A trichotomy method for defining homogeneous subgroups in a dementia population.

Area of Science:

• Neurology
• Biomarkers
• Neurodegenerative Diseases

Background:

• Diagnosing dementia in aging individuals is challenging due to multiple co-existing pathologies.
• Mixed dementia (MX), combining Alzheimer's disease (AD) and vascular disease (VD), is common but difficult to diagnose pre-mortem.
• Existing diagnostic methods struggle to differentiate between AD, VD, and their combination.

Purpose of the Study:

• To develop and validate a novel method for differentiating mixed dementia (MX) and identifying preclinical Alzheimer's disease (AD) and preclinical vascular disease (VD) subgroups.
• To improve the in-vivo diagnosis of complex dementia pathologies.
• To define distinct patient groups based on biological and cognitive factors.

Main Methods:

• A trichotomy method was developed, utilizing three diagnostic axes: AD factor (ADF) from CSF biomarkers, VD factor (VDF) from white matter integrity, and cognitive function (Cog).
• The method was applied to a cohort of 538 participants from the Alzheimer's Disease Neuroimaging Initiative (ADNI).
• Clustering analysis identified distinct patient subgroups based on ADF, VDF, and Cog.

Main Results:

• Eight biologically defined subgroups were identified, including MX (9.3%), preclinical VD (3.9%), and preclinical AD (13.6%).
• Cognition was significantly associated with both ADF and VDF, with independent contributions from each factor.
• Partial correlations confirmed the significant impact of ADF (r=0.46) and VDF (r=0.24) on cognition, even after accounting for the other variable.

Conclusions:

• The trichotomy method successfully creates eight distinct patient groups, including MX, preclinical AD, and preclinical VD.
• This approach offers a promising tool for the in-vivo characterization of dementia subtypes.
• Longitudinal studies are recommended to further validate the utility of this multimodal biomarker clustering method.