Altered Patterns of Local Connectivity in Autism Measured By Regional Homogeneity

Background: Disruption in functional connection has been a characteristic feature of brain functioning in autism spectrum disorders (ASD). Studies of regional homogeneity (ReHo), which measure local connectivity, in autism have provided somewhat inconsistent results. Previous reports of local overconnectivity in visual cortices during rest in ASD (Maximo et al., 2013; Keown et al., 2013) were not replicated in some studies using a large shared database ABIDE (Autism Brain Imaging Data Exchange) (DiMartino et al., 2013; Dajani & Uddin, 2015). However, eye status during rest was not accounted for in these studies (samples included both eyes open and eyes closed scans), and studies presented both standardized and unstandardized connectivity maps. Our recent study of subsamples from ABIDE showed variable patterns of posterior ReHo across eye status and processing pipelines (Nair et al., under review).

Objectives: To use a data-driven approach in examining the patterns of local brain connectivity in autism as a function of eye status and processing pipelines during resting state fMRI.

Methods: Resting state fMRI data from ABIDE-II database (Georgetown University sample) were preprocessed using standard processing stream, including motion correction, spatial and temporal filtering, and analysis was performed with global signal regression (GSR). Time points with motion >.5mm were censored, and participants with >80% time points remaining after censoring were included in the analysis. Participants were matched on age (8-14 years; M=11), IQ (93-147; M=116), and RMSD (.03-.14; M=.08), resulting in final samples of 31 ASD, 35 TD participants. Voxel-wise ReHo maps (AFNI’s 3dReho) were derived and standardized, and 2 two-sample t-tests were conducted for standardized and unstandardized ReHo pipelines.

Results: Standardized analysis yielded local overconnectivity in ASD in bilateral paracentral lobule, left precuneus and fusiform gyrus, left superior frontal gyrus, and left inferior parietal lobule (IPL). Local underconnectivity was detected in ASD in bilateral cuneus, right cerebellum and superior temporal gyrus. Unstandardized analysis yielded local overconnectivity in ASD in bilateral paracentral lobule and precuneus, left fusiform gyrus, IPL, middle temporal gyrus, inferior frontal gyrus, left cerebellum, and right middle frontal gyrus, and local underconnectivity in right cerebellum. Results were cluster-corrected using Monte Carlo simulations to obtain a corrected significance level of p<.05 (uncorrected p<.01) and cluster size of 41 voxels (AFNI’s 3dClustSim).

Conclusions: While DiMartino et al. and Dajani & Uddin reported overall patterns of local underconnectivity across posterior areas and overconnectivity across anterior areas (using mixed eyes open and eyes closed sample), both ReHo processing pipelines in the present study do not show such a distinct anterior-posterior pattern of local connectivity. This variable pattern of connectivity in standardized analysis is novel and interesting. Additionally, lateralization effects entailed general diffuse overconnectivity for left and underconnectivity for right hemisphere in ASD. These findings are preliminary and our future analyses will utilize low-motion, high quality eyes open and eyes closed data from ABIDE-II. Assessing differences across local connectivity studies in ASD in relation to our findings is important, as they may underscore effects of multisite data and methodological variability.