Specificity of Atypical Neural Response to Language in Infants at Risk for ASD

Background:  Language delay impairing communication is a primary feature of autism spectrum disorder (ASD). Children with ASD display social and communicative impairments, and present with varying levels of language functioning (Kjelgaard et al., 2001). One approach to studying language processing in infants at high risk for ASD is the use of auditory event-related potentials (ERPs). Previous work has demonstrated irregularities in the P150 component in infants at high risk for ASD as an indicator of abnormal speech processing (Guiraud et al., 2011). Outside of ASD, study of language development is of interest in craniosynostosis, a congenital condition of premature skull fusion in infants causing abnormal skull shape and distribution of brain volume. Craniosynostosis (CSO) has been associated with delayed speech and decreased abilities in both reading and spelling (Knight et al., 2014). This study seeks to compare language processing in infants across two disorders associated with language impairment—infants at high-risk for ASD with infants with CSO.

Objectives:  Compare auditory ERPs during an auditory oddball paradigm in infants at high risk for ASD (HR), infants with craniosynostosis (CSO), and infants at normal risk for ASD (NR).

Methods:  49 infants (12 HR, 13 CSO, and 24 NR) completed an EEG recorded with a 128-channel Hydrocel Geodesic sensor net. Infants listened to a series of 100 speech sounds – the dental /da/ versus the retroflex /da/ phoneme, each presented 50 times. Data was segmented to 100ms pre-stimulus, and ERPs were extracted over 100-300ms post-stimulus for the P150 component and 400-550ms for the N450 component. Magnitudes of the P150 and N450 component amplitudes were compared between the study groups. Statistical analyses were performed using repeated measures analysis of variance (ANOVAs) with participant group as a between-subjects factor and brain hemisphere as a within-subjects factor.

Results:  No significant differences (p>0.05) were observed in the P150 or N450 component amplitudes in response to the dental versus retroflex phonemes within each of the HR, CSO, and NR groups. At the N450, a significant hemisphere by group interaction was observed (p<0.05). NR infants displayed lateralized response to language (p<0.01), while HR and CSO infants did not evidence significant hemispheric lateralization (p=0.32 and p=0.60, respectively). At the P150 ERP, a marginal hemisphere by group interaction was observed (p=0.06). NR infants displayed lateralized response to language (p=0.04), while HR and CSO infants displayed no detectable hemispheric lateralization (p=0.25 and p=0.57, respectively).

Conclusions:  This study compared infants at high risk for autism with infants with craniosynostosis, another congenital condition characterized by language impairments. We found that both HR and CSO infants lacked hemispheric lateralization of neural response to language at multiple auditory ERPs, in contrast with control infants who displayed expected patterns of lateralization. Shared patterns of abnormal auditory processing in these two clinical populations suggest that atypical language lateralization may reflect a general disruption in language development, not specific only to ASD.