Disease timeline modelling of amyloid and tau PET reveals dynamic timescales of amyloid and tau accumulation in Alzheimer’s disease

Al Young, P. A. Wijeratne, Leon Aksman, Alexa Pichet Binette, Olof Strandberg, Neil P. Oxtoby, André Altmann, Daniel C. Alexander, Erik Stomrud, S. Palmqvist, Niklas Mattsson‐Carlgren, Jacob W. Vogel, Oskar H. HanssonAlzheimer 2025 · DOI

TL;DR

This study uses a novel temporal event-based model (T-EBM) to map the dynamic progression of amyloid and tau accumulation in Alzheimer's disease, revealing a consistent timeline of approximately 20 years. The findings suggest a slow initial amyloid buildup followed by a gradual, then accelerating, spread of tau pathology, offering crucial insights for understanding disease onset and potential intervention windows.

What Problem Does This Paper Address?

Alzheimer's disease (AD) is characterized by the accumulation of amyloid plaques and tau tangles in the brain. While it's understood that these pathologies precede cognitive decline, the precise timing and dynamics of their accumulation are not fully elucidated. This lack of detailed temporal understanding hinders the identification of optimal windows for therapeutic intervention and accurate prediction of disease progression. Existing models often struggle to account for the complex, dynamic nature of these biological processes and the variability in individual disease trajectories. This paper aims to establish a more precise timeline for amyloid and tau accumulation, providing a framework for understanding the disease's natural history.

Methods

The researchers employed a novel "explicit-duration" version of the temporal event-based model (T-EBM). This approach is designed to infer the order and timeline of biological events, specifically amyloid and tau accumulation, by accounting for the censoring of individuals' first and last visits and handling arbitrary time intervals between data points. This is a key innovation, allowing for more robust modeling of longitudinal data with irregular follow-ups. Regional tau-PET SUVR abnormality was computed using data-driven mixture modeling across five established regions. The T-EBM was then used to model the progression of global amyloid-PET and regional tau-PET abnormalities.

Datasets

The study utilized data from the Swedish BioFINDER-2 study. A total of 960 individuals were included, with varying numbers of tau-PET and amyloid-PET scans at follow-up intervals of 1, 2, 4, and 6 years. Specific subsets were analyzed: 773 individuals with at least two amyloid-PET scans for amyloid duration estimation, and 434 CSF-amyloid-positive individuals for comparison across the population.

Key Insights and Ideas

  • A Predictable Disease Timeline: The study provides a remarkably consistent estimated timeline of approximately 20 years for the combined progression of amyloid and tau pathology. This offers a concrete temporal framework for understanding AD's natural history.
  • Sequential and Accelerating Pathology: The findings strongly suggest a sequential process: amyloid accumulation occurs first, followed by tau pathology that begins in the entorhinal cortex and then spreads to other temporal lobe regions. Crucially, the tau spread is not linear; it starts slowly and then accelerates.
  • Braak-like Pattern Confirmed: The T-EBM inferred a tau accumulation pattern that closely resembles the well-established Braak staging system, lending further validation to the model and the observed pathological spread.
  • Progressors vs. Non-Progressors: Individuals who progressed in their T-EBM stage were found to be older, have more advanced symptoms, more APOE4 alleles, worse cognitive scores (MMSE), and were more frequently amyloid-positive. This highlights key demographic and genetic factors associated with faster disease progression.

Key Results

The T-EBM inferred an average timeline of global amyloid and regional tau accumulation of approximately 20 years. The progression from amyloid accumulation (Stage 1) to entorhinal tau (Stage 2) was estimated to take about 8 years. The transition from Stage 2 to temporal lobe tau (Stage 3) took approximately 5.5 years, with subsequent stages progressing in 2-3 year intervals. This timeline remained consistent even when focusing solely on amyloid-positive individuals. The study also showed that progressors were significantly older, had more APOE4 alleles, and lower MMSE scores compared to non-progressors.

Limitations and Future Work

The authors acknowledge that work is ongoing to validate these inferred timelines in additional datasets. While the T-EBM model accounts for censoring and arbitrary time intervals, the inherent complexity of biological processes and individual variability means these timelines represent averages. Future work will likely focus on refining these models, exploring the impact of other biomarkers, and applying these temporal insights to predict treatment response and identify optimal intervention strategies.

Summary updated

Abstract

Amyloid accumulates slowly, after which tau spreads from the entorhinal cortex to the temporal lobe, initially at a slower pace before accelerating to a faster rate across the cortex. This data indicates that slower rates of accumulation would be expected at earlier stages. Work is ongoing validating these timelines in additional datasets.
BibTeX
@inproceedings{young2025disease,
  author = {AL Young and P. A. Wijeratne and Leon Aksman and Alexa Pichet Binette and Olof Strandberg and Neil P. Oxtoby and André Altmann and Daniel C. Alexander and Erik Stomrud and S. Palmqvist and Niklas Mattsson‐Carlgren and Jacob W. Vogel and Oskar H. Hansson},
  title = {{Disease timeline modelling of amyloid and tau PET reveals dynamic timescales of amyloid and tau accumulation in Alzheimer’s disease}},
  booktitle = {Alzheimer},
  year = {2025},
  doi = {10.1002/alz70862\_110838},
}
Figure