Forecasting the evolution of cancer

Bethan Warman Posted in General News, Publications 28 May 2018

Predicting the trajectory of tumour growth

New research, published today in Nature Genetics, has developed a computer model that forecasts the changes that occur within tumours as they develop. In the future, it is hoped that such a model may enable the prediction of the trajectory of tumour growth in patients, allowing clinicians to pre-empt disease course and tailor treatment regimens accordingly. 

Prof Trevor Graham (left) and Mr Marc Williams (right)

The model was developed in collaboration with researchers from BCI’s Centre for Tumour Biology, led by Prof Trevor Graham (Lead for the Evolutionary and Cancer Biology Laboratory), the Institute of Cancer Research, led by Dr Andrea Sottoriva, and University College London, led by Dr Chris Barnes.

Reading the 'secret diary' of a tumour

As tumours grow and become more advanced, different mutations- changes to the genetic code- occur in different cancer cells. Some of these changes may confer an advantage for the cell, making it more suited to survive in its environment. Following the principles of Darwin’s theory of Evolution, such cells will be favoured and able to replicate, creating more cells with the same mutation. In this way, different populations of cells with different mutations build up, creating tumours that are highly genetically diverse.

Trevor Graham clonal mutations diagram

The authors simulate the process of tumour growth from the first tumour cell to a large tumour (left to right), with new mutations appearing each time the tumour cells divide (shapes).

Although it is known that such changes take place as cancer develops, knowing exactly what changes occur and when has proven difficult as tumour samples are often only analysed in the laboratory after they are large enough to be detected in a patient. Therefore, the history of how tumours grow has remained invisible to researchers. The present study, majoritively funded by The Wellcome Trust, endeavoured to fill in this missing history.

By interpreting the genetic sequences of tumour cells, the computer system is able to read the ‘secret diary’ of mutations hidden in the genome, learning how the cancer has changed over time. Notably, the study revealed that mutations acquired by cells that drive the progression of cancer- known as driver mutations- allow the cells to grow up to 30% faster in some cases than cells without the mutation.

Using a tumour's history to predict its future

The computer system used this genetic information with defined algorithms to predict future changes that may occur as the cancer continues to evolve.

How a cancer changes over time underlies a patient’s prognosis, as such changes may determine response to treatment. This study, also supported by Cancer Research UK, makes steps towards the ‘mechanistic forecasting’ of future disease course and offers great potential for personalised medicine.

Prof Graham said:

Forecasting of the future offers a real opportunity for clinical benefit that is based on anticipatory action. It would be a great advantage to know if a tumour is going to change in a certain way and so be able to administer the most effective treatment in light of the prediction. This work is in its infancy, and the possibilities that this research may have as it matures are very exciting.

Futute directions

Before this research can become a clinical reality, the team first need to verify that the predictions of the system are correct by testing it in pre-clinical models and analysing tumour samples of patients that are under long-term surveillance.

Mr Marc Williams, first author of the paper, added:

We are measuring changes that occur in cancer that have proven difficult to quantify. Such changes may be important in the evolutionary dynamics of individual cancers. Hopefully with these kinds of measurements we will be able to make important predictions of what the future evolutionary trajectory of cancer could be. This is the long-term aim.

Leave a comment

You are commenting as guest. Optional login below.

This site uses cookies in order to function properly. By continuing to browse, you agree that we can save them on your device. Privacy Policy.