Investigation of neurophysiologic and functional connectivity changes following glioma resection using magnetoencephalography

Abstract

Background: In patients with glioma, clinical manifestations of neural network disruption include behavioral changes, cognitive decline, and seizures. However, the extent of network recovery following surgery remains unclear. The aim of this study was to characterize the neurophysiologic and functional connectivity changes following glioma surgery using magnetoencephalography (MEG).

Methods: Ten patients with newly diagnosed intra-axial brain tumors undergoing surgical resection were enrolled in the study and completed at least two MEG recordings (pre-operative and immediate post-operative). An additional post-operative recording 6-8 weeks following surgery was obtained for six patients. Resting-state MEG recordings from 28 healthy controls were used for network-based comparisons. MEG data processing involved artifact suppression, high-pass filtering, and source localization. Functional connectivity between parcellated brain regions was estimated using coherence values from 116 virtual channels. Statistical analysis involved standard parametric tests.

Results: Distinct alterations in spectral power following tumor resection were observed, with at least three frequency bands affected across all study subjects. Tumor location-related changes were observed in specific frequency bands unique to each patient. Recovery of regional functional connectivity occurred following glioma resection, as determined by local coherence normalization. Changes in inter-regional functional connectivity were mapped across the brain, with comparable changes in low to mid gamma-associated functional connectivity noted in four patients.

Conclusion: Our findings provide a framework for future studies to examine other network changes in glioma patients. We demonstrate an intrinsic capacity for neural network regeneration in the post-operative setting. Further work should be aimed at correlating neurophysiologic changes with individual patients' clinical outcomes.

Keywords: functional connectivity; glioma; magnetoencephalography; neural networks.

Figure 1.

Pre- and post-resection neuroimaging. (A-J) Axial views of T2-weighted magnetic resonance imaging (MRI) (pre-) and T2-FLAIR (post) sequences for each study subject with delayed post-resection MEG recordings. Images were acquired before surgery and <48 hours following tumor resection. Note that the same imaging sequences are displayed for consistency. In some cases of high-grade tumors, edema, and post-operative mass effect may be more apparent.

Figure 2.

Changes in MEG-derived spectral power following tumor resection. (A–F) Topographical plots show sensor-level cluster-based permutation testing for delayed post-resection brain activity compared to pre-resection brain activity. Plots are shown only for frequency bands with a statistically significant change in spectral power, Bonferroni-corrected (n = 7 multiple comparisons for frequency bands), P < .0071. Green boxes highlight significant frequencies in the anatomical vicinity of the tumor. Yellow (positive T-statistic values) represent an increase in spectral power post-resection. Blue (negative T-statistic values) represent a decrease in spectral power post-resection. Frequency bands: delta (1–3 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (13–29 Hz), low gamma (30–50 Hz), mid gamma (50–70 Hz), and broadband gamma (70–170 Hz). MR images (left) show the anatomical orientation of tumor location.

Figure 3.

Recovery of regional functional connectivity following tumor resection. (A–F) Boxplots comparing local connectivity (regional coherence) between healthy controls (HC, n = 28 [56 measurements]) and glioma patients pre-resection and delayed post-resection (triplicate measurements for all conditions). Frequencies of interest are determined from significant changes in spectral power following tumor resection specific to tumor location (refer to Figure 2, green boxes). Plots are shown for brain regions with significant changes in coherence between groups, showing a recovery of abnormally low or high pre-resection coherence to levels found in healthy controls. Regional functional connectivity was estimated by averaging coherence for all brain region pairs within each of the major brain regions. Brain regions: left (L) and right (R) frontal (F), sensorimotor (SM), parietal (P), occipital (O), limbic (L), basal ganglia (BG), temporal (T), and cerebellar (C). Frequency bands: delta (1–3 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (13–29 Hz), low gamma (30–50 Hz), mid gamma (50–70 Hz), and broadband gamma (70–170 Hz). ns, not significant; *P < .05; **P < .01; ***P < .001; ****P < .0001. MR images (left) show the anatomical orientation of tumor location.

Figure 4.

Changes in inter-regional functional connectivity following tumor resection. (A–F) Functional connectivity maps (connectograms) depict changes in coherence for delayed post-resection relative to pre-resection brain activity. Frequencies of interest are determined from significant changes in spectral power following tumor resection specific to tumor location (refer to Figure 2, green boxes). Intra-regional functional connectivity was estimated by averaging coherence for all brain region pairs within each major brain region. Inter-regional (between region-of-interest (ROI)) connectivity was estimated by averaging coherence for all brain region pairs between two ROIs and included interhemispheric (i.e. left vs. right) and intrahemispheric (i.e. left vs. left, right vs. right) comparisons. Red (positive T-statistic values) represent a significant increase in coherence post-resection (“functional coupling”). Blue (negative T-statistic values) represent a significant decrease in coherence post-resection (“decoupling”). Brain regions: left (L) and right (R) frontal (F), sensorimotor (SM), parietal (P), occipital (O), limbic (L), basal ganglia (BG), temporal (T), and cerebellar (C), and midline vermis (MV). Frequency bands: delta (1–3 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (13–29 Hz), low gamma (30–50 Hz), mid gamma (50–70 Hz), and broadband gamma (70–170 Hz). MR images (left) show the anatomical orientation of tumor location.