Plasma phospho-tau217 as a predictive biomarker for Alzheimer's disease in a large south American cohort

Background: Blood-based Alzheimer's disease (AD) biomarkers have been increasingly employed for diagnostic, prognostic, and therapeutic monitoring purposes, due to accuracy in distinguishing AD pathophysiologic process. Compared to other p-tau isoforms, plasma p-tau217 exhibits stronger associations with AD hallmarks in CSF and brain. However, most studies have been conducted in non-Hispanic Whites, limiting our understanding of the performances and utility of these biomarkers across ethnicities.

Methods: We examined a cohort of Peruvians from the GAPP study, a recently established cohort of Peruvian mestizos from Lima and indigenous groups from Southern Peru (Aymaras and Quechuas). We tested plasma levels of p-tau using the Quanterix Simoa ALZpathp-tau217 assay in 525 samples and tested the association between p-tau217 and clinical diagnosis (healthy controls n = 234 vs. AD n = 113) using generalized mixed regression models, adjusting for sex, age, education, APOE-e4 allele (fixed effects) and study site (random effect). We also tested biomarker levels in MCI (n = 178) vs. other groups. The receiver operating characteristics area under the curve (ROC-AUC) was used to evaluate the biomarker's classification performances.

Result: Participants showed on average 80% Native American ancestry. p-tau217 was significantly associated with AD (β = 2.61, 95%CI = 0.61-4.29) and its levels were inversely correlated with cognitive performances; p-tau217 levels did not differ between controls and MCI (p-value > 0.05). p-tau217 levels were higher in participants carrying at least one APOE-e4 allele (OR = 2.31, 95%CI = 1.85-2.90). The ROC-AUC for p-tau217 was estimated at 82.82% in the fully adjusted model.

Conclusion: To our knowledge, this is the largest study conducted in a South American cohort phenotyped for AD with available p-tau217. Most investigations have previously focused on highly selected cohorts with established AD-endophenotypes (CSF biomarkers, autopsy report, PET etc.), while data on cohorts with clinical assessment are currently lacking, especially in non-European populations.