Sleep-related breathing and sleep-wake disturbances in ischemic stroke

Sleep-related breathing disturbances.

SDB is highly prevalent in stroke patients. Nighttime symptoms include sleep-onset insomnia, snoring and other respiratory noises, irregular breathing, shortness of breath, palpitations, and nocturia. In patients with severe hypoventilation, arousal responses may be suppressed by sleep debt and lead in conjunction with cardiac arrhythmias to sleep-related death. Daytime symptoms of SDB include headaches, fatigue, excessive daytime sleepiness, and attention and memory deficits.

In 50%–70% of acute stroke patients, SDB, defined by an apnea-hypopnea index (AHI) ≥10/hour, is found (table 1).^{1–4,10–18} Patients with recurrent stroke have a higher SDB likelihood than first-ever stroke victims.⁹ In most studies, no link was found between SDB and stroke severity, topography, or presumed etiology.^4,13,14 In single studies, a link between poststroke SDB and macroangiopathy^3,17 or patent foramen ovale¹⁹ was suggested. The frequency of SDB was similar in patients with TIA and stroke,^1,2,10,20 suggesting that SDB may be not only a consequence of brain injury but also a preexisting condition.

The most common form of SDB is obstructive sleep apnea (OSA), which is caused by cessation of nasal flow due to upper airway collapse.^1–4 Cheyne-Stokes breathing (CSB) is characterized by cyclic fluctuations in breathing drive, hyperpneas alternating with apneas or hypopneas in a gradual waxing-and-waning fashion. Not rarely, stroke patients present OSA, central sleep apnea (CSA), and CSB.^1,2,21–23 CSA/CSB was first described in patients with bilateral strokes associated with disturbed consciousness or cardiorespiratory failure.²⁴ More recently, CSB presenting only during sleep has also be found after unilateral strokes with preserved consciousness and in the absence of overt cardiorespiratory dysfunction.^22,23 Poststroke CSA/CSB was linked to injuries to central autonomic networks, i.e., insula and thalamus.²² In these patients, CSA/CSB typically improves in the subacute stroke phase. CSA/CSB in chronic stroke patients is strongly associated with heart failure.²⁴

In the subacute stroke phase SDB improves, which suggests that breathing abnormalities are exacerbated by stroke (table 1). Nonetheless, ∼50% of patients still exhibit an AHI ≥10/hour at 2–3 months.^2,3,13,25,26 CSA/CSB appears to improve more than OSA based on some,^2,23 but not all²⁷ studies, which might explain its lower frequency in studies with late patient recruitment. Recovery of neurologic deficits (e.g., motor paresis), reduction of time spent in the supine position, amelioration of lung function (recovery from pneumonia), disappearance of cardiac complications (heart failure), and patient mobilization may contribute to CSA/CSB improvement.

Disturbances of wakefulness.

The spectrum of poststroke wakefulness disturbances is broad and includes hypersomnia (i.e., abnormal sleep propensity with increased sleep/24 hours), excessive daytime sleepiness (EDS; increased tendency to fall asleep at daytime), and fatigue (physical exhaustion, lack of energy, tiredness).

Poststroke hypersomnia can be found after subcortical (in particular caudate-putamen), thalamomesencephalic, upper pontine, medial pontomedullary, and even cortical strokes. In a recent study of 285 consecutive patients, we observed that at 21 ± 18 months after stroke, hypersomnia (27% of cases with sleep needs ≥10 hours/day), EDS (28% with Epworth Sleepiness Score ≥10), and fatigue (46% with fatigue severity scale ≥3) are frequent. Although hypersomnia improves during the first months, fatigue often persists in the chronic phase.²⁸ At 6 months, fatigue is more prevalent after minor stroke than TIA (56 vs 29%),²⁸ indicating that fatigue may be a consequence of brain damage.

The most dramatic form of poststroke hypersomnia is noted after paramedian thalamic stroke.^29–31 Patients typically present with sudden onset of coma. After awakening, patients exhibit severe hypersomnia and sleep-like behavior up to 20 hours/day, associated with attention, cognition, and memory deficits.³¹ Hypersomnia gradually improves within months, whereas cognitive deficits often persist, particularly after left-sided and bilateral stroke.³¹ Patients with bilateral strokes may report increased sleep needs for several years.

Severe hypersomnia also can be seen in patients with other stroke topographies.³² Hypersomnia with hyperphagia (symptomatic Kleine-Levin syndrome) has been reported after multiple cerebral strokes.³² Hypersomnia with cataplexy-like episodes, hallucinations, sleep paralysis, and low CSF hypocretin-1 levels (symptomatic narcolepsy) has been described in patients with posterior hypothalamic injury due to cerebral hypoxia induced by cardiac arrest or diencephalic stroke after craniopharyngioma surgery.³²

Insomnia.

Insomnia is defined by difficulty initiating or maintaining sleep, early awakenings, poor sleep quality, and daytime fatigue. Often poststroke insomnia is linked with stroke complications and not due to brain damage per se. Environmental factors (including noise, light, and intensive care unit monitoring) may play a role together with comorbidities such as cardiac failure, SDB, anxiety, depression, or pain.

Similar to hypersomnias, poststroke insomnias are not rare. In a systematic study of 277 consecutive patients, insomnia was found in the first months after stroke in 57% of patients.⁵ In 18% insomnia appeared de novo after stroke.⁵ In view of the known consequences of impaired sleep, i.e., fatigue, attention and cognitive problems, insomnia is expected to impair stroke recovery. Systematic studies supporting this hypothesis are lacking, however.

Less commonly, insomnia may be related directly to brain damage, most often in the brainstem. Patients with almost complete polysomnographically proven loss of sleep EEG patterns lasting over several months have been reported after pontine and pontomesencephalic strokes.³³ Patients with paramedian thalamic stroke may also present with insomnia, possibly linked to the inability to generate sleep spindles.³¹

Sleep-related movement disorders/parasomnias.

In a series of 137 patients who were consecutively examined 1 month poststroke, Lee et al.⁶ found de novo restless legs syndrome (RLS) symptoms in 12% of patients. RLS was observed mainly in pontine, thalamic, basal ganglia, and corona radiata infarcts. About two thirds of patients had bilateral RLS symptoms; one third had unilateral complaints contralateral to stroke.⁶ RLS always appeared within 1 week poststroke and was frequently accompanied by periodic limb movements in sleep (PLMS).

Pontine tegmental strokes can lead to REM sleep behavior disorder, in which patients act out usually violent dreams and exhibit polysomnographically increased phasic muscular activity and loss of REM atonia.³⁴

Sleep-related breathing disturbances.

Several studies have shown that SDB predisposes to vascular diseases.³⁵ Seminal studies in the 1990s exhibiting a link between habitual snoring and stroke prompted studies assessing the role of SDB. Stratifying 1,022 patients admitted to a sleep laboratory into groups with an AHI ≥5/hour or <5/hour who were followed up over 6 years, Yaggi et al.⁷ found that OSA is associated with an increased risk for stroke, TIA, and death, when patients were adjusted for age, gender, body mass index (BMI), arterial hypertension, diabetes, atrial fibrillation, hyperlipidemia, and smoking habits (hazard ratio [HR] = 2.0; confidence interval [CI] = 1.1–3.5). Patients with severe SDB exhibited a particularly high HR (3.3; CI = 1.7–6.3).

Marin et al.³⁶ recruited 1,387 male OSA patients, 377 simple snorers, and 264 healthy men, matched for age and BMI. After a mean follow-up of 10.1 years, patients with untreated severe OSA (AHI >30/hour) had a higher incidence of cardiovascular events than patients with mild to moderate OSA (5/hour < AHI ≤30/hour), and patients with OSA of any severity treated with continuous positive airway pressure (CPAP), snorers, and healthy participants. This observation persisted when data were adjusted for potential confounders (hypertension, diabetes, cardiovascular disease, lipid disorders, smoking status, besides others; odds ratio [OR] for cardiovascular death = 2.9; 95% CI = 1.2–7.5). The major limitation of the latter 2 studies is that vascular risk was examined in sleep laboratory patients, and therefore their risk profile may not represent the general population.

Three population-based studies overcame this limitation. In a cross-sectional analysis analyzing 1,475 subjects, Arzt et al.³⁷ showed that patients with an AHI ≥20/hour have an increased stroke risk (OR = 3.8; CI = 1.2–12.6) compared with those without SDB (AHI <5/hour) after adjustment for risk factors. In a longitudinal study, the authors evaluated the risk of having a first-ever stroke within 4 years. In the unadjusted population, persons with SDB (AHI ≥20/hour) had an elevated stroke risk. After adjustment for age, sex, and BMI, this association was no longer significant. The observation period may have been too short in that study.

Recruiting an elderly population of 394 noninstitutionalized stroke-free subjects (>70 years), Munoz et al.³⁸ reported after a follow-up of 6 years that persons with severe SDB (AHI >30/hour) have an increased stroke risk when adjusted for sex (HR = 2.5; CI = 1.0–6.1). A limitation of this latter investigation is that 2,528 persons had to be screened to include less than 400 participants.

In the Wisconsin sleep cohort, which followed up 1,522 persons (initial age of 48 ± 8 years) over 18 years, long-term mortality significantly increased with SDB severity after adjustment for age, sex, BMI, smoking, and hypercholesterolemia.⁸ As such, OR of 3.8 (95% CI = 1.6–9.0) and 5.2 (95% CI = 1.4–19.2) were found for overall mortality and cardiovascular mortality, when subjects with severe SDB (AHI ≥30/hour) were compared with those without SDB (<5/hour).

Mechanisms underlying the elevated stroke risk.

The mechanisms underlying the elevated stroke risk in patients with SDB have recently been summarized.³⁹ Obstructive apneas are accompanied by recurrent hypoxias, intrathoracic pressure changes, and sympathetic activation. Acutely, hemodynamic instability with blood pressure swings, cardiac arrhythmias, and cerebral blood flow fluctuations are observed. Chronically, oxidative stress, inflammation, and endothelial dysfunction lead to arterial hypertension, glucose intolerance, and atherosclerosis. In patients with SDB, the risk of developing hypertension and diabetes was increased by 2.6-fold and 5.5-fold in the Wisconsin sleep cohort.⁸ SDB also results in an increased risk of myocardial ischemia and cardiac rhythm abnormalities.³⁹

CPAP therapy attenuates the vascular consequences of SDB.³⁹ In a controlled randomized trial in patients with SDB without stroke, CPAP reduced mean arterial blood pressure (MABP) by 3 mm Hg.⁴⁰ Another study obtained similar results.⁴¹ In pharmacologic studies, a relative reduction of stroke risk by ∼5% is expected for every mm Hg systolic blood pressure is decreased.⁴² Thus reducing MABP by 3 mm Hg should decrease stroke risk by ∼15%–20%. In other studies, CPAP reversed the inflammatory changes and endothelial dysfunction, and ameliorated cardiac rhythm abnormalities and ischemic ECG changes induced by SDB.³⁹

Disturbances of wakefulness.

EDS has been linked with stroke risk. In a hospital-based cohort of 181 patients with first-ever stroke who were compared with age- and sex-matched healthy subjects, prestroke EDS but not SDB was associated with stroke (OR = 3.1; 95% CI = 1.6–6.1).⁴³ This association was independent of other risk factors, namely hypertension, smoking, and alcohol. The underlying mechanisms are unknown.

Sleep-related movement disorders/parasomnias.

Whether RLS is associated with cardiovascular risk in the general population was recently assessed in a cross-sectional study comprising 3,433 individuals (mean 67.9 years).⁴⁴ Notably, the presence of self-reported RLS significantly increased the prevalence of coronary heart disease (OR = 2.0; 95% CI = 1.4–3.0) and cardiovascular diseases (OR = 2.1; 95% CI = 1.4–3.0) after adjustment for age, sex, BMI, diabetes, blood pressure, cholesterol, and smoking history, besides others.⁴⁴ The association was stronger in patients with severe than moderate RLS symptoms. In RLS, PLMS are associated with short-lasting elevations of blood pressure and sympathetic hyperactivity.⁴⁵ Whether these effects are sufficient to promote the development of secondary hypertension remains controversial.

Sleep-related breathing disturbances.

Several studies indicate that SDB negatively affects early neurologic worsening,²¹ hospitalization duration,¹⁴ and short-term^14,21 and long-term^3,46,47 outcome. The presence of SDB was associated with a worse Barthel index at 6 months after stroke⁴⁶ and predicted a higher mortality during the following 10 years.^3,46,47 Noteworthily, an elevated mortality was found only in patients with OSA but not CSA.⁴⁷

Several reasons including SDB per se, hypertension, and co-shared risk factors have been discussed for the poor recovery of SDB patients. Stroke patients with moderate to severe SDB (AHI >30/hour) exhibit higher daytime and nighttime blood pressure, compared with patients without SDB (AHI <10/hour).¹⁴ Considering the close link between blood pressure and stroke risk in hypertensive patients without stroke,⁴² the less favorable prognosis of stroke patients with SDB may be explained by hypertension and its complications, i.e., macro- and microangiopathies.

Disturbances of wakefulness.

EDS and fatigue are associated with neuropsychiatric (depression, anxiety) and cognitive disturbances, and have a negative impact on rehabilitation and quality of life.⁵ They are therefore expected to influence stroke recovery. Few studies have addressed this issue. A population-based study following up 21,268 subjects over 22 years supports the idea that increased sleep needs in the general population are associated with increased mortality.⁴⁸ In that study, long sleepers (>8 hours/day) had a significantly elevated death risk as compared with average (7–8 hours) sleepers (+24% and +17% in men and women) after adjustment for several sociodemographic and lifestyle factors, including BMI, smoking status, and depression.⁴⁸

That poststroke fatigue is an independent predictor of independence and death was shown in a cohort of 8,194 stroke patients.⁴⁹ In a questionnaire, 39% of patients reported fatigue 2 years after their stroke. Patients with fatigue were more likely to be institutionalized and had an increased mortality rate in the subsequent year.⁴⁹ The major weakness of this study is that it did not adjust for other medical conditions unrelated to the stroke (e.g., infections).

Insomnia.

Not only increased but also reduced sleep needs are considered to influence life expectancy. In fact, not only long sleepers, but also short sleepers (<7 hours sleep/day) were shown to be associated with an increased mortality in population-based studies (+26% and +21% as compared with normal sleepers in men and women).⁴⁸ Snoring as covariate did not change the results. A significantly increased mortality was noted in frequent users of hypnotics/tranquilizers (+31% and +39%).⁴⁸ A recent article has shown that insomnia with objective short sleep duration is associated with a high risk of hypertension.⁴⁹ Whether insomnia influences mortality after stroke remains unknown.

Sleep-related movement disorders/parasomnias.

That RLS/PLMS may be associated with an increased mortality was suggested by a Swedish population-based study following up 5,102 subjects (30–65 years) over 20 years.⁵⁰ In a multivariate analysis adjusted for age, sleep time, lifestyle factors, medical conditions (e.g., diabetes, hypertension), and depression, women with RLS and daytime sleepiness had an excess mortality compared with women without RLS and daytime sleepiness.⁵⁰ No association of RLS and sleepiness with mortality was found in men. Whether RLS/PLMS influences stroke outcome remains to be shown.

Sleep-related breathing disturbances.

Screening for SDB may be warranted in all stroke patients who may potentially accept CPAP treatment. SDB can be accurately diagnosed by respiratory polygraphy, in which nasal airflow, respiratory movements, and capillary oxygen saturation are monitored. Polysomnography offers additional information, but is expensive and less commonly available. It should therefore be reserved for unclear cases.

SDB treatment in stroke patients represents a technical and logistical challenge (table 2). Treatment should always include prevention and therapy of secondary complications (respiratory infections, pain) and cautious use of alcohol and sedative-hypnotic drugs, which negatively affect breathing during sleep (table 3). Patient positioning in the acute phase influences oxygen saturation as well (table 3).

CPAP is the treatment of choice for OSA (table 3).^40,41 Since SDB improves after the acute stroke phase, CPAP systems which allow automatic titration of CPAP pressure may be preferable. In patients with CSA or CSB, improvement can be achieved with oxygen (table 3).²⁴ Novel methods of ventilatory support (e.g., adaptive servoventilation) may be considered. Tracheostomy and mechanical ventilation may become necessary in patients with central hypoventilation.

A total of 10 studies investigated effects of CPAP treatment after stroke (see table 2). CPAP compliance has been reported to be as high as 70% in the rehabilitation setting.⁵¹ Other groups working in the acute^{3,13,18,52,53} but also subacute^54–57 stroke phase reported lower percentages, mostly around 25%–50%. In one randomized trial of stroke patients with severe SDB (AHI ≥30/hour), CPAP was used only 1.4 hours/night.⁵⁶ This study was prematurely stopped because of poor recruitment. In another randomized trial, CPAP use was slightly better (mean 4.1 hours/night).⁵⁵ Poor compliance was related to spontaneous improvement of SDB, lack of EDS in most patients, and motor (facial and bulbar palsy) and cognitive deficits (confusional states, dementia, aphasia, anosognosia).

Few studies assessed the efficacy of CPAP (table 2). In a group of 41 patients with stroke and SDB, who were treated with CPAP over 10 days, Wessendorf et al.⁵¹ reported an improvement of subjective well-being and nighttime blood pressure values. Sandberg et al.⁵⁵ observed an improvement of depressive symptoms in 31 stroke patients with SDB treated over 28 days. Martinez-Garcia et al.⁵⁴ studied 51 patients with stroke and an AHI ≥20/hour. Compared with noncompliant subjects, the 15 CPAP-compliant patients had a lower incidence of new vascular events (7% vs 36%) over a follow-up observation period of 18 months (table 2). In a recent large study, Martinez-Garcia et al.⁵⁷ found that 68 stroke patients with an AHI ≥20/hour who did not tolerate CPAP had a higher 5-year mortality than 28 patients who tolerated CPAP.

Future studies should probably include selected patient populations and have vascular (e.g., blood pressure, stroke recurrence risk) as well as nonvascular (e.g., overall outcome, cognitive functions, quality of life) endpoints, using sham CPAP as control condition. Because of the low CPAP compliance and the relatively low incidence of new vascular events after stroke, large study populations, long observations periods, or use of surrogate markers of vascular outcome will be needed. In view of high dropout rates, patients with severe facial and bulbar palsy, severe motor deficits, confusional states, dementia, aphasia, and anosognosia should be excluded from randomized trials.

In the absence of final proof of efficacy, there is no justification to recommend CPAP treatment systematically to all stroke patients exhibiting SDB at the present stage. We preferentially recommend CPAP to stroke and TIA patients with moderate to severe obstructive apneas, daytime symptoms, and high cardiovascular risk profile (including therapy-resistant hypertension).

Disturbances of wakefulness.

Most SWD can be suspected on clinical grounds. Not uncommonly, SWD is first noted, however, when patients leave the hospital and have to resume work. Sleep questionnaires (e.g., Epworth Sleepiness Scale score, Fatigue Severity Scale, Pittsburgh Sleep Questionnaire) should be included in the standard assessment of stroke patients. Actigraphy may be helpful to estimate changes in sleep-wake rhythms and sleep/rest needs,³¹ although a differentiation between sleep and inactivity due to severe motor deficits, apathy, or depression may be difficult. Sleep EEG and vigilance tests such as Multiple Sleep Latency Test and Maintenance of Wakefulness Test can be used in selected cases. The correlation of poststroke SWD and sleep EEG is poor when brain damage includes thalamo-cortical structures.^30,31 In poststroke hypersomnia, sleep EEG may reveal a reduction, less commonly an increase of non-REM or REM sleep.³¹

Treatment of poststroke hypersomnia is often difficult (table 3). In patients with paramedian thalamic stroke, improvement of sleep behavior was reported with 20–40 mg bromocriptine, 200 mg modafinil, and 20–50 mg methylphenidate.^29,32,33 A favorable influence on early stroke recovery was observed both after levodopa (100 mg/day) and methylphenidate (5–30 mg/day),⁵⁸ an effect that may at least partially be related to improved wakefulness. Treatment of depression with stimulating antidepressants may also improve hypersomnia.

Insomnia.

Similar to hypersomnia, the recognition of poststroke insomnia occurs mainly on clinical grounds. Actigraphy may be used to document and quantify the reduction in sleep time and quality. Polysomnography is only rarely needed, e.g., to rule out intrinsic sleep problems.

Treatment of poststroke insomnia should include placement of patients in quiet rooms at night, protection from noise and light, light exposure and physical activity during the day, and, when unavoidable, temporary use of hypnotics that are relatively free of cognitive and muscle-relaxant effects, such as zolpidem (table 3). Benzodiazepines may provoke neuropsychological deficits and result in the reemergence of motor symptoms.⁵⁹ Sedative antidepressants may also improve poststroke insomnia (table 3). In a study of 51 stroke patients, 60 mg/day of mianserin led to a better improvement of insomnia than placebo, even in patients without depression.⁶⁰

Sleep-related movement disorders/parasomnias.

RLS is a clinical diagnosis. The presence of PLMS is best documented by polysomnography.

Patients with stroke-related RLS can be treated with dopamine agonists (ropinirole 0.25–1 mg/day, pramipexole 0.125–0.5 mg/day; see table 3). Lee et al.⁶ reported marked relief in all except 2 patients, who had mild improvement. All patients had persistent RLS until the end of the study period, indicating that most patients require treatment.⁶ Spontaneous improvement was noted in only 4 out of 17 patients.⁶ Dopaminergic drugs can be used for poststroke RLS/PLMS. Antidepressants, neuroleptics, metoclopramide, and lithium should be avoided wherever possible, because they can aggravate RLS and PLMS.

REM sleep behavior disorder is diagnosed by polysomnography and can be treated with clonazepam (0.5–2.0 mg 1 hour before sleep; table 3).