Guido Sanguinetti nominated for Medical Physics breakthrough of the year for his work on AI in cancer evolution prediction

[2018] Guido Sanguinetti and Giulio Caravagna (postdoc in Informatics 2015-17) are co-leading a cancer study that got a lot of interest in the media: 64 news outlets around the world picked it up, including ITV News, Independent and Daily Express. It has been nominated as one of the 5 breakthroughs of 2018 in Medical Physics.

The work, in collaboration with colleagues at the Institute of Cancer Research in London, addresses one of the biggest challenges in treating cancers: the heterogeneous and dynamically changing nature of tumours. The team of researchers developed a new technique known as Revolver (Repeated evolution of cancer), which picks out patterns in DNA mutation within cancers to identify their evolutionary histories.

Scientists also looked at sequences of repeated tumour mutations and found a link to survival rates, which could be used as an indicator of prognosis. The fundamental insight is that cancer heterogeneity arises from the randomness of the evolutionary history of each tumour, but that prognostically relevant markers should be found by looking at multiple patients jointly.

The team then developed a new machine-learning technique which transfers knowledge about tumours across similar patients. This method identifies patterns in the order that genetic mutations occur in tumours that are repeated both within and between patients’ tumours, applying one tumour’s pattern of mutations to predict another’s. These repeating patterns constitute hallmarks of a particular cancer, and can be used to stratify more effectively patients and, ultimately, to predict the likely future development of the disease. It is hoped that this novel methodology could be used clinically in the future to allow medics to better target therapies to individual cancers.

The research is published in the journal Nature Methods.

Related links

Link to paper 'Detecting repeated cancer evolution from multi-region tumor sequencing data'