At a recent annual meeting, Friends of Cancer Research released three white papers based on discussions between FDA, NCI, industry, and academia:
- Interpreting data from dose-finding studies in early phase oncology trials to determine the optimal dose
- Incorporating pragmatic elements in study designs to enhance oncology randomized clinical trials
- Maximizing data from academic-led studies for regulatory decision-making
[This meeting] is the opportunity to bring together scientists from various different sectors, advocates, policymakers, researchers, in order to determine and identify potentially action-oriented steps,” Jeff Allen, president and CEO of Friends of Cancer Research, said at the Nov. 14 annual meeting. “A lot of work has gone into bringing these ideas together.”
The recommendations presented by the working groups come at a time when FDA and NIH—with a renewed focus on data infrastructure by newly confirmed NIH Director Monica Bertagnolli—are collaborating to transform the way clinical trials are conducted, increase enrollment, and build more complete and interoperable data sets (The Cancer Letter, Dec. 8, 2023).
“You’re operating with incomplete data constantly. And we do pretty well because our clinicians are dedicated and dedicated to their patients,” Bertagnolli said at the Nov. 14 meeting. “But we could do so much better with better evidence. And I want to get the kind of evidence that helps those doctors and patients to make their decisions together.”
Dose optimization
Drug sponsors are being directed to define optimal dosing in early-phase clinical trial designs—to balance efficacy, safety, and tolerability for newer therapeutic classes like molecularly targeted therapies and immunotherapies that may have wide separation of dose-response curves.
“A key uncertainty is how to establish the appropriate totality of evidence from these different endpoints and how to interpret the data to select optimal dose(s), which is a dose that can maximize the benefit/risk profile or provide the desired therapeutic effect while minimizing toxicity,” the authors of the Friends white paper wrote.
FDA wants the dosage of drugs to be optimized prior to approval, said Mirat Shah, the clinical lead for the FDA Oncology Center of Excellence’s Project Optimus, an initiative to reform the dose selection paradigm for oncology drugs.
“The reason this reform is needed is because currently, often oncology drug dosages are inadequately characterized prior to drug approval and this can have negative consequences for patients,” Shah, who is also a medical oncologist on the Breast, Gynecologic, and Supportive Oncology team within FDA’s Office of Oncologic Diseases, said at the meeting. “How do we look at the data to decide which dosages to evaluate, how do we design trials to evaluate those dosages?”
In early-phase clinical trials, collecting patient-reported outcomes can be beneficial to both patients and drug manufacturers.
“Drug makers are able to anticipate some of the toxicity their product may have if approved and look at some of those issues with tolerability at different dosages,” Shah said. “They can consider that information from pros as part of the totality of data for dosage selection.
“In my opinion, some of the situations where PRO information might be particularly valuable is if a product is associated with chronic symptomatic low-grade toxicities—so, things like diarrhea or blurry vision where it’s very difficult to capture the full extent and the impact on a patient’s life in the context of clinic visits,” Shah said.
“I do want to add that at the FDA, we don’t view PRO information as a replacement for standard safety measures, but rather something that can compliment standard safety measurements and enhance our understanding of a patient’s experience on taking that drug.”
When FDA reviews the totality of evidence provided by a drug sponsor, the agency focuses primarily on evaluating the risk-benefit trade-offs between the different dosages, Shah said.
“I do want to take a moment and point people towards a newer resource available through FDA oncology, which is the Oncology Dosing Tool Kit,” Shah said. “This is something that’s been posted online as its own webpage and it’s really meant to aid decision makers or drug makers and other stakeholders regarding their decision around dosage optimization and particularly what are some of the key elements and the totality of data.
“I think FDA’s recommendation for randomization as part of dosage optimization has become a bit of a hot button issue. I just want to explain why this is our strong recommendation,” Shah said. “It’s not something that’s meant to be a checkbox, but it’s something that we believe is an incredibly important tool to help us understand dose and exposure response relationships for dosage optimization.
“It’s the most powerful tool we have to limit the influence of confounders when we’re trying to interpret these relationships.”
Future research on dose optimization is needed to support approaches to data extrapolation, which information to include in dose-finding trials, and how to interpret the data to select the optimal dose, the Friends white paper states.
“It will be helpful to define the criteria necessary for extrapolating doses from one therapy, or therapeutic class, to another,” the authors of the white paper wrote. “Whether the same dose or a new dose would be necessary will depend on the available data and appropriate justification by the sponsor. In each of these situations, discussions outlined in this white paper should be considered as principles regarding what is included in the totality of evidence will remain.”
Pragmatic trials
As efforts to streamline trials and obtain fit-for-purpose endpoints and data are underway, leading clinical trialists in oncology are focused on incorporating pragmatic elements into study designs, including broadening of eligibility criteria, selection of routine clinical practice sites, and flexibility in delivery and monitoring of therapy.
“Within the continuum of trial designs, trials can include various pragmatic elements and study objectives,” the authors of the Friends white paper wrote. “The prospective nature of pragmatic trial designs is critical to address challenges typically seen in observational studies using real-world data which may include data quality, missingness, and heterogeneity of endpoints and outcomes when incorporating data collection more reflective of real-world practices and settings”
Over the past year, FDA and NCI—in what has been described as an unprecedented collaboration in oncology across government, academia, and advocacy and industry groups—showcased a new streamlined study design in a trial called Pragmatica-Lung, which focuses on overall survival and collection of essential data (The Cancer Letter, Dec. 2, 2022; Pragmatica-Lung).
“Pragmatica-Lung is sort of the far end of the spectrum,” Erin Larkins, supervisory associate director for the Division of Oncology 2, Office of Oncologic Diseases at FDA, said at the meeting. “It’s about as pared down as you can get a study and a large part of that has to do with the amount of data we have, but we don’t want Pragmatica to be a one-off or only used in this space.
“The key is to look at pragmatic design elements and every trial can be looked at to see if you can incorporate pragmatic design elements,” Larkins said. “We need buy-in. That’s more likely to come out of cooperative groups. But we would also need buy-in from the pharma companies to help support studies like that because they would need to be very large.
“That’s the other issue with pragmatic trials, because you’re expanding eligibility, you need to size up the trials. Oftentimes, the effect size may not be the same if you’re enrolling a patient population that’s not as healthy as the one that would be enrolled to a classic clinical trial.”
Drug sponsors may want to rely on validated endpoints when designing pragmatic trials, Larkins said.
“Clearly, for regulatory decision-making, overall survival is a pretty easy one,” Larkins said. “When it comes to things like the composite endpoints of time-to-next therapy, that’s a little more complicated. They haven’t really been validated as well as they would need to be to support regulatory decision making on their own.
“The problem is most of the data we see from this is from retrospective studies as well, or where someone’s comparing progression-free survival from a single-arm trial to real-world data for time-to-disease progression,” Larkins said. “If these could be incorporated into even a pragmatic trial where the primary endpoint is OS—but you’re also going to look at these endpoints prospectively randomized between two treatments that would help generate the sort of data we would need to determine at some point—can these be used as sort of an early marker of clinical benefit for regulatory decision-making, potentially, to help inform that?”
When investigators add more endpoints to a pragmatic trial, at what point does it cease being pragmatic?
“If it’s an easy one to collect, then it may make sense to put it in there,” Larkins said. “And again, not every trial is going to be as stripped down as we were able to make Pragmatica-Lung. So, what we really want to see is pragmatic elements being brought in across the board where feasible.
“Early engagement with us from the regulatory perspective is the most important. I know we’ve talked about the global nature of these trials. We engage with international regulators all the time,” Larkins said. “I think they may be more open to some of these than you think they are, particularly when we’re talking about things where if you’re really just going to try to look for overall survival—that’s what a huge part of their approvals is based on, especially in the countries where it’s national health service and the coverage is managed in a different way than it’s here.”
Additional efforts are needed to encourage and enable the uptake of trials incorporating pragmatic elements, the Friends white paper states.
“Initially, there is likely to be some burden on these trial sites while they build their infrastructure and not all sites may be feasible for a trial,” the authors of the white paper wrote. “Efforts to increase the standardization and level of structured data in the EHR, such as mCODE, may eventually support data collection.
“Alignment between clinical care and clinical research on data collection standards is needed. In addition, resources and best practices are needed for engaging sites that are not large academic centers and may not regularly conduct clinical trials.”
Academic-led studies
In addition to industry-sponsored trials, studies conducted by academic investigators are another source of data that can provide information on the safety and efficacy of a drug for the purposes of supporting a regulatory submission.
For example, NCI’s National Clinical Trials Network, which is primarily composed of SWOG, COG, ECOG-ACRIN, NRG Oncology, and Alliance—formerly known as cooperative groups—can reach a patient population that is more representative of the U.S. population to support FDA’s review of cancer drugs, according to the agency.
“For academic drug development trials where the industry partner indicates an interest in the potential use of the data to support registration or labeling updates, considerations should be given to enable proactive planning of the data collected and align with expectations of regulatory submissions,” the authors of the Friends white paper wrote. “Submitting data to the FDA as well as other health-regulatory authorities for regulatory decision-making requires data to be comprehensive and formatted in well-defined and internationally recognized standardized ways.
“This can be difficult to achieve if statistical designs, study conduct, data collection methodologies, and other processes do not meet the expectations of the FDA and other health regulatory authorities.”
Drug sponsors should engage FDA early in the review process to ensure that data generated in an academic study are appropriate for supporting approval of an investigational drug or new indication, said Christy Osgood, acting supervisor and associate director in Division of Oncology 1 at FDA’s Office of Oncologic Diseases.
“If you’re planning to submit it for regulatory purposes, I would say if we’re hearing about it at the very end, sometimes what we find is that the trial wasn’t actually adequately designed for registration purpose,” Osgood said at the meeting. “We completely understand that when you’re initially designing a trial for an academic exercise or to publish a paper or to answer a question like dose optimization, but suddenly you’re seeing these results, it may not have the appropriate endpoint given the disease that you’re looking at.
“We need to talk about the statistical design, the efficacy endpoints, the disease setting compared, or arms and all of those sorts of things. So, that’s sort of what we need to be provided with and we’ll provide you with, we can try to provide flexibility once you come and talk to us with some of this stuff. And I know that we’ll discuss later sort of the flexibility about the kind of data that’s actually submitted.”
FDA supports the use of different trial designs and submission methods in order to utilize academic data, including data from the NCTN, Osgood said.
“We do feel like it’s very important, as part of the representation of the American public in clinical trials and getting drugs out to people in the most expeditious but also safe way,” Osgood said. “We’re absolutely willing to meet not only with the industry partners, but also the academic and cooperative group partners in order to figure out where we can see flexibilities, where we can talk about submission timelines and things like that.
“It has to be done on a case-by-case basis because it is, as I said, very important to understand the context of the disease, the drug that’s being studied, the treatment landscape, what else has come out with a new safety update or something like that,” Osgood said. “We are willing to give you flexibility, but we have to have specific things that we can comment on and provide flexibility for.
“And the earlier that that happens, the easier it is for us to be flexible, because we’ve had a longer time to figure out what you’re doing and help give more advice.”
To reduce delays, industry should consider establishing collaborations with academic clinical trialists early in the drug development or registration trial process, the Friends white paper states.
“Given resource limitations for many academic drug development trials and significant efforts to streamline data collection and workflows for site staff, it is important to recognize that it may not be feasible for those conducting academic drug development trials to program every study to meet regulatory requirements (e.g., SDTM/CDISC format) due to limited resources and differing objectives,” the authors of the white paper wrote.
“As such, industry partners should consider long-term partnerships with academic investigators or U.S. network groups that allow for more sustained support for these efforts and help develop the infrastructure for these types of studies.”
