Cheyne and Stokes were the first to observe periodic breathing in patients with heart failure (Cheyne-Stokes respiration). In a subsequent systematic study, Harrison and colleagues [2] reported that periodic breathing characterized by repeated episodes of apnea and hypopnea during sleep occurred in patients with congestive heart failure. Since this early observation, several investigators have used standard polysomnography to study periodic breathing during sleep in patients with congestive heart failure [3-7]. The differences in prevalence rates in these studies (36% to 100%) might have been caused by several factors, including the few patients studied (4 to 11), the varying inclusion of patients with risk factors predisposing them to sleep apnea (such as loud snoring, witnessed apnea, and obesity), the varying severity of heart failure and systolic dysfunction, the inclusion of patients with unstable (as opposed to stable, maximally treated) congestive heart failure, and the presence of other factors that may influence periodic breathing. These important factors, which were not considered in most of previous studies, could considerably affect the prevalence and the severity of sleep apnea.

We determined the prevalence and effect of sleep-disordered breathing in a relatively large group of clinically well-defined patients with stable, optimally treated congestive heart failure. We also determined the predictors of sleep-disordered breathing in these patients.

Methods

Top Methods Results Discussion Author & Article Info References

Entry Criteria

Forty-two ambulatory patients with stable congestive heart failure (no change in signs or symptoms of congestive heart failure and no change in medications for at least 4 weeks before polysomnography) and systolic dysfunction (left ventricular ejection fraction 45%) participated in the study. Patients were recruited from the cardiology and medical clinics of the Department of Veterans Affairs Medical Center, Cincinnati, Ohio. The cardiologist coinvestigator evaluated all patients to confirm that their condition was stable and that they were receiving optimal therapy, which included digoxin (26 patients), diuretics (38 patients), angiotensin-converting enzyme inhibitors (37 patients), or hydralazine (2 patients).

At the time of recruitment, no information was sought about symptoms or risk factors for sleep apnea. The following were the exclusion criteria: unstable angina; unstable congestive heart failure; acute pulmonary edema; congenital heart disease; primary valvular heart disease; use of benzodiazepines or theophylline; intrinsic pulmonary diseases, including interstitial lung disease, moderate to severe chronic obstructive lung defect (percentage of the ratio of the predicted forced expiratory volume in 1 second and forced vital capacity < 68%); intrinsic renal and liver disorders; untreated hypothyroidism; and kyphoscoliosis. For uniformity, we studied only male patients (female patients are rarely referred to this center). Only 6 of the 48 patients who met the entry criteria and were asked to participate in the study refused. The main reasons for refusal were an unwillingness to stay in the hospital or an unwillingness to travel to the hospital because of distance. The study was approved by the Research and Development Committee of the Veterans Affairs Medical Center, Cincinnati, Ohio, and the Institutional Review Board at the University of Cincinnati College of Medicine.

Baseline Studies

After giving written informed consent, the patients were hospitalized for 2 consecutive nights. On the first day, a detailed history was obtained and physical examination and screening tests were done, including complete blood count; tests for serum electrolytes, digoxin, thyroid, and renal function; radionuclide ventriculography; determinations of arterial blood gases and pH; and pulmonary function tests. Strict criteria described elsewhere [8, 9] were used for doing pulmonary function tests and for obtaining arterial blood samples and their measurements. Because some studies [10] have suggested that a low baseline PaCO₂ is a risk factor for periodic breathing, skin over the radial artery was anesthetized with 2% lidocaine to minimize pain (which could potentially induce hyperventilation during arterial blood sampling). While the patient was sitting, arterial blood was collected anaerobically in heparinized syringes during several breath cycles. Duplicate determinations of arterial blood gases and pH were immediately made with appropriate electrodes [9].

Polysomnography

On the first night, patients were taken to the sleep laboratory. Surface electrodes were attached, but no recording was obtained. This adaptation night was used to minimize the first-night effect of sleeping in the laboratory. On the next night, polysomnography was done using standard techniques described previously [11-13]. For staging sleep, we recorded electroencephalograms (two channels), chin electromyograms (one channel), and electro-oculograms (two channels). Thoracoabdominal excursions were measured qualitatively by respiratory inductance plethysmography (Respitrace; Ambulatory Monitoring, Inc., Ardsley, New York) or by pneumatic respiration transducers (Grass Instrument Company, Quincy, Massachusetts) placed over the rib cage and abdomen. Airflow was qualitatively monitored using an oral-nasal thermocouple (Model TCT1R; Grass Instrument Company). Arterial oxyhemoglobin saturation was recorded using an ear oximeter (Biox IIA; BT, Inc., Boulder, Colorado). These variables were recorded on a multichannel polygraph (Model 78D; Grass Instrument Company). An apnea was defined as cessation of inspiratory airflow lasting 10 seconds or longer. An obstructive apnea was defined as the absence of airflow in the presence of rib cage and abdominal excursions. A central apnea was defined as the absence of airflow and of rib cage and abdominal excursions [11-13]. However, a central apnea may not be easily distinguished from an obstructive apnea if esophageal pressure is not measured. Hypopnea was defined as a reduction of airflow lasting 10 seconds or more that was associated with at least a 4% decrease in arterial oxyhemoglobin saturation or an arousal. An arousal was defined as the appearance of waves on an electroencephalogram that were at least 3 seconds in duration [14]. The number of episodes of apnea and hypopnea per hour is referred to as the apnea-hypopnea index. Scoring of polysomnograms was blinded.

The prevalence of sleep-disordered breathing in patients with congestive heart failure was determined using an apnea-hypopnea index of more than 20 episodes per hour. In a retrospective study of patients with the sleep apnea syndrome [15], an index of more than 20 apneas per hour was associated with excess mortality. Because this study was done before hypopnea was recognized, the investigators did not include it in their calculation of the index. Lower thresholds (for example, an apnea-hypopnea index of 10 episodes per hour) have been used in other studies; however, the clinical importance, particularly of low cutoff points, has not been adequately determined.

Other Studies

Holter monitoring was done during polysomnography. Three electrocardiographic channels (leads V₁, V₃, and V₅) were recorded using a Laser SxP Holter monitor system (Marquett Electronics Inc., Milwaukee, Wisconsin). The tapes were analyzed by computer and were manually overread by the cardiologist coinvestigator.

Using standard techniques [16], we calculated right and left ventricular ejection fractions from gated first-pass and multigated radionuclide ventriculograms, respectively [16].

Statistical Analysis

We used the Wilcoxon rank-sum test to assess the significance of differences between the two groups because the common variance assumption required by the t-test was not appropriate for many of the measurements. A P value of less than 0.05 was considered significant. We calculated 95% CIs using the approximate degrees of freedom for the t-statistic [17].

The relations between certain pathophysiologically important variables and the apnea-hypopnea index were examined by regression analysis and stepwise multiple regression analysis. Calculations were done using SAS software [17].

Results

Top Methods Results Discussion Author & Article Info References

The apnea-hypopnea index varied from 0.3 to 82.2 episodes per hour. The frequency histogram of the index is shown in Figure 1. In 23 patients (group I), the apnea-hypopnea indexes varied from 0.3 to 13.4 episodes per hour (mean ±SD, 4.4 ±4 episodes per hour [CI, 2.7 to 6.0 episodes per hour; median, 3.2 episodes per hour]). In the 19 patients in group II (45%), the apnea-hypopnea index varied from 26.5 to 82.2 episodes per hour (mean, 44 ±13 episodes per hour [CI, 37.6 to 50.6 episodes per hour; median, 40.4 episodes per hour].

View larger version (13K): [in this window] [in a new window]   Figure 1. Frequency distribution of the apnea-hypopnea index in 10-unit intervals in 42 patients with stable, optimally treated congestive heart failure.

The two groups did not differ significantly in demographic and historical data (Table 1). Congestive heart failure was caused by ischemic cardiomyopathy (16 patients in group I and 13 patients in group II), idiopathic cardiomyopathy (5 patients in group I and 6 patients in group II), and alcohol-related cardiomyopathy (2 patients in group I).

The mean values for selected laboratory test results for both groups are shown in Table 2. The respective sleep architecture values did not differ significantly. The mean value of the arousal index (the number of arousals per hour) was significantly greater in group II than in group I (Table 3).

In group II, the mean apnea-hypopnea index was high (44 ±13 episodes per hour [CI, 38 to 51 episodes per hour; Figure 2]). The obstructive apnea-hypopnea index (10 ±13 episodes per hour [CI, 4 to 16 episodes per hour]) accounted for only a few of these episodes; in contrast, more than 50% of the episodes were classified as central apneas (Figure 2). Approximately half of the disordered breathing episodes were associated with arousals (Figure 2).

View larger version (24K): [in this window] [in a new window]   Figure 2. Sleep-disordered breathing episodes and arterial oxyhemoglobin saturation in the two groups. AH I = apnea-hypopnea index; CA I = central apnea index; OAH I = obstructive apnea-hypopnea index; Ar I DB = arousal index associated with disordered breathing; BASE SAT = baseline arterial oxyhemoglobin saturation in supine position before sleep onset; LOW SAT = lowest arterial oxyhemoglobin saturation during sleep; SAT < 90%, min = time in sleep (in minutes) when arterial oxyhemoblobin saturation was less than 90%; SAT < 90%, TST = percentage of total sleep time when arterial oxyhemoglobin saturation was less than 90%. Values are expressed as mean ±SD.

Blood gas values did not differ significantly while the patients were awake: PaO₂, 79 ±13 mm Hg in group I compared with 81 ±12 mm Hg in group II; PaCO₂, 39 ±5 mm Hg compared with 37 ±7 mm Hg; and (H⁺), 37 ±2 mmol/L compared with 36 ±3 nmol/L. However, because of sleep-disordered breathing episodes, duration and severity of arterial oxyhemoglobin desaturation were significantly greater in group II than in group I: duration, 7 ±13 minutes (CI, 1 to 13 minutes; median, 0.4 minutes) in group I compared with 61 ±63 minutes (CI, 30 to 92 minutes; median, 42 minutes) in group II; severity, 87% ±4% (CI, 85% to 89%; median, 87%) in group I compared with 74% ±13% (CI, 68% to 80%; median, 77%) in group II (Figure 2).

Ventricular ectopy was more prevalent and left ventricular ejection fraction was lower in group II (Table 4).

Correlations

We determined regression coefficients for some potentially related variables and used disordered breathing episodes as the dependent variables. Age was not significantly correlated with any episodes of disordered breathing, including the apnea-hypopnea index (b = –0.01;CI, –0.081 to 0.79; n = 42) (b is the regression coefficient). Body mass index was significantly and positively correlated with the obstructive apnea-hypopnea index (b = 0.77 episodes per hour per unit of body mass index; CI, 0.38 to 1.16 episodes per hour) but not with the central apnea-hypopnea index (b = –0.52;CI, –1.29 to 0.25). These correlations suggested that our recognition and classification of central and obstructive disordered breathing episodes (see Methods section) were probably correct. Left ventricular ejection fraction was significantly and negatively correlated with the apnea-hypopnea index (b = –0.79 episodes per hour per unit of left ventricular ejection fraction; CI, –1.47 to –0.11).Right ventricular ejection fraction was not significantly correlated with any of the disordered breathing episodes.

In a stepwise multiple regression analysis in which the apnea-hypopnea index was used as the dependent variable and age, body mass index, left ventricular ejection fraction, PaCO₂, and PaO₂ were used as independent variables, only left ventricular ejection fraction was a significant factor.

In another stepwise multiple regression analysis, left ventricular ejection fraction, PaO₂ PaCO₂, the lowest arterial oxyhemoglobin saturation, the apnea-hypopnea index, the central apnea index, and the serum potassium level were used as independent candidates for predicting ventricular arrhythmias. The central apnea index was the only significant independent variable for the number of premature ventricular contractions (b =0.52 contractions per hour; CI, 0.48 to 0.56 contractions per hour), couplets (b = 0.76 couplets per hour; CI, 0.32 to 1.20 couplets per hour), and ventricular tachycardia (b = 4.8 episodes per hour; CI, 1.97 to 7.65 episodes per hour).

Discussion

Top Methods Results Discussion Author & Article Info References

We show that 45% of the patients with stable and optimally treated congestive heart failure had an apnea-hypopnea index of more than 26 episodes per hour (mean, 44 episodes per hour). The disordered breathing episodes were associated with excessive arousals and severe arterial oxyhemoglobin desaturation Table 3 and Figure 2. This degree of occult sleep-disordered breathing is probably associated with excess morbidity and possibly mortality in patients whose heart function is already compromised.

Our study differs from previous studies in several important ways. We approached 48 patients with well-defined criteria (stability of congestive heart failure, optimally treated congestive heart failure, absence of other medical disorders, use of drugs that could affect breathing, and so forth) and were able to recruit 42. We believe that this degree of recruitment minimized bias and therefore makes our results applicable to many male patients with stable congestive heart failure.

Furthermore, although we studied patients with stable congestive heart failure and systolic dysfunction (the latter mild to severe in degree), we found a surprisingly high prevalence of clinically unsuspected sleep-disordered breathing. According to our results, the latter may be caused, in part, by the high prevalence of central apneas. Finally, because we studied a relatively large number of patients and used several tests, including measuring left ventricular ejection fraction and monitoring for ventricular arrhythmias, we could determine some predictors and consequences of sleep apneas and hypopneas.

Predictors of periodic breathing in congestive heart failure have not been systematically studied and are unknown. Limited data suggest that periodic breathing is more prevalent during decompensation of congestive heart failure and improves with treatment [2, 18]. One of our objectives was to determine potential predictors of periodic breathing, if any, in patients with stable, optimally treated congestive heart failure.

One possible predictor of sleep-disordered breathing is loud snoring. However, the prevalence of habitual snoring was similar between the two groups (Table 1). This was not surprising because most episodes of periodic breathing noted in our patients with congestive heart failure were classified as central rather than obstructive apneas.

We had hypothesized that older patients with congestive heart failure may have a higher prevalence of sleep-disordered breathing than the younger patients, but mean values for various ages were similar between the two groups; more than 50% of the patients in group II were younger than 65 years. Furthermore, in both regression analysis and multiple regression analysis, age was not significantly correlated with any of the sleep-disordered breathing episodes.

Although Young and associates [19] showed that the prevalence of sleep-disordered breathing did not change significantly in patients aged 30 to 60 years, studies by Ancoli-Israel [20] showed that in three randomly selected groups of patients older than 65 years, 4% of the 427 independently living patients, 11% of the 235 patients living in nursing homes, and 14% of the 300 patients in medical wards had an apnea index of 20 or more episodes per hour. It was suggested that increased prevalence in patients in medical wards (14% of the patients) might be caused by the high prevalence of congestive heart failure. In our study, however, 45% of ambulatory patients with chronic stable congestive heart failure had an apnea-hypopnea index of 26.5 or more episodes per hour (mean, 44 episodes per hour), which was associated with severe arterial oxyhemoglobin desaturation. We emphasize, however, that a prospective study larger than ours is necessary to accurately estimate the prevalence of periodic breathing in patients with stable congestive heart failure.

Our results indicate that the severity of left ventricular systolic dysfunction was an important risk factor in predicting periodic breathing during sleep in patients with stable congestive heart failure. The left ventricular ejection fraction was significantly lower in group II than in group I and was significantly correlated with the apnea-hypopnea index.

Obesity is recognized as an important risk factor for the obstructive sleep apnea-hypopnea syndrome [19]. Interestingly, our study shows that body mass index correlated only with the obstructive apnea-hypopnea index, indicating that obesity remains a risk factor for sleep-disordered breathing in patients with congestive heart failure and systolic dysfunction.

Sleep-disordered breathing and associated hypoxemia (and hypercapnia) may adversely affect sleep architecture and cardiac function. In our study, patients in group II had excessive episodes of arousal that occurred approximately 24 times per hour and were directly attributable to disordered breathing (Figure 2). Excessive arousals may cause daytime fatigue, irritability, exhaustion, inability to concentrate, and sleepiness [21]. In our study, the prevalence of subjective excessive daytime sleepiness was higher in patients with sleep disordered breathing, but the difference was not significant (Table 1). However, our study was not designed to quantitatively measure daytime sleepiness [21].

Disordered breathing and nocturnal arterial oxyhemoglobin desaturation could adversely affect cardiac function by various mechanisms [22-28]. We found that left ventricular systolic function was significantly more impaired in patients with congestive heart failure and sleep-disordered breathing (group II) than in those without disordered breathing (group I) (Table 4). As also noted earlier, regression analysis showed that the left ventricular ejection fraction was the most important risk factor for periodic breathing. We cannot determine from these data whether sleep-disordered breathing exacerbated left ventricular dysfunction or whether patients with more severe left ventricular dysfunction had excessive sleep-disordered breathing. However, we speculate that interaction between sleep-disordered breathing and left ventricular dysfunction could result in a vicious cycle.

Another important finding in our study was the increased prevalence of ventricular arrhythmias in patients with sleep-disordered breathing and congestive heart failure (Table 4). In the multiple regression analysis, the left ventricular ejection fraction, serum potassium level, and arterial oxyhemoglobin desaturation were not significantly correlated with ventricular arrhythmias during sleep. Interestingly, we found central sleep apnea to be the main predictor of ventricular arrhythmias during sleep, a provocative finding that needs confirmation.

The mechanisms underlying ventricular arrhythmias in patients with congestive heart failure remain poorly understood [29, 30]. It has been suggested that activation of the sympathetic nervous system mediates the increase in the frequency of ventricular extrasystoles noted in sleep in some patients [31]. In sleep apnea, this activation may be linked to episodes of hypoxemia and hypercapnia [25] or arousals (the latter is ill-defined by electroencephalography and therefore cannot be accurately quantified). Alternatively, Massumi and Nutter [32] suggested that variation in arterial and cerebrospinal fluid PCO₂ affecting cardioregulatory centers may be a potential mechanism.

The long-term effects of sleep-disordered breathing on congestive heart failure are unknown. Because of disordered breathing, patients in our study with congestive heart failure had considerable nocturnal hypoxemia Figure 2, despite normal PaO₂ during wakefulness. Nocturnal hypoxemia of this degree may result in excess mortality, as has been shown in patients with chronic obstructive lung disease [1]. Furthermore, retrospective studies [15] of patients with the classic sleep apnea syndrome have suggested that an apnea index of more than 20 episodes per hour is associated with excess mortality. Longitudinal studies with appropriate control groups are needed to determine the potential contribution of disordered breathing to death in patients with congestive heart failure.

Mechanisms of periodic breathing have been reviewed elsewhere [32-35] and are beyond the scope of this article. However, we emphasize that in an overwhelming majority of patients with the sleep apnea-hypopnea syndrome, sleep-disordered breathing episodes were predominantly caused by upper airway occlusion (obstructive apneas). This contrasts with the prevalence of central sleep apnea found in most of our patients with congestive heart failure [6, 10, 12, 36]. Among many other factors [32-35], a prolonged circulation time (caused by pulmonary congestion and decreased cardiac output) could be an important factor contributing to periodic breathing (repetitive episodes of central apnea and hypopnea) in patients with congestive heart failure.

In summary, we showed that 45% of patients with stable, optimally treated congestive heart failure associated with systolic dysfunction experience a moderate to severe degree of sleep-disordered breathing. These episodes were associated with prolonged and severe nocturnal hypoxemia. The only risk factor that was associated with the apnea-hypopnea index was the severity of left ventricular dysfunction (obesity was only associated with the obstructive apnea-hypopnea index). Future studies emphasizing mortality rates are needed to evaluate the effects of sleep-disordered breathing on daytime sleepiness, neuropsychiatric function, and the natural history of congestive heart failure. Meanwhile, physicians caring for patients with congestive heart failure need to be aware of the magnitude and severity of sleep-disordered breathing, particularly in patients with more severe left ventricular systolic dysfunction. Furthermore, because we also found that central apneas were strongly correlated with ventricular arrhythmias, patients with stable congestive heart failure who are found to have excess arrhythmias on Holter monitoring probably should be evaluated for sleep-disordered breathing and appropriately treated [5, 12, 36] if necessary.