Skip to main content
Premium Trial:

Request an Annual Quote

Alzheimer's Disease Genes, Treatment Targets Identified With Deep Learning Method

NEW YORK – A team led by researchers at the Cleveland Clinic and Case Western Reserve University has devised a computational strategy for following up on genetic links found through genome-wide association studies, applying the approach to Alzheimer's disease to track down potential drug interventions and treatment targets.

With a deep learning framework dubbed the "network topology-based deep learning framework to identify disease-associated genes" (NETTAG) that combines multiomic data sources and human protein interaction networks, the researchers tracked down hundreds of new and known Alzheimer's disease-related genes — work they outlined in a paper appearing in Cell Reports on Tuesday.

For a set of 156 genes prioritized from their network-based prediction analyses, the team brought in electronic health record-based case-control data for millions of individuals to unearth possible drug targets, along with several drugs that could potentially be repurposed for the neurodegenerative condition.

"We showed that predicted genes are enriched with drug targets, differentially expressed in [disease-associated microglia] and astrocytes, and most importantly significantly associated with AD," senior author Feixiong Cheng, a researcher affiliated with the Cleveland Clinic and Case Western Reserve University School of Medicine, and colleagues wrote.

Although prior GWAS, genome sequencing analyses, brain sample-centered transcriptomic profiles, and other studies have demonstrated the potential for finding molecular contributors to Alzheimer's disease, the researchers noted that a functional understanding of the disease remains incomplete, despite past attempts to bring in large datasets.

"Several network-based analytic techniques have recently been developed to address the myriad different types of inputs of omics layers," the authors wrote, noting that "existing network-based approaches do not leverage the integration of multiomics profiles, such as genetics, functional genomics, transcriptomics, and proteomics, for risk gene prediction and drug target identification."

Using insights from past GWAS on Alzheimer's disease, cerebral amyloid deposits, and other traits or conditions, the team mapped SNPs across the human genome. They incorporated published data on regulatory elements, ranging from expression quantitative trait locus (QTL) and histone QTL patterns to open chromatin profiles, enhancer or promoter sequences, and transcription factor binding features.

From there, the researchers put together a human protein-protein interactome network, turning to NETTAG to find informative clusters, topological structures, and features that distinguished Alzheimer's disease risk genes, pathways, and regulators.

"The fundamental premise of NETTAG is that disease risk genes exhibit distinct functional characteristics compared with non-risk genes and therefore can be distinguished by their aggregated genomic features, converge to a limited number of pathobiological pathways captured by the human protein-protein interactome, and include multiple AD pathobiological modulators and potential therapeutic targets," they explained.

The deep learning method led the team to hundreds of apparent Alzheimer's disease risk genes, including 156 "NETTAG-inferred Alzheimer's disease risk genes," or alzRGs, prioritized with help from protein-protein interaction and differential gene expression data, along with case-control EHR data.

With gene target clues and insights from these health records, the researchers highlighted drugs with apparent ties to reduced Alzheimer's disease risk, including the anti-inflammatory drug ibuprofen, cholecalciferol or vitamin D3, an antibiotic called ceftriaxone, and the lipid-regulating drug gemfibrozil.

Analyzing nearly 3.2 million EHRs compiled in the Northwestern Medicine Enterprise Data Warehouse over a decade, from 2011 to 2021, the team saw lower-than-usual Alzheimer's disease documentation in patients taking these drugs, even after adjusting for factors such as age, sex, or history of other conditions.

In an active-comparator analysis, for example, the team's results suggested that Alzheimer's risk was roughly 43 percent lower in patients taking gemfibrozil compared to individuals taking simvastatin, another lipid drug.

"We believe that the NETTAG presented here, if broadly applied, could significantly catalyze innovation in drug discovery for AD and other neurodegenerative diseases," the authors concluded.