Clasificación ATN y fenotipos de biomarcadores de plasma aprendidos por máquina revelan patología distinta de...

The ATN (Amyloid/Tau/Neurodegeneration) framework provides a theory-driven approach to Alzheimer's disease (AD) classification using binary biomarker cutoffs, while unsupervised machine learning offers data-driven phenotyping. The concordance between these approaches in population-representative samples remains incompletely characterized.

To compare plasma ATN classification with data-driven clustering methods and evaluate their associations with cognitive outcomes in a nationally representative cohort.

We analyzed plasma biomarkers (Aβ42/40 ratio, p-tau181, NfL, GFAP) from 4,465 participants aged ≥51 years in the Health and Retirement Study 2016 Venous Blood Study. ATN profiles were classified using literature-based cutoffs. We applied k-means clustering, Gaussian mixture modeling, and variational autoencoder (VAE) dimensionality reduction to identify data-driven biomarker phenotypes. Agreement between ATN and clustering was quantified using adjusted Rand index (ARI) and normalized mutual information (NMI). Longitudinal analyses examined associations with cognitive decline over 4 years (2016-2020).

The analytic sample included 4,465 individuals (mean age 69.7±10.4 years; 58.7% female; 75.8% non-Hispanic White). ATN classification yielded 14 profiles, with A+/T-/N- (27.4%) and A-/T-/N-(22.6%) most prevalent. K-means clustering identified 4 optimal clusters with distinct biomarker signatures. Agreement between ATN and clusters was modest (ARI=0.119, NMI=0.113). Sensitivity analysis excluding GFAP from clustering improved agreement to ARI=0.187 (+57% relative increase), indicating that GFAP's orthogonal biological information accounts for approximately one-third of discordance, while binary categorization versus continuous phenotyping accounts for two-thirds.[Table S12] Additional sensitivity analyses confirmed that k=4 provides superior biological resolution over k=3 by preserving extreme phenotypes,[Table S13] and that Cluster 4 represents a stable biological structure across distance metrics[Table S14] despite its small size. Cluster 1 (n=51, 1.2%) showed severe pathology; Cluster 3 (n=3,479, 78.6%) represented the largest and most heterogeneous group, encompassing the broad spectrum of minimal to moderate pathology across all ATN profiles; Cluster 4 (n=14, 0.3%) exhibited a non-AD neurodegeneration pattern with high stability (Jaccard=0.779). VAE revealed localized nonlinear structure, though PCA achieved higher global separation (silhouette: PCA=0.671 vs VAE=0.564). Both ATN and clusters predicted 4-year cognitive decline (ATN R ² =0.024, p<0.001; Clusters R ² =0.019, p<0.001).

Theory-driven ATN classification and data-driven biomarker phenotyping capture partially overlapping but largely distinct biological information. Modest concordance (ARI=0.119) reflects three primary factors: binary cutoffs discarding continuous information (dominant), GFAP's orthogonal inflammatory signature (contributing ~one-third), and fundamental differences in biological constructs targeted. Sensitivity analyses confirmed that k=4 provides superior biological resolution over k=3, and that rare Cluster 4 represents a stable non-AD phenotype. Both approaches predict cognitive decline with modest effect sizes (R ² =1.9-2.4%) consistent with population-based studies. Integrating theory-driven and data-driven frameworks promises a more comprehensive characterization of AD-related pathology in population research.