Longitudinal Cortical Thickness Development in Relation to Changes in SRS Scores over Time in Autism

Background:  The relationship between brain development, clinical heterogeneity and outcome in autism is unknown. Our group previously described widespread abnormalities in longitudinal cortical thickness age-related changes in autism compared to a typically developing sample.  Here we expand on these initial findings and examine brain growth associated with changes in clinical characteristics within our autism sample.

Objectives:  To identify regional differences in longitudinal cortical thickness age-related changes in autism participants whose autism traits decrease over time vs. increase over time, as measured by the Social Responsiveness Scale (SRS).

Methods:  Freesurfer-derived cortical thickness measurements were examined in 73 males with autism (age 3-36 at first MRI scan) scanned up to three times, on average every 2.5 years. SRS scores obtained at the initial scan and a follow-up timepoint were compared, resulting in a slope estimate for each participant. Participants were classified as having autism traits that “decreased” vs. “increased” over time if their slopes fell outside of one standard deviation of a typically developing comparison sample. Mixed effects models were used to compare longitudinal cortical thickness changes between the autism subgroups.

Results:  Individuals whose SRS Total Score or autism traits decreased over time (n=18) had a greater rate of cortical thinning in the right lateral orbitofrontal, middle temporal, insular cortex, and left parahippocampal and left precuneus cortex.  Those whose autism traits increased over time (n=20) showed a greater rate of cortical thinning in the right rostral anterior cingulate cortex, left bank of the superior temporal sulcus and temporal pole.  A decrease in SRS Autistic Mannerisms over time (n=34) was associated with reduced cortical thinning in the cingulate (bilateral rostral anterior, right caudal anterior and isthmus), right fusiform, left bank of the superior temporal sulcus and temporal pole. 

Conclusions:  Our findings suggest regional differences in cortical thickness age-related changes may be associated with improvement or worsening of autism symptomatology in late neurodevelopment. Cortical thickness of the middle temporal and bilateral cingulate regions overlap with those identified in a genetic analysis of candidate genes for autistic traits in a typically developing sample (Hedrick et al 2012). Interestingly, only one region identified in our current analysis was also described as atypically developing in a group level analysis of our larger autism sample vs. our typically developing group (Zielinski et al. 2014). These findings highlight the importance of examining clinical heterogeneity within autism samples and accounting for changes in autism traits over time when measuring brain-based biomarkers.