PhenMap and NHS: AI will identify who will not benefit from bevacizumab in colorectal cancer

British and Irish researchers have presented an AI approach that helps assess in advance whether bevacizumab will work for a patient with advanced colorectal cancer. If the method is confirmed on large samples, doctors will be able to prescribe toxic and expensive treatment less frequently to those for whom it will almost certainly not help.

How the Method Works

The Cancer Research Institute in London and RCSI in Dublin worked on the system. Their PhenMap tool combines several types of data: the tumor's genetic profile, the patient's clinical parameters, and disease characteristics, including age, sex, and the location of the tumor in the colon. Unlike standard stratification by several subtypes, the model attempts to capture more complex combinations of features and convert them into a practical prognosis for a specific individual.

After that, researchers ran another model on top of PhenMap, which assigns a risk group to the patient after bevacizumab therapy in combination with chemotherapy. The system divides patients into high, moderate, and low risk groups and assesses not just the overall severity of the disease, but the probability that this particular treatment regimen will be useless. This is important for personalized oncology: the task is not only to find those for whom the drug will help, but also to eliminate unnecessary treatment from the pathway of others in advance.

What the Study Showed

The team analyzed data from 117 European patients with metastatic colorectal cancer who had already received bevacizumab along with chemotherapy. The work was published in Scientific Reports. The model found biomarkers associated with poor treatment response and formed prognostic groups based on them.

In the high-risk group, not a single patient responded to therapy, whereas in the low-risk group, response was observed in 10 out of 12 people. Among the signals the AI identified, the mutation in the BRAF gene was particularly notable: patients with it fell into the high-risk group and had worse outcomes. Researchers also associated poor prognosis with two chromosomal deletions and used these features as the basis for a future test.

The idea is to see not only the diagnosis before treatment starts, but also the molecular profile of resistance of a particular tumor.

For clinicians, this could look like:

• Before prescribing therapy, a doctor receives not a general "average" prognosis, but an individual risk profile
• Patients with a high probability of non-response can avoid unnecessary toxicity
• The hospital saves time and resources on treatment that will have no effect
• For some patients, the path to other regimens or clinical trials opens up faster

Why This Matters

NHS approved bevacizumab for patients with advanced colorectal cancer in December, but the drug only works for a limited number of people and can cause serious side effects, including hypertension, gastrointestinal problems, and blood clots. England identifies almost 10,000 cases of this disease stage annually, and treatment options after tumor spread are significantly fewer. Therefore, even rough preliminary filtering by probability of response already has clinical value here.

"Until now, we have not been able to identify in advance patients who are unlikely to benefit from the drug," says

Professor Anguradh Sadanandan.

The authors separately emphasize that this is not yet a finished diagnostic test for hospitals. The results need to be confirmed on a larger cohort, and then verified in a prospective clinical trial. Only after such validation can the model be used for real treatment decisions. In parallel, the team wants to understand whether the same approach can predict response to other targeted drugs and transfer the method to other types of tumors.

What This Means

AI in medicine is increasingly used not for abstract "support for the doctor," but for a very concrete decision: who should not be given heavy therapy with no chance of benefit. If PhenMap passes validation, oncologists will have a tool that simultaneously makes treatment more precise, gentler for the patient, and more rational for the healthcare system. For NHS and other systems, this is also a rare case where AI can simultaneously reduce harm, accelerate treatment selection, and reduce unnecessary expenses.