Quetiapine modestly improves sleep and breathing but impairs next-day performance in people with obstructive sleep apnea and difficulty maintaining sleep: a randomized controlled trial

Abstract

Rationale

Quetiapine is commonly prescribed “off-label” to people with insomnia symptoms. People with undiagnosed obstructive sleep apnea (OSA) frequently report insomnia symptoms, particularly difficulty maintaining sleep. In 2023, 10.7 million prescriptions were dispensed for quetiapine in the United States. Yet, there is limited information regarding the effects of quetiapine on sleep, breathing, and next-day performance in people with OSA.

Objectives

To determine the effects of 50 mg of quetiapine versus placebo on OSA severity and other polysomnographic parameters plus next-day vigilance and driving simulator performance in people with OSA who also reported difficulty maintaining sleep.

Methods

We performed a double-blind, randomized, placebo-controlled, cross-over study (NCT05303935) in 15 people with OSA and difficulty maintaining sleep. Participants were studied overnight via polysomnography twice approximately 1 week apart and received either 50 mg of quetiapine or placebo (order randomized) just prior to sleep. A 10-minute psychomotor vigilance task (PVT) and 30-minute driving simulator test were performed each morning.

Results

Compared to placebo, quetiapine reduced the apnea/hypopnea index (primary outcome) (mean ± SD: 27 ± 16 vs 20 ± 12 events/hour, P = .009), arousal index (32 ± 16 vs 25 ± 9 arousals/hour, P = .012), and increased sleep efficiency (80 ± 11 vs 87 ± 9%, P <.001) without worsening hypoxemia (mean overnight oxygen saturation: 94.7 ± 1.2% vs 94.5 ± 1.5%, P = .38). However, next-morning vigilance (median PVT reaction time: 336 ± 48 vs 382 ± 84 ms, P = .008) and driving simulator performance (steering deviation: 72 ± 38 vs 96 ± 53 cm, P <.01) were significantly impaired with quetiapine.

Conclusions

Consistent with a hypnotic effect, a single night of low-dose quetiapine reduced OSA severity as measured via the apnea/hypopnea index and increased sleep efficiency without worsening overnight hypoxemia. However, there was evidence of next-day impairment in vigilance and driving simulator performance.

Clinical Trial Registration

NCT05303935

Introduction

Quetiapine is a second-generation antipsychotic that has been approved to treat schizophrenia and bipolar disorders in several countries including the United States, United Kingdom, and Australia. However, quetiapine is now prescribed off-label more frequently than its approved use, primarily for insomnia, sleep disturbances, and anxiety by primary care practitioners in several countries.^1-3 Quetiapine prolongs sleep duration and reduces nocturnal awakenings, which may explain its growing popularity in general practice.⁴ Indeed, in 2023, quetiapine was prescribed 10.7 million times in the United States alone, 76% of prescriptions at doses intended for short-term titration or below the levels recommended for approved therapeutic indications like schizophrenia and bipolar disorder.^5,6

Insomnia and obstructive sleep apnea (OSA) are the 2 most common sleep disorders, collectively affecting an estimated 2 billion individuals worldwide.^7,8 Comorbid insomnia and sleep apnea (COMISA) is also common, with 30%-40% of people with chronic insomnia also meeting the diagnostic criteria for OSA.⁹ OSA and insomnia share several symptoms, including frequent awakenings, difficulty maintaining sleep, nonrestorative sleep, mood disturbances, and fatigue, which can complicate their diagnosis. In the United States, 17% of women and 15% of men report taking medication for insomnia.¹⁰ While cognitive behavioral therapy for insomnia (CBTi) is considered the gold standard and most efficacious treatment, “less than 1% of adults in Australia with clinical insomnia symptoms receive CBTi treatment which includes behavioral components,”¹¹ and another study indicated that 90% of insomnia treatment delivered by primary care clinicians was pharmacological.¹² Thus, many people who report sleep disturbances in primary care settings are prescribed a hypnotic, including low-dose quetiapine. This is particularly concerning as >80% of individuals with OSA are estimated to remain undiagnosed and untreated.¹³ Therefore, we anticipate that a significant number of people with undiagnosed OSA are being prescribed quetiapine. This raises important clinical questions, as its effects on people with OSA remain largely unknown.¹⁴

OSA is a heterogeneous disorder with distinct pathophysiological causes or endotypes.¹⁵ Most people with OSA have a narrow or collapsible upper airway due to factors such as excess adipose tissue, craniofacial abnormalities, muscle hypertrophy, and orthostatic fluid shifts during sleep.¹⁶ A narrow upper airway does not always result in OSA, as compensatory pharyngeal dilator muscle activity can maintain airway patency during sleep.¹⁷ Of people with OSA, 70% have one or more “nonanatomical” endotypes that contribute to their OSA pathophysiology, each of which is a potential target for pharmacotherapy.^17,18 One-third of people with OSA have a low respiratory arousal threshold endotype. Waking up too easily to minor airway narrowing events (a low arousal threshold) can fragment sleep and preclude deeper stages of sleep that are associated with more stable breathing.^15,17,19,20 Hypnotic agents such as trazodone and eszopiclone can increase the arousal threshold and reduce OSA severity.^21-23 However, while most common hypnotics promote sleep without worsening OSA, concern remains that hypnotics may worsen hypoxemia in certain patients (ie, those with a high arousal threshold or an obesity hypoventilation phenotype).²⁴ Another third of people with OSA have high loop gain, which is characterized by exaggerated responses to small fluctuations in CO₂ and unstable breathing during sleep.^15,17,19,20 In addition, almost a third of people with OSA have poor upper airway dilator muscle responsiveness during sleep.^15,17,19,20

Quetiapine may have differential effects on the key pathophysiological causes of OSA. As a histamine antagonist, quetiapine may increase the arousal threshold and help stabilize breathing in people with a low arousal threshold endotype.^14,17,24 As a dopamine antagonist, quetiapine may also reduce the sensitivity to CO₂, and thus help to reduce high loop gain.^14,17 The effect of quetiapine on the upper airway muscle endotype is unclear. As a muscarinic acetylcholine antagonist and noradrenaline transporter blocker,¹⁴ quetiapine and its active metabolite, norquetiapine, might be beneficial. Indeed, recent studies indicate that noradrenergic agents alone or combined with antimuscarinic agents reduce OSA severity.^18,25 Conversely, as an adrenergic antagonist, it may be detrimental.^14,26

More than 75% of quetiapine prescriptions are below a therapeutic dose, with the 50-mg dose accounting for 24% of all prescriptions.^5,6,14 This prescribing pattern aligns with findings reported in other countries that indicate a global trend in the off-label use of quetiapine, frequently for insomnia.^1,27-30

The effects of quetiapine on breathing during sleep, hypoxic load, sleep architecture, next-day performance, and safety in people with comorbid OSA and insomnia are unclear.¹⁴ Furthermore, there are concerns about the safety of quetiapine in special populations. For example, the risks of all-cause mortality, falls, and dementia in adults aged >65 years are all higher in people who use low-dose quetiapine versus trazodone.³⁰ Accordingly, this study aimed to determine the effects of quetiapine on sleep, breathing, and next-day performance in the clinically relevant group of people with OSA who also have difficulty maintaining sleep.

Methods

Participants

Sixteen adults aged 18 years or older, with OSA (apnea/hypopnea index [AHI] ≥5 events/hour), confirmed via in-laboratory polysomnography within the past 2 years or in-home sleep cardiorespiratory monitoring (Resmed ApneaLink) using the 3% desaturation criteria, were recruited. Participants also had moderate or high difficulty maintaining sleep, corresponding to question 2 of the Insomnia Severity Index (ISI). All participants provided informed written consent, and the study was approved by the Flinders University Human Research Ethics Committee (4965). The study was conducted at the Nick Antic Sleep Laboratory, Adelaide Institute for Sleep Health, Flinders University, and was preregistered on ClinicalTrials.gov (NCT05303935).

Participants were excluded if their body mass index was <18.5 kg/m² or >40.0 kg/m², or if their nadir oxygen saturation fell below 75% during qualifying sleep studies. Additional exclusion criteria included use of concomitant medications or herbal supplements likely to interact with quetiapine; a history of adverse reactions to antipsychotics; use of drugs or presence of medical conditions (other than OSA or insomnia) known to affect breathing, sleep, arousal, or muscle physiology; psychiatric comorbidities; or any condition deemed by the investigative team to pose a safety risk. Participants were instructed to avoid alcohol, caffeine, and cigarettes for several hours before and during the study nights.

Protocol and study design

Two in-laboratory overnight sleep studies were performed with approximately 1 week washout between visits according to a double-blind, randomized, placebo-controlled, cross-over design. On one occasion participants received placebo, the other 50 mg of quetiapine. Participants were given quetiapine or placebo just prior to lights out followed by an 8-hour sleep opportunity. All investigators, participants, and individuals involved in scoring the studies were blinded to the treatment allocation (permuted block randomization of treatment order was managed by the independent external study pharmacist, Optima Ovest). Unblinding occurred after data collection was completed and all sleep studies had been scored.

Key measurements

Polysomnographic setup

Polysomnographic measurements as recommended by the American Academy of Sleep Medicine version 2.6 with 10 electroencephalographic [EEG] leads were performed³¹ using Grael 4K Amplifier (Compumedics Ltd., Abbotsford, VIC, Australia).

Questionnaires

In the evening of the first sleep study, the ISI and the Epworth Sleepiness Scale were completed by participants prior to receiving the allocated study intervention. The Karolinska Sleepiness Scale (KSS)³² was administered after the intervention was given in the evening, and after waking the following morning.

Next-morning performance measures

Driving simulation performance

Within 30 minutes of waking up, participants completed a 30-minute AusEd driving simulator task under low light conditions of approximately 5 lux. The AusEd driving simulator replicates a “monotonous night-time drive on a rural road.”³³ The key driving performance measures were steering deviation, measured as the average deviation in centimeters from driver’s median lane position, and crashes, defined as vehicle veered completely off the road, collision with a slow-moving truck, or remained stationary for >3 seconds. All crash events were reviewed using participant video and EEG records to confirm validity of the events.

Psychomotor vigilance task

Immediately following the driving simulator task, a 10-minute psychomotor vigilance task (PVT) was performed (Script Psychomotor Vigilance Test, 2019; Inquisit Lab, Millisecond Software, version 6.3.5). Thus, driving simulator performance and PVT were performed approximately 8.5 to 9.5 hours after dose administration. The main PVT measures were reaction time (milliseconds) and lapses defined as reaction time >500 milliseconds.

Outcome measures

The primary outcome was change in AHI between quetiapine night versus placebo night. Based on a sample size of n = 15 and a SD of 10 events/hour for between-night change in AHI, we estimated that we could detect an 8 event/hour change in AHI between conditions with >80% power (α = 0.05).

Secondary outcomes included changes in measures of hypoxemia and other polysomnographic parameters such as sleep efficiency, arousal index and sleep architecture, and next-day subjective (KSS) and objective alertness (PVT and AusEd driving simulator metrics) between conditions. OSA endotypes¹⁵ were estimated using a validated quantitative algorithm applied to non–rapid eye movement (REM) sleep polysomnography data.^18,20,34-36 As described in prior studies,^18,20,34-36 this approach was implemented in MATLAB software (MathWorks, Natick, MA, USA). In addition, people with a low arousal threshold were identified using a simple validated algorithm from 3 polysomnography input variables (AHI, nadir oxygen saturation, and the fraction of apneas to hypopneas) collected during the placebo night.³⁷

Statistical methods

Continuous variables were summarized by mean and standard deviation (SD) or median and interquartile range (IQR), depending on the distribution as appropriate. Mean differences between placebo and quetiapine for all study outcomes were explored using linear mixed model analyses, with restricted maximum likelihood estimation and random effect for participants within sequence. Number of PVT lapses and driving simulator crashes were analyzed using mixed-effects negative binomial regression, including period and sequence as fixed effects, and a random intercept for each participant within sequence. A P value <0.05 was considered statistically significant. Analyses were conducted using IBM SPSS version 29 and Stata SE version 18.0.

Results

Participants

Participants were recruited from our research volunteer database, through distributing flyers at our institutions, posting announcements on social media, and engaging in local community outreach. A total of 42 individuals underwent screening for concomitant medications and medical history, and 24 ApneaLink devices were sent to potential participants for screening. Following initial evaluations, 16 eligible participants provided consent to join the study. Of these, 15 participants (7 women) completed both study nights and next-morning performance testing (Figure 1). One participant withdrew on the first night. Participant characteristics are displayed in Table 1. All participants had moderate to severe symptoms of insomnia with an ISI range of 9 to 20 (Table 1). On average, participants had moderate to severe OSA (Figure 2 and Table 2).

Figure 1.

Flow diagram of the randomized crossover clinical trial design used in this study. Forty-two participants were screened for eligibility and 24 were sent ApneaLink devices. Sixteen participants consented (n = 7 female) and were randomized to placebo (n = 8) or quetiapine 50 mg (n = 8), with one participant withdrawing from the quetiapine arm due to an adverse event. The first study period included overnight polysomnography, vital sign monitoring, and questionnaires, followed by morning assessments including vital signs, driving simulation, psychomotor vigilance testing, and questionnaires. After approximately one week washout, participants crossed over to placebo (n = 7) or quetiapine 50 mg (n = 8), repeating overnight polysomnography and morning outcome assessments.

Table 1.

Participant characteristics.

Characteristic	Value
Sex
Female	7
Male	8
Age, y	61 ± 10
Body mass index, kg/m2	28 ± 4
Neck circumference, cm	38 ± 3
Waist circumference, cm	103 ± 10
Epworth Sleepiness Scale (0- to 24-point scale)	10 ± 5
Insomnia Severity Index (0- to 28-point scale)	15 ± 4
Mallampati Scale
Class I: soft palate, uvula, tonsillar pillars visible	1
Class II: soft palate and most of the uvula visible	4
Class III: soft palate and the base of the uvula visible	8
Class IV: soft palate not visible	1
Unknown	1

Data are presented as mean ± standard deviation or No. of participants.

Figure 2.

Apnea/hypopnea index (AHI). Panel A shows Total AHI for individual participants, with paired lines connecting placebo to quetiapine conditions. Mean ± SD values were 27 ± 16 for placebo and 20 ± 12 events/hour for quetiapine with a statistically significant reduction with quetiapine (P < .01). Panel B shows Supine AHI with the same paired participant trajectories. Mean ± SD values were 48 ± 27 for placebo and 33 ± 20 events/hour for quetiapine with a significant reduction with quetiapine (P = .001). Side bars show mean ± SD. Each uniquely colored line and shape represents the same participant across all figures.

Table 2.

Effects of quetiapine on sleep parameters.

Parameter	Placebo	Quetiapine	P value
Apnea/hypopnea index	27 ± 16	20 ± 12	.009
Supine (events/h sleep)	48 ± 27	33 ± 20	.001
Non-REM (events/h sleep)	25.2 ± 16.9	18.2 ± 11.9	.017
REM (events/h sleep)	30.7 ± 21.8	27.8 ± 15.2	.49
Central sleep apnea (events/h sleep)	0.6 ± 1.4	0.5 ± 0.7	.53
SpO2, average (%)	94.7 ± 1.2	94.5 ± 1.5	.38
Average minimum (%)	85.7 ± 5.3	86.7 ± 4.1	.29
ODI (3%, number/h sleep)	15 ± 10.8	13 ± 9.9	.30
Wake after sleep onset (min)	77.5 ± 54.5	43.5 ± 43.0	<.001
Total sleep time (min)	400	441	<.001
Arousal index (arousals/h sleep)	32 ± 16	25 ± 9	.012
Sleep latency (min)	19.4 ± 21.7	13.9 ± 13	.26
Sleep stage
N1 (%)	15.4 ± 8.7	12.7 ± 7.2	.09
N2 (%)	31.1 ± 6.6	32.8 ± 11.3	.55
N3 (%)	31.8 ± 12.4	33.4 ± 12.6	.50
REM (%)	21.7 ± 6.4	21.1 ± 4.1	.77
Periodic leg movements (number/h sleep)	5.9 ± 6.5	11 ± 10	.03
Heart rate during sleep (beats/min)	62.6 ± 11.4	63.3 ± 12.1	.5

Data are presented as mean ± standard deviation unless otherwise stated. Bold values indicate statistically significant differences between groups.

Abbreviations: ODI, oxygen desaturation index (using 3% criteria); REM, rapid eye movement; SpO_2, estimated arterial blood oxygen concentration via pulse oximetry.

Effects of quetiapine on sleep and breathing

Quetiapine reduced the total AHI versus placebo (mean ± SD) by 24 ± 64% (Figure 2A). Quetiapine reduced the supine AHI by 31 ± 112% (Figure 2B). Reductions in AHI with quetiapine occurred during non-REM sleep (Table 2). Quetiapine decreased the arousal index (Figure 3A), and increased sleep efficiency versus placebo (Figure 3B). Quetiapine also increased total sleep time (Figure 3C) and reduced the amount of wake after sleep onset (Table 2). There was no difference in measures of oxygen saturation between study nights, nor sleep architecture (Table 2). The frequency of periodic leg movements increased with quetiapine (Table 2).

Figure 3.

Panel A = Total arousal index which decreased from 32 ± 16 during placebo to 25 ± 9 events/h sleep with quetiapine (P = .01). Panel B = Sleep efficiency (% time asleep while in bed) which increased from 80 ± 11 during placebo to 87 ± 9 % with quetiapine (P < .001). Panel C = Total sleep time which increased from 400 ± 51 to 440 ± 44 minutes (P < .001). Side bars show mean ± SD. Each uniquely colored line and shape represents the same participant across all figures.

OSA endotypes

All but one of the study participants was estimated to have a low arousal threshold endotype. Estimates of OSA endotypes were not systematically different between conditions (Table 3).

Table 3.

Obstructive sleep apnea endotypes.

Endotype	Placebo	Quetiapine	P value
LG1 (1 cycle/min frequency)	0.55 ± 0.17	0.53 ± 0.2	.57
LGn	0.41 (0.32-0.5)	0.43 (0.35-0.51)	.88
VRA (%eupnea)	30 ± 15	31 ± 20	.78
Arousal threshold (%eupnea)	112 (108-118)	118 (108-123)	.27
Vpassive (%eupnea)	94 (94-96)	95 (94-96)	.3
Vactive (%eupnea)	100 (94-104)	101 (94-103)	.53
Vcomp (%eupnea)	3.5 (1.6-9.7)	5.4 (−0.9 to 7.1)	.37
Vpassive1d (%eupnea)	58 (51-65)	62 (50-71)	.06

Data are presented as mean ± standard deviation or median (interquartile range: quartile 1 to quartile 3) as appropriate. Loop gain (LG) is a calculated estimate of the ventilatory drive response to a disturbance in ventilation. LG1 represents the ventilatory response to a 1-cycle-per-minute reduction in ventilation, while LGn accounts for the effects of circulatory delay. A lower loop gain indicates a more stable respiratory control system with reduced sensitivity to disturbances. In contrast, a higher loop gain reflects greater system sensitivity, which can contribute to respiratory instability and periodic breathing patterns. Ventilatory response to arousal (VRA) quantifies the degree of ventilatory overshoot during and after a transient cortical arousal from sleep. A higher VRA value indicates greater ventilatory overshoot, increasing the likelihood of subsequent respiratory instability. Arousal threshold reflects the degree of respiratory disturbance needed to trigger awakening from sleep (arousal). A higher value indicates greater tolerance of reduced ventilation before arousal occurs. Vpassive represents the estimated average ventilation during sleep at eupneic drive when the pharyngeal muscles are minimally active. A higher value indicates a less collapsible upper airway. Vactive represents the estimated ventilation under conditions of maximum ventilatory drive. Higher values indicate more muscle recruitment. Vcomp quantifies pharyngeal muscle compensation as the difference between Vactive and Vpassive (Vactive − Vpassive), reflecting the ventilation increase driven by muscle effort. A higher value indicates greater compensatory recruitment to maintain respiratory airflow. Vpassive1d, sometimes also referred to as Vmin,⁶⁰ estimates airway collapsibility overnight at the lowest decile of respiratory drive.

Next-day performance

There were no statistically significant differences in next-day performance measures based on treatment sequence, indicating that any potential effect of learned familiarity with the PVT and driving simulator tests was minimal.

Quetiapine impaired several measures of next-day alertness and performance versus placebo. PVT median reaction time was longer during the morning after taking quetiapine compared to placebo (Figure 4). PVT lapses increased from 2 (IQR, 0-6) on placebo to 10 (IQR, 2-19) following quetiapine. The incidence rate ratio (IRR) was more than thrice that of placebo (IRR, 3.22 [95% CI, 1.92-5.42]; P < .001).

Figure 4.

Reaction time from psychomotor vigilance task (PVT), shown as median for each individual in milliseconds. Reaction time increased from 336 ± 48 during placebo to 382 ± 84 ms with quetiapine (P < .01). Side bars show mean ± SD. Each uniquely colored line and shape represents the same participant across all figures.

Steering deviation from the median lane position was significantly greater after quetiapine versus placebo (Figure 5). The total number of driving simulator crashes following quetiapine was higher compared to placebo (55 vs 27), and the IRR was almost twice that of placebo (IRR, 1.95 [95% CI, 0.87-4.37]) but was not statistically significant (P = .105).

Figure 5.

Deviation from median lane position in AusEd driving simulator, median for each individual in centimeters. Steering deviation increased from 72 ± 38 during placebo to 96 ± 53 cm with quetiapine (P < .01). Side bars show mean ± SD. Steering deviation of healthy controls in previous studies using the AusEd driving simulator was 37 ± 9 cm.^46-48 Each uniquely colored line and shape represents the same participant across all figures.

All but 2 individuals were either subjectively sleepier or did not experience any change in their morning KSS values following quetiapine versus placebo night (Figure 6). The 2 individuals who subjectively felt more alert on the quetiapine arm still performed worse in the objectively measured PVT and AusEd driving simulation task.

Figure 6.

Karolinska Sleepiness Scale (KSS). Subjective sleepiness increased from 5 ± 2 during placebo to 7 ± 2 with quetiapine (P < .001). Side bars show mean ± SD. Each uniquely colored line and shape represents the same participant across all figures.

Adverse events

No serious adverse events were observed during the study. One or more mild to moderate adverse events related to quetiapine were reported by or observed in 5 participants; all were resolved within 48 hours. Two participants required medical evaluation for their adverse events (Table 4).

Table 4.

Adverse events.

Placebo (n = 15)

Quetiapine 50 mg (n = 16)

• Nausea (moderate) • Increased limb pain • Tingling on the face

• Claustrophobia (unknown), anxiety (common), and panic attack (unknown) • Excessive thirst overnight (uncommon), complained of restless legs (common), inappropriate disinhibited behavior (unknown) • Fall (unknown, US Food and Drug Administration boxed warning) (no injury, blood pressure dropped to 62/41 mm Hg, unresponsive for >10 seconds, required intervention and medical evaluation) • Orthostatic hypotension (common) (n = 2) • Difficult to wake after ≥8-hour sleep opportunity, groggy for several hours, next-day hypersomnolence (very common)

Each bullet point represents one participant, except where a number is indicated (rarity in parentheses where known).⁵ All adverse events were self-limiting and resolved by the following day, apart from one participant who experienced hypersomnolence lasting 2 days. Participants who experienced orthostatic hypotension were managed with oral fluids, and blood pressure was reassessed to confirm improvement prior to discharge from the sleep laboratory.

Discussion

Although low-dose quetiapine is frequently prescribed off-label, previous studies have not assessed its effects on sleep, breathing, or next-day performance in people with COMISA.¹⁴ The primary findings of the current study indicate that a 50-mg dose of quetiapine, taken approximately 30 minutes before sleep, reduces OSA severity as measured via the AHI and improves sleep efficiency without affecting sleep architecture or oxygen saturation in people with mild to moderately severe OSA and insomnia symptoms. While most common hypnotics do not change the AHI,^24,38 eszopiclone and trazodone have yielded similarly modest reductions in AHI in some,^21,23 but not all, previous studies.²² Consistent with the current findings, reductions in AHI with eszopiclone consistently occurred in people with a low arousal threshold endotype, although whether these participants with OSA also had difficulty maintaining sleep is unclear.²¹ Since quetiapine has the potential to modulate several brain receptors involved in the sleep-wake cycle, muscle responsiveness, and breathing, it is challenging to determine which specific receptor targets contribute to this effect.¹⁴

Given the multiple potential conflicting mechanisms combined with the relatively modest sample size in which to explore secondary mechanistic outcomes using indirect estimates rather than direct physiological measures of OSA endotypes, we did not detect any significant differences in OSA endotypes. Nonetheless, based on previous findings of hypnotics in OSA and plausible mechanistic rationale,^14,21,24 a quetiapine-induced increase in the arousal threshold remains the most likely explanation for the reduction in OSA severity in the current participants, almost all of whom had low arousal thresholds. Indeed, lower dosages of quetiapine primarily act on histaminergic H₁ and alpha-1 and alpha-2 adrenergic receptors to drive sedative effects.^39-41 Thus, the current reduction in AHI with low-dose quetiapine is likely driven by its sedative properties via histaminergic pathways, although this requires further investigation in an appropriately designed study. Fifty milligrams of quetiapine is unlikely to significantly impact dopaminergic pathways in people with schizophrenia.^41-43 However, in the current study, 50 mg of quetiapine increased periodic leg movements, consistent with a dopaminergic effect.

The approximately 7% improvement in sleep efficiency with quetiapine is consistent with the magnitude change of other common hypnotics in people with OSA, including those who do not have low sleep efficiency at baseline.²⁴ Whether a hypnotic reduces the arousal index in people with OSA typically depends on whether the frequency of sleep disordered breathing also decreases²⁴ as was the case in the current study albeit of relatively modest magnitude of 7 events/hour for both indices. Conversely, the increase in total sleep time of >40 minutes and approximately 45% reduction in wake after sleep onset with quetiapine represent large clinically relevant sleep promotion effects. The magnitude of these changes may relate, at least in part, to the concomitant insomnia symptoms in the current study. Unlike other common hypnotics, the current findings also raise the possibility that quetiapine may increase periodic leg moments in these patients. This requires further investigation.

Although sleep and breathing improved on quetiapine nights, next-day performance and alertness were impaired. Driving simulators, which have been used in research for several decades, are validated tools that provide insight into on-road behavior and impairment.^44,45 Participants with OSA and symptoms of insomnia in the current study had steering deviation impairment approximately double that of healthy controls from previous studies^46-48 in the placebo condition. Quetiapine further exacerbated this impairment to almost triple the steering deviation of healthy controls.^46-48

Being able to drive is an important function that increases a person’s independence, contributes to their identity, and facilitates their social engagement and sense of well-being.⁴⁹ This must be balanced with societal interest in safe driving. Road injuries represent a significant global public health burden. Indeed, in 2017 alone, 1.2 million people died and 54 million were injured.⁵⁰ In 2022 in the United States, there were 10.5 million vehicle accidents, with 3 million injuries and 60 000 deaths.⁵¹ In a large Canadian study that compared motor vehicle accidents (MVAs) and medication use from health records, sedating antipsychotics increased risk for MVA (adjusted odds ratio, 1.35 [98.75% CI, 1.25-1.46]) in both new and established users.⁵² Medical conditions can also adversely impact driving ability. Insomnia increases the risk of an MVA by almost 2-fold in commercial truck drivers.⁵³ People with insomnia perform worse than good sleepers after 20 minutes of simulated driving.⁵⁴ OSA has a well-established increased risk for MVA, and certain individuals who are effectively treated with the gold standard continuous positive airway pressure may still be at increased risk.⁵⁵ Further, people in whom their OSA is treated but still have residual sleepiness have a 5-fold increased risk of having an MVA.⁵⁵ This impairment is also demonstrated in driving simulations.⁴⁷

Furthermore, accurate self-assessment of alertness is vital when deciding to engage in complex tasks such as driving. Performance and subjective sleepiness do not always match,⁵⁶ potentially leaving these people with the false impression of their abilities. While consistent with impaired driving performance and vigilance, most (11/15) participants in the current study reported that they felt subjectively sleepier as measured via the KSS on the morning following quetiapine versus placebo. This serves as an important reminder to never drive while sleepy. However, 2 participants did not report subjective awareness of increased sleepiness with quetiapine versus placebo. A further 2 people subjectively felt more alert following quetiapine despite objective worsening on the PVT and AusEd driving tasks. Thus, the current findings indicate that approximately one in 4 people may either not be aware or falsely believe they are less sleepy following 50 mg of quetiapine. This represents a potential safety concern for driving and other vigilance-dependent tasks. Accordingly, while further investigation in a larger sample is required, people with COMISA should be cautioned not to drive or perform other safety-dependent tasks for at least 9.5 hours following acute use of quetiapine.

People with COMISA often will come to their primary care providers and request pharmacological treatment for their sleep disturbance, which is often attributed to insomnia, particularly if they do not have the “typical” OSA risk factors. The current next-day impairment findings highlight the need for caution when considering prescribing low-dose quetiapine in this clinical population. However, given the large number of people with undiagnosed OSA, these patients may be difficult to identify within primary care, further reinforcing the need for appropriate sleep disorder screening.

Despite our careful screening and eligibility assessment for safety criteria, approximately one-third of participants from all age groups experienced some form of adverse event on the quetiapine night. Some of the events were known to be rare or uncommon.⁵ Although our study was not powered to evaluate adverse events, it raises the possibility that people with COMISA might be particularly vulnerable to adverse effects associated with quetiapine.

Recently, the safety of low-dose quetiapine for insomnia was examined in older adults and was associated with increased risk of mortality, dementia, and falls compared to prescribing another hypnotic, trazodone.³⁰ Long-term use of low-dose quetiapine is associated with adverse metabolic effects,^57,58 increased risk of major adverse cardiovascular events,⁵⁹ and weight gain,^57,58 which may increase OSA severity.

Methodological considerations

While the current study is the first to systematically investigate the effects of quetiapine on sleep and breathing and next-day function in people with OSA and insomnia symptoms using rigorous clinical trial design, there are several methodological considerations that need to be acknowledged. First, although participants spanned both sexes and a range of ages, OSA severity, and comorbidities, the study was powered for AHI change, limiting subgroup analyses and broader generalizability for other outcomes. The current study was limited to one night in a highly controlled laboratory environment in people who were not currently taking quetiapine. Thus, the current findings may not reflect longer-term use or the more nuanced situation in off-label quetiapine prescribing patterns. In this study, only the 50-mg dose strategy was investigated, whereas quetiapine dosing for insomnia may vary from 25 mg to 100 mg. Next-day performance and side effects may differ with acclimatization to quetiapine. However, previous real-world analyses suggest that the next-day performance may not improve in established users of sedating antipsychotics.⁵² As highlighted, future mechanistic studies using gold standard physiological methodology and varying quetiapine dosing strategies are also required to investigate the role of quetiapine on the arousal threshold and other OSA endotypes to better understand the potential role of quetiapine on breathing stability in people with OSA.

Conclusions

Consistent with its hypnotic effect, a single low dose of quetiapine reduced OSA severity as measured via the AHI and increased sleep efficiency without worsening overnight hypoxemia. However, there was evidence of next-day impairment in vigilance and driving simulator performance. Accordingly, larger, long-term studies are needed to evaluate tolerance, cumulative sedation, real-world dosing (25-100 mg), and potential risk benefit profile across different age groups, sexes, and relevant comorbidities before any clinical application of quetiapine to individuals who present to care providers with complaints of poor sleep.

Author contributions

C.F., T.B., R.A., K.A.L., and D.J.E. developed the study concepts and aims. C.F., B.T., K.A.L., A.V., and D.J.E. performed data extraction and data analysis. C.F., G.R., K.A.L., R.A., T.B., G.W., and D.J.E. were responsible for participant recruitment and screening. All authors provided important insight on data interpretation and drafting of the manuscript and approved the final version of the manuscript.

Supplementary material

Supplementary material is available at Annals of the American Thoracic Society online.

Conflicts of interest

Please see the ICMJE disclosure forms, which have been provided as supplementary material. Outside the current work, D.J.E. reports research grants from Bayer, Apnimed, Eli Lilly, Invicta Medical (now Restera), Restora Medical, Withings, and Takeda and serves on scientific advisory boards or as a consultant for Restera, Restora Medical, Mosanna, SleepRes, Takeda and Apnimed.

Funding

This study was funded by a National Health and Medical Research Council of Australia Investigator Grant (1196261).

Data availability

De-identified individual data for key study outcomes are provided in the figures. Other de-identified data that support the findings of this study, are available from the corresponding author upon request and subject to ethical and data custodian approval.