NEW YORK – Researchers from Oregon Health and Science University are piloting a unique method for blood-based cancer detection and mutation analysis that uses cancer patients' radiation treatments to boost tumor DNA levels for easier detection and sequencing.
The team is currently investigating the approach, called RAMP-Seq (radiation-assisted amplification sequencing), in the context of a specific radiation protocol called stereotactic body radiation therapy, or SBRT, which is standard of care for patients that present with lung nodules suspected to be cancerous but not easily biopsied or surgically removed.
Christopher Boniface, who is leading the effort, shared early results from a small group of patients at last month's American Association for Cancer Research Advances in Liquid Biopsy Meeting. Using serial blood draws, researchers showed that timing sampling to coincide with a small window post-SBRT could enrich circulating tumor DNA levels at least 10-fold, potentially raising otherwise undetectable levels to well within the range of standard sequencing methods.
Throughout the history of ctDNA test development, the fact that some cancers, especially early-stage tumors, may shed only minute quantities of DNA into blood or other body fluids, has become a universally recognized challenge.
There have been other attempts to induce tumors to shed more ctDNA into the blood — Luis Diaz and colleagues, for example, performed some early experiments subjecting individuals to physical exertion. However, most efforts to raise test sensitivity have been in the technical aspects of sequencing or in layering different biomarkers like copy number changes, DNA methylation, and fragment patterns over mutational signatures.
Boniface said in a recent interview that there had been some evidence that radiation could increase ctDNA levels. For example, in a separate study using liquid biopsy to track individualized ctDNA signatures in women with metastatic breast cancer, he and his colleagues, had observed a single patient in whom variant frequencies surged during a course of palliative radiation treatment.
"They were undergoing palliative radiation to a lesion in their neck … so we took the opportunity to sample … more frequently," Boniface said. "And we saw a dramatic, I think a 40-fold increase from baseline prior to radiation over the course of their palliative sessions."
"It also wasn't erratic," he added. "It was a nice linear trajectory up until where it peaked, and it may have even gone further than that [but] we weren't able to sample again until about 30 days later."
SBRT presented a unique and particularly promising test case for the group to try to explore this further because it involves higher-intensity radiation, but a much narrower focus, with beams directed precisely at the site of a tumor, with minimal exposure to surrounding normal tissues. As such, Boniface and colleagues set out to consent about 20 patients receiving SBRT for suspected non-small cell lung cancer.
In this context, the ability to detect and sequence ctDNA would not only offer potentially valuable genotyping of these suspected tumors but could also aid diagnosis, since not all the lesions treated in this way turn out to have been true malignancies. "They could be fungal masses, and I'm assuming other things," Boniface said. "I don't know as much about that early disease setting [but] I think the estimates are that … somewhere around 80 percent or 90 percent of the patients [receiving SBRT in this setting] actually turn out to have cancer."
The goal was both to study patterns in how ctDNA concentration changes in response to radiation, but also to explore an actual clinical use case, he added. "We [thought this] might be an opportunity for us to understand the biology … but we also had this cohort where the radiologists are saying this would be wonderful because we have no way of [doing genomic] testing [on] these folks"
At the AACR meeting, Boniface shared some early data from eight patients recruited to the study thus far. Using an error-corrected sequencing method they had developed previously, the researchers sequenced samples at a handful of timepoints in the days after SBRT treatment, aiming for 5,000x to 20,000x coverage. Overall, they found that radiation appeared to induce about a 25-fold average ctDNA enrichment.
This wasn't uniform across the eight patients, but every individual did show a measurable bump in ctDNA, some from virtually undetectable baseline levels.
Boniface said that although the data is preliminary, it suggests possible utility for a RAMP-seq approach in early detection and mutation profiling of masses that are difficult to biopsy or characterize.
"Potentially we can genotype or do mutational profiling on any tumor that we see on imaging if we can irradiate it," he said.
In lung cancer, because SBRT is already a standard-of-care treatment for early, suspicious masses, piggybacking ctDNA sequencing to push a tumor to spill its secrets presents few hurdles. But to explore the method in other early-detection settings, investigators would need to start from scratch to make a safety and cost/benefit case for irradiating patients in the service of this goal.
"We've had people argue both sides of this," Boniface said. "I'm not a radiologist, so I stay out of it, but I've heard plenty of people say [that] we can do this with a minimal amount of radiation."
"I think in the case of an inaccessible tumor where you have no other option, it's proved to be a resilient way to do a liquid biopsy, and I can see that … pushing the case in favor of allowing that sort of treatment," he added.
Other possible application areas might be esophageal tumors that can grow close to vulnerable parts of the central cardiovascular system, or potentially pancreatic cancers.
The group is also still looking at what this might be able to offer in the later-stage setting, having had approval now to proactively apply SBRT, as opposed to the palliative radiation that they first looked at by happenstance, in a few other metastatic breast cancer subjects.
In this vein, Boniface said, the hypothetical utility of RAMP-seq would be to induce the release of a more representative, more heterogeneous population of ctDNA into circulation. This is based on the team's observation that despite only one tumor site being irradiated, DNA representing clonal populations from other distant sites also went up in the blood of that first observed patient.
"Back in the 50s, somebody observed this thing called the abscopal effect … [which is] basically that when you hit one tumor with radiation you see a response in distal tumors as well. It's thought to be mediated through some sort of immune response, so if you're triggering antigens from one tumor, then the immune system gets trained and then goes out to similar signatures and the other tumor. But [regardless of the mechanism] we think we are essentially observing the same thing," Boniface said.
In the nearer term, he and his colleagues are now focused on extending beyond the eight patients that he reported on at the meeting. They are also hoping to be able to obtain biopsies and sequence-matched tumor tissue from at least some individuals to evaluate how their de novo ctDNA mutation calls compare.
"This is something that the field has been tackling for a while [and] something I've been personally trying to get my head around … [that] there isn't really a good gold standard," Boniface said. "But obviously when we get the biopsy we'll be able to confirm that the mutations we did call were real."
He added that the group's sequencing approach has been well tested across other cohorts, and the team is confident that its "fairly robust."
"We can confidently say [from testing] patients where we already know the mutations that they should have … that we catch about 80 to 95 percent of what we knew was there," he added.
Hopefully the study will also be able to shed some light on basic questions about the sources of cell-free DNA. Boniface said the team plans to try to tackle this using size profiling and potentially also epigenetic and fragmentomic tissue of origin methods.