Objectives: The current study used fMRI to compare three groups during processing of sentences that involve low or high levels of visual imagery for reading comprehension: participants with high functioning autism (HFA), typical peers (TD), and individuals with a history of autism spectrum disorder (ASD), but who no longer meet diagnostic criteria for such a disorder (i.e., “optimal outcome” - OO).
Methods: Groups of youth with HFA (n=23), TD (n=21), and OO (n=17) of equivalent mean age (13 years) and gender proportion completed an event-related fMRI task involving Yes/No responses to low-imagery or high-imagery sentences (e.g., “the number six can be rotated to make the number nine”).
Brain regions that survived FDR “whole-brain” corrections where high imagery activation was greater than low imagery for all groups (via SPM8 conjunction analysis; p<.001 uncorrected) were selected as 5mm radius regions-of-interest (ROIs). Extracted mean data from ten ROIs (left Broca's area, left fusiform, right and left IFG/precentral, right and left IPL/postcentral gyrus, right and left SMA, and right and left posterior MTG) were examined using 2 (high vs low imagery) × 3 (group) multivariate repeated-measures ANOVA.
Results: Our comparison of high vs low imagery conditions closely replicated previous findings. There was a multivariate effect of group (p=.001), but not imagery × group (p=.521). Univariate tests showed that average task activation in left inferior frontal gyrus and supplementary motor area had abnormally high HFA activity; OO activation fell between TD and HFA, and did not significantly differ from TD or HFA. Groups differed in right postcentral/inferior parietal lobule (IPL) and bilateral posterior superior temporal gyrus (BA 37) activation. In the latter regions, TD’s either showed no activity or “deactivated” to sentence stimuli, while OO showed abnormally high activation (even greater than HFA hyper-activation). Supplemental analyses found several additional group effects throughout the brain when examining either high imagery or low imagery conditions. The majority of these differences outside of a priori ROIsoccurred near the edges of activated regions, implicating additional ASD abnormality in the extent of activated cortex.
Conclusions: In addition to partially replicating previously reported findings, novel results include the finding that ASD youth with optimal outcomes have partial normalization of abnormal activity observed in HFA, including prefrontal cortex activation in Broca’s region and supplementary motor area during sentence comprehension. In addition, OO show exaggerated hyper-activation, greater than both TD and HFA, in posterior brain regions, including those involved with visuospatial processing (right IPL) and object recognition (posterior inferior temporal). The results suggest that during language processing, regardless of visual imagery, activation in the OO group is linked to a mixture of normalization of function in frontal regions and possibly to compensatory over-activation in posterior regions.