Although blood-based biomarkers are now available for diagnosing Alzheimer’s disease (AD), the best biomarker for AD-specific neurodegeneration and cognitive decline remains unclear. This study aimed to determine the relative importance of four plasma biomarkers—phosphorylated tau (p-tau) 217, p-tau181, neurofilament light chain (NfL), and glial fibrillary acidic protein (GFAP)—in AD-specific neurodegeneration and cognition.

We analyzed cross-sectional data from two independent, ethnically distinct cohorts spanning the clinical spectrum from cognitively unimpaired to dementia: 150 participants from the SAMD cohort (100% Asian) and 284 participants from the ADNI cohort (94.0% White). Plasma biomarker levels were quantified using Single-Molecule Array (Simoa) assays. We employed dominance analysis to determine the hierarchical contributions of these biomarkers to AD signature regions of interest (ROI) thickness (entorhinal, inferior temporal, middle temporal, and fusiform regions), total cognition, and memory, stratified by amyloid-PET status. Three sensitivity analyses were further conducted to validate the findings across these cohorts, mitigating potential biases arising from differences in demographic characteristics and clinical severity. All analyses were adjusted for age, sex, education, and APOE ε4 status.

The dominance hierarchy differed markedly according to the amyloid status. Plasma GFAP and p-tau217 emerged as the dominant predictors for AD signature ROI thickness and cognitive impairment in amyloid-positive participants. Specifically, GFAP demonstrated superior dominance in explaining cortical atrophy within the SAMD cohort, whereas p-tau217 was the dominant predictor in the ADNI cohort. P-tau217 generally outperformed the others in explaining total cognition and memory in the amyloid (+) group. In contrast, among amyloid (−) participants, plasma NfL showed greater explanatory power than GFAP for both neurodegeneration and cognitive decline across both cohorts.

The efficacy of plasma biomarkers in reflecting AD-related neurodegeneration varies significantly depending on the presence of amyloid pathology. While GFAP and p-tau217 are robust indicators of AD-associated changes linked to plaque pathology, NfL better reflects non-specific neurodegeneration involving axonal damage. Consequently, a stratified approach based on amyloid status is essential for the optimal application of blood-based biomarkers in monitoring disease progression and evaluating therapeutic efficacy in future clinical trials and precision medicine.