I have started a series of posts sharing the research I did working on an Emergent Ventures grant in early 2020. This is the third post. The first post is here. Here is the second post, which this third post is part two of.
Drugs cost a lot of money to develop. It’s not that coming up with chemical compounds is expensive – every drug company has tens of thousands of otherwise unknown substances in their library. It’s that drugs cannot be sold unless they are approved by medical regulators.
Most important among medical regulators is the Food and Drug Administration, which controls what can be marketed in the USA. The FDA has very strict standards for what can be sold: you have to prove it is safe and effective in specific ways that typically involve several years and tens of thousands of subjects.
Not all drugs are going to work. So not only do you have the cost of testing a specific drug, but it has to make up for all the failed projects too, like how venture capitalists tend to make lots of bets that fail, made up for by one bet that succeeds big.
So drugs have huge fixed costs – the costs of researching and developing them – that you have to foot to bring them to market. One estimate says this cost is roughly $1 billion per ‘new chemical entity’ – which is what the industry call newly approved drugs.
But drugs are an especially strange market because they are also quite easy to copy. This could cause serious problems. Imagine I, WorldPharm, spend my $1 billion and develop my excellent new cancer drug, but my competitor XenoDrug just copies it and immediately sells it for half price the day it comes out. In this world, no one develops drugs.
Of course, we don’t live in this world. You are probably thinking that the answer is that we have patents, but not so fast. Between 1962 and 1983 the fact that your competitor XenoDrug could easily copy your drug didn’t matter, even if your drug was long past its patent or any other protection. Because remember you need approval from the FDA to sell in this market, and it’s WorldPharm that has the approval. Until 1983, the FDA required a XenoDrug to go through exactly the same clinical trials as WorldPharm had done originally, if they wanted to sell the copycat drug. Firms could effectively count on permanent trade secret protections for the results of their drug trials, even after they were divulged to the FDA to get approval.
Of course we do also have patents. Pharmaceutical firms apply for patents when they identify a new drug. The patent protects them for 20 years – but this includes the time they are studying whether the drug works, and then also the time the FDA takes to decide whether it’s going to approve the drug. This means the ‘effective’ patent term is often much less than 20 years, depending on how long it takes to prove the drug works. This is very important, as we’ll see.
In 1984 the US passed the Drug Price Competition and Patent Term Restoration Act, generally known as Hatch-Waxman. One the one hand, they wanted more generics – a full third of out-of-patent drugs faced no generic competition in 1983. But they didn’t want to overly punish drugmakers, and also wished to make effective patent terms longer.
The change on the generics side was gigantic. Firms wishing to market a generic now merely have to prove their drug is chemically identical to the original brand name version. What’s more, firms offering up the first generic got 180 days where only they and the original branded version could be sold.
By the end of the 1990s, generics were well established, with a generic competitor arising in almost every case when patents expired, and with a majority of those currently taking a drug usually switching to the generic competitor as soon as it became available.
This ten years was extended up a bit with some extra rules on market exclusivity to supplement patents, intended to ‘restore’ effective patent terms, as the bill name indicated. There is a huge amount of legal complexity here in terms of schemes that pharmaceutical firms use to try and eke out a little more exclusivity (e.g. adding different formulations, or making special versions approved for some market segment) and schemes generics producers try and use to get around this. But overall, drugmakers tend to get a bit more legal exclusivity now – about fourteen years according to one estimate – at the cost of a lot less practical exclusivity, which was effectively permanent in many cases under the old system.
One interesting upshot of this system is that drugs for certain conditions get systematically more effective protection than others. In the bulk of conditions, drugs are approved on the back of keeping people alive for longer. If your preventative drug, for example, would massively lengthen lifespan, but only if taken for 20 years, then there’s no way of making money off it at all. By contrast, if your drug is for end-stage cancer, with most of the affected patients dying within a few years, you might get almost the full two decades of protection.
This negative effect of time to impact on research incentives is called ‘commercialisation lag’. Dozens of drugs are developed every year to treat cancer, but only a handful of drugs have ever been approved designed to prevent cancer. Treatment is important, but many recent cancer drugs extend life by a matter of months, whereas prevention may have significant untapped potential.
Heidi Williams estimates how important this is in practice, by comparing what firms do when researching cancer drugs in general with how they research in areas like leukaemia where they can get drugs approved on the basis of ‘surrogate endpoints’ – proxies that happen quicker such as cancers shrinking in size.
She finds just what we’d expect. In general, there is a massive bias towards trying to find drugs for the most aggressive and rapid cancers, rather than those that kill the most overall, but more slowly. But this bias goes away in the subfields where firms can get drugs approved on the basis of these surrogate endpoints. Overall, William reckons 900,000 life-years are lost annually in the USA due to this distortion.
Now, as we will see in the next post, these endpoints have huge downsides too, at least as currently used. But one of the greatest successes in terms of outcomes in healthcare over recent decades has been the 50% drop in cardiovascular disease mortality. Again, CVD is an example of a field where most of the best treatments have only long-term impacts. However, it is also an area where the FDA approves drugs on the basis of surrogate endpoints.