Optimization and validation are needed to achieve ctDNA’s highest-priority application: surrogate endpoint for early cancer trials
There’s a compelling case for validating reduction in circulating tumor DNA (ctDNA) as a surrogate endpoint in solid tumor studies. Done right, ctDNA could shrink trials and accelerate development by providing an earlier — and potentially more sensitive — signal of clinical benefit than radiographic imaging. But in the settings where ctDNA would matter most, the work needed to validate it has barely begun.
At a Friends of Cancer Research meeting on Modernizing Oncology Endpoints this month, former government officials and industry R&D leaders outlined both the promise of ctDNA-based endpoints and the obstacles to making them reliable.
Circulating tumor DNA (ctDNA) measures DNA fragments released into the blood from tumor cells when they break down, either naturally or due to therapeutic intervention.
In general, the greater the tumor burden, the higher the concentration of ctDNA in the blood. When therapy is working, the hypothesis is that ctDNA concentration should fall. But other factors related to tumor or treatment class can also impact the amount of ctDNA shed.
Biopharmas already use ctDNA in several ways: to identify patients with targetable mutations, to inform early go/no-go decisions as an exploratory endpoint, and in screening approaches for early cancer detection. Longer term, proponents envision ctDNA as a real-world monitoring tool — a way to detect response and resistance earlier than imaging and adjust therapy sooner.
“As we move to longer maintenance therapies, where we are asking people to be on drugs for months and months and months, if you had earlier detectors of progression of disease before radiographic evidence, you could save people a tremendous amount, not only monetarily but more importantly in terms of toxicity,” Richard Pazdur, former director of FDA’s Center for Drugs Evaluation and Research (CDER), said at the meeting.
“That’s circulating tumor DNA,” Janet Woodcock, also a former CDER director, added. “It could be like HIV, where you use it to monitor treatment, to change treatment, and you can use it in chronic disease to monitor what you should do. We should be all over circulating tumor DNA, but it’s expensive. You can’t just do it by bake sales.”
Panelists agreed that ctDNA’s most consequential use would be as a surrogate endpoint, but the evidentiary package needed to show ctDNA reduction or clearance reliably predicts patient outcomes has not yet been generated.
The ctMoniTR project, led by Friends of Cancer Research, is laying the groundwork for evidence generation. It has performed retrospective analyses on over 20 advanced solid tumor trials to establish a correlation between ctDNA reduction and survival. Those results generally support the hypothesis that ctDNA dynamics can be predictive, while also exposing a limitation: across trials, ctDNA is measured at different time points, with different assays and thresholds, making it difficult to compare findings or pool data.
“The reason that prospective trials are so important is that in the pilot studies, we’ve had to overcome a degree of missing data.”
So far, ctMoniTR has focused on advanced cancer, where ctDNA levels are typically higher and assays don’t need to be as sensitive. It’s also a setting where conventional imaging and survival endpoints are often adequate, and shorter survival times allow trials to read out faster.
The bigger opportunity — and the bigger gap — is in early-stage disease. In that setting, validating evidence is almost completely lacking. As patients live longer after surgery or first-line therapy, the field increasingly needs endpoints that can credibly predict event-free survival; without them, development timelines stretch, and in some settings incentives for innovation weaken.
An early cancer focus raises the bar for assay sensitivity because complete ctDNA clearance will likely be the standard required for predicting positive long-term outcomes.
Validating ctDNA clearance as a surrogate endpoint in early-stage disease will take more than retrospective correlations. The field will need to improve assay technologies and identify optimal assay parameters, define standards and embed the standardized measures in prospective trials to support meta-analyses.
Doing all that will require funding and biopharma buy-in.
Setting standards
Friends of Cancer Research is calling on biopharmas to run prospective trials with standardized ctDNA protocols, but first the field needs to agree on what “standardized” means, from collection timing and sample handling to assay choice and reporting thresholds.
President and CEO Jeff Allen told BioCentury that ctMoniTR’s retrospective analyses were hampered by inconsistent ctDNA collection schedules and assay-to-assay variability. Different tests “have different performance characteristics,” he said, which makes meta-analysis difficult and “limits some of the analyses that can be performed, including some of those that FDA would typically request for endpoint validation.”
Even so, Allen said the project is still “seeing a strong association with ctDNA reduction and improvements in overall survival,” despite the heterogeneity across studies.
To sort out the issue of timing, Allen pointed to a resource already available: frozen plasma samples collected during clinical trials.
“There is a fair amount of plasma collection at various time points in a given trial, and as the details are being worked out around optimizing the assay or optimizing timing, a lot of these samples are essentially being stored in freezers,” he said.
Allen said analyzing banked plasma could generate correlative data to identify which collection time points best track with outcomes in specific tumor types and treatment contexts.
The next step, he said, is bringing biopharmas together to embed the aligned ctDNA measurements in prospective trials.
“The reason that prospective trials are so important is that in the pilot studies, we’ve had to overcome a degree of missing data,” Allen told BioCentury.
Assay standardization
Once timing is optimized, identifying the appropriate assay parameters will also be required. It’s possible that it could vary by treatment setting.
Two main ctDNA assay approaches are used as trial endpoints. The first, tumor-informed assays, sequence a patient’s tumor to identify individualized variants, then track those variants in blood. The second, tumor-agnostic assays, are off-the-shelf tests that look for ctDNA using predefined panels or signatures.
Because tumor-agnostic assays cast a wider net, they generally haven’t matched the sensitivity of tumor-informed tests. But they don’t require a matched tumor sample from biopsy or surgery — tissue that isn’t available for many patients — and they’re typically simpler and less expensive to run.
The problem is that only tumor-informed assays reach the sensitivity threshold that stakeholders are converging on as a standard: one part-per-million.
In the high-priority early-disease setting, panelists at last week’s meeting agreed that tumor-informed assays are preferred for now, based on their higher sensitivity and the need for confidence in ctDNA clearance.
But they acknowledged tissue availability will be a problem.
“Tissue is not always available. It takes work to find and it takes work to achieve the tumor cellularity and the quality of DNA that we need as input,” Minetta Lu, CMO of Oncology at Natera Inc. (NASDAQ:NTRA) said at the meeting. “It’s different than what we have typically done in clinical practice for 20 years,” she addd.
FDA has moved toward eliminating unnecessary biopsies in clinical trials to protect patients, making tissue less available for research. For many early-stage cancers, surgery sparing is becoming the goal to improve quality of life, which also makes tissue samples less accessible.
Liu told attendees that while tumor-informed assays are the focus, “if tissue is not available, you need to have a solution for that patient. You need to be able to reflex to a tissue-agnostic assay or a blood-only assay if necessary.”
Even if the sensitivity of tumor-agnostic approaches doesn’t improve in the near-term, Allen said the approach may be appropriate for advanced tumors where ctDNA levels are higher and the goal is reduction rather than clearance.
Heterogeneity adds complexity
The push to validate ctDNA for surrogate endpoint measurements follows the successful validation of measurable residual disease (MRD) negativity as a surrogate endpoint for multiple myeloma trials.
Establishing ctDNA endpoints will be more complex because the applications span many tumor types and clinical settings.
Multiple myeloma is a single tumor type, where the goal was establishing no trace of residual disease in the bone marrow, based on an established sensitivity threshold. After a meeting of FDA’s Oncologic Drugs Advisory Committee (ODAC) to discuss MRD-negativity in 2024, the agency released draft guidance on the endpoint this January.
By contrast, ctDNA shedding — and the relationship between ctDNA dynamics and outcomes — varies across solid tumors, which complicates threshold analyses.
“In ctMoniTR, we were in the advanced disease stage looking at a reduction of 90% of ctDNA levels following treatment,” Allen told BioCentury. “One could imagine implementing a similar approach in a prospective manner.” However, he noted “that becomes challenging because you have different cancers that have been shown to shed different amounts of ctDNA.”
One approach that has drawn interest in early-stage solid tumors, he said, is focusing on MRD-negativity in the form of ctDNA clearance rather than percent reduction, because “no detectable disease is a good thing no matter what the tumor type.”
Allen added, “From our vantage point, it’s really about trying to put together that framework of what it’s going to take to implement a validation study for the use of ctDNA. I think that commonality hopefully could be applied to different therapeutic contexts, ultimately so that not each context has to go back to start and begin validation from the beginning.”
Appetite for endpoint innovation
Identifying earlier endpoints that can reliably predict clinical outcomes becomes more important as cancer patients live longer, and as using overall survival as the primary yardstick becomes slower, costlier and harder to interpret. But it’s not yet clear how receptive current FDA leadership will be to new surrogate endpoints, particularly in oncology.
Over the last year, FDA actions and public comments have raised questions about the accessibility of accelerated approval and fueled speculation that the agency will demand more clinical outcomes data — and rely less on response rates and other intermediate measures.
At a meeting with patient advocacy groups in October 2025, then-CDER Director George Tidmarsh said the agency was reviewing the endpoints used to support accelerated approvals in oncology, signaling a possible shift away from response rate and toward outcomes such as overall survival. Those comments echoed positions Vinay Prasad advanced prior to his appointment as FDA’s chief medical and scientific officer and CBER director.
Where ctDNA fits into that framework remains unsettled: it’s not yet clear whether regulators will view ctDNA dynamics as a more credible predictor of benefit than objective response rate or progression-free survival.
Outside oncology, signs of FDA’s stance toward accelerated approvals suggest a rising bar. The agency recently issued a complete response letter to RegexBio Inc. (NASDAQ::RGNX) for its MPS II gene therapy RGX-121, citing issues that included the company’s use of an arguably well-validated surrogate endpoint, CSF heparan sulfate.
Still, FDA’s January 2026 draft guidance on minimal residual disease (MRD) and complete response in multiple myeloma is an encouraging sign that the agency remains open, in at least some contexts, to biomarker-driven endpoints when supported by an adequate evidentiary package.
“The good news is that there’s sort of a playbook to follow with the Oncology Drugs Advisory Committee and multiple myeloma. We’re hoping to build on that precedent,” Allen told BioCentury.
He noted that increased workload at FDA, given reduction in staff, could limit its ability to engage in these types of projects. He also suggested that helping FDA secure additional resources to support development program-agnostic tools is something Friends of Cancer Research is considering.
“That really could benefit across the board and ultimately expedite the development of new medicines for the people who need them,” Allen said.