Improving the Diagnostic Accuracy of Plasma Alzheimer's Disease Biomarkers With Objective and Subjective Cognitive Screening

Abstract

ABSTRACT Background Blood‐based biomarkers are increasingly used to identify Alzheimer's disease (AD) pathology and may play an important role in early disease screening and participant selection for clinical trials. However, their performance across stages of cognitive impairment remains unclear. This study examined whether plasma biomarkers differ in their ability to predict amyloid pathology by performance on cognitive screening measures. Methods Participants were older adults from the community‐based Vanderbilt Memory and Aging Project (N = 154; mean age = 72 ± 6 years; 37% MCI). Plasma phosphorylated tau (p‐tau)231 and glial fibrillary acidic protein (GFAP) were measured using Quanterix Simoa technology. Amyloid positivity was defined by cerebrospinal fluid β‐amyloid 42/40 ratio < 0.072. Objective cognitive impairment was defined as ≤ 26 on the Montreal Cognitive Assessment; cognitive complaints were assessed by a single‐item self‐report. Logistic regression and stratified ROC analyses evaluated associations between biomarkers and amyloid status, with area under the ROC curve (AUC) values compared using DeLong's test. Results Forty‐seven percent of participants were amyloid positive. Higher plasma p‐tau231 was associated with increased odds of amyloid positivity (OR = 1.18, p < 0.001; AUC = 0.73). Discriminative accuracy was significantly higher among individuals without subjective cognitive concerns (AUC = 0.85) than among those individuals with concerns (AUC = 0.67; p = 0.02). Plasma GFAP was also associated with amyloid positivity (OR = 1.01, p < 0.001; AUC = 0.75) and showed greater accuracy among participants with objective cognitive impairment (AUC = 0.82) compared to unimpaired individuals (AUC = 0.63; p = 0.03). Conclusions Plasma p‐tau231 and GFAP were associated with amyloid pathology, but their diagnostic accuracy varied by cognitive screening results, highlighting the value of integrating brief cognitive assessments when interpreting blood‐based biomarkers.