Published ahead of print on January 18, 2005, doi:10.1164/rccm.200407-910OC
© 2005 American Thoracic Society doi: 10.1164/rccm.200407-910OC
Hypocapnia Is Not a Predictor of Central Sleep Apnea in Patients with CirrhosisPulmonary and Liver Services, Departments of Veterans Affairs Medical Center and Medicine, College of Medicine, University of Cincinnati; and Sleep Care Diagnostics, Cincinnati, Ohio Correspondence and requests for reprints should be addressed to Shahrokh Javaheri, M.D., Emeritus Professor of Medicine, Pulmonary Section (111F), VA Medical Center, 3200 Vine Street, Cincinnati, OH 45220. E-mail: shahrokh.javaheri{at}med.va.gov
During sleep, maintenance of rhythmic breathing is critically dependent on the level of PCO2, such that if the prevailing spontaneous PCO2 decreases below the apneic threshold, central sleep apnea (CSA) occurs. Several studies have shown that in patients with systolic heart failure (SHF), presence of a low, awake arterial PCO2 (PaCO2) increases the likelihood of developing CSA during sleep. We therefore sought to determine if a low PaCO2 is a predictor of CSA in patients with cirrhosis of the liver and with normal left ventricular systolic function. In 13 hypocapnic (PaCO2 < 36 mm Hg, mean = 33 mm Hg) patients with SHF and a mean left ventricular ejection fraction of 23%, the mean apnea–hypopnea index, was 28/hour. CSA accounted for most of the breathing disorders. In 10 hypocapnic (PaCO2 < 36 mm Hg, mean = 32 mm Hg) patients with cirrhosis and a normal left ventricular ejection fraction (60%), the mean apnea–hypopnea index was 2/hour. The maximum central apnea index was 0.2/hour. There were no significant differences in age, demographics, pulmonary function tests, PaO2, PaCO2, minute and alveolar ventilation, and ventilatory responses to CO2 between the two groups. We conclude that, in contrast to SHF, presence of hypocapnia does not predict CSA in cirrhosis.
Key Words: apneic threshold • PCO2 • periodic breathing During sleep, ventilation decreases and partial pressure of CO2 rises to a small degree, by 3 to 6 mm Hg. Normally, this elevated level of PCO2 is necessary to maintain rhythmic breathing, and if it is lowered below a certain level, referred to as the apneic threshold, ventilation ceases (1). Central sleep apnea (CSA) restores the CO2 tension to its previous level, and rhythmic breathing resumes. However, if for any reason (e.g., an arousal) ventilation increases and the prevailing PCO2 decreases below the apneic threshold, when sleep resumes, CSA recurs. In heart failure with systolic dysfunction, awake, steady-state hypocapnia predicts CSA (2–4). In our study of a relatively large number of patients (4), the presence of arterial blood hypocapnia, defined as PaCO2 of less than 36 mm Hg during wakefulness, had a predictive value of 78% for CSA. The assumption is that a low PaCO2 that is close to the apneic PCO2 increases the likelihood of developing CSA. Furthermore, the inhalation of a small amount of supplemental CO2, which increases PCO2, eliminates CSA, although a CO2-induced rise in ventilation could make it less likely to develop CSA (5–7). Hypocapnia commonly occurs in a variety of pathologic conditions, such as congestive heart failure, interstitial lung diseases, and cirrhosis. To our knowledge, the pathogenetic role of hypocapnia in the genesis of CSA has not been systematically tested in any other conditions except heart failure, idiopathic CSA, and high altitude (8). In the present prospective study, we sought to determine the importance of hypocapnia-induced CSA in subjects with cirrhosis and normal left ventricular systolic function.
We studied 23 patients: 13 with heart failure and 10 with cirrhosis of the liver. These were consecutive patients with cirrhosis and heart failure matched for age and PaCO2. The entry criterion was an awake PaCO2 of less than 36 mm Hg. This threshold was used because it highly predicts development of CSA in subjects with systolic heart failure (SHF) (4). Within 24 hours of polysomnographic studies, a detailed history was obtained and a physical examination was performed. Radionuclide ventriculography, pulmonary function tests, venous and arterial blood samples for determination of electrolytes, and arterial blood gases were also obtained. Minute and alveolar ventilation, CO2 production, and rebreathing hypercapnic ventilatory response were also measured. Details of these tests performed in our laboratory have been described elsewhere (9–15). This protocol was approved by the University of Cincinnati Institutional Review Board, and informed consent was obtained from all study subjects.
Description of Patients The diagnosis of cirrhosis was based on compatible clinical history, liver function tests, radiographic studies, and liver biopsy (six patients). None of the patients had asterexis. The etiologic mechanisms contributing to the cirrhosis were hepatitis B viral infection (one patient), hepatitis C viral infection (two patients), and alcoholism in eight patients. One patient had both hepatitis C and alcoholic liver disease. At the time of the study, all patients had been abstinent from alcohol for at least 3 months. Plasma ethanol was undetectable in all patients. Five had ascites and were on treatment with diuretics (furosemide, spironolactone, or combination of both). Two of the patients had a past history of hepatic encephalopathy but were compensated on lactulose. Three had endoscopic evidence of esophageal varices and were on ß-blockers (propanolol and nadolol). Patients with heart failure were recruited from an outpatient cardiology clinic at the Cincinnati VA Medical Center and were matched for age and PaCO2 with patients with cirrhosis. They had moderate to severe systolic dysfunction, with a range in left ventricular ejection fraction from 9 to 37%. The patients were ambulatory and stable and were involved in prospective studies to determine the prevalence, mechanisms, and treatment of CSA in heart failure (9). Thirteen patients had a PaCO2 of less than 36 mm Hg and were enrolled into this study. Nine were using an angiotensin-converting enzyme inhibitor, 10 were using furosemide, eight were using digoxin, one was using hydralazine, and one was using a ß-blocker. In this study, only male subjects were recruited, as female subjects are rarely referred to this center.
Statistical Analysis
In comparing subjects with SHF with those with cirrhosis, there were no significant differences in demographics, hemoglobin, hematocrit, electrolytes, renal function tests, PaO2 and PaCO2, and pulmonary function tests (Tables 1 and 2). As expected, left and right ventricular ejection fractions were significantly lower in patients with heart failure than in patients with cirrhosis. Patients with cirrhosis had significantly more impaired liver function tests than patients with heart failure (albumin 3.4 ± 0.6 vs. 4.1 ± 1.4 g/dl, bilirubin 2.6 ± 2.2 vs. 0.9 ± 0.4 mg/dl, and alanine transaminase 56 ± 18 vs. 24 ± 13 U/L, respectively).
Table 3 depicts sleep characteristics, sleep-disordered breathing events, and oxyhemoglobin saturation during sleep in the two groups. Total sleep time and distribution of sleep stages did not differ significantly between patients with heart failure and those with cirrhosis. In patients with heart failure, the mean apnea–hypopnea index was 28/hour. Central apneas accounted for most of the disordered breathing events and the mean CAI was 19/hour. In patients with cirrhosis, the mean apnea–hypopnea index was 2/hour and the mean CAI was 0.05/hour.
The individual values for arterial PCO2 and disordered breathing events in patients with heart failure and cirrhosis are depicted in Table 4. CSA was highly prevalent in patients with heart failure, whereas none of the patients with cirrhosis had CSA. The range of PaCO2 varied from 22.6 to 35.7 mm Hg in patients with heart failure and 26.8 to 35.8 mm Hg in patients with cirrhosis (Table 4). There were no significant correlations between PaCO2 and apnea–hypopnea index (n = 10, r = 0.39, p = 0.3) or CAI (n = 10, r = 0.05, p = 0.9) in patients with cirrhosis. Similarly, there were no correlations between PaCO2 and apnea–hypopnea index (n = 13, r = –0.00, p = 0.9) or PaCO2 and CAI (n = 10, r = 0.04, p = 0.9).
There were no significant differences in minute ventilation, alveolar ventilation, the slopes of hypercapnic ventilatory responses to CO2, and when hypercapnic ventilatory response to CO2 was normalized for body surface area, FVC, and maximum voluntary ventilation (Table 5). The patients with heart failure had a respiratory rate that was higher than that in patients with cirrhosis (p = 0.06; Table 5).
We studied 10 consecutive hypocapnic patients with cirrhosis who were free from SHF as evidenced by history and normal left ventricular ejection fractions. We compared their demographics and polysomnograms with 13 hypocapnic patients with SHF. Both groups were matched for age and PaCO2. In the 13 hypocapnic patients with SHF, the mean CAI was 19/hour. Seven of the 13 patients (54%) had periodic breathing with a CAI of 7 or greater/hour. In contrast, none of the patients with cirrhosis had a CAI of more than 0.2/hour. Therefore, in SHF, hypocapnia highly predicts CSA. In contrast, in patients with cirrhosis and a similar degree of hypocapnia, surprisingly, CSA is absent. If a low, awake, steady-state PaCO2 is so critical in the genesis of CSA in patients with heart failure (2–4), why is it that hypocapnic patients with cirrhosis do not develop CSA? One factor that has been shown to increase the likelihood of periodic breathing is increased hypercapnic ventilatory response to CO2 (17). However, there were no significant differences in hypercapnic ventilatory responses between the two groups in this study (Table 5), although we emphasize that the number of patients who had a CO2 response was small. One important difference between the two groups of patients in this study, however, was the presence of severe impairment in left ventricular function in subjects with heart failure but not in subjects with cirrhosis (mean left ventricular ejection fraction: 23 vs. 60%, respectively). The mechanisms mediating periodic breathing in SHF are complex (17–21). However, one hemodynamic factor underlying periodic breathing relates to an increased arterial circulation time because of pathophysiologic processes in SHF. Increased arterial circulation time results in delay of transfer of information from lung to chemoreceptors and converts a negative feedback system to a positive one, destabilizing breathing (17–21). Because patients with cirrhosis and normal left ventricular ejection fraction do not have altered hemodynamic features of congestive heart failure, one possibility is that the hemodynamic impairment of SHF is a prerequisite for development of periodic breathing in SHF. The experimental importance of increased arterial circulation time, as a cause of periodic breathing, was demonstrated in canine experiments of Guyton and associates (22). However, we (17) and others (23) have questioned the importance of increased arterial circulation time, because in Guyton and coworkers' experiments (22), the circulation time was prolonged considerably and only a third of animals developed periodic breathing. In response to our suggestion (17), Cherniack (19) has pointed out that administration of anesthesia, by decreasing the gain of the chemoreceptors, would have decreased the likelihood of development of periodic breathing in canine experiments (22). We, therefore, might have underestimated the importance of prolonged circulation time in the pathogenesis of CSA, and the results of the study are consistent with that notion.
Another possibility that could account for the findings of the present study may relate to differences in the mechanisms mediating hypocapnia in patients with heart failure and cirrhosis. While our study was in progress, Nakayama and colleagues (24) reported that apneic threshold is a dynamic value. Depending on the mechanisms of hyperventilation, CSA may or may not occur. This has to do with a critical value, the difference between two PCO2 set points, the prevailing PCO2-apneic threshold PCO2 (
The mechanisms of hyperventilation in congestive heart failure and cirrhosis are not clear and could be multifactorial (26). The most commonly quoted mechanism of hyperventilation in heart failure is an increase in pulmonary capillary pressure and J receptor stimulation. In cirrhosis, increased ammonia and progesterone, two respiratory stimulants (26), are believed to contribute to hyperventilation. In this regard, it is noted that there was an almost significant (p = 0.06) difference between respiratory rates of the two groups, with patients with SHF having a relative tachypnea compared with patients with cirrhosis (Table 5). The difference in mechanisms mediating hyperventilation may have an important bearing on the likelihood of developing central apnea during sleep. In the study of Nakayama and colleagues (24), conditions of nonhypoxic chemoreceptor stimulation, specifically almitrine administration and metabolic acidosis, increased The absence of central apnea in hypocapnic patients with cirrhosis reported in this study is consistent with an observation in interstitial lung disorders. In 11 subjects with interstitial lung disease (27), three subjects had a PaCO2 of less than 35 mm Hg and none had CSA.
In summary, the results of this study show that hypocapnia does not predict CSA in patients with cirrhosis, an observation that contrasts to that in patients with heart failure. Future studies similar to those in heart failure (25) are necessary to determine if
The authors thank Candice A. Brown for technical assistance, Debra Patton for assistance in patient recruitment, Faye A. Jones for secretarial assistance, and Tim Tanner for performing sleep studies.
Supported, in part, by merit review grants from the Department of Veterans Affairs. This article has an online supplement, which is accessible from this issue's table of contents at www.atsjournals.org Conflict of Interest Statement: S.J. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; K.F.A. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; K.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript; C.L.M. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form July 14, 2004; accepted in final form January 9, 2005
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
2 analysis was used for proportions. A regression analysis was performed between steady-state, awake arterial PCO2 and apnea–hypopnea index and central apnea index (CAI). A two-sided p value of less than 0.05 was considered to indicate statistical significance. Mean values ± SD and percentages are reported as needed. All the calculations were done with In Stat software, version 2.03 (Graph Pad, San Diego, CA). 
PCO2), and the increased chemosensitivity to hypocapnia below the eucapnic PCO2. The smaller the 
