Variance in Language Abilities in Autism As a Function of Hemispheric Lateralization and Functional Connectivity

Background:  In autism spectrum disorders (ASD), the brain’s language network (including Broca’s and Wernicke’s areas) has been found to be more right lateralized than in typically developing (TD) individuals, yet the consequences of this remain unclear. While some studies have correlated greater right hemisphere (RH) language lateralization with decreased language abilities, others argue that increased RH activity is an alternate but not inferior means of language processing. Previous literature shows that a weak hemispheric dominance in TD individuals predicts greater functional connectivity between LH and RH language areas (Tzourio-Mazoyer et al., 2015); this is important considering the disruptions in connectivity reported in individuals with ASD.

Objectives:  To determine whether functional lateralization of Broca’s and Wernicke’s areas and functional connectivity between these RH and LH homologues during language processing accounts for variance in the language abilities of individuals with and without ASD.

Methods:  High-functioning ASD (N=15) and TD (N=17) adults silently read literal and figurative sentences in a Siemens 3.0 Tesla fMRI scanner. Groups were matched on age (ASD=21.21 years, TD=21.87 years) and performance IQ (ASD=106.60, TD=107.94); however, verbal IQ (VIQ) differed significantly between groups (ASD= 102.33, TD= 114.65, p<.05). Based on neural activity during sentence processing, individual region-of-interest (ROI) lateralization indices (LI) and ROI-to-ROI functional connectivity were calculated in SPM12 and the LI and CONN toolboxes. Anatomical ROIs were defined using Brodmann areas BA 44 and 45 (Broca’s area) and BA 22 (Wernicke's area). LI values and beta-weights of connectivity between RH and LH language homologues were treated as independent variables in two separate 3-way ANOVAs (group x LI x connectivity; one ANOVA for BA 44/45, one for BA 22) predicting variance in individual VIQ scores.

Results:  1) In BA 22, the three-way ANOVA yielded a significant group x LI x connectivity interaction, F(1,24)= 7.407, p<.05; 2). In BA 44/45 the same effect was not significant, F(7,24)= 2.084, p= .085. However, a two-way ANOVA testing group x LI significantly predicted VIQ, F(3,28)= 4.319, p<.05, such that in the TD group greater left lateralized activity predicted higher VIQ, while the opposite pattern was found in ASD. 3) An additional two-way ANOVA (group x BA 44/45 LI) significantly predicted interhemispheric BA44/45 connectivity F(3,28)= 3.658, p<.05, such that in the TD group connectivity increased as one’s language network became less left lateralized, while no such relationship was found in the ASD group.

Conclusions:  In BA 22, group x LI x connectivity interaction indicates that the VIQ varied extensively in the ASD group as a function of connectivity and lateralization. Most notably, in ASD a combination of weak inter-hemispheric connectivity and reduced LH dominance predicted some of the lowest VIQ’s observed. In BA 44/45, the ASD group failed to show an increase in inter-hemispheric connectivity with weaker hemispheric lateralization. Overall, our findings suggest that the wide variance in VIQ seen in ASD is significantly related to interactions of functional lateralization and connectivity in this population in a manner that is distinct from what is seen in TD individuals.