Friends of Cancer Research at a meeting earlier this month presented new data from its ctMoniTR project that it believes may help validate the use of circulating-tumor DNA as an early endpoint in clinical trials for oncology therapies.
The project began more than five years ago with an initial proof-of-concept analysis combining various independent data sets to build evidence and understand the associations between ctDNA and response to treatment for solid cancers, Jeff Allen, president and CEO of Friends of Cancer Research (FOCR), said in an interview.
In hematologic cancers, ctDNA or minimal residual disease is further along in its acceptance as a surrogate endpoint. MRD is accepted as a surrogate endpoint for developing treatments for acute lymphoblastic leukemia and multiple myeloma. The US Food and Drug Administration has approved ALL drugs based on their impact on MRD, and the biomarker is getting closer to achieving surrogate endpoint status for expediting treatments to market for acute myeloid leukemia.
The FDA hasn’t yet approved a solid tumor treatment based on MRD, though drugmakers are exploring its use as an early endpoint in trials.
Wanting to contribute to the understanding of ctDNA as a predictive biomarker in the solid tumor setting, FOCR’s original proof-of-concept analysis drew on data from five relatively small clinical trials in lung cancer and showed a “pretty strong association” between reduction of ctDNA in the first several weeks following treatment initiation and patients’ long-term outcomes, Allen said. “What we wanted to try and do was at least explore the feasibility of what it would take methodologically to combine various different independently constructed datasets from different trials in order to build a more robust body of evidence,” he said.
After this first stage of the project, the organization wanted to characterize the potential utility of ctDNA further. It identified collaborators for a pilot project and a large number of trials with common parameters, namely the measurement of ctDNA at baseline, at least one early on-treatment ctDNA measurement, and availability of long-term outcomes.
In the last couple of years, the organization has been aggregating the data from two dozen clinical trials from 12 different sponsors, breaking it into three modules to “begin to answer the questions around what drug types might be able to be characterized by ctDNA change,” as well which ctDNA parameters could best predict long-term outcomes for patients with lung and other tumor types, Allen said.
The first module analyzed eight trials in lung cancer where 1,000 patients were treated with different tyrosine kinase inhibitors. In that module, the analysis showed that for patients who cleared ctDNA within the first several weeks after treatment, there was an improvement in overall survival, Allen said.
In the second module, the organization combined results from 3,000 lung cancer patients in three different trials who had been treated with either immunotherapy or chemotherapy and found that a significant reduction in ctDNA in the first 10 weeks correlated with improvements in outcomes.
The third module, data from which were presented at the FOCR meeting earlier this month, was “more exploratory” in that researchers looked at 10 trials covering patients with tumors in other organs than the lung and observed similar patterns, Allen said. In this module, the organization gathered more aggregated data than what could have been gleaned from these individual trials, “to try to more holistically characterize the role that ctDNA could play as a drug development tool moving forward,” he said.
Névine Zariffa, principal and founder of strategic consulting firm NMD Group, presented new data from the third module at the meeting. The module included patients with breast, cervical, colorectal, and prostate cancers, as well as patients with melanoma. These patients had received a variety of treatments, including TKIs and immunotherapy. Researchers looked for an individual-level association, or whether ctDNA levels correlated with patients’ prognosis for improved survival, and a trial-level association, or how well ctDNA levels correlated with the overall treatment effect, in other words, how precisely ctDNA levels predict treatment benefit across multiple studies.
On the individual patient level, there was an association between decreased ctDNA levels and improved overall survival. While there was not enough data to generalize broadly, Zariffa noted that it presented an “encouraging picture.”
The researchers then looked at the trial-level association between data from all the randomized controlled trials that had been collected in ctMoniTR thus far. In looking at the trial-level association, researchers considered whether the effect of treatment on ctDNA levels was associated with the effect of treatment on the clinical endpoint of interest, such as overall survival. FOCR found less robust trial-level association between ctDNA levels and overall survival.
Statistical analysis exploring how well ctDNA levels predicted survival endpoints produced less clear results, showing a weaker trial-level association between ctDNA and overall survival and a somewhat stronger association between ctDNA and progression-free survival. Zariffa noted that there might be weaker correlations since the analysis combined treatments and cancer types and drew data from few studies, some of which allowed patients to crossover from the comparator to the investigational treatment arm.
The organization is packaging the analysis within this third module for publication and putting a manuscript together to submit to an as-yet-undecided journal, Allen said.
These aggregations are meant to be an intermediate step in exploring the role ctDNA could play in clinical trials, according to Allen. “Starting essentially from scratch and then waiting until the trial’s completed … without having any indication that it would be a successful association seemed like a heavy lift,” he added. “This was always meant to be a bit of a stepping stone.”
According to Allen, now that this evidence has been developed, FOCR’s next goal is to facilitate prospective trials that explore the link between ctDNA and treatment outcomes. The organization needs to “start thinking about what the formal process should be in engaging the clinical community and the FDA, particularly around the establishment of this as a particular endpoint for certain situations?’,” he said.
A prospective trial would allow FOCR to correct for some of the limitations in its retrospective analysis, Allen noted. Oftentimes, ctDNA was used as a more exploratory endpoint in the trials FOCR looked at, so there were data missing from some patients that “has left some ambiguity we would want to correct for in a prospective trial,” he said.
“Embedding these measures in a common metric across multiple trials is a goal,” Allen added.
When drugmakers do test for ctDNA in drug trials they collect data at different treatment time points and use different types of assays with varying performance characteristics. Such differences “left questions as to what’s the right limit of detection or performance specs that would be necessary,” he said, suggesting that conducting a prospective trial will bring a certain level of uniformity in these aspects.
Even if FOCR wanted to further confirm its findings from the pilot project using archived samples, they’d need access to those. “While companies may be collecting samples from a good number of patients in a clinical trial, they’re oftentimes moving to freezers as opposed to being analyzed,” Allen said, hopeful that FOCR will be able to use some of the data and archived samples sponsors have already collected to confirm the results from the project thus far.
“This may be a good time to leverage some of these archived samples to make sure that we’re trying to do a very evidence-oriented approach to identifying some of these commonalities that could inform future trials and move this field along faster,” he said.
Thus far, the project has utilized a collaborative approach, and Allen hopes a “strong consortium-type” approach will continue to serve the organization well. Ideally, FOCR would want to identify multiple trials in a common therapeutic setting where companies would be willing to embed ctDNA collection in a more uniform, structured way.
According to previous US Food and Drug Administration guidance on validating a new endpoint, Allen said, there needs to be data coming from multiple different trials so that a proper meta-analysis can be done to evaluate the associations of that new biomarker with longer-term outcomes.
“For an endpoint like this, because of its applicability, not only FDA but the field would want to know that this is a tool that’s useful for multiple different trials, across numerous different companies and contexts into the future,” he said.