Connect

Repurposing drugs for rare diseases, cancer and beyond

2024年1月30日

Ann-Marie Roche 別

SDI Productions/E+ via Getty Images

Elsevier is applying AI and data science to reinvent and accelerate the process of drug repurposing. Our team talks about how they support R&D professionals in pharma, oncology and industry.

It’s an enticing concept: recycling old drugs for new cures. It’s also viable: aspirin, Viagra and the new weight loss drug Wegovy are all repurposed drugs. And today, thanks to innovations in AI, we are accelerating the process — call it digital speed-dating that links a disease with its potential treatment.

But what does using precision medicine for crop protection have to do with it?

We talked to colleagues on Elsevier’s Professional Services team to find out.

“Swiss Army” medicine

When we asked Elsevier’s VP of Data Science Mark Sheehan(新しいタブ／ウィンドウで開く) what non-AI related innovation is exciting him the most lately, his eyes lit up: “Novo Nordisk(新しいタブ／ウィンドウで開く)’s weight loss drug Wegovy(新しいタブ／ウィンドウで開く). It’s amazing how it’s tapped into a massive new market — a rarity in the Life Sciences. You have to be impressed, especially when it’s answering a specific global crisis: obesity.”

Wegovy is also interesting because it began as a treatment for diabetes. Now it’s been repurposed — like aspirin, Viagra, thalidomide and countless other drugs before it. It’s also interesting because drug repurposing — aka drug repositioning or reprofiling — can help meet the global demand for unmet medical needs.

By reducing timelines and costs compared to creating drugs from scratch, repurposing also taps into a deeply human emotion: the love for a bargain. Why spend millions on R&D if a solution already exists?

And while repurposing enjoys a long and rich history, Elsevier is now applying AI and data science to reinvent and accelerate the process, with the Professional Services team at the forefront of this quest. In fact, the related work has expanded beyond healthcare to also take in agriculture and confronting climate change.

Solving non-standard puzzles

“Our team covers the non-standard — anything that goes beyond licensing databases like Reaxys or Embase or journal subscriptions,” says Dr Frederik van den Broek(新しいタブ／ウィンドウで開く), Senior Director of Professional Services. “This can include integrating our data into customer systems, or customer data into our systems. We also do custom projects, which can be a write-up, a scientific analysis or even creating machine learning models for chemistry or other applications.”

In the case of drug repurposing, the team has a wide array of digital tools to mine the scientific literature for molecular connection that can help predict the ability of a particular drug to target a particular disease.

“Our job is to solve puzzles. And while not being actual scientists anymore, we are still close to R&D, which in the end has the job of curing patients,” says Frederik. Meanwhile much of the team’s work remains shrouded in non-disclosure agreements and the IP considerations of clients.

Luckily, there is still more than enough to talk about.

Taming reams of knowledge into actionable insights

With a bachelor's degree in physics, a master's in organic chemistry and a PhD in molecular biology, Dr Anton Yuryev(新しいタブ／ウィンドウで開く), Consulting Director at Professional Services, is already a well-rounded individual. And he’s also a well-published data scientist with a practiced sense of humor. “Numbers just add up: The fact I’ve written over 50 scientific articles only means I am old,” he quips.

Anton was also one of the co-founders of Ariadne Genomics, which Elsevier acquired in 2011(新しいタブ／ウィンドウで開く) and which provided the tech backbone to current offering Biology Knowledge Graph(新しいタブ／ウィンドウで開く), which mines data to visualize disease biology.

“Twenty-plus years later, I am still driven by the same passion,” he says. “It’s a passion we all have here in Professional Services. It comes down to Elsevier having a huge mass of data, and many different tools to access this data. And my primary job is actually to help customers navigate this maze of products, databases, interfaces, and so on — taking this immense amount of data and knowledge, focusing it, and then applying it to cure patients.

“Look at the case of rare diseases: the research is scattered across many different journals — most of which are highly specialized and not in open access. Elsevier publishes about 25% of all the articles on rare diseases in our knowledge graph, which also contains journals from other major scientific publishers such as Wiley and Springer. As a result, we have more than 80% of all articles and data on rare diseases that’s not in open access. Now that’s a lot of data to sift through — but you also know it’s of the highest quality due to data extraction by Elsevier AI technology.”

Secrets in innovation

Dr Maria Shkrob(新しいタブ／ウィンドウで開く), a biologist-turned-bio-informaticist, is a consultant at Professional Services and a former employee at Ariadne Genomics. She switched to Professional Services when she saw how much Anton was enjoying himself. “It struck me how you get to work with customers and really build something,” she says. “Also, the diversity of projects is immense: you never know what’s coming next. And as Frederik says, it is really like solving a puzzle.”

“But it's not just a mental game. The work can potentially impact the development of new drugs and therapies. So that's exciting — and hopefully useful. Of course, because of IP issues, you are never sure because our customers don't usually come back to say how they ended up using the data or whether it sparked some big decisions. But you always want to believe your work has an impact,” she says and smiles.

Drug repurposing and Elsevier

There’s another reason why her work feels somewhat secretive: Elsevier’s ruling reputation as a publisher. “Whenever I am presenting at conferences, people are surprised I work at Elsevier — they still assume I’m dealing with journals and books. Most don’t know we also directly help pharma and biotech companies to design new drugs and to make them safer and better.”

Read ‘4 drivers of success in drug repurposing for rare diseases’

Repurposing drugs for rare diseases is nothing new. A recent article written by Anton marked the 40^th anniversary of the Orphan Drug Act(新しいタブ／ウィンドウで開く), which helped jumpstart an industry around repurposing. Meanwhile, Elsevier’s own Year of the Zebra(新しいタブ／ウィンドウで開く) initiative puts the spotlight on the 300+ million people worldwide with rare diseases, half of them children. Meanwhile, only 5%(新しいタブ／ウィンドウで開く) of 8,000 known rare diseases have licensed treatments.

So there’s still a lot of work to be done.

Anton’s article outlines what is needed to chase down treatments for untreatable diseases. This includes having diverse cross-functional teams, access to FAIR data(新しいタブ／ウィンドウで開く) sources, staying on top of the latest computational research methods, and anticipating regulatory challenges — while always putting patients at the center of your considerations.

The article also covers recent success stories from Elsevier in terms of uncovering drug repurposing candidates, such as with Diffuse Intrinsic Pontine Glioma (DIPG), a rare childhood brain cancer. “With DIPG, we wrote a paper(新しいタブ／ウィンドウで開く) about how we built a model of the disease aligning the genomic data of biopsies from brain tumors with the literature of this particular disease,” Anton explains. “Basically, we showed our algorithm could outperform a board of specialist scientists when it comes to creating a list of promising drug candidates. In fact, we could also rank this list and find more interesting candidates.”

And this was largely thanks to the ability to loop in personalized patient data, Anton says.

Drug repurposing. Or is it personalized medicine?

The future of healthcare is clearly personalized — or “precision” — medicine by the simple fact that not all medicines work for all people. Take this one sobering statistic: it’s estimated(新しいタブ／ウィンドウで開く) that the 10 top-selling drugs in the United States help just one in four patients. Meanwhile, almost 20 million people worldwide are diagnosed with cancer(新しいタブ／ウィンドウで開く) every year.

So how can you scale precision medicine for the masses? In essence, you follow the same processes as when you are trying to repurpose medicine: connecting a disease’s actions with a person’s molecular makeup and genomic profile.

“Drug repurposing and personalized medicine are really two sides of the same coin,” says Frederik. “That’s why, we are now working more directly with doctors — and getting closer to the patients themselves.”

“Every time you analyze, for example, a cancer patient’s genomic data and try to understand the particular mechanism of their cancer, you are essentially doing drug repurposing,” adds Anton. “You are trying to find the best drug for their particular disease make-up. And this has really become the most exciting part of my job: Doctors are using our suggestions for patients with stage 4 cancer when there are no longer any other options. And we’ve had some real success stories(新しいタブ／ウィンドウで開く).”

Wake Forest: in search of precision medicine at scale

Anton cites one particularly satisfying Elsevier collaboration with the oncology practice at Wake Forest University(新しいタブ／ウィンドウで開く) in North Carolina. The idea was to help doctors connect individual late-stage patients with the best choice among an ever-increasing number of anti-cancer treatments.

The project involved connecting the attending doctor’s own knowledge and experience, with various Elsevier digital tools to mine the relevant research — while also looping in as much previously unstructured patient data as possible.

The results were positive in terms of validating precision medicine. All stage 4 cancer patients treated by a drug suggested by Elsevier outlived overall standard of care survival times. But it was not yet a complete approach that can yet be scaled for daily use for millions.

“There’s still a lot of work to be done,” Anton says. “We need to know more about the mechanism of action of all the drugs out there. More efforts have to be made in understanding and managing all the additional unstructured data that may have an influence. But I really believe we are at a tipping point where rapid acceleration is just around the corner. Look at Every Cure.”

Chasing 9,000 cures

The Professional Services team is particularly excited about their work with nonprofit Every Cure(新しいタブ／ウィンドウで開く). “They have a very broad focus: finding a cure for as many rare diseases as possible,” Frederik says.

Every Cure certainly comes with a compelling back story. Co-founder Dr David Fajgenbaum(新しいタブ／ウィンドウで開く) would most likely be dead from Castleman disease if he didn’t track down a generic drug that put him in remission — for 10 years and counting. Now with Every Cure, he’s out to find cures for the remaining 9,000-plus incurable rare diseases by linking them with existing off-the-shelf drugs.

While Elsevier is donating both data and algorithms, the project is financially supported by high-profile partners such as the Clinton Foundation(新しいタブ／ウィンドウで開く) and the Chan Zuckerberg Initiative(新しいタブ／ウィンドウで開く). “And here’s a fun fact,” says Anton: “Elsevier's algorithm found the same drug that the founder found more ‘manually.’”

Every Cure founder David Fajgenbaum, MD, is Associate Professor of Medicine at the UPenn School of Medicine.

The project has once again allowed the team to work their magic across several Elsevier products. “We’re working to change this, but our data is now largely in silos,” Anton says. “For instance, Biology Knowledge Graph is extracted from the biomedical literature. It contains all kinds of relationships between all kinds of biological entities, such as between diseases and proteins, drugs and diseases, drugs and proteins. Meanwhile, information about drug-target affinity and drug biological effects is in Reaxys and PharmaPendium. Therefore, we’re automating the workflow so we can borrow data from these knowledge sources as well.

“In the end, we are able to produce a list of drug candidates along with links to the supporting research. More importantly, we are able to score them from the most hopeful to the least to help Every Cure to prioritize their work.”

An essential part of the process remains applying domain knowledge to evaluate if a potential target-drug can potentially work — and looping these observations back into the algorithm to make it stronger. “Take a compound like curcumin: This is the active compound in the Indian spice turmeric, and it gets the best score for essentially every disease,” Anton explains. “And yes, it’s a remarkably promiscuous compound: it bonds with at least 50 different proteins — and a lot of these are related to inflammation, which is involved in about half of all diseases. But it’s just not bioavailable enough: you would maybe need to intake a kilogram orally to have an effect.”

Going to the next level

Meanwhile, the idea of repurposing is going very much beyond healthcare. For example, his team collaborated with a company that’s working to genetically modify bacteria to capture carbon before it enters the atmosphere at steel plants. “We gathered all the information around a particular bacterium and made the company a custom knowledge graph linking all the known information,” Anton says. “While we can only hope we are setting them on the right track, they still do come across as happy customers.”

Based on a proof-of-concept project for a large agrochemical firm, the team was able to publish the paper: Finding novel lead compounds in pesticide discovery inspired by pharmaceutical research(新しいタブ／ウィンドウで開く). “Let’s say you are out to target a specific fungal pest. In a way you are trying to induce an adverse drug event,” Frederik explains. “So this is very much precision medicine — but then as a poison. You are still trying to understand what a certain molecule does once inside the cell of an organism. And here we chose drugs that also work against the malaria parasite to see if they affected a particular fungal pest. When Maria presented our results, one of the company’s scientists showed a lot of interest — but we don’t know if it ended up as a new pesticide treatment. “But the approach makes sense: Why take the broad-spectrum glyphosate approach when you can go precision?”

Upward and onward!

The team believes their projects will only become more diverse as awareness continues to spread about Elsevier’s singular offering. “We’re doing more than just journals and books,” says Frederik. “We have the technology to turn this knowledge into actionable insights.”

Maria says: “We have this ideal combination of accessible content, along with the chemistry and biology subject knowledge — and the data expertise to know what’s possible.”

“And these abilities fit into our larger mission,” Anton adds. “It’s not just helping pharma with their R&D and getting their drugs to market quicker. Our most important mission is for society in general: helping patients improve their lives and longevity. Everything we do, we do for people.”

Learn more about drug repurposing approaches and methods

貢献者