AbstractBackgroundThe clinical and molecular heterogeneity of Alzheimer disease (AD) is increasingly recognized as a major factor impacting diagnosis accuracy, the design of therapeutic interventions and clinical trials, and the development of effective treatments for AD. Recently, research efforts have leveraged molecular data to characterize AD patients into subgroups. However, the influence that such AD molecular profiles may have on cognitive decline trajectories remains poorly understood. Comprehensive studies that aggregate multiple modalities of omics have proven to be a powerful way to identify molecular profiles for human diseases including cancer. Here, we sought to determine whether cross‐omics integration can reveal molecular profiles of AD that have clinical and biological relevance.MethodWe leveraged machine learning approaches, digital deconvolution, and traditional statistical approaches to integrate and analyze multiple high‐throughput omics data from different cortical regions and cohorts, including the parietal cortex (N = 278) from the Knight Alzheimer Disease Research Center (Knight ADRC), the dorsolateral prefrontal cortex (N = 237) from the Religious Orders Study and Memory and Aging Project (ROSMAP) cohort, and the parahippocampal gyrus (BM36; N = 116) from The Mount Sinai Brain Bank (MSBB) study.ResultWe identified four distinct molecular profiles, one of which was associated with significantly worse Clinical Dementia Rating (CDR) at death, shorter survival after symptom onset, more severe neurodegeneration and astrogliosis, and unique metabolomic profiles. This profile, present in multiple cortical regions affected at distinct stages of AD progression, showed significant dysregulation of synaptic genes and pathways, indicating neuron/synapse losses and dysfunction at later stages of AD and associated with worse cognitive function. Among other molecules, we found that the expression of alpha‐synuclein (SNCA) and SNAP25 were downregulated in this profile. The synaptic dysregulation associated with this profile was further validated in a mouse model of pathological aSyn aggregation.ConclusionTogether, our results demonstrate that cross‐omics approaches can reveal multiple molecular pathways associated with AD progression, and reinforce that, in AD, prominent synaptic loss and dysfunction are closely associated with worse cognition. This may open the possibility for new biomarkers for the molecular staging of AD and cognition, as well as potential therapeutic targets to ameliorate cognitive decline and AD progression.