Opioid use whether acute or chronic, illicit or therapeutic is prevalent in Western societies. Opioid receptors are located in the same nuclei that are active in sleep regulation and opioid peptides are suggested to be involved in the induction and maintenance of the sleep state. mu-Opioids are the most commonly used opioids and are recognized respiratory depressants that cause abnormal awake ventilatory responses to hypercapnia and hypoxia. Abnormal sleep architecture has been reported during the process of opioids induction, maintenance and withdrawal. During induction and maintenance of opioid use there is reduction of rapid eye movement (REM) sleep and slow wave sleep. More recently, central sleep apnoea (CSA) has been reported with chronic opioid use and 30% of stable methadone maintenance treatment patients have CSA. Given these facts, it is sobering to note the paucity of human data available regarding the effects of short and long-term opioid use on sleep architecture and respiration during sleep. In this manuscript, we review the current knowledge regarding the effects of mu-opioids on sleep and respiration during sleep and suggest research pathways to advance our knowledge and to explore the possible responsible mechanisms related to these effects.

Opioid use whether therapeutic or illicit is common worldwide. In year 2000, approximately 1.2% of the American population reported heroin use at least once in their lifetime.1 In Australia, the estimate of recent illicit opioid users was 0.6% of the 14 yrs and older population in 2001 compared to 1% in 1998.2 Australian Institute of Health and Welfare. 2001 National Drug Strategy House Hold Survey: first results. Canberra, Australia, 2002.2 More than 140,000 patients were receiving methadone maintenance treatment (MMT) in 1998 in the United States,3 and in Australia 30,000 were receiving MMT in year 2000.4 Opioids are also commonly used for acute and chronic pain management and as an adjunct to anaesthesia and occasionally for the restless legs syndrome.5 and 6 An American study reported that 80% of 2118 cancer patients referred to a pain service were prescribed opioids.6 In year 2004, there were more than 410,000 registrants for opioid use in American pharmacies compared to around 390,000 in 1997.7 The average gram weight per registrant increased 7.3 fold for oxycodone, 5 fold for methadone, 4.6 fold for fentanyl base and 3 fold for hydrocodone from 1997 to 2004.7 Prescribed opioids related deaths account for most of non-illicit drug poisoning deaths in America and the problem has been increasing in the past decade.8 Abuse of opioid analgesics is very common and in 2005, 9.5% of American 12th graders reported using Vicodin and 5.5% of these students reported using OxyContin in the past year.9

Many opioid receptors are located in the same nuclei that are active in sleep regulation10 and it has been suggested opioid peptides are involved in the induction and maintenance of the sleep state.11 Chronic opioid use has been hypothesized to cause disturbed sleep as well as excessive day-time sleepiness and fatigue.12 A few studies have reported abnormal sleep architecture in opioid users but most were performed prior to 1990 and few tested breathing during sleep though opioids are well-known respiratory depressants.5 In recent studies, central sleep apnoea (CSA) has been found in stable MMT patients and in patients prescribed time-release opioid analgesic management.13, 14 and 15 Abnormal ventilatory responses to hypercapnia (HCVR) and hypoxia (HVR) have also been noted in stable MMT patients.16 Infants born to substance-abusing mothers have a higher prevalence of periodic breathing during sleep and a 5–10 times increased risk of sudden infant death syndrome (SIDS) compared to normal infants.17 and 18 The potential symptoms and other sequelae related to CSA and periodic breathing have not been discussed in reviews of opioid use for chronic pain or in methadone substitution programs.6

Morphine-like μ-opioids are clinically the most commonly used opioids and this review focuses on their effects on sleep and respiration during sleep in humans. Where evidence is available, we will discuss the pathogenesis of abnormalities described and will also discuss future research directions given the considerable lack of knowledge in this important area of medicine. The scope of this review includes sleep and respiration during sleep in acute and chronic opioid analgesic use, opioid abuse and in MMT programs.

Opioids and control of sleep

There are four major classes of endogenous opioid receptors in the central nervous system: μ, δ, κ and nociceptin/orphanin FQ (N/OFQ) receptor.5 Each of these receptor subtypes has a distinct profile in terms of its pharmacology as well as its distribution within the brain and spinal cord. Most of the clinically used opioids are relatively selective for μ receptors, such as morphine and methadone.5 It appears that REM suppression is associated primarily with the actions of μ-opioid receptor agonists.19 Three classes of opioid peptides have been identified: the enkephalins, endorphins and dynorphins.5 These have been shown to have a role in sensory modulation and analgesia and may be important in the onset and maintenance of sleep, and therefore be involved in attenuation of arousal and waking.20 Enkephalin is contained in neurons that are widely distributed through the brain and regions involved in slow wave sleep (SWS) such as the solitary tract nucleus, the preoptic area and the raphe, where it is colocalized with serotonin receptors.20 Enkephalin containing fibres innervate the locus coeruleus noradrenergic neurons which are inhibited by locally delivered opioids and produce decreased awakenings and increased SWS.20 β-Endorphin is derived from prepro-opiomelanocortin (POMC) which is also processed into the non-opioid peptides adrenocorticotropic hormone (ACTH), melanocyte-stimulating hormone (α-MSH) and β-lipotropin (β-LPH).5 ACTH is the hormone closely related to stress and α-MSH has been suggested to induce sleep and increase SWS. The association of sharing the same precursors implies a close physiological linkage between stress, sleep and the opioid systems.5

The mechanism of opioid peptide action on sleep control remains unclear. It has been hypothesized that opioid peptides in conjunction with the peptide neurohormone vasopressin are involved in the induction and maintenance of the sleep state through a complex and modifiable circadian mechanism driven by the suprachiasmatic nuclei (SCN).11 Vasopressin, one of the neurohormones in the circadian pacemaker SCN, has been shown to have a close relationship to circadian rhythms.21 Vasopressin causes the secretion of endorphins into the cerebro–spinal fluid (CSF), while pain and opioids stimulate the secretion of vasopressin from the pituitary.22 In the supraoptic and paraventricular nuclei, vasopressin is stored with the opioid peptide dynorphin which also shows circadian variability of its blood levels.23 It is possible that both vasopressin and the opioids are part of the neurochemical mechanism driven by SCN to maintain the daily sleep and wake rhythm.11

Exogenous opioids may affect the activity of opioid receptors by binding to the same sites as those of endogenous opioid peptides.24 Endogenous enkephalin production is linked by a negative feedback mechanism to the serum level of opioids which are high in chronic opioid users.25 Given the possible links between endogenous opioids and control of sleep previously discussed, it is reasonable to suggest that acute and chronic opioid use may have an effect on sleep hygiene and sleep architecture.

Opioids and control of respiration when awake and during anaesthesia

In humans, the primary opioid receptors involved in control of respiration are assumed to be μ-receptor type.5 δ-Receptors may exert modest respiratory depressant effects, whereas κ-receptors have little respiratory depressant activity.26 Acute use of μ-receptor stimulating opioids can cause dose dependant depression of respiration.5 Opioid receptors in brain stem (medulla, pons, nucleus tractus solitarius, and nucleus ambiguous), spinal cord, and peripheral sites such as lung tissue are involved in the respiratory depression. However, the brain stem respiratory centres predominate with regard to this effect.5, 26 and 27 Both HCVR and HVR can be significantly reduced by acute use of opioids.26 and 28 Opioids can also blunt the increase in respiratory drive normally associated with increased loads such as increased airway resistance.27 Acute opioid use may cause increased respiratory pauses, delays in expiration, irregular and/or periodic breathing and decreased/normal tidal volume.27 The prolonged expiratory time in the respiratory cycle induced by opioids often results in greater reductions in respiratory rate than in tidal volume.26 Increased tidal volume variability was reported to be a better predictor of respiratory depression than a fall in respiratory rate when remifentanil was infused during dental anaesthesia.29

Waters et al. studied 13 children with OSA and 24 normal subjects undergoing tonsillectomy.30 They found that under inhalational anaesthetic and spontaneous ventilation, the OSA group hypoventilated and tended to have higher end tidal CO2 levels than the normal group. Following fentanyl injection, 6 of the OSA group exhibited central apnoea compared with one of the normal subjects. The production of central apnoea post opioid injection in both groups was related to end tidal CO2 higher than 50 Torr.30 Though this study has methodological flaws, the data suggests that at least in children, subjects prone to hypoventilation may progress to central apnoea when given a μ-opioid, however, whether this equates to CSA in these children is unknown.

With long-term use of opioids, subjects have reduced HCVR although tolerance appears to develop.16 and 31 An early study assessing HVR in MMT patients suggested blunting of HVR in both the acute and chronic stages of methadone use.31 However, this study should be interpreted with caution as there was no baseline data from normal subjects available for comparison.31 In contrast, in our study of stable MMT patients, HVR appears to be increased (Fig. 1).16 The causes for this finding are not clearly known and may relate to long-term stimulation of the hypoxic response by long-term intermittent hypoxia.16 The high HVR and low HCVR we found in the stable MMT patients related to changes in respiratory rate and not tidal volume response.16

Opioid use, sleep and respiration during sleep

Given the large opioid using population and the possible close link between endogenous opioids and control of sleep, there is a scarcity of studies investigating sleep in human adult subjects using opioids.12 There are studies assessing sleep in animals that show changes in sleep architecture with acute and chronic μ-opioid use.32 and 33 We were unable to find data related to effects of chronic use of μ-opioids on respiration during sleep in animals. Breathing during sleep in human subjects using opioids has been poorly studied despite the fact that the commonly used μ-opioids are known to depress respiration and sleep-disordered breathing in its own right can significantly affect sleep architecture.5

Table 1 shows the findings and methodologies of 18 human studies available on PUBMED published between 1966 and 2005 investigating sleep and respiration during sleep in adult humans using opioids. We cite only those studies using objective measurements and reported in English. Studies assessing the effects of opioids on sleep and respiration during sleep in post-operative surgical patients and restless legs syndrome patients are not included in the table. Restless legs syndrome itself can significantly disturb sleep and the studies in the anaesthetic literature are often confounded by poor patient selection, use of concomitant anaesthetic agents, analgesics and post-operative pain. In addition, natural sleep is clearly different to anaesthesia which is a state of unrousable unconsciousness.34 During anaesthesia, there is dose-dependant depression to most of the vital functions including respiration.34 Abnormal breathing patterns in anaesthesia are different to sleep-disordered breathing although the tendency to upper airway obstruction during sleep and during anaesthesia are probably related.35 Of the studies listed in Table 1, 16 used morphine like μ-opioids and 3 used opioid antagonists.

Summary of opioid effects on sleep

Despite the inherent methodological limitations discussed above, the studies provide useful information about the effects of opioids on sleep. There are four basic phases of opioid dependence and withdrawal: drug induction phase, drug maintenance phase, acute abstinence phase and protracted abstinence phase.38 Sleep architecture changes are different for each of the 4 phases. In general, during the induction phase, the use of morphine-like opioids significantly disrupts sleep with reduced REM sleep and SWS and increased wakefulness and arousals from sleep. TST and SE are usually reduced while percentage stage 2 sleep and REM sleep latency are often increased. During the maintenance phase of μ-opioid use, the decreases in SWS and REM sleep tend to normal as do the increases of wakefulness, arousal and REM sleep latency. Vocalization during REM sleep, significant delta burst and increased daytime sleepiness may commonly appear in this phase. Limited evidence is available regarding sleep during acute withdrawal from chronic opioid use.38 Changes in sleep from withdrawal of short-term opioid administration39 may be different to the changes seen in withdrawal from chronic opioid use. Significant insomnia is the major complaint during chronic opioid withdrawal, accompanied by frequent arousals and decreased REM sleep. During the protracted abstinence phase, TST significantly increases with rebound of SWS and REM sleep. After chronic methadone use, the rebound of SWS and REM sleep usually occurs between 13 and 22 weeks following withdrawal of the opioid.40 and 41

Chronic opioid use is associated with symptoms of fatigue and excessive daytime sleepiness.12 and 14 The abnormal sleep architecture discussed above can affect daytime functioning in its own right. However, it is difficult to know how much the abnormal sleep architecture noted in these studies impacts on daytime function and excess daytime sleepiness.

Within the opioid class, morphine and methadone have comparable effects on sleep and are half as potent as heroin with regard to EEG measures.42 The difference between morphine and methadone on sleep is that chronic morphine use gives measures of persistent sleep architecture disturbances which are not found with chronic methadone use.40, 41 and 43 Further studies employing larger subject numbers and improved methodology are necessary to gain a clearer and more comprehensive understanding of opioid effects on sleep and to explore the long-term affects of sleep architecture changes on the subjects’ daytime function.

Respiration during sleep with acute opioid use

There are only two human studies assessing respiration during sleep with acute μ–opioid use. Robinson et al. assessed awake pharyngeal resistance, HCVR, HVR and respiration during sleep in 12 healthy adult humans after ingestion of 2 and 4 mg of oral hydromorphone.44 Awake pharyngeal resistance, HCVR and breathing during sleep did not change significantly following either dose of the drug, although there is a trend toward increased apneas (more than doubled) and decreased hypopneas with the 4 mg dose. Awake HVR was significantly reduced after 4 mg of the drug.44 Similarly, Shaw et al. measured breathing during sleep on 7 healthy adults after injection of morphine (0.1 mg/kg) and did not find an increase in sleep-disordered breathing compared to either baseline or placebo use.45 Further studies with larger sample size are needed to test the effects of acute opioid use on respiration during sleep.

Respiration during sleep with chronic opioid use

Few studies have investigated respiration during sleep in subjects using opioids long term.13, 14, 15 and 46 The studies include two that assessed stable MMT subjects;13 and 14 one assessed 3 subjects using chronic time release opioid analgesics;15 and one assessed subjects with restless legs syndrome.46 The stable MMT subject studies were the only studies that matched patients and normal subjects for age, sex and BMI.13 and 14 CSA was noted in 20 of 60 subjects in the MMT cohort and no CSA was noted in the normal subjects.13 and 14

In the largest cohort study assessing respiration during sleep in subjects using opioids chronically, CSA was found in 30% of 50 stable MMT patients while obstructive sleep-disordered breathing was similar in the MMT cohort and normal subjects.14 The CSA in the MMT patients is more prominent in NREM sleep than in REM sleep and did not cause increased arousals compared to normal control subjects.14 The CSA described in stable MMT patients is of periodic and non-periodic type.13 and 14 During sleep, MMT patients have only mildly reduced arterial oxygen saturation and mildly increased transcutaneous arterial carbon dioxide tension.14

CSA, periodic breathing and Biot’s breathing pattern (i.e. “ataxic breathing” with unpredictable irregular pattern) was reported in 3 females using opioids chronically for pain relief.15 This report lacks data regarding a clear definition of “ataxic breathing” and whether the irregular breathing was or was not related to arousals or transitional sleep.15 This breathing pattern appears to be similar to sub-criteria CSA or the non periodic breathing CSA we describe in the stable MMT patients14 *D. Wang, H. Teichtahl, O.H. Drummer, C. Goodman, G. Cherry and D. Cunnington et al., Central sleep apnea in stable methadone maintenance treatment patients, Opioids have been suggested to interfere with pontine and medullary respiratory centres that regulate respiratory rhytmicity based on various cats studies.27 To date there is however no evidence regarding the prevalence and possible mechanisms of the ataxic/Biot’s breathing pattern during human sleep in chronic opioid use.

Seven patients with restless legs syndrome were studied with PSG before and after long-term opioid monotherapy over an average of 7 years.46 Two of the seven patients developed sleep apnoea with respiratory disturbance index of 10 and 15 and a third patient developed worsening of pre-existing sleep apnoea. The type of sleep apnoea found in the 3 patients was not reported.46 It therefore appears that further studies with larger sample size and improved methodology are needed to elucidate if the CSA noted in stable MMT patients also exists in the patients using opioids chronically for pain relief and restless legs syndrome.

The outcome of the CSA observed in these groups of patients is unknown. For example, we do not know if these patients with CSA have higher morbidity or mortality than those patients using long-term opioids but without CSA. We also do not know whether the CSA noted in these patients contributes to daytime dysfunction, though it is clear that stable MMT patients are more depressed and sleepier during the day, and have poorer general health than normal subjects.14 and 47 What we do know is that CSA is not the sole cause of excess daytime sleepiness in the MMT patients.14

Potential mechanisms for CSA with chronic opioid use

Though the human studies showing CSA with chronic opioid use are of interest, only 2 have assessed potential mechanisms related to this finding.14 and 16 One of the major problems with using human subjects for investigating the pathogenesis of CSA in these populations is that chronic opioid use is usually associated with a number of other medical and psychiatric conditions.47 Therefore, these subjects often use concomitant therapy such as benzodiazapines and antidepressants, and many have a history of cigarette abuse.47 These confounders make it difficult to reach conclusions regarding pathogenetic mechanisms for CSA without testing large numbers of subjects and this can be difficult in these patient populations. We therefore suggest that future research be targeted at further developing animal models to better assess the mechanisms involved in CSA with chronic opioid use. However, even with the above caveats, the data obtained from our previous studies can give direction for further animal and human research and we will in detail discuss some of the information we have obtained in a cohort of stable MMT patients.14 These MMT patients were on stable doses of methadone and had been in the treatment program for a minimum of 2 months.

The CSA noted in the MMT patients appears to be different to the Cheyne–Stokes respiration seen in congestive heart failure patients.14 For example, the CSA of MMT patients shown in Fig. 2 are not of the crescendo–decrescendo type and have much shorter cycle time than the Cheyne–Stokes respiration of congestive heart failure.14 and 48 In addition these MMT patients had normal cardiac function.49 The CSA in the MMT patients is not of the idiopathic type as these subjects lacked the typical characteristics of idiopathic CSA, such as male preponderance and significant arousals during sleep.14 and 50 Hypercapnia alone does not seem to explain the CSA in these stable MMT patients as their lung function tests were only mildly abnormal and their awake arterial CO2 tension was marginally raised in only 10 of the 50 patients.48

We currently believe that no simple cause and effect relationship can explain CSA with chronic opioid use. An important clue is that methadone blood concentration is the best predicting variable for CSA in stable MMT patients.14 Another important lead is that stable MMT patients have blunted central chemosensitivity but elevated peripheral chemosensitivity.16 We therefore believe that the pathogenesis for CSA in this group is most likely multifactorial in nature and related to a variable interplay of abnormalities of central controller function and central and peripheral chemoreceptor sensitivity. μ-Opioids are well known central respiratory depressants.5 The significant association between CSA and methadone blood concentration suggests that depressed central controller plays a critical role in the genesis of CSA.14 Structural brain damage has been reported to occur secondary to cerebrovascular accidents associated with prior illicit drug use.51 This brain damage particularly if it occurs in the midbrain or brainstem would lead to central respiratory controller dysfunction in these MMT patients. Functional and structural MRI studies are required in this group of patients to assess this hypothesis.

As shown in Fig. 1, stable MMT patients have significantly reduced HCVR but increased HVR, which may suggest blunted central chemosensitivity but elevated peripheral chemosensitivity.16 An imbalance of central and peripheral chemosensitivity has been suggested to pose a greater risk for periodic breathing.52 When carotid chemoreceptor stimulation becomes the dominant sensory input to the respiratory controller relative to the input of the medullary chemoreceptors, the breathing pattern tends to become instable.52 The combination of antidepressant and methadone use may further reduce the already blunted HCVR16 and lead to an increased risk of CSA.14 We have shown that of the stable MMT patients with CSA, 57% of those receiving both antidepressant and methadone had central apnoea index >10.14 These patients had significantly reduced central chemosensitivity compared to the patients taking methadone alone.16 This mechanism is therefore similar to that of hypercapnic-type CSA.50 Acute opioids use can significantly reduce HVR, however, long-term application of opioids may lead to recurrent episodic hypoxia which may continuously stimulate peripheral chemosensitivity and lead to an increased HVR.14 and 28 It has been reported that exposing subjects to very mild and short-term hypoxia can cause an increase in HVR.53 High peripheral chemosensitivity itself is a predisposing factor for sleep-disordered breathing54 and has been shown to occur in high altitude periodic breathing55 and in CSA of congestive heart failure.48

The above mechanisms may contribute to the CSA seen in chronic opioid use. Each mechanism may occur in isolation or in variable combinations with other mechanisms.

Opioid use and SIDS

The SIDS is acknowledged as a major cause of death in infancy.56 In Australia in the late 1990s’, SIDS killed approximately one in every 1200 infants. Neonatal life of infants born to substance abusing mothers or born to those using opioids chronically is similar to that of chronic opioid use followed by a natural opioid abstinence period. Following birth there is an acute withdrawal of supply of exogenous opioids through placental circulation while endogenous opioids production is low. This may cause functional impairment in the CNS and altered sleep patterns.57 In pregnant MMT patients, foetal breathing movements and the response to carbon dioxide are significantly less than in normal subjects, and are further decreased after receiving methadone.58 Infants born to substance-abusing mothers have been shown to have an impaired repertoire of protective responses to hypoxia and hypercapnia during sleep.59 They have higher prevalence of periodic breathing during sleep and a 5–10 times increased risk of SIDS compared to normal infants.17, 18 and 59 In a population-based study, more than 1.2 million infants born in New York City between 1979 and 1989 were investigated and infants born to mothers in MMT were found to have 3.6 times increased chance of having SIDS compared to infants born to mothers not using methadone.60

Practice points

• There is increasing acute and chronic use of illicit and prescribed opioids in Western societies.
• Sleep architecture is abnormal with opioid use and the abnormalities of sleep architecture are different across the four basic phases of opioid dependence and withdrawal.

• CSA including periodic and non-periodic breathing pattern have been reported with chronic opioid use and 30% of stable MMT patients have CSA. The potential impacts on patient outcomes of these findings are unknown.

• The mechanisms producing CSA with chronic opioid use probably involve changes in central and peripheral ventilatory control mechanisms.

• Infants born to substance abusing mothers have a higher prevalence of periodic breathing during sleep than normal infants and a 5–10 times increased risk of SIDS compared to infants born to non-substance abusing mothers.

Research agenda

• Assess the prevalence of sleep-disordered breathing in patients using opioids long-term.
• Animal and human studies are required to explore the pathogenesis of sleep-disordered breathing noted with chronic opioid use.

• Assess the short and long-term effects of sleep architecture changes and sleep-disordered breathing with chronic opioid use and investigate strategies to prevent the complications.

• Data is required regarding the acute and chronic interactions of opioids, antidepressants and benzodiazapines on sleep architecture and respiration during sleep.

• Animal and human studies with improved methodology are needed to assess the effects of acute opioid use on respiration during sleep.

• Data is required for patients with OSA undergoing surgery to assess the effects of opioid anaesthesia and of opioid analgesia on post-operative respiration both awake and during sleep.

• Develop clinical guidelines as to when patients using opioids should be investigated for sleep disorders including sleep-disordered breathing.