Failure of STS Activation May Underlie Early Emerging Social Orienting Defects In Autism

Background: Unusual social responding, such as a failure to respond to name, is a red flag for autism in infants and toddlers, but the neural systems that underlie this deficit are unknown.  The superior temporal sulcus (STS) has been dubbed the “chameleon of the human brain” because of its involvement in many social tasks. It is unknown whether its dysfunction underlies social responding deficits at the time of first clinical signs of autism. If so, it may prove to be a valuable biomarker of risk for autism in infants. 

Objectives: To determine whether the STS fails to respond to social orienting sounds in ASD infants and toddlers as compared to typical infants and toddlers.

Methods:  During natural sleep, fMRI responses to social and non-social orienting sounds were evaluated in 31 ASD and 31 typically-developing toddlers ages 12 to 45 months.  Three sound conditions were delivered via headphones:  social orienting consisting of emotionally salient sentences containing each toddler’s name (e.g., “look here, Johnny, look here!”); non-social orienting consisting of environmental noises (e.g., the sound of an airplane), and language sounds consisting of simple sentences (e.g., “Doll playing with doll”).  AFNI was used for motion correction and regression analyses comparing the BOLD responses between each condition and periods of no stimulation. Whole brain group analyses as well as anatomically defined ROI analyses at the individual subject level were performed.  After spatial blurring and normalization, whole brain group maps of significant social orienting-related responses were created for each subject group as well as for direct between group comparisons.  Functional maps were corrected for multiple comparisons (individual voxel level p<.01).  STS was an a priori ROI for individual subject analyses.  To handle anatomical variability in STS morphology, particularly in the developing brain, individual STS masks were created by expert anatomists.  The number of significantly active voxels within the STS mask at p<.01 was calculated for each subject in each condition.      

Results:  Whole brain analyses revealed significantly greater activation in the left STS in TD as compared to ASD subjects. Individual subject analyses of STS revealed striking activation effects. In TD subjects, social orienting sounds produced the greatest number of significantly active STS voxels, while environmental sounds produced the least.  ASD subjects displayed the opposite activation profile: the greatest STS activation was produced by environmental sounds and the least by social orienting sounds. In fact, TD infants and toddlers had a 2.2 times greater extent of STS activation than ASD in response to social sounds (TD=801mm³ vs ASD=359mm³ , p<.001).

Conclusions:  Activity in STS is consistently recruited in response to social orienting sounds, such as the sound of each child’s name being called, in TD toddlers but not in those with an ASD.  Instead, in ASD infants and toddlers the strongest STS/STG activation was to environmental sounds.  Thus, with suitable contrasting social and environmental stimuli, it may be possible to further refine sleep fMRI tests that reliably detect an abnormal profile of STS/STG activation that distinguishes infants at risk for an ASD.