This exact point came up around here when we last discussed FDA reform, so it’s good to see it made at length in the New England Journal of Medicine. Remember solanezumab? That was the amyloid-targeting antibody that Eli Lilly kept on investigating in trial after trial, looking for some effect on Alzheimer’s. Last November, the final, final word finally came down that it really, truly, does not work. To recap, mouse model results with a similar antibody were published in 2001. Phase I results of solanezumab itself were published in 2010, and Phase II results were published in 2012.
The authors of the NEJM paper would like to point out that under the current system, the cost of investigating all this was largely borne by the drug’s developers, not the patients and not the taxpayers:
But the failure of solanezumab also offers a window into the U.S. drug regulatory system, particularly in the con- text of the recently signed 21st Century Cures Act and the ongoing national debate about the role of the Food and Drug Ad- ministration (FDA). The solanezumab story is an important case of a regulatory system that worked.
Indeed it is. Under a system designed to speed up drug approvals, people might have started taking it back in 2010-2012, when the Phase I and II results showed no adverse effects. I was actually worried about this at the time – not so much that people would get early access, but that a trial might show eventually show wispy possible little hints of efficacy, just barely enough to get the drug approved, but to no beneficial effect in the real world.
But there really is such a huge demand for something, anything, with any hint of hope. People would line up to buy anything that got FDA approval, no matter how tenuous the evidence was. And that puts the agency in a very tough position. . .You can argue that one of the main purposes of the agency is to make sure that medicines that people can be prescribed in this country will actually do some good, rather than raise hopes for nothing. You could also argue that responsible adults – and their physicians, and their insurance companies – should be able to make such choices for themselves, and should be able to spend their time and money in the ways that they best see fit. You could argue that companies with marginally effective (or ineffective) therapies face a huge moral hazard, in that their incentives are to get such treatments onto the market whether they do anyone else any good or not. None of these are foolish positions, but they are also, in places, mutually incompatible. Alzheimer’s disease might well turn into the next place in which we thrash them out.
I think all those points stand, and this latest article is basically a sigh of relief that we didn’t get to that point:
Lilly’s decision to undertake Expedition3, though logical, was not inevitable. The trial was undertaken after those reported discussions with regulators because the law requires manufacturers to provide the FDA with substantial evidence of a product’s efficacy before bringing it to market, and prospective, randomized trials are generally the best way to do so. In recent years, however, this standard has increasingly been criticized as overly stringent; some commentators have even suggested that the FDA should simply approve all drugs that are determined not to be toxic, letting “the marketplace” determine which ones work best.
Normally, as a free-market capitalist guy, I would take exception to those scare quotes around “the marketplace”, but unfortunately, in health care they’re warranted. We have a very tightly regulated and opaque market indeed in this country for prescription drugs and every other form of health care, and it’s not a very good place to discover prices or utilities. You could imagine a system where these things could be done better than we’re doing them, but such a system would be pretty far from what we have going now.
And even in an idealized situation like that, I’m not sure how well “efficacy discovery” could ever work. How would one collect all the patient data? How could that be done in a way where the data themselves could be compared? Issues of patient selection, diagnosis, compliance, measurement of outcomes, and other issues would all get much more difficult and complicated. Dealing with all those variables and getting them out of the way is exactly why we run controlled trials in the first place, so tossing all that aside seems odd, to say the least. You might imagine a world in which you could say “Well, come on now, does this drug work or doesn’t it? Simple!”, but that’s not the world we’re living in. It really isn’t.
The NEJM paper estimates, pretty conservatively, that had solanezumab been given conditional approval back in 2012 or so, that we – meaning Medicare, for the most part, which is to say all taxpayers, but also insurance companies and patients – would have spent at least ten billion dollars injecting Alzheimer’s patients with an expensive placebo. No one would have gotten the tiniest bit better. False hope all around, with no benefit, and billions of dollars down the tubes.
This would be a good time to remember Bastiat’s broken window fallacy. Actually, it’s always a good time to remember Bastiat; the man was a fountain of good sense. You hear this one trotted out after hurricanes and the like to the effect that at least all that rebuilding will stimulate the local economy. But Bastiat draws attention to “what is not seen” – the fact that the money spent doing this would have been doing something else, had all those windows not been broken and all those buildings not destroyed. Then we would have had the benefits of that activity, on top of keeping all the buildings intact. The same goes for the billions that would have been spent on solanezumab. One might say “Oh well, no harm done”, since the drug at least met the primum non nocere cutoff. But those ten billion dollars could have helped those Alzheimer’s patients (and their families) in other, more meaningful ways, and been used for other things entirely as well. Pouring money and effort (and human hopes) down ratholes is never a good thing to rationalize. Let’s not.