How Much Clinical Evidence Does FDA Require to Support a Biomarker Claim?
- 13 hours ago
- 7 min read
By David Petrich, Landrich Group Co-Founder and VP of Quality and Regulatory
The Question Every CDx Team Eventually Has to Answer
How much clinical evidence is sufficient?
It’s a common question for clinical teams working on a Companion Diagnostic (CDx) or biomarker-based In Vitro Diagnostic (IVD). There is no simple, clear-cut rulebook that provides an answer.
CDx are at the frontiers of modern science, targeting a wide range of diseases and revolutionizing treatment for many forms of cancer. They play a pivotal role in a new generation of precision medicine, where drugs and diagnostics work hand in hand to target specific genes and proteins for more effective therapies.
Yet because science is so new, the regulatory environment struggles to keep pace with innovation in the drug-diagnostic development model. The FDA does have specific requirements for clinical performance in every drug-diagnostic submission. However, the evaluation specifics depend on many variables, including the biomarker, the intended use, the clinical setting, the therapeutic context, and the evidentiary standard that FDA has applied to similar claims.
This article provides an overview of the FDA’s broad framework for evaluating clinical evidence for biomarker claims. Even though there’s no perfect guidebook, the agency clearly defines factors that influence the evidence requirement. There’s also broad clarity on the practical questions that need to be addressed to design a clinical study supporting CDx submission.
Analytical Performance
FDA evaluates clinical evidence for IVD claims based on analytical performance and clinical performance studies. First, we’ll address analytical performance, which focuses on proving the device measures what it claims to measure in a manner that is accurate, reproducible, and consistent across the range of specimens and use conditions in the intended use. For biomarker claims, this typically requires limit of detection (LOD) and precision studies that show repeatability, reproducibility, accuracy/method comparison, and interference/cross-reactivity studies.
A commonly cited example of analytical evidence is the cobas BRAF v600 Mutation Test for Vemurafenib. This is a real-time PCR assay intended to qualitatively detect the BRAF V600E mutation in DNA extracted from melanoma tissue to help identify patients eligible for the precision cancer therapy known as Vemurafenib.
In its analytical validation, the test demonstrated an analytical sensitivity of 5% mutant allele when 125 ng of DNA was used. In a comparison study with bidirectional sequencing, it achieved 97.3% positive percent agreement (PPA), 84.6% negative percent agreement (NPA), and 90.9% overall agreement for BRAF V600E detection.
This is considered an illustrative example because the analytical evidence defined the BRAF gene mutation in terms of specific specimen type, mutation target, assay platform, detection capability, and precise intended use. Understanding these details is essential, since the difference between V600E and non-V600E mutations can determine patient prognosis and drug responsiveness.
| Analytical question | BRAF example |
1 | Does it detect the intended analyte? | BRAF V600E mutation in FFPE melanoma tissue |
2 | How little analyte can it detect? | 5% mutant allele fraction under the studied input condition |
3 | Does it agree with a comparator? | 97.3% PPA and 84.6% NPA versus bidirectional sequencing |
4 | Is the intended-use specimen specified? | Yes—DNA extracted from FFPE human melanoma tissue |
Clinical Performance
While analytical performance verifies device accuracy in measuring a biomarker, clinical performance establishes that the biomarker result is clinically meaningful. For a CDx, the central question is usually predictive. Is therapeutic benefit primarily or exclusively for patients identified by the test? FDA expects evidence that links the result generated by the candidate CDx to the relevant clinical outcome in the labeled population, treatment setting, and decision point.
In the strongest clinical study designs, that evidence demonstrates treatment-by-biomarker interaction. In real-world terms, this means patients with a biomarker-positive result received a different treatment than biomarker-negative patients. A simple association between a biomarker and outcome may establish prognosis. However, clinical performance must demonstrate a direct relationship with a particular therapy.
Let’s look at a few different examples to understand scenarios for clinical evidence in biomarkers. First, the case of BRAF V600E and vemurafenib, cited earlier for analytical evidence. The studies for this drug-diagnostic submission, known as the BRIM drug-diagnostic trials, enrolled 675 patients with previously untreated, unresectable, or metastatic melanoma whose tumors were BRAF V600E-positive by the cobas BRAF test. It compared Vemurafenib with the therapy Dacarbazine. The diagnostic was clinically validated in BRIM2 and BRIM3, as the assay identified tumors carrying BRAF V600E.
This is cited as a best practice for CDx submission because it is closely aligned with the FDA expectation for these dual measures of performance. The study provided an analytically defined test result, linking it to a therapeutic trial to identify the target population and show efficacy for the intended disease.
A second commonly cited example for clinical evidence is the BRACA Analysis CDx approved with Olaparib for patients with deleterious or suspected deleterious germline BRCA-mutated advanced ovarian cancer after three or more prior attempted chemotherapy treatments.
The BRACA Analysis was supported by an international multicenter study that enrolled 137 patients with measurable gBRCA-mutated ovarian cancer. The objective response rate was 34%, with a median response duration of eight months. The evidence package submitted to the FDA demonstrated the connection between a defined germline BRCA result and clinical treatment in the exact treatment population described in the labeled indication.
Both BRAF/Vemurafenib and BRCA/Olaparib illustrate biomarker-enriched development with clear clinical performance standards. The trials enrolled patients whose tumors or germline DNA carried the relevant mutation, then quantified the therapeutic benefit against a control or comparator within that biomarker-positive population.
These designs support treatment selection for biomarker-positive patients. The fact that these trials did not conclude biomarker-negative patients is not necessarily a major design problem for the FDA. For an essential CDx, a trial limited to biomarker-positive findings can be entirely appropriate. However, it does limit the scope of conclusions of the trial and evidence-supported claims. It can support a label limiting the use to the test-positive population for clinically proven therapies.
Other tests aim to cover both biomarker-positive and biomarker-negative patients. These are sometimes referred to as “all-comers”, in terms of trial design, because they address this diverse population. Additionally, they may try to determine relative degrees of effectiveness between positive and negative patients. In the IPASS trial, for example, gefitinib improved progression-free survival versus chemotherapy in EGFR mutation-positive NSCLC but performed worse than chemotherapy in EGFR mutation-negative disease.
What Drives the Evidence Requirement?
We’ve talked about the types of evidence the FDA expects. The other question is how much? Once again, it depends on several factors. The novelty of the biomarker itself is a key aspect of this determination. CDx has the deepest body of established evidence in certain genes related to cancer, dating back to the first CDx approval for Herceptin over twenty-five years ago. Biomarkers such as KRAS, EGFR, and HER2 have an extensive body of literature and precedent CDx approvals. Therefore, they will have a lower evidentiary requirement than a novel biomarker with limited published data. For novel biomarkers, the FDA may expect prospective clinical evidence generated specifically to support the claim, not just retrospective analysis of existing samples.
Another factor is the clinical consequence of an incorrect result. A biomarker used to select patients for a therapy where the alternative is a viable, effective treatment carries a lower false-positive risk tolerance than a biomarker used as the sole basis for prescribing a therapy with significant toxicity. FDA's evidence expectations scale with the clinical consequence of error.
The consequence of error changes the required evidence. In IPASS, EGFR-mutation-positive NSCLC favored gefitinib over chemotherapy. Patients whose tumors were EGFR-mutation-negative had significantly better progression-free survival with carboplatin/paclitaxel than with gefitinib. A false-positive EGFR result could therefore expose a patient to a less effective first-line therapy. A false-negative result could deny a patient a treatment with a greater likelihood of benefit. That is why FDA’s CDx co-development guidance emphasizes the risks of both false positives and false negatives when test results determine treatment selection.
A third consideration is the intended use population. A biomarker claim covering a broad population, all solid tumor patients, for example, requires evidence across that population. A claim limited to a specific histology or disease stage can be supported by a more targeted evidence set if the claim is stated accordingly.
Another important criterion is the therapeutic relationship. For CDx claims specifically, the FDA expects clinical evidence to support the intended use as labeled for the therapeutic. It should reflect the same patient population, the same line of therapy, and the same decision point. Evidence from a different clinical scenario may not apply, even with the same biomarker and the same drug.
What 'Sufficient Evidence' Looks Like in Practice
For an established biomarker with multiple prior CDx approvals, FDA has generally accepted:
Analytical performance studies demonstrating equivalence or superiority to a predicate or comparator method
Concordance data from archived clinical specimens (retrospective), with sample size powered to the sensitivity and specificity claims
Clinical outcome data from trials of the associated therapeutic, demonstrating that the biomarker-positive population achieves the labeled therapeutic benefit
Appropriate utilization of FDA harmonized consensus documents when available
Justification when study designs are altered for specific intended uses, sample types, products, and technologies.
For a novel biomarker or a novel intended use claim, FDA has generally expected:
Prospective clinical data specifically designed to validate the biomarker-treatment relationship
Pre-specified analysis plan filed before data lock
Samples and data collected prospectively from a representative clinical population
Evidence that the analytical method used in the clinical study is the same method being submitted for approval.
It is important to remember that, in addition to the examples cited here, it is critical to research similar CDx products on the FDA device listing database. This includes Guidance Documents and Summary of Safety and Effectiveness statements. The FDA maintains a list of approved CDx products. These resources, along with harmonized consensus documents, are essential tools for designing analytical and clinical performance studies.
The Importance of the Pre-Submission Meeting
For CDx programs with any complexity in the biomarker claim, a pre-submission meeting with the FDA is among the highest-value regulatory investments you can make. The FDA will give direct feedback on proposed clinical study designs, the adequacy of the planned evidence package, and the acceptability of the intended use claim as drafted.
The common mistake is treating the pre-submission meeting as a validation step ("we'll show the FDA what we're planning") rather than obtaining scientific input on proposed clinical study designs ("we need FDA's feedback before we design and conduct the study"). Pre-submission meetings are most valuable before the clinical study early in the product development lifecycle, before initiating clinical performance studies.
The Benefit of a Pro-Active Biomarker Strategy
At Landrich, we’ve learned there are significant returns to upfront research and planning for trial design and submission of a CDx or biomarker product. Risks of generating insufficient analytical and clinical evidence can be mitigated through a series of actions. Perform research on similar products. Align analytical and clinical performance study designs with harmonized consensus standards.
"Sufficient" clinical evidence for a biomarker claim is not a fixed standard. It's a function of the biomarker, the claim, the population, and the therapeutic context. The way to determine whether your planned evidence package is sufficient is to map it against the specific factors FDA uses to evaluate those dimensions, identify any gaps, and get regulatory feedback early to shape the study design and conduct with confidence.
