Scientists have used artificial intelligence to predict how cancers will progress and evolve—so that doctors can design the most effective treatment for each patient.

They developed a new technique called REVOLVER (Repeated evolution of cancer), which picks out patterns in DNA mutation within cancers and uses the information to forecast future genetic changes.

The ever-changing nature of tumors is one of the biggest challenges in treating cancer—with cancers often evolving to a drug-resistant form.

But a team led by scientists at The Institute of Cancer Research, London, and the University of Edinburgh were able to use their analysis of genetic changes to predict cancer’s next move—allowing doctors to stay one step ahead.

The research was funded by the Wellcome Trust, the European Research Council, and Cancer Research UK and is published today in the journal Nature Methods.

If doctors can predict how a tumor will evolve, they could intervene earlier to stop cancer in its tracks before it has had a chance to evolve or develop resistance, increasing the patient’s chances of survival.

Scientists at the ICR and the University of Edinburgh working collaboratively with colleagues from the University of Birmingham, Stanford University, and Queen Mary University London, also found that there was a link between certain sequences of repeated tumor mutations and survival outcome.

This suggests that repeating patterns of DNA mutations could be used as an indicator of prognosis, helping to shape future treatment.

For example, the researchers found that breast tumors, which had a sequence of errors in the genetic material that codes for the tumor-suppressing protein p53, followed by mutations in chromosome 8, survived less time than those with other similar trajectories of genetic changes.

Predicting the trajectory of tumor development

The ICR team and their colleagues developed a new machine-learning technique that transfers knowledge about tumors across similar patients. This method identifies patterns in the order that genetic mutations occur in tumors that are repeated both within and between patients’ tumors, applying one tumor’s pattern of mutations to predict another’s.

The researchers used 768 tumor samples from 178 patients reported in previous studies for lung, breast, kidney, and bowel cancer, and analyzed the data within each cancer type respectively to accurately detect and compare changes in each tumor.

By identifying repeating patterns and combining this with current knowledge of cancer biology and evolution, the scientists could predict the future trajectory of tumor development.

If tumors with certain patterns are found to develop resistance to a particular treatment, this novel methodology could be used to predict if patients will develop resistance in the future.