Effects of Tiagabine on Slow Wave Sleep and Arousal Threshold in Patients With Obstructive Sleep Apnea

Abstract

Introduction:

Obstructive sleep apnea (OSA) severity is markedly reduced during slow-wave sleep (SWS) even in patients with a severe disease. The reason for this improvement is uncertain but likely relates to non-anatomical factors (i.e. reduced arousability, chemosensitivity, and increased dilator muscle activity). The anticonvulsant tiagabine produces a dose-dependent increase in SWS in subjects without OSA. This study aimed to test the hypothesis that tiagabine would reduce OSA severity by raising the overall arousal threshold during sleep.

Aims and Methods:

After a baseline physiology night to assess patients’ OSA phenotypic traits, a placebo-controlled, double-blind, crossover trial of tiagabine 12 mg administered before sleep was performed in 14 OSA patients. Under each condition, we assessed the effects on sleep and OSA severity using standard clinical polysomnography.

Results:

Tiagabine increased slow-wave activity (SWA) of the electroencephalogram (1–4 Hz) compared to placebo (1.8 [0.4] vs. 2.0 [0.5] LogμV², p = .04) but did not reduce OSA severity (apnea–hypopnea index [AHI] 41.5 [20.3] vs. 39.1 [16.5], p > .5). SWS duration (25 [20] vs. 26 [43] mins, p > .5) and arousal threshold (−26.5 [5.0] vs. −27.6 [5.1] cmH₂O, p = .26) were also unchanged between nights.

Conclusions:

Tiagabine modified sleep microstructure (increase in SWA) but did not change the duration of SWS, OSA severity, or arousal threshold in this group of OSA patients. Based on these findings, tiagabine should not be considered as a therapeutic option for OSA treatment.

Statement of Significance

Previous studies showed that hypnotics in obstructive sleep apnea (OSA) may stabilize breathing and improve OSA severity or, on the contrary, may worsen upper airway obstruction and oxygen desaturation because of myorelaxation and respiratory depression. The pharmacologic induction of slow-wave sleep (SWS) through specific hypnotics such as tiagabine may be an ideal mechanism for reducing OSA severity, as OSA often improves during SWS. Compared to placebo, tiagabine increased delta activity (1–4 Hz) of the electroencephalogram (EEG) during sleep but not the time spent in non-rapid eye movement stage 3, and the drug did not show an effect on apnea–hypopnea index. Despite tiagabine increasing EEG slow-wave activity, it does not appear to be useful for OSA treatment.

INTRODUCTION

Several studies have shown that obstructive sleep apnea (OSA) improves during slow-wave sleep (SWS), sometimes dramatically and even in patients with otherwise severe OSA.^1,2 Ratnavadivel et al.¹ found that 82% of patients with moderate to severe OSA achieve an apnea–hypopnea index(AHI) < 15 episodes/hr in SWS. The reason for improvement likely relates to changes in the non-anatomical factors that contribute to OSA during SWS such as reduced arousability,³ thereby enabling a higher activation of upper airway dilator muscles^4–6 yielding reduced pharyngeal collapsibility.⁷

Previous research^8–11 have established the importance of arousals in the pathophysiology of OSA, namely that a low respiratory arousal threshold can limit neuromuscular compensation of the upper airway and contribute to the development of sleep-related hypopneas and apneas in many individuals.¹² Thus, raising the arousal threshold could be a potential pharmacological treatment approach for OSA, particularly if it can be done by inducing SWS and leaving pharyngeal compensatory mechanisms intact.

Pharmacologic induction of SWS through specific hypnotics¹³ may be an ideal mechanism for raising the arousal threshold to treat OSA, particularly since SWS seems to be a “protective state” against OSA.^1–3 Historically, hypnotic use in OSA has been discouraged because of concern that delayed arousal would produce further blood gas derangement and that myorelaxation and respiratory depression would worsen obstruction. However, as underlined in recent reviews,^12,14 not all studies show worsening of OSA with hypnotics¹⁵ and some show improvement, even in individuals with moderate and severe OSA.^16–18 Moreover, recent studies indicate that certain hypnotics can actually increase genioglossus responsiveness to respiratory stimuli.^19–21 Therefore, while high doses of certain hypnotics can worsen oxygenation in some severe obese OSA patients, the belief that all hypnotics are contraindicated in all patients is not supported by the literature.

Tiagabine is a γ-aminobutyric acid (GABA) reuptake receptor (GAT-1) inhibitor that allows increased GABA concentration at the synaptic level of the central nervous system. GABA is the principal inhibitory neurotransmitter in the mammalian nervous system. GABAergic inhibitory mechanisms are crucial for the initiation and maintenance of sleep and for the generation of slow-wave activity (SWA, 1–4 Hz).²² Tiagabine has been shown to produce a dose-dependent increase in SWS in older adults²³ and in patients with primary insomnia.^24–26 Doses as low as 10 mg increase the power density of slow EEG oscillations and the percentage of SWS from 15% to 24%.²⁷ Pharmacologic SWS enhancement with tiagabine reduced several aspects of the behavioral, psychological, and physiological impact of sleep restriction in a study by Walsh et al.²⁸ but did not improve memory consolidation in a subsequent study from Feld et al.²⁷ These contrasting data raise the question of whether increased SWS with tiagabine is simply an EEG epiphenomenon or is capable of producing a physiologically meaningful change in sleep state.¹³ If the latter is the case, tiagabine-related increase in SWS could lead to a concomitant increase in the arousal threshold and an improvement in OSA severity in certain patients (i.e. those with a low arousal threshold).

Accordingly, we tested the hypothesis that administration of tiagabine in patients with OSA would increase the amount of SWS and increase the arousal threshold, thereby reducing OSA severity. We anticipated this drug to be preferentially effective in patients with a low arousal threshold. The primary outcome was the AHI (events/hour of sleep). Secondary outcomes were (1) the time spent in SWS (non-rapid eye movement stage 3, NREM3), (2) the SWA from EEG spectral power, and (3) the arousal threshold as measured by the last esophageal pressure swing before cortical arousal from sleep.

METHODS

Participants

Patients with OSA (AHI > 10 events/h sleep) aged 18–75 years were included in the study. Patients were consecutively recruited from our existing database of volunteers as well as from the community. Individuals were excluded if they were taking medications known to influence breathing or sleep–wake physiology. Subjects with a history of epilepsy were also excluded. The protocol was approved by the Partners Institutional Review Board at Brigham and Women’s Hospital. All subjects provided written informed consent prior to enrolment.

Protocol

The protocol consisted of 3 nights: a single physiology night assessing patients’ OSA phenotypic traits (described below) and 2 overnight sleep studies performed 1 week apart: a placebo night and a tiagabine (12 mg PO) night, in a double-blind, randomized, placebo-controlled design (clinical trial NCT02387710). Randomization was performed by the investigational pharmacy; all data analyses and data exclusions were performed before unblinding. For each night, the subjects arrived at the sleep laboratory at approximately 8:00 pm. Given that the time from administration of tiagabine to peak plasma concentration is 45 min due to its rapid absorption,²⁹ placebo or drug was administered approximately 30 min before lights out. Once the patient had been set up for overnight monitoring, data were collected as described below.

Measurements and Equipment

Anthropometric data were collected on all the study nights. Neck circumference was measured at the superior border of the cricothyroid cartilage and waist circumference was measured at the highest point of the iliac crest.

Baseline Phenotypic Traits Using Continuous Positive Airway Pressure Manipulation

Before the clinical polysomnography (PSG) nights, baseline phenotypic traits were measured for each patient following the phenotyping technique previously described.³⁰ Detailed methods and results of the physiology night can be found in the online data supplement.

Clinical PSG Nights

Patients were instrumented with a clinical montage for PSG in addition to esophageal pressure (Pes) measurements to assess the impact of tiagabine on the arousal threshold. Pes was determined with a thin, flexible pressure-tipped catheter (Millar Instruments, Houston, TX) that was inserted through a decongested (oxymetazoline-HCL) and anesthetized (4% lidocaine) nostril until the tip of the catheter was located 1–2 cm caudal to the base of the tongue. After visual inspection through the mouth, the patient was asked to repeatedly swallow until the catheter reached the lower third of the esophagus.

In order to minimize the effect of body position on AHI, we asked patients to sleep on their back for as much as possible. The AHI value reported all refer to sleep in the supine position.

The EEG was recorded continuously from electrodes placed according to the 10–20 System, referenced to 2 coupled electrodes attached to the mastoids. EEG signals were sampled at a rate of 250 Hz. On each night, total sleep time (TST) and time spent in the different sleep stages (wake; NREM sleep stages 1, 2, 3, and REM sleep) was calculated in minutes and as a percentage of TST.

We also aimed to estimate the upper airway collapsibility from peak flow on and off drug using a technique recently developed in our lab³¹ in a subset of patients (N = 7) in which nasal flow was assessed on both tiagabine and placebo nights using a sealed nasal mask and pneumotachometer instead of a nasal cannula.

Data Analysis

OSA Severity

Apneas, hypopneas, arousals, and sleep stages were scored using standard American Academy of Sleep Medicine guidelines³² by a technician blinded to the study randomization.

Arousal Threshold

The arousal threshold was quantified as the mean of all the nadir negative esophageal pressure swings immediately preceding an arousal at the end of an obstructive apnea or hypopnea during both placebo and tiagabine nights. Mean arousal thresholds for all sleep stages was also calculated and compared between nights.

EEG Spectral Power

Average power spectrum density was calculated from C4 for all NREM and REM sleep epochs. Power spectra were calculated by the fast Fourier Transformations with a Hanning window (4 s). In the averaged spectra, median power was determined for the delta wave bands (1–4 Hz). Windows occurring within arousals were excluded from analysis.

Peak Flow

In the subset of patients in whom nasal flow was recorded with a pneumotach, the median peak flow was measured with an automated analysis from all NREM supine sleep epochs excluding arousals. This technique was recently validated by our group as an alternative measurement of active upper airway collapsibility.³¹

Statistical Analysis

Variables were compared using a Wilcoxon matched-pairs signed-rank test, with p < .05 considered as statistically significant. Data are expressed as a median (interquartile range: 75th percentile minus 25th percentile). Statistical analyses were performed using Graph Pad Prism 6.0 (GraphPad Software, La Jolla, CA).

RESULTS

Participants

A total of 18 patients were enrolled in the study. Two patients did not tolerate the study setup and did not come back for the second study night. One patient was excluded because of predominantly central sleep apnea on both nights, and another was excluded because of low-quality EEG data. Therefore, 14 subjects were analyzed. The main characteristics of these patients are described in Table 1. Six subjects had a history of cigarette smoking, but no subject was an active smoker at the time of the study. No subjects had an history of chronic obstructive pulmonary disease or heart disease. No side effects related to tiagabine were described by the patients.

Table 1.

Characteristics of the Patients.

Female Gender, n (%)	6 (43)
Age, years	58 [12]
BMI, kg/m 2	31.4 [6.9]
Neck circumference, cm	42.0 [6.4]
Waist circumference, cm	107.5 [17.8]
Treated, n (%)
• CPAP	5 (36)
• Oral appliance	2 (14)
Comorbidities, n (%):
• Hypertension	4 (29)
• Hypercholesterolemia	2 (14)
• Osteoarthritis	1 (7)
• GERD	1 (7)

Data are presented as median [interquartile range] unless otherwise specified. BMI, body mass index; CPAP, continuous positive airway pressure; GERD, gastro-esophageal reflux disease. N = 14 patients with OSA.

Sleep Architecture

Effects of tiagabine on sleep architecture are shown in Table 2. There were no differences in sleep stages between placebo and tiagabine nights. In particular, NREM stage 3 duration was unchanged on tiagabine compared to placebo (p > .5). However, EEG spectral analysis revealed increased SWA (delta power, 1–4 Hz) and theta power (4–7 Hz) during sleep on the tiagabine night compared to the placebo night (Figure 1).

Table 2.

Sleep Architecture on Placebo and Tiagabine Nights.

	Placebo	Tiagabine	p value
TST, min	426 [55]	422 [61]	>.5
SE, %TIB	69.6 [14.9]	74.6 [15.3]	>.5
NREM1, %TST	33.6 [14.2]	26.2 [12.1]	>.5
NREM2, %TST	49.2 [15.7]	52.1 [11.6]	.46
NREM3, %TST	8.8 [10.6]	8.3 [16.0]	>.5
REM, %TST	12.7 [5.4]	6.7 [8.6]	.09
WASO, min	97.0 [50.7]	89.0 [44.4]	>.5

TIB, time in bed; TST, total sleep time; SE, sleep efficiency; NREM1, NREM2, and NREM3: non-rapid eye movement sleep stages 1, 2, and 3; REM, rapid eye movement sleep. Data are expressed as median [interquartile range]. N = 14.

Figure 1.

Group data of EEG power spectral density for the frequencies 1–32 Hz during all sleep stages (excluding arousals). Tiagabine did not have a specific effect on slow-wave activity but increased both delta (1–4 Hz) and theta (4–7 Hz) power.

OSA Severity and Arousal Threshold (Pes)

AHI, oxygen desaturation index (ODI), and arousal threshold (measured with the esophageal catheter from the clinical PSG nights) were not different on tiagabine compared to placebo (Table 3). Arousal threshold for each sleep stage are shown in Table 4. Individual patient data for AHI and overall arousal threshold changes are shown in Figure 2, A and B, respectively.

Table 3.

Obstructive Sleep Apnea Severity on Placebo and Tiagabine Nights.

	Placebo	Tiagabine	p value
AHI supine, events/h	41.5 [20.3]	39.5 [16.5]	>0.5
AHI NREM supine, events/h	40.2 [20.5]	41 [15.7]	>0.5
AHI REM supine, events/h	38.9 [42.0]	28.8 [23.7]	.08
Arousal index, events/h	44.1 [16.6]	46.7 [15.6]	>.5
ODI NREM supine events/h	19.1 [28.4]	24.5 [18.2]	>.5
Nadir SaO 2 , %	88.5 [5.8]	84.5 [2.8]	.24

AHI, apnea–hypopnea index; ArI, arousal index; ODI, oxygen desaturation index; SaO₂, oxygen hemoglobin saturation. Data are expressed as median [interquartile range]. N = 14 except during REM in which data during both conditions was available in 11 patients.

Table 4.

Arousal Threshold on Placebo and Tiagabine Nights for Each Sleep Stage.

Sleep stages	Arousal threshold, cmH2O		p value
	Placebo	Tiagabine
NREM 1	−24.5 [5.1]	−25.0 [8.8]	>.5
NREM 2	−26.6 [4.5]	−24.6 [5.8]	>.5
NREM 3	−42.3 [27.2]	−46.9 [51.0]	.25
REM	−21.7 [28.4]	−26.4 [18.2]	>.5

Data are expressed as median [interquartile range]. N = 14 except during NREM 3 and REM in which data during both conditions was available in 4 patients.

Figure 2.

Individual data showing changes in AHI (A) and arousal threshold (B) from placebo (P) to tiagabine (T) nights.

Associations with OSA Severity

Change in AHI was related to the change in NREM stage 3 (r = 0.60, p = .02) but not to the SWA variation (p > .5). A trend for a relationship between the reduction in AHI and the increase in arousal threshold (more negative Pes, r = 0.54, p = .06) was also found.

Peak Flow Analysis

In a subset of 7 patients, median peak flow (pneumotach) was calculated from all breaths during NREM supine sleep (arousal breaths excluded). Peak flow did not consistently change from placebo to tiagabine nights (0.31 [0.13] L/s vs. 0.28 [0.12] L/s, p = .47), suggesting that there was no significant change in airway collapsibility under active conditions. However, those with greater worsening of OSA severity on tiagabine (≥ 5 events/h, range +5 to +28 events/h, N = 3) had a higher peak flow on placebo (range: 0.33–0.51 L/s vs. 0.20–0.31 L/s , p = .04) and a greater reduction in peak flow on tiagabine (range: 0.17–0.26 L/s vs. 0.07–0.18 L/s, p = .01) versus the other patients (range −15 to +3 events/h, N = 4, see online supplement for further details). A possible explanation for this finding is that tiagabine might affect upper airway dilator muscle contractility or reduce respiratory drive in the patients who had higher peak flow on placebo.

DISCUSSION

The main findings of this study are that tiagabine increases SWA in the EEG power spectrum but does not increase either the amount of NREM stage 3 sleep or the arousal threshold compared to placebo in patients with OSA. Tiagabine does not change OSA severity and the responses observed are within the range of night-to-night variability.

Consistent with previous literature,²⁷ tiagabine significantly increased SWA in the EEG power spectrum by a median value of 16.4 [27.7]% compared to placebo. However, this was not sufficient to increase the amount of NREM stage 3, in contrast with previous studies performed in patients with primary insomnia or healthy controls.^23–25,28 A possible explanation is that the presence of sleep-disordered breathing prevented sleep consolidation in this group compared to normal controls, despite an increased tendency towards deep sleep shown by increased SWA.

The increase of SWA with tiagabine was not accompanied by an increase in arousal threshold or with a reduction in AHI as we had hypothesized. These data were confirmed also when arousal threshold was analyzed for each sleep stage: the highest increase in SWA was seen during NREM stage 2 on tiagabine (94 [68] μV² on placebo vs. 105 [139] μV² on tiagabine), and this was not followed by an increased arousal threshold in the same stage (Table 4). This fact suggests that the increase in EEG delta activity in this group may be due to EEG slowing, i.e., an electric epiphenomenon,³³ rather than to a physiologic increase in SWS. As shown in Figure 1, tiagabine increased both delta and theta power spectral densities, and it is possible that this drug caused a nonspecific increase in all frequencies relevant to NREM sleep (1–7 Hz) rather than a selective increase in SWA. Interestingly, recent findings support the notion that arousability is not a function of power in any specific frequency but is related to the relative powers in the different frequency ranges.³⁴ This theory would help to explain why the increase in SWA was not accompanied by an increase in the arousal threshold.

In a previous study, Sériés et al.³⁵ were able to increase SWS by 17% with γ-hydroxybutyrate (derived from GABA) at the (relatively high) dose of 60 mg/kg in patients with OSA, but the effects on AHI were small and inconsistent. However, during the treatment night there was a significant increase in central apneas and hypopneas compared to the control night. This change in the pattern of respiratory events may be related to central ventilatory depression mediated by γ-hydroxybutyrate as it has been shown that high doses of GABA can interfere with respiratory rhythm generation at the pre-Bötzinger complex,³⁶ resulting in an intermittent fall in ventilatory drive with the development of central apneas. Tiagabine increases the availability of GABA at the synaptic level of the central nervous system; however, we did not see any evidence for a systematic reduction in mean peak flow or an increase in the number of central apneas at the dose of tiagabine delivered in the current study.

Even though the small AHI changes observed in the current study were most likely due to spontaneous variability, at least 2 patients had a clear worsening of OSA on tiagabine compared to placebo, changing from a mild to a severe degree of OSA for the same body position. Both these patients were characterized by high activeV₀ during the phenotype assessment. ActiveV₀ reflects the ventilation during sleep when the upper airway muscles are maximally activated, and it is a function of baseline anatomy and of the neuromuscular compensation to increased upper airway resistance. These subjects had mild/moderate anatomic defect (passiveV₀), good muscle compensation (activeV₀ minus passiveV₀) on the phenotyping night, and they could reach stable breathing if the upper airway muscles were maximally activated during normal sleep. On placebo, they had mild OSA that worsened during REM sleep, consistent with dilator muscle dependence. On tiagabine, OSA became severe, suggesting that in these subjects the increase in AHI might be mediated by decreased contractility or by respiratory depression associated with this sedative. The peak flow analysis that was performed on a subset of patients in the study also supported this notion: while tiagabine did not change the peak flow during sleep as a group, those who experienced the highest increase in AHI on tiagabine were those with a higher peak flow on placebo and a higher reduction in peak flow on tiagabine compared to the others (for further analysis see the online data supplement). While further investigation is needed, these findings could help to identify a specific phenotype of OSA patient, characterized by high activeV₀, in which the administration of a hypnotic with a similar pharmacologic profile to tiagabine might be contraindicated.

Similar to the findings of previous studies,^11,37,38 we found a tendency for a relationship between the change in arousal threshold and change in OSA severity, whereby an increase in arousal threshold was associated with a reduction in AHI. Arousal from sleep has an important protective role in patients with blunted chemical drive (i.e., obesity hypoventilation syndrome) and in instances when other defensive mechanisms such as upper airway muscle recruitment fail. However, in about 30–50% of OSA patients,^8,39 cortical arousals are not required to recover ventilation but may actually perpetuate breathing instability and subsequent upper airway closure during sleep. A small number of hypnotics have been tested so far in OSA patients trying to increase arousal threshold: 0.25 mg of triazolam increased arousal threshold by 24%,¹⁵ 100 mg of trazodone by 33%,³⁷ 3 mg of eszopiclone by 29%,³⁸ and 7.5 mg of zopiclone by 20%.¹⁸ Tiagabine did not have a consistent effect on the arousal threshold: the change in arousal threshold between nights in most cases was minimal and unlikely to exceed night-to-night variability. These findings suggest that the ability of various hypnotics to increase the arousal threshold and their ability to promote deeper sleep in the context of OSA may differ with drug type and dose, and more work is required to characterize the patients who may benefit from hypnotics for the treatment of OSA.

Limitations

This study has several limitations. First, the small sample size does not allow for firm conclusions to be drawn about the effect of tiagabine on OSA severity. However, given that this drug failed to increase NREM 3 sleep and the arousal threshold, as well as reduce AHI in this group, it is unlikely that tiagabine will play a significant role in the development of a pharmacologic agent to treat OSA. Second, we tested only 1 dose of 12 mg of tiagabine. It is possible that higher doses would effectively increase NREM 3 sleep in some patients, although it might also generate deeper respiratory suppression and muscle relaxation in others, worsening the severity of the disease. Third, we did not directly measure upper airway dilator muscle activity in this group. Thus, we cannot conclusively determine whether major decreases in muscle contractility occur, at least in some patients with tiagabine. We felt that placement of intramuscular wires might have generated sleep disruption and discomfort, significantly affecting the possibility of finding an increase in NREM sleep stage 3. Finally, we did not measure phenotype traits on and off the drug. It would have provided useful insights on the effects of tiagabine on phenotype traits; however, we were able to test the effect of this drug on the arousal threshold in all patients and on the airway collapsibility from peak flow in a subset of them.

CONCLUSIONS

In this study, we showed that tiagabine increased SWA in the EEG power spectrum but did not increase NREM sleep stage 3 or arousal threshold, and it had no overall impact on OSA severity. Based on these findings, tiagabine should not be used for OSA treatment in the clinical setting.

FUNDING

This research project received generous philanthropic funding from Fan Hongbing, President of OMPA Corporation, Kaifeng, China. This work was also supported by the American Heart Association (15POST25480003), National Institutes of Health grants R01 HL102321 and P01 NIH HL095491 as well as the Harvard Catalyst Clinical Research Center: UL1 RR 025758-01. Dr Taranto-Montemurro is supported by the American Heart Association (15POST25480003). Dr Edwards is supported by the National Health and Medical Research Council (NHMRC) of Australia’s CJ Martin Overseas Biomedical Fellowship (1035115). Dr Sands was supported by the National Health and Medical Research Council of Australia and R.G. Menzies Foundation (1053201, 1035115) and is currently supported by the American Heart Association (15SDG25890059) and American Thoracic Society Foundation Unrestricted Grant. Dr Marques was supported by Capes Foundation, Ministry of Education of Brazil. Dr Eckert is supported by a NHMRC R.D. Wright Fellowship (1049814).

Clinical trial name: Tiagabine to Enhance Slow Wave Sleep in Patients with Sleep Apnea, NCT02387710. URL: https://clinicaltrials.gov/ct2/show/NCT02387710?term=tessa&rank=1

DISCLOSURE STATEMENT

This manuscript describes an investigational use of tiagabine for obstructive sleep apnea treatment. The authors do not report any conflict of interest related to this project. LTM receives a fee as a consultant for Novion Pharmaceuticals, Inc. DPW and AW receive a fee as consultants of Philips Respironics.