Brain network topology and its cognitive impact in adult glioma survivors

Structural brain network topology can be altered in case of a brain tumor, due to both the tumor itself and its treatment. In this study, we explored the role of structural whole-brain and nodal network metrics and their association with cognitive functioning. Fifty WHO grade 2–3 adult glioma survivors (> 1-year post-therapy) and 50 matched healthy controls underwent a cognitive assessment, covering six cognitive domains. Raw cognitive assessment scores were transformed into w-scores, corrected for age and education. Furthermore, based on multi-shell diffusion-weighted MRI, whole-brain tractography was performed to create weighted graphs and to estimate whole-brain and nodal graph metrics. Hubs were defined based on nodal strength, betweenness centrality, clustering coefficient and shortest path length in healthy controls. Significant differences in these metrics between patients and controls were tested for the hub nodes (i.e. n = 12) and non-hub nodes (i.e. n = 30) in two mixed-design ANOVAs. Group differences in whole-brain graph measures were explored using Mann–Whitney U tests. Graph metrics that significantly differed were ultimately correlated with the cognitive domain-specific w-scores. Bonferroni correction was applied to correct for multiple testing. In survivors, the bilateral putamen were significantly less frequently observed as a hub (pbonf < 0.001). These nodes’ assortativity values were positively correlated with attention (r(90) > 0.573, pbonf < 0.001), and proxy IQ (r(90) > 0.794, pbonf < 0.001). Attention and proxy IQ were significantly more often correlated with assortativity of hubs compared to non-hubs (pbonf < 0.001). Finally, the whole-brain graph measures of clustering coefficient (r = 0.685), global (r = 0.570) and local efficiency (r = 0.500) only correlated with proxy IQ (pbonf < 0.001). This study demonstrated potential reorganization of hubs in glioma survivors. Assortativity of these hubs was specifically associated with cognitive functioning, which could be important to consider in future modeling of cognitive outcomes and risk classification in glioma survivors.


Post-hoc analyses: binary networks and non-normalized graph measures
We recomputed the analyses for binary graphs at different densities, but we have to keep in mind that binary networks with low density are not reproducible.Furthermore, by keeping the density the same between patients, we may also introduce spurious connections.By using weighted graphs with no or soft thresholding, we indeed also take spurious connections into account but the lower weights in that case, diminish the effect of these connections.
1. Hub identification (hubscore ³2 in >80% healthy controls) In weighted networks, we identified 12/78 nodes as hub regions.In binary networks, we identified 9-14 hubs (for densities of 10%: n=9, 20%: n=12, 30%: n=14), which largely overlap with the hubs identified in the weighted networks.In weighted networks, the postcentral gyrus was identified as a hub, but not in the binary networks, while the precuneus was identified as a hub in the binary networks but not in the weighted networks.

Binary networks
In patients compared to HC: Higher characteristic path length: density 15% (p = 7.06E-12), density 20% (p = 8.38E-12), density 25% (p = 1.36E-09), density 30% (p = 1.92E-08), density 35% (p = 3.42E-07), density 40% (p = 5.78E-06)In conclusion, as the densities of the binary networks increased, we observed a convergence in the results of the binary networks towards those of the weighted networks.While significant differences were predominantly identified in the weighted networks and binary networks with higher network densities (30-40%), between-group disparities in local efficiency, clustering coefficient, and path length were more frequently detected in the weighted networks.Conversely, differences in nodal strength between groups were not evident in the weighted networks, whereas they were observed in the binary networks.These correlations, although slightly diminished, remain robust.Regarding local efficiency, only one significant correlation persists within the non-hubs, while the correlations within the hubs remain unchanged.Similarly, for shortest path length, the correlations remain consistent within the hubs, but there is a variation within the non-hubs (e.g., the right pars triangularis exhibits a significant association instead of the right pericalcarine).In terms of assortativity, most correlations with IQ persist, although the associations with attention are less distinct in the binary networks.Additionally, fewer significant associations were observed between assortativity of hub nodes and cognitive outcomes (8 nodes instead of 11).

Figure S1 :
Figure S1: Correlation matrices of cognitive outcomes (all subjects)

Table S1 Hub regions and likelihood of being a hub across groups
*: significant value, with threshold of p<.05

Table S2 Overview of pure non-hubs Pure non-hub Patients Controls
insula Pure non-hub (hub score of 0 in all participants of the defined group) in patients compared to controls indicated in grey.Nodes which are defined as pure non-hub in controls but not in patients are indicated in bold, while nodes which are defined as pure non-hub in patients but not in controls are indicated in italic.

Table S6 Associations between nodal graph measures and cognitive domain scores Nodal measure Node Hub (H) or non-hub (NH)
Pearson correlations of nodal graph measures of clustering coefficient, local efficiency, shortest path length and assortativity with w-scores per cognitive domain.Only significant Bonferroni-corrected values are displayed