Abstract
Introduction:
Patients with unilateral brain tumors show dramatic alterations in functional brain connectivity. These alterations are not only restricted to the tumor area, but are also thought to occur in remote, even contralateral areas. Functional reshaping is time-dependent, as recruitment of perilesional and remote brain areas is much more efficient in slow growing than acute lesions. Whether the growth-rate of a tumour modulates the functional network topology of the contralesional hemisphere remains, however, unclear. Low-grade glioma (LGG, WHO-grade I-II) and high-grade glioma (HGG, WHO-grade III-IV) patients provide an optimal window to study this. LGG grow more slowly and less aggressively with lower degrees of cell infiltration and proliferation than HGG, permitting a greater plastic reorganization of the functional networks in LGG patients. The goal of this study was therefore to examine the differences between LGG and HGG patients in functional network topology in the contralesional hemisphere.
Methods:
Resting state fMRI data were acquired in 80 glioma patients with a left hemispheric tumor (40 LGG, 40 HGG patients) before resective brain surgery. A connectivity matrix for the contralesional hemisphere was created. Based on this connectivity matrix, a multivariate pattern classification was used to classify patients as having an LGG or HGG. The following network metrics were computed: global connection strength (provides information on the total degree of connectivity); global efficiency (reflects the integration of network-wide communication); local efficiency (represents the potential for local information transfer); modularity (indicates to what extent the network can be subdivided into separate modules); intra-modular connection weight (reflects the local processing within modules) and inter-modular connection weight (reflects the distributed processing across modules). These metrics were compared between LGG and HGG patients with permutation tests.
Results:
The multivariate pattern classification based on the contralesional connectivity matrix was successful in classifying LGG and HGG patients (accuracy = 63%; p < .05 above chance). Analyses of the network metrics of the contralesional hemisphere showed a lower local efficiency, a lower intra-modular connection weight and a higher inter-modular connection weight in LGG than in HGG patients (all p’s < .05).
Conclusions:
We were able to correctly classify LGG and HGG patients based on the contralesional connectivity matrix, suggesting differences in the contralesional functional network topology between these two groups. More specifically, LGG patients showed a lower potential for local information transfer and more distributed processing across modules than HGG patients. This suggests that differences in lesion speed can lead to differences in the contralesional functional network topology.
Patients with unilateral brain tumors show dramatic alterations in functional brain connectivity. These alterations are not only restricted to the tumor area, but are also thought to occur in remote, even contralateral areas. Functional reshaping is time-dependent, as recruitment of perilesional and remote brain areas is much more efficient in slow growing than acute lesions. Whether the growth-rate of a tumour modulates the functional network topology of the contralesional hemisphere remains, however, unclear. Low-grade glioma (LGG, WHO-grade I-II) and high-grade glioma (HGG, WHO-grade III-IV) patients provide an optimal window to study this. LGG grow more slowly and less aggressively with lower degrees of cell infiltration and proliferation than HGG, permitting a greater plastic reorganization of the functional networks in LGG patients. The goal of this study was therefore to examine the differences between LGG and HGG patients in functional network topology in the contralesional hemisphere.
Methods:
Resting state fMRI data were acquired in 80 glioma patients with a left hemispheric tumor (40 LGG, 40 HGG patients) before resective brain surgery. A connectivity matrix for the contralesional hemisphere was created. Based on this connectivity matrix, a multivariate pattern classification was used to classify patients as having an LGG or HGG. The following network metrics were computed: global connection strength (provides information on the total degree of connectivity); global efficiency (reflects the integration of network-wide communication); local efficiency (represents the potential for local information transfer); modularity (indicates to what extent the network can be subdivided into separate modules); intra-modular connection weight (reflects the local processing within modules) and inter-modular connection weight (reflects the distributed processing across modules). These metrics were compared between LGG and HGG patients with permutation tests.
Results:
The multivariate pattern classification based on the contralesional connectivity matrix was successful in classifying LGG and HGG patients (accuracy = 63%; p < .05 above chance). Analyses of the network metrics of the contralesional hemisphere showed a lower local efficiency, a lower intra-modular connection weight and a higher inter-modular connection weight in LGG than in HGG patients (all p’s < .05).
Conclusions:
We were able to correctly classify LGG and HGG patients based on the contralesional connectivity matrix, suggesting differences in the contralesional functional network topology between these two groups. More specifically, LGG patients showed a lower potential for local information transfer and more distributed processing across modules than HGG patients. This suggests that differences in lesion speed can lead to differences in the contralesional functional network topology.
Original language | English |
---|---|
Pages | iii41-iii42 |
DOIs | |
Publication status | Published - 2017 |