EEG slowing in predementia Alzheimer's disease is compatible with neuronal hyperactivity: A multiscale computational modeling study

BACKGROUND: Evidence from animal models of Alzheimer's disease (AD) for neuronal hyperactivity and inhibitory dysfunction is accumulating, but the exact mechanism by which these microscale abnormalities lead to large-scale brain dysfunction and cognitive impairment in AD remains unclear. This study investigated empirically informed pathophysiological mechanisms of neuronal hyperactivity in a computational neurophysiological brain model and compared its output to EEG data of predementia AD patients. METHOD: EEG-like oscillations were simulated by a computational brain network model of 78 neural masses, each representing regional activity of a large population of interconnected excitatory and inhibitory neurons, coupled according to human DTI-derived structural topology. Several scenarios of amyloid-β mediated neuronal hyperactivity were investigated, including excitatory hyperexcitability and inhibitory neuronal dysfunction. The model offered a direct readout of neuronal activity by showing excitatory neuron pulse density. Consequently, we performed spectral analysis and analyzed absolute power and relative power in relevant frequency bands and peak frequency on the model output of each simulated AD scenario as well as for resting state EEG of predementia AD patients with positive amyloid biomarkers (n = 196) and elderly controls (n = 180). Similarity in outcome measure changes between the modelled output of each scenario and human EEG resulted in a scenario compatibility score (range 0-6). RESULT: Predementia AD patients showed increased relative theta power and decreased relative higher alpha power. The modeled scenarios involving 'inhibitory neuron hypoexcitability' and 'excessive neuronal excitation' were most compatible with the observed EEG changes in AD (score of 5), providing support for their presumed role. Other scenarios produced improbable physiologically parameter change combinations. Of the two best compatible scenarios, inhibitory hypoexcitability was the only scenario that led to neuronal hyperactivity in the model. CONCLUSION: We found that neuronal hyperactivity scenarios (i.e. increased pulse density) is compatible with large-scale oscillatory slowing of predementia AD patients. This study provides support for amyloid-driven neuronal disinhibition due to inhibitory hypoexcitability as most likely pathophysiological mechanism to explain oscillatory slowing in predementia AD.