Impaired glymphatic drainage underlying obstructive sleep apnea is associated with cognitive dysfunction

Abstract

Obstructive sleep apnea (OSA) is highly prevalent but easily undiagnosed and is an independent risk factor for cognitive impairment. However, it remains unclear how OSA is linked to cognitive impairment. In the present study, we found the correlation between morphological changes of perivascular spaces (PVSs) and cognitive impairment in OSA patients. Moreover, we developed a novel set of dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) methods to evaluate the fluid dynamics of glymphatic drainage system. We found that the inflow and outflow parameters of the glymphatic drainage system in patients with OSA were obviously changed, indicating impairment of glymphatic drainage due to excessive perfusion accompanied with deficient drainage in OSA patients. Moreover, parameters of the outflow were associated with the degree of cognitive impairment, as well as the hypoxia level. In addition, continuous positive airway pressure (CPAP) enhances performance of the glymphatic drainage system after 1 month treatment in OSA patients. We proposed that ventilation improvement might be a new strategy to ameliorate the impaired drainage of glymphatic drainage system due to OSA-induced chronic intermittent hypoxia, and consequently improved the cognitive decline.

Keywords: Cognitive impairment; Dynamic contrast-enhanced magnetic resonance imaging; Glymphatic; Obstructive sleep apnea.

Fig. 1

CONSORT diagram. Consolidated Standards of Reporting Trials flow diagram showing study participants screening, eligibility and inclusion

Fig. 2

Measurement of the PVSs and ventricles areas by MRI. (A) Representative T2 axial MRI scans of PVSs in the bilateral frontal cortex and the bilateral basal ganglia of the NCs. The irregular regions within the closed yellow contour represent the frontal cortex or bilateral basal ganglia. The irregular red regions represent the PVSs. Representative T2 axial MRI scans of the bilateral lateral ventricle of the NCs. The irregular blue regions represent the bilateral lateral ventricles. The irregular regions within the closed yellow contour represent the total brain area at the level of the lateral ventricle. Representative images of the fourth ventricle of the NCs. The irregular blue regions represent the fourth ventricle. The irregular regions within the closed yellow contour represent the total cerebellum area at the level of the fourth ventricle. Scale bar, 20 mm. (B) Representative T2 axial MRI scans of PVSs in the bilateral frontal cortex and the bilateral basal ganglia, and the bilateral lateral ventricle and the fourth ventricle of the OSA. Scale bar, 20 mm. (C) Comparison of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles between the NCs (n = 28) and OSA (n = 31) groups (Mann–Whitney U-test). (D) ROC curves of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles for distinguishing OSA (n = 31) from NCs (n = 28). (E) Comparison of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles between the mild-moderate OSA (n = 17) and severe OSA (n = 11) groups (Mann–Whitney U-test). (F) ROC curves of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles for distinguishing severe OSA (n = 11) from mild-moderate OSA (n = 17). (G) Comparison of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles between the OSA with mild hypoxemia (n = 18) and OSA with severe hypoxemia (n = 10) groups (Mann–Whitney U-test). (H) ROC curves of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles for distinguishing OSA with severe hypoxemia (n = 10) from OSA with mild hypoxemia (n = 18)

Fig. 2

Measurement of the PVSs and ventricles areas by MRI. (A) Representative T2 axial MRI scans of PVSs in the bilateral frontal cortex and the bilateral basal ganglia of the NCs. The irregular regions within the closed yellow contour represent the frontal cortex or bilateral basal ganglia. The irregular red regions represent the PVSs. Representative T2 axial MRI scans of the bilateral lateral ventricle of the NCs. The irregular blue regions represent the bilateral lateral ventricles. The irregular regions within the closed yellow contour represent the total brain area at the level of the lateral ventricle. Representative images of the fourth ventricle of the NCs. The irregular blue regions represent the fourth ventricle. The irregular regions within the closed yellow contour represent the total cerebellum area at the level of the fourth ventricle. Scale bar, 20 mm. (B) Representative T2 axial MRI scans of PVSs in the bilateral frontal cortex and the bilateral basal ganglia, and the bilateral lateral ventricle and the fourth ventricle of the OSA. Scale bar, 20 mm. (C) Comparison of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles between the NCs (n = 28) and OSA (n = 31) groups (Mann–Whitney U-test). (D) ROC curves of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles for distinguishing OSA (n = 31) from NCs (n = 28). (E) Comparison of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles between the mild-moderate OSA (n = 17) and severe OSA (n = 11) groups (Mann–Whitney U-test). (F) ROC curves of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles for distinguishing severe OSA (n = 11) from mild-moderate OSA (n = 17). (G) Comparison of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles between the OSA with mild hypoxemia (n = 18) and OSA with severe hypoxemia (n = 10) groups (Mann–Whitney U-test). (H) ROC curves of the relative area ratios of the bilateral frontal cortex, basal ganglia, lateral ventricle and the fourth ventricles for distinguishing OSA with severe hypoxemia (n = 10) from OSA with mild hypoxemia (n = 18)

Fig. 3

Measurement of fluid flow of PVSs by DCE-MRI. (A) K-means cluster outcome of all CTCs in the bilateral frontal cortex of the NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. The red drops represent type I CTCs. The blue triangles represent type II CTCs. (B) Average CTCs based on the cluster analysis of totaling 489 CTCs from the NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. (C) The proportions of the type I and type II CTCs in NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. (D) The numbers of type I CTCs for each subject of NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. (E) The average CTCs of the NCs and before-CPAP-treatment OSA groups. (F) ROC curves of peak concentration values wash-in rate values, and wash-out rate values of all CTCs in frontal cortex for distinguishing type II CTCs (n = 314) from type I CTCs (n = 175). (G) Comparison of peak concentration values wash-in rate values, and wash-out rate values of the type I CTCs in the frontal cortex between NCs (n = 70) and before-CPAP-treatment OSA (n = 105) groups (Mann–Whitney U-test). (H) Comparison of peak concentration values, wash-in rate values, and wash-out rate values of the type II CTCs in the frontal cortex between NCs (n = 166) and before-CPAP-treatment OSA (n = 148) groups (Mann–Whitney U-test). (I) K-means cluster outcome of all CTCs in the bilateral frontal cortex of the NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. The red drops represent type I CTCs. The blue triangles represent type II CTCs. (J) Average CTCs based on the cluster analysis of totaling 489 CTCs from the NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. (K) The proportions of the type I and type II CTCs in NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. (L) The numbers of type I CTCs for each subject of NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. (M) The average CTCs of the NCs and after-CPAP-treatment OSA groups. (N) ROC curves of peak concentration values wash-in rate values, and wash-out rate values of all CTCs in frontal cortex for distinguishing type II CTCs (n = 314) from type I CTCs (n = 175). (O) Comparison of peak concentration values wash-in rate values, and wash-out rate values of the type I CTCs in the frontal cortex between NCs (n = 70) and after-CPAP-treatment OSA (n = 105) groups (Mann–Whitney U-test). (P) Comparison of peak concentration values wash-in rate values, and wash-out rate values of the type II CTCs in the frontal cortex between NCs (n = 166) and after-CPAP-treatment OSA (n = 148) groups (Mann–Whitney U-test)

Fig. 3

Measurement of fluid flow of PVSs by DCE-MRI. (A) K-means cluster outcome of all CTCs in the bilateral frontal cortex of the NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. The red drops represent type I CTCs. The blue triangles represent type II CTCs. (B) Average CTCs based on the cluster analysis of totaling 489 CTCs from the NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. (C) The proportions of the type I and type II CTCs in NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. (D) The numbers of type I CTCs for each subject of NCs (n = 236) and before-CPAP-treatment OSA (n = 253) groups. (E) The average CTCs of the NCs and before-CPAP-treatment OSA groups. (F) ROC curves of peak concentration values wash-in rate values, and wash-out rate values of all CTCs in frontal cortex for distinguishing type II CTCs (n = 314) from type I CTCs (n = 175). (G) Comparison of peak concentration values wash-in rate values, and wash-out rate values of the type I CTCs in the frontal cortex between NCs (n = 70) and before-CPAP-treatment OSA (n = 105) groups (Mann–Whitney U-test). (H) Comparison of peak concentration values, wash-in rate values, and wash-out rate values of the type II CTCs in the frontal cortex between NCs (n = 166) and before-CPAP-treatment OSA (n = 148) groups (Mann–Whitney U-test). (I) K-means cluster outcome of all CTCs in the bilateral frontal cortex of the NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. The red drops represent type I CTCs. The blue triangles represent type II CTCs. (J) Average CTCs based on the cluster analysis of totaling 489 CTCs from the NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. (K) The proportions of the type I and type II CTCs in NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. (L) The numbers of type I CTCs for each subject of NCs (n = 236) and after-CPAP-treatment OSA (n = 253) groups. (M) The average CTCs of the NCs and after-CPAP-treatment OSA groups. (N) ROC curves of peak concentration values wash-in rate values, and wash-out rate values of all CTCs in frontal cortex for distinguishing type II CTCs (n = 314) from type I CTCs (n = 175). (O) Comparison of peak concentration values wash-in rate values, and wash-out rate values of the type I CTCs in the frontal cortex between NCs (n = 70) and after-CPAP-treatment OSA (n = 105) groups (Mann–Whitney U-test). (P) Comparison of peak concentration values wash-in rate values, and wash-out rate values of the type II CTCs in the frontal cortex between NCs (n = 166) and after-CPAP-treatment OSA (n = 148) groups (Mann–Whitney U-test)

Fig. 4

Spearman correlations between imaging parameters and clinical manifestation. (A) Heatmap of Spearman correlations among morphological changes of PVSs, ventricle enlargement, the AHI, the ODI, the MMSE scores, the MoCA scores, the PSQI scores and the ESS scores. The circle size and color intensity represent the magnitude of correlation. (B) Heatmap of Spearman correlations among wash-out rate values of type I CTCs of frontal cortex, the AHI, the ODI, the MMSE scores and the MoCA scores. The circle size and color intensity represent the magnitude of correlation