A multiscale brain network model links Alzheimer’s disease-mediated neuronal hyperactivity to large-scale oscillatory slowing

Abstract

Background Neuronal hyperexcitability and inhibitory interneuron dysfunction are frequently observed in preclinical animal models of Alzheimer’s disease (AD). This study investigates whether these microscale abnormalities explain characteristic large-scale magnetoencephalography (MEG) activity in human early-stage AD patients. Methods To simulate spontaneous electrophysiological activity, we used a whole-brain computational network model comprised of 78 neural masses coupled according to human structural brain topology. We modified relevant model parameters to simulate six literature-based cellular scenarios of AD and compare them to one healthy and six contrast (non-AD-like) scenarios. The parameters include excitability, postsynaptic potentials, and coupling strength of excitatory and inhibitory neuronal populations. Whole-brain spike density and spectral power analyses of the simulated data reveal mechanisms of neuronal hyperactivity that lead to oscillatory changes similar to those observed in MEG data of 18 human prodromal AD patients compared to 18 age-matched subjects with subjective cognitive decline. Results All but one of the AD-like scenarios showed higher spike density levels, and all but one of these scenarios had a lower peak frequency, higher spectral power in slower (theta, 4–8Hz) frequencies, and greater total power. Non-AD-like scenarios showed opposite patterns mainly, including reduced spike density and faster oscillatory activity. Human AD patients showed oscillatory slowing (i.e., higher relative power in the theta band mainly), a trend for lower peak frequency and higher total power compared to controls. Combining model and human data, the findings indicate that neuronal hyperactivity can lead to oscillatory slowing, likely due to hyperexcitation (by hyperexcitability of pyramidal neurons or greater long-range excitatory coupling) and/or disinhibition (by reduced excitability of inhibitory interneurons or weaker local inhibitory coupling strength) in early AD. Conclusions Using a computational brain network model, we link findings from different scales and models and support the hypothesis of early-stage neuronal hyperactivity underlying E/I imbalance and whole-brain network dysfunction in prodromal AD.

Tau in cerebrospinal fluid induces neuronal hyperexcitability and alters hippocampal theta oscillations

“…It has also been demonstrated that changes to theta oscillatory power – similar to what we observed herein – can be used to distinguish between prodromal AD and non-AD cases with cognitive decline [91]. The fact that our effects on oscillations agree with clinical human electrophysiological studies gives confidence that this is a method that could, in the future, be clinically useful as in compliment to EEG testing.…”

Combining Aperiodic 1/f Slopes and Brain Simulation: An EEG/MEG Proxy Marker of Excitation/Inhibition Imbalance in Alzheimer’s Disease

“…Previous models in AD research have lumped neural populations at each brain region into neural masses (neural mass models) (Ranasinghe et al 2022; Monteverdi et al 2022; van Nifterick et al 2022; Stefanovski et al 2021; Stefanovski et al 2019; Zimmermann et al 2018; de Haan et al 2017). Using neural mass models, recent studies have found a relationship between abnormal excitatory and inhibitory time-constants and spatial depositions of Aβ and tau (Ranasinghe et al 2022; Stefanovski et al 2019), have linked neuronal hyperactivity in preclinical AD to oscillatory slowing (van Nifterick et al 2022; Stefanovski et al 2019) and have developed a successful strategy to preserve network integrity during AD progression (de Haan et al 2017). A recent work has also used neural mass models to extract information about the excitatory/inhibitory balance in single subjects and suggested that AD subjects were characterized by increased global coupling and increased inhibition (Monteverdi et al 2022).…”

Revealing the combined roles of Aβ and tau in Alzheimer’s disease via a pathophysiological activity decoder

“…We obtained subject-specific relative contributions of the considered pathophysiological factors on neuronal activity and reconstructed proxy quantities for electro-(magneto-)encephalographic (E/MEG) sources in each brain region. Subsequently, we aimed to test the proposed pathophysiological - activity generator’s ability to validate reported spectral changes in AD, i.e., increases of theta band power (4–8 Hz) and decreases of power in the lower alpha band (alpha1, 8–10 Hz) 9–11 . Among the quantities contributing to the E/MEG model output, we also closely studied excitatory firings and changes to their magnitude given the influence of the toxic protein depositions (for definitions and further details on the calculations, see Methods, Integrative neuronal activity simulator ).…”

“…These disease processes also manifest differently given subject-specific genetic and environmental conditions 1, 8 . Models of multiple pathological markers and physiology can help reveal the connection between individual AD fingerprints and cognitive deficits 3, 9,10 .…”

“…Simulated resting-state functional MRI across the AD spectrum was used to estimate biophysical parameters associated with cognitive deterioration 29 . In addition, different intervention strategies to counter neuronal hyperactivity in AD have been tested 9, 22 , illustrating another key advantage of mechanistic causal modeling: the design of personalized treatments 1, 3, 8, 10, 25, 26 . Notably, computational approaches also offer accurate tools to quantify biological parameters that are not measurable with non-invasive techniques 30 , like neuronal excitability values, which would allow us to characterize sources of functional hyperactivity 28 .…”