PhenMap AI Tool Identifies Bowel Cancer Patients Unlikely to Benefit from Bevacizumab

TL;DR: Researchers at the Institute of Cancer Research and RCSI Dublin have used an AI tool called PhenMap to identify advanced bowel cancer patients who are unlikely to respond to bevacizumab, a drug the NHS approved in December. The approach could spare thousands from ineffective treatment and its side effects.

The study tracked 117 European patients treated with chemotherapy and bevacizumab, analysing tumour genetic profiles to find patterns in drug response. PhenMap — short for “phenotype mapping” — integrates genetic data to surface patient subgroups that standard pathology cannot.

Context and Background

Roughly 10,000 cases of advanced bowel cancer are identified in the UK each year, with young adults seeing a particular rise in diagnoses. The disease has the second-highest cancer mortality rate in the country after lung cancer. Five-year survival falls from 98% when caught early to as low as 10% for advanced presentations.

Bevacizumab slows tumour growth by restricting the proteins cancers need to develop, but benefits only a small subset of patients and carries risks including blood clots and gastrointestinal side effects. The PhenMap analysis flagged a patient group sharing the same gene mutation who all showed high risk of adverse reactions — the kind of signal that is difficult to isolate manually across multi-dimensional genomic data.

Professor Anguraj Sadanandam of the ICR said the tool 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”. He cautioned that the findings need validation on a larger cohort before clinical use.

For NHS commissioners, the wider context matters: bevacizumab was approved in December 2025 specifically for advanced bowel cancer, but without a companion diagnostic, clinicians have limited tools to target the patients most likely to benefit. An AI-derived stratification test, if validated, would fit the precision medicine approach the NHS Long Term Plan already commits to.

Looking Forward

The research team now plans to expand PhenMap to larger patient samples and test whether the method transfers to other cancer types. For practical clinical deployment, the path runs through standard NICE-style health-technology appraisal — PhenMap would need prospective validation and a defined diagnostic pathway before cancer multidisciplinary teams could use it to guide bevacizumab prescribing.