Cancer blood test uses ‘noise cancellation’ to improve sensitivity

For the longest time, scientists have been trying to find ways to use blood to better understand how cancer progresses inside the body.

As a tumour grows inside the body, it releases DNA into the bloodstream. And over the last decade, this circulating tumour DNA (ctDNA) has been the focus of researchers aiming to detect or monitor cancer.

This technique (known as a liquid biopsy) allows doctors to find out more about a patient’s cancer without the need for surgery. Although these ctDNA-based tests aren’t yet common, their use has been rapidly growing over the last few years.

The potential benefits are huge, both for designing clinical trials and for patients. Monitoring patients – particularly after they’ve received treatment – allows doctors to assess if treatment was successful and monitor for signs that the cancer might return.

Now, a team of Cancer Research UK scientists led by Dr Nitzan Rosenfeld at the University of Cambridge have developed a new experimental liquid biopsy that is up to ten times more sensitive than those being tested currently and has the potential to change how cancer is treated.

Cancelling out the noise

We’ve blogged before about the challenges with developing a blood test to pick up cancer. And a big issue is the complexity of changes to cancer cell’s DNA.

Cancer isn’t caused by just one error, it’s a culmination of many mutations that will vary from cancer to cancer and person to person. One of the ways that researchers can get around this complexity problem is to make the test more tailored. By analysing the individual genetic makeup of a person’s tumour, liquid biopsies can home in on a specific set of mutations and use them as a starting point to monitor the progression of cancer.

But knowing what to look for is just the start – because these tiny fragments of tumour DNA aren’t alone in the blood, they’re floating amongst millions of fragments of DNA from other cells.

Various methods have been proposed to reduce noise and improve signal to noise. In the last few months there’s been an explosion of methods that are based on tumour-informed analysis and personalized panels, with several exciting advances presented or published very recently.

– Dr Nitzan Rosenfeld

Scientists have worked to cut through this noise by looking for more than one DNA error, designing tests that pick up anything between 10 and 100 mutations. Using this approach, some tests are able to detect 1 mutant molecule amongst 30,000 pieces of DNA. But while the numbers seem impressive, they’re not good enough – particularly when there’s only a tube of blood to work with. This insensitivity means that even if the patient has enough cancer in their body to lead to a potential relapse, the test can come back negative.

Rosenfeld’s team thought they could do better.

According to Dr Rosenfeld this came about by “a combination of developing a lab process, which generates data from patient samples and control samples, and computational methods that have been developed to take advantage of the dataset that this generates”

The key was to combine data generated from personally profiling a patient’s tumour (looking for hundreds and sometimes thousands of mutations in each blood sample) with a clever computational solution.

It’s like listening to a quiet song on a pair of noise-cancelling headphones. Their new computational technique uses their data to “learn” the pattern of background noise – the molecular equivalent of a crowded street – in order to filter it out and better analyse the mutations that have been made clearer.

This uses control data to “learn” the pattern of background noise, and a series of filters and statistical methods that remove noise, define the detection classification algorithms and improve their confidence margins

– Dr Nitzan Rosenfeld

Putting their new techniques to the test on samples from 105 cancer patients, across 5 different cancer types and multiple stages of disease, they found the new technique was able to detect ctDNA at high sensitivity in patients with advanced breast and melanoma cancer, as well as patients with glioblastoma (a cancer traditionally difficult to detect in blood).

Compared to the traditional techniques, this new method can pick up 1 mutant molecule in amongst 100,000 – 1,000,000 pieces of DNA.

This sensitivity boost meant the test could detect tumour DNA in the blood of patients with earlier stage disease, included patients with early-stage lung and breast cancers, as well as those who had already undergone surgery for melanoma, despite the levels being much lower and more difficult to find.

Monitoring the future

Although it might be several years before this type of approach is ready to be used with patients, the team are still excited by what might come.

Further increases in sensitivity could lead to tests that would only require a drop of blood – meaning that patients could do it at home and send it off to the lab themselves. This would allow for more patients to have their cancers more continually monitored while not having to go to as many hospital appointments.

In future studies, the team and their collaborators plan to use this technique to measure ctDNA levels in people who are at high risk of developing cancer and hope that the information they will generate can be used to help refine future tests for cancer early detection.

Alex

Reference

Wan, J.C.M., et al. (2020) ctDNA monitoring using patient-specific sequencing and integration of variant reads. Science Translational Medicine. DOI: 10.1126/scitranslmed.aaz8084