of systolic Pes, systolic left ventricular transmural pressure, systolic and diastolic blood pressures, stroke volume, and cardiac indexes changed significantly from baseline. However, there was a small but significant increase in R-R interval compared with baseline over the first five beats. R-R interval returned to the baseline level by the final five beats.

During the MM 20, patients achieved the target systolic Pes and maintained it throughout the maneuver. Whereas systolic left ventricular transmural pressure increased significantly over the first five beats, it drifted back toward the baseline level over the final five beats because of a parallel downward trend in systolic blood pressure. Both systolic and diastolic blood pressure decreased significantly over the first five beats and remained at this level over the final five beats. R-R interval initially decreased slightly, but significantly, but returned toward the baseline level by the end of the MM 20. Stroke volume and cardiac output decreased immediately and remained at these lower levels until the end of the MM 20. Most subjects did not initially achieve or sustain the target systolic Pes for the MM 40 for the full 15 s. Consequently, the average systolic Pes became significantly less negative by the final five beats. Whereas the immediate increase in systolic left ventricular transmural pressure was significant, this measure of left ventricular afterload fell below its baseline level by the final five beats, primarily because of a pronounced drop in systolic blood pressure. There were significant reductions in both systolic and diastolic blood pressures during the first five beats, and blood pressure fell further during the final five beats. The mean reductions in systolic and diastolic blood pressures over the final five beats of the MM 40 were 21% (from 121 to 96 mmHg) and 16% (from 70 to 59 mmHg), respectively. R-R interval did not change significantly during either the first or final five beats. Stroke volume decreased significantly during the first five beats and fell even further by the final five beats. Cardiac output also fell significantly initially and tended to fall further toward the end of the MM 40. Average reductions in stroke volume index and cardiac index over the final five beats of MM 40 were 33% (from 27 to 18 ml/m²) and 30% (from 2.0 to 1.4 l · min¹ · m²) below baseline, respectively. During all breath holds and Mueller maneuvers, oxyhemoglobin saturation remained >91%.

Responses to negative intrathoracic pressure during apnea compared with apnea alone. Comparisons among responses to breath hold, MM 20, and MM 40 appear in Figs. 2 and 3. By design, a wide separation of systolic Pes among breath holds, MM 20, and MM 40 was achieved during both the first and final five beats (Fig. 2). During the first five beats, systolic left ventricular transmural pressure increased progressively and significantly from breath hold to MM 20 to MM 40. The magnitude of this responses was proportional to the intensity of the Pes stimulus. However, over the final five beats, systolic left ventricular transmural pressure decreased toward the breath-hold level during MM 20 and became significantly lower than breath-hold values during MM 40. Systolic and diastolic blood pressures decreased progressively and significantly from breath hold to MM 20 to MM 40 during both the first and final five beats.

View larger version (23K): [in this window] [in a new window]   Fig. 2.   Plots of changes () in Pessys, BPsys, diastolic blood pressure (BPd) and LVPtmsys, relative to baseline, which was taken as 0, during 1st and final 5 cardiac cycles of breath holds (), MM 20 (), and MM 40 (). * P < 0.01 vs. breath hold;  P < 0.01 vs. breath hold and MM 20 by Student-Newman-Keuls tests.

View larger version (14K): [in this window] [in a new window]   Fig. 3.   Plots of changes in R-R interval, SVI, and cardiac index (CI) relative to baseline during 1st and final 5 cardiac cycles of breath holds (), MM 20 (), and MM 40 (). * P < 0.01 vs. breath hold;  P < 0.01 vs. breath hold and MM 20.

As illustrated in Fig. 3, R-R interval was significantly shorter during MM 20 and MM 40 than during breath hold over both the first and final five beats, such that the magnitude of this response was proportional to the intensity of the Pes stimulus. There were, however, no significant differences in R-R intervals during either the first or final five beats between the MM 20 and MM 40. Stroke volume index decreased progressively and significantly from breath hold to MM 20 to MM 40 during both the first and final five beats. Similar reductions in cardiac index from breath hold to MM 20 to MM 40 were observed. Decreases in both stroke volume and cardiac indexes by the end of MM 40 were similar in patients with ischemic and idiopathic dilated cardiomyopathy (10.1 ± 8.1 vs. 8.7 ± 3.0 ml/m², P = 0.72, and 0.61 ± 0.34 vs. 0.60 ± 0.25 l · min¹ · m², respectively, P = 0.95). In addition, there was no significant relationship between the degree of reduction in stroke volume at the end the MM 40 and left ventricular end diastolic diameter (r = 0.006, P = 0.98).

	DISCUSSION

Top Abstract Introduction Methods Results Discussion References

The objectives of the present study were to determine whether negative intrathoracic pressure generated during apnea by patients with impaired left ventricular systolic function caused significantly greater immediate reductions in stroke volume and blood pressure than did apnea alone and, if so, to determine whether there was a stimulus-response relationship between the negative intrathoracic pressure generated and the magnitude of these initial hemodynamic responses. Because we performed beat-by-beat analyses during these interventions, we were able to characterize, for the first time, the instantaneous hemodynamic responses to apnea and negative intrathoracic pressure in CHF patients and to describe changes in these responses over the time course of each maneuver. Mueller maneuvers caused immediate increases in systolic left ventricular transmural pressure and simultaneous reductions in blood pressure, stroke volume, and cardiac output. These changes were proportional to the magnitude of the negative intrathoracic pressure generated. Examination of the hemodynamic responses to breath holds allowed us to distinguish between responses to negative intrathoracic pressure generated during apnea and responses to apnea itself. Breath holds had no effect on these hemodynamic variables. Therefore, the abrupt reductions in stroke volume and cardiac output observed in these patients must be attributed to the generation of negative intrathoracic pressure and not to apnea alone.

The immediate response to MM 40 was a profound reduction in stroke volume and cardiac output. The simultaneous increase in left ventricular transmural pressure (i.e., afterload) probably played an important role in this initial hemodynamic compromise. However, increases in afterload cannot be held accountable for subsequent reductions in stroke volume and cardiac output, since left ventricular transmural pressure was clearly below baseline values during the last five beats of the MM 40. A reduction in preload is far more likely to be responsible for these later hemodynamic responses. Thus Mueller maneuvers exert adverse hemodynamic effects on patients with CHF through several mechanisms, and the intensity of these responses and their interactions vary over time. When Mueller maneuvers are sustained, the extent to which increased afterload contributes to these responses diminishes while the effect of reduced preload likely increases.

Scharf et al. (22) studied the effects of Mueller maneuvers in healthy subjects and in patients with coronary artery disease. A small (4%), but significant, decrease in left ventricular ejection fraction, derived by radionuclide angiography, was observed during Mueller maneuvers in patients with coronary artery disease but not in healthy subjects. Reductions in ejection fraction were attributed to increases in left ventricular afterload resulting from falls in intrathoracic pressure. These results suggested that patients with underlying coronary disease are more susceptible to the adverse effects of negative intrathoracic pressure than are subjects without cardiac disease. That study differed from the present experiment in several important respects. All patients had coronary artery disease, whereas five of our eight patients did not. The great majority of those patients did not have CHF, whereas all of ours did. In contrast was to our study, neither cardiac output nor beat-to-beat blood pressure was measured, breath holds were not performed as a time control for apnea, negative intrathoracic pressure stimulus magnitude-response characteristics were not assessed, and the time courses of the changes in physiological variables were not described.

The acute physiological consequences of upper airway obstruction in humans include negative intrathoracic pressure, apnea, hypoxia, and arousal from sleep (20, 31). In the setting of normal cardiac function, obstructive apneas consistently reduce stroke volume and cardiac output (25, 28, 29) and increase sympathetic nervous system activity (26). When functionally important coronary artery stenoses are present, generation of negative intrathoracic pressure can precipitate acute myocardial ischemia and hypokinesis, even in the absence of hypoxia (22, 24).

Because the three interventions in this experiment (i.e., breath hold, MM 20, and MM 40) had distinctly different effects on systolic Pes, we were able to dissociate the effects of negative intrathoracic pressure during apnea from apnea itself, without the complicating influences of hypoxia and/or arousals from sleep. Systolic and diastolic blood pressures decreased significantly during MM 20 and MM 40 but not during breath holds. These findings are in agreement with those of Morgan et al. (15) in healthy subjects, in whom Mueller maneuvers caused greater reductions in blood pressure than breath holds did. We attribute changes in systolic blood pressure in our subjects to changes in stroke volume. This is for several reasons. The abrupt fall in systolic blood pressure observed during the first cardiac cycle of the Mueller maneuver was associated with a simultaneous decrease in stroke volume. (Our documentation of reductions in extrathoracic carotid artery blood flow during these Mueller maneuvers, which paralleled reductions in aortic blood flow measured from the suprasternal notch, validates the latter as a reasonable estimate of changes in stroke volume.) During MM 40, both systolic and diastolic blood pressures fell further over time, in parallel with reductions in stroke volume. Reductions in heart rate or systemic vascular resistance cannot account for these changes, because heart rate either did not change or it increased, and efferent sympathetic vasoconstrictor tone has been demonstrated to increase during Mueller maneuvers (15).

The mechanism or mechanisms underlying the hemodynamic response to negative intrathoracic pressure have been the subject of debate. The sudden increase in left ventricular afterload could reduce stroke volume during obstructive apneas and Mueller maneuvers (3, 29, 35), as it does when aortic impedance is increased acutely (36). The failing myocardium is particularly susceptible to reductions in stroke volume in response to increases in left ventricular afterload (21). Therefore, the immediate reduction in stroke volume in our patients, evident within the first cardiac systole of the Mueller maneuver, can be attributed to the abrupt increase in left ventricular afterload, which was proportional to the magnitude of negative Pes generated. However, when negative intrathoracic pressure was sustained, left ventricular transmural pressure fell below baseline values. Therefore, other mechanisms must come to bear, then predominate, later in the time course of the Mueller maneuver.

One possibility, which we can neither confirm nor exclude, is that cardiac contractility fell over the course of the Mueller maneuver as a result of reduced diastolic blood pressure and coronary artery perfusion (9, 35). Other studies suggest that generation of negative intrathoracic pressure will decrease cardiac output by reducing left ventricular preload (24, 31). This could result from a leftward shift of the interventricular septum, from an increased impedance to right ventricular emptying due to increased right ventricular transmural pressure, or from a reduced left ventricular diastolic relaxation rate (24, 31, 35). However, other investigators have found either no change or an increase in left ventricular end-diastolic dimensions and pressures during Mueller maneuvers and obstructive apneas in subjects with normal ventricular function (4, 6, 22, 34), and cardiac output is less dependent on preload in CHF patients than it is in subjects with normal cardiac function. Despite these objections, our data point to a decrease in left ventricular preload as the principal mechanism responsible for reductions in stroke volume during the final seconds of the Mueller maneuver in these patients. Recent observations by Atherton et al. (1) provide new insight as to how this might occur. In a subset of patients with severe CHF and ventricular dilatation, left ventricular filling appeared to be limited by diastolic pericardial constraint (1). In such patients, generation of exaggerated negative intrathoracic pressure and subsequent distension of the right ventricle could lead to leftward shift of the interventricular septum and further restrict left ventricular filling. This would render CHF patients particularly susceptible to hemodynamic compromise over the course of the Mueller maneuver. Although this is an attractive hypothesis, it should be noted that these authors required a 5-min intervention (sustained lower body negative pressure) to provide evidence for this interaction in a subset of their patients (1). Because chest wall distortion precluded the reliable measurement of changes in left ventricular end-diastolic dimensions during Mueller maneuvers, we were unable to determine whether the much briefer interventions (15 s) applied in the present study had similar effects. However, our data over the last five cardiac cycles of the Mueller maneuver are consistent with this concept.

In summary, our findings indicate that hemodynamic responses to the generation of negative intrathoracic pressure in patients with left ventricular systolic dysfunction are complex and dynamic. The immediate response is an increase in left ventricular afterload and an abrupt fall in stroke volume and blood pressure. The magnitude of this response is proportional to the negative intrathoracic pressure generated. When negative intrathoracic pressure is maintained, the influence of afterload dissipates, and additional mechanisms, such as a drop in left ventricular preload due to ventricular interaction, are engaged, then predominate, further reducing stroke volume and systolic blood pressure.

The use of intrathoracic pressure (i.e., Pes) as a measure of pericardial pressure has certain limitations, but those would pertain chiefly to diastolic rather than to systolic events (32). For example, in an acutely dilated heart, the pericardium may reach its elastic limit in diastole, such that changes in Pes at this time may overestimate change in pericardial pressure (23). However, since the circumference of the heart decreases in systole and since we quantified Pes only during systole, the pericardium could not have been at its elastic limit. In addition, as demonstrated by Scharf et al. (23) in anesthetized dogs with beating hearts, changes in Pes during Mueller maneuvers are not significantly different from changes in pericardial pressure. Subsequently, Virolainen et al. (34, 35) validated the measurement of intrathoracic pressure for determination of pericardial and systolic left ventricular transmural pressure during Mueller maneuvers in humans. In any event, left ventricular afterload will increase regardless of any effects on pericardial pressure, because negative intrathoracic pressure will increase systolic transmural pressure in the thoracic aorta. Accordingly, changes in Pes, as applied in our experiments, provide a reasonable estimate of changes in pericardial and intrathoracic aortic surface pressure during systole.

The present data suggest that obstructive apneas can have clinically important adverse effects on cardiac performance, on coronary and systemic perfusion, and on disease progression, even when such patients are on optimum contemporary medical therapy for their heart failure. There were no significant differences between patients with ischemic and idiopathic dilated cardiomyopathy in the extent to which stroke volume and cardiac output were compromised by the generation of negative intrathoracic pressure, and there was no significant relationship between left ventricular volumes and the degree of hemodynamic compromise observed. These latter findings argue for the generalizability of our observations. The net impact of these interactions in these CHF patients was substantial. By the end of MM 40, cardiac index had fallen 30% (from 2.0 to 1.4 l · min¹ · m²), systolic blood pressure decreased from 121 to 96 mmHg, and diastolic blood pressure decreased from 70 to 59 mmHg on average. These results indicate that obstructive apneas can profoundly aggravate any preexisting hemodynamic impairment in patients with CHF, even in the absence of hypoxia.

We acknowledge two limitations that may prevent us from extrapolating directly from these data to the clinical scenario of obstructive sleep apnea in the setting of CHF. Although these two stimuli have similar effects on cardiac output and blood pressure in subjects with normal ventricular function (27, 31, 35), Mueller maneuvers differ from obstructive sleep apnea in that the negative intrathoracic pressure generated is sustained and constant rather than intermittent and progressively more negative. Thus the immediate, afterload-mediated responses observed over the first five cardiac cycles in the present experiments are likely to predominate early in the obstructive cycle. Second, our experimental protocol does not address the impact of confounding factors characteristic of obstructive apneas, such as hypoxia, and arousals from sleep or termination of apneas. Each of these factors may affect cardiac output and systemic blood pressure independently of changes in intrathoracic pressure. Therefore, further experiments, although difficult, should be performed during sleep in patients with CHF to determine the functional importance of these events.

In conclusion, our data indicate that, in patients with CHF, voluntary Mueller maneuvers reduce blood pressure, stroke volume, and cardiac output in proportion to the negative intrathoracic pressure developed. Several mechanisms, with different time constants, elicit these responses. Cardiac loading conditions change, such that there is an initial increase in afterload, followed by a decrease, owing to a fall in systolic blood pressure. Thus the extent to which increased afterload contributes to decreases in stroke volume diminishes over time. The progressive reduction in stroke volume as 40 cmH₂O was sustained must, therefore, have been due to a reduction in left ventricular preload, a decrease in contractility, or both. In the absence of negative intrathoracic pressure generation, brief breath holds cause little, if any, hemodynamic change. Intrathoracic pressures of up to 90 cmH₂O have been documented in patients with obstructive sleep apnea and CHF (13). Such patients are exposed to these repetitive intrathoracic pressure oscillations hundreds of times during the night, perhaps over several years. It is highly likely, but as yet unproved, that obstructive sleep apneas produce hemodynamic compromise similar to that observed during Mueller maneuvers in the present study or lead to progressive ventricular dysfunction (13).