Authors: Atqiya Aishah (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Molly Kim (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Laura Gell (4Apnimed, Inc.), Daniel Vena (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Ali Azarbarzin (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Huy Pho (4Apnimed, Inc.), Daniel Norman (2Santa Monica Clinical Trial, Santa Monica, CA, USA), Joseph Ojile (3Clayton Sleep Institute, St Louis, MO, USA), Neda Esmaeili (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Luigi Taranto-Montemurro (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA; 4Apnimed, Inc.), Andrew Wellman (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Scott A Sands (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA), Ludovico Messineo (1Brigham and Women’s Hospital and Harvard Medical School, Boston, MA, USA)
Categories: Article
Source: Thorax
Authors: Atqiya Aishah, Molly Kim, Laura Gell, Daniel Vena, Ali Azarbarzin, Huy Pho, Daniel Norman, Joseph Ojile, Neda Esmaeili, Luigi Taranto-Montemurro, Andrew Wellman, Scott A Sands, Ludovico Messineo
Recently, a combination of the noradrenergic atomoxetine with either the anti-muscarinic aroxybutynin or trazodone demonstrated substantial effect on OSA severity. However, atomoxetine may contribute to apnea cycling and reduced dug tolerability as it is wake-promoting especially in a subgroup of patients (poor cytochrome 2D6 metabolizers), who may experience higher blood concentration of this medication. Therefore, we investigated the effect of a potentially more manageable noradrenergic, viloxazine, with and without trazodone, on OSA severity.
A double-blind, placebo-controlled, crossover study was performed; twenty-five patients with OSA aged 18–75 years were randomized to 500mg viloxazine, 500/75mg viloxazine-trazodone (vilo-trazo) or placebo. Drugs were taken before bed for two weeks with a 1-week run-in period at lower doses (200mg viloxazine with/without 50mg trazodone) and a 1-week washout between treatments. In-laboratory polysomnography was performed at the end of each crossover period. Mixed model analyses compared the effect of vilo-trazo vs. placebo on AHI4 (apnea-hypopnea index with 4% desaturations; primary-outcome) and hypoxic burden (secondary-outcome). Additional secondary outcomes examined the effects of vilo-trazo vs. viloxazine alone on total sleep time (TST) and wake-after-sleep-onset (WASO). Safety endpoints (patient-reported outcomes, heart rate, and adverse events) were also assessed.
Vilo-trazo reduced AHI4 (mean difference [95% CI]: 10.4 [6.6, 13.6], P<0.001) and hypoxic burden (16.7 [21.8, 9.6] %.min/hr, P<0.001) vs. placebo. Compared to viloxazine, TST tended to be longer on vilo-trazo (22.3 [−46.0 to 1.4] min, P=0.07), while WASO was unchanged. TST and WASO remained significantly reduced on vilo-trazo vs. placebo. Viloxazine had comparable effects to vilo-trazo on OSA severity. Both interventions worsened patient-reported outcomes, albeit to a lesser extent on vilo-trazo, and increased heart rate vs. placebo. Commonly reported adverse events were insomnia, constipation, headache, and xerostomia.
Viloxazine-trazodone reduced OSA severity. Potential deleterious effects of viloxazine on sleep quality appeared partly attenuated by trazodone.
In recent years, the spotlight has turned toward pharmacotherapy as a potential game changer in the treatment of obstructive sleep apnea (OSA). Current available therapeutics have presented several challenges including poor adherence (1), variable response rates (2, 3) or limited accessibility (4, 5) among patients. Therefore, there is an unmet need for alternative treatments that offers effectiveness and access, but also are accepted by patients, to ultimately improve cardiovascular and functional outcomes in OSA (6–9). In this regard, recent trials demonstrating the efficacy of atomoxetine and oxybutynin/aroxybutinin on OSA severity have ignited significant interest in pharmacological interventions for OSA (10–13). Atomoxetine, a noradrenergic reuptake inhibitor, is thought to exert its effect on OSA by improving upper-airway muscle responsiveness (12, 13), while the anti-muscarinic oxybutynin has been suggested to partly contribute to upper airway muscle activation and—via a modest sedative hypnotic effect (14, 15)—offset atomoxetine’s wake-promoting properties. Other studies testing similar drug combinations have corroborated the potential of this treatment approach for individuals with OSA (16–18).
However, a key aspect that could hinder the acceptance or long-term use of these drug combinations is the potential for associated side effects. For example, atomoxetine can cause sympathomimetic and gastrointestinal reactions (19), especially in ~7% of Caucasians who have little activity of the cytochrome P450 (CYP) isoenzyme 2D6 which metabolizes the drug (i.e., poor metabolizers) (20, 21). Likewise, oxybutynin reduces bladder sensitivity and will be less suitable in patients with urinary retention concerns. Therefore, combination agents with similar effects on OSA severity, but lower risks of side effects are of interest.
Viloxazine, as an alternative norepinephrine reuptake inhibitor to atomoxetine, can be metabolized by other CYPs (21), and is expected to obviate poor metabolizer issues of atomoxetine. Specifically, viloxazine’s peak blood concentration will not be as elevated as atomoxetine’s in poor metabolizers for CYP2D6 (21), lowering the likelihood for viloxazine’s blood levels to rise beyond expectations. Trazodone, as an alternative to oxybutynin/aroxybutynin, has established hypnotic properties (22), and has been shown to reduce symptoms of insomnia (23). On the basis that a main physiological mechanisms of oxybutynin/aroxybutynin is to raise the arousal threshold and consolidate sleep, trazodone is likely to provide a stronger hypnotic effect that could be more suitable for a broader range of patients including those with greater arousability.
Here we sought to investigate the effect of viloxazine, with and without trazodone, on OSA severity in this double-blinded, placebo-controlled, crossover study. The main hypothesis was that the combination of viloxazine and trazodone (vilo-trazo) would reduce OSA severity (as per apnea hypopnea index [AHI4], with hypopneas scored when flow reduction is associated with 4% oxygen desaturation) vs. placebo. Secondary endpoints were the effect of vilo-trazo on hypoxic burden with 4% oxygen desaturation (HB4) vs. placebo, and the effect of vilo-trazo on total sleep time and wake after sleep onset (WASO) vs. viloxazine alone. Additional exploratory endpoints were the across arm differences in other OSA severity parameters, sleep stage distribution, arousals, and questionnaires to measure sleepiness and sleep impairment (i.e., Epworth sleepiness scale (ESS) and patient-reported outcomes measurement information systems (PROMIS)). In exploratory analyses, predictors of treatment response and treatment efficacy were also assessed.
Twenty-five participants with OSA were recruited across three Clayton, St. Louis, MO (N=14), and Santa Monica Clinical Trials, Los Angeles, CA (N=10), Brigham and Women’s Hospital, Boston, MA (N=1). Participants were either treatment naïve or not currently on treatment for their OSA. Participants were included if they had an AHI4 of 10–45 events/h inclusive, PROMIS-Fatigue, PROMIS-sleep related impairment or PROMIS-sleep disturbance scores of > 11, and body mass index (BMI) between 18.5 and 40 kg/m^2^. Exclusion criteria uncontrolled hypertension, any underlying major organ disease, any sleep disorder besides OSA, clinically significant urinary or gastrointestinal retention, benign prostatic hypertrophy, severe or frequent gastrointestinal reflux, pregnancy, untreated narrow angle glaucoma, attempted suicide within 1 year prior to screening, or current suicidal ideation. People were also excluded if, at the baseline assessment, they exhibited a prolonged QTc interval (>450 ms in men or >470 ms in women), hypokalemia or hypomagnesemia, high hepatic transaminases, or high total bilirubin, and estimated glomerular filtration rate < 50 mL/min. The study protocol was approved by WIRB-Copernicus Group Institutional Review Board and pre-registered online at clinicaltrials.gov (NCT05793684). All participants provided informed written consent prior to study participation.
This was a randomized, placebo-controlled, multiple-dose, 3-period crossover study (Fig 1). Initially, a screening visit was conducted to establish eligibility; during this visit, informed consent, demographics, and medical history, including medications, physical examination, vitals (blood pressures and heat rate), clinical laboratory testing and 12-lead electrocardiogram (ECG) were recorded. The following questionnaires were also PROMIS-Fatigue, PROMIS-sleep impairment, PROMIS-sleep disturbance and ESS. Participants who were eligible completed a second screening visit, which consisted of an in-lab polysomnography (PSG) to confirm OSA per inclusion criteria. Subsequently, participants were randomized to three 500mg Viloxazine, 500/75mg vilo-trazo or placebo. Each arm consisted of 2 weeks of treatment, followed by a 1-week wash-out. The medications were taken nightly approximately 30 minutes prior to bed, with a first week of run-in of 200mg Viloxazine and 50mg Trazodone (if on the vilo-trazo arm). In-laboratory PSG were performed after 2 weeks (i.e., on the 14^th^ night) at the end of each cross-over period. In-laboratory PSG were recorded in accordance to the American Academy of Sleep Medicine guidelines (24). At each in-laboratory PSG visit, vital signals and questionnaires (PROMIS-Fatigue, PROMIS-sleep impairment, PROMIS-sleep disturbance, ESS) were completed in the evening. Next morning procedures included recording of vital signs, Karolinska Sleepiness Scale (KSS) questionnaire and symbol digit modalities test (SDMT), 1-hour post-awakening. During the SDMT, a cognitive assessment tool that can capture alertness deficits (as potentially observed in people after impaired sleep) (25), participants were tasked with substituting digits for abstract symbols using a reference key, and the assessment was conducted both in a written and oral format; the score was determined by summing the number of correctly matched digits, meaning a higher score indicates less impairment in the tasked participant.
Adverse events and concomitant medication monitoring was performed regularly during treatment and wash-out periods, as well as after each in-laboratory PSG, and at the end of the last study arm’s wash-out period.
PSGs were scored for study endpoints by a centralized PSG center (Sleep Strategies, Ottawa, ON), blinded to treatment assignment, following AASM scoring criteria (24). In line with the pre-specified, exploratory outcomes of the study, additional indexes of OSA severity were computed for each study besides AHI4, AHI4 in both rapid eye movement (REM) and nonREM sleep, AHI3a (apnea hypopnea index where hypopneas were scored in the presence of ≥3% oxygen desaturation from pre-event baseline or if the respiratory event was associated with an arousal), HB3 (hypoxic burden with 3% desaturations) and oxygen desaturation index based on ≥4% desaturation (ODI4).
Previously validated algorithms (26–30) were used to estimate baseline VCOMP, during nonREM sleep from the PSG-derived flow signal. VCOMP is an estimate of pharyngeal muscle responsiveness and was calculated from the baseline PSG as the difference between VACTIVE (ventilation just before arousal, when the upper airway muscle are activated) for VPASSIVE (ventilation during sleep at eupneic drive, when the pharyngeal muscles are relatively passive). VCOMP was reported in %VEUPNEA, defined as unobstructed ventilation during sleep, with higher values reflecting higher muscle compensation.
The study was powered at 80% to detect a difference in the primary endpoint (AHI4 on Vilo-trazo vs. placebo) of 11±11 events/h at α = 0.05. Data are expressed as mean ± standard deviation (SD) or median [IQR] where appropriate (Shapiro-Wilk normality test), and were analyzed using linear mixed effect analyses with the treatment arm as fixed effect and the subject as random effect. All models were adjusted by period, sequence, and body position during sleep. To address data skewedness, AHI and ODI were square-root transformed and back-transformed for presentation; HB4 was log-transformed and back-transformed for presentation in accordance with previous studies (31). Based on previous research (32, 33), specific parameters were evaluated as leading-candidate potential predictors of treatment response, including baseline VCOMP and baseline BMI. Other baseline parameters such as AHI4, ESS, loop gain, collapsibility (VPASSIVE), and arousal threshold were also evaluated. Each patient contributed a single value (baseline) for each of these parameters, which were then indexed and divided into tertiles. Subsequently, treatment efficacy was explored in individuals categorized in the top (third) and bottom (first) tertiles for each parameter according to the mixed effect analysis described above. Analysis was performed using MATLAB (Mathwork, Natick, MA).
Twenty-five participants were randomized, including 6 dropouts, one due to relocation and 5 due to adverse events (Fig 1). Baseline characteristics of all 24 participants who contributed to the final analysis (i.e. completed at least one treatment arm) are outlined in Table 1.
The primary hypothesis of the study was met, as vilo-trazo reduced the AHI4 by 10.4 [6.6, 13.6] events/h (mean difference [95%CI]) vs. placebo (P<0.001; Fig 2A, Table 2). HB4 was also reduced by 16.7 [21.8, 9.6] %.min/hr on vilo-trazo vs. placebo (P<0.001; secondary outcome; Fig 2B, Table 2). Additionally, vilo-trazo reduced nonREM and REM AHI4, as well as the AHI3a, HB3 and ODI4 (Table 2). However, vilo-trazo also reduced total sleep duration by 37.7 [14.5, 60.9] mins, (P=0.002) and increased WASO by 26.6 [4.7, 48.6] mins (P=0.02) compared to placebo.
Viloxazine had similar effects to vilo-trazo on OSA severity vs. placebo, including reductions in the AHI4 (11.4 [7.8, 14.4] events/h, P<0.001; Fig. 2A, Table 2) and the HB4 (16.5 [9.4, 21.6] %.min/hr, P<0.001; Fig 2B, Table 2). These reductions were also observed for nonREM and REM AHI4, as well as AHI3a, HB3 and ODI4. Additionally, viloxazine decreased total sleep time and raised WASO.
The reduction of total sleep time on viloxazine was of a greater magnitude than on the combination vs. placebo. In other words, total sleep time tended to be longer on vilo-trazo (22.3 [−1.4, 46.0] min, P=0.065) compared to viloxazine alone (secondary outcome). There were no differences in WASO on vilo-trazo vs. viloxazine (13.0 [−9.5 to 35.6] min, P=0.25; secondary outcome).
Sleep architecture changes were observed across different treatment arms. Overall, viloxazine alone and vilo-trazo increased N2 sleep whilst reducing REM sleep. Vilo-trazo also increased N3 sleep compared to viloxazine alone (Table 2).
Both vilo-trazo and viloxazine alone increased PROMIS-sleep disturbance compared to placebo (6.5 [2.7, 10.3], P=0.001; 6.1 [2.4, 9.8], P=0.002, respectively; higher values reflect greater disturbance), while PROMIS-sleep impairment (5.2 [1.0, 9.4], P=0.02) and PROMIS-fatigue (6.6 [2.3, 11.0], P=0.003) were only reduced on viloxazine vs. placebo (Table 3). ESS was higher (greater sleepiness) on vilo-trazo (2.1 [0.6, 3.7], p=0.008), but not on viloxazine alone (1.3 [−0.2, 2.8], p=0.08) compared to placebo (Table 3).
Both vilo-trazo (5.6 [3.6, 7.6] beats/min) and viloxazine alone (3.5 [1.5, 5.5] beats/min) increased heart rate during sleep compared to placebo. Vilo-trazo—but not viloxazine alone—increased diastolic blood pressure in the evening before the PSG by 4.0 [0.5, 7.4] mmHg vs. placebo; all other blood pressure measurements were unchanged across treatment arms (Table 3).
Unlike viloxazine alone, vilo-trazo reduced SDMT written score by 3.7 [0.8, 6.6] points (P=0.012) vs. placebo (lower values reflect greater alertness impairment; Table 3).
42%, 67% and 72% of people on placebo, viloxazine, vilo-trazo experienced at least one side effect, respectively. Commonly reported adverse events on viloxazine, vilo-trazo and placebo were headache, insomnia, constipation, and xerostomia of mild-to-moderate severity (Table 4). For both viloxazine and vilo-trazo, an estimated odds ratio above 2.0 was observed for insomnia, constipation, and xerostomia, but not for headache (Fig 3).
Five participants discontinued the study due to AEs: one participant experienced mild blurry vision with moderate fatigue after 14 days on the vilo-trazo arm and was withdrawn from the study before completing any study night. Similarly, another participant dropped out of the study while on vilo-trazo after experiencing moderate fatigue and mild blurry vision after 14 days of drug administration. Another participant dropped out due to reported moderate headache after 10 days on vilo-trazo. A participant had elevated blood pressures after 10 days of viloxazine, and experienced left upper chest pain starting after 14 days of viloxazine, which eventually led to study withdrawal. Similarly, another participant withdrew from the study due to moderate headache and mild symptoms of flushing, loss of appetite, nausea, light headedness, and chills after 2–3 days of viloxazine.
All these adverse events were deemed probably associated with the drug administration in analysis of relatedness.
In the subgroup of patients with higher baseline AHI4 (>31 events/h, top tertile), vilo-trazo and viloxazine decreased AHI4 by 60% and 57% respectively; interventions had no observable effect in the subgroup with lower baseline AHI4 (<17 events/h, bottom tertile; Table 5). In individuals with high baseline VCOMP (>7.8%eupnea), AHI4 fell by 54% on vilo-trazo and by ) were minimal. Efficacy also appeared greater in patients with higher BMI (>34 kg/m^2^), in whom AHI35% on viloxazine alone vs. placebo; effects of vilo-trazo in low Vcomp (−4.3% of VEUPNEA4 was lowered by 60% vs. placebo, but was reduced within lower BMI (<28 kg/m^2^). Similarly, vilo-trazo and viloxazine alone reduced AHI4~ by 50% and 45% respectively only in people with a high baseline ESS (≥11). Other baseline endotypic traits (loop gain, VPASSIVE~, arousal threshold) were explored but no meaningful differential effects on AHI4 between subgroups were observed.
The main finding of this study is that the combination of viloxazine and trazodone reduces OSA severity compared to placebo by roughly 54%. Despite the recognized REM suppression associated with noradrenergic drugs, this reduction in severity persisted even when the analysis was isolated to nonREM sleep, with AHI4 drops of ~50% and ~58% on vilo-trazo and viloxazine vs. placebo, respectively. In addition, vilo-trazo improved total sleep time and deep sleep compared to viloxazine alone, and, in contrast to viloxazine alone, did not exacerbate subjective symptoms such as fatigue and sleep impairment, suggesting that trazodone could somewhat mitigate the noradrenergic effects of viloxazine. However, despite meeting the study outcomes, vilo-trazo increased self-reported sleepiness (ESS) and induced several side effects, including xerostomia, insomnia and constipation. Nevertheless, this study underscores that testing noradrenergic reuptake inhibitors in combination with a hypnotic is a viable strategy for OSA suppression and should be pursued further.
For the first time, the current study demonstrated that the noradrenergic reuptake inhibitor viloxazine provides a substantial reduction in OSA severity. Given that OSA severity was reduced similarly when viloxazine was administered with or without trazodone, and that trazodone alone is not expected to have a potent effect on OSA (34), our study suggests that the main effect on OSA severity is driven by viloxazine. This finding builds on previous work describing improved OSA severity with selective noradrenergic agents combined with antimuscarinics or hypnotics (13, 16, 18, 35–38). Notably, vilo-trazo systematically reduced all indexes of OSA severity, a distinction from a previous study that examined the effect of atomoxetine and trazodone on OSA severity, finding a reduction in the AHI4, but not in AHI3A (39). Overall, the efficacy of viloxazine was similar to studies of two other selective noradrenergic reuptake inhibitors, namely atomoxetine and reboxetine (10, 40), which stand in contrast to other noradrenergic agents with a less specific action on the noradrenergic transporter, such as milnacipram or duloxetine, which appeared less efficacious (41). Taken together, available data suggests that noradrenergic reuptake inhibitors with fewer dual or off-target effects on the noradrenalin transporter inhibition, such as atomoxetine and viloxazine, are preferred for reducing OSA severity and should be prioritized for future investigation.
The addition of the hypnotic trazodone appeared to mitigate some of the wake-promoting noradrenergic effects of viloxazine that are potentially disruptive of sleep continuity. PROMIS questionnaires, total sleep time, and N3 sleep duration shifted favorably with the combination compared to viloxazine alone. The effect of trazodone was similar to that of anti-muscarinics when combined with the noradrenergic component, consistent with the notion that a key function of anti-muscarinics when paired with a noradrenergic may lie with their hypnotic effects. In the current study, unlike others, we observed that the addition of trazodone yielded increased N3 sleep duration. While total sleep time and additional sleep quality measures remained inferior to placebo on vilo-trazo, it is possible that a fine tuning of dosage of viloxazine or trazodone, or a combination with different hypnotics (i.e., oxybutynin plus zolpidem), may minimize this concern (43); further investigation is warranted.
Similarly to previous trials involving noradrenergic drugs (10, 35), vilo-trazo was associated with instances of xerostomia and insomnia, however it also exhibited a higher incidence of constipation compared to other studies (Figure 3). Of note, only 1 out of 5 patients dropped out on the low dose of therapy, suggesting a lower dosage of viloxazine could potentially lead to more favorable outcomes in terms of adverse events.
Increases in heart were anticipated and in line with findings from previous studies due to the sympathomimetic effect of viloxazine (10, 13, 35). The observed increase in evening diastolic blood pressure on vilo-trazo, was not supported by increases in morning values or systolic levels or with viloxazine alone; diastolic blood pressure changes have been absent in most (44, 45), but not all trials (10) of noradrenergic combination interventions so far.
The increase in ESS observed with vilo-trazo is comparable to the effect seen after CPAP withdrawal (46) and has not been reported in previous trials of other noradrenergic combinations. The significance of this increase is further validated by the SDMT performance on vilo-trazo, which decreased by ~4 points vs. placebo—a drop accepted as clinically significant (47). While trazodone is not typically considered to induce a “hangover effect” (23) and has been used safely at higher doses to raise the arousal threshold in individuals with OSA (34), there may be instances in which trazodone’s half-life can increase (48, 49). Moreover, side effects are dose-dependent (23), such that it is plausible that the dosage of trazodone used in this study might have been too high (previous studies on insomnia have shown that 50 mg of trazodone is sufficient for sleep maintenance). Of note, reducing the dosage of viloxazine would potentially decrease the “need” to offset its noradrenergic effects. In addition, increased morning sleepiness might be of lesser concern if investigated in a targeted subgroup of patients with comorbid OSA and insomnia.
Participants with estimated good muscle compensation (high VCOMP) showed better responses to vilo-trazo compared to those with low VCOMP. This contradicts previous assumptions suggesting that interventions targeting a specific endotype are more likely to be effective in individuals with an abnormal endotype. Recent studies, however, started to indicate that treatment responders are often those with the least severely altered endotype, since more readily “fixable”. For example, oral appliances works best in patients with milder collapsibility (50), and acetazolamide in those with lower loop gain (36), despite these characteristics not being the theoretical targets of the intervention (i.e., oral appliances and acetazolamide reduce collapsibility and loop gain, respectively). Similarly, individuals with good muscle responsiveness exhibit better response to interventions targeting pharyngeal muscles, including hypoglossal nerve stimulation, acetazolamide and atomoxetine-oxybutynin (32, 36, 51), as a functional increase in muscle force and airway patency may be more readily achieved with muscle stimulation, while people with poor muscle compensation may have difficulty translating neural activation into airway patency. Of note, vilo-trazo led to a further 30% AHI4~ drop vs. viloxazine alone in people with high VCOMP (Table 5), which implies a potential synergistic effect of trazodone on the stimulation of the upper airway muscles in this group (52), an expected effect of raising the arousal threshold. Preliminary data demonstrated reduced VCOMP is associated with complete lateral pharyngeal wall collapse (53), a site of collapse that may be particularly challenging to improve with muscle activation. Therefore, more than one intervention may be required to manage these OSA patients.
The observation that patients with higher BMI in this study responded better to both active treatments and in particular to vilo-trazo (a further 30% drop) is somewhat in contrast with previous preliminary findings indicating low BMI as a predictor of treatment response to atomoxetine-plus-oxybutynin (54). One possible explanation for this discrepancy is that non-obese OSA patients often have a more heterogeneous OSA pathogenesis, that is, characterized by other abnormal traits, such as a low arousal threshold (55). Regardless, obese OSA patients have typically more respiratory events than their non-obese counterparts, and thus are more susceptible to symptoms and cardiovascular risk, making treatment more compelling. To further support this notion, patients with more severe OSA or sleepiness at baseline greatly responded to the active treatments compared to milder/non-sleepy patients. These findings may help guide those most likely to benefit from such a therapy, and thus potential targets for future clinical studies.
This study has several its small sample size did not allow for a conclusive assessment on patients reported outcomes for the drugs tested, however there was a clear signal for objective reduction in OSA severity, and objective and subjective sleep disruption, especially when viloxazine was administered alone.
The study design facilitated examination of interventions after one week of full dose administration. This relatively-short duration was sufficient to detect an efficacy signal and characterize the early safety profile of the agents studied, and ultimately indicate that dosing could be further optimized.
Another important limitation of this current trial is the lack of an arm with trazodone alone; however based on previous trials (34) we expected minimal efficacy of this hypnotic on AHI in patients with OSA.
Additionally, the endotypes used to explore treatment response were assessed non-invasively without gold-standard procedures. However, we note that all the most important recent studies investigating the effects of pharmacotherapies on OSA severity have followed a similar approach to avoid concerns relating to effects of invasive instrumentation on sleep.
The combination of viloxazine and trazodone demonstrated significant reductions in AHI and hypoxic burden. In the combination, viloxazine, the third noradrenergic reuptake inhibitor found to be efficacious in OSA, was found to be the driver of the AHI reduction, while trazodone partially-mitigated some of the noradrenergic effects of viloxazine. People with high VCOMP and high BMI at baseline had the strongest responses to vilo-trazo. Overall, the current study supports the notion that pairing a noradrenergic agent with a hypnotic—which doesn’t necessarily need to be an antimuscarinic—is a promising strategy for OSA treatment.
However, since vilo-trazo administration also resulted in reduced total sleep time and increased ESS compared to placebo, future studies are warranted to determine whether a reduced dose of viloxazine (and potentially trazodone) in the combination could improve sleep-related measures and minimize side-effects whilst maintaining therapeutic efficacy.