Cheyne-Stokes Respiration in Stroke
http://www.100md.com
美国呼吸和危急护理医学 2005年第5期
Sleep Research Laboratory of the Toronto Rehabilitation Institute, Centre for Sleep Medicine and Circadian Biology of the University of Toronto, Toronto, Ontario, Canada
ABSTRACT
Background: Central sleep apnea (CSA) and Cheyne-Stokes respiration have been reported in association with stroke, but their pathophysiologic correlates have not been well described. Objective: To test the hypotheses that (1) CSA in patients with stroke is associated with nocturnal hypocapnia and (2) in those stroke patients with CSA and with left ventricular (LV) systolic dysfunction, periodic breathing (PB) will have a Cheyne-Stokes respiration pattern in which cycle duration is greater than in those without LV systolic dysfunction. Methods: We prospectively performed polysomnography and echocardiography in 93 patients with stroke. CSA was defined as central apneas and hypopneas occurring at a rate of 10 or more per hour of sleep. In patients with CSA, we compared PB cycle duration between those with normal and impaired LV systolic function (LV ejection fraction [LVEF] > 40% and 40%, respectively). Results: CSA was found in 19% of subjects who had lower nocturnal transcutaneous PCO2 (39.3 ± 0.9 vs. 42.8 ± 0.8 mm Hg, p = 0.015) and a higher prevalence of LVEF of 40% or less (22 vs. 5%, p = 0.043) than stroke patients without CSA. There was no significant difference in stroke location or type between the two groups. In patients with CSA, those with LVEF of 40% or less had a longer PB cycle than those with an LVEF of more than 40% (66.6 ± 5.6 vs. 46.6 ± 2.9 seconds, p = 0.006), but had no symptoms of heart failure. Conclusion: In patients with stroke, CSA is associated with hypocapnia and occult LV systolic dysfunction but is not related to the location or type of stroke. The presence of LV systolic dysfunction is associated with a Cheyne-Stokes pattern of hyperpnea.
Key Words: Cheyne-Stokes respiration sleep apnea stroke
Although central sleep apnea (CSA) has been reported in association with stroke (1, 2), its pathogenesis in this setting remains unclear. Central apneas occur when there is a temporary withdrawal of central respiratory drive causing cessation of respiratory muscle activity and airflow (3). In subjects without stroke, common predisposing factors for CSA are enhanced central and peripheral chemosensitivity (4eC6), hypocapnia, and the presence of left ventricular (LV) systolic dysfunction (7, 8). CSA in patients with both congestive heart failure (CHF) and with normal LV function (idiopathic CSA) are triggered and propagated by hyperventilation and hypocapnia, causing PCO2 to fall below the apneic threshold, precipitating a central apnea (3eC6). Similar factors may be at play in the pathogenesis of CSA in subjects with stroke. Several articles in the 1950s and 1960s described hyperventilation, respiratory alkalosis, and increased respiratory sensitivity to CO2 (9eC11) in association with recurrent central apnea observed during wakefulness. It was suggested that increased sensitivity to CO2 after a stroke resulted from bilateral supramedullary brain lesions that disinhibit neural stimulatory input into the brainstem respiratory center (neurogenic CO2 hypersensitivity) (9). However, subsequent reports have failed to confirm a relationship between the site and size of a stroke and the presence of central apneas (12eC15). Moreover, because these central apneas were studied only during wakefulness, the relevance of these associations to the pathogenesis of central apneas during sleep is not clear.
Cheyne-Stokes respiration with CSA (CSR-CSA) is a particular form of periodic breathing (PB) in which central apneas and hypopneas alternate with periods of hyperventilation having a waxing and waning pattern of VT and a consequent long PB cycle duration (16, 17). We have previously demonstrated that patients with CHF and CSR-CSA have a greater PB cycle duration than that of patients with CSA but without CHF, which is directly proportional to lung to chemoreceptor circulation time and inversely proportional to cardiac output (17). Consequently, depressed LV function appears to be an important factor in causing CSR-CSA.
CSR-CSA has been described in association with both CHF and strokes (6, 8, 16, 17). In fact, the first case report of CSR described by Cheyne (16) in 1818 was of a man who suffered from both CHF and a stroke. Subsequently, the question arose as to whether CSR was primarily related to cardiac or neurologic dysfunction (1, 5, 10, 11, 13, 14).
In view of these considerations, we hypothesized, first, that CSA in patients with stroke would be associated with nocturnal hypocapnia and, second, that in those stroke patients with CSA in whom systolic LV function was depressed, the PB cycle would be greater in duration than in those whose LV systolic function was normal. To test these hypotheses, we examined stroke location, transcutaneous PCO2, PB cycle duration, and LV systolic function in stroke patients with and without CSA.
METHODS
Subjects
We studied prospectively 93 consecutive patients after they had experienced a stroke who were admitted to the Stroke Rehabilitation Unit of the Toronto Rehabilitation Institute as part of an ongoing epidemiologic study (15). The diagnosis of completed stroke was confirmed by neurologists and was based on the following factors: (1) a history of sudden onset of a neurologic deficit that lasted more than 24 hours, (2) presence of a neurologic deficit on physical examination, and (3) brain lesion compatible with the neurologic deficit detected on computed tomography or magnetic resonance imaging of the brain. Patients with previously diagnosed sleep apnea were excluded from the study. On admission to the Stroke Rehabilitation Unit, demographic characteristics, height, weight, and body mass index were determined.
Echocardiography
All patients underwent two-dimensional echocardiography before sleep studies during admission in the Stroke Rehabilitation Unit from which LV ejection fraction (LVEF) was classified as being normal (LVEF > 40%) or depressed (LVEF 40%) (18). Echocardiography was interpreted by personnel blind to the results of polysomnography.
Polysomnography
Overnight polysomnography was performed in all patients while in the Stroke Rehabilitation Unit using standard techniques for scoring sleep stages and arousals, as previously described (19, 20). Polysomnography was performed and interpreted by personnel blind to the results of echocardiography. Thoracoabdominal movements and VT were monitored by a calibrated respiratory inductance plethysmograph (21). Transcutaneous PCO2 was recorded continuously with a transcutaneous capnograph calibrated against test gases at the beginning and end of each study, as previously described (22). The mean and lowest SaO2 and O2 desaturation index during sleep were determined, as previously described (22).
Central apneas were identified by the absence of VT for at least 10 seconds without thoracoabdominal motion. Central hypopneas were defined as a 50% or greater reduction in VT from the baseline value for at least 10 seconds with proportional in-phase reductions in thoracoabdominal movements (22eC24). Obstructive apneas and hypopneas were similarly defined, except that out-of-phase thoracoabdominal motion had to be present (4). Sleep apnea was defined as 10 apneas and hypopneas or more per hour of sleep (apneaeChypopnea index), and CSA as a central apneaeChypopnea index of 10 or more per hour of sleep. Patients without CSA and those with predominantly obstructive sleep apnea were combined into a single non-CSA group.
Analysis of PB characteristics was confined to Stage 2 sleep to control for the potential influence of different sleep stages on cycling characteristics, and because this is where the great majority of central apneas occurred. Hyperpnea duration was calculated as the time between the beginning of inspiration during the ventilatory cycle and the onset of the subsequent apnea. Apnea duration was defined as the time between the end of inspiration of the breath preceding a central apnea and the onset of inspiration of the breath terminating the apnea. Cycle duration was calculated as the sum of the hyperpnea and apnea durations (25). The circulation time from the lungs to the peripheral chemoreceptors was estimated by measuring the lung-to-ear circulation time (LECT), which is the time from the first breath after a central apnea to the subsequent nadir of SaO2 detected at the ear by an oximeter, as previously described (17, 24). To determine whether the presence of LV systolic dysfunction was associated with a CSR pattern, we measured the amount of time it took after apneas to reach peak VT during hyperpneas, and the number of breaths per hyperpnea, and compared these values between patients with CSA who had an LVEF of more than 40% and those who had an LVEF of 40% or less.
Statistical Analysis
All data are expressed as mean ± SEM unless stated otherwise. Statistical analysis was performed using SPSS version 10.0.7 (SPSS, Chicago, IL). Between-group comparisons of continuous data were by two-tailed unpaired t test or the Mann-Whitney rank sum test as appropriate. The 2 test or the Fisher's exact test was used for nominal variables as appropriate. Multiple regression analysis was performed to determine independent correlates of CSA. Because we have previously shown that the duration of hyperpnea in CSA is related to cardiac function, we also analyzed the relationship between hyperpnea duration and the presence of depressed LV systolic function (i.e., LVEF 40%) (17). Accordingly, the sensitivity and specificity of hyperpnea duration for detecting an LVEF of 40% or less were examined through use of receiver operating characteristic curve analyses.
RESULTS
As shown in Table 1, subjects were generally middle-aged or elderly, and were mildly overweight. Approximately 62% were men and 38% were women. The great majority had ischemic stroke, whereas a minority had hemorrhagic stroke or a combination of both. Most patients had supratentorial lesions, whereas a small minority had either infratentorial lesions or both supra- and infratentorial lesions. Only 4% of patients had bilateral cerebral brain lesions. Seventy-six (82%) patients suffered from their first stroke, whereas 17 (18%) had had a previous stroke. Echocardiography was performed an average of 40 ± 6 days from the time of stroke onset. Polysomnography was performed 44 ± 3 days after stroke. CSA was present in 19% of subjects. There were no significant differences in age, body mass index, Functional Independence Measure scores, Epworth Sleepiness Scale, and underlying diseases between patients with and without CSA. There was a tendency for a higher male/female ratio in the CSA group (p = 0.06).
Polysomnographic data are shown in Table 2. The total apneaeChypopnea index was greater in the CSA than in the non-CSA group, and by design, the central apneaeChypopnea index was significantly higher than in the non-CSA group. However, sleep structure did not differ significantly between the two groups. Compared with the non-CSA group, those in the CSA group had lower mean transcutaneous PCO2 during sleep (p = 0.028) and a higher prevalence of LV systolic dysfunction (p = 0.043).
Figure 1 shows polysomnographic tracings from representative subjects with CSA and an LVEF of more than 40% (Figure 1A) and with an LVEF of 40% or less (Figure 1B). Compared with the subject with normal LV function, the subject with LV dysfunction had longer hyperpnea and PB cycle durations.
In those four subjects with CSA and LV dysfunction, the mean PB cycle and hyperpnea durations were greater than in those 14 subjects with CSA and normal LV function (61.9 ± 9.9 vs. 46.6 ± 2.9, p = 0.05, and 35.0 ± 6.8 vs. 23.5 ± 1.5, p = 0.019, respectively), but the apnea duration was not greater (26.9 ± 3.2 vs. 23.1 ± 2.1, p = 0.4). Because this analysis was based on only four subjects with LV dysfunction, we included eight additional stroke patients with CSA to increase the statistical power of our analysis. They were studied consecutively between September 1995 to May 2001 and had an LVEF of less than 40% by echocardiography that was performed within 6 months of the sleep study. None had symptoms of heart failure at that time.
The inclusion of these additional subjects confirmed the findings of the original prospective data as shown in Figure 2. Figures 2A and 2B demonstrate that the 12 patients with CSA and with LV systolic dysfunction had significantly greater PB cycle and hyperpnea times than the 14 patients with CSA and with normal LV function. However, there was no difference in apnea duration between the two groups (Figure 2C). The amount of time it took to reach peak VT during hyperpneas and the number of breaths per hyperpnea were significantly greater in patients with CSA and an LVEF of 40% or less than in those with an LVEF greater than 40% (15.3 ± 1.4 vs. 7.1 ± 0.5 seconds, p = 0.011, and 10.2 ± 1.2 vs. 6.0 ± 0.8 breaths, p < 0.001, respectively). These data indicated a more gradual rise in VT to peak over more breaths in those with LV systolic dysfunction than in those without LV systolic dysfunction. In addition, in those subjects with CSA and an LVEF of 40% or less, the mean LECT was greater than in those subjects with CSA and an LVEF greater than 40% (Figure 3). The PB cycle and hyperpnea times also correlated significantly with LECT (R = 0.630, p = 0.02, and R = 0.801, p = 0.001, respectively) but not with apnea duration (R = 0.233, p = 0.44).
Multiple logistic regression analysis, based on the prospective data of the original 93 subjects, demonstrated that CSA was significantly, and independently, related to lower mean transcutaneous PCO2 during sleep, with an odds ratio of 0.8, and to the presence of an LVEF of 40% or less, with an odds ratio of 8.5 (Table 3). Regarding the ability of hyperpnea duration in patients with CSA to detect LV systolic dysfunction, we found that hyperpnea duration at 33 seconds provided the optimal diagnostic accuracy for an LVEF of 40% or less, with sensitivity 0.833 and specificity of 1.
DISCUSSION
Our study demonstrates several important findings regarding the pathophysiologic correlates of CSA in patients with stroke. First, CSA in patients with stroke was associated with nocturnal hypocapnia but was not related to either stroke type or location. Second, CSA was associated with occult LV systolic dysfunction independently of other known risk factors. Third, among stroke patients with CSA, those with evidence of LV systolic dysfunction had a longer LECT and hyperpnea time, with a waxingeCwaning pattern of VT typical of CRS, than those without LV systolic dysfunction. These data indicate that CSA in association with stroke shares several of the pathophysiologic features of CSA in patients with idiopathic CSA and those with CHF.
The 19% prevalence of CSA in the present study is in keeping with the 6 to 28% prevalence of CSA previously reported in patients with stroke (1, 26). Therefore, given the prospective nature of our epidemiologic study, our data are probably representative of stroke populations in rehabilitation units.
The observation that nocturnal hypocapnia was associated with CSA in our patients with stroke is consistent with the observations that nocturnal hypocapnia plays a key role in the pathogenesis of other forms of CSA not associated with stroke, such as idiopathic CSA, and CSA occurring in patients with CHF (3, 6, 8, 10). Hypocapnia maintains PCO2 close to the apneic threshold such that even a small increase in ventilation, such as might occur after an arousal from sleep, is sufficient to drive PCO2 below the apneic threshold and trigger central apnea (3, 6, 8, 13, 22). The tendency to hyperventilate in patients with idiopathic CSA is related to increased chemoresponsiveness, whereas in patients with CHF and CSR-CSA, it is related to both increased chemoresponsiveness and the presence of LV systolic dysfunction with vagal irritant receptor stimulation by pulmonary vascular congestion (3, 27, 28).
Although we have no information on chemoresponsiveness, our data indicate that CSA in our patients with stroke is strongly associated with the presence of LV systolic dysfunction. In subjects with an LVEF of 40% or less, the odds of having CSA were increased by 8.5-fold after controlling for potential confounding factors (Table 3). Another striking finding of our study was that, among stroke patients with CSA, those with evidence of LV systolic dysfunction had longer PB cycles, with a typical CRS pattern of waxing and waning VTs during hyperpnea, than patients with normal LV systolic function. Furthermore, none of the four patients with stroke from the prospective epidemiologic data who had CSA and LV systolic dysfunction had clinical symptoms or a history of CHF before being admitted to the Stroke Rehabilitation Unit. Thus, all of them had occult, asymptomatic LV systolic dysfunction. In addition, four of eight patients with stroke and with an LVEF of 40% or less had CSA. This 50% prevalence of CSA is in keeping with the 55% prevalence of CSA reported in patients with asymptomatic LV systolic dysfunction (LVEF 40%) but without stroke (29) and the 25 to 40% prevalence reported in patients with symptomatic CHF and impaired LV systolic function (7, 8).
In a previous study, Hall and coworkers (17) compared PB cycle and hyperpnea durations in patients with idiopathic CSA and normal systolic function to those with CHF and depressed LV systolic function. None of the patients had strokes. They found that those with CHF had longer PB and hyperpnea durations than those without CHF, and that these were inversely proportional to cardiac output and directly proportional to LECT. Stroke volume and cardiac output were lower in the patients with CHF than in those without CHF. These observations indicated that prolonged PB cycle and hyperpnea durations typical of CSR-CSA are a consequence of LV systolic dysfunction, low cardiac output, and prolonged circulation time. Our data are in keeping with those of Hall and colleagues (17), but we extend them to the stroke population. We found that, among patients with stroke and CSA, those with an LVEF of 40% or less had longer PB cycles and hyperpneas than those with an LVEF greater than 40%. We also found that a long hyperpnea duration was predictive of the presence of occult LV systolic dysfunction, such that a hyperpnea duration of 33 seconds provided the best diagnostic accuracy for detecting an LVEF of 40% or less, with a sensitivity of 0.833 and specificity of 1.
It is also noteworthy that the mean cycle duration of PB in the patients with CSA and an LVEF of 40% or less was 67 seconds. This finding is similar to the cycle duration of approximately 1 minute first described by Cheyne (16) in a patient with CSR who had both a stroke and CHF. Thus, it is likely that impaired cardiac function was a major contributor to the development of CSR both in Cheyne's case and in ours.
Studies of CSR in the 1950s and 1960s (9eC11) reported that the presence of central apneas observed during wakefulness in patients with stroke was associated with extensive bilateral brain disease, hypocapnia, respiratory alkalosis, and increased respiratory sensitivity to CO2. However, none of these studies was performed during sleep. Although, cardiovascular abnormalities and pulmonary congestion were frequently found in these populations, it was suggested that increased sensitivity to CO2 after a stroke resulted from bilateral supramedullary brain lesions that disinhibited neural stimulatory input into the brainstem respiratory center (neurogenic CO2 hypersensitivity), and that extracerebral abnormalities were not the primary cause of CSR (9). However, subsequent reports (12eC15) have failed to confirm a relationship between the site and size of a stroke and the presence of CSA. Our data are in agreement with these latter studies because we did not find any relationship between stroke location and the presence of CSA. Moreover, our study of 93 subjects is the largest study of its kind to address this issue. Only four patients in this study had bilateral cerebral brain lesions. Of these, one patient had CSA but three did not. Parra and colleagues (30) have shown that CSA is frequent immediately after a stroke, but decreases in frequency several months later. These observations suggest that strokes do predispose to CSA, at least in the immediately post-stroke period. Accordingly, because CSA was present in 14 of our patients with stroke who had normal LV systolic function, it is likely that neurologic lesions caused by the strokes contributed to the pathogenesis of their CSA. However, because we found no relationship between the type or location of strokes and the presence of CSA, it is difficult to speculate on the nature of the neurologic lesion that predisposed to CSA in these subjects. Further work will be required to determine the causes of CSA in patients with stroke and without LV dysfunction.
Our data do not allow us to say whether CSA was present in our patients before they had a stroke or whether it developed afterwards. Many of our patients had both some central and some obstructive events, raising the possibility that, in some cases, central events predisposed to the development of obstructive events and vice versa (24). However, our study was not designed to investigate this possibility. Further research will be required to determine whether there is any pathophysiologic interaction between central and obstructive respiratory events during sleep in patients with strokes.
In conclusion, our data indicate that CSA is relatively common in patients with stroke, and that its presence is related to hypocapnia and impaired LV systolic function. These findings also shed light on a long-standing controversy as to whether CSR is related primarily to cardiac or neurologic disease. We found that, among patients with stroke and CSA, the presence of a CSR pattern with long hyperpnea and cycle durations, and a gradual rise to peak VT during hyperpnea is associated with LV systolic dysfunction, but is not related to the location or type of stroke. In fact, we found that hyperpnea duration was highly sensitive and specific for detecting the presence of an LVEF of 40% or less. Moreover, in all of these patients, LV systolic dysfunction was occult. These data indicate that the presence of CSA with a CSR pattern is more closely associated with LV systolic dysfunction than it is with the stroke per se. However, in stroke patients with CSA, but without a CSR pattern, it is likely that the stroke itself is playing a role in the pathogenesis of CSA, but it is not clear how. Therefore, in patients with stroke, the presence of CSA with a CSR pattern appears to be a sign of underlying LV systolic dysfunction. This finding is of considerable clinical significance. Approximately one third of patients suffering a stroke subsequently die of heart disease (31). Thus, the presence of CSA, particularly in association with a CSR pattern, should raise suspicion of the presence of asymptomatic LV systolic dysfunction, the treatment of which could improve prognosis in such patients (32).
REFERENCES
Nachtmann A, Siebler M, Rose G, Sitzer M, Steinmetz H. Cheyne-Stokes respiration in ischemic stroke. Neurology 1995;45:820eC821.
Bassetti C, Aldrich MS, Chervin RD, Quint D. Sleep apnea in patients with transient ischemic attack and stroke: a prospective study of 59 patients. Neurology 1996;47:1167eC1173.
Bradley TD, Phillipson EA. Central sleep apnea. Clin Chest Med 1992;13:493eC505.
Solin P, Roebuck T, Johns DP, Walters EH, Naughton MT. Peripheral and central ventilatory responses in central sleep apnea with and without congestive heart failure. Am J Respir Crit Care Med 2000;162:2194eC2200.
Javaheri S. A mechanism of central sleep apnea in patients with heart failure. N Engl J Med 1999;341:949eC954.
Xie A, Rutherford R, Rankin F, Wong B, Bradley TD. Hypocapnia and increased ventilatory responsiveness in patients with idiopathic central sleep apnea. Am J Respir Crit Care Med 1995;152:1950eC1955.
Javaheri S, Parker TJ, Liming JD, Corbett WS, Nishiyama H, Wexler L, Roselle GA. Sleep apnea in 81 ambulatory male patients with stable heart failure: types and their prevalences, consequences, and presentations. Circulation 1998;97:2154eC2159.
Sin DD, Fitzgerald F, Parker JD, Newton G, Floras JS, Bradley TD. Risk factors for central and obstructive sleep apnea in 450 men and women with congestive heart failure. Am J Respir Crit Care Med 1999;160:1101eC1106.
Heyman A, Birchfield RI, Sieker HO. Effects of bilateral cerebral infarction on respiratory center sensitivity. Neurology 1958;8:694eC700.
Brown HW, Plum F. The neurologic basis of Cheyne-Stokes respiration. Am J Med 1961;30:849eC869.
Karp HR, Heyman A. Cerebral circulation and function in Cheyne-Strokes respiration. Neurology 1961;30:849eC869.
Lee MC, Klassen AC, Resch JA. Respiratory pattern disturbances in ischemic cerebral vascular disease. Stroke 1974;5:612eC616.
North JB, Jennett S. Abnormal breathing patterns associated with acute brain damage. Arch Neurol 1974;31:338eC344.
Bassetti C, Aldrich MS, Quint D. Sleep-disordered breathing in patients with acute supra- and infratentorial strokes: a prospective study of 39 patients. Stroke 1997;28:1765eC1772.
Kaneko Y, Hajek VE, Zivanovic V, Raboud J, Bradley TD. Relationship of sleep apnea to functional capacity and length of hospitalization following stroke. Sleep 2003;26:293eC297.
Cheyne J. A case of apoplexy in which the freshy part of the heart was converted into fat. Dublin Hosp Rep 1818;2:216eC223.
Hall MJ, Xie A, Rutherford R, Ando S, Floras JS, Bradley TD. Cycle length of periodic breathing in patients with and without heart failure. Am J Respir Crit Care Med 1996;154:376eC381.
Otterstad JE, Froeland G, St. John Sutton M, Holme I. Accuracy and reproducibility of biplane two-dimensional echocardiographic measurements of left ventricular dimensions and function. Eur Heart J 1997;18:507eC513.
Rechtschaffen A, Kales AA. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Los Angeles, CA: UCLA Brain Information Service/Brain Research Institute; 1968.
Sleep Disorders Atlas Task Force of the American Sleep Disorders. EEG arousals: scoring rules and examples: a preliminary report. Sleep 1992;15:173eC184.
Chadha TS, Watson H, Birch S, Jenouri GA, Schneider AW, Cohn MA, Sackner MA. Validation of respiratory inductive plethysmography using different calibration procedures. Am Rev Respir Dis 1982;125:644eC649.
Naughton M, Benard D, Tam A, Rutherford R, Bradley TD. Role of hyperventilation in the pathogenesis of central sleep apneas in patients with congestive heart failure. Am Rev Respir Dis 1993;148:330eC338.
Gould GA, Whyte KJ, Airlie AA, Cateral JR, Shapiro CM, Douglas NJ. The sleep hypopnea syndrome. Am Rev Respir Dis 1988;137:895eC898.
Tkacova R, Niroumand M, Lorenzi-Filho G, Bradley TD. Overnight shift from obstructive to central apneas in patients with heart failure: role of PCO2 and circulatory delay. Circulation 2001;103:238eC243.
Hanly P, Zuberi N, Gray R. Pathogenesis of Cheyne-Stokes respiration in patients with congestive heart failure: relationship to arterial PCO2. Chest 1993;104:1079eC1084.
Wessendorf TE, Teschler H, Wang YM, Konietzko N, Thilmann AF. Sleep-disordered breathing among patients with first-ever stroke. J Neurol 2000;247:41eC47.
Lorenzi-Filho G, Azevedo ER, Parker JD, Bradley TD. Relationship of PaCO2 to pulmonary wedge pressure in heart failure. Eur Respir J 2002;19:37eC40.
Solin P, Bergin P, Richardson M, Kaye DM, Walters EH, Naughton MT. Influence of pulmonary capillary wedge pressure on central apnea in heart failure. Circulation 1999;99:1574eC1579.
Lanfranchi PA, Somers VK, Braghiroli A, Corra U, Eleuteri E, Giannuzzi P. Central sleep apnea in left ventricular dysfunction: prevalence and implications for arrhythmic risk. Circulation 2003;107:727eC732.
Parra O, Arboix A, Bechich S, Garcia-Eroles L, Montserrat JM, Lopez JA, Ballester E, Guerra JM, Sopena JJ. Time course of sleep-related breathing disorders in first-ever stroke or transient ischemic attack. Am J Respir Crit Care Med 2000;161:375eC380.
Hardie K, Hankey GJ, Jamrozik K, Broadhurst RJ, Anderson C. Ten-year survival after first-ever stroke in the Perth Community Stroke Study. Stroke 2003;34:1842eC1846.
The SOLVD Investigators. Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. N Engl J Med 1992;327:685eC691.(Cherdchai Nopmaneejumrusl)
ABSTRACT
Background: Central sleep apnea (CSA) and Cheyne-Stokes respiration have been reported in association with stroke, but their pathophysiologic correlates have not been well described. Objective: To test the hypotheses that (1) CSA in patients with stroke is associated with nocturnal hypocapnia and (2) in those stroke patients with CSA and with left ventricular (LV) systolic dysfunction, periodic breathing (PB) will have a Cheyne-Stokes respiration pattern in which cycle duration is greater than in those without LV systolic dysfunction. Methods: We prospectively performed polysomnography and echocardiography in 93 patients with stroke. CSA was defined as central apneas and hypopneas occurring at a rate of 10 or more per hour of sleep. In patients with CSA, we compared PB cycle duration between those with normal and impaired LV systolic function (LV ejection fraction [LVEF] > 40% and 40%, respectively). Results: CSA was found in 19% of subjects who had lower nocturnal transcutaneous PCO2 (39.3 ± 0.9 vs. 42.8 ± 0.8 mm Hg, p = 0.015) and a higher prevalence of LVEF of 40% or less (22 vs. 5%, p = 0.043) than stroke patients without CSA. There was no significant difference in stroke location or type between the two groups. In patients with CSA, those with LVEF of 40% or less had a longer PB cycle than those with an LVEF of more than 40% (66.6 ± 5.6 vs. 46.6 ± 2.9 seconds, p = 0.006), but had no symptoms of heart failure. Conclusion: In patients with stroke, CSA is associated with hypocapnia and occult LV systolic dysfunction but is not related to the location or type of stroke. The presence of LV systolic dysfunction is associated with a Cheyne-Stokes pattern of hyperpnea.
Key Words: Cheyne-Stokes respiration sleep apnea stroke
Although central sleep apnea (CSA) has been reported in association with stroke (1, 2), its pathogenesis in this setting remains unclear. Central apneas occur when there is a temporary withdrawal of central respiratory drive causing cessation of respiratory muscle activity and airflow (3). In subjects without stroke, common predisposing factors for CSA are enhanced central and peripheral chemosensitivity (4eC6), hypocapnia, and the presence of left ventricular (LV) systolic dysfunction (7, 8). CSA in patients with both congestive heart failure (CHF) and with normal LV function (idiopathic CSA) are triggered and propagated by hyperventilation and hypocapnia, causing PCO2 to fall below the apneic threshold, precipitating a central apnea (3eC6). Similar factors may be at play in the pathogenesis of CSA in subjects with stroke. Several articles in the 1950s and 1960s described hyperventilation, respiratory alkalosis, and increased respiratory sensitivity to CO2 (9eC11) in association with recurrent central apnea observed during wakefulness. It was suggested that increased sensitivity to CO2 after a stroke resulted from bilateral supramedullary brain lesions that disinhibit neural stimulatory input into the brainstem respiratory center (neurogenic CO2 hypersensitivity) (9). However, subsequent reports have failed to confirm a relationship between the site and size of a stroke and the presence of central apneas (12eC15). Moreover, because these central apneas were studied only during wakefulness, the relevance of these associations to the pathogenesis of central apneas during sleep is not clear.
Cheyne-Stokes respiration with CSA (CSR-CSA) is a particular form of periodic breathing (PB) in which central apneas and hypopneas alternate with periods of hyperventilation having a waxing and waning pattern of VT and a consequent long PB cycle duration (16, 17). We have previously demonstrated that patients with CHF and CSR-CSA have a greater PB cycle duration than that of patients with CSA but without CHF, which is directly proportional to lung to chemoreceptor circulation time and inversely proportional to cardiac output (17). Consequently, depressed LV function appears to be an important factor in causing CSR-CSA.
CSR-CSA has been described in association with both CHF and strokes (6, 8, 16, 17). In fact, the first case report of CSR described by Cheyne (16) in 1818 was of a man who suffered from both CHF and a stroke. Subsequently, the question arose as to whether CSR was primarily related to cardiac or neurologic dysfunction (1, 5, 10, 11, 13, 14).
In view of these considerations, we hypothesized, first, that CSA in patients with stroke would be associated with nocturnal hypocapnia and, second, that in those stroke patients with CSA in whom systolic LV function was depressed, the PB cycle would be greater in duration than in those whose LV systolic function was normal. To test these hypotheses, we examined stroke location, transcutaneous PCO2, PB cycle duration, and LV systolic function in stroke patients with and without CSA.
METHODS
Subjects
We studied prospectively 93 consecutive patients after they had experienced a stroke who were admitted to the Stroke Rehabilitation Unit of the Toronto Rehabilitation Institute as part of an ongoing epidemiologic study (15). The diagnosis of completed stroke was confirmed by neurologists and was based on the following factors: (1) a history of sudden onset of a neurologic deficit that lasted more than 24 hours, (2) presence of a neurologic deficit on physical examination, and (3) brain lesion compatible with the neurologic deficit detected on computed tomography or magnetic resonance imaging of the brain. Patients with previously diagnosed sleep apnea were excluded from the study. On admission to the Stroke Rehabilitation Unit, demographic characteristics, height, weight, and body mass index were determined.
Echocardiography
All patients underwent two-dimensional echocardiography before sleep studies during admission in the Stroke Rehabilitation Unit from which LV ejection fraction (LVEF) was classified as being normal (LVEF > 40%) or depressed (LVEF 40%) (18). Echocardiography was interpreted by personnel blind to the results of polysomnography.
Polysomnography
Overnight polysomnography was performed in all patients while in the Stroke Rehabilitation Unit using standard techniques for scoring sleep stages and arousals, as previously described (19, 20). Polysomnography was performed and interpreted by personnel blind to the results of echocardiography. Thoracoabdominal movements and VT were monitored by a calibrated respiratory inductance plethysmograph (21). Transcutaneous PCO2 was recorded continuously with a transcutaneous capnograph calibrated against test gases at the beginning and end of each study, as previously described (22). The mean and lowest SaO2 and O2 desaturation index during sleep were determined, as previously described (22).
Central apneas were identified by the absence of VT for at least 10 seconds without thoracoabdominal motion. Central hypopneas were defined as a 50% or greater reduction in VT from the baseline value for at least 10 seconds with proportional in-phase reductions in thoracoabdominal movements (22eC24). Obstructive apneas and hypopneas were similarly defined, except that out-of-phase thoracoabdominal motion had to be present (4). Sleep apnea was defined as 10 apneas and hypopneas or more per hour of sleep (apneaeChypopnea index), and CSA as a central apneaeChypopnea index of 10 or more per hour of sleep. Patients without CSA and those with predominantly obstructive sleep apnea were combined into a single non-CSA group.
Analysis of PB characteristics was confined to Stage 2 sleep to control for the potential influence of different sleep stages on cycling characteristics, and because this is where the great majority of central apneas occurred. Hyperpnea duration was calculated as the time between the beginning of inspiration during the ventilatory cycle and the onset of the subsequent apnea. Apnea duration was defined as the time between the end of inspiration of the breath preceding a central apnea and the onset of inspiration of the breath terminating the apnea. Cycle duration was calculated as the sum of the hyperpnea and apnea durations (25). The circulation time from the lungs to the peripheral chemoreceptors was estimated by measuring the lung-to-ear circulation time (LECT), which is the time from the first breath after a central apnea to the subsequent nadir of SaO2 detected at the ear by an oximeter, as previously described (17, 24). To determine whether the presence of LV systolic dysfunction was associated with a CSR pattern, we measured the amount of time it took after apneas to reach peak VT during hyperpneas, and the number of breaths per hyperpnea, and compared these values between patients with CSA who had an LVEF of more than 40% and those who had an LVEF of 40% or less.
Statistical Analysis
All data are expressed as mean ± SEM unless stated otherwise. Statistical analysis was performed using SPSS version 10.0.7 (SPSS, Chicago, IL). Between-group comparisons of continuous data were by two-tailed unpaired t test or the Mann-Whitney rank sum test as appropriate. The 2 test or the Fisher's exact test was used for nominal variables as appropriate. Multiple regression analysis was performed to determine independent correlates of CSA. Because we have previously shown that the duration of hyperpnea in CSA is related to cardiac function, we also analyzed the relationship between hyperpnea duration and the presence of depressed LV systolic function (i.e., LVEF 40%) (17). Accordingly, the sensitivity and specificity of hyperpnea duration for detecting an LVEF of 40% or less were examined through use of receiver operating characteristic curve analyses.
RESULTS
As shown in Table 1, subjects were generally middle-aged or elderly, and were mildly overweight. Approximately 62% were men and 38% were women. The great majority had ischemic stroke, whereas a minority had hemorrhagic stroke or a combination of both. Most patients had supratentorial lesions, whereas a small minority had either infratentorial lesions or both supra- and infratentorial lesions. Only 4% of patients had bilateral cerebral brain lesions. Seventy-six (82%) patients suffered from their first stroke, whereas 17 (18%) had had a previous stroke. Echocardiography was performed an average of 40 ± 6 days from the time of stroke onset. Polysomnography was performed 44 ± 3 days after stroke. CSA was present in 19% of subjects. There were no significant differences in age, body mass index, Functional Independence Measure scores, Epworth Sleepiness Scale, and underlying diseases between patients with and without CSA. There was a tendency for a higher male/female ratio in the CSA group (p = 0.06).
Polysomnographic data are shown in Table 2. The total apneaeChypopnea index was greater in the CSA than in the non-CSA group, and by design, the central apneaeChypopnea index was significantly higher than in the non-CSA group. However, sleep structure did not differ significantly between the two groups. Compared with the non-CSA group, those in the CSA group had lower mean transcutaneous PCO2 during sleep (p = 0.028) and a higher prevalence of LV systolic dysfunction (p = 0.043).
Figure 1 shows polysomnographic tracings from representative subjects with CSA and an LVEF of more than 40% (Figure 1A) and with an LVEF of 40% or less (Figure 1B). Compared with the subject with normal LV function, the subject with LV dysfunction had longer hyperpnea and PB cycle durations.
In those four subjects with CSA and LV dysfunction, the mean PB cycle and hyperpnea durations were greater than in those 14 subjects with CSA and normal LV function (61.9 ± 9.9 vs. 46.6 ± 2.9, p = 0.05, and 35.0 ± 6.8 vs. 23.5 ± 1.5, p = 0.019, respectively), but the apnea duration was not greater (26.9 ± 3.2 vs. 23.1 ± 2.1, p = 0.4). Because this analysis was based on only four subjects with LV dysfunction, we included eight additional stroke patients with CSA to increase the statistical power of our analysis. They were studied consecutively between September 1995 to May 2001 and had an LVEF of less than 40% by echocardiography that was performed within 6 months of the sleep study. None had symptoms of heart failure at that time.
The inclusion of these additional subjects confirmed the findings of the original prospective data as shown in Figure 2. Figures 2A and 2B demonstrate that the 12 patients with CSA and with LV systolic dysfunction had significantly greater PB cycle and hyperpnea times than the 14 patients with CSA and with normal LV function. However, there was no difference in apnea duration between the two groups (Figure 2C). The amount of time it took to reach peak VT during hyperpneas and the number of breaths per hyperpnea were significantly greater in patients with CSA and an LVEF of 40% or less than in those with an LVEF greater than 40% (15.3 ± 1.4 vs. 7.1 ± 0.5 seconds, p = 0.011, and 10.2 ± 1.2 vs. 6.0 ± 0.8 breaths, p < 0.001, respectively). These data indicated a more gradual rise in VT to peak over more breaths in those with LV systolic dysfunction than in those without LV systolic dysfunction. In addition, in those subjects with CSA and an LVEF of 40% or less, the mean LECT was greater than in those subjects with CSA and an LVEF greater than 40% (Figure 3). The PB cycle and hyperpnea times also correlated significantly with LECT (R = 0.630, p = 0.02, and R = 0.801, p = 0.001, respectively) but not with apnea duration (R = 0.233, p = 0.44).
Multiple logistic regression analysis, based on the prospective data of the original 93 subjects, demonstrated that CSA was significantly, and independently, related to lower mean transcutaneous PCO2 during sleep, with an odds ratio of 0.8, and to the presence of an LVEF of 40% or less, with an odds ratio of 8.5 (Table 3). Regarding the ability of hyperpnea duration in patients with CSA to detect LV systolic dysfunction, we found that hyperpnea duration at 33 seconds provided the optimal diagnostic accuracy for an LVEF of 40% or less, with sensitivity 0.833 and specificity of 1.
DISCUSSION
Our study demonstrates several important findings regarding the pathophysiologic correlates of CSA in patients with stroke. First, CSA in patients with stroke was associated with nocturnal hypocapnia but was not related to either stroke type or location. Second, CSA was associated with occult LV systolic dysfunction independently of other known risk factors. Third, among stroke patients with CSA, those with evidence of LV systolic dysfunction had a longer LECT and hyperpnea time, with a waxingeCwaning pattern of VT typical of CRS, than those without LV systolic dysfunction. These data indicate that CSA in association with stroke shares several of the pathophysiologic features of CSA in patients with idiopathic CSA and those with CHF.
The 19% prevalence of CSA in the present study is in keeping with the 6 to 28% prevalence of CSA previously reported in patients with stroke (1, 26). Therefore, given the prospective nature of our epidemiologic study, our data are probably representative of stroke populations in rehabilitation units.
The observation that nocturnal hypocapnia was associated with CSA in our patients with stroke is consistent with the observations that nocturnal hypocapnia plays a key role in the pathogenesis of other forms of CSA not associated with stroke, such as idiopathic CSA, and CSA occurring in patients with CHF (3, 6, 8, 10). Hypocapnia maintains PCO2 close to the apneic threshold such that even a small increase in ventilation, such as might occur after an arousal from sleep, is sufficient to drive PCO2 below the apneic threshold and trigger central apnea (3, 6, 8, 13, 22). The tendency to hyperventilate in patients with idiopathic CSA is related to increased chemoresponsiveness, whereas in patients with CHF and CSR-CSA, it is related to both increased chemoresponsiveness and the presence of LV systolic dysfunction with vagal irritant receptor stimulation by pulmonary vascular congestion (3, 27, 28).
Although we have no information on chemoresponsiveness, our data indicate that CSA in our patients with stroke is strongly associated with the presence of LV systolic dysfunction. In subjects with an LVEF of 40% or less, the odds of having CSA were increased by 8.5-fold after controlling for potential confounding factors (Table 3). Another striking finding of our study was that, among stroke patients with CSA, those with evidence of LV systolic dysfunction had longer PB cycles, with a typical CRS pattern of waxing and waning VTs during hyperpnea, than patients with normal LV systolic function. Furthermore, none of the four patients with stroke from the prospective epidemiologic data who had CSA and LV systolic dysfunction had clinical symptoms or a history of CHF before being admitted to the Stroke Rehabilitation Unit. Thus, all of them had occult, asymptomatic LV systolic dysfunction. In addition, four of eight patients with stroke and with an LVEF of 40% or less had CSA. This 50% prevalence of CSA is in keeping with the 55% prevalence of CSA reported in patients with asymptomatic LV systolic dysfunction (LVEF 40%) but without stroke (29) and the 25 to 40% prevalence reported in patients with symptomatic CHF and impaired LV systolic function (7, 8).
In a previous study, Hall and coworkers (17) compared PB cycle and hyperpnea durations in patients with idiopathic CSA and normal systolic function to those with CHF and depressed LV systolic function. None of the patients had strokes. They found that those with CHF had longer PB and hyperpnea durations than those without CHF, and that these were inversely proportional to cardiac output and directly proportional to LECT. Stroke volume and cardiac output were lower in the patients with CHF than in those without CHF. These observations indicated that prolonged PB cycle and hyperpnea durations typical of CSR-CSA are a consequence of LV systolic dysfunction, low cardiac output, and prolonged circulation time. Our data are in keeping with those of Hall and colleagues (17), but we extend them to the stroke population. We found that, among patients with stroke and CSA, those with an LVEF of 40% or less had longer PB cycles and hyperpneas than those with an LVEF greater than 40%. We also found that a long hyperpnea duration was predictive of the presence of occult LV systolic dysfunction, such that a hyperpnea duration of 33 seconds provided the best diagnostic accuracy for detecting an LVEF of 40% or less, with a sensitivity of 0.833 and specificity of 1.
It is also noteworthy that the mean cycle duration of PB in the patients with CSA and an LVEF of 40% or less was 67 seconds. This finding is similar to the cycle duration of approximately 1 minute first described by Cheyne (16) in a patient with CSR who had both a stroke and CHF. Thus, it is likely that impaired cardiac function was a major contributor to the development of CSR both in Cheyne's case and in ours.
Studies of CSR in the 1950s and 1960s (9eC11) reported that the presence of central apneas observed during wakefulness in patients with stroke was associated with extensive bilateral brain disease, hypocapnia, respiratory alkalosis, and increased respiratory sensitivity to CO2. However, none of these studies was performed during sleep. Although, cardiovascular abnormalities and pulmonary congestion were frequently found in these populations, it was suggested that increased sensitivity to CO2 after a stroke resulted from bilateral supramedullary brain lesions that disinhibited neural stimulatory input into the brainstem respiratory center (neurogenic CO2 hypersensitivity), and that extracerebral abnormalities were not the primary cause of CSR (9). However, subsequent reports (12eC15) have failed to confirm a relationship between the site and size of a stroke and the presence of CSA. Our data are in agreement with these latter studies because we did not find any relationship between stroke location and the presence of CSA. Moreover, our study of 93 subjects is the largest study of its kind to address this issue. Only four patients in this study had bilateral cerebral brain lesions. Of these, one patient had CSA but three did not. Parra and colleagues (30) have shown that CSA is frequent immediately after a stroke, but decreases in frequency several months later. These observations suggest that strokes do predispose to CSA, at least in the immediately post-stroke period. Accordingly, because CSA was present in 14 of our patients with stroke who had normal LV systolic function, it is likely that neurologic lesions caused by the strokes contributed to the pathogenesis of their CSA. However, because we found no relationship between the type or location of strokes and the presence of CSA, it is difficult to speculate on the nature of the neurologic lesion that predisposed to CSA in these subjects. Further work will be required to determine the causes of CSA in patients with stroke and without LV dysfunction.
Our data do not allow us to say whether CSA was present in our patients before they had a stroke or whether it developed afterwards. Many of our patients had both some central and some obstructive events, raising the possibility that, in some cases, central events predisposed to the development of obstructive events and vice versa (24). However, our study was not designed to investigate this possibility. Further research will be required to determine whether there is any pathophysiologic interaction between central and obstructive respiratory events during sleep in patients with strokes.
In conclusion, our data indicate that CSA is relatively common in patients with stroke, and that its presence is related to hypocapnia and impaired LV systolic function. These findings also shed light on a long-standing controversy as to whether CSR is related primarily to cardiac or neurologic disease. We found that, among patients with stroke and CSA, the presence of a CSR pattern with long hyperpnea and cycle durations, and a gradual rise to peak VT during hyperpnea is associated with LV systolic dysfunction, but is not related to the location or type of stroke. In fact, we found that hyperpnea duration was highly sensitive and specific for detecting the presence of an LVEF of 40% or less. Moreover, in all of these patients, LV systolic dysfunction was occult. These data indicate that the presence of CSA with a CSR pattern is more closely associated with LV systolic dysfunction than it is with the stroke per se. However, in stroke patients with CSA, but without a CSR pattern, it is likely that the stroke itself is playing a role in the pathogenesis of CSA, but it is not clear how. Therefore, in patients with stroke, the presence of CSA with a CSR pattern appears to be a sign of underlying LV systolic dysfunction. This finding is of considerable clinical significance. Approximately one third of patients suffering a stroke subsequently die of heart disease (31). Thus, the presence of CSA, particularly in association with a CSR pattern, should raise suspicion of the presence of asymptomatic LV systolic dysfunction, the treatment of which could improve prognosis in such patients (32).
REFERENCES
Nachtmann A, Siebler M, Rose G, Sitzer M, Steinmetz H. Cheyne-Stokes respiration in ischemic stroke. Neurology 1995;45:820eC821.
Bassetti C, Aldrich MS, Chervin RD, Quint D. Sleep apnea in patients with transient ischemic attack and stroke: a prospective study of 59 patients. Neurology 1996;47:1167eC1173.
Bradley TD, Phillipson EA. Central sleep apnea. Clin Chest Med 1992;13:493eC505.
Solin P, Roebuck T, Johns DP, Walters EH, Naughton MT. Peripheral and central ventilatory responses in central sleep apnea with and without congestive heart failure. Am J Respir Crit Care Med 2000;162:2194eC2200.
Javaheri S. A mechanism of central sleep apnea in patients with heart failure. N Engl J Med 1999;341:949eC954.
Xie A, Rutherford R, Rankin F, Wong B, Bradley TD. Hypocapnia and increased ventilatory responsiveness in patients with idiopathic central sleep apnea. Am J Respir Crit Care Med 1995;152:1950eC1955.
Javaheri S, Parker TJ, Liming JD, Corbett WS, Nishiyama H, Wexler L, Roselle GA. Sleep apnea in 81 ambulatory male patients with stable heart failure: types and their prevalences, consequences, and presentations. Circulation 1998;97:2154eC2159.
Sin DD, Fitzgerald F, Parker JD, Newton G, Floras JS, Bradley TD. Risk factors for central and obstructive sleep apnea in 450 men and women with congestive heart failure. Am J Respir Crit Care Med 1999;160:1101eC1106.
Heyman A, Birchfield RI, Sieker HO. Effects of bilateral cerebral infarction on respiratory center sensitivity. Neurology 1958;8:694eC700.
Brown HW, Plum F. The neurologic basis of Cheyne-Stokes respiration. Am J Med 1961;30:849eC869.
Karp HR, Heyman A. Cerebral circulation and function in Cheyne-Strokes respiration. Neurology 1961;30:849eC869.
Lee MC, Klassen AC, Resch JA. Respiratory pattern disturbances in ischemic cerebral vascular disease. Stroke 1974;5:612eC616.
North JB, Jennett S. Abnormal breathing patterns associated with acute brain damage. Arch Neurol 1974;31:338eC344.
Bassetti C, Aldrich MS, Quint D. Sleep-disordered breathing in patients with acute supra- and infratentorial strokes: a prospective study of 39 patients. Stroke 1997;28:1765eC1772.
Kaneko Y, Hajek VE, Zivanovic V, Raboud J, Bradley TD. Relationship of sleep apnea to functional capacity and length of hospitalization following stroke. Sleep 2003;26:293eC297.
Cheyne J. A case of apoplexy in which the freshy part of the heart was converted into fat. Dublin Hosp Rep 1818;2:216eC223.
Hall MJ, Xie A, Rutherford R, Ando S, Floras JS, Bradley TD. Cycle length of periodic breathing in patients with and without heart failure. Am J Respir Crit Care Med 1996;154:376eC381.
Otterstad JE, Froeland G, St. John Sutton M, Holme I. Accuracy and reproducibility of biplane two-dimensional echocardiographic measurements of left ventricular dimensions and function. Eur Heart J 1997;18:507eC513.
Rechtschaffen A, Kales AA. A manual of standardized terminology, techniques and scoring system for sleep stages of human subjects. Los Angeles, CA: UCLA Brain Information Service/Brain Research Institute; 1968.
Sleep Disorders Atlas Task Force of the American Sleep Disorders. EEG arousals: scoring rules and examples: a preliminary report. Sleep 1992;15:173eC184.
Chadha TS, Watson H, Birch S, Jenouri GA, Schneider AW, Cohn MA, Sackner MA. Validation of respiratory inductive plethysmography using different calibration procedures. Am Rev Respir Dis 1982;125:644eC649.
Naughton M, Benard D, Tam A, Rutherford R, Bradley TD. Role of hyperventilation in the pathogenesis of central sleep apneas in patients with congestive heart failure. Am Rev Respir Dis 1993;148:330eC338.
Gould GA, Whyte KJ, Airlie AA, Cateral JR, Shapiro CM, Douglas NJ. The sleep hypopnea syndrome. Am Rev Respir Dis 1988;137:895eC898.
Tkacova R, Niroumand M, Lorenzi-Filho G, Bradley TD. Overnight shift from obstructive to central apneas in patients with heart failure: role of PCO2 and circulatory delay. Circulation 2001;103:238eC243.
Hanly P, Zuberi N, Gray R. Pathogenesis of Cheyne-Stokes respiration in patients with congestive heart failure: relationship to arterial PCO2. Chest 1993;104:1079eC1084.
Wessendorf TE, Teschler H, Wang YM, Konietzko N, Thilmann AF. Sleep-disordered breathing among patients with first-ever stroke. J Neurol 2000;247:41eC47.
Lorenzi-Filho G, Azevedo ER, Parker JD, Bradley TD. Relationship of PaCO2 to pulmonary wedge pressure in heart failure. Eur Respir J 2002;19:37eC40.
Solin P, Bergin P, Richardson M, Kaye DM, Walters EH, Naughton MT. Influence of pulmonary capillary wedge pressure on central apnea in heart failure. Circulation 1999;99:1574eC1579.
Lanfranchi PA, Somers VK, Braghiroli A, Corra U, Eleuteri E, Giannuzzi P. Central sleep apnea in left ventricular dysfunction: prevalence and implications for arrhythmic risk. Circulation 2003;107:727eC732.
Parra O, Arboix A, Bechich S, Garcia-Eroles L, Montserrat JM, Lopez JA, Ballester E, Guerra JM, Sopena JJ. Time course of sleep-related breathing disorders in first-ever stroke or transient ischemic attack. Am J Respir Crit Care Med 2000;161:375eC380.
Hardie K, Hankey GJ, Jamrozik K, Broadhurst RJ, Anderson C. Ten-year survival after first-ever stroke in the Perth Community Stroke Study. Stroke 2003;34:1842eC1846.
The SOLVD Investigators. Effect of enalapril on mortality and the development of heart failure in asymptomatic patients with reduced left ventricular ejection fractions. N Engl J Med 1992;327:685eC691.(Cherdchai Nopmaneejumrusl)