Imaging Radial Cortical Anisotropy to Measure Microstructure of the Cortex in Autism: A Novel Method for the Detection of Early Brain Changes

Background: Theories of autism have hypothesised early developmental brain over-growth, altered cortical micro-circuitry with excessive packing of minicolumns, and loss of inhibitory architecture. The differences in micro-circuitry relate to differences in cognition, for example, altered social cognition appears to be linked to differences in fusiform face processing cortex. Minicolumn structure has also been associated with hemispheric asymmetry of language function. However, seeking to explain altered cognitive functions such as face processing and language at the microscopic scale is a challenge given the current limits of brain imaging. For much of its history MRI has been concerned with the detection of volumetric differences but the cortex is not an undifferentiated, homogenous network; it consists of multiple, columnar, structural units that may constitute micro-circuits. Past studies have shown that developmental changes in DTI signal (radial cortical anisotropy) appear to correspond to the developmental expansion of these micro-circuits. Our studies of post-mortem histology have revealed a cortical signature, based on multi-regional comparison that is sensitive to early signs of pathology.

Objectives: We report here on a novel application of diffusion tensor imaging (DTI) to cerebral cortex for comparison with measurements of microstructure in post-mortem brains.

Methods: Structural MRIs, including diffusion weighted images, were acquired from a series of post-mortem human brains. The tissue collection comprised brains donated by controls (4 subjects) and two neurological/psychiatric conditions: ASD (4 subjects) and multiple sclerosis (MS) (9 subjects). The scanning protocol requires long scans with sequences devised specifically for fixed post-mortem tissue broadly as described in Miller et al (2011). The analysis of radial cortical anisotropy used the novel ‘CHIPS’ software developed for the purpose (FMRIB, Oxford, UK). After scanning, tissue samples were dissected from several regions and cryosectioned to provide Nissl stained slides for minicolumn estimation. The minicolumn analysis was similar to that described in Chance et al (2011) and provided data on minicolumn centre-to-centre spacing.

Results: Multi-region comparisons typically indicated wide spacing of minicolumns in prefrontal cortex, intermediate spacing in temporal lobe auditory association cortex (including the planum temporale which has been associated with language asymmetry) and narrow spacing in primary auditory and visual cortices. Diffusion measures of the cortex from the CHIPS analysis were significantly correlated with the minicolumn centre-to-centre spacing across both primary visual and prefrontal regions, providing an MRI index of minicolumn organisation in human brain.

Conclusions: Validation of this technique raises the prospect of measurement of (i) much more subtle cortical changes than current volumetric methods, and (ii) brain region differentiation at the microstructural level. Successful translation of this technique to in vivo imaging will aid detection and assessment of autism and other disorders.  References: Miller et al. Neuroimage 57(1):167-81 (2011).  Chance et al. Cerebral Cortex 21(8):1870-8 (2011)