Premium
Network analysis to identify proteomic markers for brain aging and dementia in healthy older adults
Author(s) -
Short Meghan I.,
Zare Habil,
Beiser Alexa S.,
Larson Martin G.,
Vasan Ramachandran S.,
Seshadri Sudha
Publication year - 2020
Publication title -
alzheimer's and dementia
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 6.713
H-Index - 118
eISSN - 1552-5279
pISSN - 1552-5260
DOI - 10.1002/alz.037711
Subject(s) - dementia , proteome , framingham heart study , cohort , framingham risk score , disease , oncology , medicine , bioinformatics , endophenotype , biology , psychology , neuroscience , cognition
Background Biological processes involved in preclinical stages of Alzheimer’s Disease and Related Dementias (ADRD) are not well understood. The plasma proteome is a promising tool to identify pathways involved in ADRD development, linking genetic and environmental risk factors with disease outcomes. The goal of this study is to leverage proteome data in a community‐based cohort to identify proteins and pathways involved in early stages of dementia. This is the first study to employ a network analysis approach, which can detect subtle but consistent variations in protein expression, to identify biological pathways associated with ADRD and endophenotypes from the plasma proteome. Method A panel of 1305 proteins was measured in plasma from 1861 participants from the Framingham Heart Study Offspring cohort (mean age 55, 54% women). Weighted Co‐expression Network Analysis (WGCNA) was used to identify groups (“modules”) of co‐expressed proteins. Summary measures of proteins in the modules (“eigenproteins”) were related to total brain volume (n = 1038), hippocampal volume (n = 1038), and white matter hyperintensity (n = 1022) using linear regression, and incident dementia (128 events, 1740 at risk) and Alzheimer’s Disease (AD; 94 events, 1740 at risk) over 20 years of follow‐up using Cox proportional hazards regression. Result Network analysis resulted in 4 modules containing 42, 67, 165, and 272 proteins, respectively. Two modules were associated with total brain volume, one directly (0.17% per standard deviation [SD] increase in eigenprotein; p = 0.005) and one inversely (‐0.26% per SD increase in eigenprotein; p = 1.3 × 10 −5 ). Of these, the first was associated with incident dementia (HR [95% CI]: 0.82 [0.68‐0.99]; p = 0.04), and the second was associated with incident dementia (HR [95% CI]: 1.22 [1.04‐1.44]; p = 0.01), and AD (HR [95% CI]: 1.26 [1.04‐1.52]; p = 0.02). Associations with brain volume remained significant after multiple testing corrections; associations with incident AD and dementia did not. KEGG pathways for axon guidance and cytokine‐cytokine receptor interaction were the most enriched from the two modules, respectively. Conclusion Network analysis identified two groups of proteins associated with a brain MRI endophenotype and suggestive of association with incident dementia. These results identify possible biological pathways implicated in early stages of dementia, which we will validate in other cohort studies.