Polysomnographic Endotypic Traits Associated with AtoOxy Efficacy in OSA

Laura Gell, PhD

Polysomnographic endotypic traits including lower collapsibility, lower loop gain, and more were associated with greater efficacy of AtoOxy - a combination therapy of atomoxetine and oxybutynin- in patients with obstructive sleep apnea (OSA). These traits were also independent of baseline disease severity.

The findings were presented at the American Thoracic Society 2022 International Conference in San Francisco.

Previously, the combination of atomoxetine plus oxybutynin had been shown to significantly reduce the severity of OSA when administered before sleep.

Despite this a recent trial - ComboPlus, NCT03892772- found the response to AtoOxy to be heterogenous.

For the present study, the same investigative team led by Laura Gell, PhD, of the Division of Sleep Medicine at Harvard Medical School, hypothesized that therapeutic efficacy would be associated with “favorable pathophysiology, greater arousal threshold and a greater capacity for pharyngeal muscle compensation”.

A tota of 19 with OSA completed a randomized placebo-controlled crossover study with polysomnographic sleep studies performed at baseline and following 3 nights of AtoOxy therapy including 80/5 mg qhs and half doses that’s were administered on the first night.

An advanced analysis of the baseline study provided endotypic traits including loop gain, arousal threshold, collapsibility, and muscle compensation. 

Meanwhile, multivariable linear regression quantified associations between treatment efficacy and the 4 traits, before adjusting for baseline AHI, and the primary analysis determined if the traits explained unique heterogeneity in efficacy beyond AHI, and subsequent analysis described associations between individual traits and efficacy.

Finally, leave-one-out cross validation provided conservative estimates of predictive value.

AtoOxy efficacy was strongly associated with the four traits according to multivariable regression, independent of baseline AHI (P=0.00025 vs. AHI alone, R2=0.75 vs. 0.05).

Gell and colleagues noted that each of the 4 traits contributed significantly to the model.

As they hypothesized, greater AtoOxy efficacy was associated with lower collapsibility, lower loop gain, higher arousal threshold, and better muscle compensation.

Notably, the model fully separated responders (>50% reduction from baseline, N=10/19) and non- responders (N=9/19).

Predictive value was “promising” following cross validation, as predicted responders lowered AHI from baseline by 74% while predicted non-responders lowered AHI by 31% (P-difference=0.02). 

Meanwhile, the odds ratio for response was 18.7[1.6,223], and prediction accuracy was 79%.

A mixed model analysis confirmed greater AtoOxy efficacy compared to placebo in predicted responders.

“Our results provide new insights into the physiological factors responsible for greater AtoOxy efficacy, and may provide a means to identify patients most suitable for this therapy,” the team wrote.