The "omics" technologies are generating highly detailed molecular atlases for neurological diseases, including Alzheimer's disease (AD) and Epilepsy. However, a single omic technology analysis captures changes in only one component of the biological cascade. Integrating multiple omic modalities is a powerful tool for identifying complex diseases' molecular subtypes.
Integrated transcriptomic (~60,000 transcripts), proteomic (1,092 proteins), metabolomic, and lipidomic (627 metabolites) profiles of AD cases (n=462) and controls (n=139) from multiple cortical regions and three different cohorts are made. Following is the use of machine learning techniques, digital deconvolution, and traditional statistical analysis to integrate and analyze multiple omics (survival analysis, differential expression, cell type-specific effect, and cell proportions inference analysis).
A molecular profile associated significant cognitive impairment at the time of death, shorter survival, higher markers of neurodegeneration and astrogliosis, and reduced levels of metabolomic profiles is identified. This profile shows a significant dysregulation of synaptic genes (p=1.0×10-15) in multiple cortical regions. Subsequent AD staging, pathways, co-expression, and CSF survival analyses identified synaptic genes dysregulated at different stages of AD and associated with dementia progression.
The results demonstrate that molecular profiles of AD with clinical and biological relevance can be found by integrating multiple omic data.