Pharyngeal Muscle Control Mechanisms of Atomoxetine-plus-oxybutynin in Obstructive Sleep Apnea

Overview

Current therapies available for obstructive sleep apnea (OSA) have varying degrees of efficacy due to the complex nature of the disorder. A reduction in pharyngeal muscle activity characterizes OSA, and recent research has shown that combining atomoxetine and oxybutynin improves OSA severity. Thus this may be a viable treatment option. However, the specific effects of these agents alone and in combination on pharyngeal muscle activity remain unknown. The current study will look at the impact of each drug on pharyngeal muscles to gain insight into the mechanisms of this combination.

The goal of this randomized controlled double-blind crossover physiology study is to examine the pathophysiological mechanisms of how atomoxetine-oxybutynin (AtoOxy) versus atomoxetine alone effect pharyngeal dilator muscle activity in people with obstructive sleep apnea (OSA). Subjects will attend a virtual Screening and Consent visit to assess eligibility for enrollment. Participants will take part in a video call with the consenting doctor to obtain consent (Zoom). After consent, eligible subjects will be randomized to receive the first period of study medication for 1 night. Allocations will be concealed from the subjects, investigators, physicians, and outcomes assessors. Patients will perform in random order the two study interventions: A) atomoxetine 80 mg plus oxybutynin 5 mg (full doses described) B) atomoxetine 80 mg plus placebo (encapsulated to mimic oxybutynin) There will be a 1-week washout between periods. Subjects will undergo acute administration of the drugs. The treatment will be administered after the first sleep cycle (60-90 minutes after sleep onset), to provide within-night control data. An additional night (open-label placebo) will be performed at the end of the first two study arms. In addition to the listed outcome, we will also examine effects of each treatment on clinical parameters: apnea hypopnea index, hypoxic burden, arousal index, within the caveats that the studies are physiological in nature (per experimental equipment). Data analysis Apneas, hypopneas, sleep stages and arousals from sleep will be scored using current AASM guidelines (hypopneas defined by at least a 30% reduction in airflow in conjunction with either 3% desaturation or arousal) by a technician blinded to the study condition. Breath-by-breath tables will be made describing ventilation, ventilatory drive (from intraesophageal diaphragm EMG), peak and tonic genioglossus and tensor palatini activity. For each night, data during pre-treatment control periods and on-treatment periods will be compiled separately. For a given patient-condition (on-treatment vs. pre-treatment control), data will be sorted based on ventilatory drive decile into 10 bins. Corresponding values of peak genioglossus activity, peak tensor palatini activity, and ventilation will be calculated for each drive decile (medians). These decile-binned data will be used in linear mixed effects models to address the hypotheses described above. Data from non-REM sleep during physiological sleep (arousals and two post-arousal breaths excluded) will be used in primary modelling. Arousal threshold will be based on the ventilatory drive swings (intraesophageal catheter) preceding arousal from sleep (presented as %eupnea). Loop gain will be calculated as the increase in ventilatory drive (ventilatory overshoot) divided by the preceding reduction in ventilation. Statistical analysis plan. A per-protocol analysis is planned on the basis that this study aims to assess the mechanisms of action of the agents when present in the circulation rather than the clinical effectiveness. For Aim 1a, the quantitative primary outcome will be the increase in peak genioglossus activity with Ato-Oxy vs. atomoxetine alone (mixed model analysis, see below), in drive-unadjusted analysis. Values will be calculated for the 1st decile ventilatory drive level. Peak genioglossus activity will be expressed using standard units (percent maximum wake levels) in standard analysis (%baseline will be used if needed to handle outliers). For Aim 2a, the quantitative primary outcome will be the increase in tonic tensor palatini activity with Ato-Oxy vs. atomoxetine alone (mixed model analysis, see below), in drive-unadjusted analysis. Values will be calculated for the 1st decile ventilatory drive level. For Aim 3a, the quantitative primary outcome will be the increase in ventilation with Ato-Oxy vs. atomoxetine alone (mixed model analysis, see below), in drive-unadjusted analysis. Values will be calculated for the 1st decile ventilatory drive level. Standard linear mixed-effects model analysis for crossover studies will be used to evaluate the effect of AtoOxy versus atomoxetine alone on the outcome (fixed effect), with each patient treated as a random effect, chosen primarily to account for incorporation of incomplete data. Models will also include control data obtained from the across-subject placebo night (fixed effect). Model will also include pre-treatment data period (within-subject control); 'SubjectNight' (unique identified for each night) will be included as a random effect to account for night-to-night measurement variability. Models will therefore include 6 data conditions per individual, and will follow the general form: Y \~ AtoOxy + Placebo + WithinControl + (1\|Subject) + (1\|SubjectNight) where atomoxetine alone is the reference condition, and thus the AtoOxy term describes the difference in outcome variable (peak genioglossus activity, Y) with Ato-Oxy versus atomoxetine alone (primary outcome analysis). For each condition, models will include 10 values of peak genioglossus data based on 10 values of ventilatory drive, i.e. one for each decile per condition; thus ventilatory drive will be additionally included as a fixed effect. For the primary drive-unadjusted analysis, "Decile" will be included to describe the continuous decile number centered (0-9) such that differences in the outcome variable are assessed at the first decile (Decile=0). Decile squared will also be included to incorporate expected non-linearity based on prior experience. An alternative form of the same model will be used to describe differences versus placebo: Y \~ AtoOxy + Atomoxetine + WithinControl + (1\|Subject) + (1\|SubjectNight) where AtoOxy describes the difference in outcome variable with Ato-Oxy versus placebo, and Atomoxetine describes the difference in outcome variable Y on atomoxetine alone versus placebo. An additional alternative will be used to describe differences versus the within-night control: Y \~ AtoOxy + Atomoxetine + Placebo + (1\|Subject) + (1\|SubjectNight) where AtoOxy describes the difference in outcome variable with Ato-Oxy versus the within-night control, and so on. Models will be expanded to include randomization sequence (AB or BA, i.e. carryover effects) and period (i.e. effect of time) as fixed effects, if considered necessary based on changes in primary model coefficients. P\<0.05 will be used to indicate statistical significance, separately for each of the three Aims a, b, c. Secondary analyses for each aim will examined hierarchically and considered significant only if each test higher in the hierarchical order is also significant, otherwise findings will be considered exploratory and hypothesis generating. Secondary analyses The above modelling approach will assess whether there is an increase in physiological variables (at 1st decile, or median in sensitivity analysis, i.e. GGmin, TPmin, Vmin, Dmin) with AtoOxy vs. atomoxetine alone in drive unadjusted analysis. These analysis will be repeated, in drive-adjusted analysis, to assess differences in muscle activity separate from the effect of drive per se (GGpassive, TPpassive, Vpassive). Instead of the Decile term, a term for "Drive" will be included in the model. Drive will be square-root transformed for analysis of ventilation, but not muscle activity, based on prior experience with this modelling. Drive data will be centered at Drive=100%eupnea to facilitate estimation of differences in activity levels at eupneic drive. We will also seek to calculate any increase in GGmin/TPmin/Vmin due to drive versus non-drive mechanisms. In an additional analysis, a AtoOxy-by-drive interaction term will be added to a modified primary model. Significant interaction will be used to quantify whether AtoOxy increases muscle responsiveness. Differential effects of AtoOxy versus atomoxetine alone on additional outcome variables will be assessed in a similar manner. Power analysis. 15 patients (as an approximation of 16 at treatment 1 and 14 at treatment 2, i.e. 30 active treatment visits) will provide approximately 80% power (alpha=0.05) to detect a physiologically-significant difference in the primary outcome between the two treatment arms (50%) based on an effect size of 1 (treatment difference of 50%, SD of each treatment effect = 50%) based on previous physiological data and computer simulations. Differences versus placebo are expected to have 2-3 times the effect size, providing more than sufficient power. Dropouts prior to the Placebo night are expected to have no meaningful impact on primary analysis power.