Demographic data is shown in . Participant groups did not differ on IQ (p = 0.93) or gender distribution (p = 0.75).
Comparison of ICA-derived DMN maps in children and young adults
We first characterized the DMN in children and young adults using ICA. ICA indicated that the DMN can be identified in both children and young adults, and consists of nodes in the mPFC, PCC, the left and right MTL, and the left and right AG nodes (). A two sample t-test revealed that compared to young adults, children showed significantly weaker DMN connectivity and decreased spatial extent (p < 0.01, FDR corrected) in the mPFC but not in the PCC, MTL or the AG. Compared to young adults, children did not show greater network connectivity nor increased spatial extent in any brain region within the DMN.
Comparison of functional connectivity within the DMN in children and young adults
To further investigate group differences in ICA maps, we analyzed regional connectivity between the PCC, mPFC, MTL and the AG nodes of the DMN. Significant functional connectivity between each of these pairs of nodes was observed for both the child and young adult groups (p < 0.001, FDR corrected). In particular, strong functional connectivity between left and right MTL (rchildren = 0.53, radults = 0.58), and between left and right AG (rchildren = 0.45, radults = 0.47) was observed in both children and young adults. However, children showed significantly lower functional connectivity between the PCC and the mPFC than young adults (p < 0.01, FDR corrected). Partial correlation analysis revealed that this difference was significant (p < 0.01, FDR corrected) even after controlling for the influence of other nodes of the DMN (bilateral MTL and bilateral AG) and five other consistently detected large-scale brain networks -- salience, right executive control, left executive control, visual and sensorimotor network, as shown in . No other DMN nodes showed significant group differences in functional connectivity (, ). Additionally, we compared within hemisphere connectivity in each group. In children and young adults, no differences between PCC-right MTL and PCC-left MTL functional connectivity were observed (pchildren = 0.38, padults = 0.83). Similar results were observed when PCC-right AG and PCC-left AG were compared in each group (pchildren = 0.12, padults = 0.34).
Maturation of DMN: PCC-mPFC connectivity
Maturation of DMN: PCC-left MTL connectivity
Maturation of DMN: PCC-right MTL connectivity
Comparison of gray and white matter within the DMN in children and young adults
VBM revealed greater gray matter and lower white matter volume in children compared to young adults in several brain areas. DMN-related gray matter differences were observed in the PCC and mPFC (). White matter differences were observed in the dorsal cingulum “cingulate gyrus“ tract, which connects the PCC and mPFC () as well as left cingulum “hippocampus“ tract, which connects the PCC and left MTL (). No group differences were found in the right cingulum tract connecting the PCC and right MTL (). Children did not show lower gray matter or greater white matter volumes in any brain region examined.
Comparison of DTI connectivity within the DMN in children and young adults
White matter tracts connecting the PCC to the mPFC and the MTL were detected in individual participants, as shown in . Tracts connecting the PCC and mPFC were detected in all 33 participants for whom DTI data was available. PCC to right MTL tracts were detected in all 15 young adults and in 17 out of 18 children. PCC to left MTL tracts were detected in all 15 young adults, but in only 7 of 18 children. No tracts directly connecting the mPFC to the MTL were detected in either group.
Individual subject fiber tracts between DMN regions in children and young adults
The fiber density (number of fibers per unit area) of the PCC-mPFC tracts was significantly lower in children compared to young adults (p < 0.001), as shown in . Compared to young adults, children also showed significantly lower fiber density in the PCC-left MTL tracts (p < 0.001) (). This difference was significant even after excluding the subjects in which the PCC-left MTL tracts were not detected. There were no group differences in the fiber density of PCC-right MTL tracts ().
The mean FA value also showed similar differences between the two subject groups. The mean FA of the PCC-mPFC tracts was significantly lower in children compared to young adults (p << 0.0001) (). Compared to young adults, children also showed significantly lower mean FA along the PCC-left MTL tracts (p < 0.01) (). There were no group differences in the mean FA of the PCC-right MTL tracts ().
Replication and robustness of structural connectivity findings within the DMN in children
To determine if our novel structural connectivity findings, particularly the laterality effects observed in PCC to MTL fiber tracts in children, were robust (reproducible across datasets) we repeated the DTI analyses described in the main text on an independent dataset. This dataset was comprised of 10 subjects (5 males, 5 females) ranging in age from 7 to 9 (mean age 8.1) with an IQ range of 91 to 131 (mean IQ: 114). These subjects did not differ from the 23 children used in the original analyses, in age or IQ. Each subject underwent a structural, DTI, and resting-state fMRI scan. The parameters used for these scans were identical to those used for scanning participants that were examined in the original analyses. Results obtained from analyzing DTI data obtained from these 10 children were consistent with the original analyses: the PCC to mPFC fiber tracts were readily detected in all the children, the PCC to right MTL fiber tracts were detected in all but one children, while less than half the children (3 out of 10) showed PCC to left MTL fiber tracts.
Comparison of function-structure relationships within the DMN in children and young adults
Next, we examined the relationship between functional and structural connectivity within the DMN. We found that resting-state functional connectivity and fiber density between the PCC-and the mPFC were significantly positively correlated in young adults (r = 0.57; p < 0.05) but not in children (r = 0.21; p = 0.4), as shown in . This correlation was significant for the two groups combined (r = 0.5; p < 0.05). No significant correlation between functional connectivity and fiber density between the PCC and left or right MTL was observed in either group (, ). This correlation was insignificant even after excluding the subjects in which the PCC-MTL fibers were not detected.
A similar relationship between DMN resting-state functional connectivity and structural connectivity was observed when mean FA was used as a measure of structural connectivity instead of fiber density. Specifically, we found that resting-state functional connectivity and mean FA between the PCC- and the mPFC were significantly positively correlated in young adults (r = 0.62; p < 0.05) but not in children (r = 0.26; p = 0.28). This correlation was significant for the two groups combined (r = 0.53; p < 0.05). No significant correlation between functional connectivity and mean FA between the PCC and left or right MTL was observed in either group. This correlation was insignificant even after excluding the subjects in which the PCC-MTL fibers were not detected.