Hemispheric Differences in Language Processing in Autism Spectrum Disorders: A Meta-Analysis of Neuroimaging Studies

Background: Language and communication impairments are a hallmark feature of autism spectrum disorders (ASD). These impairments have been attributed to brain abnormalities at anatomical as well as functional levels. Abnormal asymmetry of the language association cortex (Herbert et al., 2005) and right lateralization of the functional language network (Redcay & Courchesne, 2008; Kana & Wadsworth, 2012) have been reported in ASD. Other findings include altered connectivity of the brain areas underlying language processing (Williams et al., 2013), and an over-reliance on visual cortices (Kana et al., 2006) to aid language comprehension in autism. However, a comprehensive characterization of the functional language networks in individuals with ASD, which could aid diagnosis and design of intervention, has been lacking.   

Objectives: The aim of the present study was to quantify common and consistent patterns of neural activity differences associated with language processing in ASD as compared to typically developing control (TD) participants across a large number of neuroimaging studies of language. 

Methods: This preliminary analysis utilized activation likelihood estimation (ALE) meta-analytic approach to examine 20 previously published fMRI studies of language processing in ASD (N=311) and TD (N=307) participants. Statistically significant brain activation foci from these studies at within-group level (ASD, TD), and at between-group level (ASD>TD, TD>ASD) were subjected to quantitative voxel-based meta-analysis using ALE in GingerALE and the results are reported at a cluster-level correction of p<0.05 and a cluster forming FDR correction of p<0.05. The tasks included addressed semantic processing, sentence comprehension, processing figurative language, and speech production.  

Results: The main results of this study include: 1) ASD participants had significantly reduced brain activity (ALE values), relative to TD, in core language areas, such as the left middle/superior temporal gyri (LMTG/LSTG) and the left inferior frontal gyrus (LIFG); 2) Reduced activity in ASD in medial regions, such as the left medial prefrontal cortex (LMPFC) and the anterior cingulate cortex (ACC); and 3) Greater activity in ASD than TD participants in the right hemisphere homologues of language areas, such as the right inferior frontal gyrus (RIFG) and the right superior temporal gyrus (RSTG) as well as in the cerebellum and the posterior cingulate cortex (PCC).  

Conclusions: Findings of this study point to altered patterns of neural recruitment in individuals with ASD while engaged in language processing. An important finding is the hemispheric difference, with ASD participants relying more on the right hemisphere language areas, which has been correlated with poorer language functioning (Dawson et al., 1986; Coffey et al., 2008). This could reflect task difficulty and spill-over processing (Just et al., 1996). We also found reduced recruitment of left hemisphere language areas in ASD participants, suggesting less refined resources and approach to language comprehension (e.g., Harris et al., 2006). Furthermore, reduced activity in MPFC may suggest weak coherence in text comprehension in ASD (Mason and Just, 2006). Although preliminary, the findings of our study provide a cross-section of the functional integrity of the “language brain” in ASD. Further analyses of the data by including more studies are in progress.