Alzheimer’s disease (AD) is the most common cause of dementia, characterized by amyloid-β plaques, tau neurofibrillary tangles, and progressive neuronal loss. Current diagnostic tools, including cerebrospinal fluid (CSF) analyses and positron emission tomography (PET) imaging, are accurate but invasive, costly, and not widely available. The identification of blood-based biomarkers reflecting AD neuropathology represents a major breakthrough toward accessible and scalable diagnostics.

This systematic review and meta-analysis aimed to evaluate the diagnostic performance of four plasma biomarkers—phosphorylated tau at threonine 181 (p-tau181), phosphorylated tau at threonine 217 (p-tau217), amyloid-beta 42/40 ratio (Aβ42/Aβ40), and neurofilament light chain (NfL)—for early and prodromal stages of AD. Following PRISMA 2020 guidelines, PubMed, Scopus, and Web of Science databases were searched for studies published between 2015 and 2025. Sixty-four studies encompassing 23,456 participants were included and analyzed using random-effects models (DerSimonian–Laird method).

The pooled area under the curve (AUC) was highest for plasma p-tau217 (AUC = 0.91, 95% CI: 0.88–0.94), followed by NfL (AUC = 0.86), p-tau181 (AUC = 0.84), and Aβ42/Aβ40 (AUC = 0.82). Combining p-tau217, NfL, and Aβ42/Aβ40 improved overall diagnostic accuracy (AUC = 0.94) with a sensitivity of 90%. Heterogeneity was moderate (I² = 47%) and mainly related to assay type and cohort variability.

These results demonstrate that plasma p-tau217 and NfL can achieve near-CSF accuracy for detecting AD pathology, providing robust, non-invasive, and cost-effective biomarkers suitable for large-scale screening and early intervention. Integrating plasma biomarkers with machine learning models may further enhance diagnostic precision, supporting a paradigm shift toward accessible and personalized approaches in Alzheimer’s disease detection and monitoring.