Investigating the molecular consequences of co‐morbidites in Alzheimer’s disease

Abstract

AbstractBackgroundDue to the transient and highly sensitive nature of DNA methylation, it is important that Epigenome‐Wide Association Studies (EWAS) account for numerous variables that may influence the methylome as it is becoming increasingly clear that other diseases can influence the onset and severity of Alzheimer’s disease (AD).MethodDNA was extracted from both AD (n = 119) and cognitively normal (n = 76) post‐mortem middle‐temporal gyrus brain tissue and DNA methylation quantified using Illumina methylation arrays. Combining the methylation beta values with detailed phenotypic data collected for each patient, including data on diabetes mellitus and TDP.43 proteinopathy, Principal Component Analysis was performed to determine co‐morbidities of AD that had a confounding effect on DNA methylation across the genome. Linear regression models were then designed to compare the methylomic consequences of TDP.43 proteinopathy in AD when co‐morbidities, such as DLB, are corrected for. Additionally, Analysis of Variance was used to conduct an EWAS evidencing the additional methylomic consequences of diabetes on AD‐diagnosed patients.ResultsAn EWAS of AD and diabetes mellitus showed significant differential methylation at several sites across the genome, in particular within the gene BACE2, suppression of which promotes β‐cell survival. Further linear regression analysis of methylation data that had been corrected for co‐morbidities, including DLB, shows that much of the methylation changes seen in EWAS may be accounted for by these confounding diseases, while changes in genes associated with the pathology of interest retain their significance. In the study of TDP.43 proteinopathy and AD this can be observed at probes in the genes HAX1, CUX1, and CCDC88C.ConclusionThe next step for this project is to obtain more samples with clinical information on common co‐morbidities of AD. EWAS focussed on the effects of TDP.43 proteinopathy and diabetes mellitus can then be performed with greater power and the specific pathways identified. Nonetheless, this pilot data emphasises the need for comprehensive clinical data when performing studies on methylomic data.