By
·
3 minute read
Medicinal chemists don’t need to be convinced that there’s value in drug repurposing.
After all, they are acutely aware of the time, money, and effort it takes to bring a truly novel and effective drug to market. The prospect of spending as long as 17 years and more than $2 billion on a project, knowing the success rate for new drugs is around 10%, is daunting under the best of circumstances.
So, it comes as no surprise to anyone in the biopharma industry that effective drug repurposing can result in a very substantial reduction in cost, time to clinic, and failure risk. However, historically, luck and keen clinical observation have played a major role in identifying successful drug retargeting candidates. Many successful opportunities for drug repurposing were first discovered serendipitously through unexpected therapeutic effects observed during clinical use.
Drug Repurposing: Past Success Stories and Current Roadblocks
One of our favorite stories of drug repurposing is the story of the discovery of the antithrombotic effects of aspirin. The idea that daily, low doses of aspirin could prevent cardiovascular incidents was first reported in 1950 by Lawrence Craven, a general practitioner from suburban California. A few years previously, Craven had noticed that patients who chewed newly-launched aspirin gum for pain relief experienced significantly more hemorrhaging after a tonsillectomy or a tooth extraction. Craven realized that the right dose could help patients at risk of myocardial infarction or stroke, so he decided to test his theory for himself. Two years and 400 patients on low-dose daily aspirin later, not one of them had suffered an MI.
Famously, Viagra® was discovered as a treatment for ED during clinical trials for its intended use: treating angina. Nearly $2 billion in profits later, it’s another high-profile example of the lucrative possibilities of drug repurposing.
The problem is, of course, that we don’t want to have to rely on luck. After all, even Craven’s work lay mostly undiscovered for decades; it took over 50 years until the world caught up enough for his recommendations to become part of widespread health policy. There are currently just over 3,000 FDA-approved treatments that are approved for almost the same number of diseases, leaving almost 10,000 diseases without a single approved therapy. Waiting around for clinical inspiration to strike leaves patients suffering when it is perfectly possible (indeed likely) that effective treatments for at least some of these conditions already exist.
Drug Repurposing Technology: How We’ve Gotten Here
In the current artificial intelligence (AI) craze, it’s easy to forget that the history of computational, in silico screening for drug repurposing dates back to long before the large models and high compute deep learning methods that are getting us all excited today. As our own CSO, Kevin Brogle, can attest, computational chemistry has been around for quite some time. From the late 20th century, advancements in computer technology and bioinformatics have enabled researchers to digitally simulate biological processes and drug interactions in previously undreamed ways.
Initially, these methods relied on databases of molecular structures and simplistic algorithms. They could identify potential new uses for existing drugs by predicting their interactions with various biological targets. As technology progressed, more sophisticated computational models emerged, incorporating complex molecular dynamics and high-throughput virtual screening techniques.
While early, pre-AI/ML methods resulted in some interesting developments, they were few and far between—until more recent approaches revolutionized the field. By introducing machine learning algorithms capable of analyzing vast datasets with unprecedented speed and accuracy, ML/AI-driven in silico screening makes a lot of exciting things possible. For example, we now have the technology to identify intricate patterns and correlations within biological and chemical data, uncovering novel drug-disease relationships that would be slow, difficult, or even impossible to detect through traditional methods.
AI-Assisted Drug Repurposing: What’s Next?
Since its introduction, AI-assisted drug repurposing has resulted in some interesting successes, including baricitinib for COVID-19. The team here at Accencio® were able to use our IP-GeoScape® algorithms and technology to identify clinical-stage molecules as potential retargeting candidates as part of our COVID-19/SARS-CoV-2 initiative in 2020. There are yet more recent AI/ML-facilitated retargeting discoveries in early-stage testing, which we will watch with interest.
Even with these promising successes, we have barely scratched the surface of what is possible. The future of AI-facilitated drug repurposing is poised to be transformative, driven in no small part by the latest advancements in deep learning models. These models, which excel at processing and learning from vast and complex datasets, enable researchers to uncover previously hidden relationships between drugs and diseases with remarkable precision. They can even drill down to specific patient populations, leading to more personalized and effective treatments. Excitingly, as deep learning models continue to evolve, their ability to simulate and predict biological interactions will become increasingly sophisticated, enabling us to approach drug repurposing from new angles and further reducing the time and cost associated with drug development.
Ultimately, AI-driven drug repurposing holds the potential to revolutionize the pharmaceutical industry, bringing new hope to patients by rapidly and effectively expanding the therapeutic uses of existing medications.
Accencio’s flagship product, IP-GeoScape®, is a visual landscape of the molecular IP space representing any given chemical area. The clustering algorithm and visualization tool enables researchers to contextualize ideas and See IP Differently™. Please contact us if you are interested in a demo or speaking with an IP-GeoScape expert about how we can help you.