Treatment-Emergent Central Apnea

Abstract

The purpose of this review was to describe our management approach to patients with treatment-emergent central sleep apnea (TECSA). The emergence of central sleep apnea during positive airway pressure therapy occurs in approximately 8% of titration studies for OSA, and it has been associated with several demographic, clinical, and polysomnographic factors, as well as factors related to the titration study itself. TECSA shares similar pathophysiology with central sleep apnea. In fact, central and OSA pathophysiologic mechanisms are inextricably intertwined, with ventilatory instability and upper airway narrowing occurring in both entities. TECSA is a “dynamic” process, with spontaneous resolution with ongoing positive airway pressure therapy in most patients, persistence in some, or appearing de novo in a minority of patients. Management strategy for TECSA aims to eliminate abnormal respiratory events, stabilize sleep architecture, and improve the underlying contributing medical comorbidities. CPAP therapy remains a standard therapy for TECSA. Expectant management is appropriate given its transient nature in most cases, whereas select patients would benefit from an early switch to an alternative positive airway pressure modality. Other treatment options include supplemental oxygen and pharmacologic therapy.

Treatment-emergent central sleep apnea (TECSA), previously called complex sleep apnea, refers to the development of central apnea following the initiation of positive pressure therapy for OSA. The International Classification of Sleep Disorders-Third Edition diagnostic criteria for this condition include: (1) the presence of predominantly OSA on diagnostic polysomnography (PSG); (2) resolution of obstructive events with positive airway pressure (PAP) therapy without a backup rate; (3) the emergence or persistence of central apneas or hypopneas on PAP therapy with central apnea index > 5 events/hour of sleep, and the number of central events is ≥ 50% of the total events; and (4) the central sleep apnea is not better explained by another central sleep apnea disorder.¹

The presence of central apnea upon initiation of PAP therapy underscores the pathophysiologic overlap between OSA and CSA, including instability of the ventilatory motor output, as expressed by high loop gain in patients with OSA and the occurrence of upper airway narrowing or occlusion during central apneas and hypopneas.^{2, 3, 4} Thus, TECSA is a “dynamic” process, with spontaneous resolution with ongoing PAP therapy in most patients, persistence in some, or appearing de novo in a minority of patients.^{5, 6, 7, 8} The American Academy of Sleep Medicine practice parameters for the treatment of CSA syndromes in adults do not specifically address TECSA. More recently, TECSA has been briefly included in a European Respiratory Society Task Force document on nocturnal central breathing disturbances. Therefore, TESCA treatment remains a gray area as caution is mandated when a therapeutic approach is extrapolated from other forms of central apnea.

The current review describes common clinical scenarios in which TECSA is encountered, discusses our management method in the context of current guidelines and the relevant literature, and presents our approach to the treatment of TECSA in the clinical setting.

Case Scenarios

Case No. 1

A 53-year-old healthy man was referred in consultation for evaluation of snoring and hypersomnia. PSG revealed OSA with an apnea-hypopnea index (AHI) of 56 events/hour. Auto-PAP was prescribed at 5 to15 cm H₂O. Wireless monitoring data during a follow-up visit at 4 weeks of therapy revealed full PAP adherence. Residual AHI was elevated at 23 events/hour due to the emergence of central events. The patient was asymptomatic and sleeping well.

Case No. 2

An obese but otherwise healthy 28-year-old man was referred in consultation after being involved in a motor vehicle accident when he fell asleep while driving. A split-night PSG revealed mostly obstructive respiratory events with an AHI of 42.6 events/hour. CPAP therapy was titrated to 14 cm H₂O. Obstructive apneas and hypopneas were eliminated at CPAP of 12 cm H₂O. Central apneas appeared at CPAP of 8 cm H₂O and were reduced but not eliminated at CPAP of 14 cm H₂O. Wireless monitoring data after 2 weeks of therapy at CPAP of 14 cm H₂O revealed a suboptimal adherence of 48.5% use > 4 h and a residual AHI of 17.4 events/hour, mainly due to central apneas. The patient reported PAP intolerance and significant sleep fragmentation.

Review of Current Literature

Determinants of Breathing Instability During Sleep: Lessons From Physiologic Studies

Respiration during non-rapid eye movement sleep is critically dependent on Paco₂; the susceptibility to central apnea manifests by unmasking the hypocapnic apneic threshold.⁹ Interestingly, central apnea rarely occurs as a single event; instead, it occurs in cycles of apneas or hypopneas, alternating with hyperpnea, a reflection of the negative feedback closed-loop cycle that characterizes ventilatory control. This is often described by using the engineering concept of “loop gain,”¹⁰ combining mainly two factors: (1) chemoreflex sensitivity (controller gain) reflecting the response of the ventilatory system to changing pressure of end tidal carbon dioxide (the controller); and (2) the effectiveness of the lung/respiratory system in lowering pressure of end tidal carbon dioxide in response to hyperventilation (the plant). The overall loop gain is the multiplicative result of several distinct and interactive mechanisms (chemosensitivity, plant gain, and circulatory delay).

The loop gain is a valuable construct to understand the contribution of breathing instability to the pathogenesis of sleep-disordered breathing (SDB), especially Cheyne-Stokes respiration. However, SDB includes a multitude of physiologic derangements that defy the assumed rhythmic periodicity of loop gain, including high peripheral chemoreflex gain,¹¹ frequent transient arousals from sleep, and abnormal cerebrovascular responsiveness,¹² factors that promote further breathing instability during sleep.

The propensity to develop central apnea during sleep could be determined experimentally by inducing central apnea using nasal mechanical ventilation.¹³ The requisite magnitude hypocapnia to induce central apnea is referred to as the CO₂ reserve. A narrow reserve CO₂ indicates a high propensity to develop central apnea and vice versa. This experimental paradigm also allows for determination of the plant gain and chemoreflex sensitivity (controller gain).

Findings from experimental studies of induced central apnea have provided significant insight regarding the determinants of CSA and may inform our understanding of TECSA, as well as potential therapeutic approaches. For example, increased controller gain may explain increased propensity to central apnea in men vs premenopausal women,¹⁴ and in older adults compared with young and middle-aged adults.¹⁵ Likewise, patients with OSA also exhibit a higher propensity to induced central apnea, and higher loop gain, compared with healthy matched adults.¹³ Interestingly, PAP therapy for 4 weeks is associated with decreased controller gain and widening of the CO₂ reserve. This observation may provide the physiologic explanation for the noted resolution of TECSA in many patients following PAP therapy for 3 months.

Plasticity is a fundamental property of the ventilatory control system. Controller gain/chemoreflex sensitivity displays substantial plasticity in response to physiologic interventions or pharmacologic manipulations. For example, acute intermittent hypoxia, mimicking recurrent respiratory events on oxygenation, results in increased controller gain and subsequent narrowing of the CO₂ reserve.¹⁶ In contrast, administration of a hyperoxic gas mixture results in decreased controller gain and widening of the CO₂ reserve.¹⁷ Finally, manipulation of male sex hormones exerts a predictable effect on the controller gain and the CO₂ reserve. Specifically, administration of testosterone¹⁸ to premenopausal women was associated with increased hypocapnic chemoreflex sensitivity (controller gain), whereas administration of leuprolide¹⁹ or finasteride²⁰ to young men exerted the opposite effect. The mutability of the controller gain can be used therapeutically in the treatment of central apnea, including TECSA.

The propensity to central apnea is also influenced by the changes in the background drive to breathe. Increased ventilatory drive and ensuring low Paco₂, for a given metabolic rate, promote stability by decreasing the magnitude of hypocapnia for a given change in alveolar ventilation, whereas reduced drive and elevated Paco₂ increase the magnitude of hypocapnia for a given change in alveolar ventilation. For example, background hypoventilation, which may occur in response to opioid analgesics, increases the propensity to develop central apnea. In contrast, administration of acetazolamide is associated with decreased plant gain and mitigation of the risk for central apnea.²¹

Central apnea may also influence the development of OSA. Patients with unfavorable upper airway anatomy are dependent on ventilatory motor output to preserve upper airway patency. Studies using upper airway imaging have shown that central apnea² and hypopnea⁴ result in pharyngeal narrowing or occlusion in normal individuals and patients with SDB. Pharyngeal closure combined with mucosal and gravitational factors may impede pharyngeal opening and necessitate a substantial increase in a drive that perpetuates breathing instability. Thus, investigating determinants of central apnea may be relevant to understanding the pathogenesis of upper airway obstruction in susceptible individuals.

Prevalence and Risk Factors

The prevalence of TECSA is uncertain given the variability in applying the diagnostic criteria. A systematic review estimated a prevalence of 8% (range, 5%-20%).²² One limitation is the identification and classification of hypopnea on PSG in most clinical sleep laboratories under “obstructive” events, given the limited value of precise classification in terms of management decisions. Therefore, pre-PAP central SDB cannot be excluded in a substantial proportion of patients seen in clinical practice. TECSA risk factors are summarized in Table 1.^{6,23, 24, 25, 26, 27, 28, 29, 30, 31, 32}

Table 1.

TECSA Risk Factors

Demographic factors23, 24, 25, 26

• Male sex

• Older age

• Lower BMI

Medical comorbidities6,25,27

• Congestive heart failure

• Coronary artery disease

• Opioids use

Baseline polysomnographic factors 6,24,25,28, 29, 30, 31

• More severe OSA

• Higher central apnea index

• Higher mixed apnea index

• Higher arousal index

Titration study factors6,30,32

• Split-night study

• Hasty/excessively high titration

• Mask leak

• Higher arousal index

• Lower total sleep time

• Lower sleep efficiency

• Higher residual apnea-hypopnea index

• Bilevel positive airway pressure use

TECSA = treatment-emergent central sleep apnea.

Natural History: Spontaneous Resolution Vs Persistence

TECSA is a dynamic condition that seems to resolve after several weeks of PAP therapy,^{6, 7, 8} with a spontaneous resolution rate between 54% and 86%.³⁰ One caveat is the tendency to aggregate PAP-persistent CSA and PAP-emergent CSA under the rubric TECSA. Although the occurrence of TECSA may implicate PAP or the relief of upper airway obstruction as the “triggers,” persistent CSA, following a period of PAP therapy, may indicate PAP failure and the need for an alternative treatment of CSA. A study used PAP telemonitoring to assess CSA trajectories during PAP therapy at weeks 1 and 13 after initiating therapy in a large number of patients (n = 133,006).⁸ Overall, TECSA was noted in 3.5% of the patients, resolved in more than one-half, persisted in about one-quarter of affected patients, and was associated with a higher rate of therapy termination. Similar findings were reported in a systematic review of five studies analyzing the natural evolution of TECSA (n = 135,283).³⁰ Of note, all studies except one allowed inclusion of patients with CSA at baseline. Patients affected by TECSA may be less adherent to therapy and are at higher risk of PAP intolerance, manifesting as dyspnea, air hunger, and involuntary CPAP mask removal during the night.^6,23 Moreover, delayed TECSA is another distinct form of TECSA that insidiously manifests on a subsequent titration study despite the absence of TECSA on the first titration study. In summary, TECSA has a dynamic nature: being transient (weeks to months) in most patients, persistent over the long run, or delayed, and appearing on a subsequent titration study after being absent on baseline assessment.

Review of Guidelines

The 2012 American Academy of Sleep Medicine CSA treatment guidelines did not specifically address TECSA.³³ In contrast, the 2017 European Respiratory Society guidelines defined TECSA as CSA that emerges and persists under CPAP use, and excluded preexisting CSA and transient CSA that resolves with ongoing PAP use as well as CSA in patients with underlying cardiovascular, endocrine, renal, or neurologic diseases.³⁴ They suggested a switch to adaptive servo ventilation (ASV) in patients with TECSA who have a residual AHI > 15 events/hour on CPAP.

Management Strategy

Goals of Therapy

Management options for TECSA parallel those used for the treatment of CSA. In addition, several factors must be considered for appropriate management of TECSA. Table 2 is a summary of different treatment modalities and their mechanism(s) of action.^{30,35, 36, 37}

Table 2.

Treatment of TECSA: Summary of Different Modalities

Modality	Mechanism of Action	Special Considerations	Effectiveness	Cost ($)
PAP therapy
CPAP	Fixed pressure eliminates obstruction (optimal pressure is a challenge as higher pressures may induce/worsen TECSA and lower pressures leave residual obstructive apneas and hypopneas)	Close follow-up recommended High rate of therapy discontinuation in symptomatic patients Repeat titration required in select patients; may be associated with increased costs	Spontaneous resolution of TECSA in 53.8%-85.7% of patients after 4-28 weeks30	300-1,000
BPAP-ST	EPAP is set to relieve obstruction IPAP and backup respiratory rate mitigate hypoventilation	Avoid high IPAP-EPAP difference (PS) and backup respiratory rate to prevent hyperventilation Optimize patient-ventilator synchrony to improve comfort and adherence to therapy	No direct comparative effectiveness with CPAP Inferior to ASV35 Effectiveness is dependent on sleep technician proficiency for optimal titration results	2,000-4,000
ASV	EPAP is set to relieve obstruction PS mirrors ventilation based on breath-by-breath analysis over 3-4 min window Dampens the magnitude of hyperventilation	Limited availability secondary to high cost Contraindicated in heart failure patients with EF < 45% (increased mortality)	Superior to CPAP/BPAP-ST35,36 Effectiveness is dependent on sleep technician proficiency for optimal titration results	3,000-5,000
Oxygen	Decreases carotid body chemosensitivity, and dampens oscillations in ventilatory control	Hypoxemic patients with cardiopulmonary comorbidities may benefit from titration studies with oxygen	Most effective in patients with CSA-CSB37 In our experience, oxygen is mostly effective when added early on after TECSA appears and persists despite slow and careful upward titration and titrated to keep oxygen saturation ≥ 94%	200/mo
Acetazolamide	Widens the CO2 reserve Decreases plant gain	Limited evidence Use is extrapolated from other CSA types (primary CSA, CSA secondary to spinal cord injury/disease)	Evidence is limited	54-89/30 d

ASV = adaptive servo ventilation; BPAP-ST = bi-level positive airway pressure in spontaneous timed mode; CSA = central sleep apnea; CSB = Cheyne-Stokes breathing; EF = ejection fraction; EPAP = expiratory positive airway pressure; IPAP = inspiratory positive airway pressure; PAP = positive airway pressure; PS = pressure support; TECSA = treatment-emergent central sleep apnea.

First, the overall aim of treatment of TECSA is to reduce the AHI and improve residual symptoms. However, the International Classification of Sleep Disorders-Third Edition diagnostic criteria do not include clinical features among the diagnostic criteria, and studies investigating treatment options have used the frequency of respiratory events as the outcome variable. Thus, management strategy should be individualized based on the underlying etiology and comorbid conditions.

Second, the appearance of TECSA on PSG may reflect one or more pathophysiologic mechanisms:

• •Unmasking of central apnea: patients with central SDB and an unfavorable upper airway anatomy may develop pharyngeal narrowing and obstructive apneas during periods of central apnea or reduced ventilatory motor output,² such as during hypocapnic central hypopnea. Relief of upper airway obstruction with PAP therapy may unmask the underlying central apnea.
• •PAP-induced events: rapid changes in PAP level or mask leak may rapidly decrease arterial Pco₂ below the hypocapnic apneic threshold, leading to central apnea.
• •Effect of intermittent hypoxia: exposure to chronic intermittent hypoxia is associated with enhanced peripheral chemoreceptor activity. Likewise, acute intermittent hypoxia, as seen in OSA, is associated with increased propensity to central apnea.¹⁶

Third, the etiologic variability of TECSA may explain the variability in treatment response, as many published studies include those with PAP-refractory CSA.^6,28 In addition, most patients with CSA have comorbid OSA.³⁸ The lack of randomized controlled studies investigating the treatment of TECSA renders estimates of treatment response and natural history imprecise.

The presence of recurrent central apnea indicates elevated loop gain via one of its two components: plant gain or controller gain. Patients with persistent TECSA may be those with the highest loop gain values.^39,40 In a pilot study, Stanchina et al⁴⁰ documented that calculated loop gain was higher for patients with persistent TECSA following 1 month of CPAP therapy and that loop gain measurement may facilitate determination of patients who need alternative modes of PAP therapy. Thus, identification of the underlying abnormality may be beneficial. For instance, CPAP could decrease plant gain by increasing lung volume, whereas elevated chemosensitivity may respond to supplemental oxygen.^39,41 A combination of therapies may be necessary when several abnormalities exist in an individual patient (eg, CPAP plus oxygen).

Proposed Approach to Treatment

Optimization and Watchful Observation

This strategy is based on the premise that central apnea will resolve spontaneously in most patients following 2 to 3 months of PAP therapy.^6,30 We favor a cautious watchful waiting approach, informed by the overall clinical picture and the severity of residual AHI. This approach requires a careful assessment of comorbid conditions, appropriate adjustments of opioid analgesics, and optimization of medical management, especially for patients with heart failure. Telemonitoring of device transmission data may obviate the need for repeat PSG in the majority of patients. Patients should be counseled to continue CPAP use pending reassessment, while addressing mask leak or adjusting pressure level if needed.⁴² A combination of symptomatic improvement and low residual AHI (< 15 events/hour) supports the continuation of CPAP therapy. Figure 1 outlines a proposed treatment algorithm that we use in our sleep center.

Figure 1.

Proposed treatment algorithm. AHI = apnea-hypopnea index; ASV = adaptive servo ventilation; BPAP-ST = bi-level positive airway pressure in spontaneous timed mode; TECSA = treatment-emergent central sleep apnea.

Bilevel Therapy (BPAP or ASV)

Persistence of TECSA (AHI > 15 events/hour) may require switching to an alternative PAP mode (ASV or BPAP with a backup rate), especially in the setting of residual symptoms. Both modes provide expiratory positive airway pressure (EPAP) to eliminate OSA and an inspiratory pressure above EPAP to increase ventilation. BPAP delivers fixed IPAP and EPAP; the difference between both pressures is the magnitude of pressure support (PS). Accordingly, BPAP delivers fixed tidal volume for a given PS level. In contrast, ASV, which was originally introduced as a treatment for central apnea associated with heart failure, mitigates CSA by providing a variable magnitude of PS, above the amount of EPAP required to eliminate obstructive events, and a backup respiratory rate. The magnitude of the PS level is reciprocal to the observed respiratory effort over a 3 to 4 min window. In other words, ASV provides a higher PS level during low-flow periods and less PS when flow is high, thus dampening the magnitude of hyperventilation. Overall, ASV is more efficacious than CPAP or BPAP in eliminating respiratory events in patients with TECSA.^35,36 One potential limitation of the available literature is that the majority of studies investigating ASV have been sponsored by the device manufacturers, using proprietary algorithms and testing intermediate physiologic outcomes rather than clinical outcomes. In a direct comparison between ASV and CPAP, Morgenthaler et al⁴³ reported a higher rate of CSA resolution 90 days following initiation of ASV compared with CPAP. However, the difference of 5.5 events/hour was lower than the a priori determined clinically relevant difference of 10 events/hour, and there was no difference in PAP adherence or patient-reported outcomes such as the Epworth Sleepiness Scale, quality of life, and feeling refreshed. To our knowledge, the trial by Morgenthaler et al is the only study that compared measures of symptomatic improvement between PAP modalities, and the first to evaluate quality of life measure in patients with TECSA.

The presence of comorbid conditions may influence the response to CPAP therapy. For example, CSA associated with heart failure may be refractory to CPAP in up to 50% of patients, even with long-term use.^44,45 Select patients may need adequate care, with an early switch to alternative PAP modes.^46,47 However, ASV is contraindicated in patients with CSA associated with heart failure with reduced ejection fraction. This is based on the findings of the Treatment of Sleep-disordered Breathing with Predominant Central Sleep Apnea by Adaptive Servo Ventilation in Patients with Heart Failure (SERVE-HF), a randomized trial of ASV vs standard medical therapy in patients with predominantly CSA due to heart failure with reduced ejection fraction (ejection fraction ≤ 45%).⁴⁸ The study found that ASV was associated with a 6% absolute increase in all-cause mortality and cardiovascular mortality compared with standard medical therapy.

The aforementioned considerations underpin our approach to use ASV in an individualized manner. Specifically, we use ASV, in the absence of a contraindication, if symptomatic TECSA persists despite the use of CPAP, alone or with supplemental oxygen (discussed later).

BPAP is another option for TECSA. Nevertheless, BPAP in the spontaneous mode may worsen central apneas.^49,50 In contrast, several studies have shown an improvement in AHI with the use of BPAP with a backup rate (BPAP in spontaneous timed mode [BPAP-ST]).^35,36,49 One peculiar observation is the delayed emergence of TECSA 6 weeks following initiation of BPAP-ST.³⁵ This finding would not be expected to occur with ASV given the automated pressure support adjustment to ventilatory instability. Special care must be taken to use the least magnitude of effective PS to minimize hyperventilation and the risk of re-emergence of TECSA. Furthermore, clinicians must optimize patient-ventilator synchrony to minimize arousals and sleep state instability, as this also destabilizes ventilation.

Supplemental Oxygen

Oxygen therapy had been proven to reduce the central apnea index even in the absence of associated nocturnal hypoxemia.^51,52 Increased arterial Po₂ works by lowering carotid-body chemosensitivity, therefore buffering oscillations in ventilatory control.⁵³ The addition of oxygen to CPAP may result in better control of TECSA, via reduction of the hypoxic respiratory drive and increasing cerebral Pco₂ as CO₂ is displaced from hemoglobin by the increased oxygen level (ie, the Haldane effect).⁵⁴

Nocturnal home oxygen therapy (HOT) has been extensively studied in patients with CSA and congestive heart failure. The CHF-HOT study group assessed the efficacy of HOT on SDB and other variables in 56 patients with stable congestive heart failure and CSA/Cheyne-Stokes breathing.³⁷ The study found that HOT significantly reduced AHI. A large network meta-analysis of 14 randomized controlled trials (n = 919 patients) compared the effect of any combination of CPAP, ASV, oxygen, or inactive control on AHI in patients with congestive heart failure and CSA/Cheyne-Stokes breathing.⁵⁵ The authors found that ASV was the most efficacious therapy (87.8%), followed by oxygen (12.2%). Therefore, oxygen is reserved for patients who do not tolerate PAP therapy or used in combination with PAP when response to PAP alone is unsatisfactory.⁵⁶ Oxygen is also beneficial in hypoxemic patients with cardiac or pulmonary comorbidities requiring oxygen therapy independent of their TECSA. However, reimbursement for supplemental oxygen, in the absence of sustained hypoxia, is often a barrier.

Acetazolamide

Acetazolamide, a mild diuretic, is associated with increased ventilatory motor output by inducing metabolic acidosis. Several studies have shown the efficacy of acetazolamide in reducing the severity of central apnea. There is empiric evidence that administration of acetazolamide is associated with widening of the CO₂ reserve, likely attributed to decreased plant gain and controller gain.²¹ However, the effectiveness of acetazolamide following prolonged use has yet to be determined. We use acetazolamide on a case-by-case basis in subjects with persistent symptomatic TECSA as an adjunct to CPAP therapy.^21,57,58

Positional Therapy

Several studies have noted increased central apnea frequency in the supine position, likely attributed to passive upper airway collapse during CSA, lower lung volumes, and worsened pulmonary vascular congestion and associated hypoxia. New-generation sleep position devices may be efficacious as salvage therapy for patients with CSA who are intolerant to PAP therapy.⁵⁹ We are cautious in using this approach to treat TECSA, pending outcomes data.

Case Scenario Outcomes

The first patient had documented initial optimal adherence and reported substantial clinical improvement despite elevated residual AHI. Given the absence of TECSA risk factors and the overall favorable clinical picture, expectant management was an acceptable initial choice. Close telephone follow-up along with routine access to wireless monitoring data allowed for proper assessment of any potential symptomatologic worsening as well as the time course of TESCA evolution and guided the potential need for a repeat titration study. Residual AHI at 3 months was 14 events/hour. The patient remained asymptomatic. No further action was taken.

The second patient failed expectant management, given significant discomfort with therapy and persistence of TECSA (residual AHI > 15 events/hour). An ASV titration study was ordered, and ASV was successful in eliminating both obstructive and central respiratory events. At 4 weeks follow-up, the patient had significantly improved adherence to therapy with 76% use > 4 h and reported “quiet restful sleep.” PAP download data confirmed the absence of TECSA.

Conclusions

TECSA is a complex process that combines central breathing instability and unfavorable upper airway structure and function. It is a dynamic process with spontaneous resolution with ongoing PAP therapy in most patients (transient TECSA), persistence in some (persistent TECSA), or appearing de novo in a minority of patients (delayed TECSA). Expectant management is appropriate in asymptomatic patients with close follow-up. Switch to alternate PAP modalities (ASV, BPAP-ST) is appropriate in symptomatic patients and when AHI is > 15 events/hour on follow-up.