Functional Connectivity during Language Processing in Infants at Familial Risk for ASD

Background: Autism spectrum disorder (ASD) is a disorder of connectivity, with ASD genes converging to disrupt neural development and cortical connectivity prenatally. Familial-risk infants, defined by having at least 1 older sibling with ASD, have increased risks for ASD and language delay (Messinger 2013). Neuroimaging studies have identified abnormal structural and functional connectivity in FR infants at 6 months that relate to future ASD symptoms (Shen 2017). No prior study has examined connectivity during language processing in familial-risk infants longitudinally throughout the first year of life.

Objectives: We asked if EEG connectivity, as measured by phase coherence, during language processing, (1) differentiates infants based on risk status and ASD symptoms during the first year of life (at 3-month, 6-month, 9-month, and 12-month), and (2) relates to 18-month language ability and ASD symptoms. We hypothesized that atypical connectivity during language processing is detectable at 3 months of age in infants who will have 18-month behavioral-concerns for ASD.

Methods: Participants included 49 familial-risk and 35 low-risk infants, who were divided into ASD-concern and No-concern groups based on their 18-month ADOS-T calibrated severity score (scores ≥ 4 are ASD-concern). 18-month groupings included 17 ASD-concern and 54 No-concern infants. EEG was acquired longitudinally (at 3, 6, 9, and 12-month) while infants listened passively to a continuous stream of syllables for 2.5 minutes (EGI 128 channels). EEG was collected using NetStation and filtered at 1.5-50 Hz. To maximally reduce artifacts, EEG data were cleaned using EEGLAB artifact subspace reconstruction, 1-model adaptive mixture independent component analysis, and transformed to current source density. Using newcrossf, phase coherence was calculated in the theta (4-6 Hz), alpha (6-12 Hz) and gamma (30-50 Hz) bands between 39 electrode pairs in putative language networks (Figure 1). Language ability and ASD symptoms were assessed at 18-month using Mullen Scales of Early Learning (MSEL), MacArthur-Bates Communicative Development Inventory (MCDI), the Vineland Adaptive Behavior Scales, and Autism Diagnostic Observation Schedule-Toddler Module (ADOS-T).

Results: All statistics for coherence analyses are provided in Table 1. Phase coherence differentiated ASD-concern from No-concern infants cross-sectionally at 3-month, 6-month, 9-month, and 12-month. Compared to ASD-concern infants, No-concern infants had higher phase coherence across theta, alpha and gamma bands in multiple frontal-posterior, frontal-temporal and frontal-central electrode pairs throughout the first year of life. At electrode pairs that differentiated risk-groups, greater phase coherence predicted greater language ability and reduced ASD symptoms at 18-month.

Conclusions: Neural connectivity during auditory language processing in early infancy likely supports subsequent language development. This study documents the earliest manifestation of altered connectivity during language processing in familial-risk infants. Hypoconnectivity during language processing could serve as an early marker of atypical neurodevelopmental trajectories. Ongoing work with this study will include longitudinal modelling of 3-month through 12-month timepoints to examine potential divergent developmental trajectories in infants with atypical development.