News Centre

Predicting which cancers will return

In 2020, we awarded Dr Anna Wilkins and Dr Erik Sahai funding to develop and test software to predict which cancers will return after treatment. January 2023 marked the successful end of their PCR project, so we’re taking a look back at what their project means for patients and what they have achieved.

AI: Will my cancer return?

Dates: June 2021- January 2023

Location: Francis Crick Institute, London

Amount: £100,000

We have many successful treatments for prostate cancer which means that for many people, we can get rid of their cancer but sometimes the cancer can come back. It can return in the same area of the body, or it may return in a different area. This happens when a very small amount of cancer cells are left in the body after treatment. At the moment, we cannot predict whether someone’s cancer is likely to return which makes it difficult to know how to treat each individual’s cancer.

If we could predict this, we could make sure that those with a high risk of cancer returning could have their treatment changed to reduce the chance of the cancer coming back. We could also avoid over treatment of those patients whose cancer is unlikely to return, meaning that more people could avoid side effects such as incontinence, erectile dysfunction, and body changes such as the growth of breast tissue.

Studies have found patterns in the prostate tissue that can be used to predict whether a cancer is likely to return. Unfortunately, finding these patterns needs to be done by a highly skilled individual, and it is time consuming and expensive.

Recent advances in artificial intelligence enabled Anna and Erik to develop computer software to try to find these patterns quickly and automatically. The software can also ‘learn’ and become better with more use. Over the course of this project, they tested and refined their software’s ability to find the patterns by using it to analyse tissue samples collected from patients with prostate cancer during the CHHiP clinical trial.

Anna and Erik have now successfully completed their PCR project. They proved that their technology could identify patterns which mean someone’s cancer is more likely to return. Crucially, these patterns were the same as those identified in previous non-computational studies. This suggests that the results from Anna and Erik’s project are likely to apply more widely in prostate cancer.

These results mark a huge step forward for more personalised treatment. The researchers will now take their technology forward and prove that it can analyse a wide range of patient tissue, for example, by testing it using tissue from men who have more advanced prostate cancer. They also aim to continue to make improvements to make the software work even better.