It doesn’t seem to have generated many headlines, but the FDA recently released a report (PDF) on the pace of drug development. Titled “Targeted Drug Development: Why Are Many Diseases Lagging Behind”, it seems to be motivated, at least in part, by a desire to answer “Not Because of Us!” The agency goes into some detail about the state of knowledge in Alzheimer’s, diabetes, hepatitis C, and rare genetic diseases, as examples of target-driven drug discovery. They note, correctly, that in cases like Alzheimer’s that we have such an incomplete understanding of the causes of the disease that it’s hard to see how to speed up drug discovery (other than by continuing to figure out what’s going on).
Now, you could imagine one way that wouldn’t seem to depend on knowing what really causes Alzheimer’s: a solid phenotypic assay. Those are target-agnostic, and in fact can be one of the best ways to actually track down the targets themselves, once you screen and find some agent that gives you the right effects. But there are no appropriate small animal species that get Alzheimer’s (and there’s room to argue about the situation even with the larger primates, some of which clearly don’t have AD pathology). And despite a great deal of work over the last 25 years and more, no such animal model has really emerged. No spontaneous mutant rodents, for example, have ever shown up with AD, and the many efforts to engineer some all suffer from a circular reasoning problem: you have to know what defects to engineer in the first place. Putting human-style amyloid genes (and amyloid-processing genes) into a mouse is not a good way to track down a new Alzheimer’s target, because you have to assume that the amyloid pathways you’re putting in are the important ones. If they turn out to be, and if they express correctly, then you do have a platform for screening, but so far (to my knowledge) none of these attempts are generally accepted as appropriate models.
Part of that problem is, to be sure, that the timeline of AD research is so long that the real test of an animal model (whether a drug that works in it works in humans) is always going to be a long time coming. But remember, the point isn’t to validate the animal model. The point is to treat human disease, and no one has been able, with much confidence, to advance any small-animal system that can do that.
How about something like diabetes? That’s understood far better than Alzheimer’s; I don’t think anyone would argue that point. There are even animal models for the disease – they’re not perfect, but compared to the situation in Alzheimer’s, they’re terrific. But even there, when you get right down to the causes, things get fuzzy. Type I diabetes, for example, occurs when a patient’s immune system targets their islet cells in the pancreas. That seems well established – but what sets off that immune response? What distinguishes a baby who will go on to develop classic childhood Type I from one who won’t – can you go through a list of newborns and pick them out? Not yet, although some of that is probably because the autoimmune problems may be set off by environmental factors like an infection. But I don’t think we can even say which babies are at higher risk, in most cases. Then there’s Type II, a major feature of which is the development of insulin resistance in the peripheral tissues. But if that is the root of the problem, what molecular pathway, exactly, causes insulin resistance? There are some good candidates out there, but that crucial question still hasn’t been nailed down.
(As an aside, you may wonder where those diabetes animal models came from. There are compounds that have been found to selectively damage islet cells, so you can induce a fairly realistic insulin secretion deficiency on that end. And if you alter a rodent’s metabolism so that they put on fat (and there’s more than one way to do that), something about that condition seems to also bring on insulin resistance – quite likely via the same sorts of pathways, whatever they are, that make obesity a risk factor for Type II in humans.)
The FDA is quite right to point these things out, because it’s for sure that a lot of people don’t realize them. But as you read the report, it does start to come across as an apologia pro vita sua, a brief for the defense. Here’s what the agency is doing to speed things up, here’s where we’ve helped, here’s where it’s just not our fault that we don’t have a flippin’ Alzheimer’s drug yet. They do get a lot of flak from the people who think that the main problem is regulatory overkill, and that if the government would just get out of the way that we’d have more cures. And although I have libertarian sympathies, I know too much about drug discovery to buy into this position. (Someone unsympathetic to libertarianism in general would no doubt add that this happens every time it touches on something that you actually know something about!)
The FDA Law Blog comes away with the same opinion, and they have a theory about why this is going on:
. . .we note that, when taken in the context of the current legislation before Congress (e.g., 21st Century Cures Act), the blog post and white paper seem to indicate a degree of defensiveness on the part of FDA. The blog post, in particular, begins by stating that FDA’s drug approval process “is the fastest in the world.” It then relates that “[w]hen research does not offer answers to important scientific questions, cures cannot be developed. And when viable cures are not in the pipeline, focusing on regulation will not improve the situation . . . .”
There we are: the 21st Century Cures Act. That’s been passed by the House (and here’s the sponsors’ summary of it and case for it). And here’s policy wonk Norm Ornstein’s article on it in The Atlantic. He’s a fan:
But getting to “yes” was not so easy. First, it required a bundle of money—at a time when federal dollars are particularly difficult to come by. Many Republicans do not want to expand government of any variety, and the zeitgeist is to slash more, not spend more. The Brat Amendment reflected that view, and would have cut the innovations in the Act off at the knees. Second, the Act tries to reduce the daunting costs of bringing promising treatments through the lengthy and difficult process of laboratory work, animal trials, and several phases of human clinical trials—something that is opposed by many people who fear that any shortcuts could be dangerous, and who do not want to do anything that might reward pharmaceutical companies. Thus, such luminaries as David Kessler, the former head of the FDA, opposed the bill.
Kessler’s not alone. Here’s a worried editorial at the New England Journal of Medicine that cheers some aspects of the bill, but is not so happy about others. The authors note that the general emphasis on speeding everything up and approving more drugs runs the danger of relying on trials that are too small, too underpowered, too reliant on early-stage biomarkers, etc. And they have some strong points: if an Alzheimer’s therapy looks sort of promising in Phase II, but wipes out completely in Phase III (as has happened many times in the last few years), then the problem is not that the Phase III trial got in the way. Drugs that are going to benefit people actually work in big, well-designed Phase III trials, and a lot of things that look like they’re going to work before hitting them die off when they finally do. That’s a feature, not a bug.
So, defensive language aside, I think that the FDA has a point in its white paper. The reason we don’t have cures for a lot of diseases is not so much that pesky regulations keep great drugs from getting to the market. It’s that we don’t know enough – yep, even in 2015 – to make so many great drugs.