NEW YORK – The precision oncology community is hoping it could soon see a new landscape of drug trials using circulating tumor DNA as an early or surrogate endpoint for adjuvant therapies and other treatments in early-stage disease.
In a recent draft guidance, the US Food and Drug Administration said that while the evidence isn't there yet to support these types of approvals, it could be soon. Thus, stakeholders are hoping to be proactive in developing the right kind of foundational data.
The nonprofit Friends of Cancer Research has tasked itself with leading this charge, publishing a white paper at the end of last year that called for broad and dedicated collaboration.
During a meeting held by the organization last week, FDA Division Director Nicole Gormley said that as an early endpoint, ctDNA testing could operate similarly to measures like pathologic complete response or progression-free survival, offering a way to analyze patient responses at an interim time point without having to wait for overall survival data. Similarly, ctDNA negativity as determined by minimal residual disease assays could also prove itself as a surrogate endpoint for clinical response.
Both use cases could potentially support accelerated drug approval for early-stage and adjuvant cancer therapies, a pathway that, while not without controversy, has dramatically reduced the waiting period for metastatic patients in accessing new precision oncology drugs.
Foundation Medicine VP Geoff Oxnard highlighted the transformative impact this has had in late-stage cancers. "Patients can get commercial access to these therapies anywhere in the US and start benefiting … sometimes years before the full endpoint is available for confirmatory approval. … What we are talking about today is how we take this clearly potent learning from advanced cancer patients and apply it to this new challenging opportunity of impacting access to these therapies for earlier, curable cancer patients," he said.
FDA Medical Oncology Chief Julia Beaver stressed that the risk-benefit considerations in early-stage, curable disease — the potential harms resulting from a premature approval — are different than in the metastatic setting.
"I think it's really critical to have strong evidence for use of a novel endpoint like ctDNA in this space. That said, we recognize the imperative here and the unmet need here, and because the adjuvant setting is a potential curative setting, expediting access to novel therapies here would reach arguably the most patients and perhaps have a larger impact on cancer mortality," she said.
In addition, the FDA recognizes a "huge potential" for adjuvant drug development, Beaver added. "These adjuvant trials are powered for long-term outcomes that require large sample sizes, and they take a very long time to report out even disease-free survival [let alone] overall survival," she said.
"Having that earlier endpoint … could really cut years off of drug approval," she added. "And that's obviously the reason we're all here discussing this right now."
Participants in the meeting — from biopharma, diagnostic, regulatory, and academic sectors — agreed solidly that their efforts stand to benefit enormously from taking into consideration lessons learned in the successes and missteps of forebears like pathological complete response.
"I think that we learned a lot in trying then to take that endpoint to drug development in [breast cancer]," UPenn oncologist Angela DeMichele said. "What we've found over the course of the almost two decades that we've been evaluating this is that early on there was a lot of focus on testing drugs to see if they would improve pathologic complete response rate [pCR], but not powering the trials to ultimately be able to then link those early results to an outcome of interest."
She further noted that many drugs or regimens improved pCR, "but we couldn't go on to connect that to an improvement in event-free survival."
Another issue was the need to harmonize definitions of pCR. "When we went to do a meta-analysis to try to really understand the key relationship between [pCR] and event-free survival … it was very difficult to look at these trials in aggregate," DeMichele added.
Finally, she said, it also became clear that post-surgery treatment in some patients but not others confounded the ability to accurately measure the relationship between pCR and progression-free survival. Investigators realized they needed to come together as a group to standardize pCR methods and codevelop trial designs that would be powered for long-term clinical endpoints.
Although the breast cancer community has since seen success with neoadjuvant therapies coming to market on the back of pCR-endpoint studies, Oxnard said that the same issues have hampered similar success in other tumor types.
"In lung cancer there's been enthusiasm for a long time … but what's happened is there's not been as much adoption of the new agent paradigm. There's not been as much evidence generated," Oxnard said. "This year we had our first FDA-approved immunotherapy for the neoadjuvant treatment of non-small cell lung cancer … and while the pCR data about that trial was available more than a year prior [showing a] clear, measurable increase in pCR rates … it had not been qualified as an early endpoint for lung cancer. And it's not because of lack of enthusiasm. It's simply that the sort of systematic evidence we heard about in breast cancer hadn't been done."
"I think that's a lesson as we embark upon this for ctDNA, [to figure out] how we can be organized to make sure we can deliver on this opportunity for patients," Oxnard said.
In the group's white paper, experts convened by FOCR wrote that they believe broad collaboration will be key in validating the use of ctDNA as an early endpoint, a task that they said will require large amounts of data from multiple prospective clinical trials.
The FDA's Gormley said that there are two things the agency looks at when evaluating whether or not a particular endpoint can support a drug approval. For an assay-based endpoint, the first is whether the assay has been analytically validated.
Among key technical questions the FOCR group highlighted in its white paper is the need to establish minimum analytical performance requirements, which could differ if the treatment setting is neoadjuvant or adjuvant or based on the type of tumor.
Participants on the call also cited minimal residual disease in hematologic cancers as a source for inspiration and caution. The same questions asked in that space regarding optimizing thresholds for MRD negativity to best correlate with clinical benefit, how to time tests, and whether testing at multiple time points is needed are all directly relevant to ctDNA.
Meanwhile, various ctDNA assay platforms, both tumor-informed and fixed-content assays using both genetic and epigenetic markers have been developed and are being employed, and will potentially have to be harmonized.
The second consideration, Gormley said, is clinical validity. "Does that endpoint allow us to predict the clinical endpoint of interest?" She called it paramount that stakeholders work together, and that there be consistent collection of data and consistent use of assays across trials.
"We need to do the foundational work," DeMichele said, especially in designing current phase II studies well enough that they can then translate that into properly powered phase III trials.
"Coming together as a cancer community to define what those phase two trials look like and what kinds of data we need to generate would go a long way toward really being able to combine forces and bring this technology to our field much faster than if we're … doing it in a more piecemeal fashion," she added.
Overall, the FOCR authors expressed confidence that the field can make this happen, citing FOCR's previous collaborative efforts in the liquid biopsy space, such as the ctMoniTR project, whose first phase sought to confirm whether changes in ctDNA levels accurately reflect the therapeutic effect of immunotherapies in advanced lung cancer.
In step two of the project, investigators are already looking across more than 20 clinical trials and 16 different therapies to characterize whether changes in ctDNA are a prognostic indicator.
The FDA's guidance recognizes this growing evidence, but participants in the call all agreed that that this prognostic value will have to be borne out in prospective trial data to be considered truly predictive, or sufficiently predictive enough to serve as an early endpoint for accelerated approvals.
Alain Silk, senior director of regulatory affairs at Tempus, argued that evidentiary needs may evolve over time, so the framework for data collection should be adaptable going forward.
"I can see a couple of ways to do this," Silk said. "One might be to take a pre-planned, stepwise approach, either collecting across a broad diversity of trials and then focusing down on those areas where there's the most potential for near-term application. … Or conversely, you can think of initially focusing evidence generation on the most promising cases where you can efficiently generate information for additional analysis and then broadening out to demonstrate generalizability."
In his experience, the FDA has been able to apply notable regulatory flexibility in other settings, and "it's important to keep in mind that it's possible that the same information may not be needed for each case where ctDNA might be able to support regulatory decision-making as an early endpoint," he added.
During the call, multiple speakers suggested that establishing a correlation between ctDNA negativity and observed pCR could be a low-hanging fruit in terms of immediate next steps.
In the FDA draft guidance, the agency specifically said it encourages sponsors to "develop evidence regarding the usefulness of ctDNA response in addition to or supporting pathologic complete response information after neoadjuvant therapy."
Aadel Chaudhuri, assistant professor at Washington University in St. Louis, said he thinks more attention could definitely be paid in this vein. "The data in breast cancer is exciting, but I think there is potential in other tumor types as well," he said, citing work in his lab to correlate ctDNA with pCR in colorectal cancer and in bladder cancer.
"I would love to see that sort of framework used more as we can really generate data quickly," Chaudhuri said. "It will be, of course, extremely important to wait longer and correlate with progression-free survival and overall survival, but this is an early endpoint that we could be looking at more."