Head-to-head comparison of plasma biomarker assays across platforms for amyloid pathology in a multicenter cohort.
TL;DR
Several plasma p-tau217 assays demonstrated strong diagnostic accuracy for amyloid PET positivity in this cross-sectional memory-clinic cohort, although performance varied across platforms and cutoff transferability was limited.
Abstract
Background
With the advancement of anti-amyloid treatments for Alzheimer disease, accurate assays for amyloid-β (Aβ) pathology are essential. Plasma phosphorylated tau 217 (p-tau217) is a blood-based biomarker, but its performance varies across platforms, necessitating head-to-head comparisons.
Methods
This multicenter study evaluated 9 plasma p-tau217 assays (6 including Aβ42 measurements) for detecting amyloid positron emission tomography (PET) positivity. The final analysis included 431 participants from 10 memory clinics in China (median age, 68.0 years; 61.7% women; 64.0% amyloid PET-positive): 30 cognitively unimpaired individuals, 230 with mild cognitive impairment, and 171 with dementia. All plasma samples underwent blinded, batch-matched testing in a central laboratory across chemiluminescence immunoassay (Fujirebio, Beckman, Vazyme, manufacturer A), single-molecule immunoassay (Quanterix, Lychix, iomicsBio, manufacturer B), and multiplex bead-based flow cytometric immunoassay (CellGene).
Results
Seven of the 9 assays showed acceptable performance (area under the curve [AUC] 0.899-0.930), whereas 2 showed lower performance (AUC <0.800; P < 0.001). Under a two-cutoff approach (90% sensitivity/90% specificity), these 7 high-performing assays yielded an intermediate zone of 2.8% to 13.7%. Performance remained robust in mild cognitive impairment and dementia subgroups. Adding Aβ42 showed assay- and population-dependent effects. Manufacturer-recommended and previously published cutoffs showed variable performance in this independent cohort.
Conclusions
Several plasma p-tau217 assays demonstrated strong diagnostic accuracy for amyloid PET positivity in this cross-sectional memory-clinic cohort, although performance varied across platforms and cutoff transferability was limited. Further longitudinal studies with predefined clinical outcomes are needed to determine prognostic value and clinical utility.