DEVELOPMENT AND APPLICATION OF BRAIN TISSUE BASED MULTI-OMICS PROFILE SCORES FOR ALZHEIMER'S DISEASE.

BACKGROUND

Advances in omics technologies, such as epigenomics and metabolomics, provide novel insights into the biological mechanisms underlying Alzheimer's disease (AD). However, little is known how different omics layers interact and jointly relate to AD neuropathology.

METHODS

We performed a comprehensive single- and multi-omics analysis integrating genome-wide DNA methylation and high-resolution metabolomics data from 157 frontal cortex samples. We developed novel single and multi-omics profile scores (PS) for AD pathology, using a combination of machine learning, regression, and pathway analysis.

RESULTS

For the ABC score (Amyloid, Braak, CERAD) the PS of DNAm outperformed metabolomics-based PS (median R†: 0.11 vs. 0.04). Combining both omics layers with the best-performing multi-omics PS yielded a partial R† of 0.15 for the ABC score independent of age, sex, race and socioeconomic factors. DNAm-specific pathways highlighted redox balance, immune activation, synaptic signaling, and lipid biosynthesis, whereas metabolomics-specific pathways emphasized inflammatory, hormonal, lipid, and energy metabolism. Notably, both omics layers converged on lipid metabolism and signal transduction as shared biological systems implicated in AD neuropathology.

CONCLUSIONS

Despite limited gains in predictive accuracy, integrative pathway and network analyses of DNAm and metabolomics PS converged on lipid metabolism and signal transduction, underscoring shared biological mechanisms and the value of multi-omics approaches for biological insight rather than prediction alone.