Plasma p-tau217 as a two-cutoff approach to improve tau-PET screening and trial recruitment in Alzheimer’s disease

Abstract

Blood-based biomarkers could broaden access to Alzheimer’s disease diagnostics, but their ability to approximate tau PET, and to address gray-zone results, remains uncertain. We evaluated plasma p-tau217 against tau PET using a harmonized ADNI baseline cohort and an independently derived temporal meta-ROI tau-PET threshold. We included participants with complete baseline plasma biomarkers (p-tau217, Aβ42, Aβ40, NfL, GFAP), tau PET, and demographics, including: cognitively normal = 229, mild cognitive impairment = 149, and Alzheimer’s disease dementia = 41. ROC analyses assessed discrimination of tau-PET positivity overall and by diagnostic group. A two-cutoff strategy was implemented to minimize indeterminate classifications. Across the pooled cohort, p-tau217 demonstrated the strongest ability to discriminate tau-PET status (AUC 0.671, 95% CI 0.597–0.746), outperforming Aβ42, Aβ40, and Aβ42/Aβ40; Though overall accuracy remained moderate and performance differed based on diagnosis, with AUCs of 0.785 for Alzheimer’s disease dementia and lower values for cognitively normal and mild cognitive impairment. The two-cutoff approach substantially reduced intermediate classifications compared with conventional markers. In ADNI, plasma p-tau217 showed the highest discriminative performance among the single blood biomarker for detecting tau-PET positivity, showing moderate accuracy overall and its strongest performance in Alzheimer’s disease dementia. A two-cutoff approach substantially reduces gray-zone classifications, supporting p-tau217 as a primary screening tool before confirmatory imaging. Further work and validation are needed across different groups.