Bioinformatics investigations on multi-omics datasets of neurodegeneration

Abstract

Alzheimer’s Disease (AD) has become a rapid global health concern, due to its high associated expenses, absence of efficacious treatments and growing prevalence in aging societies. AD diagnosis remains extremely challenging due to its insidious nature of progression. Conversely, the efficacy of treatment strategies is heavily constrained by the extent of AD progression, highlighting the importance of early AD prediction. Efforts to stratify patients in the prodromal stage of AD, otherwise known as patients with Mild Cognitive Impairment (MCI), using neuropsychological presentations for early AD prediction remains inadequate in capturing the full extent of heterogeneity present in MCI. In this study, we demonstrate that the heterogeneous MCI cohort could be stratified into meaningful subclusters using biomarkers from blood transcriptomics and structural MRI imaging (sMRI) data. Our study also reveals an increased likelihood of AD conversion as the panel of biomarkers exhibits an increased correspondence to AD. Through multi-omics factor analysis, we also identified a key latent factor (LF) that is strongly correlated to AD conversion status which further revealed a prioritized sequence of sMRI features critical for the prediction of AD onset. Overall, our study has discovered a novel approach in predicting AD onset from a previously ambiguous MCI cohort.