Who are our customers in this drug business? Well, sick patients, naturally. But their physicians, too, since they're the ones who will be writing the prescriptions. And the insurance companies, of course, since in most cases they're the ones who will be paying at least some of the bill. But the customer before we get to all that is the FDA.
Not many other industries have a gatekeeper that absolute. Every product has to be submitted and given an explicit, detailed review, with a thumbs-up or thumbs-down at the end of it. Imagine a car maker putting together a New Model Authorization package for each new model of light truck for Approval To Sell in the fall, or waiting for the Committee On SUVs to define the review criteria for Truck-Like Four-Door Crossovers before any of them can be sent to the dealers. A fast food chain wants to offer a new Double Taco Burger? Fine - submit the paperwork, and the agency will get right on it. The review committee on Fake Mexican Entrees meets in March.
The reasons for all this are no mystery. Anything that can directly affect a person's health, in either direction, is going to have a lot more oversight than the new Double Taco Burger, which will probably not kill you, although you may wonder about that forty minutes later. But given that we're going to have a large, complicated regulatory regime for new drugs, do we have the right large, complicated regulatory regime?
Steve Usdin and the team at BioCentury have a good article out that asks just this question. Since we were talking around here about the conditional approval for Avastin in metastatic breast cancer, that's a good example of how this new system might work: instead of binary decisions, the whole thing is adaptive.
The idea is to set things up so that decision-making data can be generated more quickly, and so that these decisions can be modified based on later findings. The big push in the early phases of the clinic would involve biomarkers - and yes, I know that everyone's been trying to do that, with rather mixed success. But the plan here is not to add on biomarker work, but to make it an integral part of every clinical program, with stored samples (and incentives to share them), and a clear regulatory framework for what the FDA wants to see in each case.
But since biomarkers aren't easy to come by, the next part of the plan is wider use of conditional approval and adaptive clinical trials. Another way to speed things up with adaptive designs would be to run several new therapies in a given space simultaneously, re-assigning patients as the more effective candidates show themselves. If the trials are going on continuously, the barriers to getting in on them would be lower than they are under the current system, where everyone has to start their own work from the ground up. Again, the idea is to be able to make some sort of decision as early as feasible, with the option of going back if later data don't pan out. (That's the key mental adjustment in the whole thing, actually - the willingness to act on the belief that, if done well, enough of these early decisions will turn out to be the right ones to outweigh the ones that aren't).
Conditional approval would have to be coupled with restrictions on marketing until more data came in - you couldn't just go crazy as soon as possible. But it would work both ways - a company would get wider authorization as the numbers got better, or would have to narrow things down or even pull a compound altogether. This would be a big adjustment for the public to make, frankly - I can already see the editorials going on about making the entire American public a group of test subjects, and so on. But you know, they already are: every compound that makes it to market is still an investigational drug, no matter what anyone might think.
There's more at that BioCentury link, and I encourage anyone who's interested to read it all. I think that there's a lot of merit in these ideas myself, although getting them implemented in the real world won't be easy. There's also the worry that half-implementing them would leave us with a system that's no better (or subtly worse) than the one we have now. Thoughts?