Disrupted emergence of networks in ASD: Evidence from fcMRI and DTI

Background:  There is broad consensus that ASD requires investigation at the brain network (rather than local) level. Huge amounts of data, mostly from functional connectivity MRI (fcMRI) and diffusion tensor imaging (DTI), published in the past decade support abnormalities of functional and anatomical network connectivity in ASD. While quantitatively reports of underconnectivity predominate, empirical inconsistencies, methodological challenges, and conceptual issues today highlight the need for a more nuanced interpretation of findings.

Objectives:   To outline a developmental model reconciling some of the most problematic inconsistencies in the fcMRI and DTI literature on ASD, based on methodological and conceptual considerations.

Methods:   Exemplary findings from literature and from own fcMRI and DTI studies (including total cohorts of over 60 children and adolescents with ASD, ages 7-18 years, and an age, sex, handedness and nonverbal IQ-matched typically developing [TD] cohort) will be reviewed.

Results:  FCMRI findings in ASD range from ‘general underconnectivity’ to ‘general overconnectivity’ across different studies. Factors such as participant age, head motion, and effects of global signal fluctuations have been debated, but are unlikely to provide an explanation of the inconsistencies. Other methodological factors, in particular those relating to the difference between task-driven vs. intrinsic signal correlations, account for larger amounts of variability in between-group effects, based on a literature survey as well as empirical demonstration in three independent datasets including >100 participants.

     DTI results for ASD show a different pattern of inconsistencies. While findings indicating white matter compromise (e.g., reduced fractional anisotropy [FA], increased mean diffusion [MD]) predominate, regional or tract-specific patterns have varied much across studies. There are also notable exceptions of studies with inverse effects (e.g., increased FA in ASD). Most of these included infants and children under the age of 8 years.

Conclusions:   Given relatively firm evidence of early with matter overgrowth in ASD, the DTI evidence from infants suggests that at some point in the first two years of life, at least some parameters of anatomical network organization (e.g., axonal numbers, myelination) progress precociously and out of step with experiential input available from emerging sensorimotor systems. This asynchrony subsequently dampens progressive and regressive mechanisms of neurotypical plasticity and the emergence of distinctly sculpted functional networks, with the result of inefficient (‘noisy’) information transfer. FCMRI can detect both aspects of developmental network abnormalities in ASD, with underconnectivity within neurotypical networks reflecting impaired progressive and diffuse (out-of-network) overconnectivity reflecting impaired regressive mechanisms. DTI findings in participants >8 years of age further support impaired progressive mechanisms. However, the technique’s currently limited ability to resolve multiple fiber directions within a voxel make conventional DTI protocols largely insensitive to the detection of residual excessive overconnectivity.