Beyond statistical thresholds: Mapping sequential ATN cascade transitions via brain structural topography in...

Alzheimer's disease is widely conceptualized as a biological continuum initiated by amyloid-β (Aβ) accumulation, followed by tau aggregation and neurodegeneration. Although the ATN framework is commonly used to define disease stages, its static biomarker categories may not fully capture the dynamic pathological transitions that determine the optimal window for anti-Aβ therapies. Here, we introduce a Biological Continuum framework that distinguishes dynamic pathological cascades from static biomarker categories. Using neuroimaging-derived brain network topology, we identify four reproducible subgroups that exhibit distinct biological processes, biomarker signatures, and longitudinal cognitive trajectories. Integrated proteomic analyses suggest that these subgroups reflect sequential molecular transitions corresponding to early vulnerability, initiation, amplification, and terminal stages. Post hoc analysis indicates that slower cognitive decline after amyloid-targeting treatment was observed mainly in the initiation stage group. These findings provide a biologically grounded stratification system that captures disease dynamics overlooked by threshold-based models and offers mechanistic insight into the timing of anti-Aβ therapies.