Tau Progression Index (TPI): An individualized, clinically applicable, multimodally-derived score to predict AD progression

While amyloid-beta (A{beta}) and tau are known to be the key neuropathological hallmarks of Alzheimers disease (AD), imaging and postmortem studies highlight significant variability in disease severity and progression even in subjects with similar burdens of A{beta} and tau accumulation. This uncertainty in the trajectory of disease progression results in substantial challenges for patients and their families in understanding disease prognosis, as well as the research community in understanding heterogeneity in the rate of A{beta} and tau accumulation, cognitive decline, and likelihood of progressing to dementia. In this study, we developed and validated the Tau Progression Index (TPI); a multimodally derived score that leverages strongly supported hypotheses and models of tau spread, to serve as a reliable, and clinically applicable biomarker of AD progression. TPI has 4 components: 1) Remote interaction (TPI-ri) interactions between A{beta} and tau pathologies in regions that spatially separate; 2) Local interaction (TPI-li) between co-localized A{beta} and tau pathology; 3) Functional connectivity (TPI- fc) is based on functional magnetic resonance imaging (fMRI)-measured inter-regional functional connections facilitate the spread of tau in the brain; and 4) Structural connectivity (TPI-sc) based on diffusion MRI (dMRI)-measured inter-regional structural connectivity. Importantly, all prior models of tau spread in AD were based on group-averaged structural and functional connectivity, derived from healthy and often young subjects. However, brain connectivity patterns are extremely subject-specific, often referred to as a human connectome fingerprint, and are significantly altered by aging and neurodegenerative diseases. Therefore, we have shown that TPI computed by subject-specific connectomes is crucial to accurately model subject-specific trajectories of tau spread.