The story
- Bevacizumab is a targeted therapy that interferes with the blood supply tumours use to grow. It was approved on the NHS for advanced bowel cancer in December 2025. It works for some patients and not for others — and the ones it doesn't work for still get its side effects: high blood pressure, gastrointestinal trouble, blood clots.
- Researchers at the Institute of Cancer Research in London and RCSI University of Medicine and Health Sciences in Dublin built an AI tool called PhenMap ("phenotype mapping") that combines a tumour's genetic profile with clinical data — age, sex, and which side of the colon the cancer started — and produces a 1-to-100 score for likelihood of benefit.
- In a study of 117 European patients published in Scientific Reports, none of the patients PhenMap flagged as highest-risk responded to bevacizumab. Among that group, every patient with a particular BRAF gene mutation had poor outcomes.
- About 10,000 people in England are diagnosed with advanced bowel cancer each year, and around 7,000 are eligible for bevacizumab according to NICE. A reliable filter could spare a meaningful number from a treatment that was never going to work for them.
- The 117-person study is small. It will need to be validated in a larger cohort before clinicians could use PhenMap to guide actual prescribing, and the path through NICE-style appraisal is a long one. The tool won't be at the bedside next year.
- The principle is the interesting part. For most of the history of cancer treatment, "which patients will benefit from this drug?" has been answered by trying it and seeing. A tool that can answer that question before the trying begins is the kind of small, structural change that compounds.
"
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.
— Prof. Anguraj Sadanandam, Institute of Cancer Research, London
Sources