Hypoxic burden predicts weaker coordination between brain pulsations and CSF flow independent of non-hypoxic arousals: Implications for glymphatic activity

Objective: To examine whether hypoxic burden is distinctly associated with key drivers of glymphatic activity, independent of sleep fragmentation. Background: OSA is a robust risk factor for multiple neurodegenerative conditions, including Alzheimer's disease, with glymphatic impairment representing a potential mechanistic pathway. However, it remains unknown whether distinct pathological features of OSA, including hypoxic burden and sleep fragmentation, are differentially associated with the two key physiological drivers of glymphatic activity: (1) coordination between brain pulsations and CSF flow and (2) brain pulsation strength. Method: Twenty-eight individuals with newly identified OSA and eight healthy individuals without OSA completed either in-lab polysomnography or WatchPAT, providing estimates of hypoxic burden, quantified as time spent below 90% oxygen saturation (T90), and sleep fragmentation, quantified as non-hypoxic respiratory effort-related arousals (RERAs). Participants also completed 7T resting-state fMRI to quantify the coordination between brain pulsations and CSF flow using gBOLD-CSF coupling (i.e., cross-correlation between g ray matter b lood o xygen level d ependent signal pulsations and CSF inflow) and brain pulsation strength using gBOLD amplitude. Hierarchical regression and Pearson correlations were used to examine associations between sleep measures and fMRI-derived metrics. Results: =0.008). Contrary to expectations, greater T90 was associated with higher gBOLD amplitude across the temporal lobe and multiple frontal and parietal regions. Within regions showing T90-linked elevations in gBOLD amplitude, higher gBOLD amplitude was not associated with stronger region-specific gBOLD-CSF coupling. This contrasted with regions not associated with T90, where higher gBOLD amplitude was associated with stronger gBOLD-CSF coupling (difference in β=0.0006, p<0.001). RERAs were not associated with gBOLD-CSF coupling and gBOLD amplitude throughout the cerebral cortex. Conclusions: In OSA, hypoxic burden, rather than respiratory effort-related sleep fragmentation, may be the primary pathological feature associated with impaired brain pulsation and CSF dynamics, both of which are key drivers of glymphatic activity. These alterations may be most prominent in the temporal lobe, potentially reflecting its elevated metabolic demand and vulnerability to hypoxemia.