UK scientists have developed an AI-driven method to gauge which patients will respond to a bowel cancer drug now on the NHS.

PhenMap, an AI tool part developed in London, is poised to assess which patients with advanced bowel cancer are likeliest to respond to a targeted drug new to the NHS.

It will also determine those least likely to respond.

The treatment, Bevacizumab, was approved in December last year for NHS use. It targets advanced bowel cancer by slowing its growth, but currently only works for a small patient group. It also comes with risks of serious side effects, including high blood pressure, gastrointestinal problems and blood clots.

PhenMap could spare thousands of patients from treatments that would be ineffective for them. The researchers behind it, from London’s Institute of Cancer Research (ICR) and RCSI University of Medicine and Health Sciences, Dublin, hope its approach could also result in new treatments in the future, including for other types of cancers.

They say PhenMap allowed them to “integrate complex data on the genetic makeup of the tumour”, including clinical information on gender, age and the side the tumour was on.

From this, they searched for new biological signals – patterns relevant to a patient’s response to Bevacizumab.

Phenotype mapping

Until now, scientists have grouped cancers into a small number of subtypes. But PhenMap, short for “phenotype mapping”, can pick up more complicated patterns and narrow these groups, such as by putting patients on a scale of one to 100.

Based on its patterns, another AI tool generated a score to indicate the risk of dying after treatment with bevacizumab and chemotherapy.

Each patient was allocated ‘high’, ‘moderate’, or ‘low’ risk. The highest 10% of risk scores were placed in ‘high risk’, the lowest 10% ‘low risk’, and the rest ‘moderate risk’.

From analysing the clinical outcomes, the research team established that none of the patients in the ‘high risk’ group responded to the treatment.

The complex pattern of features present within ‘high risk’ patients could be used as a biomarker, enabling doctors to identify patients unlikely to respond to Bevacizumab.

The tool also spotted that patients with a mutation in the BRAF gene were all in the high-risk group and had poor outcomes.

Anguraj Sadananda, professor in stratification and precision medicine at the ICR, said: “Once bowel cancer spreads to other parts of the body, there are very few treatment options available for patients. It is therefore positive that patients can now access the targeted drug Bevacizumab on the NHS. However, we know that the majority of patients won’t benefit from the drug, meaning thousands of people in England could be facing unpleasant side effects unnecessarily. Until now, we haven’t been able to identify these patients.”

He continued: “Our research uses advanced AI methods to pull together large amounts of complex data, helping us to spot patterns that would otherwise be impossible for a human to see, and to uncover the clues hidden within a patient’s tumour. In our research, we have shown that this allows us to identify the patients least likely to respond to treatment with Bevacizumab. While these findings are encouraging, they will need to be validated in a larger cohort to ensure they are applicable to all patients.”

Bowel cancer is particularly curable if diagnosed early, but is a major concern in the health service, with almost 50,000 cases diagnosed per year in the UK. Last month, it was also identified as one of 11 cancer types becoming more common in young people in England.

The full research was published in the journal Scientific Reports in April.