If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Division of Developmental Cognitive Neuroscience, Institute of Development, Aging and Cancer, Tohoku University, Sendai, JapanDivision of Medical Neuroimaging Analysis, Department of Community Medical Supports, Tohoku Medical Megabank Organization, Tohoku University, Sendai, JapanDepartment of Nuclear Medicine & Radiology, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan
Division of Developmental Cognitive Neuroscience, Institute of Development, Aging and Cancer, Tohoku University, Sendai, JapanSmart Ageing International Research Centre, Institute of Development, Aging and Cancer, Tohoku University, Sendai, Japan
Childhood is an extremely important time for neural development that has a critical role in human intelligence. Efficient information processing is crucial for higher intelligence, so the intra- or inter-hemispheric interaction is vital. However, the relationship between neuroanatomical connections and intelligence in typically developing children, as well as sex differences in this relationship, remains unknown.
Methods
Participants were 253 typically developing children (121 boys and 132 girls) aged 5–18. We acquired diffusion tensor imaging data and intelligence using an age-appropriate version of the IQ test; Wechsler Intelligence Scale for Children (WISC) or Wechsler Adult Intelligence Scale (WAIS). We conducted whole-brain multiple regression analysis to investigate the association between fractional anisotropy (FA), which reflects white matter microstructural properties, and each composite score of IQ test (full-scale IQ, performance IQ, and verbal IQ).
Results
FA was positively correlated with full-scale IQ in bilateral inferior occipitofrontal fasciculus, genu, and splenium of corpus callosum (CC). FA in the right superior longitudinal fasciculus, bilateral inferior longitudinal fasciculus, and splenium of CC were also positively correlated with performance IQ. Furthermore, we found significant sex interaction between FA in the CC and verbal IQ. FA was positively correlated in boys, and negatively correlated in girls.
Conclusion
Results suggest that efficient anatomical connectivity between parietal and frontal regions is crucial for children’s intelligence. Moreover, inter-hemispheric connections play a critical role in verbal abilities in boys.
Intelligence has played a significant role for human development. MRI studies have helped elucidate the relationship between human intelligence versus brain structure and function in children and adolescents. The development of brain structures, including both gray and white matter, is an essential process of neural development in childhood and plays an important role in cognitive function. Several previous studies have revealed that level of intelligence was correlated with brain structures, such as total brain volume, cortical thickness, or gray matter volume [
Diffusion tensor imaging (DTI) is a powerful modality for investigating white matter microstructure in vivo. This technique makes it possible to track brain white matter development. Previous studies have revealed white matter microstructural changes in development using metrics like fractional anisotropy (FA) or mean diffusivity (MD). Brain development during late childhood and adolescence is characterized by decreasing gray matter and increasing white matter volume [
]. For example, the corpus callosum (CC) and fornix develop earlier than association fibers such as the arcuate fasciculus or superior longitudinal fasciculus [
], which could explain subtle cognitive differences.
Whilst the relationship between white matter microstructure and cognitive functions are understudied, previous research has revealed that gray matter metrics such as volume or cortical thickness in frontal and parietal or temporal regions are commonly correlated with intelligence [
] found a positive correlation between FA in frontal and occipito-parietal areas and full-scale IQ (FSIQ) in children aged 5–18. More recently, Muetzel, Mous [
] also found a positive correlation between white matter microstructure and FSIQ in the right uncinate fasciculus in a larger cohort of 6- to 10-year-old children. Chiang, McMahon [
] investigated genetic influence over white matter structure (FA) in 705 children and adolescence twins and their siblings. They found significant association between heritability and intellectual performance in the thalamus, genu and posterior limb of internal capsule, and corona radiata. Regarding sex differences in the relationship between cognitive function and white matter microstructure there was a positive correlation between verbal IQ, which reflects verbal abilities such as verbal reasoning, vocabulary, or verbal working memory, and FA in girls in the superior longitudinal fasciculus (SLF), inferior fronto-occipital fasciculus, and cortico-spinal tract, though the sample size for this study was relatively small [
These studies have clarified the relationship between FA and general cognitive ability. However, there are few previous studies examining the relationship between FA and more specific cognitive abilities such as non-verbal visuospatial abilities, or verbal abilities, and the interaction with sex using larger samples.
This study investigated the relationship between white matter microstructure and cognitive abilities, also considering sex difference in those relationships in a large number of typically developing children. Considering parieto-frontal integration theory of intelligence [
], we hypothesized that FA in the SLF, which connects frontal and parietal regions, would correlate with cognitive functions. We also predicted that there would be sex differences in these associations.
2. Method
2.1 Participants
We acquired brain MRI images from 298 typically developing Japanese children (152 boys and 146 girls; age range, 5.6–18.4 years). The details relating to their initial recruitment were described in our previous study [
]. In brief, we recruited only right-handed children who did not have any neurodevelopmental disorders, history of malignant tumors or head traumas involving loss of consciousness using an advertisement in local schools. We also confirmed that all subjects were right-handed using the self-report questionnaire, the “Edinburgh Handedness Inventory” [
]. After we had explained the purpose and procedures of the study, written informed consent was obtained from each subject and a parent prior to participation in this study. Because DTI data were obtained from only parts of the entire subjects, the imaging analyses were performed with 253 subjects. Final sample comprised 121 boys (mean age = 11.2 ± 2.8 years old) and 132 girls (mean age = 11.9 ± 3.3 years old). This study was approved by the Institutional Review Board of Tohoku University and conducted in accordance to the Declaration of Helsinki (1991).
2.2 Cognitive measures
On the same day as the MRI scan, we measured intelligence using an age-appropriate version of an IQ test (Wechsler Intelligence Scale for Children (WISC)-III for participants whose age was under 16 years old, Wechsler Adult Intelligence Scale (WAIS)-III for participants whose age was over 16 years). Trained examiners administered these tests. Three IQ scores: FSIQ, Verbal IQ (VIQ), and Performance IQ (PIQ), were computed according to the standard formulae. See Table 1 for a summary of these data.
MRI data acquisition was conducted using a 3T Philips Achieva scanner (Royal Philips, Amsterdam, Nederland). Using a spin-echo echo planer imaging (EPI) sequence (repetition time = 10,293 ms, echo time = 55 ms, big delta (Δ) = 26.3 ms, little delta (δ) = 12.2 ms, field of view = 22.4 cm, 2 × 2 × 2 mm3 voxels, 60 slices, SENSE reduction factor = 2, number of acquisitions = 1), diffusion-weighted data were collected. The diffusion weighting was isotropically distributed along 32 directions (b value = 1000 s/mm2). Additionally, a single image with no diffusion weighting (b value = 0 s/mm2; b0 image) was acquired. The total scan time was 7 min 17 s. Acquisitions for phase correction and signal stabilization were not used as reconstructed images. FA maps were calculated from the collected images using a commercially available diffusion tensor analysis package on the MR console as has been described previously [
]. These procedures involved correction for motion and distortion caused by eddy currents. Calculations were performed according to a previously proposed method [
Three-dimensional (3D) high-resolution T1-weighted images (T1WI) were also collected using a magnetization-prepared rapid gradient-echo (MPRAGE) sequence. The parameters are as follows: 240 × 240 matrix, repetition time = 6.5 ms, echo time = 3 ms, inversion time = 711 ms, field of view = 24 cm, 162 slices, 1.0 mm slice thickness, and scan duration of 8 min and 3 s. Although we did not collect any clinical brain MRI sequences due to scan time constraints for young children, a radiologist (YT) had checked potential brain lesions using T1WIs.
2.4 Image preprocessing
Preprocessing and analysis of MRI data were performed using statistical parametric mapping 8 (SPM8; Wellcome Department of Cognitive Neurology, London, UK) implemented in Matlab (Mathworks Inc., Natick, MA, USA). The skull of the unsmoothed b = 0 images of all the subjects in this study was stripped by masking the images using the threshold of a given signal intensity from spatially smoothed (using 8 mm Full Width at Half Maxim; FWHM) b = 0 images of each participant. Then, using the coregister option, this skull-stripped unsmoothed b = 0 image was coregistered to a skull-stripped b = 0 image template that was created previously [
]. Using this parameter, other diffusion imaging data were aligned to the template, too.
Subsequently, using a previously validated two-step new segmentation algorithm of diffusion images and the previously validated diffeomorphic anatomical registration through exponentiated lie algebra (DARTEL)-based registration process [
], all images, including gray matter segments [regional gray matter density (rGMD) map], white matter segments [regional white matter density (rWMD) map], and cerebrospinal fluid (CSF) segments [regional CSF density (rCSFD) map] of diffusion images, were normalized. The voxel size of these normalized images was 1.5 × 1.5 × 1.5 mm3.
Next, from the average image of the normalized white matter segmentation (rWMD) images of all the subjects, we created a mask image consisting of voxels with a white matter signal intensity > 0.99. We then applied this mask image to the normalized FA image, therefore retaining only areas that were highly likely to contain white matter from the normalized FA images. These images were smoothed (6 mm FWHM) and carried through to second-level analyses of FA. The description in this subsection has been mostly reproduced from our previous study using an identical methodology [
]. By taking FA signal variabilities within the white matter areas into account in the DARTEL procedures, the misalignment of the tracts was prevented using stringent masking. Signal contamination from other tissues was prevented thereby effectively solving or alleviating the major problems of voxel based analyses of FA images [
T1WIs were used to calculate total intracranial volume. Using the new segmentation algorithm implemented in SPM8, T1WIs of each individual were segmented into six tissues. Then we summed segmented gray matter, white matter, and cerebral spinal fluid volume as total intracranial volume which was treated as covariate in following analyses.
2.5 Data analysis
2.5.1 Behavioral data analyses
Behavioral data analyses were performed using statistical analysis software package, SPSS version 20 (IBM, Japan). We examined sex differences across age, FSIQ, VIQ and PIQ. Because age and PIQ displayed non-parametric distributions, we tested for sex differences using Mann-Whitney U tests. T-tests were performed for the FSIQ and VIQ as these displayed parametric distributions. The statistical threshold was set at P < 0.05, with false discovery rate (FDR) correction. The effect size for each statistical tests was calculated using Cohen’s d for T-tests and r for Mann-Whitney U tests.
2.5.2 Brain imaging data analyses
We conducted three separate whole-brain multiple regression analyses to investigate the association between FA and each composite score of IQ test (FSIQ, VIQ, or PIQ). Sex and age (days after birth) were assigned as covariates. Next, we investigated the interaction between sex and FA and IQ composite scores. To examine the effect, age, and one of three composite scores were modeled so that each covariate had a unique relationship with FA for each sex (using the interaction option in SPM), which enabled investigation of the effects of interactions between sex and each covariate. Due to there is a known association between brain volume and IQ or diffusion parameters [
], total intracranial volume was also assigned as a covariate in all regression analyses.
A multiple comparison correction was performed using threshold-free cluster enhancement (TFCE) with randomized (5,000 permutations) non-parametric testing using the TFCE toolbox (https://dbm.neuro.uni-jena.de/tfce/). We applied a threshold of family-wise error (FWE) corrected P < 0.05. In order to specify the tracts from the coordinates for significant clusters, we used the International Consortium for Brain Mapping (ICBM) DTI-81 Atlas (https://www.loni.usc.edu/).
3. Results
3.1 Sample characteristics
The sample characteristic of age is illustrated in Fig. 1. The means and standard deviations for age and IQ scores (FSIQ, VIQ, and PIQ) are presented in Table 1. We did not find any significant sex differences in these values (age: U = 7034.5, p (uncorrected) = 0.04, p (FDR) = 0.2, r = 0.1; FSIQ; t (251) = 2.17, p (uncorrected) = 0.031, p (FDR) = 0.124, d = 0.27; VIQ: t (251) = 2.18, p (uncorrected) = 0.031, p (FDR) = 0.124, d = 0.27; PIQ: U = 6794, p (uncorrected) = 0.1, p (FDR) = 0.053, r = 0.13).
Fig. 1Distribution of age in years in (a) boys and (b) girls.
We found a significant positive correlation between FSIQ and FA in bilateral inferior occipitofrontal fasciculus, genu and splenium of the CC (Table 2, Fig. 2a). Moreover, PIQ was positively correlated with FA in the right SLF, bilateral inferior longitudinal fasciculus (ILF), and splenium of the CC (Table 2, Fig. 2b). We did not find any correlation between VIQ and FA in whole group analyses.
Table 2Clusters with positive correlations of FA with Full-Scale IQ and Performance IQ over the whole cohort.
White matter location
voxels
MNI coordinates
TFCE value
p
x (mm)
y (mm)
z(mm)
FSIQ
Genu of corpus callosum
154
24
30
8
490.89
0.018
Splenium of corpus callosum
38
−21
−51
14
435.72
0.033
Right Inferior occipitofrontal fasciculus
36
30
−68
8
417.35
0.049
Left inferior occipitofrontal fasciculus
79
−38
−46
−6
434.93
0.033
PIQ
Right superior longitudinal fasciculus
1193
38
−8
26
553.68
0.011
Right Inferior longitudinal fasciculus
403
32
−21
−8
475.86
0.023
Splenium of corpus callosum
819
−20
−51
12
480.25
0.022
Left inferior longitudinal fasciculus
78
−34
−32
3
466.2
0.025
376
−38
−46
−8
460.75
0.026
p: TFCE-corrected p value.
FSIQ: full-scale IQ, MNI: Montreal Neurological Institute, PIQ: performance IQ, TFCE: Threshold Free Cluster Enhancement.
Fig. 2Positive FA correlates of Full-Scale IQ and performance IQ in the whole-group analysis. (a) FA values in the genu and splenium of the corpus callosum and bilateral occipito-frontal fasciculus were positively correlated with Full-Scale IQ. (b) FA values in right superior longitudinal fasciculus, bilateral inferior longitudinal fasciculus, and splenium of the corpus callosum were positively correlated with Performance IQ. The result was obtained using Threshold Free Cluster Enhancement, P < 0.05 based on 5,000 permutations.
We found a significant interaction between sex and VIQ in the CC (TFCE value = 483.9, TFCE corrected P value = 0.02, cluster size = 105 voxels, Montreal Neurological Institute (MNI) coordinate; x = −9 mm, y = 0 mm, z = 27 mm, Fig. 3). FA in this cluster was positively correlated with VIQ score in boys and negatively correlated in girls. We did not find any significant sex interaction between other composite scores (FSIQ and PIQ) and FA.
Fig. 3Location of brain region where a sex interaction was seen between verbal IQ and FA. (a) There was a significant positive correlation in boys and negative correlation in girls in the corpus callosum. This result was obtained using Threshold Free Cluster Enhancement, P < 0.05 based on 5,000 permutations. (b, c) Scatter plots depicting the correlation between verbal IQ score and parameter estimates of the significant cluster in the corpus callosum in boys (b) and girls (c).
In order to check the possibility that both of the VIQ and FA correlated with the age, we calculated Pearson’s correlation coefficients. We did not find any significant correlations (VIQ and age: r = 0.075, p = 0.23, FA and age: r = 0.007, p = 0.91, Fig. 4).
Fig. 4Scatter plots of (a)verbal IQ and age, (b)parameter estimates of significant cluster in corpus callosum and age in whole group.
In the present study, we explored the relationship between intelligence and white matter microstructure in typically developing children, also considering interactions with sex. Our hypothesis that FA in the SLF would correlate with cognitive functions was partially accepted. PIQ score correlated with FA mainly in the right SLF. Additionally, we found that in the whole sample, FSIQ score correlated with FA in the bilateral occipitofrontal fasciculus, genu and splenium of the CC. Moreover, consistent with our hypothesis, we found a significant sex interaction between FA in the CC and VIQ score.
The findings of a positive correlation between FA in CC and bilateral occipitofrontal fasciculus and FSIQ were largely consistent with previous researches. The CC is the most important structure for inter-hemispheric connectivity [
]. As we found a positive correlation between these fibers and FSIQ score, it suggests that efficient anatomical inter-hemispheric connections and associations between frontal and occipital regions play a great role in human intelligence. In line with this interpretation, previous studies on brain structure and cognitive function have indicated that gray or white matter metrics in specifically inferior frontal gyrus, ventro- and dorsolateral prefrontal cortex, and parietal regions have been associated with higher intelligence [
]. According to this theory, sensory information is processed by temporal and occipital regions and integrated within parietal regions. The frontal regions such as anterior cingulate cortex and dorsolateral prefrontal regions are associated with higher order processing. Previous studies on cortical thickness [
Additionally, previous studies that investigated the association of brain and cognitive development using multiple imaging modalities, including cortical thickness, white matter volume and microstructure, have shown that cortical thickness and mean diffusivity (MD) (which reflects average magnitude of water diffusion) showed negative correlations, and FA and white matter volume showed positive correlations with intellectual abilities [
]. From these previous findings, regional gray matter metrics (volume or cortical thickness) and anatomical connectivity (which is measured by FA) seem to be closely related. In line with these findings, our results indicate that it is not simply regional gray matter volume in posterior and frontal regions that are crucial for intelligence, but also efficient anatomical connections between the two.
With regard to the positive correlation between FA and PIQ, previous study found that visuospatial ability was associated with global white matter microstructure, but they did not find any significant associations between certain fiber tracts and cognitive abilities [
]. Extending this result, we found that the white matter properties of certain fibers were associated with cognitive function. In this study, we found a significant positive correlation between PIQ and FA mainly in the right SLF. PIQ reflects visuospatial processing ability, response speed, or visuomotor association [
]. SLF is a major association fiber pathway that connects postrolandic regions and frontal lobe. The SLF is separated into a dorsal SLF I, middle SLF II, ventral SLF III, and arcuate fascicle [
]. The significant regions, with which we observed a positive correlation with PIQ in this study, mainly overlapped the ventral SLF III. This pathway originates from the supramarginal gyrus and terminates in ventral premotor and prefrontal areas [
]. As for the functional role of SLF, since this pathway connects the rostral part of the inferior parietal lobule with lateral inferior frontal lobe in a bidirectional way [
Connections of the ventral granular frontal cortex of macaques with perisylvian premotor and somatosensory areas: anatomical evidence for somatic representation in primate frontal association cortex.
]. Therefore, it is quite plausible that the larger FA value in the right ventral SFL was associated with higher PIQ score.
Our second finding was the interaction between sex and VIQ and FA. We observed a significant positive correlation in boys and a negative correlation in girls between VIQ and FA in the CC. These findings indicate that efficient information transfer between hemispheres plays a critical role in higher cognitive abilities. The CC shows a protracted development pattern that continues into adolescence [
]. Previous studies on sexual dimorphism of the CC have revealed that males showed a higher callosal anisotropy than females in anterior and posterior genu, as well as in the truncus region [
Tractography-guided statistics (TGIS) in diffusion tensor imaging for the detection of gender difference of fiber integrity in the midsagittal and parasagittal corpora callosa.
], the faster and stronger connectivity in males might have seen as the results of an adjustment for the longer distance of inter-hemisphere connectivity [
Developmental stages and sex differences of white matter and behavioral development through adolescence: a longitudinal diffusion tensor imaging (DTI) study.
], we did not find any positive correlation between FA and IQ indices in girls, which may be attributable to differences in the subjects’ age range, sample size or methods of the DTI analysis. The second limitation regards the DTI technique itself. Since the specific neurobiological meaning of DTI parameters is unclear, differences in FA are mostly thought to be the result of differences in fiber organization, but they could also be related to myelination, fiber density, axonal diameter, and ratio of intracellular/extracellular space. Multiple imaging modalities combined could help to delineate the association between brain development and cognitive abilities. Third, the present study was cross-sectional. To capture development of brain structure and cognitive ability, longitudinal data are needed to validate the results. Fourth, we did not consider the positive (such as studying, reading, or exercising) and negative (maltreatment from parents, such as neglect or verbal abuse) factors associated with brain development. Finally, due to the DTI imaging acquisition protocol, we only collected a single image with no diffusion weighting. Therefore, the EPI distortion artifact may not be completely corrected.
In conclusion, we found that white matter microstructure in (1) CC and bilateral occipitofrontal fasciculus and FSIQ score, and (2) the right SLF and PIQ score were positively correlated in typically developing children. These results suggest that efficient anatomical connectivity between parietal and frontal regions is crucial for children’s intelligence. Moreover, FA in the left CC and VIQ were positively correlated in boys, suggesting that efficient inter-hemispheric connections play a role for verbal abilities in boys. The present results should be validated using longitudinal data.
Acknowledgements
We would like to thank Y. Yamada, Y. Kotozaki, and R. Nouchi for operating the MRI scanner and for administering the psychological tests.
Conflicts of Interest Disclosures
The authors declare no competing interests.
Funding
This study was funded by JSPS KAKENHI Grant Number JP15K17418 and JP18K13221 .
References
Shaw P.
Greenstein D.
Lerch J.
Clasen L.
Lenroot R.
Gogtay N.
et al.
Intellectual ability and cortical development in children and adolescents.
Connections of the ventral granular frontal cortex of macaques with perisylvian premotor and somatosensory areas: anatomical evidence for somatic representation in primate frontal association cortex.
Tractography-guided statistics (TGIS) in diffusion tensor imaging for the detection of gender difference of fiber integrity in the midsagittal and parasagittal corpora callosa.
Developmental stages and sex differences of white matter and behavioral development through adolescence: a longitudinal diffusion tensor imaging (DTI) study.
00061-4&id=gr1.jpg){kind=link}
00061-4&id=gr2.jpg){kind=link}
00061-4&id=gr3.jpg){kind=link}
00061-4&id=gr4.jpg){kind=link}