Autism spectrum disorder (ASD) is characterized by core sociocommunicative impairments. Atypical intrinsic functional connectivity (iFC) has been reported in numerous studies of ASD. A majority of findings has indicated long-distance underconnectivity. However, fMRI studies have thus far exclusively examined static iFC across several minutes of scanning. We examined temporal variability of iFC, using sliding window analyses in selected high-quality (low-motion) consortium datasets from 76 ASD and 76 matched typically developing (TD) participants (Study 1) and in-house data from 32 ASD and 32 TD participants. Mean iFC and standard deviation of the sliding window correlation (SD-iFC) were computed for regions of interest (ROIs) from default mode and salience networks, as well as amygdala and thalamus. In both studies, ROI pairings with significant underconnectivity (ASD<TD) were identified. Mediation analyses showed that decreased mean iFC in the ASD groups was significantly affected by increased SD-iFC. Our study is the first to identify temporal variability across time as a significant contributing factor to the common finding of static underconnectivity in ASD. Since peak connectivity across time was not significantly reduced in ASD, static underconnectivity findings may have to be reinterpreted, suggesting that connections are not actually "broken" in ASD, but subject to greater intra-individual variability across time. Our findings indicate the need for dynamic approaches to iFC in clinical functional connectivity MRI (fcMRI) investigations.
Keywords: autism spectrum disorder; default mode network; fMRI; functional connectivity; intra-individual variability; mediation analysis; resting state.