Sleep architecture and serum biomarker heterogeneity in obstructive sleep apnea: a cross-sectional study.

STUDY

OBJECTIVES

To determine whether sleep architecture is associated with heterogeneity in serum biomarkers of neuronal injury, astroglial stress, tau-related pathology, and neuroplasticity in adults with obstructive sleep apnea (OSA).

METHODS

Adults referred for suspected OSA underwent overnight polysomnography and next-morning fasting serum sampling. Primary analyses were restricted to participants with OSA (apnea-hypopnea index [AHI] ≥5 events/h; n = 82). Serum neuron-specific enolase (NSE), neurofilament light chain (NfL), brain-derived neurotrophic factor (BDNF), cAMP response element-binding protein 1 (CREB1), S100 calcium-binding protein B (S100B), glial fibrillary acidic protein (GFAP), and (phosphorylated microtubule-associated protein tau) p-Tau/human pMAPT/pTAU were measured by enzyme-linked immunosorbent assay (ELISA). We first examined whole-group correlations between respiratory burden indices (AHI, ODI, and T90) and biomarkers, then fit age- and sex-adjusted linear models with continuous sleep-stage percentages. Median-split stage analyses and stage-stratified correlations were treated as exploratory sensitivity analyses.

RESULTS

Biomarker availability ranged from 70 to 71 participants. Lower rapid eye movement (REM) percentage was independently associated with higher ln(NSE) (beta = -0.0138 per 1 per cent REM; 95% CI = -0.026 to -0.001; p = .033). Whole-group BDNF correlated inversely with ODI (rho = -0.299, p = .011) and AHI (rho = -0.277, p = .020). In sensitivity analyses, REM-AHI was positively associated with p-Tau/human pMAPT/pTAU (beta = 1.799, p = .006) and GFAP (beta = 2.094, p = .024). Interaction terms testing formal effect modification by REM per cent or N3 per cent were directionally consistent with the hypothesis but did not reach conventional significance. After Bonferroni correction, only the short-REM AHI-BDNF correlation remained significant among exploratory stage-stratified correlations.

CONCLUSIONS

Reduced REM sleep showed the clearest and most reproducible relationship with an adverse peripheral biomarker profile in OSA, specifically higher serum NSE. Sleep architecture may contribute to biomarker heterogeneity, but evidence for stage-dependent modification of hypoxemia-biomarker relationships remains exploratory.