Tumour

Photo by Mary Lucien Lynch, MD

Cancer research that focuses on very specific tumours types may be limiting what we can learn, according to three presentations at the American Association for Cancer Research’s annual meeting on Monday. Most modern oncology research is aimed at a certain cancer type or tissue, but these presentations argue that we might be missing important information by not looking at the bigger picture.

The first was presented by Peter Van Loo from the Francis Crick Institute in London, England on behalf of his team. Describing cancer genomes as an ‘archaeological record of tumours past,’ Van Loo described how his team examined more than 2,700 cancers across more than 30 different types of cancer for somatic changes in their DNA. He demonstrated how his team had used these changes to infer the evolutionary history of the tumours by determining when they had occurred in tumour development.

Their results showed that the majority of these somatic mutations occurred in later stages of the tumour’s development, which might offer an explanation as to why late-stage cancer is more difficult to treat than cases identified earlier on. As more of these changes occur, more genetic mutations begin to contribute to the overall disease, making it far more complex and better at avoiding anti-cancer agents. The team also found that, in contrast, breast and ovarian cancers developed somatic mutations very early in tumour development.

With further work, the team were also able to determine that in most cases, the common ancestor cell that gives rise to the tumour appears 1 to 2 years before the patient’s initial diagnosis. Other cases saw the cell appearing as much as 5 years prior, whereas in certain lung cancer cases, whole genome duplications could occur up to 20 years before diagnosis.

The team then used 1,900 cancer samples to monitor the frequency of mutation types across different cancer forms, finding distinct changes in the quantity of elements like single nucleotide variants.

The second presentation was by Gad Getz, from the Massachusetts General Hospital Cancer Centre. His team have been working on using multiple modelling tools to identify the driver mutations in different forms of cancer. Using different modelling tools to examine the mutational burden, clustering of mutations, and impact on cell function across multiple cancer types, the team developed a system to detect variants of significance that offers improved sensitivity and specificity over other techniques.

The work has involved 3,000 whole genomes but Getz has said that they are a long way from characterising a complete landscape of driver mutations. However, their work has been able to detect mutations that other tools would have overlooked and looks promising for future development. Getz hopes that building a better understanding of the non-coding regions of the genome will help the team to improve their system.

Alexander Baras, researcher at Johns Hopkins, gave a third presentation detailing his team’s work comparing different sizes of gene panels to analyse the mutational burden in tumours. Previous work has shown that total mutational burden correlates with responses to immune checkpoint inhibition.

The team used de-identified sequences through whole exome sequencing (WES), large gene panels, and small, more targeted panels to determine the non-silent total somatic mutational burden of the samples. Overall they were able to characterise the mutational burden from 14,000 samples. Of these, 2,000 were found to have high mutational burden and yet only roughly 130 were detected by the smaller gene panels. In contrast, the large gene panels were found to be almost equivalent with WES in terms of detection power.

Baras concluded that small gene panels lack the power to determine mutational burden, but large gene panels could be used as effective tools across a diverse range of cancer types.

 

More on these topics