Life in the Fast Lane: Navigating a New Era in FDA's Accelerated Approval Program in Oncology
The agency's preference for randomized clinical trials over single arm studies can have major implications for clinical development programs and innovations in external controls and digital twins
The Accelerated Approval (AA) program, established by the United States Food and Drug Administration (FDA) in 1992, aims to expedite drug approvals for serious or life-threatening conditions. Initiated primarily in response to the HIV/AIDS crisis, the program has since expanded to cover other critical areas of therapeutic development, particularly oncology, which today represents about 85% of all accelerated approvals. By allowing faster approvals based on surrogate endpoints that are reasonably likely to predict clinical benefit, the AA program strives to improve patient outcomes by providing earlier access to promising new therapies (Code of Federal Regulations, Title 21).
In 2018, the FDA's Oncology Center of Excellence published a report in JAMA Oncology examining the AA program's impact on malignant hematology and oncology approvals between December 11, 1992 and May 31, 2017. During this period, 64 products received AA, resulting in 93 new indications, with the majority (87%) of endpoints based on overall response rate (ORR). Single-arm trial designs contributed data to most (72%) of the initial AA indications. Out of the 93 new indications, 51 (55%) fulfilled post-marketing requirements and verified their clinical benefits within a median of 3.4 years after receiving AA. Meanwhile, 37 (40%) had not completed confirmatory trials or verified clinical benefit and 5 (5%) of the indications were withdrawn from the market. In this report, the FDA argues that the AA program has been highly successful with only a small percentage failing to confirm clinical benefits after AA approval.
In a move that may be interpreted as a more cautious view on the success of the AA program in oncology, in March 2023 the FDA released a draft guidance document titled "Clinical Trial Considerations to Support Accelerated Approval of Oncology Therapeutics." The document signals a shift in FDA’s comfort level with the routine use of single-arm trials for accelerated approvals of oncology therapies towards greater use of randomized controlled clinical trials (RCTs). The guidance outlines several known limitations of single-arm trials, such as the small size of safety databases, challenges in attributing adverse events to the drug under study, and difficulties in interpreting common time-to-event efficacy endpoints. Furthermore, the FDA expresses concerns about the potential for drawing inaccurate conclusions regarding treatment effects and determining the contribution of each component of therapy in single-arms studies examining the safety and efficacy of combination therapies.
Almost concurrently with the release of the draft guidance document, the FDA published an article in the Journal of Clinical Oncology (JCO) called "Irreconcilable Differences: The Divorce Between Response Rates, Progression-Free Survival, and Overall Survival." This article emphasizes the weak connection between early efficacy endpoints, typically used in single-arm studies to support AA, and overall survival (OS). The authors point out that factors such as drug class, the natural history of the disease, and subsequent therapies play a significant role in determining the relationship between early efficacy endpoints and OS. The FDA backs up their assertions by citing RCTs involving non-Hodgkin lymphomas (NHLs) and chronic lymphocytic leukemia (CLL) that despite showing improvements in progression-free survival (PFS) have demonstrated potential negative effects on OS. Additionally, the FDA refers to three trials in recurrent ovarian cancer where an initial PFS advantage was later found to be associated with potential detriments in OS with longer follow-up. Furthermore, the authors mention several immunotherapy trials that displayed improvements in OS without corresponding gains in PFS and/or ORR. Remarkably, the FDA concedes in the article that many early intermediate endpoints, such as ORR and PFS, have not been officially established as surrogate endpoints for OS (despite the fact they have been routinely used not only in the AA program but also in full approvals and confirmatory studies).
The FDA’s announcement of the new guidance document mentions a 2022 article published in the New England Journal of Medicine (NEJM) titled "The On- and Off-Ramps of Oncology Accelerated Approval," where the FDA emphasizes the development of a comprehensive AA strategy in oncology addressing both "on-ramp" and "off-ramp" considerations. On-ramp considerations involve trial design, end points, study population, and data collection timelines. Off-ramp considerations include confirming clinical benefit for traditional approval or removing indications if clinical benefit is not verified. The FDA points to the fact that AA in oncology often uses durable ORR rate as the efficacy endpoint but again acknowledges that the relationship to OS varies across cancer types and drug classes. In this article, the FDA proposes alternative strategies for AA, including conducting a single RCT that could support AA and verify clinical benefit or running two concurrent (vs sequential) studies: a single-arm AA study and a randomized trial to demonstrate clinical benefit.
The FDA is undoubtedly placing a very strong emphasis on reducing the time between AA and verification of clinical benefit in confirmatory studies as an important measure of a successful AA pathway while making additional strategic maneuvers to articulate its evolving stance on the topic. This includes a recently launched initiative called Project Confirm, which strives to address several challenges in the AA process for oncology indications by focusing on enhancing transparency, striking a balance between early access to life-saving therapies and verification of clinical benefit via confirmatory studies, fostering collaboration and innovation among stakeholders, and supporting ongoing monitoring and evaluation of patient outcomes.
The FDA's guidance and the publications collectively speak to the complexities and critiques of the AA pathway, as well as the challenges of conducting timely confirmatory trials, with important implications for clinical development programs. Moreover, the FDA's growing preference for RCTs over single-arm studies within the AA program may have significant ramifications for organizations developing external controls and digital twins. External control arms and digital twins can, in theory, be employed in single-arm studies as comparative benchmarks for assessing the safety and efficacy of new therapies. However, it is not clear whether the FDA and other regulatory authorities will promote or call for exercising caution in using these innovations for tangible confirmation of clinical benefit. Regardless, it is important to keep in mind that appropriately validated and "regulatory grade" (a commonly used term with no standard definition) external controls and digital twin methodologies could potentially strengthen the FDA's case for broadening the use of randomization within the AA program.
In the post-market setting, external control arms and digital twins can be pivotal components of assessing the real-world safety and efficacy of approved therapies by providing a comparative basis for evaluation of treatment outcomes in real-world patient populations. However, assurance of model accuracy, which necessitates faithful representation of the complex and heterogeneous patient populations encountered in real-world settings, adds an additional layer of intricacy to this paradigm.
The FDA's evolving stance on the AA program highlights the need for a more rigorous and balanced approach to expediting drug approvals, one that can ensure the safety and efficacy of new therapies while promoting innovations that can advance the objectives. These dynamics underscore the importance of going beyond standard best practices to incorporate more innovative clinical trial designs, methodologies such as external control arms and digital twins, and comprehensive multidisciplinary strategies based on resources such as real-world data, to advance clinical development programs in the context of the AA program and beyond. As the landscape of drug development continues to evolve, success will increasingly depend on accurately predicting and appropriately adapting to the perpetually shifting realities of the healthcare system, the biomedical research enterprise, and the corresponding regulatory frameworks.