Latest News

Pan-Cancer Testing: A Closer Look at Different Technology Approaches

By Paul Nicolaus

April 5, 2022 | Diagnostics World recently took a broad look at the emerging field of pan-cancer testing, outlining some of the key players, main market drivers, and lingering challenges. Here we focus on different technology approaches that are being pursued. 

Researchers are currently exploring various approaches, including the measurement of changes in DNA or RNA sequences, patterns of DNA methylation or fragmentation, levels of protein biomarkers, and antibodies, among others. Some are striving to come up with a multi-layered approach in hopes of providing the most comprehensive picture of early-stage cancers. For now, though, it remains to be seen which of these differing routes will lead to the most effective method or methods. In the meantime, scientists continue to refine existing strategies and search for new possibilities. 

DNA Methylation-Based Approaches  

Perhaps the highest-profile player in the pan-cancer space, GRAIL is pursuing the promise of its multi-cancer early detection blood test, Galleri, which uses machine learning and cell-free DNA to search for abnormal patterns of DNA methylation. Methylation signatures help shed light on the tissue of origin, not only showing there is a tumor with a mutation that is potentially harmful but also revealing where that signal is coming from. 

Findings from its Circulating Cell-free Genome Atlas (CCGA) study appeared in Annals of Oncology in 2020 (DOI: 10.1016/j.annonc.2020.02.011). The study demonstrated the test’s ability to detect over 50 cancer types and localize the tissue of origin with over 90% accuracy. Across all cancer types, sensitivity reached over 90% in stage IV cancers but fell to under 20% for stage I cancers.  

Similarly, results of the third and final phase of the CCGA study published in 2021 (DOI: 10.1016/j.annonc.2021.05.806) revealed that that test’s sensitivity improves with cancer stage: 17% for stage I; 40% for stage II; 77% for stage III; and 90% for stage IV. Interim results of the company’s PATHFINDER trial were presented in June of 2021, with final results anticipated in the first half of this year. 

England’s National Health Service (NHS) has launched a trial aiming to recruit 140,000 people to further explore Galleri’s performance. The NHS-Galleri trial is run by Cancer Research UK and King’s College London Cancer Prevention Trials Unit, along with NHS and GRAIL. Initial results are anticipated next year. If successful, NHS intends to expand the rollout to one million more people in 2024 and 2025. 

Several other companies are pursuing methylation-based approaches, too. A 2020 Nature Communications study (DOI: 10.1038/s41467-020-17316-z) reported preliminary results of Singlera’s circulating tumor DNA methylation-based test, for instance. Findings showed that the PanSeer test detected five types of cancer in 88% of post-diagnosis patients with a specificity of 96% and detected cancer in 95% of asymptomatic people who were later diagnosed. 

Whereas methylation-based approaches have emerged as one of the leading technologies for early detection, Adela has pointed out that existing approaches tend to tap into only a subset of the methylome. Its platform, on the other hand, uses genome-wide methylation enrichment technology to mine the methylome more completely, distinguish the most highly informative regions, and target those regions for sequencing. 

The company, which launched last year, indicated that it has demonstrated its technology across ten different cancer types, and its approach does not require bisulfite conversion—a chemical treatment that leads to the loss of genomic material.  

Coupling DNA Mutations with Protein Biomarkers

CancerSEEK, a multi-analyte test under development by Exact Sciences, is designed to pick up on the presence of cancer-specific DNA mutations and protein markers. It takes very different technologies to evaluate those analytes, said Isaac Kinde, VP of technology assessment at Exact Sciences, but the thought is that this leads to improved performance.  

On the DNA side, “we’re looking for mutations in the most commonly mutated genes across all cancers,” said Kinde, a co-founder of Thrive—the company acquired by Exact Sciences. He noted that sequencing technology licensed from Johns Hopkins improves detection accuracy, which is crucial considering the DNA shed from tumors can be present in very low amounts. On the protein side, they are searching for “highly over-abundant proteins” associated with cancers.  

“So we have some technology for the DNA analysis, for the DNA that’s shed from cancers. We have a separate suite of technology for protein detection. And then we combine the results from the DNA and protein outputs with a machine learning classifier, which gives a single result, whether it be a positive or negative call,” Kinde explained.

Following a positive result, whole-body PET-CT imaging can be used as a follow-on to CancerSEEK to help identify where a cancer may be located and to help inform treatment decisions, he explained. 

A prototype version of CancerSEEK was evaluated in DETECT-A, a study of over 10,000 women between 65 and 75 without a history of cancer. In 2020, results published in Science (DOI: 10.1126/science.abb9601) revealed that 24 cancers were found by standard of care screening, and an additional 26 were found using CancerSEEK.  

The test detected cancers across ten organs, including seven that lack standard-of-care screening. Seventeen of the 26 cancers (65%) were localized or regional (stage I-III), including five with stage I cancer. The positive predictive value (PPV) of blood testing alone was 19%; when combined with PET-CT, that figure rose to 28%.

One percent of the study participants underwent PET-CT imaging based on false positives. Most of the time, a single imaging test could rule out the false positive, Kinde said, but in a small number of patients, additional follow-up was needed.  

Moving forward, the intent is to initiate a more extensive study with an eye on eventually securing FDA approval. “We want to continue to generate the data that’s going to be convincing to physicians and patients and also be reassuring to other aspects of the whole landscape, including insurance,” he added.  

Detection Roads Less Traveled 

Some others have set off in their own direction, making their way down lesser-pursued pan-cancer paths. The main drawback of most liquid biopsy approaches is that they depend on the presence of tumor-derived biomarkers in the blood, according to James Howard-Tripp, CEO of Canadian company StageZero Life Sciences.  

For cell-free DNA, cell-free RNA, and circulating tumor cells, signals tend to be stronger as cancer progresses and the primary tumor sheds more cells. In early-stage cancers, though, there tends to be a scarcity of these circulating fragments. “This makes early detection very difficult with much of today’s technology,” he said. And although there has been some demonstrable success, it is not consistent across tumor types. 

Howard-Tripp believes this weakness of many approaches is the strength of his company’s technology. StageZero uses messenger RNA (mRNA) technology to detect multiple cancer signatures from a blood sample to screen for several of the most common cancers, including breast, prostate, and colorectal.

Its Aristotle test looks at the activity of cells that make up the immune response in the blood and at interactions taking place between blood and tumor cells. He explained that these interactions amplify the tumor signal because a much larger number of modulated blood cells are being examined. 

According to Howard-Tripp, the blood transcriptome—the complete set of RNA transcripts present in a tissue or cell at any one time—can be used to come up with highly discriminative multi-gene panels for disease detection. Although a tissue or cell’s DNA is essentially unchanging, its transcriptome varies based upon the current physiological status of the tissue or cell. “These differences can be used to define specific gene expression patterns,” he added.

Findings published in the Journal of Clinical Oncology in 2020 (DOI: 10.1200/JCO.2020.38.15_suppl.e15037) showed that Aristotle classified cancer profiles with sensitivities ranging from 56 to 100% and PPVs spanning a range of 6 to 78% at a specificity of 99%. The mean false-positive rate for the 11 cancer classes fell between 0.3 and 7%.

Oxomics is another example of a company that is chasing after a less common approach. Rather than detecting genetic material from tumors the way many blood-based cancer tests tend to approach the task, this University of Oxford spin-out relies upon nuclear magnetic resonance (NMR) metabolomics.

The company combines the analysis of metabolites present in blood samples with machine learning to identify patterns of change indicating disease. The technique involves high magnetic fields and high-powered radio waves to profile metabolite levels in the blood. 

Healthy people, individuals with localized cancer, and those with metastatic cancer all have different blood metabolite profiles. Oxomics distinguishes between these states by focusing on roughly 90 small molecules, or metabolites, such as glucose and lactic acid. 

If you strike a tuning fork, it makes a note, and that note is unique to that tuning fork, explained James Larkin, CEO of Oxomics. “If you have one thousand different tuning forks, all with slightly different frequencies, and you struck them all at once, you could listen to that sound and carefully work out what tuning forks were present,” he said. “That is almost exactly the same approach that we’re using here.” 

The difference is that they aren’t using tuning forks; they are hitting molecules. And they are not hitting them in an audible range; they are doing so in the radio frequency range using radio waves. This can be converted into an information-rich spectrum, or fingerprint, revealing what metabolites and small molecules are in a person’s blood. From there, machine learning identifies patterns of change to determine whether a patient has cancer. 

In early 2022, he and colleagues published findings in Clinical Cancer Research (DOI: 10.1158/1078-0432.CCR-21-2855) detailing how their blood test can be used to detect a range of cancers and determine if they have metastasized. The study analyzed samples from 300 patients with non-specific symptoms of cancer and demonstrated a maximum sensitivity and specificity of 94% and 82%, respectively. In those with cancer, metastatic disease was identified with 94 percent sensitivity 88% specificity.  

Some challenges are unique to this approach, Larkin acknowledged. There are not many NMR-based techniques that are currently being used in the clinical setting, for instance. And the machinery required is not common, which presents a business scaling challenge. 

Even so, he sees plenty of upsides. “We don’t need to understand anything about the tumor before we start looking for it,” he said. In addition, the signal is amplified, so “we can see relatively small tumors with this systemic whole-body amplification of metabolism.” 

Beyond that, there are some logistical advantages. According to Larkin, this type of test is relatively inexpensive and doesn’t require much blood. In addition, the platform is universal, so the spectrum can be stored in a fixed manner. From there, a series of questions can be asked of it, like does this patient have cancer? And if so, has the cancer spread? “You can get a second answer out of the same data set, just by asking the right questions at the right time,” he said. 

The next step is to take what they’ve accomplished in the academic center and work on pushing it down the regulatory pathway. Future studies involving larger cohorts will continue to evaluate the technology.  

As he envisions the future of this field, Larkin believes that different techniques will wind up fitting in at different points in the patient pathway—and working in tandem. 

He sees his company’s approach as a nice fit for patients experiencing nonspecific symptoms of cancer, such as weight loss or fatigue. That could be useful for an initial diagnosis, he explained, but a circulating tumor DNA analysis might then be used throughout the course of treatment to help monitor progress. 

There is a whole lot of development going on currently, and it is hard to know where the final form of any of these technologies will end up. “It’s going to be very interesting over the next five or 10 years,” Larkin added. 


Paul Nicolaus is a freelance writer specializing in science, nature, and health. Learn more at