Cancer Grand Challenges scientists and collaborators have created the first map of the major chromosomal changes that occur during cancer. This information could help physicians better understand how individual tumours behave and select the most effective therapies for people with cancer.
Copy-number alterations are major changes to chromosome structure that result in large gains or losses of swaths of DNA. These changes contribute to cancer development and growth, and poorer patient survival. A global team of researchers, including those working on our Unusual Mutation Patterns challenge, has created a map of copy-number alterations in human cancers – the first of its kind for this type of DNA damage. Their findings were published today (Wednesday 15 June) in Nature.
To generate this map, the team first developed a unique set of mathematical features to capture salient copy-number changes, which they implemented in a suite of artificial intelligence techniques called SigProfiler, developed through Cancer Grand Challenges. With the map, they then assessed how copy-number alterations influence outcomes for people with different types of cancer. “We’ve shown that it’s usable in both clinical and research settings, potentially to guide clinical decision-making or unlock important information about tumour biology, respectively, and we’ve already used it to identify potential biomarkers for drug therapies,” says clinician scientist Dr Nischalan Pillay, a pathologist and group leader at University College London (UCL), and co-senior author of the study.
The bigger picture
For many years, cancer research has focused on simple mutations – single- and double-base substitutions, and small insertions and deletions – in individual genes. However, because genes make up only about 2% of the human genome, this narrow approach has left the full landscape of genomic alterations unaccounted for. Furthermore, this approach overlooks copy-number signatures, important mutational patterns that affect biological processes but were not previously represented in the main mutational variant classes because they are challenging to interpret.
According to Nischalan, the challenge of unpicking the genomic complexity imparted by copy-number changes has meant that genomic scientists have been “unable to fully capture the bigger picture of how vast swaths of genes and other DNA can be copied, moved around or deleted, without catastrophic consequences for the tumour.” But copy-number signatures can now be used to facilitate such analysis by using the SigProfiler suite, developed in the laboratory of Dr Ludmil Alexandrov, Cancer Grand Challenges co-investigator and associate professor at the University of California, San Diego.
In prior research funded by Cancer Research UK, Nischalan, Ludmil and colleagues used SigProfiler and other algorithms to identify copy-number alterations in a rare and aggressive type of cancer known as undifferentiated soft tissue sarcoma. With knowledge from that study, the team wondered what more they might learn if they applied this method across cancer types.
In their new study, the research team used SigProfiler to analyse the genomes of more than 9,800 people with 33 different types of cancer, and identified 21 copy-number signatures that were common to all cancer types, and resulted in either too many or too few chromosomes. Together, these signatures were used to construct a map, or blueprint, that researchers and clinicians can now use to predict the progression of a person’s cancer, including its aggressiveness, and potentially provide tailored treatment.
In particular, the team discovered that tumours in which chromosomes have shattered and reformed – a phenomenon known as chromothripsis – are associated with poorer survival outcomes for patients. For example, people with glioblastoma, a brain tumour that is particularly difficult to treat, were found to live for six months fewer if their tumours had undergone chromothripsis.
Broad applications from the lab to the clinic
To ensure the applicability of their method in both clinical and research settings, Nischalan, Ludmil and the team tested its application with several DNA-sequencing technologies. These technologies included whole-genome sequencing, which is increasingly being clinically used in the UK to determine the exact DNA sequence genome at the highest possible resolution, and single-cell DNA sequencing, which is often used in discovery research to enable the isolation of individual cells, and determination of their function and behaviour. Their method was found to be applicable across the spectrum of sequencing techniques.
Looking ahead, the scientists aim to refine the algorithm – which is akin to algorithms used by online streaming services to recommend movies and products – to help physicians predict how a patient’s cancer is likely to behave according to the genetic changes that drive the tumour’s growth.
The copy-number map will be hosted on COSMIC, the world’s largest, most comprehensive resource for exploring the effects of somatic (acquired) mutations in cancer. Additionally, the scientists have made the SigProfiler artificial intelligence toolkit freely available to others in the research community.
“Making these powerful computing tools free to others will accelerate progress towards a personalised blueprint for patients, based on the copy number signature of their tumour, giving them the best chances of survival,” says Dr Christopher Steele, first author of the Nature article and postdoctoral researcher in Nischalan’s laboratory at UCL.
Written by Scott Edwards
(Image: microarray; credit: NCI)