Public Health

Area of Science:

• Neuroscience
• Genetics
• Biostatistics

Background:

• Unsupervised clustering of omics data for Alzheimer's Disease (AD) subtyping yields clinically irrelevant subtypes.
• A novel supervised clustering method was developed to integrate longitudinal clinical and omics data.
• This approach captures age-dependent effects of AD risk factors like ApoE4, improving subtype clinical relevance.

Purpose of the Study:

• To develop and apply a supervised clustering method for identifying clinically meaningful Alzheimer's Disease (AD) subtypes.
• To integrate longitudinal clinical data with high-dimensional omics data for improved AD subtyping.
• To characterize AD subtypes based on cognitive trajectories, clinico-pathological features, and molecular signatures.

Main Methods:

• Applied a supervised clustering method to 1,020 adults from the ROS/MAP study.
• Integrated longitudinal global cognition trajectories and differentially expressed genes from the prefrontal cortex.
• Performed survival analysis, clinico-pathological assessments, and differential gene expression/pathway analyses.

Main Results:

• Identified four distinct AD subtypes with a gradient of cognitive decline from Subtype 1 (slowest) to Subtype 4 (fastest).
• Subtype 4 exhibited the highest prevalence of AD dementia, beta-amyloid and tau pathology, and vascular pathologies, with the steepest cognitive decline.
• Survival analysis indicated increasing AD dementia risk across subtypes, with Subtype 4 showing a 77.0% probability at age 85.

Conclusions:

• An integrative approach successfully identified clinically meaningful AD subtypes with distinct cognitive, vascular, and molecular profiles.
• These subtypes offer critical insights for advancing precision medicine and developing targeted interventions for Alzheimer's Disease.
• The findings highlight the heterogeneity of AD and the potential for tailored therapeutic strategies.