Artificial intelligence (AI) has helped scientists propose a promising new drug combination for DIPG brain tumours. Prof Chris Jones, Director of the Brain Tumour Research Centre of Excellence at the Institute of Cancer Research (ICR), and Dr Fernando Carceller, Paediatric Neuro-Oncology Consultant at The Royal Marsden Hospital, tell us more.
Diffuse intrinsic pontine glioma (DIPG) is a challenging brain tumour affecting the brainstem of children, usually around five to seven years of age. Unfortunately, for many years we knew little of the underlying biology of these tumours because, due to their location, obtaining tumour samples wasn’t common practice, and clinical trials based upon similar-looking gliomas in adults made little difference to the outcome of these patients.
The last 10-12 years have led to an enormous increase in our understanding of what makes DIPG so different to other cancers. However, for many of these new biological targets we’ve uncovered, there aren’t yet any drugs available in the clinic that can be given to these children.
An example of this is a gene called ACVR1, which was discovered to be associated with around a quarter of DIPGs nearly 10 years ago by Chris’ team at the ICR in London, along with several other groups worldwide. This gene had never been linked to human cancer but was instead known to play a key role in the development of a congenital malformation syndrome called fibrodysplasia ossificans progressiva (FOP). This exceptionally rare disease sees the soft tissue of an affected patient gradually turn to bone. Working with researchers in the FOP field, Chris’ team demonstrated that laboratory models of DIPG may respond to drugs targeting ACVR1, but none of these were yet suitable for administration to patients.
How AI helped us
Hearing this story presented at an international meeting in the US, an AI company called BenevolentAI approached Chris with a proposal to use their proprietary technology to search for drugs already in the clinic which might also target ACVR1.
Their platform scoured all available drug databases and scientific publications to look for such a drug, but there was a problem. As well as hitting ACVR1, any drug to be ‘repurposed’ in this way would also have to reach the tumour tissue, crossing the protective bloodbrain barrier – and it appeared there weren’t any such drugs available.
It’s a discovery that we wouldn’t have made without AI.
The flexibility and benefits of such a computational approach compared to manual sifting of the vast amount of information out there then became clear. The platform was subsequently tweaked to look for a combination of drugs such that one may increase the brain levels of the other.
The combination proposed were two drugs readily available for other indications – one of which targeted ACVR1, and another which blocked a cellular ‘pump’ which would otherwise limit the ability of the first to reach the brain, like bailing water from a boat. Most promisingly, this combination was already being tested in adults with brain metastases from certain forms of lung cancer.
Chris’ team first validated the use of these drugs in the lab, and then joined forces with colleagues in the Paediatric & TYA Neuro-Oncology and Drug Development Unit at the Royal Marsden Hospital, led by Fernando, to see if this was something that could be safely given to children. Both teams are currently working to translate these findings into a clinical trial to assess the efficacy of this drug combination in DIPG with mutations in ACVR1.
A glimpse of the future?
This drug combination is extremely promising for these children, who otherwise have few options for these types of tumours. It’s a discovery that we wouldn’t have made without AI. It’s exciting to think that there may be other new treatment possibilities for children with these tumours that we may uncover this way.
As AI gains a foothold in various parts of a patient’s pathway from helping analyse pathological specimens to reading radiological scans – it’s tantalising to think it might make the biggest difference in uncovering new and effective treatments for children with cancer.
From Contact magazine issue 102 - Spring 2024
