Prognostic Value of A/T/N Biomarkers: Comparing Plasma and Imaging Modalities in Alzheimer Disease

Clin Nucl Med. 2025 Sep 1. doi: 10.1097/RLU.0000000000006088. Online ahead of print.

ABSTRACT

BACKGROUND: Alzheimer disease (AD) is characterized by amyloid-β plaques (A), tau tangles (T), and neurodegeneration (N), collectively defining the ATN framework. While imaging biomarkers are well-established, the prognostic value of plasma biomarkers in predicting cognitive decline remains underexplored. This study compares plasma and imaging A/T/N biomarkers in predicting cognitive decline and evaluate the impact of combining biomarkers across modalities.

PATIENTS AND METHODS: We conducted a longitudinal study using K-ROAD cohort participants who underwent at least 2 cognitive assessments. All participants had plasma biomarker testing (Aβ ratio, p-tau181, p-tau231, p-tau217, NfL), and a subset with imaging biomarker assessments (Aβ PET, tau PET, structural MRI) formed an imaging subcohort. Multiple linear regression models identified the most predictive markers within each modality and evaluated the effect of combining A/T/N biomarkers.

RESULTS: Among 1,614 plasma cohort and 130 imaging subcohort participants, tau markers demonstrated the strongest predictive value. p-tau217MSD outperforming other plasma biomarkers, and the neo-temporal ROI showing the highest predictive power among imaging biomarkers. In plasma-based model, adding neurodegeneration markers to combination of amyloid and tau biomarkers improved the performance. In imaging-based models, same strategy decreased the performance, suggesting that combinations of amyloid and tau PET captures the most relevant prognostic information.

CONCLUSIONS: Imaging biomarkers, particularly tau PET, show superior prognostic accuracy compared with plasma biomarkers, whereas plasma biomarkers offer advantages in combination models through neurodegeneration markers. These findings underscore the complementary roles of plasma and imaging biomarkers and emphasize the need for tailored strategies for prognostic modeling in AD.

PMID:40910876 | DOI:10.1097/RLU.0000000000006088