Almost 30,000 women a year in the UK could have their lives saved or extended by a new AI breast cancer diagnosis tool that could be available on the NHS within three years, developers say.
Researchers from King’s College London are working with doctors at Guy’s and St Thomas’ hospitals and Google-backed startup Owkin to finalise and trial the new device. it is being trained to spot a sub-category of breast cancer that is not picked up with conventional diagnostic tests but could be treated much more effectively if it was.
This kind of cancer, known as HER2-low, is thought to be present in about half of the 55,000 new cases of breast cancer that emerge in the UK each year.
The early signs are that an effective new drug for another form of breast cancer could also work well against HER2-low – although more research is needed to confirm this.
“This has the potential to radically improve outcomes for patients across the NHS and beyond,” said Thomas Clozel, co-founder and chief executive of Owkin.
“Research suggests that many more women could benefit from targeted breast cancer treatments – we just need to find them. We hope to help thousands more women to benefit from targeted anti-HER2 treatments in the UK every year, with the transformative drugs able to extend and save patients’ lives.”
The HER2 proteins behind this form of cancer are overproduced by faulty genes. The new AI models will analyse hundreds of retrospective tissue samples, meaning that the amount of HER2 proteins present in the cancer can be determined more quickly and accurately from biopsy samples.
The developers of the device stress it is still very early days but that the indications are that their AI model will prove to be effective at picking up cases of HER2-low cancer, and that patients will, in turn, benefit from new drugs.
Dr Sheeba Irshad, breast cancer medical oncologist at Guy’s and St Thomas’, said: “If these methods prove to be successful, use of AI-aided assessment of tumour tissue could be rolled out to more quickly and accurately screen patients for these markers.”
Experts not directly involved in the project were also excited by its potential – again cautioning that much more work is needed to confirm its potential.
“The new AI models hold great promise for extending the benefits of targeted treatments to more women, especially those with lower levels of HER2 proteins,” said Dr Nigel Blackburn, director of drug development at Cancer Research UK, speaking in a personal capacity.
Dr Jakob Nikolas Kather, assistant professor at the University of Aachen and visiting professor at the University of Leeds, said: “This could speed up the detection of aggressive subtypes of cancer and help more women to receive optimal treatment.”