FDA extends comment period for real-time clinical trials pilot, AI use in studies (updated)

RTCT pilot, and RFI on plans to scale the program

On April 29, FDA issued a press release on the implementation of a “real-time clinical trials” (RTCT) program, unveiling the initiation of two proof-of-concept clinical trials that FDA said will report endpoints and data signals to agency reviewers from companies/investigators collecting the data “in real time.” Under the pilot, clinical trial data is captured directly from electronic study records and other sources, and analyzed through an automated platform.

FDA further announced plans to build on the two “proofs-of-concept” with a broader pilot program, about which FDA issued an RFI, seeking industry comments on how AI-enabled technologies can improve efficiency, speed, and quality of decision-making in early stage clinical trials. For the broader proposed pilot program, FDA said it intends to disseminate final selection criteria in July, and complete pilot selections in August.

Announcing the pilot program, FDA Commissioner Marty Makary said, “improvements in AI and data science” now enable sponsors and trial sites to conduct real-time trials “in a way that enhances safety monitoring and radically increases efficiency.” He also told reporters that the initiative is part of a “broader push” to make the U.S. more competitive in drug development compared to countries like China, and he said that more early-trial changes will be announced in “the coming months.” For example, FDA is looking to expand the use of Bayesian and adaptive clinical trial designs, which permit trials to incorporate accumulating data and adjust over time, as we described in January online here.

FDA Chief AI Officer Jeremy Walsh said during a press call about the initiative that the agency is not planning to “immediately” replace established regulatory touchpoints – such as pre-IND and end-of-phase meetings – instead, the pilot aims to help determine how scientific review teams choose to engage with sponsors. “We’re giving [FDA reviewers] the opportunity and the tools and new capabilities in order to redefine what their [regulatory review] process is,” Walsh said. Regarding the timeline for review processes, Makary clarified that the pilot program will initially run in parallel with traditional data submissions.

Questions remain

Taking into account the information disseminated on the press call and in the RFI, questions remain about how the FDA pilot program may be implemented, including:

• Program inclusion. What criteria will FDA use to select sponsors, trials, or technologies for the broader RTCT pilot program? Are there study characteristics that would render real-time review problematic (e.g., blinded trials)?
• Data protection. How can companies participating in the program ensure proprietary information and data privacy are safeguarded? Will concerned sponsors be able to opt out of participation? Will interested sponsors be able to volunteer?
• Patient privacy. In instances where FDA is given access to an EHR system, how can FDA’s access to a patient’s medical data be adequately limited to the scope of the investigation?
• Data quality/management. How will FDA interpret and utilize information that has not undergone monitoring, quality review, or, if applicable, independent review/adjudication as may be prespecified in the study plan? How will sponsors be able to question or dispute resulting FDA determinations?
• Safety and efficacy decisions. How will FDA evaluate the data? Will the agency make early, perhaps premature, judgments about safety? Will FDA suggest alterations to study design or procedures while the study is ongoing? Should there be concern about FDA inadvertently introducing bias through decisions/recommendations made in light of accruing clinical information?
• Reality of agency workload demands. Given the already substantial workload of agency review staff, will they realistically have sufficient time to dedicate to the review real-time study data, rather than waiting for the sponsors to submit final clinical study reports?
• AI reliability. What evidence must be collected to demonstrate that AI systems are valid and reliable across different clinical trial contexts?

Next steps

As we discussed online here, the RFI presents a unique, time-sensitive opportunity for clinical trial sponsors to shape FDA’s approach to use of AI in studies, and in regulatory decision-making more broadly. The RFI solicits industry feedback on important topics including system performance, trustworthiness, comparative evaluation, decision quality, data integrity and patient safety.

FDA will accept comments on the RFI until June 29, a 30-day extension on the initial comment deadline. If you may be interested in commenting on the RFI, or have any questions about clinical trials more generally, feel free to reach out to any of the authors of this alert or the Hogan Lovells attorneys with whom you regularly work.

Authored by Heidi Gertner, Elizabeth Jungman, Blake Wilson, Robert Church, and Eman Al-Hassan