CSF Proteomics and Machine Learning Reveal Distinct Stages Across the Alzheimer's Disease Continuum

Area of Science:

• Neuroscience
• Proteomics
• Biomarker Discovery

Background:

• Alzheimer's disease (AD) involves complex, early pathophysiological changes not fully captured by current biomarkers like Aβ and pTau.
• Existing diagnostic and therapeutic approaches are limited by the incomplete understanding of AD's molecular heterogeneity.
• High-resolution cerebrospinal fluid (CSF) proteomics offers a deeper insight into AD pathogenesis.

Purpose of the Study:

• To identify novel protein signatures in CSF associated with Alzheimer's disease (AD) pathogenesis and progression.
• To develop machine learning models for accurate staging of AD and estimation of pathological burden.
• To uncover stage-specific molecular events and pathways across the AD continuum.

Main Methods:

• Quantified 2,492 proteins in CSF from 1,104 ADNI participants using tandem-mass-tag mass spectrometry (TMT-MS).
• Analyzed protein abundance changes across asymptomatic AD, MCI (due-to-AD), and AD Dementia stages.
• Developed machine learning models for disease stage classification and pathological burden estimation (Aβ-PET, tau-PET).

Main Results:

• Identified 92 differentially abundant proteins across the AD continuum, revealing stage-specific pathway alterations.
• Observed upregulated neuropeptide signaling, G-protein-coupled receptors, and synaptic remodeling in MCI (due-to-AD) and AD Dementia.
• Machine learning models achieved high accuracy in distinguishing disease stages (AUC=0.92 for asymptomatic vs. MCI, AUC=0.87 for MCI vs. Dementia) and estimating pathological burden.

Conclusions:

• CSF proteomic signatures reflect the progressive nature of AD, from early pathway disruptions to later neuronal degeneration.
• Novel protein panels and machine learning models show promise for improved AD diagnosis, staging, and patient stratification.
• Findings support a continuum model of AD and provide a foundation for developing stage-specific therapeutic strategies.