Gray matter microstructure differences in autistic males: A gray matter based spatial statistics study

Abstract

Background: Autism spectrum disorder (ASD) is a complex neurodevelopmental condition. Understanding the brain's microstructure and its relationship to clinical characteristics is important to advance our understanding of the neural supports underlying ASD. In the current work, we implemented Gray-Matter Based Spatial Statistics (GBSS) to examine and characterize cortical microstructure and assess differences between typically developing (TD) and autistic males.

Methods: A multi-shell diffusion MRI (dMRI) protocol was acquired from 83 TD and 70 autistic males (5-to-21-years) and fit to the DTI and NODDI models. GBSS was performed for voxelwise analysis of cortical gray matter (GM). General linear models were used to investigate group differences, while age-by-group interactions assessed age-related differences between groups. Within the ASD group, relationships between cortical microstructure and measures of autistic symptoms were investigated.

Results: All dMRI measures were significantly associated with age across the GM skeleton. Group differences and age-by-group interactions are reported. Group-wise increases in neurite density in autistic individuals were observed across frontal, temporal, and occipital regions of the right hemisphere. Significant age-by-group interactions of neurite density were observed within the middle frontal gyrus, precentral gyrus, and frontal pole. Negative relationships between neurite dispersion and the ADOS-2 Calibrated Severity Scores (CSS) were observed within the ASD group.

Discussion: Findings demonstrate group and age-related differences between groups in neurite density in ASD across right-hemisphere brain regions supporting cognitive processes. Results provide evidence of altered neurodevelopmental processes affecting GM microstructure in autistic males with implications for the role of cortical microstructure in the level of autistic symptoms.

Conclusion: Using dMRI and GBSS, our findings provide new insights into group and age-related differences of the GM microstructure in autistic males. Defining where and when these cortical GM differences arise will contribute to our understanding of brain-behavior relationships of ASD and may aid in the development and monitoring of targeted and individualized interventions.

Keywords: Adolescence; Autism; Childhood; DTI; GBSS; Gray matter microstructure; NODDI.

Fig. 1

GBSS GM skeleton projected on study-specific template. GM fraction maps were first averaged across subjects and the mean GM image was skeletonized (red). All DTI and NODDI metrics and GM fraction were projected onto the skeleton from the local GM fraction maxima. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2

Cortical GM microstructure age relationships. Logarithmic and linear fit lines applied per BIC and AIC model selection in Table 2. Scatter points represent mean dMRI measures across the GM skeleton shown in Fig. 1. Bands represent confidence intervals.

Fig. 3

Voxel-based age relationships of cortical GM microstructure. Logarithmic and linear fits applied per BIC and AIC model selection in Table 2. Significant positive (Yellow/Red) and negative (Light blue/Dark Blue) voxels are shown on the dMRI maps for each measure. Voxels showing significant negative relationships are shown in blue while voxels showing significant positive relationships are shown in red. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Group differences in NODDI and DTI measures. Level of significance and neuroanatomical location of voxels from group difference model are displayed on the mean dMRI maps from all participants. Scatter points represent mean dMRI values of significant voxels for each measure. Trendlines show model fit and confidence intervals for group difference on dMRI measures when accounting for age.

Fig. 5

NODDI FICVF age by group interactions. Level of significance and neuroanatomical location of voxels from interaction model are displayed on the mean FICVF map across all participants. Scatter plots represent mean FICVF values of significant voxels from each individual. Trendlines show model fit and confidence intervals for age by group interaction on dMRI measures when accounting for the effects of age and group.

Fig. 6

ADOS-2 CSS by DWI Relationships in ASD Accounting for IQ as a Covariate. Level of significance and neuroanatomical location of significant voxels are displayed on the dMRI maps. Scatter plots represent mean dMRI values of significant voxels for each measure. Trendlines show model fit and confidence intervals for relationship between dMRI measures and ADOS CSS when accounting for the effects of age and IQ. Boxplots show TD range for DWI measures averaged over significant voxels.