Clinical trial simulations and predictive modeling will be important for improving regulatory decisions about oncology drugs showing strong efficacy on early endpoints but with uncertain overall survival benefits.

Panelists at the recent Friends of Cancer Research Annual Meeting discussed a new white paper on the use of interim overall survival data in oncology trials. The FOCR project builds on a 2023 FDA workshop to discuss issues in clinical trials where OS is a secondary endpoint and how to proceed in cases where the primary endpoint, such as progression-free survival, is clearly positive, but early OS data may show no difference or trend in the wrong direction.

FDA and industry representatives on the panel agreed that simulations may be useful in the trial design stage to help lay the groundwork for appropriate interpretation of early OS results. They also said predictive modeling techniques could help analyze results once the trial is complete.

Division of Hematologic Malignancies II Director Nicole Gormley said the goal is to have a “more rigorous assessment” of OS as a safety endpoint.

“Overall survival has always been a key endpoint that we have cared about in oncology,” Gormley noted. “In recent years, we have had just such improvement with our therapies that we have gotten to a point where we can’t rely on it as a primary endpoint.”

She highlighted multiple myeloma as an example, where “it is now thought that patients that are newly diagnosed will have an overall survival approaching 10 to 15 years … As a result, we can’t rely on overall survival in that disease as a primary endpoint.”

“That being said, we always look at overall survival because of its importance and its role in having confidence in the risk-benefit of any given product,” Gormley added.

But now that often means a very haphazard approach, she said.

“So that really prompted us … to look at overall survival as a safety endpoint,” Gormley said. “Are there ways that we can pre-specify the assessment of overall survival from a safety perspective that would provide meaningful information when we’re using earlier endpoints to make a regulatory decision?”

Lisa Rodriquez, deputy division director in the Division of Biometrics IX, said an appendix to the white paper that “talked about some approaches to doing simulations and evaluating the hazard ratio in a variety of circumstances” was important.

Simulations allow sponsors to consider “what we’re able to evaluate in terms of overall survival at different information fractions, perhaps in different scenarios, like with or without crossover, in different disease areas and different expected levels of maturity by the end of the study,” Rodriguez said.

Thinking about “how are you defining harm in each of these situations” can in turn help determine when “you might do interim analyses” and to be able to pre-specify the degree of certainty in the OS data that “might change your conclusions,” she added.

Patient advocate David Mitchell also said bringing an OS measurement into the discussion is not necessary if it is truly not meaningful in the context of a potential accelerated approval decision.

“If a drug can be approved based on surrogate endpoints and the OS just mixes everything up, then don’t use it,” he said.

Mitchell said that he does not support ignoring OS trends in the context of adverse event signals or known problems in the class. But “I don’t want it introduced if it is not useful.”

“If it is just confounding, or it is weak enough that we shouldn’t draw conclusions from it, then it doesn’t belong in the discussion at the time of accelerated approval is sought,” he said. “Leave it out.”

Gormley said the issue is complex and includes considerations of the disease context, prior therapies and totality of the data for the product.

“That information really should be considered on a case-by-case basis,” she said “Those are really complex discussions so I would encourage sponsors to come in and engage with us.”

https://insights.citeline.com/pink-sheet/r-and-d/clinical-trials/us-fda-wants-more-rigor-in-assessing-survival-as-an-oncology-safety-endpoint-5MDKJECT2NFVDPEYAPRPRLG3XI/