The goal of this detailed randomized controlled mechanistic clinical study, with
gold-standard measurements of ventilatory drive and dilator muscle activity, is to test the
hypotheses that acetazolamide improves OSA in patients with (N=18) but not without (N=18)
drive-dependent OSA (i.e. drive-dependent status explains treatment efficacy). We will also
show that acetazolamide efficacy is explained by mitigating drive-related reduction in
pharyngeal obstruction. A 4-wk open-label extension will explore repeated-dose efficacy
without invasive measurements in both subgroups.
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, subjects will first attend a baseline routine sleep study to confirm
eligibility (apnea-hypopnea index >15 events/hr) and establish the baseline characteristics.
Patients will subsequently attend a specialized physiology night with additional
gold-standard measurement of ventilation, ventilatory drive, genioglossus muscle activity,
which will serve to establish drive-dependent status.
Those eligible for the study will receive in randomized order the following:
A) Acetazolamide (2x250mg) B) Placebo
Each agent will be administered for 3 nights (2×250 mg pills or 2 matching placebo pills),
with a half dose on night 1 (1 pill, same as our preliminary study). Patients will return to
the sleep laboratory on the 3rd night of each period for outcomes data collection.
Investigators, physicians, participants and outcomes assessors will be blinded to
intervention will be blinded to allocation until study completion.
Studies will be 1 week apart (i.e. 3-day treatment, 4-day washout). Adverse event data will
be carefully collated.
An open-label extension period (4-weeks) will examine whether differences in acetazolamide
efficacy persist with repeated doses between drive-dependent and classic OSA, in the absence
of invasive instrumentation (routine sleep equipment only). A 4-week duration is considered
the minimum duration necessary for reliable assessment of short-term repeated-dose efficacy
and effects on symptoms (e.g. Epworth Sleepiness scale assesses sleepiness over the last
month).
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.
Briefly, breath-by-breath values of ventilation and ventilatory drive during sleep are
tabulated for the entire night (arousals excluded). Data are sorted into 10-bins (deciles) of
ascending drive, and are plotted graphically to show how ventilation ("Flow", median[IQR])
changes with increasing ventilatory drive ("Drive"). We note that functional collapsibility
is represented as the value of Flow at normal Drive (100%eupnea), and functional upper airway
muscle effectiveness is represented by the slope of the curve. The arousal threshold is
represented as the median value of drive preceding scored arousals. Loop gain, not drawn for
focus, is calculated as the reduction in Drive that occurs in response to a prior increase in
Flow). Plots will be constructed at Baseline, on Acetazolamide, and Placebo for each patient.
These plots provide a means to interpret the physiological mechanisms of action of any
intervention.
Statistical Analysis Plan
Efficacy by subgroup. Although the study is a mechanistic trial, an intention-to-treat (ITT)
analysis will be performed; results for individuals who started a period of treatment and
discontinued will be imputed using multiple-imputation with chained equations ("MICE").
Reasons for dropouts will be carefully documented to discriminate between potential dropouts
due to AEs as opposed to discontinuation for reasons that are clearly unrelated to study
medication (scheduling, intolerance of study equipment, COVID-19, family emergency); an
independent DSMB will facilitate adjudication.
The quantitative primary outcome variable will be the continuous percent change in
apnea-hypopnea index (AHI) from baseline; the difference in this outcome variable between
intervention and placebo (i.e. placebo-corrected effect) will be evaluated using linear mixed
model analysis. The primary comparison will be whether the placebo-corrected reduction in AHI
from baseline with acetazolamide is greater in drive-dependent OSA vs. classic OSA subgroups
(per intervention × subgroup interaction, fixed effects). Subject will be included as a
random effect, and models will adjust for period and sequence effects (AB or BA, i.e.
carryover effects); the approach also enables incorporation of incomplete data. P<0.05 will
indicate statistical significance. As appropriate, percent change from baseline will be
transformed to provide a symmetric (normal) distribution using y=x/(2-x) whereby a halving
and doubling of the outcome variable have the same magnitude of effect on the transformed
variable; results will be back-transformed for presentation.
In the absence of a significant interaction, a significant and clinically-meaningful main
effect of intervention within drive-dependent OSA, but not within classic OSA, will also be
interpreted as a sign of a meaningful subgroup effect, noting that an important goal is to
provide a subgroup enriched for responders in a future trial.
We will also report the response rate i.e. percentage of patients who had a 50% percent
reduction in AHI. Secondary outcomes will also be assessed using the same mixed model
approach (percent change in hypoxic burden, arousal index; absolute change for N1%);
variables will be transformed if necessary to provide a normal distribution before analysis.
Similar analysis will be performed for the open-label extension period (absolute change from
baseline will be used for each subjective outcome variable); changes in subjective outcomes
are exploratory and will be used to provide power analysis for a future trial.
Mechanistic analysis: A per-protocol analysis is planned on the basis that the goal is to
assess the mechanisms of action of the agents when present in the circulation. The primary
quantitative outcome variable is the increase in ventilation measured at "Nadir Drive" (i.e.
1st decile = 5th centile of drive). The absolute change in this variable with acetazolamide
versus baseline will be calculated, and the above analysis approach will examine whether
acetazolamide-induced increases in drive promote a greater increase in ventilation in
drive-dependent vs classic OSA (per intervention × subgroup interaction). We will also use
the same approach to describe the changes in ventilation at Median Drive. We will also
examine whether traits causing OSA changed similarly with treatment within each subgroup
(collapsibility, compensation, loop gain, arousal threshold).
Power analysis: Sample size is based on the primary outcome model: 36 patients will provide
80% power to detect a 25% mean increase in acetazolamide efficacy in drive-dependent vs.
classic OSA subgroups (e.g. average responses = 50% and 25% reduction from baseline, placebo
corrected). Power is based on an estimated uncertainty of SD=18% (using model residuals from
preliminary data; and N=5000 simulations). Although power analysis was based on N=18 patients
with drive-dependent OSA and N=18 patients with classic OSA, in the case of a study
imbalance, we will seek to study at least N=13 individuals from each subgroup (>1:2 ratio),
leaving the total sample at 36. N=13 and N=23 patients respectively will provide 80% power to
detect a 26.5% mean increase in acetazolamide efficacy in drive-dependent vs. classic OSA
subgroups (e.g. average responses = 50% and 23.5% reduction from baseline, placebo
corrected). Power to detect significant efficacy within drive-dependent OSA is >99%
(%reduction in AHI from baseline with acetazolamide vs. placebo). Power is >90% to detect a
greater increase in ventilation (1st decile value during sleep) in drive-dependent OSA vs.
non-drive dependent OSA. Analysis of the open-label extension period will quantify whether
the change in outcomes with acetazolamide vs. placebo persists within each subgroup; Power
for the drive-dependent group is >99%.