Regulatory Arbitrage for Medical Research: What I Know So Far

Epistemic status: pretty ignorant. I’m sharing now because I believe in transparency.

I’ve been interested in the potential of regulatory arbitrage (that is, relocating to less regulated polities) for medical research for a while. Getting drugs or devices FDA-approved is expensive and extremely slow.  What if you could speed it up by going abroad to do your research?

I talked to some people who work in the field, and so far this is my distillation of what I got out of those conversations.  It’s a very rough draft and I expect to learn more.

Q: Why don’t pharma companies already run trials in developing countries?

A: They do! A third of clinical trials run by US-based pharma companies are outside the US, and that number is rapidly growing — a more than 2000% increase over the past two decades. Labor costs in India, China, and Russia are much lower, and it’s easier to recruit participants in countries where a clinical trial may be the only chance people have to get access to the latest treatments.

But in order to sell to American markets, those overseas trials still have to be conducted to FDA standards (with correspondingly onerous reporting requirements.) Many countries, like China, are starting to harmonize their regulatory standards with the FDA.  It’s not the Wild West.

Q: Ok, but why not sell drugs to foreign countries and bypass the US entirely?

A: The US is by far the biggest pharmaceutical market. As of 2014, US sales made up about 38% of global pharmaceutical sales; the European market was about 31%, and is roughly as tightly regulated. The money in pharma comes from selling to the developed world, which has strict standards for demonstrating safety and efficacy.

Q: Makes sense. But why not run cheap, preliminary, unofficial trials just to confirm for yourself whether drugs work, before investing in bigger and more systematic FDA-compliant trials for the successful ones?

A: I don’t know for sure, but it seems like pharma companies are generally not very interested in choosing their drug portfolio based on the likely efficacy of early-stage drug candidates.  When I’ve tried to do research into how they decide which drug candidates to pursue through clinical trials, what I found was that there’s a lot of portfolio management: mathematical models, sometimes quite complex, based on discounted cash flow analysis.  A drug candidate is treated as a random variable which has some distribution over future returns, based on the market size and the average success rate of trials.

What doesn’t seem to be involved in the decision-making process is analysis of which drug candidates are more likely to succeed in trials than others. Most drug candidates don’t work: 92% of preclinical drug candidates fail to be efficacious when tested in humans, and that attrition rate is only growing.  As clinical trials grow more expensive, failed trials are a serious and increasing drag on the pharma industry, but I’m not sure there’s interest in trying to cut those costs by choosing drug candidates more selectively.

On the few occasions when I’ve tried to pitch to large pharma companies the idea of trying to “pick winners” among early-stage drugs based on data analysis (of preclinical results, the past performance of the drug class, whatever), the idea was rejected.

Investors in biotech startups, of course, do try to pick winners among preclinical drug candidates; but an investor told me that, based on his experience, it wouldn’t be much easier to raise money if you had a successful but non-FDA-compliant preliminary human trial than if you had no human trials at all.

My impression is that (perhaps as a rational reaction to high rates of noise or fraud) decisionmakers in the industry aren’t very interested in making bets based on weak or preliminary evidence, and tend to round it down to no evidence at all.

Q: So are there any options left for trying to do medical research outside of an onerous regulatory environment?

A: Well, one option is legal exemptions. For example, the FDA’s Rare Disease Program can offer faster options for reviewing applications for a drug candidate that treats a life-threatening disease where no adequate treatment exists.

Another option is selling supplements, which do not need FDA approval. You need to make sure they’re safe, you can’t sell controlled substances, and you can’t claim that supplements treat any disease, but other than that, for better or worse, you can sell what you want.  One company, Elysium Health, is actually trying to develop serious anti-aging therapies and market them as supplements; Leonard Guarente, one of the pioneers of geroscience and the head of MIT’s aging lab, is the co-founder.

The problem with supplements, of course, is that you can’t sell them as treatments. Aging isn’t legally a disease, and the FDA is not approving anti-aging therapies, so Elysium’s model makes sense. But if you had a cure for cancer, you’d have a hard time selling it as a supplement without running afoul of the law.

There’s also medical tourism, which is a $10bn industry as of 2012, and expected to reach $32bn by 2019.  Most medical tourism is for conventional medical procedures, especially cosmetic surgery and dentistry, as customers seek cheaper options abroad.  Sometimes there are also experimental procedures, like stem cell therapies, though a lot of those are fraudulent and dangerous.  It might be possible to open a high-quality translational-research clinic in a developing country, and eventually collect enough successful results to advertise it globally as a medical tourism destination.  The key challenge, from what people in the field tell me, is to get the official blessing of the local government.

Q: Could you do it on a ship?

A: Maybe, but it would be hard.

Yes, technically international waters are not under any country’s jurisdiction.  But if a government really doesn’t want you doing your thing, they can still stop you. Pirate radio (unlicensed radio broadcasting from ships in international waters) was technically legal in the 1960’s, where it was very popular in the UK, but by 1967 the legal loophole had been shut down.

Also, ships are in the water. If you compare a cruise ship to a building of equivalent square-footage, the ship needs to be staffed with people with nautical expertise, and it needs more regular maintenance.  In most situations, I’d expect it to be much more expensive to run a ship clinic than a land clinic.

There’s also the sobering example of BlueSeed, which was to be a cruise ship where international entrepreneurs could live and work in international waters, without the need for a US visa. It was put “on hold” in 2013 due to lack of investor funding.  And, obviously, a “floating condo/office” is a much easier goal than a “floating clinic.”

Q: Would cryptocurrencies help?

A: Noooooo. No no no no no.

You’re probably thinking about black markets, which are risky in themselves; and anyway, cryptocurrencies do not help with black markets because they are not anonymous.

Bitcoin helpfully points out that Bitcoin is not anonymous.  It is incredibly not anonymous.  It is literally a public record of all your transactions.  Ross Ulbricht of Silk Road, tragically, didn’t understand this.

Q: So, can regulatory arbitrage work?

A: It’s definitely not trivial, but I haven’t ruled it out yet. The medical tourism model currently seems like the most promising method.

I think that transparency would be essential to any big win — yes, there’s lots of shady gray-market stuff out there, but even aside from ethical concerns, if you have to fly under the radar, it’s hard to grow big.  If you’re doing clinical research, it’s impossible to get anything done unless you’re transparent with the scientific community.  If you’re trying to push medical tourism towards the mainstream, you have to inspire trust in patients.  Controversy is inevitable, but if a model like this can work at all, the results would have to be good enough to speak for themselves.

17 thoughts on “Regulatory Arbitrage for Medical Research: What I Know So Far

  1. “cryptocurrencies do not help with black markets because they are not anonymous.” — that is true of Bitcoin, but there are newer currencies that are anonymous. Notable examples are ZCash (which features anonymous and non-anonymous transaction options), Monero (which is used a lot now on the dark markets), and ZenCash (an upstart).

    As a side note there is also a blockchain startup working on storing medical records on the blockchain:

  2. Cryptocurrencies do help with black markets, even though if someone wants to do the footwork they can probably unmask you. If someone wants to do the footwork they can unmask you a dozen different ways. As I understood Ulbrecht’s story he didn’t meet his doom through the nature of the blockchain but was unmasked in a different way.

    They help by being a way of transferring money online. Even if it was still easier to unmask users it would still be tremendously useful to black markets just because it *works* to get the money from point a to point b over the internet. Black-grey markets have tremendous problems with conventional online payment processors. As I understand it, though, simply mailing money works fine at the moment. Just like simply mailing drugs works. Mail very un-anonymous, – yet marijuana seeds, illegal nootropics, super-potent opioids and more find their ways to customers that way.

    The real problem with bitcoin is that the problem it tries to solve isn’t actually that severe, and if it were to become severe that same impulse would likely crush your ability to exchange bitcoin-fiat alongside. If you can’t exchange bitcoin-fiat or use white-market third party assists bitcoin is shit, people won’t switch over and the blockchain is just too cumbersome to use for every little transaction anyway. I’ve been thinking of a much more lightweight distributed cryptographic ledger system…

  3. To be meaningfully free you have to do an end-run around the entities that stand between you and the customer. If this is to be an initiative of any significant size, you pretty much need to found a new and separate society. Luckily this is very possible, happens all the time. The Church of Scientology and the Rajneesh movement are some recent examples. Gangs and cartels are more mundane and depressing examples. History is littered with them. You could say they are mostly failures, but I would say they just have a short average lifespan. Mayflies still live. Sometimes, they ‘die’ by being subsumed by a larger society that takes on some of their traits. Sometimes they take over!

    I like the idea of such an initiative having a naval component, not just because it echoes the successful L. Ron Hubbard and allows you to be technically legal. It gives you options. Lets you move around the world, maybe cut your teeth on simpler problems like disaster relief or just providing really basic medical care to some fraction of the unserved population that are a stone’s throw from the sea. Build reputation and relationships.

    Any conventional organization will dissolve when real pressure is brought to bear.

  4. My impression is that (perhaps as a rational reaction to high rates of noise or fraud) decisionmakers in the industry aren’t very interested in making bets based on weak or preliminary evidence, and tend to round it down to no evidence at all.

    Is this about drug companies deciding which drugs to buy, or invest in drugs that they already own? If they are deciding which drugs to buy, then they certainly should worry about fraud (same for VCs deciding to invest). But if these potential small trials or data analyses are in-house, it seems a strange worry. And if there are a lot of preliminary drugs to choose between, that will otherwise be completely neglected, they should be cheap to bring in-house before doing the cheap studies.

  5. If you think you can pick winners among new drugs before the huge trials are complete, big phrama seems like the wrong group of people to pitch that idea to. The natural home for an incremental improvement in picking winners (or really any kind of discernment) is almost always the financial industry.

      • Are you familiar with Mr. Shkreli? He made a significant portion of his fortune shorting stocks of small companies whose main product was in clinical trials, correctly predicting that their product wouldn’t work. Picking losers (knowing that their product can’t work) is a lot easier than picking winners (even if their product could work, they could still fuck it up somehow.)

  6. MAPS does pick winners ahead of time, which is why they can get drugs approved with a much lower cost per drug than pharma companies.

  7. I’m working in a Russian biotech startup, and have some thoughts on your take of the situation.

    1. Where does this figure (92% attrition rate) come from? I’ve got this [] review which breaks down attrition rate into a therapeutic area and phase of clinical trials, but it doesn’t have the information about the class of the molecule (first-in-class vs mee-too) or the reason for attrition (efficacy vs safety vs economy / other).

    2. It always struck me as strange how little attention is paid to drug manufacturing process. It’s ok to forget about the price of a substance when you are inventing an oncology drug with a price tag of $200,000 per patient a year, but if you are doing CNS, the realization that you have to optimize the manufacturing process sometimes comes too late.

    3. The important issue is a breakdown of costs for clinical trials. AFAIK, the average cost of phase I in the USA is about 10 M$, for phase II is $ 100 M$, and for phase III is $ 300 M$. The bottleneck is then a phase II – you have to invest $100 M with 15% chance of success. To increase the efficacy of drug development process, you have to reduce cost and efficacy of phase II clinical trials.

    4. An important insight from my friend is that no magic happens in VC headquarters. Imagine you should go through 3000 participants of BIO2017 to pick up 100 or so with whom to engage in detailed discussion of their project. Junior analyst spends 5-10 min to evaluate a potential prospect and then moves on to another candidate, so a lot of decision making happens at this level. Thus a widely disseminated information about attrition rate and factors contributing to a success can make a huge difference.

  8. What reasons did the pharma people give you for the rejection?
    Maybe it’s about patent time. If it would require 3 years to run the studies in a Pakistan to get more information before running clinical trials, that means that the time that the drug company can sell the drug with the patent protection intact goes down. It might be 6 years instead of 9 years.

  9. When it comes to medical tourism the question of how to do transparency in a way that’s trustworthy isn’t trivial. I wrote the article Prediction-based Medicine on LessWrong to propose a system of how it’s possible to be transparent about the effectivity of a treatment without running all those expensive studies.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s