The structural connectome of the human brain in agenesis of the corpus callosum

Abstract

Adopting a network perspective, the structural connectome reveals the large-scale white matter connectivity of the human brain, yielding insights into cerebral organization otherwise inaccessible to researchers and clinicians. Connectomics has great potential for elucidating abnormal connectivity in congenital brain malformations, especially axonal pathfinding disorders. Agenesis of the corpus callosum (AgCC) is one of the most common brain malformations and can also be considered a prototypical genetic disorder of axonal guidance in humans. In this exploratory study, the structural connectome of AgCC is mapped and compared to that of the normal human brain. Multiple levels of granularity of the AgCC connectome are investigated, including summary network metrics, modularity analysis, and network consistency measures, with comparison to the normal structural connectome after simulated removal of all callosal connections ("virtual callostomy"). These investigations reveal four major findings. First, global connectivity is abnormally reduced in AgCC, but local connectivity is increased. Second, the network topology of AgCC is more variable than that of the normal human connectome, contradicting the predictions of the virtual callostomy model. Third, modularity analysis reveals that many of the tracts that comprise the structural core of the cerebral cortex have relatively weak connectivity in AgCC, especially the cingulate bundles bilaterally. Finally, virtual lesions of the Probst bundles in the AgCC connectome demonstrate that there is consistency across subjects in many of the connections generated by these ectopic white matter tracts, and that they are a mixture of cortical and subcortical fibers. These results go beyond prior diffusion tractography studies to provide a systems-level perspective on anomalous connectivity in AgCC. Furthermore, this work offers a proof of principle for the utility of the connectome framework in neurodevelopmental disorders.

Copyright © 2012 Elsevier Inc. All rights reserved.

Figures

Figure 1

The connectome processing pipeline utilized in this paper. FreeSurfer was used to parcellate the T1 MRI and FSL’s probtrackx2 was used to perform probabilistic tractography. The individual connectomes were combined to create a consensus connectome for each group.

Figure 2

Example midline sagittal and coronal color fractional anisotropy (FA) images for a control subject (top row) and an AgCC subject (bottom row).

Figure 3

Consensus connectomes for the a) control group, b) controls after virtual callosotomy, and c) AgCC group. The connectomes are displayed in the neurological convention and anterior is up and posterior is down. The 82 nodes are plotted with a circle scaled and colored according to the degree of the node (legend). A line between two nodes represents a suprathreshold connection, where the weight of the lines scales with the strength, although unweighted connectomes were used for the degree calculation.

Figure 4

Degree distribution across nodes for the consensus connectomes. The bars in red demonstrate the nodes with degree greater than mean plus one standard deviation for the control consensus connectome in (a). The dotted lines in (b) and (c) demonstrate the cutoff for mean plus one standard deviation in the virtual callostomy controls and AgCC. The red bars in (b) and (c) show the redistribution of regions in the virtual callostomy and AgCC consensus connectomes, respectively, that are hubs in the control consensus connectome.

Figure 5

Three example individual connectomes for the controls (left) and for AgCC (right). The connectomes are displayed in the neurological convention and anterior is up and posterior is down. The 82 nodes are plotted with a circle scaled and colored according to the degree of the node (see legend). A line between two nodes represents a suprathreshold connection, where the weight of the lines scales with the strength, but unweighted connectomes were used for the degree calculation. The controls exhibit some variability, but the overall distribution of degree is constant. The AgCC connectomes demonstrate more variability in the location and number of hub regions.

Figure 6

Mean degree distribution across nodes for the individual connectomes with standard deviation error bars. The bars colored red demonstrate the nodes with degree greater than mean plus one standard deviation for the controls. The dotted lines in (b) and (c) demonstrate the cutoff for mean plus one standard deviation for the virtual callostomy controls and for AgCC, respectively. The red bars in (b) and (c) show the redistribution of regions in the individual connectomes of the virtual callostomy and AgCC groups, respectively, that are hubs in the individual connectomes of the control subjects.

Figure 7

The module assignments for the consensus connectomes. The connectomes are displayed in the neurological convention and anterior is up and posterior is down. The 82 nodes are plotted with a circle colored according to the community to which it was assigned (legend). A line between two nodes represents a suprathreshold connection, where the weight of the lines scales with the strength. The weighted connectomes were used for modularity analysis.

Figure 8

The medial cortical modules ("structural core") as identified in the control consensus connectome. The edges colored red were significantly weaker in the AgCC group compared to the controls at p

Figure 9

The virtual Probstotomy results show that excluding the Probst bundle fibers reliably affects the corticocortical connections, as well as cortical-subcortical and subcortical-subcortical connections. Figure 9 (a) plots the number of subjects for which each pair-wise connection strength was reduced. Reduced connections were defined as those with percent differences in connection strength, before and after the Probstotomy, greater than the mean plus one standard deviation. Figure 9(b) displays the connections where at least three of seven subjects showed reduced connection strengths after the Probstotomy.