At a glance
New method helps researchers distinguish genetic mutations that cause cancer from those that merely accumulate during aging.
Results suggest that only a subset of non-inherited genetic mutations believed to cause cancer actually do so.
Findings also show that the average age at which patients are diagnosed with cancers containing a specific mutation provides a clue as to how dangerous that mutation is.
The work offers insight into the molecular roots of cancer and has implications for risk screening and drug development.
It’s a fundamental principle of science: Correlation does not equal causation. Every cancer cell has genetic mutations, but not all of those mutations necessarily drive the cancer.
A Harvard Medical School-led research team has just developed a statistical method that helps distinguish mutations that cause cancer from those that simply accumulate as we age.
In doing so, they’ve found that only a subset of non-inherited, or somatic, mutations believed to be carcinogenic actually drive cancer — supporting and quantifying a growing suspicion in the field.
They’ve also shown that the average age at which patients are diagnosed with cancers containing a specific mutation predicts how dangerous that mutation is, with mutations seen in younger patient groups being more carcinogenic.
The results, published May 5 in Nature Genetics, offer a new way to understand the molecular roots of cancer — an especially urgent need as rates of certain cancers rise in young people. Age is not currently taken into account in efforts to define cancer mutations.
The findings could also inform cancer screening methods and drug development.
“It’s a provocative paper, I think. It formalizes the idea that the currently accepted list of cancer-causing somatic mutations may include some false positives,” said senior author Kamila Naxerova, associate professor of genetics in the Blavatnik Institute at HMS. “It also demonstrates that patient age should be taken much more seriously as a factor in cancer genetics.”
Trimming the cancer mutation list
Over the past decade or so, genomic sequencing of hundreds of thousands of cancers has produced a comprehensive list of mutations that are enriched in various cancers. Researchers deduced that each of those mutations contributes to the development of at least some cancers in some large or small way.
“The question of which mutations cause cancer appeared to be largely solved,” said David Cheek, research associate in genetics in the Naxerova Lab and first author of the new study.
But in more recent years, a practice called normal-tissue sequencing has revealed that many of those supposed cancer-causing mutations also occur in the tissues of people without cancer as they age.
Cheek wanted to address what he felt was the biggest question these two avenues of research raised: Which of the mutations that had previously been established as cancer drivers truly contribute to cancer risk, and which are merely features of normal tissue aging without causing cancer?
An Oxford- and Edinburgh University-trained mathematician, he combined his proficiency in statistics and probability with Naxerova’s expertise in cancer biology and genetics to develop a framework that can measure not only which somatic mutations drive cancer but also how powerfully each one contributes.
The importance of age
The team first applied its statistical method to existing cancer and normal-tissue genomics databases to estimate mutations’ cancer-causing effects. Specifically, they analyzed mutations seen in esophageal squamous cell carcinoma, acute myeloid leukemia, and colorectal cancer.
These estimates then made it possible to test a hypothesis Cheek developed early in the project: that more powerful cancer-causing mutations occur disproportionately in the cancer genomes of relatively young patients.
Indeed, he found that mutations that are more often found in the cancers of younger people are more likely to cause cancer than mutations that are enriched in older people’s tumors. He also found that the stronger a mutation’s carcinogenic effect, the younger the patients in which it’s typically seen.
“In the first part of this work, we developed a rigorous mathematical method to help distinguish cancer-causing from normal-aging mutations that we hope will do the community a service. The second part is arguably more important and creative — the thought that age actually encodes information about causality,” said Naxerova.
Naxerova and Cheek hope the work alerts more cancer geneticists to the importance of age. The authors note that nearly 30 percent of records in the major cancer sequence database COSMIC lack information about the patient’s age.
What’s next
Everything the team found about the disease contributions of individual mutations is still considered a prediction and will need to be confirmed in animal and human tissue studies.
Meanwhile, the methods can also be applied to predict which mutations are more or less harmful in a broader range of cancers. That includes cancers in tissues that don’t yet have sufficient normal-tissue sequencing data for comparison, such as the breast, lung, and kidney.
Naxerova’s lab has already begun such follow-up studies in colorectal cancer at Massachusetts General Hospital and in breast cancer with the Susan G. Komen Tissue Bank at Indiana University.
Although the gold standard requires waiting for complete normal-tissue comparison data and laboratory validation, the new methods and concepts offer information that researchers and clinicians can consider in current efforts to improve cancer prevention and treatment, the authors said.
For instance, mutations flagged as highly causative could indicate what is most important to look for when assessing a person’s cancer risk or trying to detect a cancer early.
They could also help pharmaceutical companies prioritize which mutations to target with anticancer therapies. A mutation doesn’t need to have immense cancer-driving power to be worth targeting, Naxerova said, but in theory, the more causal it is, the more likely a cancer can’t survive without it and won’t develop resistance to the therapy.
“Big questions in the field right now are: Can we predict whether a cancer is likely to arise in a given tissue? As we study the genetics of an individual’s tissue biopsy, what specifically should we be looking for to gauge cancer risk?” Cheek said. “Our work can contribute meaningful answers.”
Authorship, funding, disclosures
Additional authors on the paper are Martin Blohmer, Martin Nowak, and Tibor Antal.
This work was supported by the National Institutes of Health (grants R37CA225655, R01CA279054, R01CA269281, and P01HL142494), the Glenn Foundation for Medical Research, and an Emerging Leader Award from the Mark Foundation for Cancer Research.
