Disruption of Functional Organization within the Primary Motor Cortex in Children with Autism

Background: Clinical observation and empirical studies suggest that children with autism spectrum disorder (ASD) exhibit impairments in motor abilities that may reflect abnormal connectivity within networks underlying motor control and learning. Recently, several groups have used patterns of correlations in spontaneous BOLD activity, referred to as resting-state functional connectivity (FC), to localize functionally relevant subdivisions of anatomically defined regions.

Objectives: Motivated by the utility of these methods in establishing functional organization, this study had two aims: (1) parcellate a key area of the motor network, the precentral gyrus (M1), in neurotypical adults and typically developing (TD) children; (2) apply this approach to children with ASD to determine if M1 functional organization differs in children with ASD compared to TD children.

Methods: Resting state fMRI and anatomical images were collected from 32 children with ASD (8-12 years) and 33 TD children balanced for age, gender, handedness and perceptual reasoning. Scan-rescan reliability data from twenty neurotypical adults were also used for parameter estimation. Initial preprocessing steps of slice time adjustment, motion correction and spatial normalization were performed using SPM5, after which motion parameters, global signal and nuisance covariates were extracted. The data were then temporally filtered (0.01-0.1 Hz pass-band).

For each M1 voxel, an FC map with all voxels outside of M1 was generated. Similarity of FC maps for every pair of M1 voxels was computed using η², and a spectral clustering algorithm was applied to the resulting η² matrix. Adjacency matrices were constructed for each participant and averaged to generate a consensus matrix for each group; the elements of the consensus matrix corresponded to the proportion of times a given pair of voxels was assigned to the same parcel across participants. Spectral clustering was reapplied to the group consensus matrices, and the labels of the clustering solutions were matched by maximizing the average overlap of similarly-labeled parcels between groups.

TD and ASD parcels were compared by volume and spatial overlap. The significance of the observed group differences was assessed by permutation: diagnosis labels were randomly assigned, consensus matrices were recalculated, and second-level spectral clustering was reapplied 10,000 times.

Results: A gross dorsomedial to ventrolateral organization emerged bilaterally within M1 in both groups. However, the dorsomedial-most (DM) parcel was significantly larger in ASDs than in TDs and encompassed much of the space occupied by the adjacent dorsolateral (DL) parcel in TDs (+10.6 cm³, p=.05). The displacement of the DL parcel in ASD caused additional functional segregation differences between groups in the region of M1 closest to the hand knob; group overlaps for the DL and anterior lateral parcels were significantly worse than predicted by permutation (.24, p=.02; .51, p=.02, respectively).

Conclusions: Given the organization of the motor homunculus, the observed differences between M1 functional subunits in TDs and ASDs may have interesting developmental implications. The enlarged ASD parcel included regions of M1 normally reserved for lower limb control but also areas normally recruited by the upper limbs, suggesting that developmental segregation of upper and lower limb control may be delayed in ASD.