Neuronal synchrony abnormalities associated with subclinical epileptiform activity in early-onset Alzheimer's disease

Abstract

Since the first demonstrations of network hyperexcitability in scientific models of Alzheimer's disease, a growing body of clinical studies have identified subclinical epileptiform activity and associated cognitive decline in patients with Alzheimer's disease. An obvious problem presented in these studies is lack of sensitive measures to detect and quantify network hyperexcitability in human subjects. In this study we examined whether altered neuronal synchrony can be a surrogate marker to quantify network hyperexcitability in patients with Alzheimer's disease. Using magnetoencephalography (MEG) at rest, we studied 30 Alzheimer's disease patients without subclinical epileptiform activity, 20 Alzheimer's disease patients with subclinical epileptiform activity and 35 age-matched controls. Presence of subclinical epileptiform activity was assessed in patients with Alzheimer's disease by long-term video-EEG and a 1-h resting MEG with simultaneous EEG. Using the resting-state source-space reconstructed MEG signal, in patients and controls we computed the global imaginary coherence in alpha (8-12 Hz) and delta-theta (2-8 Hz) oscillatory frequencies. We found that Alzheimer's disease patients with subclinical epileptiform activity have greater reductions in alpha imaginary coherence and greater enhancements in delta-theta imaginary coherence than Alzheimer's disease patients without subclinical epileptiform activity, and that these changes can distinguish between Alzheimer's disease patients with subclinical epileptiform activity and Alzheimer's disease patients without subclinical epileptiform activity with high accuracy. Finally, a principal component regression analysis showed that the variance of frequency-specific neuronal synchrony predicts longitudinal changes in Mini-Mental State Examination in patients and controls. Our results demonstrate that quantitative neurophysiological measures are sensitive biomarkers of network hyperexcitability and can be used to improve diagnosis and to select appropriate patients for the right therapy in the next-generation clinical trials. The current results provide an integrative framework for investigating network hyperexcitability and network dysfunction together with cognitive and clinical correlates in patients with Alzheimer's disease.

Keywords: epileptiform activity in Alzheimer’s disease; imaginary coherence; magnetoencephalography; network hyperexcitability; neuronal synchrony.

Figure 1

Frequency-specific neuronal synchrony abnormalities in Alzheimer’s disease patients with and without epileptiform activity. (A–F) The strength of functional connectivity as global imaginary coherence (IC) within alpha (8–12 Hz) and delta–theta (2–8 Hz) frequency oscillation bands. Controls showed highest imaginary coherence within the alpha band (8–12 Hz) in occipitoparietal cortices (A). AD-EPI− and AD-EPI+ patients showed reduced imaginary coherence within the alpha band, where the latter group showed a greater degree of reductions (B and C). Within the delta–theta band, controls showed highest imaginary coherence in frontal and parietal cortices (D). AD-EPI− and AD-EPI+ both showed enhanced delta–theta imaginary coherence where the latter group showed greater degree of enhancement (E and F). The colour maps are thresholded with a cluster correction of 30 voxels (P < 0.01) and at 5% false discovery rate (n = 30, AD-EPI+; n = 20 AD-EPI−; n = 35 age-matched controls). Alzheimer’s disease = Alzheimer’s disease; AD-EPI− = Alzheimer’s disease patients without epileptiform activity; AD-EPI+ = Alzheimer’s disease patients with epileptiform activity; IC = imaginary coherence.

Figure 2

Distinctive regional patterns of neuronal synchrony deficits in Alzheimer’s disease patients with and without epileptiform activity. (A–F) Statistical comparisons of imaginary coherence between groups. Within the alpha band, AD-EPI− and AD-EPI+ showed significantly reduced imaginary coherence compared to controls, over bilateral occipital cortices, with more extensive regional involvement in AD-EPI+ (A and B). Pairwise comparison showed AD-EPI+ with significantly lower alpha imaginary coherence than AD-EPI− (C). AD-EPI− and AD-EPI+ showed significantly increased delta–theta imaginary coherence compared to controls, over bilateral frontal and parietal cortices, with more extensive involvement in AD-EPI+ (D and E) and pairwise comparison showed AD-EPI+ with significantly higher delta–theta imaginary coherence than AD-EPI− (F). Each brain rendering depicts the t-maps from voxelwise comparison of global imaginary coherence between groups. The colour maps are thresholded with a cluster correction of 30 voxels (P < 0.01) and at 5% FDR. (C and F) depict least squares (LS)-means and 95% confidence-limits (n = 30, AD-EPI+; n = 20 AD-EPI−; n = 35 age-matched controls). Alzheimer’s disease = Alzheimer’s disease; AD-EPI− = Alzheimer’s disease patients without epileptiform activity; AD-EPI+ = Alzheimer’s disease patients with epileptiform activity; IC = imaginary coherence.

Figure 3

Neuronal synchrony indices distinguish between AD-EPI+ from AD-EPI− and are associated with rate of change in MMSE. (A) The ROC curves derived from logistic regression analyses for the discriminability between AD-EPI− and AD-EPI+ based on alpha and delta–theta imaginary coherence values from the most affected voxel-level regions of interest (varying from 1, 3, or 5 per frequency band). Longitudinal change in MMSE was significantly predicted by the first two principal components of the imaginary coherence matrix including alpha and delta–theta imaginary coherence in the full cohort (B). (A) n = 30, 20, 35, AD-EPI−, AD-EPI+ and controls; (B) n = 17, 24, 29, AD-EPI+, AD-EPI− and controls). Alzheimer’s disease = Alzheimer’s disease; AD-EPI− = Alzheimer’s disease patients without epileptiform activity; AD-EPI+ = Alzheimer’s disease patients with epileptiform activity; IC = imaginary coherence; MMSE = Mini Mental State Examination; ROC = receiver operating characteristic.