The Impact of Artificial Intelligence on Cell and Gene Therapy

This is the tenth part of a transcript of a Special Report with Deborah Phippard, PhD, and Renier Brentjens, MD, PhD. For the ninth part, click here.

The past 25 years, from 2000 to 2025, have been an unprecedented and rapid period of development of the field of cell and gene therapy. To get a perspective on how far we've come, and how far we have yet to go, CGTLive® reached out to Deborah Phippard, PhD, the chief scientific officer of Precision for Medicine, and Renier Brentjens, MD, PhD, the chair of the department of medicine and the deputy director at Roswell Park Comprehensive Cancer Center, to hold a Special Report discussion on the topic entitled: "Quarter Century Update: What’s Holding up Progress in Development? Where Have We Seen the Most?"

In this transcript of the tenth and final episode, Brentjens and Phippard discussed their thoughts on the potential impact of artificial intelligence (AI) on the field of cell and gene therapy. Brentjens emphasized that because of the complexity of AI technology and the wide range of its potential impact, it is impossible to make very specific predictions about how it will affect the field over the next 5 to 10 years, but expressed enthusiasm about its potential. Phippard agreed, and discussed a few possible use cases for AI tools.

CGTLive: Is there anything else we haven't covered yet?

Renier Brentjens, MD, PhD: Alright, I'll mention the gorilla in the room because someone has to: AI.

Deborah Phippard, PhD: I was so not opening that can of worms!

Renier Brentjens, MD, PhD: I'm kind of loosening the cap on it. I'm not opening it.

How will AI affect cell and gene therapy?

Renier Brentjens, MD, PhD: I think most people that blurt out AI have no idea what they're talking about. I've read several books on AI, and I still really, really don't know what I'm talking about. That having been said, we had some people that did know what they were talking about, and we utilized AI to create binders. I think that that is probably the tip of the iceberg. The amount of big data that we are now collecting that we still don't really know what to do with, but probably some type of AI engine will help us to make sense of it all, is one of the things that takes us out of our comfort zone and removes us from having as much control over what we're doing scientifically as we would like.

I think, and I'm not unique in this at all, that it's extremely difficult to see what impact that type of technology will have in the next 5 to 10 years. I think Deb and I were kind of talking as if everything else stays static and our technology moves forward, but you can imagine with AI and the ability to rapidly make binders, the ability to rapidly identify neoantigens, the ability to rapidly identify what the tumor microenvironment means for the design of a cell therapy, even designing in vivo delivery methods—I think that those are things that will markedly shape where we're going to be.

If we're saying that AI is going to make an impact, and then acknowledging that we don't know exactly what AI is, that means that our opinions are probably not as valid as we would like them to be, but at least we're here for the ride, and it's going to be awfully exciting. As I said, we used AI to make an optimized binder, and it was a remarkably fast and remarkably satisfying experience working with these AI guys because it works! Now, what it can do and what it can't do, we'll find out, but it's going to be very, very cool.

Deborah Phippard, PhD: Yeah, and I concur. So I am glad you opened that Pandora's box, because we're using AI in terms of patient identification. I touched on it a little bit, that when you really do phenotype and genotype diseases, they're incredibly complex. They present as a particular solid tumor, but what's driving those tumors? And it's not the same. Within a tumor, you've got a lot of heterogeneity, so there's a lot of work right now with using AI tools to select the right patient and to constantly monitor those patients and all the omics assays, as that patient goes through a cancer journey, should they be changing to different treatment modalities. With a one-and-done CAR-T maybe not, but getting these patients into the first phases of who should most directly benefit from this treatment? Particularly when you think about the fact that no treatment is 100% risk-free, being able to truly say to a patient your risk-benefit profile is this, I think helps drive whether that patient is willing to take that particular treatment. I'm not downplaying how nuanced that is, how hard it is to explain and convey to a patient, but AI tools have a lot of value when they look at patients across hospital systems. It sort of speaks back to access. I hear the big tertiary hospitals are reaching out to the community to support CAR-T-type programs and have a tertiary center on standby if a patient does have a bad cytokine release syndrome—we're getting a lot better at managing that—but thinking about finding the patients, getting them to the right treatment centers, on the right treatment. There's so many different ways we can use AI. We have barely scratched the surface. It's going to be the generation behind us I'm afraid that's going to work on that. But, yeah, there's just so much promise.

Renier Brentjens, MD, PhD: I mean, we're getting into Star Trek-type technologies. Again, my prediction for something like cell therapy for solid tumors is that AI will analyze the tumor and then it will spit out what target the cells should have, what armors the cells should have, and what type of cells or how many cells you need, or whatever. It's going to be very, very cool.

Deborah Phippard, PhD: Yeah. Preferably before either you or I die of cancer, right?

Renier Brentjens, MD, PhD: Knock on wood.

Deborah Phippard, PhD: Yeah, there you go.

This transcript has been edited for clarity.