SAN DIEGO -- FDA’s top biologics official recently told a drug industry audience that it is challenging for the agency to keep up with the pace of innovation in the field of gene editing, along with artificial intelligence and machine learning technologies being used in drug development.
“This concept of genome editing is one where we have to be able to keep up with a regulatory environment that is conducive to the scientific environment and how it's evolving, and be able to help keep up the pace with bringing these advances to people, and that is very challenging,” Center for Biologics Evaluation and Research (CBER) Director Peter Marks said during a panel at the Drug Information Association (DIA) annual meeting.
The remarks come as FDA struggles to review Moderna and Merck’s mRNA-4157 melanoma cancer vaccine candidate as the companies used AI to determine the formulation of the individualized vaccine, according to Marks. In May, he suggested FDA may not yet be in a position to address whether AI algorithm alterations make a vaccine an entirely new product.
"We have an AI committee that's really trying to think about how we actually do think about this, and how we can potentially incorporate this. That question may be a little bit of a step ahead of where we are right now,” Marks said at an event hosted by Friends of Cancer Research.
Dean Kamen, founder of technology company DEKA Research & Development Corporation, noted at the panel that the regulatory environment for pharmaceuticals hasn’t had to face a change as significant as AI technology.
“The regulatory environment hasn't changed lots since 1938,” Kamen said. “It turns out you need industry and government to get together and figure out the best path forward based on what we now know.”
When it comes to AI/ML technologies in drug development, Boston-based company PhaseV is developing a platform for adaptive clinical trial design and its execution and analysis that can increase the adoption of adaptive trials and enhance the dialogue between sponsors and regulators by providing statistical guarantees, according to its CEO Raviv Pryluk.
“There are layers of the very complex models, and then before it reaches the regulator or physician, it has to go through another layer that is the interpretability layer, and statistical guarantee layer so those layers are sometimes skipped,” Pryluk told Inside Health Policy. “PhaseV plays an active role in bridging the gap between complex machine learning and conventional biostatistics or regulatory science.”
FDA Commissioner Robert Califf said at another DIA panel that AI is currently being integrated in the background. He reiterated his previous point that a comprehensive regulatory ecosystem is necessary for overseeing AI.
“What's happening very rapidly right now is, AI is really just being incorporated into the background,” Califf said. “As far as I'm concerned, a lot is at stake in getting this right, and no one knows how to regulate it at this point. I would just say here at the FDA, there is going to have to be a regulatory ecosystem that's way beyond what any individual regulatory agency can do.” -- Maaisha Osman (mosman@iwpnews.com)