Multiple Alzheimer's disease progression pathways inferred from transcriptome data of the dorsolateral prefrontal cortex

Late-onset Alzheimer's disease (AD) is a typical type of dementia for which therapeutic strategies have not yet been established. The heterogeneity in symptom by the variability of pathogenic factors may account for the difficulties in preventing and treating this disease. In this study, based on an analysis of publicly available data, the pathology of AD was suggested to be non-uniform, but there were several different typical forms of cognitive deterioration. First, a cluster analysis of subjects diagnosed with "No Cognitive Impairment (NCI)," "Mild Cognitive Impairment (MCI)," and "Alzheimer's Disease (AD)" stages from clinical trials was performed using gene expression data. NCI was found to contain at least two substages, and MCI and AD were found to contain four and six substages, respectively. An inference of adjacency networks among substages, evaluated via partition-based graph abstraction using the gene expression profiles of subjects, suggested multiple typical disease progression pathways from NCI through different MCI substages and then to different AD substages. These results will aid future research on the pathological features and mechanisms of AD and induce changes in the strategic bases of conventional AD prevention and treatment.