a16z Podcast - a16z Podcast: Mindsets for Engineering Biology

Episode Date: October 6, 2017

Head of the largest bioengineering lab in the world, former chairman of the FDA and one of the few recipients of the National Medals of Science and of Technology and Innovation, Bob Langer's work has spanned multiple fields and settings and has been applied across numerous fields, from pharmaceutical to chemical, biotechnology to medical device companies. What does it mean to move across disciplines like this, from science to engineering, both in the lab and into the field? In this conversation with general partner and head of the a16z bio fund Vijay Pande (with Hanne Tidnam), Langer and Pande share the challenges and opportunities as people move across different disciplines, as well as the changing mindsets for innovation as applied to biotech: first principles, "rational" biology, do no harm, and others. At the heart of it all is "the interface of engineering and materials" in biology and healthcare innovation. Especially as, thanks to tech, biology shifts from empirical study to engineering -- not just in startups but in academia too. Yet does that make the work too "translational"? And what of regulation? The guests on this episode explore all of these themes, and more.

Hi, everyone. Welcome to the A16Z podcast. I'm Hannah. And today's episode continues our ongoing series on how science becomes engineering and all the new challenges and opportunities that presents as people move across fields and into new mindsets. It's a conversation between A16Z general partner Vijay, who heads up our biofund and Bob Langer, Institute professor at MIT, whose work has really been all about the interface of engineering and materials in biology and medicine. So, Bob, your work has been a very unusual blend, right, from chemical engineering to medicine to biotech. What are some of the ways or some of the problems that you thought about and approached differently, given that less traditional background?

One of the things that I noticed because I was working with materials myself were how did materials find their way into medicine? And what I'd see is almost all the time, what happened is somebody was, some clinician wanted to urgently solve a medical problem. And what they do is they'd go to their house to find some object that would kind of resemble the organ or tissue they wanted to fix. And then they'd use it in a person. Amazing. If you wanted to make artificial heart, what the clinicians did in 1967 is they said, well, what object in your house has a good flex life like a heart? And they came up with the idea of a lady's girdle. So they looked at what that was made out of, and it's a polyether urethane, and then they decided they'd use the

exact same material to make the artificial heart. That was 1967, but now 50 years later, that's still what it's made out of, because once you start down that path from like a FDA regulatory standpoint, it's hard to change. And of course, the artificial heart has run into different kinds of problems, in particular one of the problems has been when blood hits the surface of the artificial heart, the lady's girdle material, it can form a clot, and that clot can go to the patient's brain and give them a stroke and they can die. So, you know, but to me, it doesn't seem that surprising that something that was designed to be a lady's girdle isn't the optimal material to put into contact with blood. Right. Isn't the best heart necessarily.

Why not ask the question, what do you really want in a biomaterial from an engineering standpoint, chemistry standpoint, and biology standpoint? And then could you synthesize it from first principles? These kinds of things, these are they the way most decisions used to be made for these kinds of materials? Or was there, was there no sense of choosing optimal material first? Or was it always sort of happenstance? For most of the 20th century, clinicians would take off-the-shelf materials and use them in a patient. Another example is breast implants. And again, they wanted to go to their house and find what object would kind of resemble a breast implant. And they... I mean, just, that's such a crazy idea to just go home and like, look around your living room or

your bedroom, be like, what would work? I'm going to pull out my wife's girdle here. You know, that's astonishing. You're absolutely right. But, and then in the case of breast implants, one of those was a mattress stuffing because it was squishing. That's unbelievable. How long until that changed, the mattress? That's still one of the two main breast implant materials. One is a mattress stuffing and the other is actually a lubricant, which is silicone. That theme of taking materials off the shelf is still been one of the major themes of biomaterials for the 20th century.

when I started my career, you know, in the 70s and early 80s, then that's one of the things I started to realize and began to change. It's a major paradigm shift, but it's not a simple shift. Material science has blossomed so much, you know, as a discipline over the last 30, 40 years. I mean, in many ways, there weren't material science departments, no, not that long ago. Yeah. And so it was really a dramatic shift. Well, that's actually right. In fact, when a lot of the material science departments were, you know, really aimed at, at,

totally different kinds of materials. I'd love to hear some of your discoveries and the breakthroughs that you've thought will continue to have, you know, really profound implications going forward. When I was much younger, one of the challenges that we've looked at to try to isolate the first what I'll call angiogenesis inhibitors, blood vessel inhibitors, was we had to create a bioassay and have a way to slowly release these substances, which were large molecules, which could be peptides or proteins, for a long time in a way. where they would not be destroyed. And that led us to create really the first controlled release polymer systems for ionic substances and peptides and proteins. Slow release, basically?

Yes, new kinds of systems like microspheres and nanospheres and different kinds of coatings that enabled you to have release for a long time. There's certain drugs, for example, that they use today to treat advanced prostate cancer or endometriosis that are large molecules. And because they're so large, you can't swallow them because they won't get absorbed and they will get destroyed. You can't take them nasally or anyway. And then if you inject them, they're destroyed in seconds also. But when they're put in these microspheres, you can give them an injection and they can last for a month, now even six months. And that's led to new ways of treating prostate cancer endometriosis, type 2 diabetes, schizophrenia, you know, narcotic addiction,

all kinds of things. There are newer molecules that also face terrible or very tough delivery challenges and short interfering RNAs, S-I-R-N-A, messenger RNA, you know, new gene editing approaches, DNA. And these, I think, delivery ends up being so critical because if you can't deliver them, you know, probably those drugs will not work. So you have to figure out a way to get them to the target that you want. So that's a thread that's probably going through over 40. years of our research. It's kind of amazing that you can think you can just swallow something

or inject something. Right. That we take that for granted. If you think of other analogies like the postal system or whatever, you know, things don't just magically get to where they're supposed to go. And so the ability to actually start to engineer this aspect, as Bob was alluded to, is very transformative. And in some ways, there's still a lot that one can do. In what areas? Well, I mean, the key thing to think about is that when you take a drug, typically it goes everywhere. And so let's say you only need to target one pretty particular part. If you can target one particular part, you could up the dose, maybe, the effect of local dose, instead of it going everywhere. Like the drug-alluding stent, which is a terrific example of what you

just said. If you tried to take those drugs like taxol and give the amount that you need it all over the body, you'd take it orally or some other ways, I mean, it would probably kill somebody. But when you deliver it locally, you put it in a little polymer and you put it on the stent and it delivers it locally, the systemic dose is maybe one-one thousandth of what it would be as if you swallowed it. So by deliver it locally, you can totally change the safety and dosage profile. So it's not just like personalizing medicine for one person. It's like personalizing it for your liver or like for your... Yeah, or for very specific parts. And so I think a fantasy many people had was this old movie, was it fantastic forage? Or like you go

inside someone's body in a little submarine and you're driving around and you're getting right to where you need to go to be able to heal. I mean, that's the fantasy. But this is starting to get much closer to that. And you can imagine if you could, as you get closer and closer, that you can have this surgical strike, you know, so to speak, you know, almost like a SWAT team coming and doing what it's doing. And this is another, I think, trend that we see is that we've got one baseline thing,

which is a small molecule drugs, and then you've got delivery. Then you keep on stacking things on top. Yeah. One thing makes each other better and better, safer, more efficacy. So this is such a beautiful arc because it also empowers and sort of, you know, sort of enables all the other things that are done. How is the drug industry itself changing as these kinds of breakthroughs start rolling out into the world and into markets?

Well, I do think that the drug industry, what we're seeing, you know, 40 years ago, we saw the evolution of protein medicines, which are now, you know, which have, I think, transformed the pharmaceutical industry and the kinds of drugs we pick. Like, for example, if you look at the top 10 selling best drugs, you know, this past year, seven of them are protein drugs. If you looked at it even 15 years ago, at best it would have been one. But now that's really the majority. And the sales are over $200 billion. I think what we're seeing now are ways to possibly have DNA drugs and RNA drugs. And I think that will be a huge revolution. The great advance in genetic engineering is people figured out how you could, you know,

engineer a cell to make proteins relatively quickly. In fact, even engineer bacteria to do that, and that was a key advance, and that was the launch of genetic engineering. But you could actually go further back and maybe affect RNA or affect DNA, and that might even be more advantageous because that really gets to the heart of where the problems might occur. So DNA, say, somebody has an enzyme deficiency disease, gauchase disease as an example. It took years to come up with the drugs that could help that. You might do DNA therapy, gene therapy, to actually give somebody the gene that could make that enzyme. Or you might do gene editing to correct the gene that caused that problem in the first place. Right. DNA, you actually have to not only get it into the cell,

you have to get it into the nucleus. And there, so that means the delivery problems are even tougher. With RNA, you don't need to get it into the nucleus. If somebody has a medical issue, you can shut that issue down by giving S-I-R-N-A. That could sort of stop the RNA from working, and basically, and then the protein would never, you know, wouldn't occur. And messenger RNA would be exactly the opposite. Let's say, again, you had an enzyme in deficiency disease. That would enable you to give the RNA that would make that enzyme. And there, you'd have the advantage over DNA that you wouldn't have to get it into the nucleus. And it'd have the advantage over proteins that you can make the, you know, you don't have to spend nine months to a year to

manufacture the protein and also the, you don't have the challenge of protecting the protein all away through the body. You know, one of the key things that he mentioned there was, that Bob mentioned, was this concept of gain of function. That, you know, there's some activators, you know, and other small molecules drugs that can do this, but this is usually not what happens. Most drugs inhibit something and so on. And so this ability to gain function, which maybe was lost for some reason, or even put in a function that was never there in the first place. Yeah, is an incredible shift. That's an incredible shift. And again, speaks to something that is, again, more in this sort of spirit of engineering that we've been sort of alluding to in a couple different directions.

Biopharma, as we have right now, is development of small molecules and protein biologics and follows a given path. And I think what we're seeing is something which, doesn't really have a name yet. It's maybe technology in bio or something like that. Bob, how are you starting to see technological advances like machine learning or robotics affect how this kind of research in the biospace is done? Do you use some of these tools to increase throughput? Are they game-changing? Are they just kind of increasing speed? That I think is definitely game-changing. High-throughput involves doing things much, much faster by robots and new technologies like new chemistries that we've developed. Before this, if somebody wanted

to do what I'll call formulation, like get the drug to have the right solubility or the right crystal form, it took off in many, many years. And there were some huge problems. We would sometimes be able to solve these problems that people couldn't solve for years. We could solve them in a matter of hours or days. The example that we often give, which is a true story, is that just in 1996, Abbott had this drug called Norveir. It was an AIDS drug, a protease inhibitor, and they had it in what's called crystal polymorph crystal form one. But for some reason, even after it got FDA approved, after about a year and a half, it changed to polymorph crystal form two. And of course, every time you make a different crystal form, it has a different solubility

and other different properties. And so appropriately, the FDA, because it changed, told them it became a different thing. Exactly. Exactly. And so the FDA told them, to pull it off the market. And so they did, yet even after a long time, years, they couldn't. They could never get it back to the right crystal form. So they actually had a, so it was off the market, and it was, you know, worth hundreds of millions of dollars in people's lives. And then eventually they just had to make it as a solution because they could never

get the crystal form back. And so it didn't do very well and wasn't that helpful. Now, seven years later, what we did is just in two weeks using this high throughput approach, not only created crystal form 2, which is where they were Abbott when they stopped, but we also created crystal form 1 and discovered three new crystal forms that had never been discovered before. And like I said, that's in two weeks just because of the power of the technologies that we created. I have kind of a love-hate relationship with high-through-proof methods. The love part, you know, as everything Bob mentioned is spot on. And you've seen in other

spaces like in protein crystallography and in many aspects of drug screening, tons of things that used to be about almost like people, really just literally people pipeting with their hands in a sort of pre-industrial revolution way now is automated and then you can do all these amazing things. But there's a philosophical issue, which I think is a really interesting one, is that is there going to be a point where biology can move from something where we really have no choice but to just empirically try lots of things versus can we start to engineer? And the question is in what areas will we see this? I mean, it's funny, we would all rather our mecky colleagues understand how to build bridges rather than doing high throughput bridge design and then seeing which ones fail and screening them and so on.

And the problem with biology is that the biology is so complicated. And so some parts will be amenable to sort of a more rational engineering approach and we're seeing more and more parts of it. And the parts that can't, these will be the solutions. I'm really torn because I would love to see more of the rational approach, but clearly rationality has had its limits and so these things can sort of fill in those gaps. Well, and how do you know which are the areas that are right for that kind of design thinking and which are the ones where it's better to crunch and to stumble on discovery? It's all about can you make predictions.

So if you think you can engineer something, you should be able to make 10 predictions and have three work, not making 100,000 guesses and seeing what pans out. I agree. People in synthetic biology are really pushing the envelope for what you can do in this sort of designed, predicted, engineering way. So I think it's a mistake to think that, you know, we can go after anything in biology rationally right now. Even when you do high throughput types of things, you really want to bring in rational thinking to how you do it best. And then when you get the data, you want to really see if you can use that database as a way to really understand what's going on and hopefully allow you to do better and better. before this, you couldn't learn much because you wouldn't have that much data.

Now, there was so much more data you could start analyzing it and try to make predictions from the data about how you do the next generation, even of high throughput things. Yeah, no, that's a great point is that, you know, even when you do high throughput on 100,000, let's say, drug, drunk candidates, you know, there's like billions, trillions of things that you in principle could do, you know, just in the chemical space. it's not even just pure empirical approach versus a rational approach. It's just to what fraction can we do rationally. Yeah, and in which areas and which parts.

How to use it as a lever. Yeah, that's right. To just be more efficient than you could be just knowing nothing. Yeah. It goes back and forth. I think one thing feeds on the other. If you're able to do the high throughput things and start to learn more, like I say, you can then do better and better because you begin to understand what, you know,

structure function relationships or other things. things that might be valuable. We often look at what should be done where. So questions where there's fundamental science that needs to be done. Academia is fantastic for that. On the other hand, questions that are about engineering a product and scaling it up. That's great for a startup and maybe not the domain of academia. And so what we're seeing is a lot of the shift in that in areas of biology, so much science has been learned that there's now new opportunities to think about biology as being so much more on the engineering side than never before. I was curious to get your take on that, Bob. You know, you've seen this arc. Where are the areas do you think that are still

science and what parts do you think have the opportunity now to be novelty thought of engineering just in the last few years? The timelines that one has an academia course are very different than what one might have in an industry and also the difference in focus. I mean, in this case of academia, you're often trying to, you know, invent things or discover things. but you don't necessarily need to have a product in 10 years or 20 years or ever. But if you've learned some things, that can be transformative. And if you invent some things, that could be transformative. Materials is one big area that engineering continues to make advances in, like new materials, nanotechnology.

But there's lots of other kinds of aspects of biology, I think, at least, that are starting to merge with engineering. and you have all these bioengineering or biomedical engineering departments starting up. So better ways of, you know, ways that engineers can contribute to immunology, to, you know, to pharmacology, neuroscience. I think all those are beginning to happen more and more. A lot of startup, especially in biopharmine, still have to do something scientifically. They have maybe a target or a new technique, but they have to develop a small molecule and just, you know, do discovery. I think what at least we're starting to see here is the shift towards more of these things being just pure engineering and that they can sort of get out the gate and almost build startups in a sort of a fundamentally different way than you would maybe 30 years ago. What happens in academia is you can, you know, make some significant advances. You can get a proof of principle in animals, let's say, and you can even design prototypes. But there's a lot of things you can't do.

I mean, I don't think that the kind of large-scale manufacturing that ultimately needs to be done to create, say, a medical device or a drug delivery system or even, you know, the synthesis or production of a large amount of a new pharmaceutical is going to be done in academia. Similarly, I don't think the clinical trials that need to be done are going to be done, at least in the MITs of the world. there's almost a natural handoff where a lot of basic research without a timeline will happen in academia, but then more applied or focused research will probably happen at a company with the goal of ultimately getting out products that can change people's lives. And how do you see that handoff happening now? Is it a smooth process or is there still a lot of challenge and a lot of opportunities for mistakes and failure? I like to think that when we do it, it's reasonably smooth. But there's always enormous mistakes that can be made. And every story is different.

Failure can happen anywhere. And I wish we were smart enough to know. You know, sometimes failure can come later with a clinical trial not working well. Sometimes failure comes because you don't have the right CEOs. You know, and I think that, you know, and they haven't done a good job of either in the organization or raising enough money. You know, sometimes it happens because, you know, of bad partnerships that are set up. and you run into issues and hopefully you can solve them everywhere, actually. I mean, there's almost no place that you can't mess up.

You know, there's one other shift, which is that academia is trying to be more and more translational. And so, I mean, while there's been a Bob's real pioneer in this in terms of really pushing towards translation, it's starting to become more common and almost incented, which is, I think, a wonderful thing. But also on the other side, for the handoff, there are more and more incubators that are just bio-incubators. And I think it's that transition part that's really tricky. And there's only so much you can do on each side. And it must also be kind of like a culture moment, like a weird

thing, you know, where people have to have a different mindset. And now is the mindset shift kind of the difference between discovery to engineering at heart? What's the mindset shift, actually? And sort of pure research. You know, there's that Einstein quote, you know, if we knew what we're doing, we wouldn't call it research. And the idea is that there's, it's exploratory. we just don't know. And so that can't be done on a timetable, on a roadmap, or any of those things. But then once it starts all clicking, and I've seen this with people, you know, then you start realizing, oh, there's something really here. Right. And then the mindset changes, if you really want to take it to the next step. What happens in companies now can be much more

engineering than pure discovery. I think that's part of the shift. I do think things are more translational. If I go back to when I started, you know, doing this in the, you know, early 80s, that I think in academia it was somewhat frowned upon of getting involved with very translational work and applied work in companies. And I don't think that that's, and I should say that's a lot less true today. People see that there's been great value and I don't just mean monetary value in the fact that products and companies have come from, you know, really the basic research. And now there are hundreds and hundreds of companies within just a few blocks of MIT and some have market capitalizations in the hundreds of billions of dollars.

I'd love to shift gears a little bit and talk a bit about regulation. Having worked on the FDA science board, the advisory board, then as its chairman, and then also, you know, having been involved in startups and in academia, how did that change what you thought about the way regulation happens? Are there things that need to change? I sort of see two pivotal time points for me looking at regulations. The first thing was the AIDS epidemic. And I think the AIDS activist, I think, did a very good job of convincing the FDA

that if they waited too long on some treatments to be ultra-safe, then, you know, patients would die in the meantime. The second one that I think had a negative effect was the Vioxx situation in 2005, and, you know, there was just a small increase in deaths with this drug, but there were so many lawsuits over it that people at the FDA would get called on the carpet for approving it. A lot of clinicians, I know, felt to take that off the market was wrong because it really would relieve suffering and pain. That's very complicated because, like I say, the increase in deaths was very, very tiny, whereas the decrease in pain was enormous. These have been sort of pivotal moments in some way for the FDA. Now I think it's hard to say. You have to find the right balance of safety and cautiousness on the one hand, and yet really trying to get drugs out as fast as you can if people are dying and suffering on the other.

It's interesting that this concept of the Hippocratic Oath that we want to do no harm is, the guiding principle versus doing the most good. And I think do no harm makes sense in an ancient Greek world where, you know, you understand so little about biology and health care. And most of what they would do probably would be harmful. Right. A lot of the solutions. Optimize what's best for the population is a kind of a different thing. And there's different aspects of health care that I think as we get better at this, it's interesting to even sort of, I would say, hold no secret cows, rethink just how we think about health care broadly. I mean, I think, Vij, you have this in terms, this impacts how we think about everything

from diagnosis to treatment, right? And as the possibilities shift towards what we can diagnose and the kind of prevention that we're thinking about, how do, how will that play out in thinking about regulation? We usually don't talk about therapeutics, uh, in a sense of just trying to keep you healthy. We talk about it as trying to avoid an indication, avoid a disease, such as, you know, making sure you're avoiding getting type 2 diabetes versus a very different model where your healthcare professionals are trying to keep you healthy. Right. You probably couldn't even get it through FDA if you didn't have an indication. Like staying healthy longer, it's not an indication.

And so it's interesting as we see more things like in areas of longevity, in areas of prevention, that it's an exciting time when we even just really fundamentally. rethink how we can help people the most. What advice would you have for startups as they start to think about? When do you start thinking about regulation? You want to start thinking about it early on. What is the path that makes the most sense? We created this aerosol company with David Edwards Air, which later became civetus. Basically, all we did was change in geometry.

And we were able to go into clinical trials extremely fast because there was basically no new chemistry, even though I love chemistry, but there we were able to get into the clinic within a year on multiple drugs. Another case was Momenta. There, we were able to lower the bar by creating sort of a biogeneric, in this case, Heparin. And that also, I think, facilitated, you know, even though it was a brand new technology, polysaccharide sequencing for the first time, the drug we picked actually ended up making things happen faster. The work we did with with Henry Bram that Guilford initially commercialized. We invented a new polymer, synthesized a new polymer, but we picked brain cancer as a treatment. And the FDA moved through that very

quickly because that was a lethal disease. And that system's now been used in over 30 countries for the last 21 years. What do you think will fundamentally change in the next 20 years and how we think about health care or medical treatment or even the way we do science? I mentioned genetic medicines, more personalized kinds of medicines. I also think that one of the other exciting areas is regenerative medicine and cell therapies of different kinds. You know, I think have the potential to be revolutionary. We're starting to see cell-based companies, you know, doing whether it's CAR-T cells or circulating red cells, and then there's many companies doing regenerative medicine. Those kinds of things, I think, at least over the next 20 years, I think, can have a transformative

effect on enabling therapies that you can't now do with single molecules. I think it has to start with this shift in thinking. And it's too easy to think, well, once you get to elderly points in your ADA plus and you're going to get Alzheimer's, then we deal with it. We're going to start to see drugs. And some of them are associated with what we see people talking about with longevity. But really, it's not about living longer. It's about just pushing back those problems further and further, such that we have

a longer, longer span where we're as healthy as possible. What's the hardest problem that you want to see solved that cracked in the next era? Being able to make virtually any tissue or organ to help patients. And also, the other aspect of that is that it could revolutionize drug testing, minimize, you know, killing animals and minimize testing on humans to the extent that we can make tissues or organs in a dish as authentic as possible. We have so many new ways to read biology, whether they're talking about genomics, pronomics, metabolomics, all these different new assays, these new insights into biology we didn't have before, but that creates this new challenge of like, what does it all mean and what can we do with it? And once we understand something, that doesn't mean we know what to do about it. And then, you know, there's all the generalities and therapeutic modalities that will come up. We were living in a great time because we're right at that point where we can do many things, but yet there's still many things left to do to really. truly have a impact on human health. So we're not building 100 bridges to see what, which ones fall down. Thank you so much. Thank you. Thank you.