Alzheimer's disease (AD) is characterized by amyloid-β plaques and abnormal phosphorylated tau protein-containing neurofibrillary tangles, yet it shows marked heterogeneity in clinical presentation, neuroanatomical involvement, and progression rate. This variability challenges the traditional view of AD as a single disease entity and has prompted efforts to define patient subgroups with shared characteristics who might benefit from targeted treatment strategies. Here, we provide a cross-disciplinary overview of current subtyping strategies. We describe both traditional stratification approaches, including those based on clinical AD syndromes, neuropathological staging, and age of onset, as well as emerging data-driven methods that utilize clinical information, neuroimaging, omics data, and multimodal data integration. Subtypes derived using these data-driven methods overlap to some extent with hypothesis-driven classifications but also uncover additional axes of heterogeneity, including distinct anatomical patterns and molecular signatures. Furthermore, we highlight that most AD patients exhibit co-pathologies such as transactive response DNA-binding protein 43, α-synuclein, or cerebrovascular changes, which are often overlooked in existing stratification systems, despite their potential to affect atrophy patterns and progression rate and, hence, influence subtype interpretation. Finally, we outline future directions for developing unified stratification frameworks that link clinical features with underlying biology, including co-pathologies, aiming to enhance diagnostic precision and enable future personalized therapeutic interventions.
Gawor et al. (Fri,) studied this question.