AI Tool Developed by London and Dublin Researchers Could Spare Bowel Cancer Patients Needless Treatment
Researchers at London's Institute of Cancer Research and the RCSI University of Medicine and Health Sciences in Dublin have developed a groundbreaking artificial intelligence tool that could spare bowel cancer patients from gruelling and ultimately ineffective treatment, in a breakthrough that represents a significant step towards genuinely personalised cancer care.
The tool, named PhenMap, is designed to identify which patients with advanced bowel cancer are unlikely to respond to bevacizumab, a targeted drug used in combination with chemotherapy that is effective for only a minority of patients but carries a risk of serious side effects. In an initial study of 117 European patients, the tool successfully identified a high-risk group in which not a single patient responded to the therapy.
Background
Bowel cancer is the fourth most common cancer in the UK, with approximately 48,200 new cases diagnosed each year — 12% of all new cancer cases — and the second leading cause of cancer-related death, responsible for around 17,700 deaths per year. In Ireland, approximately 2,500 people are diagnosed annually, making it the second most common cancer in men and the third in women. Survival rates have improved significantly over the past 50 years, but remain highly dependent on the stage at diagnosis: over 50% of patients in Ireland are diagnosed at Stage 3 or 4, where treatment is considerably more challenging.
Bevacizumab, marketed as Avastin, is not a traditional chemotherapy drug that directly kills cancer cells. It is a monoclonal antibody that functions as an anti-angiogenic therapy, working by binding to a protein called VEGF-A that cancer cells produce in excessive quantities to stimulate the formation of new blood vessels. By blocking this signalling pathway, bevacizumab effectively starves the tumour of the oxygen and nutrients it needs to grow and spread. It can also help normalise the chaotic blood vessel structure within a tumour, potentially improving the delivery of co-administered chemotherapy drugs.
The core clinical problem is that a significant majority of patients with advanced bowel cancer do not benefit from bevacizumab, and there has been no reliable method to determine in advance which patients will respond and which will not. This means thousands of patients could be undergoing treatment that is unlikely to work for them while being exposed to unnecessary side effects — including high blood pressure, blood clots, and gastrointestinal complications — and potentially delaying the use of alternative therapies.
Key Developments
The PhenMap tool addresses this critical gap by integrating complex genetic and clinical data to generate a patient-specific risk score. Unlike traditional approaches that group cancers into a few broad subtypes, PhenMap analyses highly complex datasets combining genomic data — the genetic makeup of the patient's tumour — with clinical information such as the patient's age, gender, and the location of the primary tumour. The AI processes this information to identify intricate biological patterns that would be impossible for a human to discern, placing patients on a detailed response spectrum rather than into crude categories.
In the study of 117 patients, a second AI tool used the patterns identified by PhenMap to generate a risk score for each patient, stratifying them into high, moderate, or low risk groups. The most striking finding was that none of the patients in the high-risk group — the top 10% of risk scores — responded to bevacizumab treatment. The AI also identified that patients with a mutation in the BRAF gene were consistently placed in the high-risk group, confirming known associations while integrating this finding into a more sophisticated predictive model. This AI-defined high-risk signature could serve as a future clinical biomarker to identify patients who are highly unlikely to benefit from the drug before treatment begins.
The development of PhenMap is part of a broader trend of integrating AI into personalised cancer care within the NHS. Successful precedents, such as the OSAIRIS tool for radiotherapy planning, demonstrate a clear pathway for adopting AI technologies that enhance efficiency and patient outcomes.
Why It Matters
The development of this AI tool is a genuinely exciting advance for cancer patients in the UK and Ireland. If validated in larger clinical trials and adopted by the NHS and the Irish health service, it could spare thousands of patients each year from unnecessary suffering while freeing up NHS resources for treatments more likely to be effective. The tool represents a significant step towards personalised medicine — where treatment decisions are tailored to the individual characteristics of each patient's cancer rather than following a one-size-fits-all approach. For a disease as common and as deadly as bowel cancer, even modest improvements in treatment selection could translate into thousands of lives saved or improved each year.
Local Impact
The collaboration between the Institute of Cancer Research in London and RCSI in Dublin is a powerful example of the UK-Ireland research partnerships producing world-class science. For patients in Northern Ireland, who access NHS services but live on the island of Ireland, the development of tools that could be adopted across both health systems is particularly significant. The RCSI's involvement reflects the growing strength of Irish medical research, with Dublin increasingly recognised as a centre of excellence in oncology and biomedical science. The potential for this tool to be adopted by both the NHS and the Irish health service means that patients on both sides of the border could benefit from its insights.
What's Next
The researchers are now seeking funding for larger clinical trials to validate the tool's performance in a broader patient population. If successful, they hope to work with the NHS and the Irish health service to develop a pathway for clinical adoption. The journey from research breakthrough to routine clinical use typically takes several years, but the strength of the initial findings and the clear clinical need provide a compelling case for prioritising this work. More from Pharmaphorum. Institute of Cancer Research.
