Analysis of Alzheimer's disease severity across brain regions by topological analysis of gene co-expression networks

Abstract

BackgroundAlzheimer's disease (AD) is a progressive neurodegenerative disorder involving variations in the transcriptome of many genes. AD does not affect all brain regions simultaneously. Identifying the differences among the affected regions may shed more light onto the disease progression. We developed a novel method involving the differential topology of gene coexpression networks to understand the association among affected regions and disease severity.MethodsWe analysed microarray data of four regions - entorhinal cortex (EC), hippocampus (HIP), posterior cingulate cortex (PCC) and middle temporal gyrus (MTG) from AD affected and normal subjects. A coexpression network was built for each region and the topological overlap between them was examined. Genes with zero topological overlap between two region-specific networks were used to characterise the differences between the two regions.Results and conclusionResults indicate that MTG shows early AD pathology compared to the other regions. We postulate that if the MTG gets affected later in the disease, post-mortem analyses of individuals with end-stage AD will show signs of early AD in the MTG, while the EC, HIP and PCC will have severe pathology. Such knowledge is useful for data collection in clinical studies where sample selection is a limiting factor as well as highlighting the underlying biology of disease progression.

Exploring Different Stages of Alzheimer's Disease through Topological Analysis of Differentially Expressed Genetic Networks

“…Our method shows better result because the existing method [20] focuses on local (differential connectivity which is the local difference between two networks calculated by the number of genes associated with a particular gene) as well as on global concept (between centrality: which calculate the change in the expression levels of central genes), but as given in [13] A.D does not affect all the brain regions at a time but there are differences in AD severity across regions. So it shows more local phenomena than the global one.…”

“…So the optimality of the result is solely dependent on this parameter, making the problem more parameter driven. Here we proposed an extended version of an existing method [13] and we have shown that it outperforms other existing method [20].…”

“…Other existing method [13] focuses only on unweighted T.O measure which takes into account only of some significant gene to gene interactions. So the overall gene dependency and gene significance in a particular network is not totally highlighted.…”

“…In brain it does not affect different regions simultaneously. AD progresses in stages and is described in terms of incipient (Braak stages III-IV), mild/moderate (Braak stages IV-V) and severe AD (Braak stages V-VI) [12].While some previous works are focused on genes that have differential topology in gene co expression networks corresponding to different brain regions and to observe the difference in AD severity across regions [13], our objective is to find genes that have differential topology in gene co expression networks corresponding to different stages of AD progress and to find out genes/pathways which are responsible for AD spreading.…”

“…Let for gene i in network 1 X no of interaction(s) is/are significant, and in network 2 it is Y. So as in [13] the T.O. of gene i between two networks can be defined as…”