Bayesian graphical network analyses reveal complex biological interactions specific to Alzheimer's disease

Area of Science:

• Neuroscience
• Biochemistry
• Computational Biology

Background:

• Alzheimer's disease (AD) biomarker discovery often focuses on limited protein sets, potentially overlooking complex biological interactions.
• Understanding the connectivity of blood-based proteins is crucial for developing accurate diagnostic and prognostic tools for AD.

Purpose of the Study:

• To investigate the biological connectivity between Alzheimer's disease-associated blood-based proteins using a novel Bayesian graphical network method.
• To assess how network similarity and protein group size influence the ability to distinguish between clinical classifications in aging.

Main Methods:

• Employed a Bayesian graphical network approach on data from the Australian Imaging, Biomarkers and Lifestyle (AIBL) study.
• Analyzed three distinct groups of AD-associated proteins (18, 37, and 48) for biological connections within and between clinical groups (hpHC, HC, MCI, AD).
• Calculated posterior probabilities of network similarity to quantify differences in biological connectivity across clinical classifications.

Main Results:

• Initial analysis with smaller protein groups showed high network similarity across all clinical classifications, indicating no significant differences.
• Increasing the number of proteins analyzed enhanced the ability to separate high-performing healthy controls (hpHC) and healthy controls (HC) from the AD group.
• Posterior probabilities of separation decreased from 0.89 (18 proteins) to 0.54 (37 proteins) and 0.28 (48 proteins) as protein group size increased.
• Identified beta-2 microglobulin (β2M) as a potential master regulator of multiple proteins across all clinical groups.

Conclusions:

• The Bayesian graphical network approach offers a powerful method for uncovering novel biological insights from complex protein interaction data in AD.
• Analyzing larger sets of correlated blood-based proteins can improve the differentiation between Alzheimer's disease patients and healthy individuals.
• Beta-2 microglobulin (β2M) emerges as a key protein potentially involved in regulating multiple pathways relevant to Alzheimer's disease progression.