Skip to main content

Could a new study give hope for earlier diagnosis of cancer?

Thursday 6 February 2020

"Signs of cancer can appear long before diagnosis," reports The Guardian.

Most cells in the body divide and reproduce constantly, picking up replication errors in their DNA over time as we age. Many of these errors may be harmless, but some can cause or increase the risk of cancer.

Cancers begin when harmful errors, or mutations, cause our cells to divide in an uncontrolled way. It's usually impossible to tell if this is happening, until the cancer starts to cause physical signs or symptoms.

In this new study, an international team of researchers sequenced the genomes (the entire DNA and genetic material) of 2,658 tumour samples.

They used the information to work out the order in which mutations and copying of mutations happened, because usually more than one mutation is needed before cells become cancerous. The researchers then modelled how different types of cancer develop over time.

They found that harmful mutations for some types of cancer, such as ovarian cancer, characteristically happen very early, in some cases decades before people have any physical signs of the disease. The findings raise hopes that some cancers could be detected and treated much earlier.

However, at present it's not clear whether this research could lead to a cancer screening system based on checking for "genetic early warning signs", both in terms of effectiveness and feasibility.

At present, the best way to detect cancer early is to be alert to the possible signs and symptoms, attend cancer screening when invited, and know about your family history of the disease.

Find out more about:

Where did the story come from?

The research was carried out by the Pan-Cancer Analysis of Whole Genomes (PCAWG) Consortium, an enormous collaboration between hundreds of scientists from 4 continents. 46 scientists worked on this particular paper, from 38 universities or research institutes.

The PCAWG group published 6 papers this week, but we're focusing on just 1, which looked at the way cancers evolve over time. The study was published in the peer-reviewed journal Nature on an open-access basis, so it is free to read online.

The Guardian, BBC News and Mail Online focused on the discovery that DNA changes to cells may happen many years before cancer can currently be diagnosed, and the reporting was generally accurate.

What kind of research was this?

This was a modelling study, using data from the whole genome sequencing of 2,658 cancers to reconstruct the likely evolution of DNA in these cancers over time. The study helps scientists to better understand how cancers begin and evolve.

However, at this stage, the results cannot be used to test for cancers in people.

What did the research involve?

A team of scientists worldwide worked with 2,778 samples of cancers, taken from 2,658 people with cancer. Some people gave just 1 sample, while others gave a sample of newly diagnosed primary tumours, and later, a sample of a metastatic cancer (when cancer has spread to another part of the body). 38 cancer types were represented in the samples.

The scientists carried out whole genome sequencing of the samples. This showed where DNA mutations arose, and whether they had been copied and duplicated as more DNA changes accumulated.

Researchers could look for so-called "driver" mutations, which are known to be linked to cancer, and see whether they happened early or late in the cancer's evolution.

They used this information to model a typical "life history" for each of the 38 types of cancer. This showed whether important mutations happened early or late in the cancer's development. They then estimated how that mapped against a person's life. For example, whether cancer-causing mutations happened a short time before cancer was diagnosed, or whether they had been present for years or decades before cancer was detected.

What were the basic results?

The researchers found that the time between cancer-driving mutations and diagnosis varies a lot between cancers. Some (such as liver and cervical cancer) happen 1 to 5 years before the cancer was diagnosed. By contrast, ovarian cancers showed significant mutations 10 to 40 years before diagnosis. This suggests the original mutations that lead to some adult cancers could happen during childhood or adolescence.

Other results included:

  • a small number of mutations in genes known to drive cancer development happen early in the development of several cancers
  • a larger variety of rarer mutations are responsible for driving later cancer development
  • some mutations that are linked to outside influences (such as those caused by tobacco smoking in lung cancer and UV light exposure in skin melanoma) happen early in cancer development
  • some mutations, including signs of attempts by the cells to repair their DNA, tend to happen later in cancer development

How did the researchers interpret the results?

The researchers said: "Our study sheds light on the typical timescale of… tumour development, with initial driver events seemingly occurring up to decades before diagnosis." They say the results "highlight opportunities for early cancer detection".


This study represents an enormous achievement by many scientists working together to find out more about how cancers develop over time. This type of work is likely to be important in developing future tests for cancers, and possibly new treatments that can target cancers at a very early stage.

However, the study does not change how cancer is diagnosed or treated at present. It can take years before early-stage research like this leads to changes in clinical practice.

As one of the scientists involved in the study told journalists, the idea of being able to target mutations by doing blood tests during childhood, then eliminate dangerous mutations, is "science fiction".

This research is very complex and, as with all modelling, it relies on some assumptions about the time it takes for mutations to arise, be duplicated and copied. The accuracy of the findings will depend on the accuracy of these assumptions.

All samples in the study came from people who had developed cancer. It would be interesting to compare findings with non-cancerous tissue samples from these people, or samples from people who did not develop cancer.

It's good news that DNA sequencing technology now allows scientists to work on such a large scale, and that they're able to work together to find out more detail about the way that cancers evolve. This type of work could make a big difference to the way doctors approach cancer in future.

Analysis by Bazian
Edited by NHS Website

Links to the headlines

Signs of cancer can appear long before diagnosis, study shows

The Guardian, 5 February 2020

Landmark study to transform cancer treatment

BBC News, 5 February 2020

Cancer breakthrough as scientists find tumours give off traceable 'signals' up to DECADES before they grow to be malignant

Mail Online, 5 February 2020

Links to the science

Gerstung M, Jolly C, Leshchiner I, et al.

The evolutionary history of 2,658 cancers

Nature. Published online 6 February 2020