# Plasma biomarkers predict long‐term longitudinal cognitive decline in individuals at‐risk for Alzheimer's disease: differences across demographic groups

## TL;DR

Plasma biomarkers can predict cognitive decline in people at risk for Alzheimer's, with some differences based on sex and other factors.

## Contribution

This study identifies plasma p-tau biomarkers as strong predictors of cognitive decline and reveals sex-specific effects of GFAP.

## Key findings

- Plasma p-tau217, p-tau181, and p-tau231 showed the strongest associations with cognitive decline.
- Plasma GFAP was linked to cognitive decline in men but not in women.
- Most biomarkers were robust across demographic factors like age, BMI, and education.

## Abstract

Plasma biomarkers are promising tools for detecting Ab pathology in cognitively unimpaired (CU) individuals. However, their association with cognitive trajectories and how this associations vary across demographic subgroups remain unclear. This study investigates a range of plasma biomarkers to determine whether baseline levels predict cognitive trajectories in CU individuals at‐risk of Alzheimer's disease (AD) and explores differences in these associations across age, sex, Body Mass Index (BMI), APOE‐ε4 carriership and years of education, regardless of amyloid status.

We included 235 CU participants at‐risk of AD from the ALFA+ study, who are visited every 3 years, and had cognitive trajectories assessed using a sensitive modified preclinical Alzheimer's cognitive composite (mPACC) over three visits (follow‐up: 6.56±0.52 years; Table 1). Baseline plasma biomarkers (p‐tau181 [Quanterix Advantage V2.1], p‐tau217 [Janssen], p‐tau231 [in‐house], Aβ42/Aβ40 [Quanterix N4PE], GFAP [N4PE], NfL [N4PE] and all p‐tau/Aβ42 ratios) were obtained ∼3.75 years before the first cognitive visit. Associations between these biomarkers and cognitive trajectories were analysed using linear mixed‐effects models, corrected by age, sex, BMI, years of education and time gap between biomarker and cognitive measurements. Interaction models assessed the impact of demographic characteristics (age, sex, BMI, APOE‐ε4 carriership and years of education) on these associations.

Six out of the nine plasma biomarkers were associated with mPACC changes (Figure 1), with plasma p‐tau fragments showing the strongest coefficients (βp‐tau217 = ‐0.07; βp‐tau181 = ‐0.06; βp‐tau231 = ‐0.05). Plasma GFAP showed a sex‐specific effect (Figure 2), with higher plasma GFAP being associated with mPACC decline in men only (βGFAP_men = ‐0.10, p =  0.010). The remaining demographic variables did not show any significant interaction for any biomarker. Notably, plasma p‐tau/Aβ42 ratios provided no additional predictive value beyond plasma p‐tau biomarkers alone.

Baseline plasma biomarkers, except for plasma Aβ42/Aβ40, predict longitudinal cognitive changes in CU individuals at‐risk of AD, with plasma p‐tau biomarkers exhibiting the strongest associations. Plasma GFAP exhibited sex‐specific effects, suggesting the need of personalized prediction models for specific biomarker uses. Importantly, though, most plasma biomarkers were generally robust across diverse demographic characteristics, supporting their general applicability for tracking cognitive decline in preclinical AD.