Summary: An international research team has identified four distinct subtypes of Alzheimer’s disease based on patterns of tau protein accumulation. The discovery could enable more personalized diagnosis and treatment strategies for people with this neurodegenerative disorder.
Source: Lund University
Alzheimer’s disease is marked by the abnormal accumulation and spread of tau protein across the cerebral cortex. A large international study now shows that tau follows at least four different spatiotemporal trajectories, each linked to different clinical profiles and likely different prognoses.
The findings were published in Nature Medicine.
“Until now, many models have assumed a single, typical pattern for how tau spreads in Alzheimer’s. Our results show that tau pathology actually follows multiple, consistent patterns across individuals. This means Alzheimer’s is more heterogeneous than commonly believed and suggests we should reconsider how we define ‘typical’ Alzheimer’s and how we stage disease progression,” says Jacob Vogel from McGill University, the study’s lead author.
Tracking tau build-up in living brains has become possible with tau-PET imaging, an advanced positron emission tomography technique. Using these scans, the researchers analyzed the largest and most diverse tau-PET dataset assembled to date, drawn from sites in Sweden, Canada, the United States and Korea. The sample included people across the full clinical spectrum: those without symptoms but with early biomarker evidence of disease, individuals with mild cognitive impairment, and people with established Alzheimer’s dementia.
Longitudinal data were gathered from 1,612 participants enrolled in five independent multicenter studies. From these, 1,143 individuals who were either cognitively normal or at different stages of Alzheimer’s disease were selected for detailed analysis. To identify unbiased subtypes, the team applied a machine-learning algorithm called SuStaIn (Subtype and Staging Inference) to the tau-PET images. This automated approach separates distinct patterns and stages of pathology without relying on preconceived clinical categories.
Many participants showed no abnormal tau signal and were categorized as tau-negative. For the remaining individuals, cross-validation against a sixth independent cohort and roughly two years of follow-up data supported a solution of four reproducible tau trajectories. Each subtype appeared across cohorts and remained stable in longitudinal follow-up.
“We discovered four clear patterns of cortical tau deposition that become increasingly distinct over time. Each pattern was relatively common, with subgroup prevalence ranging from about 18 to 33 percent. No single pattern accounted for most cases, which challenges the idea of a uniform disease course,” says Oskar Hansson, professor of neurology at Lund University and senior author of the study.
- Variant one: Tau accumulation is concentrated mainly in the medial temporal lobe, producing a primarily amnestic presentation with prominent memory impairment. This variant accounted for roughly 33% of cases.
- Variant two: Tau spreads more broadly across neocortical regions outside the temporal lobe. Patients show fewer memory problems but greater impairment in executive functions, such as planning and complex task execution. This variant represented about 18% of cases.
- Variant three: Tau is concentrated in posterior regions including the visual cortex, leading to visuospatial and perceptual difficulties. Affected individuals may struggle with orientation, shape recognition, depth perception and locating objects in space. This pattern occurred in about 30% of cases.
- Variant four: Tau deposition is asymmetric and predominantly affects the left hemisphere, with language dysfunction as the principal clinical feature. This subtype represented about 19% of cases.
The researchers also compared their tau-based subtypes with demographic and cognitive measures, finding distinct profiles and differing outcomes over time. Network diffusion models further suggested that pathology in each subtype originates and propagates along different corticolimbic networks, indicating biologically plausible mechanisms for the distinct patterns.
“Large, diverse tau-PET datasets combined with modern machine-learning tools made it possible to detect and characterize these four subtypes. Confirming and refining these trajectories will require longer follow-up—ideally five to ten years—but the current evidence is robust and consistent across independent cohorts,” says Oskar Hansson.
Understanding these subtypes has immediate clinical relevance: it can help clinicians better interpret symptoms and prognosis and could guide more tailored therapeutic approaches. As multiple tau-reducing treatments are developed and tested, subtype-specific responses may emerge, highlighting the importance of stratifying patients by tau trajectory in clinical trials.
About this Alzheimer’s disease research news
Source: Lund University
Contact: Oskar Hansson – Lund University
Image: The image is in the public domain
Original Research: Closed access. “Four distinct trajectories of tau deposition identified in Alzheimer’s disease” by Jacob W. Vogel, Alexandra L. Young, Neil P. Oxtoby, Ruben Smith, Rik Ossenkoppele, Olof T. Strandberg, Renaud La Joie, Leon M. Aksman, Michel J. Grothe, Yasser Iturria-Medina, the Alzheimer’s Disease Neuroimaging Initiative, Michael J. Pontecorvo, Michael D. Devous, Gil D. Rabinovici, Daniel C. Alexander, Chul Hyoung Lyoo, Alan C. Evans & Oskar Hansson. Published in Nature Medicine.
Abstract
Four distinct trajectories of tau deposition identified in Alzheimer’s disease
Alzheimer’s disease is defined by the spread of tau pathology through the cerebral cortex. Prior models treated this spread as largely uniform across individuals, but recent evidence has pointed to substantial variability. Using tau-PET scans from 1,612 individuals, the study identified four reproducible spatiotemporal tau trajectories with prevalences between 18% and 33%.
The analysis replicated previously known limbic-predominant and medial temporal lobe-sparing patterns and additionally revealed posterior and lateral temporal trajectories resembling atypical clinical variants. These subtypes were stable over time, replicated in independent samples and associated with distinct demographic, cognitive and longitudinal outcome profiles. Network diffusion modeling suggested that each subtype reflects propagation through different brain networks.
Overall, the results indicate that variation in tau pathology is common and systematic, supporting a re-evaluation of the traditional notion of ‘typical Alzheimer’s’ and prompting reconsideration of tau staging approaches.