The Hidden Cancer: Cancer of Unknown Primary
This article was originally published in Issue One of Discover: Cancer Research In Manchester. All articles are available to read on the MCRC website and a PDF version can be accessed through the links at the end of the page.
A cancer of unknown primary, also called CUP, is a type of rare cancer that affects around 8,100 people in the UK each year.
In patients with CUP, cancer cells are found, but doctors cannot tell where in the body the cancer started. Even after performing scans and other tests like biopsies, the source of the cancer remains unknown. Detailed analysis of the biopsied cancer cells looking for a telltale appearance or markers confirm that they are metastatic, meaning they have come from another part of the body, however the site of the original cancer remains unknown.
Because doctors cannot tell where the cancer started (the primary cancer) CUP are particularly hard to treat.
In April 2024, researchers at the Cancer Research UK National Biomarker Centre including Drs Alicia-Marie Conway, Alexandra Clipson and Steven Hill published research on a new tool called CUPiD, which leverages biomarkers to help identify the primary site of a CUP.
A cfDNA methylation-based tissue-of-origin classifier for cancers of unknown primary
Instead of relying on tissue from invasive biopsies to detect the primary cancer, the National Biomarker Centre team use a liquid biopsy, a specific kind of blood test that can detect signs of cancer, such as cancer cell DNA.
All cells in our body release DNA into our blood as they grow and die, and cancer cells are no exception. This creates a mixture of normal and cancer cell DNA known as circulating-free DNA, or cfDNA in our blood.
The researchers at the National Biomarker Centre were specifically interested in a type of alteration that happens in normal and cancer cell DNA called methylation, which is where chemical tags called methyl groups are added to DNA, instructing the body to either ignore or listen to certain genes.
The addition of these methyl groups, called methylation patterns, are consistent within the same cancer type, but vary between different types of cancer. They are like unique postcodes for each type of cancer that can pinpoint where that piece of DNA has come from, making them very useful to identify CUP.
CUPiD is a machine learning tool- a form of artificial intelligence, called a classifier that analyses data and classifies it into categories.
The team trained CUPiD using existing methylation datasets of various cancers to teach it how to identify and classify different cancers. By comparing its predictions to known cancer types in the database, the model gets more and more accurate, and with repeated training, CUPiD is able to predict the cancer type of previously unseen samples.
When CUPiD was tested on liquid biopsies from patients with CUP, it correctly predicted the type of primary tumour in 88.5% of cases.
What’s next?
Although further testing and refinement of the tool on larger CUP datasets is needed, CUPiD’s approach to classify CUP is promising.
“The next steps are to test CUPiD in a larger number of samples from patients. This will enable us to assess how well the test works and who might benefit most before designing a trial where we use the test to make treatment decisions. Being sure the test is robust and accurate is an important step before it can reach patients.” Dr Conway explains.
“We have secured funding from a Cancer Research UK Biomarker Project Award supported by Stand Up To Cancer® to do this, we will refine the tool and then test it on over 500 samples from patients where we know where the cancer started.” Dr Clipson says.
The hope is that the liquid biopsy approach would be introduced for all patients with CUP at the beginning of their diagnostic journey, so that their cancer can be compared to known samples to identify the primary cancer as early as possible.
The team has also recently secured funding to adapt CUPiD to identify different types of biliary tract cancer (BTC). Diagnosing BTC often requires an endoscopic biopsy, but a limited amount of tissue can be obtained. So, the thought is that similar to its use in CUP, CUPiD could help detect unique methylation patterns in BTCs and differentiate the subtypes, enabling patients to be diagnosed earlier without the need for repeat invasive biopsies.
How could this impact patients?
Normally, a lab technique called immunohistochemistry is used to identify a CUP, where biopsied patient tissue is stained to try to identify proteins specific to a particular type of cancer. But, as there are so many different types of cancer, patients often have to undergo multiple, invasive biopsies to collect tissue for testing. These biopsies are an additional, unnecessary strain on patients and the testing is time consuming.
Also, as doctors cannot identify a patient’s primary cancer, they end up being treated with chemotherapy which often does not work well. This is because the newer, more tailored therapies like immunotherapies are not currently available without knowing what the primary tumour is.
If widely adopted, this liquid biopsy method could reduce the need for invasive procedures, shorten the time it takes to diagnose patients, and increase the chances that patients receive a more targeted, effective treatment.
To find out more about CUP, you can visit the National Cancer Institute and Macmillan websites.