The impact of gliomas on resting-state oscillatory activity and connectivity: A magnetoencephalography study

Abstract

Gliomas disturb brain functions, contributing to abnormal neuronal activity and connectivity changes that can be effectively investigated using magnetoencephalography (MEG). In this work, we used MEG to evaluate frequency-specific alterations of brain activity and functional connectivity in glioma patients. Two consecutive 5-min sessions of eyes-closed resting-state brain activity were recorded from ten glioma patients and ten age-matched healthy subjects. Modulations of power and functional connectivity, within the patient group and between patients and the healthy control group, were assessed in terms of the Neural Activity Index and the Multivariate Interaction Measure, respectively. These quantities were calculated in individualized frequency bands (delta, theta, lower alpha, upper alpha and beta) to account for changes in spectral peaks in patients. We report a decrease in power in tumor and peri-tumor regions in the upper-alpha and beta bands with respect to the rest of the brain, paired to a decrease in lower and upper-alpha band functional connectivity of the tumor and peri-tumor regions. In comparison with healthy subjects, we observe a global enhancement in power in lower-alpha in patients. Overall, our study shows that glioma infiltration can widely influence brain local and long-range synchrony in an individualized frequency-specific manner.

Keywords: Functional connectivity; Glioma patients; Magnetoencephalography; Resting-state.

Fig. 1

Illustrative example of regions of interest (ROIs) in one representative patient with a tumor in the left temporo-occipital lobe. The figure shows an axial cross-sectional image of the ROIs superimposed on the T1-weighted MR image with contrast enhancement.

Fig. 2

(a) Violin plots of the relative difference for Neural Activity Index (NAI) in the pooled tumor- and peritumor regions ‘Tumor and Peritumor’ compared to the rest of the brain ‘Rest of the brain’ and to contralateral control regions ‘Contralateral’, and in the contralateral control regions compared to the rest of the brain; the white dot and the thick whiskers denote the median value across patients and the range from the 25th to the 75th percentile, respectively; the symbol * denotes a significant difference at the p < 0.05 level (paired-sample permutation test, one-tail, FDR corrected). (b, c, d): axial cross-sectional maps of the relative difference for NAI at each brain location compared to the ‘Rest of the brain’, for the lower alpha, upper alpha and beta frequency bands, in one representative patient with a glioma in the left temporo-occipital lobe; the maps are superimposed on the T1-weighted MR image with contrast enhancement. (e, f, g): axial cross-sectional maps from the same patient of the relative difference for NAI in each location within the hemisphere ipsilateral to the tumor compared to its homologous location in the hemisphere contralateral to the tumor. A substantial decrease in NAI can be observed in the region within and surrounding the tumor compared to the rest of the brain, as well as to the contralateral region, in the lower alpha, upper alpha and beta frequency bands.

Fig. 3

(a) Violin plots of the relative difference for the Global Multivariate Interaction Measure (GMIM) in the pooled tumor and peritumor regions ‘Tumor and Peritumor’ compared to the rest of the brain ‘Rest of the brain’ and to contralateral control regions ‘Contralateral’, and in the contralateral control regions compared to the rest of the brain; the white dot and the thick whiskers denote the median value across patients and the range from the 25th to the 75th percentile, respectively; the symbol * denotes a significant difference at the p < 0.05 level (paired-sample permutation test, one-tail, FDR corrected). (b, c): axial cross-sectional maps of the relative difference for GMIM in each brain location compared to the rest of the brain, for the lower alpha and upper alpha frequency bands, in one representative patient with a glioma in the left temporo-occipital lobe; the maps are superimposed on the T1-weighted MR image with contrast enhancement. A substantial decrease of GMIM can be observed in the region within and surrounding the tumor compared to rest of the brain in the lower alpha and upper alpha frequency bands.

Fig. 4

(a) Violin plots of the average band-relative Neural Activity Index (NAI) in the whole brain for healthy subjects and tumor patients. The white dot and the thick whiskers denote the median value across patients and the range from the 25th to the 75th percentile, respectively. The symbol * denotes a significant difference at the p < 0.05 level (independent-sample permutation test, one-tail, FDR corrected). A significant increase of average band-relative NAI in the lower-alpha band can be observed in the whole brain of tumor patients compared to healthy subjects. (b) Average Global Multivariate Interaction Measure (GMIM) in the whole brain for healthy subjects and tumor patients.

Fig. 5

(a) Violin plots of the GMIM hemispheric difference for healthy subjects and tumor patients. (b) Negative correlation of GMIM laterality index in the theta band with the tumor volume (Spearman's rho = −0.62, p = 0.03, one-sided Spearman's rank correlation test).