Abbie Fisher, associate at European Intellectual Property Firm, Withers & Rogers, discusses patenting a new frontier in personalised medicine with AI-powered biomarker discovery.
Key insights:
- By harnessing the power of AI approaches such as deep learning, it has been possible to accelerate the process of biomarker discovery.
- Patents can often relate to a novel correlation between a biomarker and a specific medical condition, this can lead to opportunities to patent drugs for use in the treatment of a specific disease or patient sub-group.
- Pembrolizumab marked the first example of regulatory approval by the US Food and Drink Administration for a treatment defined by pan-tumour predictive biomarkers.
In the past five years, there has been a boom in the patenting of biomarkers as diagnostics and targets for use in personalised medicine. Research scientists have been capitalising on advances in the application of AI, especially deep learning, to find new, improved, or repurposed treatments and therapeutics for specific diseases.
A biomarker is a naturally occurring molecule, gene or clinical characteristic that may denote a pathological or physiological process. These biomarkers can be an indicator of disease, but they can also inform clinicians about disease progression and clinical efficacy. Once identified, they can be used to develop treatment pathways for specific patient groups, facilitating a more personalised treatment plan.
Prior to the advances in AI, the process of finding new drugs and treatments, or repurposing ones already in circulation, was time-consuming and expensive. The learnings from focused clinical trials tended to be more limited and often geographically spread, which made it difficult to spot signs that certain treatments were, for example, more effective in groups of patients with a particular gene or other physiological identifier. This led to a ‘one size fits all’ approach to medicine, which is not always effective or optimal for everyone.
Accelerating discovery with AI
By harnessing the power of AI approaches such as deep learning, it has been possible to accelerate the process of biomarker discovery. AI-powered approaches can lead to more accurate and reliable outcomes than is possible from a more traditional analysis of the fewer data points available from patients in separate clinical trials. AI algorithms developed for specific purposes are able to analyse large datasets from a variety of sources, correlating multiple biomarkers, confounding variables and different outcomes simultaneously. Such algorithms do not require these correlations to be pre-programmed, and so are able to find features currently unknown or poorly utilised in science. For example, they might be used to find out why a particular group of patients might react to a drug in a certain way and expose potential pathways for more effective treatments.
The sheer volume of data that is now available to biotech companies, pulled from many different resources, can help to de-risk the costly process of finding new drugs and therapies. It can also help to identify opportunities to repurpose existing ones. Through the analysis of many layers of data, AI algorithms can extract the fine details of where and how a drug works in a treatment pathway, providing better insights about what dosages or other clinical interventions will be most effective.
In this field, patents can often relate to a novel correlation between a biomarker and a specific medical condition. Once identified, this can lead to further opportunities to patent new or repurposed drugs for use in the treatment of a specific disease or patient sub-group.
A patent should never impede a medical professional from providing a diagnosis or treatment, and strategies for patenting, for example, new biomarkers and repurposed drugs targeting biomarkers, can be directed by your patent attorney. AI-powered biomarker and drug discovery brings advantages such as an increased success rate of drug development, improved efficacy and safety outcomes, and a smooth manufacturing process as the infrastructure for a known drug will already be in place.
Pembrolizumab (Keytruda), is a type of cancer immunotherapy that is effective against several cancers including melanoma and classical Hodgkin lymphoma. Pembrolizumab has recently been approved in Europe for use in advanced solid tumours in patients with microsatellite instability-high or mismatch repair-deficient biomarkers. This previously marked the first example of regulatory approval by the US Food and Drink Administration for a treatment defined by pan-tumour predictive biomarkers. Another example of a biomarker-related patent is EP3036007B1, for Trichostatin A, and its use in the treatment of cancer. In this case, the drug is used if the patient has a raised level of aurora kinase A (AURKA), as part of a personalised treatment plan.
Conclusion
As these examples show, AI-powered biomarker discovery is breaking new ground and generating a wave of safe and effective treatments and therapies for patients around the world. This is a particularly hopeful time for patients with rare diseases, where initial investment, risk and available data were stumbling blocks. The fact that this can now be done much more quickly and cost-effectively than before is a bonus for both the biotech industry and patients.
As the scale tips further towards personalised medicine, it is likely that applications for AI-powered biomarker related patents will soar, bringing in a new and exciting future of healthcare for all.