Sleep Science and Practice (May 2017)

Distinct polysomnographic and ECG-spectrographic phenotypes embedded within obstructive sleep apnea

  • Robert Joseph Thomas,
  • Chol Shin,
  • Matt Travis Bianchi,
  • Clete Kushida,
  • Chang-Ho Yun

DOI
https://doi.org/10.1186/s41606-017-0012-9
Journal volume & issue
Vol. 1, no. 1
pp. 1 – 13

Abstract

Read online

Abstract Background The primary metric extracted from the polysomnogram in patients with sleep apnea is the apnea-hypopnea index (or respiratory disturbance index) and its derivatives. Other phenomena of possible importance such as periods of stable breathing, features suggestive of high respiratory control loop gain, and sleep fragmentation phenotypes are not commonly generated in clinical practice or research. A broader phenotype designation can provide insights into biological processes, and possibly clinical therapy outcome effects. Methods The dataset used for this study was the archived baseline diagnostic polysomnograms from the Apnea Positive Pressure Long-term Efficacy Study (APPLES). The electrocardiogram (ECG)-derived cardiopulmonary coupling sleep spectrogram was computed from the polysomnogram. Sleep fragmentation phenotypes used thresholds of sleep efficiency (SE) ≤ 70%, non-rapid eye movement (NREM) sleep N1 ≥ 30%, wake after sleep onset (WASO) ≥ 60 min, and high frequency coupling (HFC) on the ECG-spectrogram ≤ 30%. Sleep consolidation phenotypes used thresholds of SE ≥ 90%, WASO ≤ 30 min, HFC ≥ 50% and N1 ≤ 10%. Multiple and logistic regression analysis explored cross-sectional associations with covariates and across phenotype categories. NREM vs. REM dominant apnea categories were identified when the NREM divided by REM respiratory disturbance index (RDI) was > 1. Results The data was binned first into mild, moderate, severe and extreme categories based on the respiratory disturbance index of < 10, 10–30, 30–60, and greater than 60, per hour of sleep. Using these criteria, 70, 394, 320 and 188 for polysomnogram, and 54, 296, 209 and 112 subjects for ECG-spectrogram analysis groups. All phenotypes were seen at all severity levels. There was a higher correlation of NREM-RDI with the amount of ECG-spectrogram narrow band coupling, vs. REM-RDI, 0.41 vs 0.14, respectively. NREM dominance was associated with male gender and higher mixed/central apnea indices. Absence of the ECG-spectrogram sleep consolidated phenotype was associated with an increased odds of being on antihypertensive medications, OR 2.65 [CI: 1.64–4.26], p = < 0.001. Conclusions Distinct phenotypes are readily seen at all severities of sleep apnea, and can be identified from conventional polysomnography. The ECG-spectrogram analysis provides further phenotypic differentiation.

Keywords