Randomized Controlled Trial of Noninvasive Positive Pressure Ventilation (NPPV) Versus Servoventilation in Patients with CPAP-Induced Central Sleep Apnea (Complex Sleep Apnea)

Abstract

Study Objectives:

To compare the treatment effect of noninvasive positive pressure ventilation (NPPV) and anticyclic servoventilation in patients with continuous positive airway pressure (CPAP)-induced central sleep apnea (complex sleep apnea).

Design:

Randomized controlled trial.

Setting:

Sleep center.

Patients:

Thirty patients who developed complex sleep apnea syndrome (CompSAS) during CPAP treatment.

Interventions:

NPPV or servoventilation.

Measurements and Results:

Patients were randomized to NPPV or servo-ventilation. Full polysomnography (PSG) was performed after 6 weeks. On CPAP prior to randomization, patients in the NPPV and servoventilator arm had comparable apnea-hypopnea indices (AHI, 28.6 ± 6.5 versus 27.7 ± 9.7 events/h (mean ± standard deviation [SD])), apnea indices (AI,19 ± 5.6 versus 21.1 ± 8.6 events/h), central apnea indices (CAI, 16.7 ± 5.4 versus 18.2 ± 7.1 events/h), oxygen desaturation indices (ODI,17.5 ± 13.1 versus 24.3 ± 11.9 events/h). During initial titration NPPV and servoventilation significantly improved the AHI (9.1 ± 4.3 versus 9 ± 6.4 events/h), AI (2 ± 3.1 versus 3.5 ± 4.5 events/h) CAI (2 ± 3.1 versus 2.5 ± 3.9 events/h) and ODI (10.1 ± 4.5 versus 8.9 ± 8.4 events/h) when compared to CPAP treatment (all P < 0.05). After 6 weeks we observed the following differences: AHI (16.5 ± 8 versus 7.4 ± 4.2 events/h, P = 0.027), AI (10.4 ± 5.9 versus 1.7 ± 1.9 events/h, P = 0.001), CAI (10.2 ± 5.1 versus 1.5 ± 1.7 events/h, P < 0.0001)) and ODI (21.1 ± 9.2 versus 4.8 ± 3.4 events/h, P < 0.0001) for NPPV and servoventilation, respectively. Other sleep parameters were unaffected by any form of treatment.

Conclusions:

After 6 weeks, servoventilation treated respiratory events more effectively than NPPV in patients with complex sleep apnea syndrome.

Citation:

Dellweg D; Kerl J; Hoehn E; Wenzel M; Koehler D. Randomized controlled trial of noninvasive positive pressure ventilation (NPPV) versus servoventilation in patients with CPAP-induced central sleep apnea (complex sleep apnea). SLEEP 2013;36(8):1163-1171.

INTRODUCTION

Complex sleep apnea syndrome (CompSAS) is characterized by the development of frequent central apneas or a Cheyne-Stokes respiratory pattern after initial application of continuous positive airway pressure (CPAP)^1–4 and thus is sometimes also referred to as CPAP emergent central apnea.⁵ This phenomenon does not only appear after commencement of CPAP therapy, but might also emerge after resolution of the obstructive sleep apnea syndrome (OSAS) achieved by tracheostomy⁶ or the use of mandibular advancement devices.⁷ The syndrome is thought to be caused by unstable, chemosensitive ventilatory control with a change of ventilatory drive.⁸ CompSAS appears to resolve spontaneously over time in some but not all patients during continued CPAP therapy.^9–12 The prevalence of CompSAS is thought to be within a range of 1.5-20% of patients who receive treatment for OSAS.^{1,4,9,11,13–16}

For the treatment of central sleep apnea, noninvasive positive pressure ventilation (NPPV) as well as flow-targeted anticyclic servoventilation have been investigated. In night-to-night comparisons either form of therapy reduces the number of central apneas, with a significant superiority of the latter form of treatment.^14,17,18 One study, however, described an increase in the number of central events when NPPV ventilation is compared to CPAP treatment in patients with central sleep apnea.¹⁹ The optimal treatment for CompSAS is therefore still controversial because comparative long-term studies to date are missing.^2,20 The goal of the current study was to compare NPPV with flow-targeted anticyclic servo-ventilation in controlling sleep disordered breathing in patients with CompSAS. We hypothesized that either form of treatment is equally effective, based on improvement of apnea-hypopnea index (AHI) as the primary end point.

METHODS

Study Patients

This was a prospective randomized controlled trial. The trial was approved by the responsible Institutional Review Board and consent was obtained for each enrolled patient prior to randomization. Patients were informed to receive one of two optional therapies by means of randomization. In addition the trial was registered in the National Institutes of Health-associated database (NCT01609244). This study recruited 37 eligible patients in whom CompSAS developed while they were being treated with CPAP. CompSAS was diagnosed on routine follow-up polysomnography (PSG), 6 weeks after commencement of CPAP therapy.

Inclusion criteria

1. AHI ≥ 15 during the initial PSG with a predominance of obstructive events according to American Academy of Sleep Medicine (AASM) criteria.²¹

2. AHI ≥ 15 on CPAP therapy after 6 weeks of CPAP treatment during a follow-up PSG with a predominance of central events according to AASM criteria.²¹

Exclusion criteria

1. Significant comorbidity (severe or unstable neurological, metabolic, respiratory, or cardiac disease), such as unstable chronic obstructive pulmonary disease, congestive heart failure, or renal insufficiency

2. Patients on rotating shift work

3. Opioid use

4. Restless legs syndrome

CPAP Treatment

All patients who were later included were initially titrated to CPAP by use of an autotitration mode (ResMed S8 autoset spirit II, ResMed, San Diego, California, USA). During a second therapeutic night CPAP was set to a fix mode using the 95^th pressure percentile of the autotitration night as recommended by Teschler et al.²² during the first half of the night followed by a pressure reduction of 2 cm H₂O during the second half of the night. Patients were discharged with a pressure selected as high as possible to efficiently treat obstructive events and as low as possible in order to avoid the emergence of central apneas.³

Study Protocol

Patients in whom CompSAS refractory to spontaneous resolution was diagnosed on follow- up PSG at 6 weeks were eligible for this study. Patients who consented to participate were randomized to receive NPPV or servoventilation (Figure 1). After titration to the respective device, patients were told to use their device nightly during sleep. Patients could contact our sleep center if they perceived treatment-related problems. Patients were not actively contacted during the treatment period and were reevaluated by PSG at the end of the 6^th week. Therapy compliance was recorded from the built-in h-meter of the different ventilators, which were zeroed at study entry. Titration for either device was performed during overnight in-house attended polygraphy that included all standard PSG leads with the exception of electroencephalograph, muscle, and video recordings.

Figure 1.

Patient flow, allocation to and dropout from the two treatment arms (noninvasive positive pressure ventilation (NPPV) and servoventilation).

Patients randomized to the NPPV arm received a standard NPPV ventilator (Somnovent ST22, Weinmann, Hamburg, Germany) programmed to deliver bilevel positive airway pressure ventilation with a backup rate just below (one to two breaths) the average sleep related respiratory rate, registered during CPAP therapy. The inspiratory trigger of the device was set to the most sensitive level, but was lowered in case of autotriggering. The expiratory trigger was set to comfort the patient during a breathing trial while awake. The expiratory positive airway pressure (EPAP) was manually titrated to keep the upper airway open during sleep. Patients' breathing efforts, not responded to by the ventilator (untriggered breath), were used as an additional criterion for obstruction of the upper airway. The inspiratory pressure was increased manually until optimal treatment of central apneas was achieved or the patient did not tolerate any further increase. The inspiratory pressure was manually titrated by the sleep technician during the polygraphic measurements at night. Measurements were reviewed on the following morning by a sleep physician who either confirmed a sufficient treatment effect or recommended different settings for the next treatment night. Patients were discharged if they subjectively were able to sleep with the selected settings and settings were effective according to overnight measurements.

Patients randomized to the servoventilation arm received a dynamic servoventilator (Somnovent CR, Weinmann, Hamburg, Germany). This device works on three different pressure levels: the inspiratory positive airway pressure (IPAP), which represents the pressure level during inspiration, the EPAP, which represents the pressure level during the first phase of expiration, and the end-expiratory positive airway pressure (EEPAP). The ventilator has the capability to compensate central apneas and periodic breathing by anticyclic regulation of the inspiratory and expiratory pressure and to treat upper airway obstruction by autoadjustment of the EEPAP. Pressure and flow is continuously measured with a pressure transducer and a pneumotachygraph function of the turbine within the device. Snoring is detected by high-frequency variations of the pressure signal. To detect changes in ventilation, the ventilator software compares the actual breath with the average value of two moving time windows; both are equally weighted. The reference is calculated as the average value of the last 2-min epoch prior to the actual breath equally weighed with the average value of the total previous therapy time. The differentiation between central and obstructive events without spontaneous breathing occurs on the flow response to mandatory ventilation. Discrimination of obstructive from central events during spontaneous breathing occurs on the basis of the snoring signal or the recognition of flattening within the flow signal.

The device is programmed to keep flow and minute ventilation constant. In case of obstructive events the ventilator will increase the EEPAP in order to splint the upper airway, similar to an automatic CPAP (APAP) device. In case of central flow limitations the device will increase the inspiratory support by increasing IPAP and decreasing EPAP. During phases of hyper-ventilation IPAP can be decreased to zero above end-expiratory level in which case the ventilator behaves in a manner similar to a pure APAP device with pressure relief during early expiration. The backup frequency can be set to automatic detection or might be chosen manually.

Sleep Studies

Standard PSG was performed during the diagnostic and follow-up studies, whereas polygraphy was performed during the NPPV and servoventilation titration nights. We used a digital diagnostic system (Somnoscreen Plus PSG, Somnomedics, Randersacker, Germany). PSG measurements included all standard PSG leads as recommended by the AASM,²³ and polygraphic measurements included all standard leads with the exception of EEG, muscle, and video recordings. Sleep staging and respiratory events were scored according to standard methods.²¹ An apnea was defined as decrease of airflow of more than 90% relative to baseline for more than 10 sec. This included isolated apneas as well as apneas occurring during periods of periodic breathing (Figure 2). Obstructive apnea was defined as the absence of airflow associated with continued thoracoabdominal excursions. Apneas that were usually defined as mixed apneas according to AASM criteria²¹ were counted as obstructive apneas if the phase angle of the thoracoabdominal belts was > 45 degrees, and they were counted as central apneas if the phase angle was ≤ 45 degree.^24,25 Hypopneas were defined as a reduction in airflow of more than 30% and associated with more than 4% drop in arterial oxygen saturation. All studies were scored by a board-certified technician and reviewed by a board-certified physician. Blinding was attempted; however, different characteristics of the pressure curves disclosed the different treatment forms upon analysis.

Figure 2.

Polysomnographic tracing of a patient who developed complex sleep apnea during continuous positive airway pressure treatment. The screenshot represents a period of 5 min where periodic central apneas are marked in red within the flow channel, corresponding desaturations are displayed in the oxygen saturation (SaO₂) channel and are marked in red.

Statistical Analysis

We assumed noninferiority of any of the two therapies. Power analysis based on the results on short-term comparisons of NPPV versus servoventilation²⁶ suggested a sample size of N = 28 patients (effect size 1.45, alpha-error 0.05, power 0.95). We assumed a dropout rate during the 6-week observation period of 20% and included 37 patients into the study. Differences between the two study groups prior to randomization were evaluated by means of an unpaired t-test. To compare longitudinal results (CPAP after 6 weeks, NPPV/servoventilation during titration and 6-week follow-up) between the two groups we used one-way analysis of variance (ANOVA). Post hoc analysis was carried out using the Scheffé procedure in case of equal variances and the Games-Howell test in case of unequal variances. A P-value less than 0.05 was considered significant. Statistical analysis was performed with the SPSS software version 20 (International Business Machines Corporation (IBM), Armonk, NY, USA).

RESULTS

Basic demographic data and information about pertinent comorbidities are shown in Table 1. Table 2 shows PSG data of the 30 patients who completed the study. All patients showed almost complete resolution of their obstructive events after 6 weeks of CPAP therapy but developed a considerable number of central apneas (Table 2). The applied CPAP pressure was 9.9 cm H₂O and 9.3 cm H₂O for the patients randomized to NPPV and servoventilation, respectively (P = 0.39). Thirty-seven eligible patients were randomized to NPPV or servoventilation (Figure 1). Three patients in the NPPV group and one patient in the servoventilation group did not tolerate their treatment and withdrew from further participation in this study (Figure 1). One patient in the NPPV group withdrew during the treatment period due to skin lesions from the mask interface, two patients in the servoventilation group withdrew during treatment (one due to aerophagia and one due to problems with the mask interface). A total of 30 patients completed the study (15 patients in each group) and their data were used for further analysis.

Table 1.

Basic demographic data

Table 2.

Polysomnographic data prior to and 6 weeks after commencement of continuous positive airway pressure therapy for the 30 patients who completed the study

The inspiratory and expiratory pressures applied in the NPPV group were on average 15.8 ± 2.7 cm H₂O and 7.7 ± 1.5 cm H₂O, respectively. Backup respiratory rate was set to 13.2 ± 2.5 breaths per min. For all patients in the servoventilation arm, the upper limit of the inspiratory pressure was set to 20.4 cm H₂O, and the lower level of the expiratory pressure was set to 8.3 ± 1.6 cm H₂O. Respiratory rate was set to the automatic mode. Titration in the NPPV arm required a titration period of 2.3 (± 0.7) nights whereas servoventilation titration was accomplished in 1.4 (± 0.6) nights (P = 0.001).

Either group achieved a significant reduction of their AHI, their apnea index, central apnea index, and desaturation index during the titration studies (Table 3). Upon follow-up investigation at the end of the treatment period we observed a significant increase in the AHI, apnea index, central apnea index, and desaturation index in the NPPV group when compared with the NPPV titration study or any of the studies, where servoventilation was applied (Table 3 and Figure 3). Tendentially respiratory arousal indices improved during NPPV (P = 0.19) and servoventilation (P = 0.09) when compared with CPAP therapy, but changes failed to reach a significant level. Overall sleep parameters were not affected by any of the treatment forms (Table 3).

Table 3.

Polygraphic and polysomnographic data of the two treatment arms during continuous positive airway pressure treatment prior to randomization as well as during titration to the randomized treatment and 6 weeks after commencement of noninvasive positive pressure ventilation or servoventilation

Figure 3.

Trend of the respiratory parameters of the two randomized arms that differed significantly at the end of the treatment period. Sigificance was determined by one-way analysis of variance. Post hoc analysis was carried out using the Scheffé procedure in case of equal variances and the Games-Howell test in case of unequal variances. CPAP, continuous positive airway pressure; NPPV, noninvasive positive pressure ventilation.

DISCUSSION

Respiratory Parameters

The results of this study show that NPPV as well as servoventilation can significantly compensate for central apneas and sufficiently treat obstructive apneas during initial application. After a 6-week treatment period, however, NPPV treatment is inferior to servoventilation in terms of suppressing central events as well as oxygen desaturations (Table 3 and Figure 3). Previous studies in patients with heart failure and CompSAS were able to show that central apneas are sufficiently treatable with NPPV during initial titration.^14,17,18,27 Fietze et al.²⁸ measured a remaining central apnea index of 16.1/h after 6 weeks of NPPV therapy in patients with chronic heart failure; indices from the initial titration period, however, were not reported. Our data are suggestive of some dynamic process during NPPV treatment in our patient cohort that does not occur during servoventilation.

Central apneas occur whenever the level of carbon dioxide falls below the apneic threshold ²⁹ that is mediated by carotid body chemoreceptors.³⁰ During sleep there is complete loss of the wakefulness stimulus to breathing.³¹ Thus, breathing regulation after sleep onset is exclusively regulated by the feedback mechanisms of the autonomous nervous system. In this context the term ‘loop gain’ has been introduced.^32,33 Loop gain can be described mathematically as the ratio of magnitude of the (hyperpneic) response to the perturbation over the (hypopneic) perturbation itself.³² This ratio has to be lower than 1 in order to achieve a steady correction. Loop gain describes the sensitivity of the entire feedback loop controlling ventilation and is influenced by the two variables: controller gain and plant gain. Controller gain refers to the chemorespon-siveness to oxygen and carbon dioxide levels in the blood, whereas plant gain describes how effective the response (here ventilation) affects the control variables carbon dioxide and oxygen within the blood. Patients with untreated OSAS increase their ventilation more briskly and pronouncedly than treated patients when given a normoxemic hypercapnic challenge, thereby exhibiting an increased controller gain.^16,34 An increased ventilatory response to a carbon dioxide rebreathing challenge in CPAP users has been proposed by Moura et al.⁸ but could not be seen in a study by Loewen et al.³⁴ where a reduction in the ventilatory response to carbon dioxide was observed after CPAP treatment of 1 month duration. Although Moura et al.⁸ conducted their experiments in awake patients, Loewen et al.³⁴ used a setup whereby patients were given different reduced oxygen and/or enriched carbon dioxide gas mixtures while sleeping with a CPAP mask. The study by Loewen et al.³⁴ also suggests that correction of hypoxemia by means of CPAP in patients with OSAS reduces the controller gain, thus highlighting the importance of carbon dioxide and oxygen tension in the regulation of the loop gain mechanisms. To date no study has evaluated such mechanisms in patients who developed CompSAS upon CPAP commencement. It is conceivable that patients with CompSAS have a very high controller gain prior to CPAP treatment and this gain does not sufficiently decrease as described in patients who do not develop CompSAS during CPAP treatment. ^34,35 In conjunction with the decreased airway resistance upon CPAP application³⁶ this could be a potential mechanism or cofactor for the ventilatory instability and emergence of CompSAS.

Servoventilation differs from NPPV in respect to the method of pressure application. Although servoventilation is targeted to deliver increasing pressurization levels during apneic events, NPPV delivers the programmed inspiratory pressure invariably.

Experimental augmentation of the controller gain by means of proportional assist ventilation during NREM sleep enables generation of periodic breathing.³⁷ NPPV and servoventilation behave differently within and outside the perturbation period. Either form of ventilation maintains lung insufflation during apneic episodes. At the end of apnea and with commencement of respiration, however, NPPV will continue to augment the patient's respiratory response, resulting in hyperventilation and thus provoking the next apneic event. The adverse effect of NPPV therefore does not occur during the perturbation period (apnea) but rather beforehand by augmentation of the controller gain in the nonapneic phase.

The fact that the difference we observed between NPPV and servoventilation did not appear until 6 weeks of treatment merits further discussion. We did not measure sleep during the titration studies. Although patients reported that they were able to sleep with their devices, we cannot completely exclude a difference in sleep quality and a lack of central apneas due to a lack of sleep. Other investigators, however, have shown that breathing and sleep parameters improved already during the first night of NPPV or servoventilation.^17,18,38 More likely are alterations in the carbon dioxide level. The delta between the eupneic carbon dioxide and the carbon dioxide level causing apnea is called carbon dioxide reserve and is the major determinant of ventilatory stability.³ NPPV is known to lower carbon dioxide levels in patients with OSAS and mild hypercapnea^39,40 and will very likely do so in nonhypercapnic individuals. Blood gas homeostasis is maintained by the interaction of the respiratory and the renal system. Therapeutic reduction of carbon dioxide over periods of weeks is usually accompanied by a reduction in bicarbonate as well that might even persist during daytime.⁴⁰ Hypocapnia elicts not only bicarbonate loss within the blood but also causes loss of intracellular bicarbonate over time⁴¹ and a lower carbon dioxide equilibrium. This translates into a decreased carbon dioxide reserve and a higher propensity to generate central apneas. In conditions with permanent carbon dioxide reduction such as high altitude residence or the presence of chronic heart failure, it is known that the apnea threshold can reset to lower levels.^4,6,42 The carbon dioxide reducing effect of NPPV however does not last for 24 h and it is known that after NPPV induced overnight carbon dioxide reduction, carbon dioxide levels increase again during the following day.⁴³ Because bicarbonate compensation of shifting carbon dioxide levels is known to occur slowly,⁴⁴ patients are at risk of having an alkalotic state during NPPV followed by a rising pH throughout the following day. Although induction of acidosis is known to decrease the incidence of central apneas,^45,46 induction of alkalosis by NPPV is likely to carry the risk of inducing central apneas. Worsening of central sleep apnea, especially with increasing inspiratory pressure levels, has been described by other investigators¹⁹ and a decrease in carbon dioxide below the apnea threshold was the proposed mechanism.⁴⁷ Our study design did not include determination of carbon dioxide levels, thus a proven concept to explain the increasing number of central events during NPPV cannot be given currently and our explanations of apnea emergence after 6 weeks of NPPV remain speculative.

Sleep

Overall we observed no differences in any of the sleep parameters (Table 3). Sleep stage distribution and arousal indices are comparable to previous investigations looking at NPPV,²⁷ servoventilation,^5,48 or both^{14,17,18,28,38} in different patient populations. The proportion of slow wave sleep during NPPV varied extensively (1-33% of total sleep time) in two other studies.^49,50 To some extent it appears surprising that a reduction of respiratory events does not decrease the number of respiratory and number of overall arousals. Our arousal indices during NPPV and servoventilation treatment are comparable to results of previous investigations.^14,18 We suspect that therapy -associated effects such as pressure changes within the mask,⁵¹ alternating pressure changes between mask and skin,⁵² or noise from the ventilator or mask interface have caused additional arousals. Our suspicion goes along with investigations of Javaheri et al.,⁵ who described an increased arousal index during servoventilation when compared with CPAP treatment despite a reduction in respiratory events and improved slow wave and REM sleep

Limitations

We measured respiratory parameters but not sleep parameters during titration nights to NPPV and servoventilation, respectively. Therefore, we cannot exclude that manual titration occurred in response to apneas while the patient was awake. Although Cheyne-Stokes respiration might occur while the patient is awake,⁵³ CompSAS is a phenomenon that during the correction of sleep related apneas by means of positive pressure treatment and does not resemble Cheyne-Stokes breathing.⁹ Thus, we deem that it is unlikely that patients were overtreated due to central apneas while awake.

The size of our study is relatively small and we cannot exclude that nonprimary parameters had shown significant differences if the sample size had been larger.

This study was randomized and blinding was impossible because different characteristics of the pressure curves disclosed the type of treatment during analysis.

CONCLUSION

In this randomized controlled trial of NPPV versus servoven-tilation over an observational period of 6 weeks involving patients with CompSAS, both NPPV and servo-ventilation were able to suppress central and obstructive events during initial titration. However at the 6-week follow-up servoventilation was superior in this regard. Sleep was not affected by any of the interventions. We suspect changes in carbon dioxide homeostasis induced by NPPV but not by servoventilation for the different outcome after 6 weeks.

DISCLOSURE STATEMENT

This was not an industry supported study. Dr. Dellweg has consulted for Weinmann (Hamburg, Germany) and has participated in speaking engagements for ResMed (Martinried, Germany). Dr. Kerl has participated in speaking engagements for Weinmann. The other authors have indicated no financial conflicts of interest.