A multiscale model to explain the spatiotemporal progression of amyloid beta and tau pathology in Alzheimer's disease

Amyloid-beta (Aβ) and tubulin-associated unit (tau) proteins are key biomarkers of Alzheimer's disease (AD), detectable by Positron Emission Tomography (PET) imaging and Cerebrospinal Fluid (CSF) assays. They reflect insoluble fibrils in the brain and soluble monomers in the cerebrospinal fluid, respectively. PET and CSF biomarkers have been utilized in diagnosing AD; however, their incomplete agreement significantly confounds the early detection. Additionally, the molecular mechanisms underlying the dynamics of AD biomarkers remain elusive and are yet to be quantitatively revealed. To answer these questions, we develop a multiscale mathematical model that characterizes various forms of AD biomarkers, including soluble molecules in cerebrospinal fluid, diffusive biomarkers across brain regions, and insoluble fibrils in the brain. Mathematical modeling of soluble and insoluble molecules enables the explanation of the asynchronous trajectory of AD biomarkers. Our model captures the spatiotemporal dynamics of Aβ and tau with neurodegeneration in AD. Simulation results demonstrate that the PET-CSF discordance is a typical stage in the natural history of protein aggregation, with CSF becoming abnormal before the onset of PET abnormality. Furthermore, correlation analysis reveals that neurodegeneration is more strongly associated with tau-PET than Aβ-PET. These findings suggest CSF Aβ is recognized as a biomarker at the early stage of AD, while tau-PET is more suitable for neurodegeneration assessment. The proposed multiscale model explains the underlying neurobiological factors contributing to neurodegeneration and offers a valuable tool for improving early detection and treatment strategies in clinical trials.