Basal Ganglia Shape Predicts Social and Motor Dysfunction in Boys with Autism

Background: Autism is associated with abnormalities suggestive of basal ganglia dysfunction, including stereotyped behavior and abnormalities on motor examination.  Neuroimaging investigations of basal ganglia structure have thus far been limited to examination of whole volumes.  Large deformation diffeomorphic metric mapping (LDDMM), a powerful computational tool used for detailed analysis of morphology (i.e., shape), can provide a more comprehensive understanding of basal ganglia contributions to the behavioral features of autism.

Objectives: To apply LDDMM to detailed analysis of basal ganglia structure and its association with the motor and behavioral features of autism. 

Methods: Basal ganglia structures (caudate, putamen, globus pallidus) were manually delineated on high-resolution MR images in 32 boys with high-functioning autism (HFA) and 45 typically developing (TD) controls, ages 8-12 years.  LDDMM was then used to map between-group differences in shape across each structure and to examine correlations with measures of basic motor skill (assessed using the PANESS – Physical and Neurologic Examination of Subtle Signs), praxis (assessed using a standardized videotaped praxis examination), and core social/communicative/behavior features of autism (assessed using the ADOS-G, Module 3). We examined the group shape differences using principal component analysis and correlation of the basal ganglia shape with behavioral measures using point-wise Pearson’s correlation analysis. The overall significance was confirmed via permutation tests.

Results: Group shape differences were principally localized to the right striatum: compared to TD boys boys with HFA showed significant compression, in the right anterior-ventral and posterior-dorsal putamen and in the right anterior caudate and caudate tail; they showed significant expansion in the mid-dorsal putamen, middle caudate, and posterior globus pallidus.  Behavioral analyses revealed the HFA group showed significantly poorer performance on the PANESS (p<0.001) and praxis examination (p<0.001).  Brain-behavior correlations revealed that, across the groups of subjects, compression of the bilateral posterior putamen was a significant predictor of poorer performance on the PANESS, while compression of the bilateral anterior putamen was a significant predictor of poorer praxis performance; in both cases these correlations were driven by significant findings in the HFA group (but not controls).  For the HFA group, bilateral expansion in the ventral caudate head (nucleus accumbens) was a significant predictor of higher ADOS social domain scores.

Conclusions: For boys with HFA, shape abnormalities in the basal ganglia are associated with motor and social dysfunction.  Consistent with established mapping of striatal function, basic motor skill impairment was associated with compression in posterior putaminal regions in circuit with primary sensorimotor cortex, while impairment in performance of skilled gestures (assessed on praxis examination) was associated with compression in dorsal anterior putaminal regions in circuit with premotor cortex.  Impaired social function characteristic of autism was associated with expansion in the nucleus accumbens, in circuit with orbitofrontal cortex.  The pattern of findings suggests that abnormalities within parallel subcortical circuits may contribute to the motor, social and behavioral features of autism.  The detailed level analysis offered by LDDMM may therefore prove valuable in identifying neuroanatomic biomarkers of autism.