NEW YORK – Newly published data have provided a glimpse at the performance of Guardant Health's noninvasive and tissue-agnostic test, Guardant Reveal, for minimal residual disease detection in colorectal cancer patients. The results, though limited, suggest the assay has sensitivity and specificity comparable to competitive products in a field that is still in its earliest stages, but also the subject of increasing clinical and economic attention.
Published last week in Clinical Cancer Research, the study was spearheaded by clinicians and researchers at Massachusetts General Hospital Cancer Center, using blood samples collected over a period of several years from patients with stage I-IV colorectal cancer after curative intent therapy — either surgery or adjuvant therapy.
The concept that circulating tumor DNA could indicate residual disease in early-stage solid tumor patients has accelerated rapidly in recent years from a scientific hypothesis to the commercialization of clinical tests. In colorectal cancer especially, clinicians have begun to embrace the approach as a tool to assess patients' likelihood of being cured after their initial treatment and to direct additional adjuvant therapy for those who show signs of lingering disease.
"The only way we can cure solid tumors is by surgically researching them [early]," Ryan Corcoran, the study's senior author and a gastrointestinal oncologist at Massachusetts General Hospital, said this week. "But unfortunately, it doesn't matter what tumor type or what stage, some percentage of patients, even if a surgery goes well, will recur, and often die of their disease."
However, Corcoran added, "if we had a test that could tell us early who still has a disease or who doesn't … we would have the ability to give additional treatment to salvage the chance of cure. … Or, for patients with a high likelihood of being cured, they could avoid unnecessary and potentially toxic therapy."
Hoping to bring this dream to fruition, Corcoran said he and his team have evaluated both in-house ctDNA detection methods and have worked with a variety of commercial platforms. "We actually collaborate with almost every company who's developing one of these technologies to try to figure out the best way to do this," he said.
"We were intrigued by this specific assay … [because] unlike other devices which are tumor-informed, meaning that they require tumor sequencing, … this assay is blood-only," he added. The most prominent tumor-informed player in the space currently is Natera, which has been offering its Signatera personalized MRD testing for several years already and received a local coverage determination for use of the technology in CRC from Medicare contractor Palmetto GBA late last year.
According to Corcoran, the emergence of a blood-only option has raised the question of whether this simpler approach can detect residual disease as effectively as tumor-informed, personalized methods.
According to Corcoran, his team didn't necessarily expect that the results would be comparable, but in this initial set of patients, they did indeed see that Guardant's test was "very much in line with the numbers that other both academic and commercial tumor-informed assays have produced in terms of sensitivity and specificity."
The study cohort, 103 patients in total, was recruited and followed for several years, with blood samples collected one month after patients completed treatment and then at several surveillance timepoints afterward. Investigators then used Guardant's test on these samples, comparing the results to patients' clinical outcomes for the 84 in whom testing was successful.
Corcoran and his colleagues reported that among a subset of 15 individuals with at least one year of clinical follow-up, the test demonstrated 100 percent specificity. In other words, all the patients with a positive ctDNA test on their primary post-treatment sample recurred.
Sensitivity, meanwhile, was about 56 percent overall and up to 63 percent for stage II and III tumors. This falls well short of perfect prediction but matches closely with what other genomic technologies being advanced for this niche have claimed, at least for single-timepoint sampling.
Tissue-informed MRD tests, like Natera's Signatera, have shown higher sensitivity with longitudinal monitoring of ctDNA, but so did Guardant's. When samples from additional time points up to 4 months before recurrence were included, sensitivity improved to 91 percent, the MGH study authors reported.
The researchers also assessed how two versions of Guardant's technology performed, one looking only at mutations and the other incorporating DNA methylation data, as is used in the company's current commercial assay. The authors concluded that the addition of methylation improved the assay's performance by about 35 percent.
In the wake of the publication, Guardant's commercial competitors have sought to downplay the MGH results, highlighting the small size of the study and the heterogeneity of the cohort.
"I think one of the limitations that we discussed in our paper was this is not a pure population of only stage II or III. It is a mix of all four stages," Corcoran contended. But he said what unifies the group is that the study harmonized them by testing all patients after completion of definitive therapy, whether that was surgery or adjuvant treatment.
"I think it's a really important question that maybe we only detected those recurrences sensitively in one stage or not others," he added. "But actually, in one of our supplemental tables we do a multivariate analysis where we show that the hazard ratio … really doesn't differ across the different stages."
Critiques of the data have also focused on the high failure rate of Guardant's technology. Among 103 patients recruited to the cohort, only 84 were successfully tested.
During a call this week discussing the company's quarterly financial results, Guardant's president, AmirAli Talasaz, addressed this, saying that the sample size for the MGH cohort was not ideal, and that in its clinical lab, the firm is now seeing a less than 1 percent failure rate.
Talasaz said Guardant recognizes that results from larger studies, which it is now conducting, will be necessary to solidly support use and reimbursement of the Reveal test.
"Is this something which would convince all major national payers to cover this test? No. If we thought that was the case, we wouldn't be doing the clinical utility studies that we have been working on for almost two years," he said.
Although the company is seeing "exciting signals" in current payer conversations, "we think the readout of those [clinical utility] studies will be what impacts long-term adoption and reimbursement," he added.
The field of blood-based MRD testing is expected to expand, but most other entrants have focused or are focusing on tissue-informed platforms, including Natera, Inivata, and most recently, Personalis, which said recently that it plans to launch an MRD test later this year that will combine tumor-informed content and a fixed gene panel.
During that firm's Q1 earnings call this week, Personalis CEO John West argued that tumor-agnostic approaches like Guardant's are inefficient compared to tumor-informed assays.
"When you sequence the DNA from the plasma … you might only have one mutation out of every million bases across the genome. [That] means that [for] something like 99.9 percent of the sequence reads, you don't have any useful information. They don't catch a variant."
"We think [tumor-informed] is the only way to achieve the kind of sensitivity that we're targeting," West said.
According to Corcoran, data is still very much emerging for the clinical paradigm for MRD as a whole and individual oncologists are not yet in a place to weigh specific technologies against one another.
In general, though, the high specificity of ctDNA, regardless of the specific technology, has meant that the road into more widespread clinical practice is being laid rapidly.
"Obviously, we want both sensitivity and specificity to be high, but one of the reasons that we actually think these tests are even usable today to guide certain treatment decisions is really because the specificity is so high and so reliable," he said.
"Yes, we are missing some patients who are eventually going to recur … and that's unfortunate. … But if the test is positive, it's a virtual certainty that that patient has residual disease and that they're going to recur."
"We can actually feel comfortable to what we call 'escalate therapy' … and not worry that we're taking a patient who is cured and giving them unnecessary and harmful treatment, because the last thing we want to do is to cause harm in that situation," Corcoran said.
At MGH, he added, clinicians have begun to use ctDNA MRD testing "cautiously and responsibly in very specific settings where we think the data supports their use as a potential adjunct to help clinical decision-making for adjuvant therapy."
"We really do need some of the larger prospective randomized studies that are ongoing right now to tell us best how to use and act on these tests, but there are specific scenarios where I think you can make an argument [for cautious implementation now]," he said.