A Joint Tau Propagation and Neuroinflammation Model Reinforces Inflammatory Modulation of Network-Driven Spread in Alzheimer's Disease

Abstract

Alzheimers disease (AD) is a progressive neurodegenerative disorder and the leading cause of dementia. Despite over a century of research, AD remains untreatable due to an incomplete understanding of its underlying mechanisms. While amyloid beta has dominated therapeutic efforts, tau pathology and neuroinflammation represent critical disease drivers and intriguing therapeutic targets. We developed an extended computational model building upon the open-source Aggregation Network Diffusion (AND) framework by coupling spatial tau propagation and aggregation with ordinary differential equations representing key inflammatory cascades. The model incorporates M1/M2 microglia and astrocytic activation, cytokine-mediated feedback loops, and neuronal loss, all modulated via region-specific genetic expression matrices of ApoE and TREM2 with variant-specific weighting. The model maintains biologically plausible tau dynamics while generating robust inflammatory marker trends, serving as a computational testbed for hypothesis generation and mechanistic exploration. The inflammatory components independently capture experimental observations and display a marked M1/M2 microglia divergence. The model demonstrates enhanced similarity to regional tau propagation patterns due to inflammatory and genetic components, reinforcing neuroinflammations role in tau spread and highlighting the need to incorporate these processes in modeling efforts. Finally, through parsimony analysis, we identify microglia and pro-inflammatory rates (including microglia-facilitated tau spread) as key contributors to improved model accuracy, informing future modeling approaches.