Combinations of blood biomarkers in a real‐life memory clinic in Thailand

Abstract

AbstractBackgroundDifferent Alzheimer’s disease (AD) blood‐based biomarkers (BBM) exhibit unique patterns and trajectories throughout the natural history of AD, and thus, their combination may result in a higher diagnostic accuracy. While the usefulness of BBM combinations had been demonstrated in research cohorts, their performance in real‐life memory clinics is unclear. Additionally, previous BBM studies have been largely confined to high‐income countries and have often excluded non‐white populations. We aimed to evaluate the performance of BBM combinations in a real‐world clinical setting, specifically in a memory clinic in ThailandMethodThe study enrolled 64 memory clinic patients who had undergone amyloid‐positron emission tomography (PET) or cerebrospinal fluid (CSF) assessment for AD biomarkers. Plasma amyloid‐β42 (Aβ42), Aβ40, p‐tau181 were quantified for each participant using single molecule array as well as APOE genotype determined. The accuracy of individual biomarkers in discriminating amyloid status (as confirmed by PET or CSF) were assessed using receiver operating characteristic (ROC) analysis. Logistic regression models were then created for different BBM combinations, and their performance was determined using ROC analysis. To ensure robustness and generalizability of the results, two‐fold cross‐validation was repeated 10 times on the best‐performing modelResultThe study consisted of 47 (73.4%) female participants with a median age of 70.0 years (interquartile range: 63.75‐75.25). Amyloid status was positive in 43 (67.2%) of the participants. Plasma Aβ42/Aβ40 ratio, Aβ42/p‐tau181 ratio and p‐tau181 had areas under the curve (AUC) of 0.68 (95% confidence interval (CI) 0.52‐0.84), 0.88 (95%CI 0.79‐0.98) and 0.87 (95%CI 0.79‐0.96) for detecting amyloid status, respectively. Logistic regression models that included APOE genotype, Aβ, and p‐tau181 had the highest AUC of 0.90‐0.91. Two‐fold cross‐validation performed on the model that included APOE genotype and the Aβ42/p‐tau181 ratio revealed an average AUC of 0.90 (standard deviation: 0.029) over 10 repeatsConclusionThe combination of APOE genotype, plasma Aβ, and p‐tau181 displayed an exceptionally high diagnostic accuracy in discriminating amyloid status in a memory clinic setting. However, further studies are needed to evaluate the cost‐effectiveness of various ways to implement BBMs into real‐world clinical practice, particularly in low‐ and middle‐income countries