Intellectual abilities and white matter microstructure in development: a diffusion tensor imaging study

Abstract

Higher-order cognitive functions are supported by distributed networks of multiple interconnected cortical and subcortical regions. Efficient cognitive processing depends on fast communication between these regions, so the integrity of the connections between them is of great importance. It is known that white matter (WM) development is a slow process, continuing into adulthood. While the significance of cortical maturation for intellectual development is described, less is known about the relationships between cognitive functions and maturation of WM connectivity. In this cross-sectional study, we investigated the associations between intellectual abilities and development of diffusion tensor imaging (DTI) derived measures of WM microstructure in 168 right-handed participants aged 8-30 years. Independently of age and sex, both verbal and performance abilities were positively related to fractional anisotropy (FA) and negatively related to mean diffusivity (MD) and radial diffusivity (RD), predominantly in the left hemisphere. Further, verbal, but not performance abilities, were associated with developmental differences in DTI indices in widespread regions in both hemispheres. Regional analyses showed relations with both FA and RD bilaterally in the anterior thalamic radiation and the cortico-spinal tract and in the right superior longitudinal fasciculus. In these regions, our results suggest that participants with high verbal abilities may show accelerated WM development in late childhood and a subsequent earlier developmental plateau, in contrast to a steadier and prolonged development in participants with average verbal abilities. Longitudinal data are needed to validate these interpretations. The results provide insight into the neurobiological underpinnings of intellectual development.

Figure 1

Horizontal slices of DTI data from two participants. (A) FA map and color‐coded vectormap from a representative 10‐year‐old male. Images are shown in original resolution to give a sense of data quality. (B) The same maps from a representative 29‐year‐old male. (C) The TBSS mean FA skeleton for the entire sample, overlaid on the standard MNI152 T1 template (top) and on the mean FA map of the entire sample (bottom). The skeleton is thresholded at FA > 0.25. All images are shown in radiological convention.

Figure 2

Age‐related changes in intellectual abilities and WM microstructure. Z‐transformed raw verbal and performance abilities scores and average fractional anisotropy (FA), mean diffusivity (MD), radial diffusivity (RD), and axial diffusivity (AD) from the skeleton voxels plotted as a function of age. Both linear and exponential models were fitted and explained variance of both models is shown in the figure.

Figure 3

Relationships between verbal abilities and WM microstructure. Results from GLMs testing linear effects of verbal abilities on fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD), while regressing out the effects of age and sex. Results are shown as 3D‐renderings in anterior, dorsal, and left views of the statistical P‐maps of the TBSS skeleton, displayed on a semitransparent template brain from FreeSurfer (fsaverage). Red areas indicate voxels with positive significant relationships between verbal abilities and FA and negative significant relationships between verbal abilities and MD and RD (P < 0.05, corrected for multiple comparisons across space). In total, FA, MD, and RD were correlated with verbal abilities in 4.6, 18.2, and 15.0% of the skeleton voxels, respectively.

Figure 4

Relationships between performance abilities and WM microstructure. Results from GLMs testing linear effects of performance abilities on fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD), while regressing out the effects of age and sex. Red areas indicate voxels with positive significant relationships between performance abilities and FA and negative significant relationships between performance abilities and MD and RD (P < 0.05, corrected for multiple comparisons across space). In total, FA, MD, and RD were correlated with performance abilities in 1.6, 9.5, and 11.0% of the skeleton voxels, respectively.

Figure 5

Relationships between verbal abilities and development of WM microstructure. Results from GLMs testing linear unique effects of the interaction term verbal abilities × age on fractional anisotropy (FA), mean diffusivity (MD), and radial diffusivity (RD), while regressing out the effects of age, sex, and verbal abilities. Red areas indicate voxels with negative significant relationships between the interaction term and FA and positive significant relationships between the interaction term and MD and RD (P < 0.05, corrected for multiple comparisons across space). In total, FA, MD, and RD were correlated with verbal abilities × age in 18.8, 32.3, and 31.9% of the skeleton voxels, respectively.

Figure 6

Regional relationships between verbal abilities and development of WM microstructure: Fractional anisotropy (FA). Individual raw FA from atlas tracts and skeleton overlaps are plotted as a function of age. For visualization purposes, the sample was split into two based on raw verbal abilities scores. Group assignment was done for each year separately in the age range 8–20 and for each 2 years in the age range 21–30 and by splitting by the median within these brackets. Exponential models were then fitted for both groups (see Methods for details). The average verbal abilities group is shown in blue, and the high verbal abilities group in red. Only TOIs with significant contributions of the continuous interaction term verbal abilities × age in previous multiple regressions are shown. TOIs shown are: anterior thalamic radiation (ATR), cortico‐spinal tract (CST), inferior longitudinal fasciculus (ILF), and superior longitudinal fasciculus (SLF).

Figure 7

Regional relationships between verbal abilities and development of WM microstructure: radial diffusivity (RD). Individual raw RD from atlas tracts and skeleton overlaps plotted as a function of age. For visualization purposes, the sample was split in two based on raw verbal abilities scores. Group assignment was done for each year separately in the age range 8–20 and for each 2 years in the age range 21–30 and by splitting by the median within these brackets. Exponential models were then fitted for both groups (see Methods for details). The average verbal abilities group is shown in blue, and the high verbal abilities group in red. Only TOIs with significant contributions of the continuous interaction term verbal abilities × age in previous multiple regressions are shown. TOIs shown are: anterior thalamic radiation (ATR), cortico‐spinal tract (CST), cingulum‐cingulate gyrus (CG), superior longitudinal fasciculus (SLF), and cingulum‐hippocampus gyrus (CH).