Researchers have developed a machine-learning framework that can identify candidates for drug repurposing in Alzheimer’s disease.
Alzheimer’s disease
Due to longer life expectancy, Alzheimer’s disease (AD) is becoming an increasingly growing healthcare crisis. Without effective preventative and treatment options, estimates indicate that disease prevalence will more than double over the next several decades. In addition to its direct impact on human health, AD imposes a substantial economic burden.
Most efforts to develop a disease-modifying therapy have largely been negative, with many of the failures due to lack of efficacy or excess toxicity. The failure of new molecular entities (NMEs) in clinical trials consumes substantial time and resources. As a result, a lot of attention has shifted to repurposing drugs already approved by the FDA. This approach is less expensive, involves already defined possible toxicities and can have higher success rates.
Traditional repurposing approaches involve using an existing drug in a new indication. An alternative approach involves testing a therapeutic concept that could then be advanced with additional chemistry and functional testing to become an NME. This option is valuable in the case of AD where the underlying disease mechanisms remain poorly understood and where there are potentially multiple drivers.
DRIAD
In this paper, published in Nature Communications, researchers presented DRIAD – Drug Repurposing In AD – a machine-learning framework that quantifies potential associations between the stage of AD and any biological process or response that can be characterised by a list of gene names. These names were obtained by using RNAseq to measure the responses of human neuronal cells to small molecular drugs and then by identifying differentially expressed genes. This generated drug associated gene lists.
Researchers applied DRIAD to lists of genes that arise from perturbations in human neural cell cultures by 80 FDA-approved drugs. These compounds were predominately kinase inhibitors with anti-cancer activity as they are the largest class of targeted drugs currently available. This produced a ranked list of possible repurposing candidates. Those that were top scoring were inspected for common trends among their targets. Among the top performers were several drugs whose primary targets are JAK kinases.
This framework allows for unbiased assessment of biological processes or drug candidates even when disease mechanisms are not explicitly known. The team propose that this method could be used to nominate drugs that, after additional validation and identification of relevant pharmacodynamic biomarkers, could be readily evaluated in a clinical trial.
Image credit: Robina Weermeijer – Unsplash
