NEW YORK – A new study published today has added to a growing body of evidence that searching the blood for a patient-specific panel of circulating tumor DNA can identify which early cancer patients — in this case breast cancer patients — are at risk for an impending recurrence and which are not.
The study has also provided new indication that the sensitivity of this type of residual disease detection benefits greatly from increasing the number of mutations targeted.
Researchers from the Broad Institute and Dana-Farber Cancer Institute published their findings today in Clinical Cancer Research, describing a method they developed to monitor individualized panels of up to hundreds of different cancer mutations in blood samples, and reported that, in a cohort of early breast cancer patients, their assays could detect cancer DNA in patient blood samples collected an average of 18 months, and up to three years, before metastatic recurrence was diagnosed.
The concept of creating bespoke, or personalized mutation panels to use for cancer recurrence detection and residual disease monitoring is not new at this point. Researchers began piloting these methods several years ago, initially tracking single mutations and then collections of alterations derived from tumor sequencing data. More recently, such methods have even begun to be commercialized, most notably by Natera, which launched its Signatera ctDNA monitoring assay service in 2017.
The Broad/Dana Farber group hopes to develop its own approach, ideally one that could greatly boost the sensitivity of this type of testing — in turn, lengthening the lead time in detecting, or predicting recurrence — by using potentially orders of magnitude more mutations.
"Our goal is to be able to turn patients who would have developed metastatic disease into patients who won't," the study's co-first author Heather Parsons, a Dana Farber medical oncologist and associated scientist at Broad, said in a statement. "In the future, if we can find those patients with residual cancer early enough, determine whether they would benefit from another course of therapy, and give them an effective additional treatment, we could potentially change the course of their disease," she added.
In an email this week, Viktor Adalsteinsson, the study's senior author and associate director of the Broad Institute's Gerstner Center for Cancer Diagnostics, said that because sequencing cost scales linearly with the number of mutations tracked, there is currently a significant bump in expense with turning to large panels as opposed to narrower methods.
However, he wrote the team is working on ways to drive down cost, for instance by adding methods to deplete the "massive excess of normal cell-free DNA [in] blood," so that their more intensive approach may still remain practical as they aim to advance to the clinic.
In their study the team retrospectively designed custom liquid biopsy assays for a cohort of 142 early breast cancer patients who had banked blood and tissue samples and had been diagnosed, treated, and monitored for disease recurrence over a 13-year period.
Across the group, patients had varying numbers of blood sample timepoints, but all had at least one immediate post-surgery sample and another taken roughly 12 months after surgery, when their initial treatment plan had been completed.
Starting with tumor tissue exome sequencing data investigators defined personalized panels of genetic mutations for each patient that they would then try to track in blood samples. The largest panel they were able to generate was 346 mutations, although the range was large —down to just two mutations for another individual, and about 57 on average across the cohort.
The researchers looked at how minimal residual disease (MRD) detection at a variety of timepoints corresponded to whether an individual was later diagnosed with a recurrence.
Notably, the liquid biopsy picked up signs of cancer in banked samples taken a year after surgery for six patients, all of whom were later diagnosed with metastatic recurrence, on average about 18 months, but in one case a full three years later.
According to the authors, although the results should be taken with caution because of the size of the current study, these lead times seem to outperform what has been seen in prior studies using narrower mutation panels.
Twenty-six other patients tested negative in their blood samples one year after surgery but still developed recurrence. Overall, though, the authors wrote that cancer in these patients tended to reappear on a much longer average timeline, as many as 10 years later. It is possible that for these longer-term recurrences, there may not be any ctDNA signals present at this early one-year testing timepoint, the group hypothesized.
The remaining patients in the cohort had no record of their disease returning.
According to Adalsteinsson and his colleagues, it was also clear that, as expected, they had a notably better chance of anticipating recurrence in cases where a patient's tumor biopsy yielded a larger number of mutations to search for in blood.
Before they set out to test the clinical samples in the study, the researchers first performed a set of dilution experiments to model how increasing the number of mutations being tracked might improve their sensitivity in detecting MRD. Compared to a single-mutation droplet digital PCR assay, which yielded detection down to a tumor fraction of 1 mutation per 1,000, the authors concluded that sequencing a panel of 97 mutations could boost sensitivity to 1 in 10,000, and 488 mutations would offer reliable detection down to 1 in 100,000.
However, most patients in the clinical cohort had a much smaller number of identifiable tumor mutations. The median in the group was 57 with a range of just 2 up to 346.
"Expectedly, clinical sensitivity was highest among patients in whom we tracked the most mutations," the authors wrote. Based on their calculations, the investigators projected that had they tracked fewer mutations per patient, they would have failed to detect MRD in 36 percent of the study samples that they could detect with their broader method.
And, reassuringly, using larger panels did not seem to result in any increase in false positives. "Tracking more mutations was not associated with increased MRD detection in the 105 patients who did not experience distant recurrence," the group reported.
However, even though they were able to detect MRD very sensitively, at least for the cases where they could glean large mutation panels from tumor tissue, the "tumor fractions in cfDNA of patients who experienced recurrence were [still] close to the detection limits." This suggests that it may be necessary to track even more mutations per patient in future studies, the authors wrote.
In this vein, Adalsteinsson and his colleagues wrote that they are now hoping to use whole-genome sequencing, rather than exome sequencing, to try to build even larger patient-specific panels for blood-based monitoring.
"If whole-genome sequencing were to be applied to each patient's tumor biopsy, we would expect to find several thousand mutations to track per patient in cfDNA," Adalsteinsson said in his email.
According to the study authors, if a prospective validation can confirm the lead times they saw for MRD prediction at a one year post-surgery timepoint in this initial study, their broad panel approach might offer an opportunity to detect and treat emerging recurrences long before the development of overt metastatic disease — potentially offering curative treatment to patients for whom these are not currently an option.
That said, there is not yet a clinical standard, or persuasive evidence that there are specific interventions that can improve outcomes for patients in whom an incipient recurrence is detected early. Such clinical utility remains to be proven.
In his email, Adalsteinsson wrote that he and his team are making plans for further prospective validation "and are open to opportunities for collaboration in these directions." However, their immediate next step is to push forward with an attempt to glean MRD tracking signatures into the thousands of mutations by starting with whole-genome sequencing in patients' tumor tissue.