Print a Personalized Cancer Vaccine

I ask Juan Enriquez how the FDA can approve a treatment built for one person. He walks me through the stack: sequence the tumor, read the immune system, use AI to find the target, and print a personalized cancer vaccine.

Then we follow those same genomic tools into the ocean. Juan explains the Global Ocean Sampling Expedition, environmental sequencing, microbial biorefineries, and why plunging solar costs give him hope.

Transcript

Pablos: How does the FDA approve something personalized? I don't even understand that.

Juan: It's an N-of-one trial.

Pablos: You do N-of-one trial for every patient. And does the FDA have a website where you can click through to do that?

Juan: They're excited enough about this that they say just keep going.

Pablos: Oh, God, that's such a relief. How many patients do you think you can do before they start to get involved?

Juan: They are involved. We went to them and said, here's what we're doing. It's very cool.

Pablos: If that's true, you have high quality sequencing; it's not extremely expensive. You can sequence my tumor. I don't understand what it means to sequence my immune system. Does that mean you're looking at the antibodies my body makes anyway? Or what are you looking for?

Juan: The immune system is a very weird thing. The person least likely to be an organ donor to you is your spouse. You almost never hear of one spouse giving another a kidney. And the reason for that is somehow, as humans, we tend to pick somebody with the complete opposite immune system from us to give our children the best chance of survival.

Pablos: Mate selection matters.

Juan: There's also bizarre distortions. Apparently their marriages break up just after the honeymoon. You think that's incompatibility after the honeymoon. But sometimes it's happened with people who've been living together for years. It turns out that birth control pills can hide smell—or whatever method we know to figure out the immune system. When you go off the pill, all of a sudden the person who you absolutely adore, who's your life, becomes somebody that you can't stand physically.

Pablos: Oh, man. It's so wild.

Juan: Part of the way in which we're going to start treating cancer is we're going to figure out: what does your immune system look like? How does it react to this particular tumor? What is this tumor doing? And then use AI to go after that tumor and then print a personalized cancer vaccine.

Pablos: How important is it to find the thing that your immune system would be trying to do to it?

Juan: Enormously.

Pablos: Because we can't just make that up.

Juan: All of us are born with cancer, because cancer is basically cells that don't know their place. When you're born, you become a pluripotent stem cell. And that becomes two pluripotent stem cells and four and 16 and 32 and everything else. For the first 16 divisions, it's like everybody's at the top of the ski slope. And then what happens is these cells decide to ski down here and become bone. Once you start going down that mountain, you've got your entire gene code, but you differentiate into very specific body parts.

What you're watching in your body is you're taking one cell, and it's going down this mountain to become every part of your body. Trillions of cells. But they each know their place. With cancer, what happens is two things. One, they don't know their place. And two, they don't know when to stop. It's a little bit like your nose will keep growing. And that's what a tumor is.

Your immune system throughout your life is basically going after these things and playing space invaders. Your immune system gets weaker as you get older. Or, in the measure that those cancers are particularly nasty genetic or radiation cancers, it overwhelms your immune system. What we're going to try and do is what your immune system does all your life, which is to fight these cancer cells by reactivating that immune system to target the specific tumor.

Pablos: Maybe what's happening is there's a tumor that your immune system started to think: that one's not a problem, and it stopped trying to take care of it or get rid of it, that specific kind of cancer.

Juan: That's part of it. The other part of it is some tumors can go cold so they don't get recognized by your immune system. The third thing is, your body gets old, your muscles get tired, your eyes get tired, your immune system gets tired. And that's the reason why most cancers tend to appear later in life.

Almost all males will get prostate cancer at some point. The question: is it aggressive enough to kill you, or will something else get you first?

Pablos: In that case, the immune system tried to fight it for most of your life, staved it off as long as possible, just got worn out. The idea with these therapies is to say: We're gonna find the thing that was working before, drive that to go harder, faster, stronger—keep going.

Juan: To this specific threat. You're not trying to redo the whole of the system. You're trying to say, ‘Hey, wake up. This thing's really bad. Go after this.’

Pablos: We don't need the machine learning model to figure out what to create to drive that. We're trying to discover it in your body. What you're using machine learning for is to figure out what direction: if this thing mutates in one way or another, how do we still get it?

Juan: It's a little bit like COVID vaccine. Or flu vaccine. Something mutates. You keep adapting the vaccine, which is what's stimulating your immune system, except in this case, it's an antigen that you're using for the immune system.

Pablos: What's happening in those eight weeks?

Juan: You're trying to figure out what type of cancer this person has. Why is this tumor growing? What is the immune system doing? What kind of stuff would reactivate the immune system in the patient?

Pablos: Are there some tests you can do for a given person?

Juan: You have standardized sequencing and typing. It's a system that we've known for years works. You could never get those treatments down into a personalized level. What's different today, for better and for worse—in cancer it's great, and places like national security it's not great—is you can type the very specific stuff and build it. The good news is, if you ever do get a threat, you can also create the counter to that threat pretty quickly.

Pablos: I'm imagining that, using a computer security analogy, we have anomaly detection on networks. You watch and see if there's something new you haven't seen before—a PowerPoint with extra executable code stuck on it. You notice anomalies; sometimes you quarantine them. Then you figure out: what is it going to take to stop this thing in future?

That's where we get to: sequencing all over the place. Anytime something novel shows up that we haven't seen before, it immediately gets investigated. If it's something that can get into humans, if it can become a real pathogen, then we start immediately designing the antigen, or whatever we need to go after it. Does that sound plausible?

Juan: That's been happening for a long time. Back when we founded Synthetic Genomics with Venter, we were sampling surface seawater all over the ocean. We got into a sailboat. It was horrible: the Galapagos to Fiji to Seychelles to the Indian Ocean. It was just terrible work.

Pablos: Especially, an “uncruel” thing to do to scientists. They're not built for that environment.

Juan: It was just awful. We were sampling surface seawater, and the weird thing about the oceans is we think of them as giant Cuisinarts, but they're not. Every 200 miles, you have more biodiversity than the Amazon. You keep going 200 miles, and it doesn't repeat. There's this incredible variety of stuff in the ocean. And the machinery and the equipment that was used to sequence the human genome, we then applied to environmental sequencing. We increased the number of known genes by thousand-folds from all species.

Then we started applying that technology to air, putting these air things in different cities to capture if something was there. Now you can do the same thing we did with the oceans. There's a weird virus, or there's a weird bacteria, or there's a weird chemical.

Occasionally—I don't know anymore—but you used to see weird setups in New York City or in DC. It would look like little posts, and they'd be nondescript, and they were sampling the air. That technology is going to become commonplace across all of our lives. You'll be able to see if there's X floating around your building or Y.

Pablos: I had a conversation with a buddy of mine named Kevin Slavin, who you've probably met before.

Juan: I remember him.

Pablos: He's trying to build a system where it incentivizes everybody around the world to go sequence everything they've got and put in a big library for everyone.

Juan: That's what we did with the global ocean sequencing, on a massive scale.

Pablos: Where's that data now?

Juan: It's all public.

Pablos: Is it a job for AI to pick through it and figure out if there's another Gila monster, the lizard that made Ozempic?

Juan: There's a whole series of ways in which life forms take sunlight and turn it into power. Leaves are giant solar panels. What this thing does is it takes these photons and absorbs them and allows the tree to live and process and grow.

When you concentrate sunlight in trees, then you can cut it up and burn it in your fireplace. What you're doing is burning concentrated sunlight in your fireplace. If you rot that tree for thousands of years, then you get oil. And then if you burn it and crunch it, then you get coal. In essence, hydrocarbons and fossil fuels are concentrated sunlight.

What you can do is disintermediate that without having to grow the entire tree stage. What we found on this expedition is thousands-fold more ways of transferring energy from sunlight into microorganisms. That means, for instance, that you can disintermediate things like corn products. Instead of having to grow a corn plant and process the plant, you get the biorefinery: you go direct into the tiny microorganism, and that makes the stuff you want.

Pablos: In some sense, photosynthesis is a thousand different things, or is it all photosynthesis? The result of that is being used to create these different operations?

Juan: There's many different pathways to take energy from the Sun and make stuff from it.

Pablos: I didn't know that. I assumed it was just photosynthesis: you take the Sun and make something, and then from there it's all the usual biological stuff.

Juan: There's thousands of different biological pathways to do that. That means that you can create biorefineries, energy storage, many different ways of structuring energy from sunlight.

Pablos: Other than photovoltaics to make electricity from sunlight, do we have any man-made schemes? I guess you use Sun to heat up water. But we're not using photosynthesis for anything other than what nature gave us already, right?

Juan: All of agriculture.

Pablos: Agriculture, but that's just redirecting what nature gave us.

Juan: Except that more than 90% of the stuff we grow to eat is unnatural selection.

Hybrid corn is sterile, and what you've done is something completely unnatural so that the energy doesn't go into reproduction. It goes into feeding cows or your weekend cookout. We've taken photosynthesis and massively repurposed it.

One of the things that gives me hope in this world is that the solar cost curves are coming down so quickly that they're already cheaper in almost all the world than heavy oil, and they're already cheaper than coal. What that means is you're gonna get close to net zero. Isn't it amazing?

Transcript

Links

Thanks, we need good co-conspirators.