Basic Science and Pathogenesis.

Area of Science:

• Neuroscience
• Biophysics
• Computational Biology

Background:

• Alzheimer's disease (AD) affects millions globally, characterized by tau protein accumulation.
• In vivo tau-PET imaging shows significant individual variability in tau progression, unlike traditional Braak staging.
• Existing models lack biophysical basis or require longitudinal data.

Purpose of the Study:

• To develop a hybrid modeling approach integrating Event Based Models (EBMs) and Network Diffusion Models (NDMs) for predictive, individualized tau spread modeling.
• To overcome limitations of current modeling techniques by combining longitudinal data insights with biophysical principles.

Main Methods:

• Developed a hybrid model combining EBMs and NDMs for tau spread prediction.
• Applied EBM to 650 ADNI subjects for disease staging and generating longitudinal trajectories.
• Utilized an extended NDM (eNDM) to model tau spread as a diffusive process on brain networks, optimizing individual seed and kinetic parameters.

Main Results:

• Individual seed optimization yielded superior model fit (mean R=0.85) compared to parameter optimization and prior benchmarks.
• Model predictions strongly correlated with longitudinal tau-PET data (mean R=0.81).
• Tau patterns exhibit maximal heterogeneity at onset, converging over time, and two distinct seeding archetypes (entorhinal-dominant and diffuse temporal lobe) were identified.

Conclusions:

• The hybrid EBM-NDM approach enables accurate, individualized prediction of tau spread from cross-sectional data.
• Findings challenge classical Braak staging, indicating diverse tau initiation pathways that converge over time.
• This framework holds potential for personalized Alzheimer's treatment strategies and application to other neurodegenerative diseases.