Sleep Apnea
Conditions
Brief summary
This study seeks to employ advanced methods to estimate the individual factors contributing to sleep apnea from standard recordings made during routine clinical sleep studies. This study focuses on breathing control or loop gain as one of the factors contributing to sleep apnea. Increased levels of oxygen in the air is known to make breathing more stable by lowering loop gain. Here, our goal is to use a new method capable of detecting a reduction in loop gain with oxygen. The investigators also aim to test whether a high loop gain measured at baseline/placebo predicts a greater improvement in sleep apnea with oxygen therapy.
Detailed description
In a single-blinded randomized crossover study, inspired oxygen/air (40%/21%) is delivered on two separate nights. Loop gain is measured from routine polysomnography using a novel mathematical method. A value of loop gain \>1 reflects unstable breathing, and a value less than but approaching 1 denotes a system more prone to oscillate. Loop gain is measured as the changes in ventilatory drive/effort that arises subsequent to changes in ventilation (e.g. due to obstructive apnea). A simple chemoreflex model (gain, time constant, delay) is fit to surrogate ventilation data (derived from airflow) during sleep. The best model is one that best matches the elevated ventilatory drive (measured as ventilation in the absence of airflow obstruction) based on the prior apneic/hypopneic fall in ventilation. Loop gain is calculated from this model. We aim to use loop gain measured on and off oxygen to determine whether a strong response (reduction in apnea severity) can be predicted by a higher loop gain (in the sham arm) using our method. We also assessed whether assessing upper airway anatomy/collapsibility, dilator muscle responsiveness, and the arousal threshold helped to predict responses to treatment.
Interventions
DRUGInspired oxygen (40%)
Supplemental oxygen at 40% inspired via venturi mask (15 L/min). Equivalent to 5 L/min via nasal cannula.
OTHERSham
Medical air with 21% oxygen via venturi mask (15 L/min).
Sponsors
National Heart, Lung, and Blood Institute (NHLBI)
American Heart Association
Brigham and Women's Hospital
Study design
Allocation
RANDOMIZED
Intervention model
CROSSOVER
Primary purpose
TREATMENT
Masking
SINGLE (Subject)
Masking description
Participants were not informed whether they were breathing supplemental inspired oxygen or sham (air) on the study night. Technicians and investigators performing the study were not masked in order to carefully ensure delivery of treatment.
Eligibility
Sex/Gender
ALL
Age
20 Years to 79 Years
Healthy volunteers
No
Inclusion criteria
* Apnea/hypopnea index \>20 events per hour * Age 20-79 years
Exclusion criteria
* COPD with desaturation (resting SpO2\<96%) * Use of respiratory stimulants or depressants * Pregnancy
Design outcomes
Primary
| Measure | Time frame | Description |
|---|---|---|
| Apnea-hypopnea Index | 1 night | Apnea-hypopnea index (AHI) will be compared between oxygen and sham nights. Hypopneas are based on 30% reduction in airflow (no desaturation or arousal criteria). AHI data are exclusive to non-REM supine sleep. The results presented here are for the AHI at each intervention (per intervention) regardless of the sequence (preferred clinicaltrials.gov format). Please note, however, that the a priori outcome measure was the reduction in AHI with oxygen as a percent of sham values, i.e. (AHI on sham - AHI on oxygen)/(AHI on sham) % (a comparison with greater statistical power), compared between patient subgroups (see Statistical Analysis section). Subgroups were defined a priori as higher (\>=0.7) versus lower loop gain (\<0.7), but tests were also performed in subgroups defined by favorable versus unfavorable pathophysiology. |
Secondary
| Measure | Time frame | Description |
|---|---|---|
| Frequency of EEG Arousals (Events Per Hour) | 1 night | Frequency of scored EEG arousals per hour of non-REM sleep. Note: Our objective was to describe changes in secondary outcomes within phenotypic subgroups. Overall effects (unselected patients / ignoring phenotypic subgroups) are first presented below, followed by effects in favorable vs. unfavorable subgroups. |
| Overnight Change in Systolic Blood Pressure | 1 night | The change in systolic blood pressure overnight. Two supine oscillometric measurements of blood pressure are made: just prior to lights out (evening), and after lights on (morning). |
| Overnight Change in Diastolic Blood Pressure | 1 night | The change in diastolic blood pressure overnight. Two supine oscillometric measurements of blood pressure are made: just prior to lights out (evening), and after lights on (morning). |
| Subjective Sleep Quality (Oxygen vs Sham) | 1 night | Better(+1)/Same(0)/Worse(-1) on oxygen vs sham, i.e. a relative comparison between arms. When subjects had completed the entire study, they were asked to compare subjectively their sleep quality on the first versus second study. |
| Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | 1 night | Assessed in the morning after the single night of treatment. Minimum score: 1 (alert), maximum score: 7 (not alert). |
Countries
United States
Participant flow
Participants by arm
| Arm | Count |
|---|---|
| Oxygen First Participants participate in inspired oxygen arm (40% oxygen) first, then the air night (21% oxygen) second
Inspired oxygen (40%): Supplemental oxygen at approximately 40% e.g. via Pink venturi mask | 22 |
| Air First Participants participate in the air night (sham) first, then the oxygen night second.
Sham: Medical air with 21% oxygen e.g. via Pink venturi mask | 25 |
| Total | 47 |
Baseline characteristics
| Characteristic | Oxygen First | Air First | Total |
|---|---|---|---|
| Age, Continuous | 56 years STANDARD_DEVIATION 13 | 49 years STANDARD_DEVIATION 22 | 52 years STANDARD_DEVIATION 19 |
| Sex: Female, Male Female | 5 Participants | 10 Participants | 15 Participants |
| Sex: Female, Male Male | 17 Participants | 15 Participants | 32 Participants |
Adverse events
| Event type | EG000 affected / at risk | EG001 affected / at risk |
|---|---|---|
| deaths Total, all-cause mortality | 0 / 46 | 0 / 45 |
| other Total, other adverse events | 0 / 46 | 0 / 45 |
| serious Total, serious adverse events | 0 / 46 | 0 / 45 |
Outcome results
Primary
Apnea-hypopnea Index
Apnea-hypopnea index (AHI) will be compared between oxygen and sham nights. Hypopneas are based on 30% reduction in airflow (no desaturation or arousal criteria). AHI data are exclusive to non-REM supine sleep. The results presented here are for the AHI at each intervention (per intervention) regardless of the sequence (preferred clinicaltrials.gov format). Please note, however, that the a priori outcome measure was the reduction in AHI with oxygen as a percent of sham values, i.e. (AHI on sham - AHI on oxygen)/(AHI on sham) % (a comparison with greater statistical power), compared between patient subgroups (see Statistical Analysis section). Subgroups were defined a priori as higher (\>=0.7) versus lower loop gain (\<0.7), but tests were also performed in subgroups defined by favorable versus unfavorable pathophysiology.
Time frame: 1 night
Population: Patients with AHI\>20
| Arm | Measure | Group | Value (MEAN) | Dispersion |
|---|---|---|---|---|
| Inspired Oxygen | Apnea-hypopnea Index | Patients with high loop gain (LG1) | 48.5 events/hour | Standard Error 6.4 |
| Inspired Oxygen | Apnea-hypopnea Index | Patients with favorable pathophysiology | 22.5 events/hour | Standard Error 4.3 |
| Inspired Oxygen | Apnea-hypopnea Index | Patients with low loop gain (LG1) | 35.9 events/hour | Standard Error 4.6 |
| Inspired Oxygen | Apnea-hypopnea Index | Patients with unfavorable pathophysiology | 51.9 events/hour | Standard Error 4.1 |
| Inspired Oxygen | Apnea-hypopnea Index | All patients | 40.5 events/hour | Standard Error 3.8 |
| Sham | Apnea-hypopnea Index | Patients with unfavorable pathophysiology | 63.2 events/hour | Standard Error 5 |
| Sham | Apnea-hypopnea Index | All patients | 57.9 events/hour | Standard Error 3.7 |
| Sham | Apnea-hypopnea Index | Patients with high loop gain (LG1) | 74.3 events/hour | Standard Error 6.1 |
| Sham | Apnea-hypopnea Index | Patients with low loop gain (LG1) | 48.5 events/hour | Standard Error 3.4 |
| Sham | Apnea-hypopnea Index | Patients with favorable pathophysiology | 49.5 events/hour | Standard Error 4.8 |
Comparison: Primary statistical comparison was the percent reduction in AHI----\[AHI(sham)-AHI(oxygen)\]/AHI(sham) %----between two phenotypic patient subgroups. Patient subgroups were defined by the loop gain (LG1) measured on the sham night as high or low (a priori cutoff LG1=0.7).p-value: 0.495% CI: [-16, 37.1]t-test, 2 sided
Comparison: Here we aimed to confirm that there was a difference between AHI on oxygen vs sham (overall, i.e. in unselected patients). This test, however was not part of our primary objective.p-value: <0.00195% CI: [10.7, 24.1]t-test, 2 sided
Comparison: The primary goal of the study was to identify a phenotypic subgroup of patients with sleep apnea that responds preferentially to oxygen (percent reduction in AHI----\[AHI(sham)-AHI(oxygen)\]/AHI(sham) %). We defined patient subgroups (favorable versus unfavorable) based on the four key phenotypic traits (loop gain, collapsibility, arousal threshold, muscle responses) measured on the sham night, with the use of multiple logistic regression and leave-one-out cross validation.p-value: 0.00195% CI: [21, 64.6]t-test, 2 sided
Secondary
Frequency of EEG Arousals (Events Per Hour)
Frequency of scored EEG arousals per hour of non-REM sleep. Note: Our objective was to describe changes in secondary outcomes within phenotypic subgroups. Overall effects (unselected patients / ignoring phenotypic subgroups) are first presented below, followed by effects in favorable vs. unfavorable subgroups.
Time frame: 1 night
Population: Patients with AHI\>20
| Arm | Measure | Group | Value (MEAN) | Dispersion |
|---|---|---|---|---|
| Inspired Oxygen | Frequency of EEG Arousals (Events Per Hour) | All patients | 35.9 events/hr | Standard Error 3.4 |
| Inspired Oxygen | Frequency of EEG Arousals (Events Per Hour) | Patients with favorable pathophysiology | 20.8 events/hr | Standard Error 3 |
| Inspired Oxygen | Frequency of EEG Arousals (Events Per Hour) | Patients with unfavorable pathophysiology | 45.4 events/hr | Standard Error 4.1 |
| Sham | Frequency of EEG Arousals (Events Per Hour) | All patients | 50.3 events/hr | Standard Error 3.7 |
| Sham | Frequency of EEG Arousals (Events Per Hour) | Patients with favorable pathophysiology | 39.7 events/hr | Standard Error 5.1 |
| Sham | Frequency of EEG Arousals (Events Per Hour) | Patients with unfavorable pathophysiology | 57.0 events/hr | Standard Error 4.6 |
Secondary
Overnight Change in Diastolic Blood Pressure
The change in diastolic blood pressure overnight. Two supine oscillometric measurements of blood pressure are made: just prior to lights out (evening), and after lights on (morning).
Time frame: 1 night
Population: Patients with AHI\>20
| Arm | Measure | Group | Value (MEAN) | Dispersion |
|---|---|---|---|---|
| Inspired Oxygen | Overnight Change in Diastolic Blood Pressure | All patients | 0.9 mmHg | Standard Error 0.9 |
| Inspired Oxygen | Overnight Change in Diastolic Blood Pressure | Patients with favorable pathophysiology | -0.7 mmHg | Standard Error 1.2 |
| Inspired Oxygen | Overnight Change in Diastolic Blood Pressure | Patients with unfavorable pathophysiology | 2.0 mmHg | Standard Error 1.2 |
| Sham | Overnight Change in Diastolic Blood Pressure | All patients | 4.1 mmHg | Standard Error 1.5 |
| Sham | Overnight Change in Diastolic Blood Pressure | Patients with favorable pathophysiology | 6.4 mmHg | Standard Error 2.3 |
| Sham | Overnight Change in Diastolic Blood Pressure | Patients with unfavorable pathophysiology | 2.5 mmHg | Standard Error 2 |
Secondary
Overnight Change in Systolic Blood Pressure
The change in systolic blood pressure overnight. Two supine oscillometric measurements of blood pressure are made: just prior to lights out (evening), and after lights on (morning).
Time frame: 1 night
Population: Patients with AHI\>20
| Arm | Measure | Group | Value (MEAN) | Dispersion |
|---|---|---|---|---|
| Inspired Oxygen | Overnight Change in Systolic Blood Pressure | All patients | -0.8 mmHg | Standard Error 1.4 |
| Inspired Oxygen | Overnight Change in Systolic Blood Pressure | Patients with favorable pathophysiology | -2.8 mmHg | Standard Error 1.9 |
| Inspired Oxygen | Overnight Change in Systolic Blood Pressure | Patients with unfavorable pathophysiology | 0.5 mmHg | Standard Error 1.9 |
| Sham | Overnight Change in Systolic Blood Pressure | All patients | 3.0 mmHg | Standard Error 1.9 |
| Sham | Overnight Change in Systolic Blood Pressure | Patients with favorable pathophysiology | 3.4 mmHg | Standard Error 2.9 |
| Sham | Overnight Change in Systolic Blood Pressure | Patients with unfavorable pathophysiology | 2.8 mmHg | Standard Error 2.5 |
Secondary
Subjective Sleepiness/Alertness (Stanford Sleepiness Scale)
Assessed in the morning after the single night of treatment. Minimum score: 1 (alert), maximum score: 7 (not alert).
Time frame: 1 night
Population: Patients with AHI\>20
| Arm | Measure | Group | Value (MEAN) | Dispersion |
|---|---|---|---|---|
| Inspired Oxygen | Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | All patients | 2.13 Scores on a scale | Standard Error 0.19 |
| Inspired Oxygen | Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | Patients with favorable pathophysiology | 2.14 Scores on a scale | Standard Error 0.32 |
| Inspired Oxygen | Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | Patients with unfavorable pathophysiology | 2.13 Scores on a scale | Standard Error 0.25 |
| Sham | Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | All patients | 2.03 Scores on a scale | Standard Error 0.18 |
| Sham | Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | Patients with favorable pathophysiology | 2.11 Scores on a scale | Standard Error 0.26 |
| Sham | Subjective Sleepiness/Alertness (Stanford Sleepiness Scale) | Patients with unfavorable pathophysiology | 1.99 Scores on a scale | Standard Error 0.25 |
Secondary
Subjective Sleep Quality (Oxygen vs Sham)
Better(+1)/Same(0)/Worse(-1) on oxygen vs sham, i.e. a relative comparison between arms. When subjects had completed the entire study, they were asked to compare subjectively their sleep quality on the first versus second study.
Time frame: 1 night
Population: Patients with AHI\>20. One patient, whose treatment and sham nights were separated by an extended duration (due to scheduling), did not provide data for this outcome measure.
| Arm | Measure | Group | Value (MEAN) | Dispersion |
|---|---|---|---|---|
| Inspired Oxygen | Subjective Sleep Quality (Oxygen vs Sham) | All patients | 0.34 units on a scale | Standard Deviation 0.13 |
| Inspired Oxygen | Subjective Sleep Quality (Oxygen vs Sham) | Patients with favorable pathophysiology | 0.69 units on a scale | Standard Deviation 0.17 |
| Inspired Oxygen | Subjective Sleep Quality (Oxygen vs Sham) | Patients with unfavorable pathophysiology | 0.14 units on a scale | Standard Deviation 0.18 |