Validation of plasma core biomarkers of Alzheimer's disease: A multicenter study in usual practice conditions

Background The new anti-amyloid therapies have brought the challenge of early and feasible identification of Alzheimer's disease (AD). Plasma biomarkers are promising tools, but real-world evidence remains limited. Objective We aimed to evaluate the diagnostic performance of plasma core AD biomarkers, while accounting for potential confounding factors. Methods Cross-sectional study of 285 patients (mean age 69.9 [range 50–85] years, 50.9% female) from five centers of the ReDeMa cohort. The diagnostic performance of plasma Aβ 42 /Aβ 40 , p-tau181, p-tau217, and p-tau217/Aβ 42 was tested against cerebrospinal fluid (CSF) amyloid (A) and tau (T) pathology using a centralized, chemiluminescence-based platform (Lumipulse © ). The potential influence of comorbidities, medications, neuropsychological variables, and apolipoprotein E gene ( APOE ) ε4 allele was analyzed, and center-related variability was examined. Results CSF amyloid (A+) was present in 216/285 (75.8%), while amyloid and tau pathology (A + T+) occurred in 191/283 (67.5%) patients. p-Tau217 displayed the best performance for detecting both A + (AUC 0.956) and A + T + (AUC 0.903). The optimal cutoff for A + was 0.234 pg/mL (95% CI 0.168–0.237), with an overall agreement of 95.8%. p-Tau217 performance showed some variability across centers (AUCs 0.916 to 1.000), but confidence intervals overlapped. APOE ε4 status ( r = 0.272), female sex ( r = 0.231), and cognitive performance ( r = −0.399) were associated with p-tau217, but multivariate models did not improve the diagnostic performance of p-tau217 alone. Significant associations were not found between p-tau217 and comorbidities or medications. Conclusions Plasma p-tau217 is established as a first-choice biomarker for the early detection of AD pathology in specialized clinical settings, displaying excellent diagnostic accuracy and minimal site-related variability.