Defining Survival: FDA's New Requirements for Oncology Trial Design

Harpreet Singh, MD, former director of the FDA's Division of Oncology 2 and chief medical officer of Precision for Medicine, discusses the FDA's draft guidance on the use of overall survival (OS) as an end point in clinical trials.

The new draft guidance from the FDA outlines 3 key elements. Importantly, these elements do not fundamentally change established practices or expectations for sponsors; rather, they serve to clarify existing expectations regarding the collection and analysis of OS data.

1. Prespecified Planning and Long-Term Follow-up

The first element mandates prespecified planning. If OS is not the primary end point of a trial, the sponsor must include a detailed plan in the protocol for capturing long-term survival data.

In the past, sponsors sometimes ceased following a patient once they experienced a progression-free survival (PFS) event (ie, when their disease progressed). The FDA is now explicitly stating that this practice can no longer occur. Once a patient progresses and meets the PFS end point, they must be followed out for survival. This requirement represents an operational effort for sponsors but is considered crucial to rule out potential harm associated with the novel therapy.

2. Subgroups and Biomarkers

The second element addresses the analysis of subgroups and biomarkers. The FDA intends to look at clinical trial results differentially based on biologically plausible subgroups. For instance, in a study targeting a specific mutation (Biomarker A) that enrolls both patients with and without that biomarker, the agency will analyze both PFS and OS in both groups.

If the results show that patients lacking the biomarker do not derive the same survival benefit as those who possess it—and especially if the drug's toxicity poses too great a risk to the biomarker-negative group—the FDA may restrict the final indication to only those patients with the target biomarker. This policy emphasizes the importance of robust survival data, as it can ultimately determine the approved patient population, even if the overall study is considered "positive."

3. Trial Design Considerations

The third element involves trial design considerations. The guidance suggests that practices like crossover and unequal randomization may be potentially limited to avoid obscuring the true survival benefit of the experimental drug.

Singh notes that this topic warrants more robust discussion with regulatory authorities. Limiting or not offering crossover can make a trial less attractive to both patients and investigators, potentially hampering enrollment. Patients are generally more willing to enroll if they know they might receive the experimental drug upon progression, even if they are initially randomized to the control arm.

For the sake of patient-centric trial design and good practice, Singh argues that instead of limiting crossover, sponsors should collaborate with their in-house and FDA statisticians to prospectively account for expected crossover. This can be achieved by modeling the expected impact on OS and using appropriate mitigation strategies, a point that is suggested to require further discussion on a trial-by-trial basis.