Plasma proteomics signatures in Alzheimer's Disease
File(s)
Author(s)
Hu, Manyue
Type
Thesis or dissertation
Abstract
Background: Alzheimer’s disease (AD) is the most common neurodegenerative disorder worldwide, yet early diagnosis and intervention remain major challenges. Plasma proteomics offers a minimally invasive approach for capturing molecular signatures associated with brain pathology and disease progression. However, the complex and heterogeneous nature of AD limits the utility of individual biomarkers. This thesis integrates large-scale proteomic datasets with machine learning and network-based approaches to investigate AD-related molecular mechanisms from a systems biology perspective.
Methods: Three independent cohorts (ADNI, ACE, and CHARIOT PRO) were analysed using Elastic Net modelling, protein–protein interaction networks, and network centrality measures. Plasma proteins were used to classify diagnostic and cognitive decline groups, while organism-level protein systems were examined for associations with APOE4 status, amyloid positivity, and cognitive trajectories. Candidate proteins were prioritised based on both model importance and network relevance.
Results: In ADNI and ACE, an 11-analyte plasma signature achieved 95% accuracy in distinguishing AD from cognitively normal individuals. A separate 17-analyte model showed substantially lower performance for classifying stable and declining mild cognitive impairment groups, highlighting the biological heterogeneity of prodromal disease stages. In CHARIOT PRO, immune and nervous system protein panels demonstrated the strongest associations with APOE4 carriage and amyloid positivity. Longitudinal analyses further identified digestive and nervous system proteins as significant contributors to cognitive decline. Across multiple analyses, glutamate decarboxylase 1 (GAD1) emerged as a recurring key protein, suggesting potential involvement of GABAergic and metabolic pathways in AD pathophysiology.
Conclusions: This thesis establishes an integrated organism-level proteomic framework for investigating AD-related biological processes. The findings demonstrate that plasma proteomics can capture systemic molecular signatures associated with genetic risk, amyloid pathology, and cognitive decline, supporting its potential for early risk stratification. Future multi-omics studies and independent cohort validation may further clarify causal mechanisms and facilitate the development of personalised interventions for AD.
Methods: Three independent cohorts (ADNI, ACE, and CHARIOT PRO) were analysed using Elastic Net modelling, protein–protein interaction networks, and network centrality measures. Plasma proteins were used to classify diagnostic and cognitive decline groups, while organism-level protein systems were examined for associations with APOE4 status, amyloid positivity, and cognitive trajectories. Candidate proteins were prioritised based on both model importance and network relevance.
Results: In ADNI and ACE, an 11-analyte plasma signature achieved 95% accuracy in distinguishing AD from cognitively normal individuals. A separate 17-analyte model showed substantially lower performance for classifying stable and declining mild cognitive impairment groups, highlighting the biological heterogeneity of prodromal disease stages. In CHARIOT PRO, immune and nervous system protein panels demonstrated the strongest associations with APOE4 carriage and amyloid positivity. Longitudinal analyses further identified digestive and nervous system proteins as significant contributors to cognitive decline. Across multiple analyses, glutamate decarboxylase 1 (GAD1) emerged as a recurring key protein, suggesting potential involvement of GABAergic and metabolic pathways in AD pathophysiology.
Conclusions: This thesis establishes an integrated organism-level proteomic framework for investigating AD-related biological processes. The findings demonstrate that plasma proteomics can capture systemic molecular signatures associated with genetic risk, amyloid pathology, and cognitive decline, supporting its potential for early risk stratification. Future multi-omics studies and independent cohort validation may further clarify causal mechanisms and facilitate the development of personalised interventions for AD.
Version
Open Access
Date Issued
2025-10-31
Date Awarded
2026-07-01
Copyright Statement
Attribution-NonCommercial 4.0 International Licence (CC BY-NC)
License URL
Advisor
Middleton, Lefkos
Robinson, Oliver
Baker, Susan
Lill, Christina
Matton, Anna
Publisher Department
School of Public Health
Publisher Institution
Imperial College London
Qualification Level
Doctoral
Qualification Name
Doctor of Philosophy (PhD)