31548
Longitudinal EEG Markers of Language Development in Infants and Toddlers at Risk for Autism Spectrum Disorder
Language development in children with autism spectrum disorder (ASD) varies greatly. While many children first present with delayed language skills, roughly one quarter go on to have age-appropriate language skills by school age, and an estimated 30% will be minimal verbal. One of the best predictors of later achievement in children with ASD is language acquisition. Despite this, we know little about the neurobiological correlates of language development in infants at high risk for autism. Here we present the results from baseline EEG data longitudinally collected from infants, as part of the Infant Sibling Project, aimed at comparing infants with familial risk of developing ASD with low risk controls.
Objectives:
This study aims to characterize longitudinal EEG data collected over the first two years of life in children at low (LR) and high risk (HR) for autism, and specifically determine (1) which EEG measures correlate with later language development and (2) whether there are differences between LR and HR infants.
Methods:
Each infant was seen at multiple points between 3 and 36 months for EEG collection, developmental evaluation using the Mullen Scales of Early Learning (MSEL), and eventual ASD evaluation at 18, 24, or 36 months. Data used for this analysis was collected from 112 infants (54LR, 58HR). Ordinary least squares modeling of longitudinal baseline power from 3 to 24 or 3 to 12 months over several frequency bands was used to determine each infant’s estimated 6-month intercept and slope. Multivariate linear regression was then used to characterize the relationship between EEG measures and the MSEL verbal quotient at 24 months.
Results:
Model estimated language scores significantly correlated with actual scores (Model Adjusted R2 = 0.329; Pearson r = 0.70, 95% CI: 0.59-0.78,P=1x10-18). Two-way interactions between risk and EEG measures were assessed to characterize differences in brain-language associations between LR and HR infants. Here we found that estimated 6-month intercept in low frequency bands delta and theta significantly contributed to estimated language scores onlyin the HR group, while estimated beta slope significantly contributed to estimated language only in the LR group. A second model, limited to EEG data collected between 3-12 months of age estimated language scores with similar accuracy (Pearson r = 0.66, 95% CI: 0.54-0.76,P=2.5x10).
Conclusions:
These data support the potential use of longitudinal EEG data in providing estimates of future language development in HR infants. Furthermore, these data support a growing body of research showing early brain differences between LR and HR infants regardless of ASD diagnosis, and emphasizes the need to characterize EEG biomarkers of language development in ASD within a high-risk population. Future work will further characterize these differences between risk groups, and determine the effect of ASD as a possible mediator of EEG predictors of language within the high-risk group.