the billion dollar molecule review

 

Now that I got out my insecurities about my own laziness in the post on work, we can focus on the real book review.

The Billion Dollar Molecule traces the first few years of a 1990s East Coast pharma startup called Vertex. Vertex is still alive and kicking today; they have several drugs at various stages of discovery and clinical trials, and a few on the market for cystic fibrosis. After reading the book, that strikes me as miraculous. Even if some of the drama was exaggerated to carry the story, Vertex was really pulling stuff out of their ass. As with many scientific stories, I find this pretty comforting (others might find it horrifying) - you just have to keep trying to pull stuff out of your ass until something works! The base rate is only slightly better than utter failure, so you have to be a little crazy and extremely gritty to have a chance. (You also have to be smart and it would be good if you could minimize the lying to make it happen, but more on that later). 

There was a lot of interesting stuff in this book about starting up a company, doing pharma research and the interplay between pharma and industry. There’s also a lot that can be learned about startups in general. I recently read Liftoff (about SpaceX) and Zero to One - both were great and covered a lot of similar themes. I probably found The Billion Dollar Molecule more interesting because I work at a biotech startup, and there were definitely some bio-heavy parts that might make it a bit of a slog for non-bio-oriented people. But if the pharma world interests you at all, I highly recommend it. 

The story

Vertex was started up by Josh Boger and some biotech finance folks, with the intent to do “rational drug design”. This was kind of a radical idea at the time. Before Vertex, Boger tried to start up rational drug design at his old company, Merck (a card-holding member of Big Pharma) and was turned down. The classic approach back then (and still somewhat now) was brute-force screening: you gather a bunch of compounds, put the protein you’re targeting on a plate, and see which compounds can stick to the protein. If the compound and the protein bind together, you start moving forward with those molecules, make small tweaks to the structure, then try the binding experiment again. 

Boger thought this was all wrong. If you have the structure of a protein to target you should be able to design a compound, molecule by molecule, to perfectly fit in the pocket of your target and do what you want, no guesswork. It’s the ultimate mastery of nature. As always, and especially in biology, easier said than done. 

Their first program focused on a compound called FK-506 which targeted a protein called FKBP (acronym for the super original name “FK-binding protein”). The compound was being used as an immunosuppressant for organ transplants, and there weren’t really great alternatives at the time. Surgeons also used cyclosporin, but it caused kidney damage and other serious side effects. Immunosuppression is a big deal for organ transplants - if you don’t tone down your immune system, your body will completely reject the new organ that it doesn’t recognize as its own. There was a guaranteed market for the drug if they succeeded.

The goal was to rationally design a better version of FK-506 with even fewer side effects that would bind even tighter to FKBP. To do that they needed to know what FKBP looked like - no one had yet figured out the protein’s structure. To do that they needed a shit-ton of the protein - no one had yet acquired large enough quantities to do the crystallization studies to get at the structure. To get the shit-ton of protein they needed John Thomson to spend weeks on end in the lab grinding up cow thymus and filtering the bovine slurry down to pure protein. At the point Vertex started looking for a partner (another pharmaceutical company) to license and handle the later stages of development for their future perfect compound for organ transplantation, they hadn’t even gotten a pure protein mixture, let alone a protein structure, let alone a rationally designed drug.

I guess this is how it all works (and I’ve seen it work this way in real life), but it’s just insane to me. People must buy into the scientific abilities of a company so much to partner when so little is known. In the end, it’s probably a good thing for innovation and I’m glad there are people with reasonably high risk tolerance (thank you, pharma partners and VCs).

In the end, Vertex partnered with a company in Japan called Chugai, which gave them enough runway to hire more scientists and go all in on the protein structure. Determining a protein structure is really difficult. In a very dramatic moment, they got scooped by a former member of their scientific advisory board publishing the structural results (more on these weird dynamics later). At least this twist didn’t matter much for the company’s prospects, it was just a hit to the scientists’ egos.

Then in a twist that did matter for the company’s prospects, it turned out FKBP might not even be the most important protein to focus on after all! They were pouring time into binding better to FKBP, but the binding of FKBP and FK-506 actually activates some other downstream pathway. It might not even make sense to spend rational design efforts on FK-506 if you could get at the important pathway more directly… yikes!

In the meantime, Vertex had a few scientists getting excited about targeting HIV protease. HIV/AIDS was a hot disease, tens of thousands of people were dying from it in the US alone. Originally the company decided not to pursue HIV because the field was too competitive and it didn’t seem like the right fit as a first or second project. But the potential for rational drug design showed some promise, so they also started working on targeting HIV protease with new molecules.

Soon after Vertex got rolling on these two projects, the market for small biotechs started heating up and Boger decided to take the company public. Which turned into a complete shitshow. They timed it a bit poorly and ended up on the tail end of a biotech bubble, when people were no longer willing to automatically pour money into anything that used test tubes. Still, the IPO money gave them more runway and they got to keep rolling.

The scientists at Vertex were working non-stop and they still had weeks where nothing progressed. There would be a sudden breakthrough, like finally isolating enough protein that the crystallographers could start their work. But then all company progress was concentrated in two crystallographers working through the night, running into wall after wall, with everyone else sitting around twiddling their thumbs.

I really appreciated this description of progress. It’s been my experience as well, and the slow, running-into-wall weeks can feel really discouraging. Things move slowly, or feel like they could fail at any moment, or you feel like you aren’t getting anywhere. But it’s pretty natural to see this rhythm for any project and probably better to not freak out about it.

The ending was a bit anticlimactic - neither of the two programs they start with end up panning out. As of today, all of their approved treatments are for cystic fibrosis and nothing in their pipeline built off their initial projects (as far as I can tell). But apparently they were doing something right. I’d really like to read a sequel about their transition from “pulling stuff out of our ass to keep buying runway” to “pulling an actual drug out of our ass.” [Edit: I did a poor job of following up after reading this book and my Supreme No-Nonsense Editor pointed out there is a sequel, which I’ll be adding to my reading list]

It’s pretty insane that Vertex worked out (so far). They are is a $65 billion company, with four FDA-approved drugs (though they all seem to be riffs on the original ivacaftor, so I’m not sure how much of this is clever FDA manipulation) and four more in Phase 3 trials (only one of these is a riff on their previous approval). It’s good to remember that a lot of companies will not succeed, even when they have people just as smart and dedicated. There’s a lot of luck involved, and Vertex almost didn’t make it either. And they certainly didn’t make it by succeeding on the first two (probably more) tries.

A few more thoughts inspired by Vertex’s story:

Picking a company focus

Wow, how do people do this? It’s the balancing act to find something that’s important enough that people will get hyped about your company, but not too competitive to get drowned out by the better resourced, more established players. It also needs to be somewhat tractable unless you want to be lying constantly about your plan and rate of progress. And especially with a drug company, when your first project has an almost certain chance of failure, you need to be hedging your bets with the scientific prowess you can show along the way. Then if your drug fails, at least you can show you have a process with potential. 

(Actually, that last bit isn’t something talked about in the book and there’s never really a hint that Boger had it in mind at all. There’s quite a bit of repetition around the theme “he never even considered they would fail.”)

For Vertex, they chose to pursue this immunosuppression target, FKBP. There was already a compound to target FKBP, but it had some problems. This sounds like a fairly tractable problem. As far as competition, Merck (a pharma giant) was going after the same target. And there was a Harvard chemist, Stuart Schreiber, researching FKBP, who was originally on Vertex’s scientific advisory board and then kicked off because he kept sharing scientific information with competitors (this man was 100% chaotic good).

Their second project was a little more contentious - HIV. It also appeared somewhat tractable because they had a clear protein in mind (HIV protease) and a mechanism to inhibit it. But the field was highly competitive because there was a huge market for any successful treatment - everyone had HIV in the 90s! Despite the obvious dollar signs (and human good, if you care about that), the scale of competition made Boger pretty hesitant about venturing into their second program. Every pharma company you can name had resources several orders of magnitude beyond Vertex invested in HIV/AIDS research.

For both of these projects, there were already people in the company with some expertise. Because Vertex recruited heavily from Merck, people came in with some knowledge of their FKBP program and just had to go about it a slightly different way (rational drug design). There was a similar dynamic for HIV research. This is probably a good place to start, but what happens when you burn through your first couple projects? All that valuable expertise from your scientists may not be as useful when you pivot to the next set of programs.

Between what I consider the “old” and “new” models of biotech startups, Vertex definitely leans new. The “old” model typically involves a company spinning out from an academic lab, based on a narrow discovery like a promising target or a promising compound. This model definitely isn’t dead. There are plenty of companies starting up this way every year. But the key here is narrow interest. The new model is about developing a platform. The goal is a system to Discover Drugs, for a wide range of diseases. You still have to pick specific projects to pursue, but the plan is to continually start up programs that can reuse the process and expertise from earlier programs. Boger’s aspiration was a company based on rational drug design - this is much more about process and much less about the particular disease you pursue… maybe! I’ll have to revisit this impression once I read the sequel. 

Picking a project is much easier in the old model, because you wouldn’t start up a company if you didn’t already have a project picked out. Prioritizing within these newer companies, where you could theoretically pursue anything within the constraints of your platform, is a definite challenge. It requires a lot more strategy around markets and chances of success in different areas and the time it will take to get all the way to a trial (though maybe this sort of strategizing was also part of the old model, in deciding whether it was even worth the effort of starting something up). I don’t have a good comprehensive theory about how to do it well, but the more business-y decisions are becoming increasingly more important and I think it’s a pretty interesting problem.

The strangeness of academia and industry

Academia and industry have very distinct sets of goals that make cooperation between the two an interesting problem. Even more interesting is the way both underdeliver on their goals and end up in mirror-image coordination problems.

Academia is going for basic research, and often makes claims of freeing up information for public use. I wonder if this is something that has changed a lot over time, because in my experience, universities act much more like businesses now with their intellectual property. In grad school, we had several talks about the proper procedures for releasing code/datasets/methods from our research that were ripe for patenting, because the university had claim to them. That doesn’t sound like free knowledge to me.

Another example I recently ran into at work: a dataset was published “for public use,” with the intent to help with drug discovery, but the license required any IP generated using the dataset to go to the university. No one’s going to discover a drug for you and then hand it over, sorry! No one can actually use this! You’re not helping anyone! You didn’t even respond to our email!

And of course we all know about the issues around publishing incentives and replication and “good science.” I won’t waste time repeating it.

Industry, on the other hand, wants to use science (and free knowledge, if they can find it) to generate something they can charge people for. In theory, industry research has an end product in mind. In practice, companies are not perfectly coordinated entities and lots of research gets done just because people want to do it that way and they aren’t really trying to go anywhere in particular. Some of the reason for this struggle is probably the way people go from academia to industry. Pharma companies hire a lot of PhDs who come in with pet interests and do not always move toward coordinated goals with the pace and focus managers might desire.

Ok so both academia and industry have issues with their own goals. What happens if we try to put them together? Unfortunately problems do not cancel each other out. In the book, Boger talks about this idealized world where academics make a lot of the scientific discoveries and publicize them, then industry goes into a locked lab and tries to figure out how to build the science into something useful and profitable. This should be how it works. Then academia can keep their altruistic incentives (HAHAHA) and industry can keep their profit incentives without dicking around trying to publish shit and it’s all great.

But everything sucks and of course it can’t work like this. Pharma companies tend to have scientific advisory boards filled by academics. In the case of Vertex, they had a board member, Stuart Schreiber, who studied the target they were going after. So he had plenty of advice to give, but he was also giving the advice to anyone who would listen! There was no way to get a competitive advantage from him (besides using him as a token when investors visited, which Vertex gladly did). He also knew confidential information from the work at Vertex, but he would immediately turn around and incorporate it into his own research and spread it to everyone. Not exactly ideal.

The whole problem around intellectual property is just a very hairy one. Within industry it can be a huge deal when people change companies. For example, one guy Vertex hired, Mark Murcko, had expertise in HIV protease inhibition, one of their founding programs, but was so worried about violating his non-disclosure agreements with Merck that he struggled to contribute. 

He was privy to all Merck’s strategies, leads, data, talking points - and legally sworn not to disclose them…Murcko hurled himself into modeling HIV-protease inhibitors that were determinedly unlike any he’d designed, or seen, at Merck. He placed himself in a kind of deep denial, ‘lobotomizing’ himself to forget what he knew, a tortured bit of self-surgery. ‘At one point, one of the chemists showed me an idea for a compound, and I knew before he was done talking exactly how to synthesize it and how it would perform in several assays. I put my hands over my ears and walked out. It hurt. That really hurt.’ (Later, [a colleague] concluded that Murcko’s repeated self-disqualifications cost the company up to six months).

Seems pretty bad for progress. The rationale for all those restrictions on intellectual property are that they protect companies trying to do new things. Would all these innovative companies really try less hard if they were under constant threat of people leaving to do the exact same thing somewhere else? It seems to me that this already happens, and it’s just the most rule-abiding people who let it cripple them and their contributions when they change companies.

So there are lots of IP issues. The old dynamic of academia providing free knowledge and industry converting it into practical products is dying of slow strangulation by university legal departments (though I’m not sure the ideal was alive at any point in history?). Non-disclosure agreements in industry make it really hard for people to switch companies and productively contribute in their areas of expertise. I know the classic line about patents driving innovation (and there’s a lot of research on that question that I haven’t read), but I see so many motivated people who are more blocked by patents/IP issues than driven by them. I guess you have to consider that companies (rather than individuals) are motivated by IP monopolies - but then they should be in competition to retain the experts to make that possible. After writing all that out, I’m still not sure where I land on this question.

Hubris and motivation

It’s pretty hard to read this book and not think about Theranos. What percent of hubris shown by Elizabeth Holmes shows up in every other company that actually succeeds? I don’t doubt that she was knowingly defrauding people. But I think she also believed her own spiel about the company’s potential and some part of her act is considered in bounds for biotech startups. Vertex was selling people on rational drug design before they even knew what they were trying to design. You have to start somewhere, but I have a hard time believing they didn’t oversell themselves. At the beginning of the book Boger presents the company to a group of non-technical investors - he could say anything and as long as he had sciency-looking diagrams in his slides, they would believe him.

I wonder how much hubris is required for motivation. It’s hard to be motivated about things you expect to fail. Start-ups must require a remarkable amount of hubris and pushing a start-up through desperate times must require even more. “We don’t have a target yet, but we’re going to make it happen.” “We can’t figure out the structure, but we’ll still be able to rationally design a drug.” You have to actually believe “this will work,” just repeating the incantations won’t be enough. I feel uncomfortable with it, but it seems necessary. We need risk takers.

But I would like another breed of people to start companies. People who can be more realistic about the chances of success and make a judgment on hope and expected value rather than what they believe is coming to them. I don’t know what you have to change about the culture to make that happen. I’d like to think having a healthy understanding about the chance of failure and that you’re going to try anyway would create healthier businesses that are less likely to lie. It’s a pipe dream because they still have to go up against liars. But we’d all be better for it.

Conclusion

Biotech is hard and cool and chaotic and inspiring. I love to read about companies that stare in the face of failure and come out the other side (SpaceX and Tesla are a couple non-biotechs that come to mind). Let’s keep building them.