NEW YORK – Researchers stopped a Phase III trial testing GlaxoSmithKline's PARP inhibitor Zejula (niraparib) as an adjuvant treatment of early-stage breast cancer due to low detection of circulating tumor DNA (ctDNA) in patients after their initial treatment, investigators said Friday at the San Antonio Breast Cancer Symposium.
The ZEST trial was designed to randomize patients with stage I to stage III triple-negative or BRCA1/2-positive breast cancer to either adjuvant Zejula or placebo, if they tested positive for ctDNA after the end of their neoadjuvant therapy and surgery but showed no signs of radiographic recurrence. Patients received ctDNA testing every two to three months, beginning after their initial treatment.
Nicholas Turner, director of clinical research and development at the Royal Marsden Hospital and Institute of Cancer Research, London, who presented the trial data at the symposium, said that the plan had been to enroll and randomize 800 patients into ZEST. However, out of more than 1,900 individuals screened for ctDNA, only 147 tested positive at the end of their initial treatment. Turner's group ended up randomizing 40 ctDNA-positive patients to Zejula or placebo, not enough to tease out a statistically meaningful difference in median disease-free survival. Because of this, they revised the primary endpoints in the trial to safety and tolerability.
The ZEST clinical trial is among a growing collection of studies designed to accrue participants based on lingering cancer signals detected via liquid biopsy testing. Turner estimated that up to 30 percent of patients treated for early-stage breast cancer will relapse. Unfortunately, oncologists don't yet have a reliable tool to detect the persistence or reemergence of these cancers, and thus, they lack the ability to intervene early.
In recent years, various ctDNA technologies have advanced in the clinic, with several now widely available for detection of minimal residual, or recurrent, disease (MRD). In ZEST, Turner and colleagues used Natera's tumor-informed, personalized assay system Signatera, which involves upfront tumor tissue sequencing, data from which is used to develop a bespoke panel of up to 16 genomic variants, which is then used to monitor a patient's blood samples for recurrence.
Although Turner and colleagues ended up only randomizing 40 out of the 147 ctDNA-positive patients, they reported a median recurrence-free survival of 11.4 months in the Zejula arm and 5.4 months in the placebo arm.
And not only was the ctDNA detection rate low, but "of the patients who were ctDNA-positive, 50 percent [already] had a radiologic recurrence when they were detected" using ctDNA, said Turner. According to Turner, ctDNA detection rates were highest in patients who had their ctDNA test within three months of the end of treatment.
The use of ctDNA testing to support clinical trials has increasingly captured the attention of drugmakers who are advancing adjuvant breast cancer treatments. The US Food and Drug Administration recently finalized a guidance document on how ctDNA could be used in drug trials. In it, the agency endorsed using MRD tests, but recommended sponsors make sure ctDNA assays have high sensitivity and negative-predictive value if they are going to use them to de-escalate treatment and have high specificity and positive-predictive value if they are going to use them to escalate treatment.
Turner acknowledged during his presentation at SABCS that a more sensitive ctDNA assay could potentially have identified disease recurrence in patients earlier in the ZEST trial.
In the study, many patients simultaneously tested positive for ctDNA and had a radiographic recurrence, and 63 percent of patients had evidence of recurrence on their first test. Forty-four percent of patients had a radiographic recurrence evident at their second visit.
"Some of these patients had a very high level of ctDNA [on their positive test], even though their test was negative only approximately three months earlier," Turner said. "This really tells us something about the pace of disease in these patients, and therefore, the difficulty of trying to intercept that with ctDNA."
In his view, ZEST's failure may be attributed, in part, to the broad entry criteria for the trial, which allowed low-risk patients to participate, and hence, may have added to the difficulty of accruing ctDNA-positive cases.
"Lots of low-risk patients still have the fear of recurrence and really wanted to get on the study, and that's really good in terms of how we would want to use these tests in clinical practice," Turner said. "But it's a real problem for the study if you have a big cohort of low-risk patients who aren't going to convert to being ctDNA positive, so you don't end up with enough patients to randomize."
This illustrates the challenge researchers face in studying technologies like ctDNA testing in clinical trials designed to reflect real-world practice. "We do have this tension in clinical trials between how we might want to use [a test] in clinical practice, but in a trial, we need a much higher-risk patient population," Turner said.
Meanwhile, the relatively high rate of metastatic disease at the time of ctDNA detection was also a problem in the study.
Among triple-negative breast cancer patients, who comprised the majority of the cohort, ctDNA was most often detected on the first test, with 60 percent of patients having ctDNA detected within the first six months after the end of treatment. Turner said that this is consistent with the known clinical presentation of these tumors, as they have a tendency to recur much more swiftly than other breast cancer subtypes.
"It can be so fast that it's actually difficult to capture it and intervene with a ctDNA assay," he said.
Because it has been shown that patients who are ctDNA positive after their neoadjuvant chemotherapy and before surgery have a very poor prognosis, it would make sense in the future to shift trial designs to start ctDNA testing earlier in the treatment process and use that earlier positivity to direct treatment.
Breast cancer is also known to shed less DNA into the bloodstream than other tumor types. As such, assay sensitivity could also have played a role, and Turner noted that there is "continual evolution" of commercial assays toward higher sensitivity.
Meanwhile, MRD tests like Signatera are already available clinically, and oncologists have reported that patients are increasingly aware of them via direct marketing and sometimes ask for testing on their own accord.
Future trials like ZEST will have to adapt, taking into account technological advances and market realities, to recruit the patients needed to actually test the therapies they are designed for.
"We've shown here that ctDNA can pick up nearly all of the people who are going to relapse, but we don't really know yet how to use that information to guide therapy and whether identifying it early can improve outcomes," Turner said. "At the moment, if you do one of these tests, and you find ctDNA, but you can't see anything on the scan, we don't actually know how to use that information to treat those people."
A positive ctDNA test when there is no evidence of radiological progression can be difficult for patients to understand, too, but lessons from trials like ZEST are helping the field better understand how to use ctDNA testing in precision oncology. "We're on the way to identifying the clinical situations where, hopefully, we can use these tests to improve outcomes in the future," Turner said.
There are multiple studies underway, he pointed out, including in other breast cancer subtypes like estrogen receptor-positive breast cancer, where it's likely much easier to intercept recurrence because the pace of disease is slower. "We are really hopeful that those studies will come through," Turner added.