The use of artificial intelligence in the development of cancer vaccines allows for individualized therapy, but the prospect of an ever-changing product poses new challenges for drug developers and regulators.
Immunotherapies have become a key treatment area within oncology. In 2023 the global market size of this type of treatment was estimated to be worth $118.9 billion, and this is expected to more than double over the next decade, suggests Statista. Many of the big pharma companies that own immunotherapies are now looking to combination treatments to boost their efficacy, providing them with an edge in the highly competitive space. This has led some companies to explore cancer vaccines, in combination with immunotherapies, to create more targeted therapies that increase or enable a greater immune reaction to cancer cells than would immunotherapy alone.
This approach has been aided by the rise of artificial intelligence and machine learning, which can digest data from cancer biopsies to design vaccines that target patient-specific mutations. The ability to target mutations specific to individual patients is not new; targeted cancer drugs, such as anti-HER2 treatments and CDK4/6 inhibitors, have become big sellers for the industry. However, AI’s potential to identify from each patient’s biopsy the neoantigens most likely to prove an effective target for a therapeutic vaccine adds a layer of specificity—and complexity—to the process.
Advances with AI
The use of AI has become a major talking point across many industries, with pharma being no exception. When combined with advances in tumor and blood biopsies, it has allowed cancer vaccine developers to home in on specific neoantigens as targets. Neoantigens are new proteins that appear on cancer cells when certain mutations occur in the tumor DNA.
Merck and Moderna are working on one of the most advanced cancer vaccines in the clinic, which they are investigating in combination with the immunotherapy Keytruda. “Mutations [detected in patient biopsies] can be fed into an algorithm that will predict what mutations are the most immunogenic,” Scott Ebbinghaus, vice president of clinical research at Merck, told BioSpace. “From there, we can synthesize an RNA that’s made to encode each of the mutated cancer genes, which can then be created into a protein and presented to the immune system. Each vaccine is going to be very close to unique for each individual.”
Unlike treatments that have been developed against a single, fixed antigen, the AI system being employed by Moderna will seek to improve its neoantigen selection. The algorithm reviews the genetic mutations present in a patient’s tumor and predicts up to 34 neoantigens most likely to elicit an immune response, Kyle Holen, head of development, therapeutics and oncology at Moderna, explained in an email to BioSpace. “The algorithm has the potential to learn over time through pairing clinical and immunogenicity data and will hopefully become better at selecting the most clinically active neoantigens,” Holen said.
Ebbinghaus noted that certain neoantigens are likely to produce stronger reactions and therefore feature in more patients’ individualized vaccines. This presented certain challenges to Moderna and Merck when designing clinical trials, as it would not be possible to ascertain efficacy from an AI system and cancer vaccine that was constantly changing as it learned from incoming data. To mitigate the issue, the algorithm was “locked down” prior to the beginning of clinical studies to avoid any changes to its calculations during trials, Holen told Fierce Biotech.
Another company using an AI approach is Transgene, which is partnered with NEC Corporation to design an individualized therapeutic cancer vaccine. Rather than an mRNA vaccine, Transgene is working with viral vectors to educate patients’ immune systems against specific cancer targets. Transgene CEO Alessandro Riva told BioSpace by email that NEC’s AI-enabled neoantigen selection tool allows the company to identify the most immunogenic peptides for a given patient.
“Looking for immunogenic mutations in tumors is like looking for a needle in a haystack. AI enables us to analyze an incredible amount of data; screen more than 10 parameters, such as HLA binding, antigen processing or RNA expression; adjust their weight depending on the data and propose a list of relevant neoantigens in a very short time frame,” Riva explained. “The beauty of this approach is that it could be applied to most solid tumors, as long as they display meaningful mutations.”
Transgene is responsible for selecting the “most appropriate” neoantigens from those discovered by NEC’s tool before preparing the patient-specific genetic cassettes that are inserted into a viral vector, Riva said. Transgene enrolled the first patient in the second part of a Phase I/II trial for its lead therapeutic vaccine, TG4050, at the beginning of June, with the last patient expected to enroll by Q4 2025. An additional Phase I trial is also set to start in 2025 for another, undisclosed indication, Riva added.
Conversations with Regulators
With cancer vaccines now reaching later-stage trials, one of the challenges is going to be regulating the potential products. The FDA has preempted some of the discussions in this area by releasing a guidance document on the subject of clinical considerations for therapeutic cancer vaccines. An area covered in the document is multi-antigen vaccines; the agency states that each component of these jabs may not need to be individually evaluated for safety and activity. However, the document does state that this will be “considered on a case-by-case basis.”
An FDA spokesperson provided an emailed statement to BioSpace regarding the agency’s work on AI approaches for new treatments. “The FDA recognizes the potential for AI/ML to enhance the development of personalized treatment approaches. Although we can’t predict future growth, the ability of AI/ML software to learn from real-world feedback and data analytics is an important component of scientific advancements that may lead to the development of novel products across a wide range of therapeutic areas,” the spokesperson said.
When asked about the regulatory process for Transgene’s potential cancer vaccine, Riva responded that it would be similar to the FDA’s guidance on cell and gene therapy. The expectation is that approval will be based on a process that will be applied to all patients rather than a single, unvarying product, he added.
Moderna refused to comment on regulatory expectations for its cancer vaccines. Previously, Holen told Fierce Biotech that the company expects to be required to present its algorithm to the FDA as part of the approval process. Merck and Moderna’s cancer vaccine, mRNA-4137 (V940), has entered Phase III studies and has been granted breakthrough therapy designation by the FDA, so open questions about regulation of such individualized therapies will soon need answers.
Ben Hargreaves is a freelance science journalist based in Tosse, France. Reach him on LinkedIn.
