At a recent meeting hosted by the Friends of Cancer Research, experts from industry talked through next steps for development of companion diagnostics for rare diseases, including in oncology. The ideas come as FDA’s diagnostics regulatory reforms face an uncertain future with the new administration and agency staffing changes.
Introduction: FDA’s diagnostics regulatory reform and the impacts for companion diagnostics development
The FDA’s Laboratory Developed Tests (LDTs) rule: In May 2024, FDA finalized a long-expected rule that would enforce the medical device regulations over a subset of in vitro diagnostics (IVDs) known as LDTs. While the FDA has long maintained that LDTs are medical devices (in the U.S., IVDs are regulated as medical devices), it has historically expressed an “enforcement discretion” for these types of tests – in effect, not enforcing regular regulatory requirements – citing their low volume and relatively low patient risk. However, in the last decade the agency has been increasingly indicating concern that the risk profile of these tests has changed, and the 2024 final rule phases out the enforcement discretion approach over a five-stage, four-year process bringing LDT developers and distributors into compliance with the device regulations. [Read AgencyIQ’s full analysis of the final LDT rule here.]
The five stages of implementation: Stage 1 includes requirements for medical device reporting, reports of correction and removal, and the quality system (QS) requirements related to complaint files. Stage 2 includes the requirements that are neither QS nor premarket review but not addressed in Stage 1. These include registration and listing; labeling, including, where applicable, Unique Device Identifier (UDI) requirements; and investigational use requirements. Stage 3 is the QS requirements, excluding the parts required under Stage 1. Stages 4 and 5 are premarket submission requirements. Stage 4 covers high-risk products; by Stage 5 all manufacturers of IVDs currently offered as LDTs would need to submit a premarket application to FDA (unless they’re under a new enforcement discretion, see here).
The rule is a monumental shift for laboratory business units, but the impact of the regulatory shift extends beyond just laboratories. Drug developers have extensive questions about what the implementation of the final LDT rule means for drug development research programs. These include the enforcement of the investigational device regulations (21 CFR 812, also known as “Part 812”) for LDTs that meet the definition of “investigational IVDs” that are used in drug development. The FDA had previously issued guidance intended to assist different stakeholders in determining whether the tests they use in studies are investigational IVDs, but this is currently a gray area in regulation.
The LDT rule and the impacts for companion diagnostics (CDx): CDx are a type of IVD that are intended for use with a corresponding therapeutic drug, specifically to provide “information that is essential for the safe and effective use” of that therapeutic. A CDx can be an entirely new kind of IVD, a new version of an existing product or an existing product that has been previously authorized for a different indication. Ideally, the CDx and its corresponding therapeutic should be ready for authorization in tandem; in practice, however, that has been extremely challenging.
Therefore, the FDA allows for some therapeutics that would require a CDx to be authorized in advance of the CDx. This flexibility, formally established in FDA guidance from 2014, is typically used for drug products intended to be used in more targeted patient populations, including precision medicine products. While the drug is approved and the CDx is not yet authorized, LDTs fill the gap – a flexibility that would be closed under the full implementation of the rule, in which LDTs would need to seek market access authorization from the FDA to effectively become authorized CDx.
The approval of treatment products without the authorization of their associated CDx have become such an issue that the FDA launched a related pilot program in 2023 – the output of which highlighted the serious issues developers face. This program sought a workaround so that certain new oncology drugs could be approved with labeling that describes the performance specifications for LDTs being used in lieu of a CDx. This would ultimately serve to standardize the performance expectations for these LDTs. To do this, the pilot program was intended to allow the drug developer to provide information about the version of the IVD used during the pivotal clinical trial (the clinical trial assay, or CTA), which the FDA would in turn use to “recommend minimum analytical performance characteristics for other tests.”
However, a key output of the pilot was that the drug developers simply do not have sufficient information about the CTAs. COURTNEY LIAS, head of the Center for Diagnostic and Radiological Health’s (CDRH) diagnostics office (OHT7) put it this way at a speaking event in 2024: “We’ve noticed … that it has been difficult for drug manufacturers to get information on the clinical trial assays that they’re using” and that “a lot of clinical trial assay laboratories are unwilling to provide information on the test performance for the clinical trial assays, and that makes it difficult for us to accept them into a pilot.” In short, it seems that the drug developers can’t get enough information on their CTAs to be able to put together a packet for submission to the FDA in the pilot that would eventually let them put that information on labeling.
The FDA has long acknowledged the challenges of CDx development, including disparate expectations and timelines in drug and test development, the issues that developers face in understanding if a CDx is needed (and during what stage of development), and situations in which a test is updated during trials, which impacts their readiness for authorization. The FDA has recommended that any test that will become a CDx “should be analytically validated,” “locked down before using in a clinical trial,” and generally not updated during the trial. However, given the unknowns about a still-investigational therapeutic product, it can be difficult to finalize an IVD before it’s used in a clinical trial. Further, the way clinical trials are designed can lead to roadblocks for CDx development, especially concerns about “bias from prescreening” where early screening of the trial population for eligibility will, by definition, limit the representativeness of the data generated on an investigational CDx.
With the well-established challenges in CDx development still in place and the LDT rule implementation underway, what options do developers have?
In February 2025, the Friends of Cancer Research (FOCR) hosted a meeting on the future of diagnostics regulation. This meeting focused on new ideas in the diagnostics space, including using novel data sources, such as digital pathology and artificial intelligence (AI), and next steps for those seeking to develop CDx for rare diseases.
A quick note: No FDA staff were present at the meeting. While CDRH Director MICHELLE TARVER was expected to be the keynote speaker, and several regulators from the agency were on the working groups that presented at the meeting, the FDA was under a communications freeze at the time, so no agency staff from FDA were able to present or attend in person.
At the meeting, MEGAN DOYLE of Eli Lilly presented the work of the FOCR’s Rare Biomarker Working Group, which issued an accompanying paper on the validation of biomarkers for CDx in rare disease populations, including in oncology. As the working group wrote in the paper, “Especially for rare biomarkers and indications affecting a small subset of the population diagnosed with cancer …, there can be inherent challenges for validating assays as companion diagnostics (CDx), such as difficulty obtaining sufficient quantities of well-characterized samples and limited established reference materials.” Doyle said at the meeting that the co-development approach is “an idealized scenario; in many cases, th[e] timelines don’t align this cleanly.”
The paper calls for “flexible validation approaches” within the existing diagnostics frameworks. The paper doesn’t propose an alteration of the diagnostics frameworks themselves but instead identifies specific areas within the diagnostics review and authorization pathways in which flexibility could be found and identified for rare disease CDx. The working group defined a series of variables that should be considered as to whether regulatory flexibility is warranted, including biomarker and indication prevalence, sample availability, and unmet needs and expedited review pathways.
The key is for therapeutic sponsors to prospectively plan for CDx development during their trials while also justifying using flexible approaches in sourcing samples and conducting validation (analytical, clinical) studies for the diagnostic, according to the Rare Biomarker Working Group’s proposals. The topline conclusions from the group’s paper are that drug development for cancers with rare biomarkers continues to expand, which calls for a more consistent approach from the FDA on how to consider these flexibilities, and increased collaboration between the drug and device sponsors to “establish proactive plans for preserving samples from all phases of clinical trials” and build out a repertoire of established validated reference materials and methods. The paper also includes a “proposed snapshot” that sponsors can use when seeking to justify their approaches to regulators. The snapshot is a template that outlines the specifics of sponsors’ approach and why it was undertaken.
The size of the population being considered is a key aspect informing whether flexibility may be needed. The paper proposes that to qualify as a “rare” cancer of biomarker, “these flexibilities be considered for biomarker-defined subsets of cancer types with an estimated prevalence of 1% or lower (in the population of patients with that specific cancer type in the U.S.) or for rare cancer types with an estimated total prevalence of 1% or lower (in the overall population of patients with cancer in the U.S.).” However, the working group noted that prevalence or incidence can change over time, including as technology advances or as screening becomes more available. Doyle said at the meeting: “We had a lot of conversation about this [definition] in the working group … Ultimately, where we landed was to try to really just define some examples of where we know it would be challenging to obtain the samples you need to conduct validation of the test.” As such, this isn’t supposed to be “a strict threshold, but really to identify places where we know there’s going to be a challenge in having enough samples to conduct the validation,” she explained.
Similarly, the paper cites a litany of challenges with sample availability. This is influenced by the complexity of sample types (for example, locations from which samples are acquired and potential degradation) and study design, which may exhaust clinical samples before CDx validation is completed. The authors wrote that developers face challenges with both availability and sufficiency of samples. One scenario is a study design that involves prescreening patients. If samples are exhausted (or are insufficient for validation), this raises the question of what to do, especially given ethical questions about obtaining new or additional samples – for example, conducting an additional biopsy on a cancer patient “for the sole or primary purpose of supporting analytical testing.” Finally, the paper includes a short note on the timing of the drug and device pathways, including when the drug application proceeds via the (condensed) Accelerated Approval pathway, which “can be challenging, particularly within the expedited timelines of FDA’s drug development programs.”
The paper walks through the challenges that typical clinical trial practices can present for CDx development. These challenges include access to clinical trial samples – especially when there are differences between enrollment tests, local tests and the candidate CDx – and when bridging studies may face additional complexities, for example, if the sample method used for a candidate CDx is different than that used by a local test for the trial. The paper recommends that clinical trial sponsors “prospectively plan for storing archival tissues or nucleic acids from the pivotal clinical trial” and collect appropriate consent from research participants to support CDx development. However, the paper also notes that CDx development needs biomarker-negative samples, which “may not be included in the target trial design due to the lack of anticipated effect in patients without the biomarker, raising ethical questions about the enrollment of biomarker-negative patients.” In these situations, the paper supports active development plans that actively consider the need for biomarker-negative specimens from patients who aren’t enrolled in the trial or “well-characterized negative samples from related studies or normal healthy donors from related studies or normal healthy donors for blood based biomarkers” – although disease staging may add complexity here.
Other data sources and their possibilities for validation: Clinical approaches, real-world evidence (RWE), procured human specimens, cell lines, contrived samples, and in silico datasets. The paper walks through a series of different sources of samples or data that could be used in validation, including trial samples from related studies or routine testing, RWE from “a diagnostic that could be developed as a CDx and [patients] who have also received a therapy of interest” (in effect, an LDT), human specimens from a vendor or data repository, contrived samples (such as spike-in) and in silico data sets. Each of these data sources could contribute to CDx development, including bridging studies (to a CTA, for example), or analytical or clinical validation, but they also all come with regulatory and practical challenges. The paper from the working group includes a chart recapping these approaches. As the Rare Biomarker Working Group members noted in their panel discussion at the FOCR meeting, the matrix is intended to provide a hierarchy in which developers may need to prioritize the types of things that they would be “willing to spend clinical samples on, but then we would establish a priority for when to use certain samples,” LabCorp Oncology’s JENNIFER DICKEY said.
When there’s a dearth of clinical samples, when should sponsors prioritize their use and when should other sample types be considered – and on what activities? When developing the snapshot, “I think the way that we really thought about it was holistically so that in combination, when you’re thinking about what samples you have available, given that some are prioritized for clinical testing, how do you allocate the rest, and how do you assemble other sources of information and data to supplement what it is you have?” Tempus AI’s ALAN SILK said. For example, cell line materials can “provide a way to access biomarkers,” he explained, but sponsors will need to consider “how well do they represent the patient samples, and how well do they need to represent the patient samples in order to be used in that study,” Silk said.
Analysis and what’s next
The challenges that face CDx and drug developers flagged in the paper may cascade going forward. FOCR published a similar paper in 2021, but as panelists at the meeting noted, the area has evolved rapidly. “I think we’ve witnessed some amazing strides in cancer drug development that are leaning towards more and more biomarker recognition within cancer subtypes and development of drugs to target those smaller populations,” Foundation Medicine’s ELIZABETH MANSFIELD said. While “we’ve seen FDA be somewhat open to using more alternative strategies for validation…, I think we can expect this to happen more and more often,” she said. This emphasizes the need for consistency in approach and best practices developed by – and distributed among – both researchers and regulators.
As methods keep emerging, “we emphasize the need to have an ongoing discussion with the agency,” Dickey said, and “how we’ve got to devise a validation strategy and then defend it to the agency either in the pre-submission process, ideally … before you’ve spent millions of dollars on studies and in the marketing submission, at least.” The tables cited above in the paper are intended to help with that: “Our intent was not to create a document that everyone has to use by any means, but something that could help streamline those conversations, provide the information to the agency upfront about your strategy and your rationale and your justification, so that you can have a productive conversation with the agency,” Doyle said. Still, these remain templates to serve in productive conversations, not a unilateral path forward.
Challenges with CDx validation aren’t just a diagnostics developer problem: “From the therapeutic sponsor perspective, I think we have to balance and juggle a lot of competing priorities in this space,” Doyle said. These include the regulatory and evidentiary needs of diagnostic validation but also patient-centric trials for areas with high unmet need. Because of the smaller patient population and the potentially expedited timeline, “we may have to do things like enroll on local tests” or “go where the patients are,” she said, while minimizing re-biopsying patients. With global development, “we may have to develop multiple companion diagnostics for different geographies, depending on what platforms are available in that geography, in the labs there,” she added.
As noted above, FDA has guidance on CDx development that was finalized in 2014. In that guidance, FDA wrote: “Ideally, a therapeutic product and its corresponding IVD companion diagnostic device should be developed contemporaneously, with the clinical performance and clinical significance of the IVD companion diagnostic device established using data from the clinical development program of the corresponding therapeutic product. However, FDA recognizes there may be cases when contemporaneous development may not be possible.” The guidance therefore provides the flexibility that would allow a drug to be authorized in advance of its CDx, which will likely need to be adjusted under the provisions of the LDT rule.
In 2016, the agency put out a draft guidance on CDx co-development. That document has never been finalized. The document was intended to be a “practical guide to assist therapeutic product sponsors and IVD sponsors” working on their CDx. It points to many of the challenges the FOCR working group also discusses in its paper, including changes in CTA design along the way and data generation and test validation method issues. However, as the FOCR paper and panel noted, these issues persist and have not been comprehensively addressed in the nearly 10 years since the draft guidance was (and more than 10 years since the 2014 CDx guidance was finalized), even as the role of CDx continues to grow.
The overarching advice for researchers continues to be: Do what you can to plan ahead, including trying to parse out where and when clinical samples will be available. Then prioritize their use for CDx validation purposes in a way that doesn’t infringe upon patient care – and get creative with the rest. Still, there is not yet a consistent regulatory strategy available to developers, nor best practices or expectations for those designing and deploying trials, as evidenced by FDA’s still-in-draft-format 2016 co-development guidance.
And, of course, the diagnostics regulatory trajectory is in flux in the U.S.: The FDA’s LDT rule and its implementation put an enhanced focus on the need for clearer guidance from the FDA about what sponsors – of both drugs and diagnostics – should do to better support co-development processes whenever possible. With LDTs filling the testing gap, there are also increased questions about where the line is for CDx as a requirement for use, and for what populations. See, for example, AgencyIQ’s recent analysis on a meeting about DPD testing for chemotherapy administration. All of this will be brought to the forefront should FDA’s LDT rule go forward. Still, there aren’t yet consistent answers to the questions of how studies to validate CDx should be designed that account for the practical complexities of fitting a diagnostics validation program into a drug trial and the small populations for rare disease drugs.
The future of that rule remains in question under the new administration, especially as the first Trump presidency saw a disagreement between the White House and the FDA about the regulatory status of LDTs. The FDA’s acting commissioner is former diagnostics regulator SARA BRENNER, who has not yet spoken publicly on her view on the rule but is now likely to take a higher-profile role in any future diagnostics reform policies in her elevated position (principal deputy commissioner, acting FDA commissioner).
What will happen with the LDT rule going forward is still not clear, even with leadership that has comprehensive background in diagnostics regulation at the FDA, but researchers and developers should watch closely. Even if the LDT rule doesn’t get implemented as finalized, the paradigm for testing associated with the safe use of specific drugs continues to be a point of complexity for both drug and diagnostic developers – and will remain a key issue for researchers seeking to design programs that can support both products in increasingly targeted populations.
https://home.agencyiq.com/article/00000194-f62f-d6bc-afdd-ff3f8a3c0000
