Alzheimer's disease disrupts alpha and beta-band resting-state oscillatory network connectivity

Abstract

Objective: Neuroimaging studies in Alzheimer's disease (AD) yield conflicting results due to selective investigation. We conducted a comprehensive magnetoencephalography study of connectivity changes in AD and healthy ageing in the resting-state.

Methods: We performed a whole-brain, source-space assessment of oscillatory neural signalling in multiple frequencies comparing AD patients, elderly and young controls. We compared eyes-open and closed group oscillatory envelope activity in networks obtained through temporal independent component analysis, and calculated whole-brain node-based amplitude and phase connectivity.

Results: In bilateral parietotemporal areas, oscillatory envelope amplitude increased with healthy ageing, whereas both local amplitude and node-to-global connectivity decreased with AD. AD-related decreases were spatially specific and restricted to the alpha and beta bands. A significant proportion of the variance in areas of peak group difference was explained by cognitive integrity, in addition to group. None of the groups differed in phase connectivity. Results were highly similar for eyes-open and closed resting-state.

Conclusions: These results support the disconnection syndrome hypothesis and suggest that AD shows distinct and unique patterns of disrupted neural functioning, rather than accelerated healthy ageing.

Significance: Whole-brain assessments show that disrupted regional oscillatory envelope amplitude and connectivity in the alpha and beta bands play a key role in AD.

Keywords: Alzheimer’s disease; Default-mode network; Functional connectivity; Magnetoencephalography; Neural oscillations; Resting state.

Fig. 1

Temporal oscillatory ICA network results generated on all data (all groups and both eye states). (A) Spatial maps of four consistent network components in the beta band transformed to MNI space. Images show absolute ICA weights (in A.U.) thresholded above 0.4, scaled to a maximum of 1. (B) SDs of beta-band oscillatory envelopes averaged within the spatial boundaries of each network, plotted over age per network and eye state. Black circles: control data, grey crosses: AD. AD = Alzheimer’s disease patients, EC = Elderly controls, YC = Young controls. Statistics indicate ANOVA post hoc t test results: ^*p < 0.05 for the relevant 2-group comparison (EC > YC, black; AD < EC, grey), ^**p < 0.05 for AD < EC&YC, Bonferroni corrected. Per plot, a least-squares line is fitted for all relevant control data over age for illustration of the group difference direction only.

Fig. 2

Per-group comparable ICA spatial network components transformed to MNI space. Each image displays an axial slice positioned at component maximum. Each ICA was performed on eyes-open and closed data combined. AD = Alzheimer’s disease patients, EC = Elderly controls, YC = Young controls. The bottom row (All) shows the original ICA analysis performed on all groups together equivalent to Fig. 1. Images show absolute ICA weights (in A.U.), scaled to a maximum of 1.

Fig. 3

Voxel-based group comparisons of beta-band oscillatory envelope SD transformed to MNI space. Per comparison and eye state, images display four sagittal (top) and axial (bottom) slices, representing t values thresholded at voxel-corrected p < 0.05, averaged over the x and z Talairach coordinates as indicated (coordinates in mm). AD = Alzheimer’s disease patients, EC = Elderly controls, YC = Young controls. Orange/Yellow colours represent regions where EC had greater SDs than AD (upper panels), or than YC (lower panels). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 4

Node-based connectivity results for the EC vs. AD comparison. Orange/Yellow colours represent regions where EC had greater node strength than AD. Only group comparisons with significant differences are displayed. Each AAL atlas region represents the results for each respective node. Per comparison and eye state, images display four sagittal (top) and axial (bottom) slices, representing t values thresholded at corrected p < 0.05 for correlation strength (the sum over nodes), averaged over the x and z Talairach coordinates as indicated (coordinates in mm). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Supplementary Fig. S1

Overview of all 15 ICA components for each of the six frequency bands. A single slice is displayed in the plane providing the best overview of the activity foci in the respective component. The colours reflect positive (yellow/red) and negative (white/blue) areas of temporal correlation. Irrespective of direction, areas with the same colour are correlated, whereas areas of opposite colour are anti-correlated. Candidate components for each of the four networks of interest are indicated (SM = sensorimotor network, V = visual network, LPF/RPF = left/right parietofrontal network), as judged by visual interpretation. A question mark indicates questionable evidence. All four components can only clearly be identified in the beta band.

Supplementary Fig. S2

Voxel-based group comparisons of oscillatory envelope SD transformed to MNI space for the EC vs. YC comparison for all frequency bands, except the beta band, that showed significant differences. Per comparison and eye state, images display four sagittal (top) and axial (bottom) slices, representing t values thresholded at corrected p < 0.05 for correlation strength (the sum over nodes), averaged over the x and z Talairach coordinates as indicated (coordinates in mm). Gamma 1 = 30–50 Hz, Gamma 2 = 50–90 Hz. Orange/Yellow colours represent regions where EC had greater SDs than YC, blue/white colours represent regions where EC had lower SDs than YC.

Supplementary Fig. S3

Voxel-based group comparisons of oscillatory envelope SD transformed to MNI space for the EC vs. AD comparison for all frequency bands, except the beta band, that showed significant differences. Per comparison and eye state, images display four sagittal (top) and axial (bottom) slices, representing t values thresholded at corrected p < 0.05 for correlation strength (the sum over nodes), averaged over the x and z Talairach coordinates as indicated (coordinates in mm). Gamma 2 = 50–90 Hz. Orange/Yellow colours represent regions where EC had greater SDs than AD, blue/white colours represent regions where EC had lower SD than AD.