Face Processing in Infants Demonstrating Early Signs of ASD
With earlier detection methods and known benefits of early intervention in Autism Spectrum Disorders (ASD), there is increasing demand for objective measures of risk for ASD and, in turn, of response to intervention. Processing of socially salient information, such as faces, has been studied as a means to understand social cognition in children showing early signs of ASD (Klin, 2003). In children with ASD and in infants at familial risk for ASD, EEG studies have revealed atypical face processing when compared to children without social communication delays (Dawson, 2002).
We investigated 12-24 month old children at high-risk (HR) for developing ASD based on evidence of early social communication delays, before they began an early behavioral intervention program (JASPER: Joint attention symbolic play engagement regulation; Kasari). We asked whether EEG correlates of face processing differentiated this group from a cohort of typically developing (TD) children and whether there was heterogeneity in face processing that may serve as a useful intermediate phenotype for predicting response to intervention. We focused on familiarity processing because the intervention is parent-mediated and targets joint attention.
A cohort of 15 HR infants and 7 age-matched TD children (mean age = 19 months) were presented with images of their caregiver and a stranger in a randomized order while EEG was recorded using 128-channel sensor caps (Electrical Geodesics, Inc.). Data were artifact detected, interpolated, and averaged before analysis, with data rejected for insufficient number of trials, leaving 10 HR and 5 TD children with data for analysis. Repeated measures ANOVA was performed for group (HR, TD), condition (caregiver, stranger) and region (right, left) for face sensitive EEG components, with focus on the Nc (negative component) based on the age range being studied.
The HR group scored significantly lower than the TD group on the Mullen Scales of Early Learning developmental quotient (DQ) and verbal DQ sub-scores. All children in the HR group scored in the at-risk range on the Autism Diagnostic Observation Schedule-Toddler Version (ADOS-T). There was a significant condition effect across groups for Nc mean amplitude, with a greater response to the stranger condition (stranger: -10.01 mV; caregiver: -6.91 mV; p<0.01). This effect did not differ between groups, such that HR and TD children both showed a larger response to the stranger condition. There were no correlations between differentiation of conditions (Nc mean amplitude difference) and Mullen subscores or ADOS-T scores in the HR group.
Despite evidence of delays in social communication skills, infants at HR for ASD demonstrated typical neural responses to faces. Within this small cohort, it seems that face processing may not be the most sensitive biomarker of risk for ASD. However, there still may be individual differences in face processing and overall attention to faces that relate to baseline joint attention skills or that may predict response to an intervention that targets eye contact and attention to social cues. Continued data collection will facilitate analysis of individual and subgroup differences that may inform predictors of outcome in this cohort.