The FDA previewed an upcoming pilot project that offers sponsors initial reviews of protocols for clinical trials with highly complex, innovative designs — including adaptive studies, master protocols and basket trials, as well as those using Bayesian statistics or computer simulations to determine operations.
Participating sponsors would be offered more frequent interactions with the agency, including opportunities to gather advice and early regulatory buy-in on their study proposals, during the IND or pre-IND stages.
In return, sponsors would agree to publicly disclose some details of their study designs, to be used as case studies to further educate agency staff and the industry, and to provide concrete examples of the ideas that can pass FDA muster.
The FDA will begin by selecting up to two submitted designs per quarter for the pilot, following the formal announcement of the program in the Federal Register. Sponsors will be afforded two multidisciplinary meetings led by CDER or CBER statistics reviewers — including one initial and one follow-up meeting, both held within about 120 days.
Negotiated as an agency commitment under PDUFA VI, one of the end goals of the pilot is to demonstrate the FDA’s willingness to consider innovative designs — and to hopefully correct industry misperceptions about what they will and will not consider when evaluating a product for approval.
“The number of comments I hear about what the FDA definitely will not accept — that have never been stated by anybody who has actually been employed or formerly employed by the FDA — is just phenomenal,” said Roger Lewis, a professor at the UCLA David Geffen School of Medicine, during an FDA meeting to gather stakeholder input on the pilot.
Those rumors may proliferate because of the agency’s legal restrictions from discussing confidential product and protocol information while under review, they said.
“The FDA is tied with what they can talk about,” said Lisa LaVange, associate chair of the Department of Biostatistics at UNC Chapel Hill and formerly director of CDER’s Office of Biostatistics from 2011 to 2017. “Until the drug is approved, you’re not going to know that the FDA accepted that design,” a process that can take years, she said.
As an example, during the negotiations for PDUFA VI, there were concerns that the FDA would not accept adaptive trial designs that require computer simulations, over concerns of type I error rates, said LaVange.
“It’s very easy to ding that design… So how do sponsors get in the door?” she said, adding that real-world evidence, Bayesian trials and other methods are also real possibilities, both in and outside of the pilot.
Sponsors have questioned the public disclosure of possibly competitive study information through the pilot program, said John Zhong, senior director of biostatistics at Biogen. While most companies are open to disclosing statistical elements of the study design, aspects such as the names of the product and company should be redacted, he said.
“A lot of companies are not even comfortable with disclosing the disease they are targeting,” Zhong said. “That kind of information is not necessary… You can also say it is a continuous endpoint, an ordinal endpoint, or a common endpoint. You can disclose information like that.”
“We hope to disclose as much as possible, but it has to comply with the sensitivity and confidentiality of the companies,” he said.
However, Lewis described how the nature of those details is fundamental to the design of adaptive studies.
“This is an area of clinical trial design where the clinical details are supposed to inform the design, because the design is supposed to be customized to those details,” he said. “When you anonymize them, you break the link between the very information that is intended to design and the design itself.”
“A lot of people are looking for very specific examples of something that was a successful design effort,” Lewis said.
LaVange said disclosures would be decided upon on a case-by-case basis, and said the pilot’s procedures would be more fully explained in the future Federal Register announcement. “It may be that nothing about the drug, or the sponsor, or even the disease needs to be disclosed,” she said.
“It may be just the elements of an adaptation or something else. On the other hand, if there is use of a patient registry, then you’ll have to disclose something about the disease.”
“Put yourself in the FDA’s shoes, and think about what would be most beneficial to the world of drug development for us to disclose,” LaVange said.
“Maybe it’s the fact that we are seriously considering a design with this particular adaptation, at these particular times, based on this information, these decision-making criteria, and this level of evidence.”
LaVange said she hopes the pilot project will spur more discussion about certain trial designs, including at industry meetings and through white papers.
“I have sat in meetings where sponsors have put up on the screen all these different types of adaptive designs — some of which are not complicated at all — and said, ‘This is what the FDA doesn’t like, because they called them less-well understood in their 2010 guidance.’ And I wanted to say, ‘No! No, we didn’t mean that!’” LaVange said. “I think we’re just trying to move the needle in this way.”