Brain white matter damage and its association with neuronal synchrony during sleep

Abstract

The restorative function of sleep partly relies on its ability to deeply synchronize cerebral networks to create large slow oscillations observable with EEG. However, whether a brain can properly synchronize and produce a restorative sleep when it undergoes massive and widespread white matter damage is unknown. Here, we answer this question by testing 23 patients with various levels of white matter damage secondary to moderate to severe traumatic brain injuries (ages 18-56; 17 males, six females, 11-39 months post-injury) and compared them to 27 healthy subjects of similar age and sex. We used MRI and diffusion tensor imaging metrics (e.g. fractional anisotropy as well as mean, axial and radial diffusivities) to characterize voxel-wise white matter damage. We measured the following slow wave characteristics for all slow waves detected in N2 and N3 sleep stages: peak-to-peak amplitude, negative-to-positive slope, negative and positive phase durations, oscillation frequency, and slow wave density. Correlation analyses were performed in traumatic brain injury and control participants separately, with age as a covariate. Contrary to our hypotheses, we found that greater white matter damage mainly over the frontal and temporal brain regions was strongly correlated with a pattern of higher neuronal synchrony characterized by slow waves of larger amplitudes and steeper negative-to-positive slopes during non-rapid eye movement sleep. The same pattern of associations with white matter damage was also observed with markers of high homeostatic sleep pressure. More specifically, higher white matter damage was associated with higher slow-wave activity power, as well as with more severe complaints of cognitive fatigue. These associations between white matter damage and sleep were found only in our traumatic brain injured participants, with no such correlation in controls. Our results suggest that, contrary to previous observations in healthy controls, white matter damage does not prevent the expected high cerebral synchrony during sleep. Moreover, our observations challenge the current line of hypotheses that white matter microstructure deterioration reduces cerebral synchrony during sleep. Our results showed that the relationship between white matter and the brain's ability to synchronize during sleep is neither linear nor simple.

Keywords: NREM sleep; sleep; traumatic brain injury; white matter.

Figure 1

Group differences on diffusion metrics. TBSS voxel-wise contrasts between TBI and control (CTL) groups (blue, TBI < CTL; red to yellow, TBI > CTL) for (A) fractional anisotropy (FA), (B) mean diffusivity (MD), (C) axial diffusivity (AD), and (D) radial diffusivity (RD). Significant results are overlaid over the MNI152 T1 1 mm brain and the mean fractional anisotropy skeleton (in green). Significant increases in the TBI group compared to the Control group are shown in the red to yellow scale and significant decreases are shown in light blue. The mean value of all significant clusters is represented on the graph for each subject. Significant results were thresholded at P < 0.05 controlled for age and corrected for multiple comparisons.

Figure 2

Slow wave amplitude and white matter damage. Areas in the TBI group where slow wave amplitude is positively correlated (red to yellow coloured areas) with (A) mean diffusivity (r = 0.81), (B) axial diffusivity (r = 0.74), and (C) radial diffusivity (r = 0.78). Significant results are overlaid over the MNI152 T₁ 1 mm brain and the mean fractional anisotropy skeleton (in green). The correlation between the mean value of all significant clusters and slow wave amplitude is represented on the graphs. No area of negative correlation was found in the TBI group and no significant correlation was found for the control group (CTL). Results are thresholded at P < 0.05, adjusted for age and corrected for multiple comparisons.

Figure 3

Slow wave slope and white matter damage. Areas in the TBI group where slow wave negative-to-positive slope is correlated (red to yellow, positive correlation) with axial diffusivity (r = 0.64). Significant results are overlaid over the MNI152 T₁ 1 mm brain and the mean fractional anisotropy skeleton (in green). The correlation between the mean value of all significant clusters and slow wave N-to-P slope is represented on the graph. No area of negative correlation was found in the TBI group and no significant correlation was found for the control group (CTL). Results are thresholded at P < 0.05, adjusted for age and corrected for multiple comparisons.

Figure 4

Slow wave negative phase duration and white matter damage. Areas in the TBI group where slow wave negative phase duration is correlated (blue, negative correlation; red to yellow, positive correlation) with (A) fractional anisotropy (r = −0.69) and (B) mean diffusivity (r = 0.73). Significant results are overlaid over the MNI152 T₁ 1 mm brain and the mean fractional anisotropy skeleton (in green). The correlation between the mean value of all significant clusters and slow wave negative phase duration is represented on the graphs. No significant correlation was found for the control group (CTL). Results are thresholded at P < 0.05, adjusted for age and corrected for multiple comparisons.

Figure 5

Relative slow-wave activity power in the first sleep cycle and white matter damage. Areas in the TBI group where slow wave activity power is correlated (blue, negative correlation; red to yellow, positive correlation) with (A) fractional anisotropy (r = −0.63), (B) mean diffusivity (r = 0.70), and (C) radial diffusivity (r = 0.65). Significant results are overlaid over the MNI152 T₁ 1 mm brain and the mean fractional anisotropy skeleton (in green). The correlation between the mean value of all significant clusters and relative slow-wave activity power is represented on the graphs. No significant correlation was found for the control group (CTL). Results are thresholded at P < 0.05, adjusted for age and corrected for multiple comparisons.

Figure 6

Self-reported fatigue and white matter damage. Areas in the TBI group where fatigue is correlated (red to yellow, positive correlation) with axial diffusivity (r = 0.66). Significant results are overlaid over the MNI152 T₁ 1 mm brain and the mean fractional anisotropy skeleton (in green). The correlation between the mean value of all significant clusters and fatigue is represented on the graph. No area of negative correlation was found in the TBI group, and no significant correlation was found for the control group (CTL). Results are thresholded at P < 0.05, adjusted for age and corrected for multiple comparisons.