Abstract
BACKGROUND: Tau neurofibrillary pathology is a hallmark of Alzheimer's disease (AD) and can be quantified in vivo using tau-selective positron emission tomography (tau PET). Tau PET signal closely correlates with cognitive decline and disease stage, yet the molecular networks underpinning tau accumulation remain incompletely defined. METHODS: We performed multi-omics integration of proteomics, transcriptomics, and tau PET standardized uptake value ratios (SUVRs), and clinical assessments data from cognitively normal and cognitively impaired individuals. Using Light Gradient Boosting Machine (LightGBM), two-way orthogonal partial least squares, and network-based approaches, we explored key tau-associated proteomic signatures and constructed protein-protein interaction (PPI) modules. Module activities were quantified by gene set variation analysis and related to tau PET and cognition. RESULTS: Among 60 regions, 15 tau PET imaging biomarkers were selected based on group differences, LightGBM importance, and cognitive relevance. Fifty key tau-associated proteins were identified and organized into four functional modules. PPI modules 1 (metabolic-cytoskeletal) and 3 (adhesion-nutrient sensing) exhibited strong associations with elevated tau PET uptake across selected cortical and limbic regions, as well as with cognitive impairment. CONCLUSION: Distinct modules reflected regional tau PET burden and cognitive outcomes in AD, highlighting convergent disruptions in energy metabolism, cytoskeletal stability, and intercellular signaling. HIGHLIGHTS: Integration of proteomics, transcriptomics, tau positron emission tomography (PET) imaging, and cognition in Alzheimer's disease.Fifteen key tau PET imaging biomarkers were prioritized.Fifty key tau-associated proteins were identified.Four distinct molecular networks contribute to regional tau pathology and cognition.Modules 1 (metabolic-cytoskeletal) and 3 (adhesion-nutrient sensing) strongly associated with tau PET burden and cognitive impairment.