iwNHL 2025 | The potential for AI-driven protein engineering to develop individualized treatment for lymphoma

We’re rapidly entering the era where ChatGPT and AI is something that’s no longer kind of a cool thing you play with online, but it’s something that’s actually influencing reality and how people use this in their jobs. It’s changing the way that we work, and I think that will likely be true in oncology as well, both in terms of how do we better diagnose or classify patients, but also in terms of how do we develop drugs. There are efforts from many AI companies, and I’m thinking specifically of the Google DeepMind team, but there are many others that are doing similar things to really understand how proteins fold, how mutations can influence how proteins fold, and then potentially design drugs that are customized for a given patient. The DeepMind team founded a spinoff company called Isomorphic Labs, where they’re making customized protein-targeting therapies that are unique to what genetic abnormalities one patient has. So not a drug that treats a disease, meaning that thousands of patients get the same drug, but perhaps a drug that is customized for what is specifically wrong with a given patient’s disease. And that’s not yet resonated in cancer late-stage clinical trials. But I do think the field is moving so quickly from AI being kind of a, is this a human or a chatbot I’m talking to, to moving into actually changing the way people work. I don’t think that’s that far off in the future that we’ll start to see customized drugs, which makes complications even worse in terms of drug development when we’re not even talking about one drug, but customized drugs that are kind of lumped into a category where you’re targeting an antigen, but it may be different for patient A versus patient B, what part of that antigen is being targeted. This is a bit similar to how CAR T-cells have worked in that we think of a CAR-T cell as a monolithic, if you get an axi-cel or a liso-cel, you’re getting the same product. Of course, that’s not true. That’s the patient’s T cells that are then being modified. And one patient might have a different T-cell phenotype or fitness than the next patient. So there is a precedent for a customized therapy in an immune sense on a T-cell phenotype. I don’t think that would be that big of a leap for us to make customized protein-targeting therapies as well.

This transcript is AI-generated. While we strive for accuracy, please verify this copy with the video.