Blood-based biomarkers have emerged as promising tools for detecting Alzheimer’s disease (AD) pathology, but validation of automated plasma assays in Chinese clinical populations remains limited. This study evaluated the diagnostic performance of a fully automated chemiluminescent plasma biomarker assay for detecting amyloid-β (Aβ) pathology in a Chinese memory clinic cohort under different pre-analytical conditions.

We enrolled 409 cognitively impaired participants from a single-center memory clinic, using amyloid-β positron emission tomography (Aβ-PET) as the reference standard. Plasma samples were analyzed under two pre-analytical conditions: frozen batch-processed samples from a historical cohort (n = 198) and freshly collected samples analyzed in real time in a prospective cohort (n = 211). Additionally, 95 participants underwent tau-PET imaging. Six plasma biomarkers were quantified using the Vazyme® AD Assay.

Across cohorts, p-tau217, p-tau217/Aβ42 ratio, and NfL/p-tau217 ratio consistently achieved excellent diagnostic performance (AUC 0.92–0.95), followed by p-tau181 (AUC 0.86–0.90). GFAP (AUC 0.82–0.83) and the Aβ42/40 ratio (AUC 0.76–0.81) showed moderate discriminative performance. Plasma p-tau217 alone achieved diagnostic accuracy comparable to composite biomarker models. A dual cut-point strategy reduced the indeterminate zone to <30%, with positive predictive values of 0.97–0.99 and negative predictive values of 0.86–0.87. Plasma p-tau217 was also significantly associated with tau-PET burden in both meta-temporal and neocortical regions (P < 0.001).

This automated chemiluminescent plasma biomarker assay demonstrated high diagnostic accuracy for detecting Aβ pathology in a Chinese memory clinic cohort under different pre-analytical conditions. The findings support its potential utility as a practical blood-based biomarker approach in specialized clinical settings, while further multicenter studies are needed to confirm its generalizability across broader populations and healthcare environments.