Abnormal Brain Response to Language Stimuli in Sleeping Infants and Toddlers with ASD

Background: Delays and deviances in language acquisition are among the early signs of autism, but alone are insufficient to clearly identify risk for ASD in infants or toddlers.  Recently, there has been great interest in identifying biological markers of early ASD to be used alone or in combination with behavioral markers to better characterize risk.  Using sleep fMRI, our laboratory reported abnormal laterality and extent of brain response to language in 3 year olds with ASD (Redcay and Courchesne, 2009).  We sought to replicate this novel finding, determine whether this neurofunctional abnormality occurs in ASD 1 year olds, and profile the autistic brain’s response to language across early development.   

Objectives: To utilize sleep fMRI to reveal similarities and differences in brain response to language in infants at-risk for ASD and typically developing controls.

Methods: During natural sleep, brain response to language was evaluated in N=30 ASD (ages 14 to 46 mos) and 14 typical (ages 13 to 41 mos) infants and toddlers using fMRI.  Two language tasks were used:  prosodic speech consisting of segments of children’s stories as in Redcay and Courchesne (2009) and single word stimuli from a list of words familiar to each participant based on the MacArthur Bates CDI.  AFNI software was used to detect and correct for head motion and carry out regression analyses that compared BOLD responses between language blocks and periods of no language stimulation while controlling for potential confounds.  After spatial blurring and normalization to standard atlas space, maps of significant language-related response were created for ASD and typical groups and for between groups comparisons, each corrected for multiple comparisons.

Results: In response to prosodic speech, typical toddlers showed positive activation in the primary language processing region, the left superior temporal gyrus, and deactivations in several brain regions such as thalamus.   In contrast, ASD toddlers showed activation in midline cuneus /lingual gyrus and in left middle occipital gyrus and bilateral deactivation in angular gyrus and several right frontal regions.  In a direct between groups comparison, ASD children had significantly less activation in the left superior temporal gyrus in response to prosodic speech.  Single word and non-word stimuli were less potent activators in the typical group, with only subthreshold responses noted in superior temporal gyrus.  Strong and widespread deactivations of midline cortical and subcortical structures were observed during this task in the ASD group.  On direct comparison, bilateral superior temporal gyrus was significantly under-responsive in the ASD group.  Within the ASD group, strong positive correlations were observed with age: Older ASD toddlers showed more widespread and cortical responses compared to younger ASD infants and toddlers.

Conclusions: The superior temporal gyrus is consistently recruited in response to both stories and single words in infants and toddlers that are typically developing but not in those with ASD.  Thus, with suitable stimuli, it is possible to detect differences in the brain response to language in infants and toddlers at-risk for ASD and holds promise that fMRI may be used to evaluate risk for autism in the future.