As new cancer vaccines—led by Moderna and Merck’s mRNA-4157—near pivotal trial readouts, the FDA’s vaccines czar Peter Marks, M.D., Ph.D., said the agency is ready to review the shots despite AI-related unknowns.
“We are ready to review—we’re open for business,” Marks, director of the FDA’s Center for Biologics Evaluation and Research, said of cancer vaccines at the 2024 World Vaccine Congress (WVC).
“We have therapeutic cancer vaccines coming in; I think we would review them very much like we could review potentially a CAR-T cell or other therapeutic products,” Marks said.
The FDA has approved a cancer vaccine before, namely Dendreon’s prostate cancer shot Provenge. Although the next wave of therapeutic cancer vaccines aim to prove efficacy through familiar clinical endpoints—such as tumor progression and overall survival—the novelty of their tech raises new regulatory questions. And neither the FDA nor the biopharma industry seem to have the full set of answers right now.
For the production of CAR-T therapies and Provenge, a patient’s own immune cells are collected and then manipulated to recognize cancer cells. In both cases, although the final products differ based on the patient’s cells, the starting components and the processing steps are the same in nature.
For Moderna and Merck’s mRNA candidate mRNA-4157, though, AI dictates how the individualized vaccine should be formulated to attack cancer.
The platform sequences a patient’s tumor and healthy tissue and then uses AI to capture the most relevant tumor-specific mutational signatures. Neoantigens identified this way form the basis of the vaccine construct, and the resulting proteins can be very different between individuals. The goal is to induce an immune response against tumors bearing those biomarkers.
As AI is a core component of the program, Moderna expects the FDA will need to inspect its algorithm as part of the review of vaccines like mRNA-4157, also known as V940, Kyle Holen, Moderna’s head of development of therapeutics and oncology, told Fierce Biotech on the sidelines of WVC.
“We have to put our algorithm on a hard drive, put it in armored truck and have it shipped to the FDA,” Holen said.
AI algorithms are known to evolve as they absorb new information. But to protect the integrity of a clinical trial for an AI-based vaccine, Moderna had to “lock down” the algorithm before the study to avoid any changes that may have confounded the results, Holen added.
All of this raises a new regulatory question: How much change to the AI algorithm is allowed before the vaccine should be considered a completely new product?
“We don’t know,” Holen said. “Nobody knows. We’re in, like, no-man’s land here, because no one’s ever done this.”
Moderna has started these discussions internally, but the company doesn’t yet have a conclusion, he added.
“I think when the phase 3 trial is complete, you should understand all the data that we generated and use that to train the model and to create a new model,” Holen said during a panel discussion at WVC. “But ultimately what that’s going to result in is the requirement for another randomized trial of the new model versus the old model.”
Moderna has talked to the FDA and learned that changing the vendor for genome sequencing and the range in the number of antigens would be considered major changes to the program, Holen said. But it’s unclear whether those adjustments would require a supplemental application or a brand-new product filing if they came after an initial approval, he said.
“It’s an unexplored space in drug development, when you have the ability to change your drug and how it works and how you manufacture your drug continuously,” Holen said.
FDA’s Marks appears to have a more stringent criterion for considering updates to cancer vaccines as new products.
“I think the assumption is that most things are going to be new products unless they actually are similar enough to an old one that you can consider for a vaccine, that it’s like a strain change,” Marks told Fierce Biotech.
It would be hard to imagine requiring completely new reviews or large randomized clinical trials for every update to a cancer vaccine. That’s why Marks believes the FDA’s yet-to-be-formalized platform technology designation could be helpful.
The FDA recently issued draft guidance on the “advanced manufacturing technology” designation program, which was developed as part of the Food and Drug Omnibus Reform Act of 2022. The initiative is viewed as a step to facilitate the development of products based on novel technologies to improve the drug production process. The agency is still separately working on the specifics of the “platform” designation. The idea would be that products developed with the same platform technologies may enjoy streamlined reviews for subsequent advances.
The FDA has had the “platform” idea for some time. But it hasn’t really applied the idea to the fullest extent possible, Marks said.
“I think the mRNA platforms are going to force us to really try to lean into that,” Marks added.