Published ahead of print on August 19, 2003, doi:10.1164/rccm.200212-1476OC
© 2003 American Thoracic Society A Randomized Controlled Trial of Adaptive Ventilation for Cheyne–Stokes Breathing in Heart FailureOxford Sleep Unit and Respiratory Trials Unit, Oxford Centre for Respiratory Medicine, Churchill Hospital Site, Oxford Radcliffe Hospital, Oxford, United Kingdom Correspondence and requests for reprints should be addressed to Justin C. T. Pepperell, M.R.C.P., D.A., Taunton and Somerset NHS Trust, Musgrove Park Hospital, Taunton, Somerset, TA1 5DA, U.K. E-mail: justinpepperell{at}doctors.org.uk
Heart failure is associated with Cheyne–Stokes breathing, which fragments patients' sleep. Correction of respiratory disturbance may reduce sleep fragmentation and excessive daytime sleepiness. This randomized prospective parallel trial assesses whether nocturnal-assist servoventilation improves daytime sleepiness compared with the control. A total of 30 subjects (29 male) with Cheyne–Stokes breathing (mean apnea–hypopnea index 19.8 [SD 2.6] and stable symptomatic chronic heart failure (New York Heart Association Class II–IV) were treated with 1 month's therapeutic (n = 15) or subtherapeutic adaptive servoventilation. Daytime sleepiness (Osler test) was measured before and after the trial with change in measured sleepiness the primary endpoint. Secondary endpoints included brain natriuretic peptide levels and catecholamine excretion. Active treatment reduced excessive daytime sleepiness; the mean Osler change was +7.9 minutes (SEM 2.9), when compared with the control, the change was -1.0 minutes (SEM, 1.7), and the difference was 8.9 minutes (95% confidence interval, 1.9–15.9 minutes; p = 0.014, unpaired t test). Significant falls occurred in plasma brain natriuretic peptide and urinary metadrenaline excretion. We conclude that adaptive servoventilation produces an improvement in excessive daytime sleepiness in patients with Cheyne–Stokes breathing and chronic heart failure. This study suggests improvements in neurohormonal activation with this treatment.
Key Words: sleep apnea • brain natriuretic peptide • placebo • positive airway pressure • randomized controlled trial Cheyne–Stokes breathing is detectable in between 36 and 57% of patients with chronic heart failure (1, 2) and is associated with a number of secondary consequences. Each apnea is associated with arousal from sleep (3), which fragments the normal sleep process (2, 4). The syndrome of heart failure includes lethargy, fatigue and sleepiness, which are plausibly related to this. The repetitive apnea and arousal cycle also induces cyclic sympatheticoadrenal activation (5), an accepted predictor of mortality in heart failure (6). It is possible that correction of Cheyne–Stokes breathing may improve both daytime symptoms and the heart failure syndrome itself. Adaptive servoventilation (ASV, Autoset-CS; ResMed Corp, San Diego, CA) is a novel therapy that could produce these benefits by correcting Cheyne–Stokes breathing. The ventilator aims to correct episodes of apnea and hypoventilation and subsequently reduce arousal, sleep fragmentation, and sympathetic activation. In observational studies, ASV is technically effective in controlling central apneas during sleep (7). However, there have been no randomized trials to confirm that controlling Cheyne–Stokes breathing is associated with clinical patient benefit. We have performed a randomized prospective comparison of therapeutic ASV with a subtherapeutic control in patients with stable chronic heart failure and Cheyne–Stokes breathing to assess whether ASV improves objectively measured excessive daytime sleepiness (primary end point), or plasma brain (B-type) natriuretic peptide (BNP) levels (a marker of heart failure severity), urinary catecholamine excretion, self-reported health status, and steering simulator performance (secondary end points). Statistical analysis was performed on "intention to treat" basis using SPSS version 10 (SPSS Inc., Chicago, IL). Unpaired t tests were used for normally distributed data and Mann–Whitney U tests were used for nonnormally distributed data. Results are mean (SEM) for normally distributed data and median (interquartile range) for nonnormally distributed data. Some of the results have been previously reported in an abstract (8).
This study was a prospective, parallel, randomized, double-blind trial performed at the Sleep and Respiratory Trials Units, Oxford Radcliffe Hospitals, UK. Additional detail on the methods is provided in an online supplement. Patients identified with a primary diagnosis of heart failure and stable drug therapy were contacted to invite them to participate. Current heart failure symptoms (New York Heart Association Class II–IV) were confirmed in subjects with more than 10 saturation dips greater than 3% per hour on overnight home pulse oximetry (Minolta 3i; AVL, Shaffhausen, Switzerland). These subjects then consented for the main study (approved by the Central Oxford Research Ethics Committee, number 99.218) and underwent overnight polysomnography performed by standard methods (9), health status questionnaires, and 24-hour urine collections. Additional details are provided in the online supplement. After polysomnography, subjects fulfilling the study entry criteria remained in hospital for a further 2 days to complete baseline assessments and institute the trial therapy. Blood samples were taken at 8:30 A.M., and patients underwent Osler tests of daytime vigilance at 9:00 A.M. and 2:00 P.M. After the 9:00 A.M. Osler test, subjects underwent training on the steering simulator before proceeding to a 30-minute test run. Two-dimensional and M-mode echocardiography was performed. During their stay, subjects learnt about the ventilator and were fitted with a mask before a daytime acclimatization session and then received subtherapeutic and therapeutic Autoset CS (ResMed Corp) on alternate nights in random order. After the second night, 4 weeks treatment with therapeutic or subtherapeutic ASV was allocated by the selection of a presealed and numbered envelope. At 4 weeks, subjects reattended for repeat assessments, and, ASV hour meters were read to calculate mean nightly usage (see further details in the online supplement). Therapeutic ASV was delivered from the Autoset CS, set to its default settings (expiratory pressure 5 cm H2O, inspiratory pressure support between 3 and 10 cm H2O, backup respiratory rate 15 breaths per minute). Subtherapeutic ASV was delivered from an identical machine, adapted to deliver a nominal 1.75 cm H2O pressure with little pressure support ventilation (minimum, 0.75; maximum, 2.75). The primary outcome was the change in the patients' objective degree of sleepiness measured using a behavioral sleep resistance challenge, the Osler test (10). Secondary outcome measures included BNP, urinary catecholamine excretion, self-reported health status (Epworth sleepiness score, short-form 36, and the chronic heart failure questionnaire), and steering simulator performance (Stowood Scientific Instruments, Oxford, UK). To ensure the study remained double blind, the specialist nurse who performed outcome assessments was unaware of the treatment assignment.
Subjects Figure 1 shows the patient flow diagram for the study. A total of 325 patients were identified by the original case screening process. After the initial telephone conversation, 223 of these patients agreed to take part in the study and underwent home overnight oximetry. Of these, 78 had more than 10 dips in SaO2 greater than 3% per hour of recording. Forty-nine of these subjects consented participate in the main study but 11 subjects withdrew before attendance (three died, two were admitted to hospital with decompensated heart failure, and six withdrew consent). Thirty-eight subjects underwent polysomnography, 8 did not have sufficient Cheyne–Stokes breathing, and the remaining 30 were randomized into the trial.
Of the 30 patients randomized, 20 had ischemic or coronary heart disease and 10 had dilated cardiomyopathy (four of these presumed secondary to hypertension), with 12 having atrial fibrillation and 14 having a past history of hypertension (Table E1). No changes were made to patients' medications during the trial. The mean apnea–hypopnea index per hour of recording was similar between groups (Tables 1 and E2). In all, 77% of apneas and hypopneas were central with a smaller proportion of mixed (19%) and obstructive events (3.1%). The majority of subjects had severely disrupted sleep with reduced amounts of sleep overall and reduced proportions of slow wave sleep, similar to previous studies using continuous positive airway pressure (CPAP) to treat Cheyne–Stokes breathing in heart failure (11) (Tables 1 and E2).
During the trial, four subjects withdrew in the subtherapeutic group (detailed in Figure 1) and none withdrew in the therapeutic group. The data for all outcomes were analyzed on intention to treat with baseline data used for follow-up with no change assumed in subjects who withdrew. The subjects' baseline characteristics and compliance data are shown in Tables 1 and E1. The study groups were well matched for baseline characteristics.
Primary Outcome Analysis: Change in Objective Sleepiness
Secondary Outcome Analyses Serum BNP levels. Therapeutic ASV produced a reduction in serum BNP when compared with the subtherapeutic control. The median change in BNP with therapeutic ASV was -56 pg/ml (interquartile range, -238 to -16), with no change in subtherapeutic ASV (0, interquartile -24 to +71; p = 0.001, Mann–Whitney U test).
Urinary catecholamine excretion.
Self-reported health status, subjective sleepiness, and steering simulator performance. There were no statistically significant changes in any of the questionnaire assessments of health status or sleepiness with either therapeutic or subtherapeutic treatment (Table E3). Two of the 30 subjects did not drive, and these subjects did not perform the driving simulator assessments. There were no statistically significant changes in any of the driving assessments (Table E3).
Compliance and technical efficacy of ASV. With control of the nocturnal Cheyne–Stokes ventilation there was a reduction in daytime hyperventilation, and daytime CO2 rose in the therapeutic group from 4.9 (0.2) kPa at baseline to 5.4 (0.2) kPa at follow-up (p = 0.002, paired t test). There was no difference in 3% SaO2 dip rate with subtherapeutic ASV, with baseline value of 20.7 (2.9), follow-up value of 21.8 (4.8), and p value of 0.77, and daytime ventilation was unchanged, with baseline CO2 value of 5.0 (0.2), follow-up value of 4.9 (0.2), and p value of 0.57. No differences were seen in primary or secondary endpoints on paired samples t tests with the subtherapeutic control machine (Tables 2, E3, and E4; Figure 2).
The Changes in Objectively Measured Sleepiness This study demonstrates that patients with Cheyne–Stokes breathing and symptomatic chronic heart failure improve the time they can remain awake when they are given therapeutic ASV. This improvement of 8.9 minutes is confirmed to be a treatment effect by comparison with a robust control, subtherapeutic ASV, which did not change measured sleepiness. The change before and after therapeutic ASV treatment was proved to be significant with paired t tests. The improvement seen with therapeutic ASV is unlikely to be due to increased compliance as there was no relationship between machine use and sleepiness improvement overall or within either the therapeutic or the subtherapeutic groups. The magnitude of this therapeutic effect is large. It is similar to the 7 minutes improvement produced by the treatment of obstructive sleep apnea with nasal CPAP therapy (12) and substantially larger than the 3 minute improvement seen with patients with narcolepsy treated with the routinely used alerting agent, Modafinil (13). In all, 77% of the 30 subjects at baseline had an abnormal Osler test result of less than 40 minutes. After therapeutic ASV, this had fallen to 40% of subjects, implying a "number needed to treat" to restore one subject to normality of only 2.5. The size of the improvement in objectively measured sleepiness with therapeutic ASV has been calculated on an intention to treat basis, with no change in sleepiness from baseline being assumed in subjects with incomplete follow up data. If the results are analyzed only in the 26 subjects with complete Osler test data, the benefit from therapeutic ASV is little changed. Change in Osler test values with therapeutic ASV was +7.9 (2.9) and with subtherapeutic ASV was -1.3 (2.4), and the difference was 9.2 (95% confidence interval, 1–14.5; p = 0.03).
It is disappointing that the large primary endpoint benefit is not associated with subjective symptom benefit on the general or disease-specific health status questionnaires or in the driving simulator performance. The energy and vitality domain of the short-form 36 is sensitive to treatment effects in obstructive sleep apnea and might have been expected to be informative in Cheyne–Stokes breathing. In fact, in this study there is a weak trend toward this index showing improvement, with an effect size of 0.3 (mean change with therapeutic ASV, 3.3; SD of baseline, 11). Unfortunately, this trial is too small to assess whether this improvement is real, and it is possible that the duration of this trial was too short to have improved these quality of life endpoints. A larger study to clarify whether such a benefit is real would be clinically important as health status improvements with an effect size of above 0.2 are generally considered worthwhile (14). Such a study would require 340 subjects ( This primary endpoint result is supported by improvements in two secondary endpoints, BNP and urinary metadrenaline excretion. At baseline the elevated BNP levels confirmed moderate to severe cardiac failure (16) (Tables 2 and E4), and the reduction in BNP with therapeutic ASV (median, -56 pg/ml; mean, -310 pg/ml) was comparable with that achieved in patients with similar severity of heart failure after their medical therapy is optimized (mean, -79.0 pg/ml) (17). These results suggest that as well as improving symptoms, ASV may improve cardiac volume loading conditions and so might lead to a reduction in cardiac morbidity (18). The change in adrenaline metabolite excretion was surprising as previous trials in this area suggest that noradrenaline metabolites are more specifically increased with sleep-disordered breathing. With this small trial and the number of comparisons performed on the secondary endpoints, this may represent a chance finding. In support of previous trials, however, there was a weak trend toward reduced noradrenaline metabolite excretion in the therapeutic group, -36.6 (21.3) µmol/mmol/creatinine, versus the subtherapeutic controls, -2.3 (13.7) µmol/mmol/creatinine, and the p value was 0.19. The improvements seen in BNP did not correlate with improved echocardiographic function. The echocardiography examinations used standard M Mode and two-dimensional views, but due to technical difficulties, ejection fraction was reported for both baseline and follow-up in 25 patients and fractional shortening in 23. Mitral regurgitation was reported in 28 of the patients: 3 severe, 10 moderate, and 15 mild. The reported ejection fraction was higher than expected compared with other trials in this area (11, 19–21) and when compared with functional class and BNP levels of the subjects studied. This may represent a bias with the methods used by the technicians performing the studies and could have limited the ability of the trial to demonstrate a change. Other trials have shown improved cardiac function over 3 months (11, 19), and it is possible that this study was too short to have shown a difference. Further large trials with more sophisticated cardiac imaging and functional assessment are needed in this area.
The Characteristics of the Studied Patients The patients studied here are typical of those with chronic heart failure. They are elderly, breathless, have objectively impaired left ventricular function, high levels of BNP, and require diuretic and vasodilator medication for symptom control (Table E1). Despite this typical clinical profile and clinically characteristic Cheyne–Stokes breathing, a significant minority of the episodes of sleeping apnea were associated with polysomnographic evidence of some pharyngeal airway collapse. Presumably, this airway collapse is due to falling pharyngeal muscle tone as the phasic afferent breathing–associated neural drive to these muscles declines as a central apnea develops (22, 23). As well as being physiologically interesting, the presence of airway collapse raises the possibility that some of the therapeutic benefits we have shown with ASV could be achieved by simply maintaining airway patency with positive airway pressure. There is a substantial literature to support positive airway pressure as a therapeutic candidate in chronic heart failure. It can sometimes improve cardiac loading conditions and ejection fraction, reduce the severity of Cheyne–Stokes breathing, and perhaps reduce clinical symptoms (11, 19–21, 24, 25). However, not all investigators have demonstrated such improvements (26). The therapeutic role on CPAP in heart failure is currently being assessed in a clinical trial comparing it with ‘best supportive care’ (27). If this trial shows a therapeutic benefit from CPAP, then comparative studies of ASV versus CPAP will be required to define their relative therapeutic efficacies and to compare their cost/benefit ratios. The patients studied here were identified by screening a population not primarily complaining of excessive sleepiness, though they all described the heart failure syndrome, which includes tiredness and exhaustion, possibly attributable to sleep disturbance. This contrasts with most trials in obstructive sleep apnea, where subjects have presented symptomatic and requesting treatment. This is confirmed by the baseline Epworth sleepiness score in this trial of 10, which is only mildly abnormal and substantially less than is seen in most obstructive sleep apnea trials (12, 28). In view of this mild sleepiness, the size of the effect we have seen on objective sleepiness is surprisingly large and might have been greater still had the trial focused on subjects complaining of greater sleepiness at baseline.
The Clinical Significance of Cheyne–Stokes Breathing This trial now shows that ASV produces clinically significant symptom improvements in heart failure patients and may also improve the cardiac loading conditions. Thus, as well as showing some therapeutic efficacy for ASV, this trial contributes to the evidence that Cheyne–Stokes breathing is responsible for significant clinical symptomatology in its own right.
Conclusions
The authors wish to thank Professor H. Watkins, Dr. O. Ormerod, Dr. Y. Bashir, Dr. A. P. Banning, Dr. K. M. Channon, Dr. I. R. Arnold, Dr. C. P. Clifford, and Dr. P. Davey for allowing us to approach their patients, and Mr. A. Prothero for his assistance with the echocardiograms.
Supported by charitable trust funds including a noncommercial donation made by ResMed UK to support research in the Oxford Sleep Unit in 1998. This article has an online supplement, which is accessible from this issue's table of contents online at www.atsjournals.org Conflict of Interest Statement: J.C.T.P. was reimbursed, by ResMed, for travel expenses (£480) to attend the American Thoracic Society Annual Conference 2002; N.A.M. has no declared conflict of interest; D.R.J. has no declared conflict of interest; B.A.L-W. has no declared conflict of interest; N.C. has no declared conflict of interest; J.R.S. has no declared conflict of interest; R.J.O.D. has no declared conflict of interest. Received in original form December 16, 2002; accepted in final form August 13, 2003
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
METHODS
= 0.05; power, 90%). If ASV does improve health status to this extent, this would be a larger "quality of life" benefit than is seen with other well-established heart failure therapies such as angiotensin-converting enzyme ACE inhibitors (
