20 July 2021
Oncode collaboration results in a new mathematical method that can predict cancer survival based on genetic changes
As cancer survival rates vary widely, Oncode Researcher Daniel Miedema from Oncode Investigator Louis Vermeulen’s lab (Amsterdam UMC), in collaboration with the teams of Oncode Investigators Hugo Snippert (UMC Utrecht) and Geert Kops (Hubrecht Institute) has developed a mathematical method to predict survival based on genetic changes in cancer. The results of this study are now published in Nature Communications.
The study started with some questions. ‘We were wondering if genomic aberrations in the primary cancer can help to explain the chances of survival for patients, and if so - which feature of the genome in particular? A large body of research has shown that each patient has a unique tumor, and in particular tumors from different cancer types have huge differences. We hence reasoned that instead of looking for one feature, it might be more useful to explore the effect of the overall genetic diversity in the primary cancer on survival’ says Daniel Miedema, the principal investigator in this study.
The team’s first step was to obtain a comprehensive and scalable measure of genomic diversity. They developed a mathematical trick to calculate the genomic heterogeneity from a single measurement of chromosomal copy numbers. This tool then allowed them to study more than 10,000 tumors from 37 different types of cancer.
‘We found that for most cancer types, patients with a genomic diverse tumor have a poor prognosis. Interestingly, our approach also gives an indication why prognosis varies between cancer types: we for instance found that esophageal cancers have a high genomic diversity and patients have poor survival compared to, for example, thyroid cancers that have low genomic diversity and the prognosis for patients is relatively good’ says Miedema.
The tool they developed is simple to apply, cheap and very effective at predicting survival for cancer patients. ‘The development and validation of this tool was effectively done within the Oncode network. Bringing together the expertise of mathematicians, oncologists and genome biologists allowed us to develop a robust and well characterized method’ says Miedema.
And there is more to come. ‘Now that the tool is published, we are excited to look at data from a more clinically relevant setting, to see if the method can be used for clinical decision making. And together with the business developers from Oncode, we are exploring options to valorize this method’ he adds.