Life After FDA

You have to admit: Mark McClellan, MD, has quite a resumé. But after you've been the commissioner of FDA and the administrator of the Centers for Medicare and Medicaid Services (CMS), what comes next?

For McClellan, who was awarded a fellowship at the AEI-Brookings Joint Center for Regulatory Studies after leaving CMS, it's about finishing what he started. Under his leadership, the think tank launched the new Engelberg Center for Health Care Reform, which seeks to find practical solutions to the very big questions surrounding the safety, affordability, and quality of medical care.

Industry insiders see McClellan as a "less talk, more action" kind of guy and hope his new post will add horsepower to moving discussions about public policy reforms into action. (After all, Medicare Part D was enacted under his watch.) Here, Pharm Exec talks to McClellan about what it will take to improve the drug safety system and the challenges that face industry and regulators alike.

From your new position, how do you view the issues associated with drug safety?

This has been a big year for FDA reform with the FDA Revitalization Act. And one of the more challenging elements in there has been what to do about gaps in the nation's drug safety system. The usual debate at FDA is that you need more regulation because you need oversight of the drug companies and further government assurances to make sure that drugs are being used appropriately. On the flip side, if you put in too many regulatory barriers, it raises costs and reduces access to cures.

Part of the problem with drug safety is that we don't have a good system for learning about safety problems in a timely way. FDA has tried very hard, but it's difficult to rely on very busy doctors and hospitals and patients who don't always call into the so-called spontaneous reporting system to report adverse drug events. Take the case of Vioxx: It's unlikely that the doctor whose patients have heart disease and heart attacks would see that as a reportable adverse event for the drug. The result has been that we don't find out nearly as quickly as we need to about potential safety signals—and that's what's led to calls for alternative approaches for more intensive regulation.

What role could the Engelberg Center play?

We developed some ideas with researchers at Harvard, like Richard Platt, on how we could implement an active drug-surveillance system. If we had an active surveillance system, there could be more confidence that if there were an adverse event associated with a drug in actual use, it would be detected quickly.

These are ideas that get broad bipartisan support. On the one hand, it gives some real assurances that the safety problems are going to be identified and can be acted upon more quickly and reliably. On the other hand, it's about gaining better evidence about drugs using the electronic data systems that are already out there.

What would such a system look like?

There are pieces of it in place now. A number of health insurance plans have already been collecting electronic population-based data on the utilization of drugs and subsequent results for patients. Just look at the recent Avandia advisory-panel meeting. WellPoint presented data on something like 160,000 beneficiaries who have been on Avandia in comparison with other treatments for diabetes and found that there wasn't a higher rate of adverse cardiovascular occurrences.

It is possible now to pool together not just the experience of individual health plans like WellPoint, but also a much broader range of health plans that collectively cover more than 100 million patients. That's a much stronger database to use for surveillance of these safety signals.

This approach is focused on data mining, rather than conducting clinical trials. What are the potential issues there?

These are observational data sets, so just because you see an association does not mean that it's a causal relationship. It will be very important for there to be careful oversight of active surveillance systems by the FDA and expert advisors to the FDA to help identify which signals really are likely to be meaningful.

I mean, there are some adverse events that just shouldn't happen. Maybe you start seeing a significant number of cases of aplastic anemia or liver failure. That's a sign that we ought to be able to detect quickly and take action on. But often, there may be an association between a drug and adverse event that may just occur by chance. If you have large enough sample sizes, lots of things look statistically significant but may not be clinically meaningful.

Any safety surveillance system will require careful expert oversight to make sure that we're using these new data opportunities appropriately. Sometimes—or maybe often—this will be just a signal detection process. It won't resolve the issue, but it will help us identify where further clinical studies and more careful clinical evaluation is needed.

The problem with follow-on studies is that they take time—and leave physicians and patients hanging on without answers. My hope is that we'll be able to do that much more quickly than in the past and, as a result, be able to more effectively target the postmarket follow-up studies that are needed.

We shouldn't have to wait five years, until a drug is used by millions of patients, and still not be able to get to a more definitive answer—like what happened with Vioxx. So I don't think we should expect to wait a long time to get the needed information. We ought to have a mechanism in place to quickly conduct any follow-up confirmatory clinical studies that are needed.

What are the most promising ways to integrate surveillance systems with what's already in place today?

You can begin the surveillance process at the time when the drug is approved. For example, in the case of Vioxx, there were concerns about possible cardiovascular adverse events during development and at the time of FDA approval. There was a relative signal that [made it seem] plausible that events could occur with long-term use and maybe in certain patients. There was also the possibility that it may not have been a real effect at all. If we had this system in place when the drug was approved, the FDA could have highlighted the importance of tracking the occurrence of cardiovascular events in patients using the drug who had heart disease.

According to some simulations done by Richard Platt in conjunction with this work we're doing, the safety signal for a higher rate of serious cardiovascular events with Vioxx could have been detected in three months—rather than four to five years—with this kind of large-scale population network in place.

When it comes to execution of a drug safety surveillance system, where are the places FDA can fall down?

We're entering a new era here—and not just because FDA is now moving to active surveillance and much more routine and systemic oversight. But we're entering into a new era in terms of the ability of anyone to look on the Web or run an analysis and find a statistically significant association between a drug used and an adverse event. A lot of times, that won't be a causal association. Did it occur by chance?

And if FDA doesn't have an aggressive mechanism in place to help the public distinguish between adverse-event associations that occur by chance and those that may really mean something, then it doesn't have a way of giving the public confidence that they will be able to get to the bottom of the issue in a timely and effective way. That's going to lead to real challenges for effective drug development.

Does FDA have what it needs to make the change?

I think the good news is that the agency is finally getting a down payment on all the additional resources that it desperately needs [through the FDA Revitalization Act] and all the additional opportunities to develop new scientific and technical capabilities to deal with these new challenges. But it also presents some major adjustment challenges.

That will include developing new skills. It will also include developing new ways of interacting with companies and the approval process, especially in the postmarketing process. Things are changing. It will not be business as usual in terms of drug approval and even postmarket planning.

What impact do you predict on new systems for adverse-event reporting and detection?

There may not be any immediate effect—but there's at least the potential for some significant improvements for a couple of reasons. If FDA is able to implement an effective active surveillance follow-up system, issues that are properly dealt with in the postmarket setting will be dealt with in the postmarket setting. In other words, you won't see as much reluctance to approve a new drug when there are some slight doubts that are, again, best resolved when you can see how a drug is used in practice. I think it might actually speed up the approval process.

Also, the new Revitalization Act includes a framework for FDA to develop the premarket science of showing that drugs are safe and effective and can be used reliably.

Part of that gets to the heart of an initiative you started at FDA—Critical Path. Where do Critical Path and the issue of safety intersect?

When we started the Critical Path Initiative, it was about trying to bring some of the new sciences, like genomics, information technology, and imaging technology, to the forefront to find better ways to predict which drugs are going to work in which patients. And the new legislation includes strong bipartisan support and some funding and new opportunities for public and private collaboration to develop this science.

There are some Critical Path Projects that are now up and running. There's a biomarker initiative that's operating through the support of the Foundation for the National Institutes of Health. It's leading to some good new opportunities for improving the development process for cancer drugs. It's been an initiative that Ray Woosley's group at the Critical Path Institute in Arizona has supported for doing a better job of predicting which patients are going to have sensitive responses to what.

There's also a project looking at warfarin and how to optimally dose patients, which uses genomic information on liver-enzyme activity. It potentially has implications for predicting adverse events for other drugs that involve liver metabolism.

Those are just a couple of the examples that are in place now. With the new law, there's an opportunity for funding, for public/private collaboration, for major new initiatives to bring better science to many other areas of product development and use. There's certainly been a lot for industry to do, as well as new steps at FDA that will be needed to take advantage of that.

Despite these successes, it seems like the Critical Path has a large potential to stall out. What do you think?

Well, I'm an optimist at heart. And I think there's too much at stake for it not to work. It's certainly easy to think about barriers to more effective innovation and the needed regulatory enhancements. But what's really at stake is nothing short of major improvements of the health of the population. And there are so many opportunities in biotechnology today for developing better cures and more personalized treatments, for preventing the complications and the chronic diseases, that I don't think we can miss out on those opportunities for too long.

It is going to be some real work to take advantage of these opportunities. But, you know, the fact there is such strong bipartisan support for the legislation at a time when there are lots of disagreements on healthcare generally and lots of rancor in Washington shows that there's some real potential for doing things differently.