Clinical predictors of the respiratory arousal threshold in patients with obstructive sleep apnea

Abstract

Rationale: A low respiratory arousal threshold (ArTH) is one of several traits involved in obstructive sleep apnea pathogenesis and may be a therapeutic target; however, there is no simple way to identify patients without invasive measurements.

Objectives: To determine the physiologic determinates of the ArTH and develop a clinical tool that can identify patients with low ArTH.

Methods: Anthropometric data were collected in 146 participants who underwent overnight polysomnography with an epiglottic catheter to measure the ArTH (nadir epiglottic pressure before arousal). The ArTH was measured from up to 20 non-REM and REM respiratory events selected randomly. Multiple linear regression was used to determine the independent predictors of the ArTH. Logistic regression was used to develop a clinical scoring system.

Measurements and main results: Nadir oxygen saturation as measured by pulse oximetry, apnea-hypopnea index, and the fraction of events that were hypopneas (Fhypopneas) were independent predictors of the ArTH (r(2) = 0.59; P < 0.001). Using this information, we used receiver operating characteristic analysis and logistic regression to develop a clinical score to predict a low ArTH, which allocated a score of 1 to each criterion that was satisfied: (apnea-hypopnea index, <30 events per hour) + (nadir oxygen saturation as measured by pulse oximetry >82.5%) + (Fhypopneas >58.3%). A score of 2 or above correctly predicted a low arousal threshold in 84.1% of participants with a sensitivity of 80.4% and a specificity of 88.0%, a finding that was confirmed using leave-one-out cross-validation analysis.

Conclusions: Our results demonstrate that individuals with a low ArTH can be identified from standard, clinically available variables. This finding could facilitate larger interventional studies targeting the ArTH.

Keywords: arousal threshold; lung; phenotype traits; respiratory-induced arousals; sleep apnea.

Figure 1.

Determining the arousal threshold during an in-laboratory polysomnogram. (A) Representative example taken from an overnight polysomnogram recording in a male patient with severe obstructive sleep apnea (apnea-hypopnea index, 56 events per hour) to demonstrate how the arousal threshold was calculated. Note the progressive increase in the negative pressure swings on the epiglottic pressure trace (Pepi) during periods of obstruction indicating an increased effort to breathe. (B) The respiratory arousal threshold was taken as the delta pressure swing (−16.7 cm H₂O in this example) immediately before arousal (gray shading). C3-A2 and O2-A1 are electroencephalograms.

Figure 2.

Proportions of individuals with a low arousal threshold. (A) Most control subjects (86.4%, gray bar) had a low respiratory arousal threshold (defined as a peak epiglottic pressure of −15 cm H₂O or less) compared with the 50.5% of patients with obstructive sleep apnea (OSA) (black bar). (B) When dividing up the patients with OSA according to apnea severity, the proportions of patients with a low arousal threshold decreased as severity increased; mild = 88% (light gray bar), moderate = 72.7% (dark gray bar), and severe = 23.1% (black bar). *Significantly reduced compared with control subjects (chi-square test, P = 0.005). ^†Significant difference compared with control subjects (multiple logistic regression, P < 0.001).

Figure 3.

Univariate associations between age, body mass index (BMI), and Epworth Sleepiness Scale (ESS) and the arousal threshold. (A) Age was not related to the arousal threshold (r² = 0.002; P = NS), whereas both (B) BMI (r² = 0.103; P = 0.001) and (C) ESS (r² = 0.078; P = 0.005) were weakly correlated with the arousal threshold. Similar correlations were seen whether the non-REM arousal threshold (age [r² = 0.001; P = 0.72], BMI [r² = 0.08; P = 0.005], ESS [r² = 0.09; P = 0.003]) or the REM arousal threshold (age [r² = 0.017; P = 0.76], BMI [r² = 0.08; P = 0.02], ESS [r² = 0.09; P = 0.013]) were used.

Figure 4.

Clinical determinants of the overall arousal threshold. (A) Apnea–hypopnea index (AHI) (r² = 0.37; P < 0.0001), (B) nadir oxygen saturation (r² = 0.43; P < 0.0001), (C) the arousal index (r² = 0.29; P < 0.001), and (D) percentage of respiratory events that were hypopneas (r² = 0.35; P < 0.001) were the strongest variables from the clinical polysomnogram report that were correlated with the overall arousal threshold. Similar correlations were seen whether the non-REM arousal threshold (AHI [r² = 0.36; P < 0.001], nadir oxygen saturation [r² = 0.42; P < 0.001], arousal index [r² = 0.26; P < 0.001], % hypopneas [r² = 0.35; P < 0.001]) or the REM arousal threshold (AHI [r² = 0.45; P < 0.001], nadir oxygen saturation [r² = 0.43; P < 0.001], arousal index [r² = 0.37; P < 0.001], % hypopneas [r² = 0.35; P < 0.001]) were used. Sp_O2 = oxygen saturation as measured by pulse oximetry.