Skip to main content
Premium Trial:

Request an Annual Quote

Studies Set Stage for Further Clinical Development of Circulating Tumor DNA


As research reveals tumor features that can affect circulating tumor DNA levels, investigators are considering several sequencing and PCR-based clinical applications for the bloodstream-borne nucleic acids — from early detection and diagnosis to minimal residual disease monitoring and treatment response tracking.

At the moment, many applications of ctDNA involve targeted testing for point mutations or rearrangements that have been detected through targeted, exome, or genome sequencing of biopsied tumor tissue. As sequencing prices dip and increased coverage can be obtained more cheaply, though, researchers are optimistic about the prospect of taking a more widespread look at tumor genomes by directly sequencing ctDNA.

A study published in Science Translational Medicine last week highlighted the feasibility of monitoring treatment resistance-related mutations with ctDNA, for example. There, Italian researchers used PCR-based ctDNA testing to keep tabs on mutations in a handful of genes that they linked to acquired EGFR-targeted treatment resistance in four individuals with metastatic colorectal cancer.

"At the beginning we were unable to design a clinical trial to obtain bioptic material at progression," senior author Alberto Bardelli of the University of Torino told Clinical Sequencing News. "In order to understand why the patients would relapse — these are metastatic colorectal patients — we would have to do individual biopsy at progression to see what was going on."

"One way around this was to consider the possibility of using circulating tumor DNA as a proxy to see mutations at resistance," Bardelli explained. "That turned out to be quite effective."

In an accompanying study in the same issue of STM, a Johns Hopkins University-led group, which included Bardelli, examined the prevalence of ctDNA in blood samples from hundreds of individuals with different cancer types at a range of clinical stages. That analysis highlighted the inter-individual and tumor variation associated with ctDNA — patterns expected to inform the development of effective clinical applications of ctDNA.

"The main goal is [understanding] how we apply this clinically," co-corresponding author Luis Diaz, director of translational medicine at the Ludwig Center for Cancer Genetics and Therapeutics at Johns Hopkins, told CSN.

While he pointed to the particular potential of using ctDNA in early detection schemes down the road, Diaz said that "from an investigational standpoint [all applications] are very important and that should be embedded in targeted clinical trials."

He and his colleagues were among the first to explore tumor DNA in blood samples of individuals with cancer: The group has been investigating ctDNA and its potential clinical utility for nearly a decade.

Together with several other co-corresponding authors on the study, Diaz founded the Baltimore, Md.-based spinout Personal Genome Diagnostics (PGDx), where Diaz currently serves as chief medical officer.

As CSN reported in 2012, the company is focused on developing technologies to clinically translate findings from cancer genomics, ctDNA, and other studies done at John's Hopkins.

Earlier this year, PGDx announced that it had secured exclusive rights to technology being developed at Johns Hopkins for the analysis of cell-free ctDNA — in particular, a "personalized analysis of rearranged ends," or PARE, approach for assessing structural rearrangements in cancer.

That proprietary technique contributed to the John's Hopkins-led team's current analysis of ctDNA in blood samples from hundreds of individuals with more than a dozen tumor types and at various stages of disease.

For their study, the researchers interrogated tumor samples from 136 individuals with metastatic cancers originating in 14 tissue types, using targeted sequencing, whole-exome sequencing, whole-genome sequencing, and/or PARE to find somatic tumor mutations.

They also assessed samples from dozens more individuals whose tumors were deemed dangerous despite the lack of metastasis, including 41 individuals with primary glioma or glioblastoma brain tumors and 10 individuals with stage III ovarian or liver cancers.

The team took a tiered approach to unearthing the informative mutations that would ultimately be used to search for ctDNA in the blood. For instance, for cancers with a handful of common cancer contributors such as colorectal cancer, the researchers began with targeted gene sequencing. In other cases, they started by sequencing much larger sets of cancer-associated gene sets before moving on to exome- or whole-genome sequencing and PARE.

Once specific point mutations or rearrangements had been defined in the tumor tissues, the team came up with the proper primers for taking a PCR-based look at whether the tumor-specific glitches were present in DNA found in each patient's blood samples.

As reported in GenomeWeb Daily News last week, the researchers saw variability in individuals' levels of ctDNA that seemed to coincide with both tumor type and cancer stage.

In particular, the team was adept at detecting ctDNA in blood samples from individuals with pancreas, colon, breast, ovarian and other cancer. On the other hand, ctDNA detection tended to be far trickier in those with glioma or glioblastoma brain cancers or with metastatic kidney, prostate, or thyroid cancers.

"We were kind of lucky that we began working on ctDNA in colon cancer, in a way," Diaz said. "Imagine if we'd started in brain cancer? We would have said it didn't exist."

The clinical utility of ctDNA in different cancer types is expected to vary as a result of these differences in tumor nucleic acids in the blood.

At the moment, there appears to be a lack of promising non-invasive methods for detecting and/or following cancer types that shed little or no detectable DNA into the bloodstream, according to Diaz, particularly because findings from the current study suggest intact tumor cells are even trickier to see in patient blood samples.

He and his co-authors of that study also saw ctDNA in some patient blood samples that do not contain detectable circulating tumor cells. Alternatively, though, they did not pick up CTCs in samples without detectable ctDNA.

Still, Diaz said he was encouraged that ctDNA could be detected as often as it was in individuals whose tumors were still localized to their tissue of origin. When the team tested for ctDNA in blood samples from more than 200 individuals with localized cancers, it picked up blood-based tumor markers in 122 of the patients.

Again, ctDNA levels varied depending on the tumor type and stage, though the researchers were able to detect it in some 47 percent of stage I cases.

"We can detect almost half the tumors we tested — which actually span more than 14 tumor types — as early as stage I, when it's easily curable by surgery," Diaz argued. "This is the first large-scale study describing the sensitivity and specificity of this technology, not only for advanced disease but also for early disease."

Bardelli remains somewhat more circumspect about the possibility of doing early detection with ctDNA, at least for now.

"In principle, that would be the most efficacious way of using liquid biopsies because we would be able to do early diagnosis — and early diagnosis is the most effective way to cure the beast," he said. "I think we have a long way to go to use this for early diagnosis, but the potential is there."

He believes the most immediate application of this blood-borne genetic material may be for monitoring minimal residual disease and following treatment response and resistance in individuals already diagnosed with certain cancers.

"Patients undergo surgery and their oncologists and pathologists may think that the patient is cured — or that there's no disease," Bardelli explained, "but there's no form of proof, so they end up getting [chemotherapy] anyway."

"If we could show that ctDNA is a good marker of minimal residual disease, then maybe we would be able to spare some of the patients from this therapy," he said.

Similarly, blood markers of cancer could give clinicians a window into cancer recurrence, Bardelli said, noting that he and his team have found some preliminary evidence suggesting that colorectal cancer relapse might be detectable using ctDNA from blood samples even before recurrent tumors can be seen with conventional imaging methods.

The researchers are part of an ongoing effort to collect longitudinal blood samples from individuals with colorectal cancer, which has been underway since around 2008. They are also getting ready to kick off a clinical trial focused on a potential therapy for overcoming acquired resistance to the EGFR-targeting drugs, which will include ctDNA-based treatment monitoring.

"We think that plasma is much better than a single biopsy," Bardelli said.

"A patient with metastatic cancer will have multiple lesions," he explained. "So even if we can biopsy [metastatic tumor sample(s)] we'll never capture the full complexity of the disease."

Because the team is taking blood samples from patients every 15 days or so for their upcoming trial of a treatment for overcoming acquired resistance to cetuximab and panitumumab, Bardelli noted that it would not currently be cost effective to do exome sequencing of ctDNA and matched normal samples in that trial.

"While we realize the relevance of doing the exome [sequencing] study at the end, when patients relapse, we see no way that we could do this every 15 days — it's simply too costly," he said. "In the future, it would be ideal if we could do this."

Part of the cost comes from the frequency of sampling, but the depth of sequencing needed to deal with the relatively scarce ctDNA in plasma is another consideration.

In situations where particular genes are either prone to mutation or most likely to provide clinically useful information, Bardelli said PCR or other targeted technologies may be sufficient for finding a few key tumor-tracking mutations for some patients, even as sequencing prices dip.

In the case of colorectal cancer, for instance, roughly 80 percent of patients carry mutations affecting the APC gene, he noted, so that may be a useful marker for detecting cases of colon cancer early on.

On the other hand, next-generation sequencing or other methods for interrogating large stretches of tumor DNA would likely be useful in instances where less is known about a primary tumor or its metastases.

The extent to which a given tumor genome is interrogated —in biopsied samples and/or with cancer DNA from the plasma — is also apt to depend on the type of information being sought.

While a small handful of markers might be sufficient for seeing the presence or absence of residual disease in a patient, for example, more comprehensive analyses would be needed to characterize the cancer involved and, in some cases, to come up with possible treatment targets.

"Point mutations only tell one part of the story," Diaz said. "Structural changes, where you do need whole-genome sequencing-like approaches, tell another part of the story."

Along with their work on ctDNA and its potential applications in the clinic, the team is exploring the possibility of exploiting circulating DNA for non-cancerous conditions.