Resting Gamma Power Predicts Language Ability in Infants at Risk for ASD

Background:  High frequency electrophysiological oscillations (gamma band: 30-50 Hz) are involved in binding local neural circuitry during language and cognitive processing (Basar 2013). During typical development, spontaneous gamma oscillations increase between 3-4 years of age and peak at 4-5 years (Takano 1998). While higher gamma power during toddlerhood has been shown to predict greater language and cognitive abilities (Gou 2011), the development and significance of gamma power in the first year of life, in the context of typical development and autism risk, is unknown. Infants are at high risk (HR) of developing ASD if they have an older sibling with ASD. Given the high rates of language difficulties in HR infants (Messinger 2013), early predictors of language development are important to establish.

Objectives:  To examine whether baseline EEG power in the gamma frequency band measured at 6 months of age predicts language abilities at 12 and 18 months in infants who are at high risk (HR) and low risk (LR) of developing ASD. We hypothesized that higher gamma power will correlate positively with later language abilities, with a stronger positive correlation in LR infants compared to HR infants.

Methods:  The sample included 28 HR infants and 17 LR infants. EEG was recorded at 6 months and behavioral testing was conducted at 12 and 18 months (m12, m18) using the Mullen Scales of Early Learning (Mullen) and MacArthur-Bates Communicative Development Inventory (MCDI). Baseline EEG was recorded using a high-density system for 2 minutes while infants watched a video of bouncing soap bubbles. EEG data were processed using Net Station 4.4.5 software (McEvoy 2015), and independent component analysis was done using EEGlab. Relative spectral power for gamma frequency band was calculated using Welch’s method and averaged across 3 frontal regions.

Results:  HR and LR infants did not differ in frontal gamma power. LR infants trended towards having higher language scores than HR infants, but the difference was not significant (Table 1). When HR and LR groups were combined together as one cohort, frontal gamma power did not correlate with language scores. Within the HR group, frontal gamma power positively correlated with m12 Mullen verbal developmental quotient (DQ), m18 Mullen verbal DQ, m18 Mullen receptive and expressive language, and m18 MCDI words understood. In the LR group, frontal gamma power did not correlate with language scores (Table 2).

Conclusions:  Findings in the HR group support our hypothesis that higher gamma power positively relates to later language abilities. Our study is the first to examine the relationship between baseline gamma power and language ability in early infancy. In our HR infants, we replicated the positive correlation between baseline frontal gamma power and language ability as was shown previously by Gou et al in typically developing toddlers. Lack of differences in frontal gamma power between HR and LR groups suggest that at 6 months of age, differences in high frequency cortical activity between these two groups have not emerged. Lack of significant findings in the LR group may be due to inadequate statistical power due to sample size.