The central sleep apnea (CSA) syndromes include a diverse group of disorders associated with the presence of central apnea during sleep (Fig. 21–1).¹–³ In some of the disorders discussed in this chapter, the patients have primarily nocturnal hypoventilation (increased arterial partial pressure of carbon dioxide [PaCO₂]) due to inadequate tidal volume and/or respiratory rate with relatively few discrete central apneas. However, discussion of patients with central apnea and hypoventilation syndromes together is useful clinically because they have many similar aspects of pathophysiology and treatment.

FIGURE 21–1 A central sleep apnea event in a patient with the idiopathic central sleep apnea syndrome (primary central sleep apnea). Note that the central apnea followed a large breath. EMG = electromyogram; SpO₂ = pulse oximetry. Adapted from Berry RB: Sleep Medicine Pearls, 2nd ed. Philadelphia: Hanley & Belfus, 2003, p. 237.

Central apnea in adults is defined as a cessation in airflow of 10 seconds or longer that is associated with an absence of respiratory effort.⁴ During diagnostic studies oronasal thermal sensor signal is used to detect apnea. During a positive airway pressure (PAP) titration the PAP flow signal is used. The diagnosis of a CSA syndrome requires that the majority of apneic events be central in nature. The exact proportion of central events required is not clear, with various authors diagnosing the CSA syndromes when 50% to 80% of the events are central. Traditionally, patients with CSA have accounted for less than 5% to 15% of patients with sleep apnea evaluated at most sleep centers. However, two factors have increased the number of CSA patients being evaluated. First, there has been an increased recognition of sleep-disordered breathing in congestive heart failure (CHF) and a substantial number of patients with systolic heart failure have Cheyne-Stokes breathing central sleep apnea (CSB-CSA) (Fig. 21–2). Second, recently, there has been more aggressive use of opiates to control pain. As discussed in this chapter, many patients develop central apnea as a result of the use of potent narcotics. Central apnea may occur either at baseline or once patients are exposed to PAP treatment. Thus, sleep centers can expect to see more patients with CSA.

FIGURE 21–2 Cheyne-Stokes breathing with central hypopnea at the nadir of ventilatory effort. This patient had obstructive and mixed apneas during the diagnostic portion of a sleep study. When placed on continuous positive airway pressure (CPAP), the underlying ventilatory instability was uncovered. Here, Cflow is the signal from the accurate flow transducer in the positive airway pressure (PAP) device. SpO₂ = pulse oximetry.

Many patients who exhibit central apneas also exhibit central hypopneas (see Fig. 21–2). The challenges in defining central hypopnea were discussed in earlier chapters. The American Academy of Sleep Medicine (AASM) scoring manual discourages identification of central hypopnea unless an accurate method to quantify respiratory effort is being used (esophageal pressure or respiratory inductance plethysmography).⁴ In general, central hypopneas are characterized by a proportionate decrease in both airflow and respiratory effort (Fig. 21–3).⁵ Usually, there is no snoring or chest-abdominal paradox and the nasal pressure or PAP device flow signal is fairly rounded (minimal or no signs of airflow limitation).

FIGURE 21–3 Central, obstructive, and mixed hypopneas. In central hypopnea, the flow falls in proportion to the respiratory effort. The airflow profile using an accurate measure of airflow (including nasal pressure or CPAP flow) shows a round contour. In obstructive hypopnea, there is evidence of airflow limitation (flat airflow shape) and flow falls even though respiratory effort stays the same or increases. In mixed hypopnea, there is a fall in respiratory effort but the fall in flow is proportionately greater and there is evidence of airflow limitation. Adapted from Berry RB: Sleep Medicine Pearls, 2nd ed. Philadelphia: Hanley & Belfus, 2003, p. 83.

Classification of Csa Syndromes

There is a number of classifications of CSA and the hypoventilation syndrome but none of these classifications is entirely satisfactory.1,6 The International Classification of Sleep Disorders, 2nd edition (ICSD-2) lists five CSA syndromes (Box 21–1).⁶ Primary CSA is termed idiopathic central sleep apnea (ICSA) in this chapter, in keeping with much of the literature. Primary CSA, CSB-CSA, and high-altitude periodic breathing (HAPB) are hypocapnic forms of CSA. Patients with these syndromes have a normal or low PaCO₂ during wakefulness. During sleep, these patients do not develop hypercapnia. In contrast, patients with CSA due to drug or substance and primary sleep apnea of infancy have normal or increased daytime PaCO₂ and may develop or have worsening hypercapnia during sleep. The ICSD-2 lists five categories of hypoventilation syndromes (Box 21–2). Idiopathic nonobstructive hypoventilation is due to an abnormality of ventilatory control of unknown etiology. Congenital hypoventilation syndrome is due to abnormal ventilatory control due to a genetic abnormality. Three groups of disorders with hypoxemia or hypercapnia during sleep are listed. The first is due to lower airways obstruction (e.g., chronic obstructive pulmonary disease), the second is due to pulmonary parenchymal or vascular disorders (e.g., pulmonary fibrosis, interstitial lung disease), and the third is due to abnormality of the chest wall or neuromuscular disorders (e.g., kyphoscoliosis, amyotrophic lateral sclerosis [ALS], obesity hypoventilation syndrome [OHS]). Patients with the OHS are discussed in Chapter 15. The effects of chronic obstructive pulmonary disease on sleep (including hypercapnia) are discussed in Chapter 22.

Box 21–1

Central Sleep Apnea Syndromes

1. Primary central sleep apnea

2. Cheyne-Stokes breathing pattern

3. High-altitude periodic breathing

4. Central sleep apnea due to drug or substance

5. Primary sleep apnea of infancy

From American Academy of Sleep Medicine: ICSD-2 International Classification of Sleep Disorders, 2nd ed. Diagnostic and Coding Manual. Westchester, IL: American Academy of Sleep Medicine, 2005.

Box 21–2

Hypoventilation Syndromes

Comment: Rare, usually case reports

2. Congenital central alveolar hypoventilation syndrome

Example: Central congenital hypoventilation syndrome

Examples: Hypercapnic COPD, bronchiectasis, or cystic fibrosis

Example: Sleep-related hypoventilation with idiopathic pulmonary fibrosis or other interstitial lung diseases or pulmonary vascular disease associated with end-stage lung disease

Examples: Obesity hypoventilation syndrome, neuromuscular disease, kyphoscoliosis

See Appendix 21–1 for diagnostic criteria.

COPD = chronic obstructive pulmonary disease.

From American Academy of Sleep Medicine: ICSD-2 International Classification of Sleep Disorders, 2nd ed. Diagnostic and Coding Manual. American Academy of Sleep Medicine. Westchester, IL: American Academy of Sleep Medicine, 2005.

This chapter utilizes a classification of the CSA syndromes adapted from one proposed by Bradley and coworkers ¹ (Box 21–3) that subdivides patients into hypocapnic and hypercapnic groups. Patients with hypocapnic CSA tend to have low to normal daytime PaCO₂ values. The disorders in this group include ICSA, CSB-CSA, HAPB, and treatment-persistent/-emergent CSA.

Box 21–3 Central Sleep Apnea Syndromes

Hypocapnic (normal or low daytime pcO₂)

1. Idiopathic central sleep apnea

2. Cheyne-Stokes breathing—central sleep apnea

Associated with congestive heart failure

Associated with neurologic disease

3. Periodic breathing at high altitude

4. Complex sleep apnea (treatment emergent or persistent sleep apnea)

• Idiopathic complex sleep apnea *

• Combined OSA and CSB (CPAP eliminates obstruction and unmasks CSB–CSA)

Hypercapnic ^†

1. Won’t breathe

A Central hypoventilation

• Congenital central hypoventilation syndrome

• Idiopathic central hypoventilation syndrome

• Brain tumors, cerebrovascular disease

• Structural brain disorders—Chiari’s syndrome

• Apnea of infancy

B Medication-induced central sleep apnea (narcotics/opiates)

• Central sleep apnea with normal or increased daytime PCO₂

• Complex sleep apnea (treatment emergent or persistent central sleep apnea)

C Obesity hypoventilation syndrome

2. “Can’t Breathe”

A Restrictive thoracic cage disorders

B Neuromuscular disorders

i. Motor neuron disease including poliomyelitis

ii. Neuropathy

iii. Neuromuscular junction disorders (myasthenia gravis)

iv. Myopathy (muscular dystrophy)

CompSA = complex sleep apnea; CPAP = continuous positive airway pressure; CSA = central sleep apnea; CSB = Cheyne-Stokes breathing; OSA = obstructive sleep apnea; PaCO₂ = arterial partial pressure of carbon dioxide; PCO₂ = partial pressure of carbon dioxide.

^*Idiopathic complex sleep apnea refers to patients with CompSA who do not have Cheyne-Stokes breathing or are not taking narcotics or other medications altering ventilatory drive.

^†Some patients with these disorders may have normal daytime PaCO₂.

The term complex sleep apnea (CompSA) has been used to identify patients who have primarily obstructive or mixed events during diagnostic studies but develop central apneas on PAP treatment (treatment-emergent central apneas) or have significant persistent central apneas on PAP treatment (treatment-persistent central sleep apnea).⁷ It is not unusual for patients with obstructive sleep apnea (OSA; predominantly obstructive apneas and hypopneas) to have some mixed or central apneas during diagnostic studies.^8,9 PAP will eliminate central and mixed as well as obstructive apneas in many of these patients on the first treatment night (or during the initial PAP treatment).⁹ In other patients, central apneas will persist after airway obstruction is eliminated. In the early studies of the effect of treatment of OSA patients with tracheostomy (elimination of obstruction), patients were found to have residual central apneas that tended to resolve with chronic treatment.⁸ The term CompSA is also sometimes used to include patients with narcotic-associated sleep-disordered breathing who have persistent or emergent central apneas on PAP. Patients with this condition usually have high normal or mildly increased PaCO₂ during wakefulness. In this chapter, this group is discussed under the hypercapnic CSA group. It is also not uncommon for patients with CHF to have a combination of obstructive or mixed apnea and CSB-CSA (Fig. 21–4).¹⁰ In some of these patients, application of PAP will eliminate obstruction but frequent central apneas of the Cheyne-Stokes type may persist or emerge. For this reason, the term CompSA is sometimes applied. These patients are discussed with the nonhypercapnic CSA disorders. Patients with CompSA without an obvious etiology (no narcotics or heart failure) are termed idiopathic CompSA for lack of a better terminology. These patients have an instability in ventilatory control either at baseline or due to PAP treatment. Because they have normal or low daytime PaCO₂, they are discussed in the hypocapnic CSA sections.

FIGURE 21–4 A mixed apnea. Note the crescendo-decrescendo pattern of ventilatory effort. When the patient was placed on CPAP, the mixed apneas were converted to central apneas of the Cheyne-Stokes type. Another hint that the patient has congestive heart failure is the delay in the arterial oxygen saturation (SaO₂) nadir. The 88% occurs before the termination of the event pictured. In fact, this desaturation is not due to the event pictured but to the previous event. Airflow = nasal pressure. From Malhotra A, Berry RB, White DP: Central sleep apnea. In Carney P, Berry RB, Geyer JW (eds): Clinical Sleep Medicine. Philadelphia: Lippincott Williams & Wilkins, 2005, p. 341.

The hypercapnic CSA syndromes associated with hypoventilation include those with a defect in ventilatory control (congenital central alveolar hypoventilation, acquired central alveolar hypoventilation, CSA due to medication/substance), patients with the OHS, thoracic cage disorders, and neuromuscular disorders. The hypercapnic CSA group can be divided into a “won’t breathe” group (can reduce their PaCO₂ with voluntary increases in ventilation) and a “can’t breathe” group due to an abnormal thoracic cage or neuromuscular weakness. This classification is not entirely satisfactory because patients with OHS have various combinations of OSA, abnormal ventilatory control, and respiratory pump abnormality (mass loading due to obesity). Of note, early in the disease course, some patients with chronic hypoventilation syndromes will have an increased PaCO₂ only during sleep. They may present with complaints of disturbed sleep or nocturnal dyspnea, morning headaches, daytime sleepiness, or insomnia. Later in the disease course, these may present with hypercapnic respiratory failure and cor pulmonale.

Pathophysiology of Csa

Effects of Normal Physiologic Changes

In normal individuals, there is a 2 to 8 mm Hg rise in PaCO₂ during non–rapid eye movement (NREM) sleep. This is thought due to loss of the wakefulness drive, reduction of the hypercapnic and hypoxic ventilatory drives, and increased upper airway resistance (Table 21–1).¹¹ The wakefulness drive is a poorly understood generalized augmentation of ventilation associated with the wakefulness state. During NREM sleep, ventilation is totally under metabolic control (chemoreceptors). Ventilatory control centers respond to information from the peripheral chemoreceptors including the carotid body (arterial partial pressure of oxygen [PaO₂] and PaCO₂) and medullary chemoreceptors (H⁺ due to changes in PaCO₂).¹¹ During rapid eye movement (REM) sleep, ventilation is irregular and nonmetabolic factors also affect ventilation. The hypercapnic and hypoxic ventilatory drives are lower during REM than during NREM sleep. In addition, during REM sleep, there is generalized skeletal muscle hypotonia.¹¹–¹³ The contribution of accessory muscles of ventilation is either reduced or absent and ventilation depends entirely on the diaphragm. The loss of accessory inspiratory muscles can compromise the ability to maintain adequate ventilation, especially in patients with muscle weakness or a high work of breathing. During the phasic changes of REM sleep (associated with bursts of eye movements), there is often additional inhibition of upper airway muscles and diaphragmatic activity resulting in episodes of central apnea or reduced tidal volume (Fig. 21–5).^12,13 Patients with hypercapnic CSA/hypoventilation syndromes usually have worse oxygenation and highest PaCO₂ during REM sleep. It is of interest that some patients with hypocapnic CSA due to ventilatory control instability may actually have better oxygen saturation during REM than during NREM sleep. For example, in the ICSA or CSB-CSA syndromes, central apneas generally do not occur during REM sleep. Some patients with CompSA will also have a much better response to CPAP during REM sleep than during NREM sleep.

TABLE 21–1

Effect of Sleep on Ventilatory Control

	WAKE	NREM	REM
Wakefulness drive	Present	Absent	Absent
Central chemoreceptors (H⁺, PCO₂)	Intact	Reduced	Very reduced
Peripheral chemoreceptors (PO₂, PCO₂ [H⁺])	Intact	Reduced	Very reduced
Apneic threshold	Not present	Present	N/A
Behavioral and other nonventilatory control center influences	Present	Absent	Periods of reduced tidal volume during phasic REM sleep often associated with bursts of eye movements

N/A = not applicable; NREM = non-rapid eye movement; PCO₂ = partial pressure of carbon dioxide; PO₂ = partial pressure of oxygen; REM = rapid eye movement.

Ventilatory Control in Hypocapnic CSA

A number of factors contribute to the occurrence of hypocapnic CSA (Table 21–2). Patients with hypocapnic CSA have a normal or low daytime PaCO₂ and high ventilatory responses to hypercapnia. They develop central apnea because the PaCO₂ falls below the apneic threshold (AT) to be discussed in the following section.^14,15 From a ventilatory control model standpoint, they have high system loop gain and ventilatory instability.^16,17 Central apneas are more common in stage N1 and N2 rather than stage N3 sleep because ventilation is less stable on transition from wakefulness to sleep. Sleep stage changes and arousal predispose to the occurrence of central apneas in patients with hypocapnic CSA.¹⁸ Hypocapnic central apneas are uncommon during REM sleep because ventilation is not totally under metabolic control and the ventilatory response to hypercapnia is lower than during NREM sleep, making ventilatory instability less likely. Short central apneas can occur during REM sleep during bursts of eye movements but these are likely not due to a lowering of PaCO₂ below the AT.

TABLE 21–2

Mechanisms Inducing Central Apnea

IDIOPATHIC CENTRAL SLEEP APNEA	CHEYNE-STOKES BREATHING
High hypercapnic ventilatory drive (high controller gain) Small sleeping PCO₂—AT difference Sleep state instability (arousals)	• High hypercapnic ventilatory response (high controller gain) • High sympathetic activity • Pulmonary congestion—high PCWP (J receptor stimulation) • Delay in arterial blood reaching chemoreceptors • Small sleeping PCO₂—AT difference

AT = apneic threshold; PCO₂ = partial pressure of carbon dioxide; PCWP = pulmonary capillary wedge pressure.

Apneic Threshold

During wakefulness, hypocapnia does not cause cessation of breathing due to the presence of the wakefulness stimulus to breathing. During NREM sleep, ventilation depends completely on metabolic control. If the PaCO₂ falls below a characteristic value (AT) for each individual, a central apnea occurs 14,15 (Fig. 21–6). Ventilation does not resume until the PaCO₂ climbs above (and actually slightly higher than) the AT. The AT can be experimentally determined by serial runs of hyperventilation with positive-pressure ventilation that progressively drop the PaCO₂ to various levels below the eucapnic PaCO₂ level during sleep. The positive-pressure ventilation is terminated suddenly, and if the PaCO₂ has dropped below the AT, a central apnea occurs. For example, suppose the spontaneous sleeping PaCO₂ is 42 mm Hg, then trials could progressively induce PaCO₂ values of 41, 40, 39, 38, 37, and so on. The level at which a central apnea occurs when the positive-pressure device is turned off is the AT. As noted previously, the sleeping PaCO₂ is normally about 2 to 8 mm Hg above waking value. The typical AT is usually at or 1 to 2 mm Hg lower than the waking PaCO₂. In Figure 21–7, during normoxia, an individual end-tidal partial pressure of carbon dioxide (P_ETCO₂) increases with sleep from 40 to 44 mm Hg. Hyperventilation trials with positive-pressure ventilation resulted in central apnea (apnea duration > 0 sec) in a range of P_ETCO₂ values from about 38 to 42 mm Hg.

Of note, it is the difference between the sleeping PaCO₂ and the AT rather than the position of the AT that is the critical determinant of the propensity to develop central apnea. The smaller the PaCO₂-AT difference the more likely CSA is to occur. Figure 21–8 shows awake and sleeping PaCO₂ values for CHF patients with and without periodic breathing. Those with periodic breathing have a small difference between the sleeping PaCO₂ and the AT. This is because the PaCO₂ increased relatively little with transition from wake to sleep but the AT did not show a corresponding decrease.¹⁹

FIGURE 21–8 The awake eupnea end-tidal partial pressure of carbon dioxide (PaCO₂), sleeping P_ETCO₂, and apneic threshold (AT) for patients with congestive heart failure (CHF) with and without periodic breathing. The patients with periodic breathing had a small sleeping P_ETCO₂—AT difference. From Xie A, Skatrud JG, Puelo DS, et al: Apnea-hypopnea threshold for CO₂ in patients with congestive heart failure. Am Respir Crit Care Med 2002;165:1245–1250.

Studies have shown that the PaCO₂-AT difference can vary with changes in ventilatory drive.²⁰ Specifically, the PaCO₂-AT difference increases with the induction of metabolic acidosis (acetazolamide) and decreases with metabolic alkalosis. This is somewhat counterintuitive because one might expect that further increases in ventilatory drive due to metabolic acidosis would increase the propensity for central apnea. In contrast to metabolic acidosis, hypocapnic hypoxia decreases the PaCO₂-AT difference. Thus, hypoxemia can increase the likelihood of central apnea. A detailed description of the factors that affect the PaCO₂-AT difference is found in references 15 and 20.

Loop Gain

Ventilatory control stability has also been analyzed using feedback control theory focusing on the loop gain (LG) of the respiratory system.2,16,17 The LG is determined by the plant gain (ability of increases in ventilation by the lungs/respiratory muscles to reduce the PaCO₂) and the controller gain (change in ventilation induced by a change in PaCO₂) (Fig. 21–9). High plant gain (Δ PaCO₂/Δ ventilation) is associated with hypercapnia and low physiologic dead space. Owing to the hyperbolic relationship between ventilation and PaCO₂, small changes in ventilation induce larger changes in PaCO₂ if hypercapnia is present (Fig. 21–10A). Controller gain depends on the sensitivity of the ventilatory control centers to changes in PaCO₂. Controller gain can be expressed as Δ Ventilation/Δ PaCO₂. The LG = plant gain × controller gain. A system with a high LG (LG > 1) is unstable (see Fig. 21–10B). High controller gain or high plant gain can destabilize the system. In hypocapnic CSA, the major factors causing instability are the high controller gain and the effect of wake-to-sleep transitions.

FIGURE 21–9 The interaction between controller gain and plant gain for the respiratory system. High controller gain or high plant gain results in a large response to a small disturbance (change in ventilation). D = disturbance; PCO₂ = partial pressure of carbon dioxide; R = response; V_E = volume of expired gas. From White DP: Central sleep apnea. In Kryger MH, Roth T, Dement WC (eds): Principles and Practice of Sleep Medicine. Philadelphia: Elsevier, 2005, pp. 969–981.

Upper Airway and Posture Effects

Both ICSA and CSB-CSA tend to be worse in the supine position.21,22 In ICSA patients, Issa and Sullivan²¹ found central apnea to be more frequent in the supine position. They hypothesized that upper airway reflexes may trigger central apnea because upper airway anesthesia abolished central apnea in two patients. In addition, high levels of continuous positive airway pressure (CPAP) abolished central apnea. CSB-CSA has also been noted to be more prominent in the supine position.²² It is not known whether this is due to upper airway factors or to changes in oxygenation or pulmonary congestion.

Hypocapnic Csa Syndromes

Patterns of Ventilation in ICSA and CSB

The patterns of ventilation differ between patients with CSB-CSA and patients with ICSA. In both groups, the central apneas typically are 20 to 40 seconds in duration. In ICSA, there is typically a short ventilatory phase of 2 to 4 breaths between central apneas, and arousals tend to occur at apnea termination (Fig. 21–11). Patients with CSB have a longer ventilatory phase between consecutive apneas, and the ventilatory phase has a characteristic crescendo-decrescendo morphology. Hall and colleagues²³ compared cycle length between patients with ICSA and CSB-CSA due to heart failure (Table 21–3). The apnea durations were similar but cycle length was longer in CSB-CSA due to a long ventilatory phase. The CSA-CSB patients also had a longer delay in the arterial oxygen saturation (SaO₂) nadir. The lower the cardiac output, the longer the ventilatory phase and the delay in SaO₂ nadir. This is believed due to a long circulation time. The patterns of ventilation in treatment-emergent CSA (nonhypercapnic, non-CSB CompSA) resembles that of ICSA.

TABLE 21–3

Differences in Respiration between Patients with Idiopathic Central Sleep Apnea and Cheyne-Stokes Breathing

	CENTRAL APNEA DURATION (sec)	CYCLE LENGTH (sec)	VENTILATORY PHASE (BETWEEN APNEAS) (sec)	DELAY IN SaO₂ NADIR (sec)
ICSA	20.9	37.3	16.7	10.3
CSB-CSA (CHF)	22.3	59.0	36.7	24.3

CHF = congestive heart failure; CSA = central sleep apnea; CSB = Cheyne-Stokes breathing; ICSA = idiopathic central sleep apnea; SaO₂ = arterial oxygen saturation.

From Hall MJ, Xie A, Rutherford R, Bradley TD: Cycle length of periodic breathing with and without heart failure. Am J Respir Crit Care Med 1996;154:376–381.

Idiopathic CSA (Primary CSA)

These patients have central apneas of unknown etiology. CHF, neurologic disorders, or medications suppressing ventilation are not present. The prevalence varies but is thought to be 5% to 10% of patients being studied in sleep centers. Patients may present with symptoms similar to those of patients with OSA. These may include daytime sleepiness or insomnia, snoring, witnessed breathing pauses, and disturbed sleep. They tend to be thinner than patients with OSA and have less prominent snoring. In the ICSD-2, the terminology for this patient group is primary CSA (Box 21–4).

Box 21–4

Primary Central Sleep Apnea—Diagnostic Criteria

A The patient reports at least one of the following:

i. Excessive daytime sleepiness.

ii. Frequent arousals and awakenings during sleep or insomnia complaints.

iii. Awakening short of breath.

B Polysomnography shows five or more central apneas per hour of sleep.

C The disorder is not better explained by another current sleep disorder, medical or neurologic disorder, medication use, or substance use disorder.

From American Academy of Sleep Medicine: ICSD-2 International Classification of Sleep Disorders, 2nd ed. Diagnostic and Coding Manual. Westchester, IL: American Academy of Sleep Medicine, 2005.

Pathophysiology of Idiopathic CSA

As reflected in the term “idiopathic,” the etiology of primary CSA is unknown. However, the patients have ventilatory instability due to a high hypercapnic ventilatory response ²⁴ or sleep state instability.¹⁸ These patients tend to have decreased PaCO₂ values during wakefulness. Often, central apnea can be triggered by only one or two large breaths (see Fig. 21–1). The periods of increased ventilation triggering central apneas often are associated with arousal. Arousal may trigger a transient increase in ventilation and a fall in PaCO₂. This transient fall in PaCO₂ is then associated with a central apnea as the patient returns to sleep. As discussed previously, in some patients with ICSA, central apnea occurs mainly in the supine position.²¹ Studies have documented a response to CPAP in some ICSA patients.^21,25 In research studies, the addition of dead space or inhalation of PaCO₂ with the goal of increasing and/or stabilizing the PaCO₂ has been shown to reduce central apnea in ICSA patients.²⁶ However, these interventions have not been tried for long-term treatment nor are they practical.

Polysomnography

Sleep studies in patients with ICSA typically reveal frequent, isolated central apneas or runs of central apneas (a form of periodic breathing). A run of central apneas may follow arousal from a nonrespiratory stimulus. Central apneas in ICSA patients occur during NREM sleep, most commonly in stage 1 or 2 sleep. Central apnea in these patients is much less common during stage N3 and REM sleep. The percentage of total respiratory events that must be central is not well defined but is usually taken as greater than 50%.

Treatment of ICSA

Because idiopathic CSA is rare, most treatment information comes from small case series (no randomized, controlled studies) and no long-term studies of the effectiveness of any treatment have been published. There is no uniform consensus about the best treatment for patients with ICSA. This group is heterogeneous, and treatment must be individualized (Box 21–5). Various respiratory stimulants have been tried as treatments for idiopathic CSA, with variable amounts of success.²⁷–²⁹ The best evidence is for use of acetazolamide (Diamox), which is a carbonic anhydrase inhibitor. Acetazolamide induces a metabolic acidosis and reduces the pH even if the PaCO₂ also decreases slightly. DeBacker and associates²⁷ studied the effects of acetazolamide 250 mg 1 hour before sleep after 1 month of treatment in ICSA patients. The apnea-hypopnea index (AHI) was reduced by about 50% and symptoms improved, although sleep efficiency was not significantly better.²⁷

Box 21–5 Possible Treatments for Idiopathic Central Sleep Apnea

CPAP

Hypnotics

Respiratory stimulants (acetazolamide)

Oxygen therapy??

CPAP = continuous positive airway pressure.

Another treatment approach is to decrease arousals and promote stage N3 sleep where ventilation is more stable and the PaCO₂ likely slightly higher. Hypnotics including triazolam and zolpidem have been tried.30–³² A recent trial of zolpidem 10 mg in ICSA was reported by Quadri and coworkers.³¹ The central AHI decreased from 30 to 13.5/hr and subjective sleepiness as measured by the Epworth Sleepiness Scale improved (from 13 to 8). However, in some patients, obstructive events increased with hypnotic treatment. Of note, hypnotics are relatively contraindicated in hypercapnic CSA.

As noted previously, CPAP also appears to be effective treatment in some patients with ICSA.21,25 The mechanisms by which CPAP works are unknown. Four possibilities are that nasal CPAP minimizes overshoots in PaCO₂ after arousal, slightly increases the sleeping PaCO₂ in patients who are hypocapnic at baseline, prevents high upper airway resistance from inducing arousals, or prevents significant negative upper airway pressure (which may trigger reflex central apnea). Patients who snore or have central apnea mainly in the supine position might be assumed to be the best candidates for CPAP treatment. Of note, one case report found that CPAP helped but bilevel positive airway pressure (BPAP) worsened CSA.³³ BPAP may destabilize the system by augmenting ventilation. No study evaluating the effect of adaptive servo-ventilation (ASV), a mode designed to stabilize ventilation, in ICSA has been published but a trial of that treatment might be reasonable. Supplemental oxygen has been shown to improve CSA in patients with CSB.^34,35 However, an evaluation of supplemental oxygen in ICSA patients has not been published. Supplemental oxygen might work by decreasing ventilatory instability by blunting the hypercapnic ventilatory response. The hypercapnic ventilatory response is increased by hypoxia and decreased by hyperoxia.

Csb with Csa

CSB occurs most commonly in patients with left ventricular systolic dysfunction 36–³⁸ but also can occur in patients with diastolic CHF or neurologic disorders.^39,40 The neurologic disorders associated with CSB include a prior cerebrovascular accident (CVA) and neurodegenerative disorders. CSB may be present in up to 30% of patients in the first few days after stroke.^39,41 Central apnea tends to resolve and is less commonly seen in studies of patients studied several months after stroke. OSA is the predominant form of sleep apnea present in patients after CVA. Of interest, a recent study suggested that a significant proportion of CSB-CSA in stroke patients was due to occult cardiovascular disease.⁴² CSA-CSB has also recently been reported in patients with heart failure with normal systolic function (diastolic heart failure).⁴⁰

CSB-CSA due to systolic CHF is probably the most common cause of CSA seen today. Javaheri and colleagues ³⁶ found occult sleep-disordered breathing in 45% of a group with stable heart failure and an ejection fraction less than 45% and CSA was common in the affected individuals. MacDonald and associates³⁷ studied a group of patients in stable CHF with maximal modern medical management and found 61% to have some form of sleep-disordered breathing (31% central apnea and 30% OSA). Oldenberg and coworkers³⁸ studied 700 patients with CHF and found similar results. It should be noted that many patients in these studies had various amounts of both central and obstructive apneas. Defining what percentage of events qualified the patient for the central apnea group was somewhat arbitrary. In addition, patients can vary in the percentage of central events overnight⁴³ or from night to night.⁴⁴ In any case, the prevalence of CSB-CSA is very high in patients with significant systolic heart failure.

Patients with CSA-CSB may complain of the typical symptoms of sleep apnea including disturbed sleep or daytime sleepiness. However, in most studies, the majority of patients do not complain of subjective excessive daytime sleepiness. Thus, a high index of suspicion is needed to suspect the presence of CSB. Nocturnal sleep complaints are often assumed to be secondary to CHF rather than to co-morbid sleep apnea. Despite the lack of subjective sleepiness complaints in CSB-CSA patients, studies have documented improvement in sleep quality and objective daytime sleepiness with successful treatment.⁴⁵

Polysomnography in CSB-CSA

The crescendo-decrescendo morphology of CSB-CSA has previously been discussed (see Figs. 21–2, 21–4, and 21–11). Another characteristic to appreciate is the long delay in the nadir in the SaO₂ tracing after event termination due to a prolonged circulation time (low cardiac output) (Fig. 21–12). The distinctive pattern of CSB may not be recognized during a diagnostic study in a patient who has both OSA and underlying CSB.¹⁰ However, during a subsequent PAP titration, pure CSB-CSA may emerge when upper airway obstruction is eliminated (Fig. 21–13). The possibility of underlying CSB should be considered if mixed apneas during the diagnostic study have a long central component or if the nadir in the SaO₂ following obstructive event termination is delayed. If central hypopneas rather than central apneas are present, this may also make CSB more difficult to recognize. As noted previously, central hypopneas rather than apneas can occur at the nadirs in ventilatory effort (see Fig. 21–2). The typical CSB cycle time is relatively long at 60 to 90 seconds and may be missed if a short time window is used to monitor sleep during a PAP titration. Figure 21–13 also illustrates the typical cessation of CSB-CSA when a patient transitions from NREM to REM sleep.