Lack of Evidence for Neural Underconnectivity in High-Functioning Adults with Autism

Background:  It is now widely believed that autism is a disorder of connectivity.  Numerous studies have found altered white matter structure, suggestive of connectivity abnormalities, and studies examining the functional interactions between brain regions have generally found reduced correlation (i.e., functional connectivity) in autism.  However, the majority of evidence for functional underconnectivity in autism comes from task-based functional imaging experiments, and so differences in task-related factors (e.g., performance, attention, etc.) might account for this apparent group difference.  An alternative approach is to examine the intrinsic dynamics of functional coupling between brain regions during the resting-state (i.e., in the absence of specific task demands).  

Objectives:  To examine the underconnectivity hypothesis of autism across the whole brain during the resting state.  

Methods:  Resting-state fMRI data were acquired from 19 adults with high-functioning autism and 20 matched controls.  Data were preprocessed using conventional corrections (transient signal, motion, slice timing, B0 unwarping, detrending), and nuisance signals were removed by automated independent component filtering.  Two complementary whole-brain approaches were used to examine resting-state functional connectivity:  (1) a data-driven method using dual regression of a group level Independent Component Analysis (ICA), in order to identify and compare specific functional networks at the individual and group levels; and (2) a pairwise correlation method using probabilistically-defined anatomic regions of interest covering the entire brain.

Results:  In both methods, the resting-state connectivity maps were strikingly similar between autism and control groups, both in terms of spatial organization and inter-regional temporal correlation.  No group differences in interregional correlations survived false discovery rate correction (q < 0.05).  These results were found to be robust across various preprocessing conditions, including regression of global signal.  Similarly, no inter-group differences were observed in the dual regression of the group-level ICA.  Only one other study has examined whole-brain resting-state functional connectivity in autism, and they found that groups were distinguishable based on measures of connectivity (Anderson et al., 2011).  However, their analysis was quite different from that of the present study, in that it employed a machine learning approach focused on group classification, and so our studies may not be directly comparable.

Conclusions:  These results challenge the underconnectivity theory of autism.  The lack of a group difference may be a consequence of examining intrinsic functional connectivity during a resting-state, as opposed to during task-driven experimental paradigms.  It may also be a consequence of sample size, though equivalent and even smaller sample sizes have identified group differences, mainly in the default mode network.  Additional replication studies in a larger sample and that include younger children with autism are warranted.

See more of: Brain Imaging: Resting State fcfMRI and Structural Imaging
See more of: Brain Imaging: fMRI-Social Cognition and Emotion Perception
See more of: Brain Structure & Function

| More