NEW YORK – Although liquid biopsy — mostly circulating tumor DNA assays — is increasingly being investigated for early-stage cancer, the jury remains very much out on whether it can actually improve patient outcomes by guiding the right individuals to adjuvant therapies or helping them avoid unnecessary treatments.
However, a report published earlier this month in Nature Cancer has added important new evidence that this might be possible, in at least some settings.
In the study, investigators from a number of institutions used a circulating tumor DNA sequencing assay, Capp-seq, to test blood sample from a contrived cohort combining patients who had post-operative chemoradiation with those also receiving consolidation immune checkpoint inhibition therapy.
Among other observations, the study found that patients with ctDNA evidence of minimal residual disease (MRD) after chemoradiation therapy who then received consolidation immune checkpoint inhibition (ICI) had significantly better outcomes than patients with signs of MRD who did not receive consolidation ICI therapy.
In contrast patients with undetectable ctDNA after chemoradiation therapy had what the authors called "excellent outcomes" whether or not they received consolidation ICI. Importantly though, one of these patients who showed complete clearance of ctDNA also suffered a significant adverse event — pneumonitis — related to the consolidation ICI.
According to authors, the results as a whole provide evidence for the potential utility of using ctDNA tests to define which patients should get immunotherapy treatment, and which might safely avoid it and its potential complications.
Maximillian Diehn, associated professor of radiation oncology at Stanford University and a senior author of the paper, said that in general the field is waiting for prospective trial results to prove more definitively that ctDNA-based MRD assessment can identify patients that might benefit from IO or other adjuvant therapies. However, his team strove to provide a preview of what the field can reasonably expect.
Researchers are examining a variety of MRD assessment methods, including patient-specific assays derived from whole-genome or exome tumor tissue data, like Capp-seq, as well as broader approaches that look for epigenetic signals.
Diehn said he suspects that a variety of approaches will eventually be used, each with pros and cons.
"Our Capp-seq method appears to be very sensitive and to have excellent clinical performance in these locally advanced lung cancer patients, where with a panel design we can find a significant number of mutations [in] the average patient," he said. However, applying the same method in sarcoma, for example, would likely fail, because these tumors don't show many recurrent mutations.
"Maybe in the sarcoma setting you'd want a personalized approach, a bespoke approach, which … we've also done some of in the past. The downside of that [though] is that one has to perform whole-genome sequencing, which actually is not trivial to do from clinical specimens. There is a significant subset of patients where [biopsy] tumor content is too low to get a reliable exome sequence," Diehn added.
Methylation assays, he argued, are "intriguing," but don't have as much sensitivity and specificity data in the MRD context, so it may take more time and data to know if and how they can play a role.
In a blog post discussing the new publication on the Nature Research Cancer Community site this month, Everett Moding, the study's first author and a Stanford radiation oncology resident, wrote that with more than 20 studies having already demonstrated negative prognostic potential for ctDNA MRD, the "natural inclination would be to add … treatment in hopes of improving outcomes."
Whether this can actually improve outcomes has been an ongoing area of debate. Those "choosing to view the glass half full," Moding wrote, have pointed to data from the hematological cancer space, where it is established that escalating therapy improves outcomes in patients with MRD, as well as other recent studies showing chemotherapy can decrease and sometimes eliminate (at least temporarily) ctDNA MRD in patients with solid tumors.
In contrast, the "glass half empty view" has been that patients with lingering ctDNA may have levels of residual disease that are too high for them to benefit from additional therapy, and that it is actually the patients who are missed by ctDNA MRD assays for whom adjuvant/consolidation therapy would be effective.
"We wanted to try to generate some evidence [for the glass-half full view] that further treatment in patients who have ctDNA MRD could potentially improve outcomes. No prior study that been able to show that … so this is a [first] step." Diehn explained.
To do so, the research team built what Diehn and Moding call an "in silico clinical trial."
February 2018 saw the FDA approval of consolidation immunotherapy with the anti-PD1 immunotherapy durvalumab (AstraZeneca's Imfinzi) for non-small cell lung cancer, and this provided the team the opportunity to retrospectively assemble two groups of patients: one treated before the regulatory approval with just chemoradiation, and another from after the approval who also received the durvalumab regimen.
Using their CAPP-Seq panel on samples from these two groups, the researchers could measure if and how outcomes differed between patients who had detectable ctDNA MRD after chemoradiation based on whether or not they received consolidation immunotherapy or not.
Diehn described the findings as several distinct groups. Patients who showed no detectable ctDNA after chemoradiation made up one group, and had significantly better outcomes than the others whether they had additional immunotherapy or not.
"In the discussion we do some extrapolation to [calculate] how many patients … given the benefit of immunotherapy in this setting, we would need to treat to benefit one ctDNA-negative individual. And it is well over 100 patients before we could help one … because the majority of those patients seem to not have disease left," Diehn said.
"Of course, this is preliminary data and its an extrapolation. But if that were to be validated, that means it may not make sense to treat those patients because these immunotherapy drugs, as great as they are, have a rate of bad toxicities in the single digit percent range. So if you have to treat 100 to get one to benefit but … one or two is going to have a terrible toxicity, is that really worth it?" he added.
Meanwhile, for the subset of patients who had signs of ctDNA after chemoradiation, the "in silico" cohort comparison showed a clear benefit from added immunotherapy.
According to Diehn, while the evidence does not replace the need for prospective trials, it does "strongly indicate that the glass is likely to be half full."
In other words, he added, the study doesn't definitively answer the question of whether ctDNA MRD can serve as a new tool for precision oncology in the adjuvant setting, but suggests its utility, at least for this lung cancer indication and potentially others like it.
"Right now, for many adjuvant therapies, we are basically committing patients to one or two years of treatment, blind to whether or not they need the treatment or whether the treatment is doing anything for them," Diehn said. "That of course, is not precision medicine."
Importantly, the analysis also showed evidence for a third group of patients within the overall ctDNA-positive group: those whose ctDNA levels increased in blood draws taken in the early weeks of their immunotherapy treatment.
Compared to those whose ctDNA levels dropped, these individuals had significantly worse outcomes, potentially indicating a need for (and hopefully a benefit from) even greater adjuvant therapy acceleration.
Diehn said that this question of on-treatment ctDNA measurements is something that will have to be studied much more. In the nearer term, results like these have begun to attract the attention of pharmaceutical firms for their potential to accelerate adjuvant drug development.
"Currently, doing studies now in the adjuvant setting we have to wait years for survival data, because unlike the advanced setting where we can use progression-free survival and response as surrogate endpoints, we don't have surrogate endpoints in the adjuvant setting because there's no disease to see on the scans," Diehn said.
"Prospective data will bear that out. Either [ctDNA] is going to be a good surrogate or it's not. My strong suspicion is that it will be a good surrogate. But we'll have to wait and see. If that's the case, then that really could bring effective drugs to patients much faster."
Having such drugs available will also be necessary if there is ever to be a case made for on-treatment testing in clinical practice, he added. "It only makes sense routinely in the clinic if there's something else you can switch to."
Diehn and his collborators are now formulatingtheir own prospective trials based on the recently published results, though said he declined to discuss specific plans.