Relationships Between Resting State Neural Connectivity and Individual Differences in ASD Symptoms

Background:  

Studies increasingly implicate neural short-range overconnectivity and long-range underconnectivity in the pathogenesis of autism. Research on resting state connectivity in autism has revealed functional underconnectivity within and between resting state regions implicated in social processes, such as the medial temporal lobe and “salience” networks. However, the specificity of resting state network abnormalities and how they relate to the specific core symptoms and attributes of autism spectrum disorder (ASD) has yet to be fully established.  In particular, the majority of resting state ASD studies to date have not benefitted from the large sample sizes needed to map connectivity networks onto individual differences in symptom profiles.

Objectives:  

To examine and compare resting state networks between large samples of individuals with ASD and typically developing controls (TDC). Although our a priori hypotheses focus on default mode, “salience”/reward circuitry, and medial temporal lobe networks, a variety of cortical networks are considered. Dimensional (ASD symptom dimensions) and categorical (group) approaches to isolating abnormalities in resting state networks are used.

Methods:  Participants in this study complete an extensive battery of psychodiagnostic and cognitive assessments (core ASD assessments include ADOS, ADI-R, Social Responsiveness Scale, and Social Communication Questionnaire). They also complete six minutes of resting state fMRI scanning (full head coverage, TR=2.340 ms).  Data analysis follows three approaches. First, general linear model (GLM) examinations of connectivity are carried out on all gray matter voxels using a priori seeds within networks of interest. Second, spatio-temporal independent components analyses (ICA) are implemented to isolate resting state networks without a priori assumptions. Third, matrices of correlations between a priori brain areas are submitted graph theory-based analyses, including calculations of clustering coefficients, global and local efficiency. Each of these analyses yields statistics that are used as dependent variables in analyses on clinical variables (i.e., scores on the ADI, Social Communication Questionnaire, and Social Responsiveness Scale).

Results:  

To date, resting state fMRI data have been collected from 45 participants with ASD (5 female) and 58 TDC participants (3 female) matched on age (TDC=11.5 years, ASD=11.3 years), and cognitive ability (Differential Abilities Scale GCA, TDC=109, ASD=105). Data collection for this project is ongoing, and will likely make this the largest single study reported to date on resting state functional connectivity in ASD.

Conclusions:  

This study will provide preliminary results of selectivity of specific brain networks to ASD symptom profiles. They will also help determine optimal data analysis path (GLM, ICA, or graph theory) for identifying abnormal cortical networks in ASD.  Ultimately, the identification of both global and local functional separations of components responsible for social processes, when measured as functions of ASD symptomatology, will refine our understanding of the relationships between structural/functional connections and core autism deficits.