Biomarkers of Outcome with Intervention in Toddlers at Risk for ASD

Background: Autism Spectrum Disorder (ASD) is a neurodevelopmental disorder characterized by core impairments in social communication skills. While behavioral intervention is the mainstay of treatment, the heterogeneity of the disorder precludes a “one size fits all” treatment approach and, in turn, makes it challenging to predict individual responses to treatment. Efforts have been made in genetics, electrophysiology (event related and resting state), and neuroimaging to quantify biomarkers that may inform treatment targets and outcomes in children with ASD.

 

Objectives: We focus on electrophysiological (EEG) biomarkers and review studies that have sought to characterize clinically meaningful (1) subgroups within the autism spectrum and (2) change with development and/or intervention, using event related and resting state EEG. We then present early data from a behavioral intervention study of toddlers, ages 12-24 months, who are showing early signs of social communication deficits (HR=high risk for ASD). EEG defined domains at baseline and after treatment include face processing, learning from social engagement and resting state oscillations.

 

Methods: Toddlers ages 12-24 months who are showing signs of social communication impairment, as quantified by a battery of clinical assessments including the Mullen Scales of Early Learning, Early Social Communication Scales (ESCS), and the Autism Diagnostic Observation Schedule-Toddler Version (ADOS-T), are being enrolled in a behavioral intervention study targeting joint attention and engagement (JASPER: Joint Attention Symbolic Play Engagement Regulation). At baseline and then at the end of the two-month intervention, children are studied with comprehensive behavioral measures and EEG measures of: familiar vs. unfamiliar faces, object processing after social engagement, and resting state (video of soap bubbles). EEG measures were chosen based on the putative developmental targets of the JASPER intervention, namely joint attention, engagement and behavior regulation.

 

Results: There was tremendous heterogeneity in resting state EEG power in the at-risk group, particularly in gamma and theta power. EEG power in both frequency bands did not correlate with language or cognitive ability but did relate to overall social engagement at baseline. On the other hand, there was consistent evidence of differentiation of familiar and unfamiliar faces, as quantified by a larger Nc and N290 amplitude to stranger compared to caregiver. Additionally, data quality was much improved at the second visit, quantified by percent good trials in event related paradigms, and in the amount of physiological artifact in resting state.

 

Conclusions: These findings suggest that processing of familiar and unfamiliar faces may not be as useful of a stratification marker as resting state oscillations in infants and toddlers with ASD symptoms. Moreover, the difference in data quality after treatment may reflect improved behavior due to intervention (ie. treatment effect on the clinical status of child) or simply a desensitization effect from repeat testing, both of which have to be considered when identifying quantifiable markers of change with intervention. We place these early results in the context of our overall discussion about choice of biomarkers in defining treatment outcomes in ASD.