Because heart failure is highly prevalent, performance of sleep studies on all patients with heart failure is not practical. However, it is difficult to predict which patients may develop periodic breathing during sleep [1]. Therefore, simple laboratory tests that could predict periodic breathing during sleep would be helpful screening tools.

Arterial PCO₂ has a dominant influence on breathing. A carefully performed study [2] done during sleep in normal humans showed that central apnea may be induced when the PaCO₂ is experimentally lowered by 1 to 3 mm Hg below the resting PaCO₂ while patients are awake. Therefore, a low PaCO₂ while awake may predispose to ventilatory instability and development of central sleep apnea.

To examine the predictive value of resting PaCO₂ while awake for central sleep apnea in heart failure, we studied patients with stable heart failure and no major comorbid conditions.

Methods

Top Methods Results Discussion Author & Article Info References

Ambulatory male patients with stable, medically treated left heart failure (left ventricular ejection fraction 45%) took part in this study. Details on these patients have been published elsewhere [1, 3]. Patients were clinically stable (symptoms or signs of heart failure had not changed in the preceding 4 weeks) and received standard therapy; no change had been made in cardiac medications in the preceding 4 weeks. Exclusion criteria were major comorbid disorders or use of morphine derivatives, benzodiazepines, or respiratory stimulants, such as theophylline and acetazolamide [1-3]. We studied only men because women are seldom referred to our center.

After patients spent an adaptation night in the sleep laboratory, polysomnography was performed by using standard techniques, as detailed elsewhere [1, 3, 4]. Apnea was defined as cessation of inspiratory airflow for 10 seconds or more. Obstructive apnea was defined as the absence of airflow in the presence of rib cage and abdominal excursions. Central apnea was defined as the absence of rib cage and abdominal excursions and absence of airflow [1, 3, 4]. Hypopnea was defined as a reduction of airflow lasting 10 seconds or more associated with a decrease of 4% or more in arterial oxyhemoglobin saturation or an arousal [5].

In the absence of measurements of esophageal pressure, however, differentiation of central apnea and hypopnea from obstructive events can be difficult. We classified hypopnea as obstructive if paradoxical thoracoabdominal excursions occurred or if the airflow decreased out of proportion to the reduction in the thoracoabdominal excursion. The apnea-hypopnea index was the number of episodes of apnea and hypopnea per hour.

Polysomnograms were scored in a blinded manner. Central sleep apnea was defined polysomnographically by the presence of 10 or more hourly episodes of apnea and hypopnea and 5 or more hourly episodes of central apnea. The number of hourly episodes of central disordered breathing had to be more than 50% of the overall total number of episodes of apnea and hypopnea. We defined absence of sleep apnea as fewer than 10 hourly episodes of apnea and hypopnea.

Arterial blood samples were obtained with the patient in a sitting position after he had rested for 15 minutes. To minimize pain, we used 2% lidocaine to anesthetize the skin where the radial artery was punctured. Afterward, by touching the skin with a sterile needle, we assured the patient that the procedure was painless. Our intent was to minimize changes in PaCO₂ caused by pain and anxiety. We performed pulmonary function tests, measured left ventricular ejection fraction, and did Holter monitoring as detailed elsewhere [1, 3, 6]. Patients were classified as eucapnic (PaCO₂ > 35 and < 44 mm Hg [n = 41]) or hypocapnic (PaCO₂ 35 mm Hg [n = 18]).

We used the Wilcoxon rank-sum test to assess significant differences between the two groups and chi-square analysis for proportions. Stepwise least-squares multiple regression analysis was used to determine the independent effects of certain variables on log transformation of the hourly rate of apnea and hypopnea. A two-sided P value less than 0.05 was considered statistically significant. Mean values ±SDs and percentages are reported as needed. We calculated 95% CIs by using t statistics. All calculations were done by using SAS software [7].

Results

Top Methods Results Discussion Author & Article Info References

In eucapnic and hypocapnic patients, the mean PaCO₂ and plasma concentrations of hydrogen and bicarbonate ions differed significantly, but demographic characteristics, results of pulmonary function tests, and left ventricular ejection fraction did not (Table 1).

Arterial PCO₂ ranged from 35.2 to 43.6 mm Hg in eucapnic patients (mean, 39 mm Hg [95% CI, 38 to 40 mm Hg]) and 23.0 to 35.0 mm Hg in hypocapnic patients (mean, 32.5 mm Hg [CI, 30.8 to 34.2 mm Hg]). Of the 18 hypocapnic patients, 14 (78% [CI, 52% to 93%]) had central sleep apnea; this prevalence was significantly higher than that in eucapnic patients (16 of 41 [39%; CI, 25% to 55%]; P = 0.01). Although the prevalence of subjective habitual snoring was lower and the prevalence of excessive daytime sleepiness was higher in the hypocapnic patients than in the eucapnic patients, the differences were not significant (Table 1). In eucapnic patients and hypocapnic patients, the use of vasodilators (93% and 89%), digoxin (73% and 67%), isosorbide dinitrate (41% and 39%), and diuretics (80% and 89%) did not significantly differ. However, significantly more hypocapnic patients were New York Heart Association class III and fewer were class I and class II compared with eucapnic patients (Table 1).

Hypocapnic patients had significantly more arousals and decreased sleep efficiency due to excessive periodic breathing (Table 2). The number of hourly episodes of apnea and hypopnea and central apnea was significantly greater in hypocapnic patients than in eucapnic patients (Table 2). Consistent with the difference in the number of hourly episodes of apnea and hypopnea, arterial oxyhemoglobin desaturation was more severe in hypocapnic patients than in eucapnic patients, although the differences were not statistically significant (Table 2). Hypocapnic patients had a significantly higher prevalence of ventricular irritability (Table 2). Ventricular tachycardia (defined as three premature ventricular depolarizations in a row) was 20 times more prevalent in hypocapnic patients than in eucapnic patients. The mean serum potassium level and concentrations of sodium and digoxin did not significantly differ between the two groups.

In a stepwise least-squares multiple regression analysis, we assessed the independent contribution of PaCO₂, hydrogen ion concentration ( 35 or >35 nmol/L), and New York Heart Association functional classes (III or I and II) with the apnea-hypopnea index as the dependent variable. The P values for PaCO₂ and hydrogen ion concentration were 0.014 and 0.034, respectively. When New York Heart Association functional classes were added, only PaCO₂ remained significant (P = 0.04 [for hydrogen ion concentration, P = 0.06]).

Discussion

Top Methods Results Discussion Author & Article Info References

Among 59 patients with stable heart failure who did not have other comorbid disorders, 18 (31%) were hypocapnic while awake. Only 4 of the 18 patients did not have central sleep apnea. Therefore, given the prevalence of central sleep apnea in study patients, 78% of patients with heart failure and low PaCO₂ had central sleep apnea.

As shown by our results, most hypocapnic patients will develop relatively severe periodic breathing during sleep, with an average of 36 hourly episodes of apnea and hypopnea (Table 2). Furthermore, in stepwise multiple regression analysis, PaCO₂ was associated with hourly episodes of apnea and hypopnea (although this may have been confounded by hydrogen ion concentrations). These episodes of apnea and hypopnea resulted in a moderate degree of arterial oxyhemoglobin desaturation, excessive arousals, and disrupted sleep (Table 2).

However, our results also show that in patients with stable heart failure, a low PaCO₂ while awake is not a prerequisite for development of central sleep apnea. Of 41 eucapnic patients, 16 (39%) had central sleep apnea. Because only 14 of the 30 patients with central sleep apnea were hypocapnic, the sensitivity [8] of a low PaCO₂ was 47%.

In the absence of systematic, large, longitudinal studies, we speculate that the degree of periodic breathing with associated arterial oxyhemoglobin desaturation and arousals, if left untreated, may adversely affect cardiac function and result in excessive illness and death [1, 9]. The results of our current study and two previous studies [10, 11] suggest that performing sleep studies on hypocapnic patients with heart failure may identify patients who require sleep apnea therapy.

Another important finding was the significantly greater number of hourly episodes of ventricular arrhythmias during sleep in hypocapnic patients with heart failure (Table 2). The prevalence of ventricular tachycardia was 20 times greater in the hypocapnic patients than in the eucapnic patients. A low PaCO₂ while awake may also indicate the need for Holter monitoring because detection and appropriate treatment of ventricular tachycardia may improve survival [12].

Although a previous report [13] suggested that hypocapnia may contribute to the development of arrhythmias in patients with coronary artery disease, our report describes the largest systematic study to date to show the relation between hypocapnia and ventricular tachycardia in patients with left heart failure. We could not determine whether hypocapnia was the cause of ventricular arrhythmias, but hypocapnia may be associated with changes in calcium binding and shifts of sodium, potassium, and hydrogen ions across cell membranes [14]. We observed no significant differences between eucapnic and hypocapnic patients in serum concentrations of sodium, potassium, and digoxin; PaO₂; and left ventricular ejection fraction, but serum hydrogen ion concentrations differed significantly (Table 1).

Several factors may result in periodic breathing during sleep [10, 11, 15, 16]. The observation that a low PaCO₂ is commonly associated with central sleep apnea in patients with stable heart failure is consistent with the physiologic notion of the apneic threshold. Accordingly, a low PaCO₂ promotes ventilatory instability, and the data obtained from normal persons while they are asleep show that lowering the PCO₂ by passive hyperventilation induces central sleep apnea [2].

Our study had some limitations. First, it was not designed to determine the mechanisms of the hypocapnia that occurred in 30% of study patients. However, hypocapnic patients were in the higher categories of the New York Heart Association classification and had a higher (although not a statistically significantly higher) prevalence of paroxysmal nocturnal dyspnea (Table 1). These patients may have had chronic pulmonary congestion that caused hyperventilation and dyspnea. However, patients in both groups were clinically stable, and the incidence of orthopnea and crackles, mean values for pulmonary function tests (including diffusing capacity for carbon dioxide), and left ventricular ejection fraction were similar in both groups. Second, because we studied only male patients and patients without comorbid disorders, our data may not be generalizable to all patients with left heart failure.

In summary, we found a high prevalence of central sleep apnea and ventricular tachycardia in hypocapnic patients with stable heart failure. We suggest that such patients be considered for sleep study and Holter monitoring because testing may guide appropriate therapy [3, 9, 12, 16-18].

Mr. Corbett: Phoenix International, 5642 Hamilton Avenue, Cincinnati, OH 45224.