Changes in rest-activity rhythms in adolescents as they age: associations with brain changes and behavior in the ABCD study

Abstract

Background: Adolescents with disrupted rest-activity rhythms (RAR) including shorter sleep duration, later sleep timing and low physical activity levels have higher risk for mental and behavioral problems. However, it remains unclear whether the same associations can be observed for within-subject changes in RAR.

Methods: Our longitudinal investigation on RAR used Fitbit data from the Adolescent Brain Cognitive Development (ABCD) Study at the 2-year (FL2: aged 10-13 years) and 4-year follow-up (FL4: aged 13-16 years). 963 youths had good-quality Fitbit data at both time points. In this study we examined changes in RAR from FL2 to FL4, their environmental and demographic contributors as well as brain and behavioral correlates.

Results: From FL2 to FL4, adolescents showed decreases in sleep duration and physical activity as well as delayed sleep timing (Cohen's d .44-.75). The contributions of environmental and demographic factors to RAR changes were greatest to sleep timing (explained 10% variance) and least to sleep duration (explained 1% variance). Delays in sleep timing had stronger correlations with behavioral problems including greater impulsivity and poor academic performance than reductions in sleep duration or physical activity. Additionally, the various brain measures differed in their sensitivity to RAR changes. Reductions in sleep duration were associated with decreased brain functional connectivity between subcortical regions and sensorimotor and cingulo-opercular networks and with enhanced functional connectivity between sensorimotor, visual and auditory networks. Delays in sleep timing were mainly associated with grey matter changes in subcortical regions.

Conclusions: The current findings corroborate the role of sleep and physical activity in adolescent's brain neurodevelopment and behavior problems. RAR might serve as biomarkers for monitoring behavioral problems in adolescents and to serve as potential therapeutic targets for mental disorders.

Keywords: actigraphy; environmental factors; grey matter; physical activity; resting state functional connectivity; sleep duration; sleep timing; white matter integrity.

Figure 1. Environmental and demographic contributors to longitudinal RAR changes

Predictors included in the best model are colored. The color represents the standardized coefficients for each predictor. * denotes significant predictors for RAR changes. ** p_FDR <.05, * p_uncorr<.05, diff: differences between FL2 and FL4.

Figure 2. RSFC (FL4-FL2) component associated with longitudinal changes in sleep duration

Loadings of RSFC component 1 are illustrated. Left: A chord chart depicts RSFC between cortical networks. |loadings|>0.04 are illustrated. Right: RSFC between subcortical regions and cortical networks. Red indicates positive loading while blue indicates negative loadings. |loadings|>0.1 are illustrated. ngd: Gordon cortical networks, scs: subcortical regions, vta: ventral attention, vs: visual, smm: sensorimotor mouth, smh: sensorimotor hand, sa: salience, rst: retrosplenial temporal, fopa: fronto-parietal, dsa: dorsal attention, df: default mode, copa: cingulo-parietal, cerc: cingulo-opercular, au: auditory, aa: accumbens-area, ag: amygdala, bs: brain-stem, cde: caudate, crc: cerebellum, hp: hippocampus, pl: pallidum, pt: putamen, thp: thalamus-proper, vtdc: ventral diencephalon.

Figure 3. GM (FL4-FL2) component associated with longitudinal changes in sleep timing

Loadings of GM component 6 associated with changes in sleep timing are illustrated. Loadings on cortical thickness (upper), loading on subcortical grey matter volumes (bottom).

Figure 4. Behavioral correlates of longitudinal RAR changes

Correlations between changes in RAR and behavioral problems after regressing out sex age and race. The color in the heatmap represents the standardized coefficients. ** p_FDR <.05, * p_uncorr<.05. Caff_ss_sum: weekly caffeine consumptions, pps_y_severity_score: severity score for early risks for psychosis, diff: differences between FL2 and FL4.