Altered fornix integrity is associated with sleep apnea-related hypoxemia in mild cognitive impairment

Abstract

Introduction: The limbic system is critical for memory function and degenerates early in the Alzheimer's disease continuum. Whether obstructive sleep apnea (OSA) is associated with alterations in the limbic white matter tracts remains understudied.

Methods: Polysomnography, neurocognitive assessment, and brain magnetic resonance imaging (MRI) were performed in 126 individuals aged 55-86 years, including 70 cognitively unimpaired participants and 56 participants with mild cognitive impairment (MCI). OSA measures of interest were the apnea-hypopnea index and composite variables of sleep fragmentation and hypoxemia. Microstructural properties of the cingulum, fornix, and uncinate fasciculus were estimated using free water-corrected diffusion tensor imaging.

Results: Higher levels of OSA-related hypoxemia were associated with higher left fornix diffusivities only in participants with MCI. Microstructure of the other white matter tracts was not associated with OSA measures. Higher left fornix diffusivities correlated with poorer episodic verbal memory.

Discussion: OSA may contribute to fornix damage and memory dysfunction in MCI.

Highlights: Sleep apnea-related hypoxemia was associated with altered fornix integrity in MCI. Altered fornix integrity correlated with poorer memory function. Sleep apnea may contribute to fornix damage and memory dysfunction in MCI.

Keywords: Alzheimer's disease; dementia; fornix; free water‐diffusion tensor imaging; hypoxia; limbic system; memory; mild cognitive impairment; sleep; sleep apnea; white matter.

FIGURE 1

Overview of the methods. (A) Polysomnographic variables related to obstructive sleep apnea (OSA) were included in a principal component analysis (PCA) to extract the composite variables PCA_sleep fragmentation and PCA_hypoxemia, while the apnea‐hypopnea index was analyzed as a single variable. (B) The neurocognitive assessment allowed us to dichotomize the sample into cognitively unimpaired (CU) or mild cognitive impairment (MCI), and episodic verbal memory was tested with the Rey Auditory Verbal Learning Test (RAVLT). (C) We performed whole brain tractography, extracted the limbic white matter tracts of interest in each hemisphere, and calculated averaged free water (FW)‐corrected diffusion tensor imaging (DTI) measures for each tract. (D) The primary analyses tested for associations between OSA measures and FW‐DTI measures in the whole sample (if there was no significant interaction with cognitive status) or in the CU group and the MCI group separately (if there was a significant interaction with cognitive status). The images in (A) and (B) were adapted from BioRender.com.

FIGURE 2

Associations between OSA‐related hypoxemia and left fornix microstructure. (A–E) Scatter plots showing the associations between PCA_hypoxemia and FW‐DTI measures of the left fornix. The y‐axis shows the predicted values from the regression analyses adjusted for age (continuous), sex (men vs. women), body mass index (continuous), Epworth Sleepiness Scale (continuous), Vascular Burden Index (<2 vs. ≥2 points), left fornix volume (divided by total intracranial volume; continuous), and cognitive status (CU vs. MCI). The solid line is the regression line and the gray area represents the 95% confidence interval. (F) Changes in the diffusion ellipsoid shape of the tissue compartment (gray ellipsoid) and the extracellular/FW compartment (blue sphere) associated with higher levels of hypoxemia in participants with MCI. Our results indicate that hypoxemia is associated with an increase in the three diffusivities (MD_T, AxD_T, and RD_T), a constant FA_T (which is consistent with a proportional increase in λ1, λ2, and λ3), and a constant FW index (which is consistent with a proportional increase in the tissue compartment and the FW/extracellular comportment). Note that the changes in the shape of the diffusion ellipsoid are shown here for illustrative and conceptual purposes (i.e., these changes are conceptually consistent with our results, but the magnitudes of the changes are not scaled or calculated based on our data). ^* p < 0.01. AxD_T, tissue axial diffusivity; CU, cognitively unimpaired; DTI, diffusion tensor imaging; FW, free water; MCI, mild cognitive impairment; MD_T, tissue mean diffusivity; OSA, obstructive sleep apnea; PCA, principal component analysis; RD_T, tissue radial diffusivity; λ, eigenvalue.

FIGURE 3

Associations between OSA‐related hypoxemia and left fornix microstructure along its rostro‐caudal sections in participants with MCI. (A) Sagittal view of the left fornix showing the five rostro‐caudal sections. (B–F) Associations between PCA_hypoxemia and FW‐DTI measures of the left fornix along the rostro‐caudal sections in participants with MCI, adjusted for age (continuous), sex (men vs. women), body mass index (continuous), Epworth Sleepiness Scale (continuous), Vascular Burden Index (<2 vs. ≥2 points), and left fornix volume (divided by total intracranial volume; continuous). The central line represents the standardized beta coefficient (β), while the upper and lower lines represent the 95% confidence interval (CI). ^* p < 0.05. AxD_T, tissue axial diffusivity; DTI, diffusion tensor imaging; FA_T, tissue fractional anisotropy; FW, free water; MCI, mild cognitive impairment; MD_T, tissue mean diffusivity; OSA, obstructive sleep apnea; PCA, principal component analysis; RD_T, tissue radial diffusivity.