Clinico-biological trajectories stratified by combined tau biomarkers in preclinical Alzheimer's disease.

Abstract

Background: Tau pathology biomarkers provide prognostic indicators of neurodegeneration and cognitive decline in Alzheimer's disease (AD), making them crucial to early patient stratification for disease-modifying interventions. Plasma p-tau217 indexes early tau pathophysiology, whereas FDA/EMA-approved [¹⁸F]flortaucipir PET (tau-PET) visual assessments indicate advanced neurofibrillary tangle pathology. Here, we identified concordant and discordant profiles based on combined plasma p-tau217 status and tau-PET visual assessments in cognitively unimpaired amyloid-β (Aβ)-positive older adults and investigated whether these profiles delineate distinct longitudinal trajectories of Aβ and tau accumulation, neurodegeneration, and cognitive decline.

Methods: We included 330 cognitively unimpaired Aβ-PET-positive participants (72.2 ± 4.8 years; 58% female; 48% randomized to receive solanezumab) from the A4 Study, who underwent plasma p-tau217 and tau-PET at baseline. Plasma positivity (T₁ +) followed A4 core criteria as previously reported (≥ 0.28 U/mL), while tau-PET positivity (T₂ +) was determined by three expert visual readers with strong inter-rater agreement (κ = 0.954). Participants were stratified as T₁[- / +]T₂[- / +] at baseline and were followed for 5.0 ± 1.7 years. Baseline and longitudinal differences in regional Aβ- and tau-PET SUVR, MRI-measured gray-matter (GM) volume, and cognitive performance were examined using ANCOVA and mixed-effects models. All models accounted for treatment, with additional sensitivity analyses excluding treated participants.

Results: At baseline, 57% were negative-concordant (T₁ - T₂ - : n = 187), 24% were discordant (T₁ + T₂ - : n = 53; T₁ - T₂ + : n = 27), and 19% were positive-concordant (T₁ + T₂ + : n = 63). T₁ + T₂ + profile constituted the highest-risk state, showing the greatest baseline cortical Aβ burden, the strongest neocortical tau progression (fronto-temporo-parietal pattern; p(FDR) < 0.05), marked baseline atrophy with accelerated longitudinal GM loss in overlapping regions, and the fastest cognitive decline (PACC: d = -1.84, p < 0.001). Consistently, T₁ + T₂ + exhibited the highest hazard of progression to more advanced clinico-biological stages over follow-up (HR_Clinical = 3.03, HR_Biological = 9.98; p < 0.001). Discordant profiles showed comparatively limited progression, suggesting earlier or low-tau states. Results were essentially unchanged after excluding treated participants.

Conclusions: Integrating plasma p-tau217 with tau-PET visual assessment reveals clinically meaningful tau-biomarker heterogeneity in preclinical AD. Our findings highlight the value of combining these biomarkers to refine early risk prediction and support prioritization strategies for prevention trials. The frequency of discordance also motivates refining tau-PET visual assessments beyond binary classification (e.g., ordinal/semi-quantitative staging) to better capture subtle early tau signal.