Abstract
OBJECTIVE: This study aimed to compare positron emission tomography (PET) and plasma-based temporal modeling of amyloid and tau biomarkers in Alzheimer's disease.
METHODS: Longitudinal amyloid PET (n = 1,097, mean age ± SD = 72.5 ± 7.38 year, 51.4% male), 18F-flortaucipir tau-PET (n = 230, 74.3 ± 7.18 year, 52.2% female), and Fujirebio Lumipulse plasma p-tau217 (n = 752, 72.8 ± 6.93 year, 51.3% male) from the Alzheimer's Disease Neuroimaging Initiative (ADNI) and University of Pennsylvania Alzheimer's Disease Research Center (Penn ADRC) were used to generate biomarker trajectory models using sampled-iterative Local approximation (SILA). SILA models using plasma p-tau217 were compared to amyloid and tau PET-based models to estimate amyloid and tau onset, and factors influencing tau onset and time from tau onset to dementia were evaluated for PET and plasma models.
RESULTS: Plasma and PET models generated similar results for estimated amyloid and tau onset, with stronger model agreement for tau (r = 0.88[0.86, 0.89], t = 57.4, p < 0.001) than amyloid (r = 0.75[0.72, 0.77], t = 37.4, p < 0.001) onset. Accuracy of estimated onset compared to actual onset was high within modality (mean absolute error [MAE] ≤ 2.03) with slightly greater error (MAE 3.09-3.42) when comparing across modalities (ie, plasma to PET). For both plasma and PET, earlier tau onset was associated with younger amyloid onset, female sex, and ≥1 apolipoprotein (ApoE) ε4 allele. Earlier dementia onset after tau was associated with later tau onset for both plasma and PET, while male sex was associated with shorter tau to dementia gap in plasma models.
INTERPRETATION: Temporal modeling of plasma biomarkers provides comparable information to PET-based models, particularly for tau onset age, and can serve as a widely accessible tool for clinical assessment of biological disease severity. ANN NEUROL 2026.