A clinically feasible framework to estimate tau pathology and clinical-biological discordance in the Alzheimer's disease spectrum.
BACKGROUND: Tau positron emission tomography (PET) is critical for biological staging and treatment stratification in Alzheimer's disease (AD), particularly in the era of anti-amyloid therapies where lower tau burden predicts greater clinical benefit. However, tau PET remains costly and inaccessible in many clinical settings. We aimed to develop and validate a clinically feasible framework to estimate global and regional tau burden using routinely available amyloid PET and clinical measures, aligned with the 2024 Alzheimer's Association (AA-2024) diagnostic framework, and to characterize clinical-biological discordance across the AD continuum. METHODS: We conducted a cross-sectional study of 229 individuals spanning cognitively unimpaired, mild cognitive impairment, and dementia stages who underwent [18F]florbetapir amyloid PET, [18F]florzolotau tau PET, structural MRI, and cognitive assessment. Amyloid burden was quantified using the Centiloid (CL) scale (A+ defined as CL > 20). Tracer-specific tau thresholds for global and Braak-stage volumes were derived using two-component Gaussian mixture modeling. Logistic regression models incorporating CL, age, Mini-Mental State Examination (MMSE), and medial temporal lobe (MTL) volume were developed to classify high global tau burden and neocortical tau involvement. Among amyloid-positive individuals, biological staging was compared with clinical stage to assess discordance patterns. RESULTS: Global and regional tau burden increased stepwise across clinical severity and amyloid strata. High global tau burden was uncommon in individuals with CL 21-60 (6.1%) but increased in CL 61-100 (22.6%) and > 100 (36.2%). A multivariable model integrating CL, age, MMSE, and relative MTL volume demonstrated good discrimination for high global tau burden (AUC = 0.87) and neocortical involvement (AUC = 0.84). Model robustness was confirmed by bootstrap resampling. Among amyloid-positive participants, 53.1% exhibited clinical-predominant AD, characterized by older age and higher cardiovascular risk despite relatively modest tau burden, indicating substantial clinical-biological discordance. CONCLUSION: Routinely obtainable amyloid PET and clinical measures can approximate global and topographical tau burden with good accuracy and identify frequent clinical-biological discordance within the AD spectrum. This scalable framework provides a practical surrogate for tau PET in resource-limited settings and may support biological staging, therapeutic eligibility assessment, and precision treatment decision-making.