Abstract
INTRODUCTION: Cross-cohort validation studies for plasma p-tau217 are limited. We evaluated the Janssen plasma p-tau217+ assay and proposed a cutpoint value in three independent community-based cohorts. METHODS: We included n=441 participants (age=70.3±7.3), with Aβ-PET, tau-PET, clinical and cognitive information. RESULTS: The cohorts had low pre-test probability (%Aβ positivity=14.9-24.7) and were predominantly cognitively normal (>73%). Plasma p-tau217+ had high accuracy for abnormal Aβ PET (AUCs=81-86%), good correlation with Aβ-PET burden (0.336-0.397) that was highest in the cohort with the most Aβ-PET-positive participants, and the biomarker concentrations were highest in the joint Aβ-PET and tau-PET positive group. NPV was high across cohorts (≤93%) but PPV was consistently poor (<57%). Sensitivity and specificity averaged 75% and 84% respectively. A combined cohort cutpoint of 0.05pg/ml gave AUC=84.5%, NPV=94%, PPV=50%, sensitivity=75%, and specificity=84%. DISCUSSION: Plasma p-tau217+ can rule out Aβ pathophysiology due to Alzheimer's disease at the population level. Cohort-level %Aβ-PET positivity influences accuracies.