[JamesBarney]

But of that 100 million dollars the bulk was safety testing due to regulation.

So you're right it doesn't make economics sense but for a specific reason. As a society we have to strike a balance between safety and efficiency. And I don't think we've struck the right balance. Many times erroring too much on the side of safety, especially with rarer and more fatal diseases without good current treatments.

[irq11]

You’re focusing on the wrong thing.

Even if you ignored the cost of regulation and development, the company who was producing this drug would need to spend millions of dollars a year on production just to keep a safe supply available. With a pool of a few hundred patients, all of whom are “cured” for a decade, it’s going to be an expensive drug no matter what regulatory system is in place.

[mistermann]

How do they make it in a lab to develop it initially? How is that possible, but it isn't possible to make small batches on demand?

[irq11]

They had a staff of skilled people who had been working on the project for years, most likely in an (expensive) laboratory that was purpose-built for the task of producing a transgenic virus destined for humans.

It’s not cheap to produce at any scale, and it’s typically more expensive to do small-scale production.

A million dollars can barely fund a startup for a year; it’s insane to think you can maintain a drug production facility on a tiny budget like that.

[mistermann]

But if making a small batch costs $1M, how do they afford to do the research for years in an academic setting?

> It’s not cheap to produce at any scale, and it’s typically more expensive to do small-scale production.

This is correct, but we're not speaking in relative terms, we're speaking in absolute terms. How can they afford to produce it in a lab (unless each research project has many, many millions to play with each year)? It doesn't seem logical to me.

[irq11]

Per the article, the development project cost hundreds of millions of dollars. It wasn’t cheap to produce.

[rileymat2]

"But of that 100 million dollars the bulk was safety testing due to regulation."

How do you differentiate the cost of safety testing that should be done and the cost of safety testing due to regulation?

[Nasrudith]

I think the answer is risk tolerance related for how much they would be sued without the testing. Many companies do risk analysis above and beyond what the minimal regulatory requirement if they feel it is more cost effective. That would be the number used in comparison.

That said the regulatory premium may still be the right thing to do given horrifying human costs and potential to leave everybody else holding the bag when they go bankrupt with expenses far beyond what they can pay. Even if you seize assets of the entire board in a criminal proceeding it is still possible to do more harm that can be repaired by the courts and thus make regulation the rational thing to do in addition to the right thing for a circumstance. Or the regulations could do more harm than good. In general regulations themselves aren't inherently good or evil but circumstantial.

[JamesBarney]

The cost of safety testing due to regulation us the cost they spent. How much you think should be done would be based on your models.

[gus_massa]

Also the 100 million is used to distinguish between real medicine and snake oil.

Also, is this treatment better than the current treatment? How can you be sure without a clinical trial?

[ekianjo]

You cant even do a proper clinical trial if there are only 300 patients in the whole of Europe who have the disease. You are missing the point.

[gus_massa]

The point is to discover real medicine that really cures the people.

If there is no clinical trial, how are you so sure that this treatment is better than the previous one?

You can trust the gut feeling of the experts, but that is not enough. The rules for a clinical trial are complex, but they are created to avoid some mistakes (or frauds) of the past.

In particular of the recent changes is traying to force preregistration. https://en.wikipedia.org/wiki/Clinical_trials_registry Because it's too easy to massage the results after the trial to get a lot of results and then report only the "good" result and hide the others. The "good" result is an statistical fluke, so any attempt to reproduce the trial will fail.

Another older requisite is having a control group with a double blind setup if possible. For example, they tried this drug in 31 people. Did this 31 people get better average results than the other patients with the old treatment? They are probably applying the experimental treatment only to people that is not near death. If someone is going to die in a month and is very weak, you don't want to kill him because the new treatment has an unexpected side effect, or get blamed that the treatment killed him instead of the previous condition, or that the family later says that he was too weak to give an informed consent, or get random problems because the organs are too broken, or ... So there are some explicit and implicit filtration rules. Does the patient has any additional illness? Is s/he to young or too old? When you take the average result of the patients in the study and the patients with the normal treatment, are you using the same explicit and implicit filtration rules. What about long term effects? What about the periodic medical checks during the trial that detect early signs of other problems? Does they eat more healthy food during the trial?

So you need a control group chosen at random after you decide to include or exclude the patient in the study, so you try ensure that you have the same population. Preferably a double blind experiment because the patients may give inaccurate reports if they thing that the treatment is good, and the medics and nurses also are influenced by their expectations.

And in these trials, the control don't get a placebo, but the usual treatment, because using a placebo when there is some good treatment is unethical. This is a rule, because some people try to do this anyway, because it's easier to show that the new treatment is better than the placebo than to show that it's better than the current treatment.

Also, they have only 31 patients, that is barely usable to make statistics. The error bars for 31 subjects are huge (something like 5%). The difference must be huge to get a statistical significant result.

You can trust the gut feeling of the experts to think new treatments, but they must prove that the treatment is effective. Otherwise, you can't distinguish between a expert with good ideas, a moron and someone who wants to sell you snake oil.

[Fomite]

You can indeed do a proper clinical trial. Effect size matters. Something that provides a complete cure will have a commensurately big effect size, especially since in this case, you have pretty clear markers of cure.

An RCT comparing fecal transplant vs. vancomycin for recurrent C. difficile infection, for example, was halted after recruiting only 43 patients because it was judged no longer ethical to keep people in the vancomycin arm.