Cerebrovascular pathology mediates associations between hypoxemia during rapid eye movement sleep and medial temporal lobe structure and function in older adults

Abstract

Obstructive sleep apnea (OSA) is common in older adults and is associated with medial temporal lobe (MTL) degeneration and memory decline in aging and Alzheimer's disease (AD). However, the underlying mechanisms linking OSA to MTL degeneration and impaired memory remains unclear. By combining magnetic resonance imaging (MRI) assessments of cerebrovascular pathology and MTL structure with clinical polysomnography and assessment of overnight emotional memory retention in older adults at risk for AD, cerebrovascular pathology in fronto-parietal brain regions was shown to statistically mediate the relationship between OSA-related hypoxemia, particularly during rapid eye movement (REM) sleep, and entorhinal cortical thickness. Reduced entorhinal cortical thickness was, in turn, associated with impaired overnight retention in mnemonic discrimination ability across emotional valences for high similarity lures. These findings identify cerebrovascular pathology as a contributing mechanism linking hypoxemia to MTL degeneration and impaired sleep-dependent memory in older adults.

Figure 1:. Regional specificity of WMH associations with REM-related hypoxemia.

Associations between hypoxemia severity during REM sleep and MRI-measured white matter hyperintensity (WMH) volume in frontal (red), parietal (magenta), temporal (blue), and occipital (orange) cortex. Sagittal slices depicting the region of interest used to assess regional WMH volumes are inset within each respective plot. Hypoxemia severity was computed using a square-root transform of the minimum SpO₂ values during REM sleep (sqrt(max(values + 1) - values) due to the negative skew in the across subjects (see Methods for details).

Figure 2:. Frontal white matter hyperintensity burden and medial temporal lobe structure and function.

Frontal WMH burden associations with bilateral hippocampal volume and bilateral entorhinal cortical thickess, shown in the plots on the left column. On the right, associations between bilateral hippocampal volume and bilateral entorhinal cortical thickess with overnight change in the lure discrimination index (LDI) are shown. LDI was calculated using: p(‘New’|Lure) - p(‘New’|Target). Overnight change was calculated as: overnight change in LDI (delayed test-immediate test/immediate test).

Figure 3:. Frontal white matter hyperintensity burden indirectly mediates the relationship between REM-related hypoxia and left entorhinal thickness.

A schematic depicting the medial model in which REM-related hypoxemia severity was the independent variable (X), frontal WMH burden was the mediator variable (M), and left ERC thickness was the dependent variable (Y). No direct effect was observed (c’=−0.046, p=0.35). A significant indirect effect was observed (indirect=−0.043, 95% CI −0.1174 −0.00015), with the indirect effect accounting for 77.3% of the total effect. REM-related hypoxemia was positively associated with frontal WMH burden (a=0.101, p=0.042), which in turn was negatively associated with left entorhinal thickness (b=−0.430, p=0.015).