"Does anyone know an approximate cost breakdown of a typical clinical trial say for a new drug?"

No one knows exactly because drug companies are allowed to count many of their marketing expenses are part of their R&D. For example, phase IV trials are mostly done as marketing as an excuse to bribe doctors with tens of thousands of dollars to put patients on the drugs in exchange for answering one or two questions about their opinions on the drug that are completely unscientific and unpublishable.

That said to get a drug through phase III trials shouldn't cost more than about $20 million. You can go to MAPS webpage and see their budget and how much they are projecting it's going to cost to get MDMA approved. The reason the pharma companies claim that it takes billions per new drug to develop is that they're basically counting marketing expenses, other failed drugs, regulatory fines, etc.

"That said to get a drug through phase III trials shouldn't cost more than about $20 million."

I would have to seriously disagree with that statement.

I've spoken to a number of CROs that run clinical trials. They estimate that it's somewhere between $10K and $20K per patient per year to run a phase III clinical trial (the exact number depends on what kind of data you are collecting).

For example, if we look at a cardiovascular drug where you're trying to get data to support improved patient outcomes (not just reduced cholesterol, but fewer deaths), you'd need to run a trial of 20,000 patients or more, for several (think 3 to 5) years.

For example, look at the JUPITER trial http://en.wikipedia.org/wiki/JUPITER_trial

Figure out the math on that: 20,000 patients * $10K/pt/yr * 3 years. I get $600M for ONE trial.

Keep in mind that the FDA requires TWO pivotal trials to gain approval for a drug.

Of course, this is an extreme example. CV drugs and the trial described above are pretty much the largest trials you'll ever see. However, to say that getting a drug approved through phase III should cost more than $20M is clearly wrong.

"They estimate that it's somewhere between $10K and $20K per patient per year to run a phase III clinical trial"

CROs pay enormous sums of money to doctors to get them to enroll patients. The reason they do this is because the patent clock is ticking, so no matter what it costs it's worth it. But if you were actually testing a drug that's not under patent, then there is no reason to pay doctors to finish the study as fast as possible. Of that 10k per patient, 8k of that could easily be the fee that the doctor gets for signing up each patient.

Also, the majority of drug approval studies only have a couple hundred patients. You'd be hard pressed to find many phase III trials with over 500 patients. IIRC the Jupiter study was a phase IV study that was done for marketing reasons to promote what was already a blockbuster drug, so it's not really relevant

Again, I'd have to disagree with you.

$8K of the $10K is not money used to grease the wheels to get studies done. Think about it: you're treating a patient for a whole year with a new drug. You're dosing the drug, taking blood samples, monitoring the patient closely, running tests to determine endpoints, collecting and processing data. You think that would only cost $2K/yr/pt?

Pharma companies may incentivize physicians to participate in clinical trials, but that $10K-$20K is the cost of running the actual trial. Doctors, physicians, technicians, data analysts don't come cheap.

And no, the majority of drug approval studies don't have only a couple hundred patients. If you're doing research for an orphan drug or an extremely rare cancer, that might be true, but those hardly make up a majority of the drug trials that happen.

A great example, Contrave an obesity drug went in front of the FDA. The FDA wanted more data (on top of the phase III studies that were already done). How many patients for the additional trial? 10,000.

Shouldn't other failed drugs count, though? If you have to develop e.g. five drugs for each one that you find to be successful, the R&D costs for the four that failed should be counted as part of the costs for the successful one.

"Shouldn't other failed drugs count, though?"

Yes and no. Part of the issue is the pharma companies are only interested in testing new drugs that are under patent. Whereas these are by far the riskiest drugs to develop since much less is known about them. In contrast there are hundreds of other chemicals that have the potential to cure everything from cancer to PTSD, and they would cost only a few million dollars to turn into commercial drugs, but it's never done because there would be no profit in it. So yes failed drugs make sense to count from the viewpoint of the balance sheet of a pharma company, but it's not a very good answer to the question about how much it actually costs to develop an individual drug.

I was taught it costs 100million dollars with one in ten chance of success

That's inherently wrong, if only because there's no one who knows the actual statistics. We know that it's possible to develop a drug for less than $20 million, how much pharma actually spends is indeterminate.

That's really interesting. Thanks. So in theory a startup might be able to get that 20 million figure down to 5 million if they're lean?

"So in theory a startup might be able to get that 20 million figure down to 5 million if they're lean?"

Not likely, MAPS is already as lean as you can get. And they have a big advantage since there have already been dozens or hundreds of previous studies done on MDMA to establish safety.

Sorry if I offended anyone with this comment ...

The cost is millions or billions. Pharma companies develop molecules that never even make it to the animal testing stage. Only one out of so many molecules makes it to the market and brings in money. If it's discovered that, after perhaps a decade of animal, volunteer and patient testing the drug can not be marketed, the cost that went in to that needs to be recuperated from the next one that can be marketed. It's a very expensive process.

"Pharma companies develop molecules that never even make it to the animal testing stage."

In the US the vast majority of pre-clinical work is done by universities. Pharma companies usually license the patent from the universities once it is ready for clinical testing, under the Bayh-Dohl act:

This is incorrect.

I would agree that universities do a large part of the basic science behind new drugs, but they do VERY little pre-clinical work (pretty much anything beyond simple receptor/enzyme bind studies).

Pre-clinical work is done by CROs and biotech/pharma. The resources required to do pre-clinical work is enormous. Membrane barrier penetration, microsomal metabolism of drug product require a very large investment and there aren't any universities I know of that have large enough R&D programs to support it.

Interesting. I'm guessing the pharma industry uses the term pre-clinical differently, I was just using it to mean anything that happens before clinical testing, which is how it's used in Wikipedia, Merriam-Websters, etc.

No, that's correct, pre-clinical is everything before actual clinical testing in humans.

However, pre-clinical has many phases, of which university research institutes only participate in the earlier ones.

Basic Research -> Lead -> Screen -> Hit -> Tox/ADME -> Formulation -> Phase I

I would argue that education research institutes only participate in the first four activities. It's very rare for them to work at all in the Tox/ADME and formulation phases, which are incredibly complex and require a lot of resources and expertise they don't have.

I'm guessing a combination of liability insurance cover for the body conducting the trial and health insurance for the trial participants would drive up the costs significantly.