Opinion | Competing Incentives of the FDA’s Accelerated Drug Approval

In part 1 of this Instagram Live interview, Jeremy Faust, MD, editor-in-chief of MedPage Today, sits down to talk with Holly Fernandez Lynch, JD, MBe, assistant professor of medical ethics and health policy at the Perelman School of Medicine at the University of Pennsylvania in Philadelphia, and Aaron Kesselheim, MD, JD, MPH, director of the Program on Regulation, Therapeutics, and Law (PORTAL) of Harvard Medical School and Brigham & Women’s Hospital in Boston, about the FDA’s accelerated drug approval process.

These experts discuss successes of accelerated approval, challenges of confirmatory trials, and the competing incentives faced by patients, pharmaceutical companies, and the FDA itself.

The following is a transcript of their remarks:

Fernandez Lynch: The idea behind the accelerated approval pathway is that it’s a compromise that’s intended to allow products to come on the market more quickly before we are totally confident that they are effective. The compromise part of it is early access and then later confirmation of benefit.

So when a product gets FDA approval, it has to be demonstrated [to be] safe and effective. That requirement stands for the accelerated approval pathway as well.

But rather than relying on measures of how patients feel, function, or survive — those clinical endpoints that really matter to patients — accelerated approval can be based on surrogate markers that are predicted to demonstrate clinical benefit, but not yet proven to do so. So it might be something [like a] biomarker, it might be some level of protein in the blood or some result on a scan that is not a direct measure of how the patient feels, functions, or survives, but gives some signal, we hope.

But because it’s not a proven surrogate measure, the requirement is for companies to continue studying their accelerated-approval drugs after they’re approved for marketing to get confirmation of benefit. Now, we’ll talk a lot more about this, if a company fails to do its confirmatory studies with due diligence or the results of those studies fail to confirm benefit, the accelerated approval pathway also offers the possibility of an expedited withdrawal process so that FDA could pull back that accelerated approval.

Faust: Perfect. I know you’ve both written about some of the problems with this pathway. If we were here as just big cheerleaders, I don’t think that would be all that interesting.

Aaron, you’ve written a lot about some of the problems, but let’s start with a win. Tell people maybe the first example of how this pathway came about. I think it’s famous and known to some of us, but not to everyone. Or maybe other examples where you think this process actually worked.

Kesselheim: Well, I think there are actually a lot of good examples where the pathway works.

The pathway was originally created after the HIV epidemic of the 1980s as a way of trying to expedite availability of drugs that would reduce the amount of HIV in the blood. But we didn’t know yet whether or not they actually helped people with HIV live longer. And so in that context, it made sense to allow these drugs — the risk was worth it, so to speak — to allow these drugs on the market sooner when all we knew was their effect on the surrogate measure rather than the effect on the ultimate clinical endpoint of the patient.

For a lot of those early HIV medications, it did end up turning out that in subsequent studies if you reduce HIV virus in the blood, that you reduce patients’ morbidity and mortality with the disease.

Then subsequent to that, it was applied much more commonly among cancer, where in some cancers if you did not show progression of the cancer on the new drug, it would be presumed that that would mean you would live longer.

Then there are some really good examples, like Gleevec [imatinib] is a really good example, of a drug where the drug [initially] showed major improvement for patients who didn’t have any other treatments available to them. Those were subsequently shown to actually translate to extending patients’ lives.

So, there are a lot of good examples out there of drugs approved through this pathway that end up turning out to be transformative products that really [are] important for the care of patients.

Faust: One thing that I think is interesting to think about is the scale of conditions that we’re considering. On one hand, you mentioned the first one, the AZT [zidovudine] for HIV/AIDS, and this was something where you had thousands and thousands of people dying a year for want of anything that worked. And then on the other side of the coin, this pathway is often used almost for the exact opposite situation, which is that these are rare diseases in which a large clinical trial really isn’t practical or it’s hard to control.

Is that a problem, Holly, just in terms of the quality of evidence that we’re relying on? Because yeah, it’s easy to run a trial and just expedite it when you’ve got a denominator of a million people, but for something with a really small population, is it problematic to rely on this kind of a thing?

Fernandez Lynch: You know, I’m not sure that the size of the patient population is especially more challenging for the accelerated approval pathway versus the traditional approval pathway. The challenge that you’re describing — small patient populations, not enough really to kind of have the statistical power that you might be looking for — that would apply regardless of the endpoints that you’re looking at.

The difficulty is that when we’re dealing with these unproven surrogates and small patient populations, it’s kind of a double whammy of uncertainty, right? You have low-powered studies and you have endpoints that we’re not totally confident about. So we may be making additional trade-offs in those domains, but they could be reasonable trade-offs.

I mean, the main issue — well, I would say [there are] two main issues with accelerated approval. One, we are not going to be looking for perfect endpoints. If they were perfect, then we would be relying on traditional approval. But, do we have some reasonable confidence in the surrogates that are being used for accelerated approval on the front end?

And then the other challenge is on the back end, making sure that those confirmatory studies are well-designed and completed in an appropriate time frame. That timeframe is going to be different depending on how large the patient population is, right? So again, if you have a very rare disease, you might need more time to do your follow-up confirmatory studies.

But just recognizing that this is a trade-off when we’re having lower quality evidence on the front end, we have to be really vigilant to make sure we are getting that confirmatory evidence on the back end.

Faust: Can we pick up the confirmatory trial piece right there? Because that is such an important part of this, right?

The idea of the pathway, and please correct me if I’m wrong, is that the FDA will accept something on the basis of something other than a clinical outcome: a biomarker, some proxy, or a “surrogate endpoint” is the official terminology. But then at the same time, the companies are required to eventually show some kind of clinical benefit, ideally in a randomized trial, although not always.

Can we talk a little bit about the problem of that part of it, the confirmatory trials? I believe the federal government itself said that we’ve got something like 34% of drugs that have gone through this pathway whose confirmatory clinical trials haven’t even been completed on time, let alone some that were done and had to be withdrawn. Can you talk a little bit about where things stand on the confirmatory trial piece?

Kesselheim: Sure, I can jump in. As you said, part of the accelerated approval quid pro quo is that you let drugs get on the market quicker based on changes to surrogate measures that are not well-validated in exchange for the conduct of a confirmatory study that goes on while the drug is on the market and is being made available to patients.

I think the two major issues with those are that those confirmatory trials can take a long time — sometimes they take longer than what the FDA expected, in which case, as you point out, they can be delayed. Then the other question is what those confirmatory trials do. Unfortunately, what we found in prior research in our group is that some of these confirmatory trials don’t actually test clinical measures. They continue to test surrogate measures.

Then even more importantly, some of these confirmatory trials turn out to be negative, and then nothing happens. The indication then isn’t taken off of the drug labeling or some other action taken.

All of those things are where this kind of quid pro quo confirmatory trial conduct issue runs into trouble.

So I think that generally speaking, it might make sense to allow faster access to some of these products on the basis of uncertain evidence. If everybody knew that there was going to be a follow-up study coming along relatively soon that was actually testing meaningful endpoints and that was going to be acted on, where if it’s positive the information would be integrated into the labeling and patients could continue to use it, or if it’s negative the approval of the drug for that use could be reexamined.

I think on all of those points, there have been some suboptimal outcomes in the last few years.

Faust: Is there a shock clock thing happening here, Holly, where the companies are dragging their feet knowing that they’ve got a patent for a certain amount of time and they just want to maximize market penetration as fast and as much as they can? And if they drag their feet on the confirmation trial, then even if eventually it doesn’t work out, they got most of their upside by just having it out there before the trial actually got done?

Fernandez Lynch: Yeah, I think that there is an incentive to challenge here.

Once a product gets accelerated approval, it’s on the market and companies can charge whatever they want and, in most cases, payers will pay it, depending on who the payer is, right? That’s a key incentive.

If the product is already approved and you’re profiting at the maximal level, completion of the confirmatory trial does not stand to benefit the sponsor. At best it confirms benefit, and then that maintains the status quo. At worst, it will damage what the company was pursuing and show that this product doesn’t actually work. So why would we move quickly to show that the drug doesn’t actually work? Similarly, maybe it will be ambivalent or the results will be ambiguous, right? That also could only be a downside for the company.

So if you just think objectively about the incentive structure for companies to pursue these on time, they don’t have much kind of internal incentive. Now, I don’t mean to paint them all as bad guys. Presumably, they’re hoping that their products are truly beneficial. But from an economic perspective, the incentives are misaligned.

So how could you address that incentive? Well, FDA could really hold companies’ feet to the fire to say, “This is your deadline. You failed to meet your deadline, or if the product does not actually confirm benefit, we’re going to yank approval.” That would be a way to actually push companies to do these studies on time.

But for a variety of reasons, in the past, FDA has not always been willing or able to put pressure on companies. That’s often because it’s hard to get the genie back in the bottle once a product is on the market for one of these conditions. That accelerated approval can only be used for serious or life-threatening diseases and conditions with unmet treatment needs.

So if nothing better has come along in the meantime, you’re going to have patients and clinicians who say, “This drug seems to be having an impact on my patient. Or maybe it is, maybe it isn’t, but we don’t have anything better.” That can create pressure on FDA to leave these products that are kind of lingering on the market; we see some pushback in that regard.

Then the last challenge in terms of incentives is that you need patients to be willing to participate in confirmatory studies, but participating in a well-designed, rigorous, randomized controlled trial can be burdensome for patients. If I am a patient in need and I have an accelerated-approval drug that’s on the market, my choice is maybe I can get my doctor to prescribe this and my insurance company to pay; that seems like the path of least resistance. Why am I going to enroll in a study and have additional visits and all the rigmarole that comes along with that? That creates another incentive structure challenge.

To the extent that we could help patients understand that there is residual uncertainty about these products, that might help. On the other hand, again, if there’s nothing better, maybe that uncertainty wouldn’t actually change people’s minds and they’d still rather just take the approved drug.

So we kind of have this trifecta of concerns in terms of incentives for patients, sponsors, and FDA that make it difficult to collect data that we need on the post-approval side.

Please enable JavaScript to view the

comments powered by Disqus.