Increased Sleep Depth in Developing Neural Networks: New Insights from Sleep Restriction in Children

Abstract

Brain networks respond to sleep deprivation or restriction with increased sleep depth, which is quantified as slow-wave activity (SWA) in the sleep electroencephalogram (EEG). When adults are sleep deprived, this homeostatic response is most pronounced over prefrontal brain regions. However, it is unknown how children's developing brain networks respond to acute sleep restriction, and whether this response is linked to myelination, an ongoing process in childhood that is critical for brain development and cortical integration. We implemented a bedtime delay protocol in 5- to 12-year-old children to obtain partial sleep restriction (1-night; 50% of their habitual sleep). High-density sleep EEG was assessed during habitual and restricted sleep and brain myelin content was obtained using mcDESPOT magnetic resonance imaging. The effect of sleep restriction was analyzed using statistical non-parametric mapping with supra-threshold cluster analysis. We observed a localized homeostatic SWA response following sleep restriction in a specific parieto-occipital region. The restricted/habitual SWA ratio was negatively associated with myelin water fraction in the optic radiation, a developing fiber bundle. This relationship occurred bilaterally over parieto-temporal areas and was adjacent to, but did not overlap with the parieto-occipital region showing the most pronounced homeostatic SWA response. These results provide evidence for increased sleep need in posterior neural networks in children. Sleep need in parieto-temporal areas is related to myelin content, yet it remains speculative whether age-related myelin growth drives the fading of the posterior homeostatic SWA response during the transition to adulthood. Whether chronic insufficient sleep in the sensitive period of early life alters the anatomical generators of deep sleep slow-waves is an important unanswered question.

Keywords: brain development; brain maturation; high density EEG; mcDESPOT; myelin; sleep EEG; sleep deprivation; slow wave activity.

FIGURE 1

Study Protocol. Study measures for each subject involved: (i) One night of high-density EEG during habitual sleep (HS) and restricted sleep (RS; blue boxes) to assess slow-wave activity (SWA, 1–4.5 Hz) topography and to compare SWA between conditions and (ii) mcDESPOT magnetic resonance imaging scan in the morning either in the HS or the RS condition to assess myelin water fraction (MWF) masks. The two conditions were counterbalanced and separated by at least 7 days.

FIGURE 2

Method illustrations. (A) Layout of high-density EEG electrode net in top view (adapted from Electrical Geodesics, Inc.). Marked electrodes (numbered 43, 48, 49, 56, 63, 68, 73, 81, 88, 94, 99, 107, 113, 119, 120, 125, 126, 127, and 128) were excluded (marked as black circles). The remaining 109 electrodes were included in the analysis. (B) Illustration of the time windows used for the three different comparisons in relation to the homeostatic decrease of sleep pressure. Each time window included 60 min artifact-free non-REM sleep (stages N2 and N3), and was identified for each individual separately. The time point of the last common 60 min in HS was determined according to minutes of non-REM sleep passed to the corresponding 60-min-window in RS. (C) Illustration of the three time windows exemplified in typical hypnograms for the two sleep conditions.

FIGURE 3

Homeostatic sleep SWA response to sleep restriction in children: effect of RS (50% reduction of HS) on the sleep EEG in the first 60 min of artifact-free non-REM sleep. (A) Effects of RS on EEG power density spectra presented as RS/HS in percentage, averaged across all 109 channels. (B) Percentage of channels reaching significant differences between RS and HS as a function of frequency (paired t-test, df [mean across 0.25-Hz frequency bins and 109 electrodes] = 23, p < 0.05).

FIGURE 4

Local sleep homeostasis. (A) Topographical SWA maps of HS and RS (group average data of 109 channels in both conditions). Values are color coded (maxima in brown, minima in blue) and plotted on the planar projection of a hemispheric scalp model. Values between electrodes were interpolated. (B) Percentage change in SWA as RS/HS. Statistical non-parametric mapping (SnPM) of the homeostatic SWA response resulted in a cluster consisting of 13 electrodes exhibiting increased SWA after RS (white circles; critical t-value = 2.18). Three electrodes in the cluster survived single-threshold tests (white circles with thicker black boundaries). (C) Topographical SWA maps of HS and RS for the last 60 min non-REM sleep. (D) Percentage change in SWA as RS/HS for the last 60 min of non-REM sleep. (E) Topographical SWA maps of HS and RS for the last common 60 min non-REM sleep (approximately last hour of RS and 4th hour of HS). (F) Percentage change in SWA as RS/HS for the last common 60 min of non-REM sleep.

FIGURE 5

(Left) MWF in the optic radiation is related to the homeostatic SWA response (Pearson correlation at each electrode, two-tailed, p < 0.05; SnPM corrections for minimal cluster size; correlation coefficients r shown in map). Bilaterally, parieto-temporal areas revealed a negative relationship with the homeostatic SWA response (RS/HS, first 60 min non-REM sleep). Electrode clusters consist of 7 and 5 electrodes (cluster size threshold 5). Electrodes within clusters did not survive single threshold tests (critical value r = ± 0.84). (Right) Sagittal view of optic radiation.