Neural mass framework for decoding amyloid and Tau impact on neuronal activity under Alzheimer’s Disease

Abstract Background We introduce a computational brain network model to understand the effect of abnormal microscopical processes into non‐invasive macroscopic neuronal activity observations and cognitive impairment across the Alzheimer’s disease (AD) spectrum. Amyloid‐beta (Aβ) plaques and tau protein aggregates (τ) modulate neuronal firing with the former likely increasing firing (hyperactivity) and the latter inhibiting activity (hypoactivity), especially in the abundant pyramidal populations ‐see (Maestu et al., 2021) for more details. We assume that such interactions scale up to macroscopic regions in humans, specifically influencing resting‐state fMRI signals. Method Multimodal data, including amyloid ([18F]Florbetapir) and tau ([18F]flortaucipir) PET, baseline resting‐state fMRI, structural and diffusion MRI, and clinical/cognitive indicators were obtained from the Alzheimer’s Disease Neuroimaging Initiative. The imaging modalities were processed to produce values for a parcellation consisting of 66 brain regions. We considered four neural masses interacting within a region. The regional excitatory and inhibitory activities then transform into BOLD signals (Valdes‐Sosa et al., 2009) ‐see Fig. 1A. Pyramidal excitabilities (v 0E,k ) contain the Aβ‐ and τ‐ parameters of interest, and their combined effect (Fig. 1B), which we estimated by minimizing the spectral distance between simulated signals and observations. Result The synergy between disease factors seems to change with disease states and subjects (Fig. 1C). Most of the considered healthy controls (HC) presented relatively low excitability change by Aβ and by τ, albeit having a greater, generally positive or ‘towards hypoactivity’ combined action. Mild cognitive impairment subjects appeared in two sub‐groups, one significantly close to the HC and another with stronger individual effects and smaller joined action. All AD patients presented positive synergistic effects and relatively high independent contributions to pyramidal population firing, as estimated from the data. Conclusion The obtained significant, subject‐specific influences of pathological factors on neural activity verify the interactions reported by (Busche et al., 2020) in animal models experiments at the mesoscopic scale. There exist causal relationships between the amyloid and tau influence parameters and clinical variables. Further research must be performed to clarify characteristic disease trajectories as effective sub‐typing in early stages could improve therapeutic interventions.