P301: FDA Biomarker Qualification Program: Analysis of Review Issues Underpinning Qualification Success Factors and Lessons Learned
Poster Presenter
Susan Chong
Senior Manager, Global Regulatory Policy and Intelligence
Amgen United States
Objectives
Analyze biomarkers that FDA qualified in BQP in 2025 and the key review issues. Reviewed the LOIs and QPs that were not accepted into the program between 2023-2025 to observe patterns for progression as opposed to non-acceptance and translating them into recommendations.
Method
Publicly available FDA determination letters for LOIs and QPs were systematically reviewed and thematically analyzed using AI. Findings were independently verified against source documents.
Results
As of March 3, 2026, there are a total of 108 programs in FDA’s Biomarker Qualification Program (BQP). Preliminary analysis identified recurring FDA review themes across the three biomarkers qualified in 2025 and non-acceptance decisions for Letters of Intent (LOIs) and Qualification Plans (QPs) between 2023-2025.
For the three biomarkers that were fully qualified in 2025, FDA’s key considerations broadly centered on whether the biomarker’s performance and evidentiary base were sufficient for the intended setting. FDA noted that the main issues generalized to threshold/score-dependent behavior that could affect reproducibility and interpretability, and limitations in dataset representativeness, to uncertainty around clinical meaningfulness of the proposed endpoints (e.g., whether the totality of evidence adequately supported the proposed context of use (COU), and concern that biomarker changes may not reliably predict clinical benefit). Time from LOI submission to full qualification for the three fully qualified biomarkers averaged 8 years 4 months (median 9 years 7 months).
Across the 10 LOIs not accepted (2023–2025) , high-level drivers included lack of a clear drug development need in the stated diagnostic area and COU-related deficiencies (e.g., COU was misaligned with study design or unlikely to influence regulatory decision-making). For QPs that were not accepted, a prominent reason was failure to adequately address prior FDA feedback; for example, in one case, FDA declined to accept the QP after the sponsor indicated plans to submit new/updated protocols, and FDA emphasized that prior comments from an earlier determination remained applicable and should be fully addressed in an amended QP with updated protocols to enable an efficient review.
Conclusion
When appropriately validated and with robust evidence, drug development tools, such as biomarkers, may enable earlier, more efficient decision-making across the development lifecycle. FDA’s BQP provides a pathway to establish a biomarker for a specified COU, making lessons from qualification successes and non-acceptance decisions informative for sponsors seeking to develop other types of biomarkers.
Across the three biomarkers fully qualified in 2025, FDA’s review emphasis consistently focused on whether analytic performance and the evidentiary base were sufficient for the proposed COU, with recurring concerns related to threshold/score-dependent behavior (impacting reproducibility and interpretability), dataset representativeness, and the clinical meaningfulness of biomarker-defined endpoints. For these qualified biomarkers, time from LOI submission to qualification averaged 8 years 4 months, underscoring challenges with resource commitment, the risk of program attrition, and the long development horizon required to generate fit-for-purpose evidence. This may also suggest the need for a clear regulatory framework.
In contrast, LOI and QP non-acceptance decisions (2023–2025) often reflected an unclear drug development need, misalignment between the stated COU and the planned study design, and insufficient incorporation of prior FDA feedback, particularly at the QP stage.
These findings suggest sponsors may improve the likelihood of progression by clearly articulating the regulatory decision the biomarker will support, tightly aligning COU with study design and representative datasets, proactively mitigating threshold/scoring-related performance risks, and fully addressing FDA comments before advancing to the next BQP stage. As noted in FDA determination letters, “not accept” decisions are not final and can support iterative refinement of the biomarker and COU as additional evidence is generated.
