Sleep Problems, Brain Development, and ASD Risk in the First Two Years of Life

Background: Children with ASD experience sleep problems at 2-3 times the rate of typically developing (TD) children (Liu, 2006; Maskey, 2013). Sleep problems in ASD may emerge as early as 30 months, and persist through adulthood (Humphreys, 2014; Croen, 2015); however, little is known about sleep patterns in infants at high familial risk for ASD or about possible effects of inadequate sleep on early brain development (Graven, 2008).

Objectives: Using data from a multi-site, longitudinal neuroimaging study of a large cohort (N=440) of infants at high risk for ASD (HR; with an older sibling with ASD) and low risk (LR; with a TD older sibling and no family hx of ASD), we examined parent-reported sleep problems from 3-12 months of age in relation to brain structure from 6-24 months of age.

Methods: Participants were assessed for ASD by expert clinicians at 24 months and underwent MRI (high-resolution 3T T1 & T2-weighted imaging data; 1mm³ voxels) at 6, 12, and 24 months. An automated segmentation algorithm derived hippocampal volume (HCV), total cerebral volume (TCV), caudate and amygdala volumes (Fig. 1). Children were grouped according to diagnostic outcome at 24 months (HR-ASD=72, HR-NonASD=241, and LR=127). Sleep problems were characterized from 3-12 months of age with the Infant Behavior Questionnaire (IBQ), a measure of infant temperament with 5 items related to sleep initiation and maintenance. IBQ sleep scores demonstrated adequate validity when compared with scores from the Brief Infant Sleep Questionnaire, available for a subgroup of the sample (n=98; Fig. 2). Linear mixed models were used to predict HPC, amygdala, and caudate volumes (individual intercepts included as a random effect; outcome group, sleep score, age, quadratic effect of age, and group interactions with each included as fixed effects; TCV and scan site included as covariates).

Results: Infant sleep scores differed across the groups (F(2,397)= 4.7, p<.01); post hoc testing revealed greater sleep problems for HR-ASD vs TD and a non-significant trend for HR-ASD > HR-NonASD infants. Group moderated the relationship between sleep problems and hippocampal volume in infancy. Worse sleep in infancy was related to increased hippocampal volume from 6-24 months only for HR siblings who went on to develop ASD (HR-ASD; β=57.16; t=3.15, p=.002; Fig. 1). No relationship between poor sleep and hippocampal volume was found in the HR-NonASD and LR groups. The IBQ-HPC relationship was unique to the IBQ sleep items; no other IBQ subscales predicted HPC volume, and no significant sleep-amygdala or sleep-caudate relationships were found.

Conclusions: Sleep in the first 12 months of life may be related to hippocampal development in HR infants who go on to develop ASD. In contrast to prior studies with typically-developing older children and sleep-deprived animals (Taki, 2012; Guzman-Marin, 2003), sleep problems in HR-ASD infants were associated with increased (rather than decreased) HPC volume. The relationship between sleep and HPC volume is likely to be dynamic across development and this relationship may begin to differ early in the course of ASD in HR infants.