Local Brain Connectivity Across the Lifespan in Autism Spectrum Disorder and Typical Development

Background:  

          Autism spectrum disorders are characterized by abnormal brain connectivity (Kana et al., 2005), with some evidence for short-range overconnectivity and long-range underconnectivity (Courchesne & Pierce, 2004).  Regional homogeneity (ReHo) is an analytic approach that can be applied to resting-state fMRI data to measure short-range (local) functional connectivity within a discrete region of the brain (Zang et al., 2004). There are few studies that have examined local connectivity in ASD using ReHo (e.g., Paakki et al., 2010), and none have examined whether local connectivity changes with age in ASD, nor whether these changes parallel those observed in typical development.

Objectives:  

          We aimed to characterize the developmental changes in ReHo across children, adolescents, and adults with high-functioning autism (HFA) and a typically developing (TD) group. We further aimed to characterize relationships between local connectivity and symptom severity in the HFA group.

Methods:  

          Data from the Autism Brain Imaging Data Exchange (Di Martino et al., 2014) contributed by NYU was used. Data were stratified into Child (< 11 years, n=36), Adolescent (11-18 years, n=40), and Adult (≥18 years, n=30) groups. The HFA and TD participants were matched within age group on IQ, mean framewise displacement (Power et al., 2012), and gender. After pre-processing the data using DPARSF-A (Yan & Zeng, 2010), ReHo values were calculated for every voxel within an eroded brain mask to remove invalid data at the brain edges.  ReHo was calculated by computing Kendall’s coefficient of concordance for one voxel and the surrounding 26 voxels (Kendall & Gibbons, 1990). Global mean ReHo was calculated by computing the mean ReHo for the entire brain for each subject. Differences in ReHo between HFA and TD groups within each age group were assessed with an independent samples t-test using SPM8. A linear regression model was used to assess the relationship between global mean ReHo and social communication questionnaire values (Berument et al., 1999), using mean framewise displacement as a covariate. 

Results:  

          Patterns of both increases and decreases in ReHo were observed in comparisons across all age groups. The TD group showed consistently higher ReHo in the occipital and cerebellar regions compared with HFA across all age groups. When comparing the effect of diagnosis and age group on global mean ReHo measures, there was a significant interaction of diagnosis and age group, F(2,100)=3.39, p=.038, partial η²=.064.  Global mean ReHo values were higher for TD (M=.435, SD=.03) than HFA (M=.419, SD=.03) in children, but this did not reach significance, t(34)=-1.72, p=.095. Global mean ReHo values were similar between diagnostic groups in adolescents and adults. Global mean ReHo values positively predicted SCQ total score in children and adolescents with HFA, t(50)=2.15, p=.039, β=.45.   

Conclusions:  

          Contrary to the short-range overconnectivity hypothesis, children with HFA demonstrated lower ReHo across the brain compared with TD children. Global mean ReHo was similar between HFA and TD groups across adolescence and adulthood, supporting a “normalization” of brain connectivity in adulthood (Tyzka et al., 2013). Individual differences in regional homogeneity, indicative of short-range overconnectivity, may drive poor social communication in autism.