NEW YORK – A genomic analysis tool often used to find new ways to treat cancer is finding new use in the search for more COVID-19 treatments.
Gene set enrichment analysis, or GSEA, is a computational method of identifying the genes most likely to functionally associate with a given cellular state, such as disease, developmental stage, or drug response.
By working with sets instead of individual genes, GSEA provides a more robust picture of how cells respond to various conditions. This network-level approach lends itself easily to drug repositioning studies, in which scientists seek new uses for currently approved medicines.
The global COVID-19 pandemic has triggered an uptick in such studies. Their increase, measured by the number of papers per year matching "drug repositioning" in the PubMed database, roughly doubled between 2020 and this year, compared to the annual change from 2017 to 2020.
While an exact count of the number of these studies relying upon GSEA and the tools that it has enabled was less clear, these methodologies certainly underlie many of them.
The underlying logic of GSEA is that differences between cells, whether different types of cells or cells in different states of disease and health, can be described by unique patterns of gene expression.
The most differentially expressed genes — those most and least expressed — are the ones most likely to play a significant role in maintaining a given cellular state. It follows from this that by reversing that gene "signature," such as one related to cancer, could push those cells back into a healthy state.
This methodology plays a significant role in drug repositioning because it shifts the focus from the singular molecular targets of a given drug to the network effect that a drug can have based on its mechanism of action, or MoA.
Gideon Bosker, the CEO and cofounder of DarwinHealth, a biotechnology company focused on drug repositioning, describes these as "field effects," acting upon diverse cellular targets.
"A systems approach to deconvoluting the field effect — what previously might have been called off target effects — may be a much more effective way of determining a drug's final application to a particular therapy or biological question," Bosker stated.
"Repurposing drugs is just a way of understanding that drugs by definition have not been preordained to only do one thing," he added.
DarwinHealth recently adapted its signature DarwinOncoTreat tool, which matches FDA-approved drugs to druggable cancer-related targets, to finding a slate of potential new COVID-19 therapies.
DarwinOncoTreat builds upon GSEA analysis by inferring the activity of "master regulators" — key proteins that control genetic programs underlying transcriptional cellular states — from the differential enrichment of their transcriptional targets in disease-relevant gene expression profiles.
The company's VIPER algorithm then prioritizes drugs whose MoA can abolish the disease-causing regulatory programs that these master regulators control.
DarwinHealth has applied this approach to various tumor subtypes and its proprietary technology is currently being deployed in seven clinical trials evaluating repositioned drugs across a spectrum of cancers.
In a preprint of a study currently under review, scientists from DarwinHealth, along with academic collaborators at Columbia University, used an adapted version of this tool, dubbed ViroTreat, to identify drugs potentially capable of treating COVID-19.
Their analysis showed a set of differentially regulated genes, some related to cellular regulatory programs involved in promoting viral replication and others in activating host immune response, along with a list of possible medications for targeting these genes, either to impede viral replication or help boost the immune response.
Encouragingly, several of the drugs they identified in this unbiased manner have also emerged as potential antiviral medications in separate studies.
DarwinHealth plans to follow up this proof-of-concept study with validation studies in physiologic models.
Other GSEA-based studies have applied similar strategies to the task of finding COVID-19 treatments.
In one, researchers used GSEA to identify upstream regulatory elements of proteases known to facilitate the entry of SARS-CoV-2 — the virus causing COVID-19 — into host cells. They discovered that estradiol and retinoic acid, two FDA-approved medications, could potentially help treat COVID-19 by modulating the activity of several regulators of those virus-associated proteases. Importantly, this finding agrees with at least two other studies that noted potential therapeutic effects of estradiol and retinoic acid in treating COVID-19 patients.
Another study, also employing GSEA in combination with other network biology tools such as weighted correlation network analysis, identified another set of potential COVID-19 therapies, including the TNF-alpha inhibitor etanercept and the GABABR agonist baclofen.
Importantly, real-world data from case studies support these predictions.
In one study that employed the basic idea of a drug's network effect without computational methods — essentially performing a proof of concept of the more unbiased GSEA-based approach to drug repositioning — a Chinese group predicted that etanercept might treat COVID-19 because of the similarity between inflammatory cytokine activities in toxic epidermal necrolysis and COVID-19.
It is worth noting that despite advances being made in drug repositioning efforts, laws and regulations surrounding repurposed medicines need to adapt as well, in order for patients to fully reap the benefits of this research.
Rules governing intellectual property, for instance, do not currently offer provisions for IP protection of drug discovery through repositioning, and protection for such drugs remains limited. Similarly, insurance companies may not reimburse for drugs prescribed off-label.
Andrea Califano, cofounder of DarwinHealth and professor of systems biology at Columbia University, referenced this dilemma in relation to drug repositioning studies involving carfilzomib, used in treating multiple myeloma.
"We predict very consistently carfilzomib for ovarian cancer," Califano said. "The drug costs $120,000 a year, so unless insurance reimburses it, it's not gonna happen."