Remarks by

David A. Kessler, M.D.

IOM 25th Anniversary Lecture

Seattle, Washington

November 7, 1994

I'm honored to be here today to inaugurate the Institute of Medicine's 25th anniversary lecture series and to celebrate the extraordinary contributions of the IOM. For this nation has a scientific and medical genius unparalleled in the world, and it is writ large in the accomplishments of the very distinguished members of the IOM who are in this audience today.

I am going to begin my remarks by telling you about a meeting that took place a few months ago in a conference room on the third floor of the FDA building in Rockville, Maryland.

I sat and listened to a group of AIDS activists make an extraordinary plea: they urged me not to approve HIV drugs too quickly. It was a remarkable meeting--one that I don't think anyone following events over the past decade would have predicted.

As I sat there, I thought back to a scene only a few years earlier when there were AIDS activists literally scaling the walls of the FDA building in Rockville, Maryland, demanding access to potential therapies that had barely moved out of the test tube.

In stark, personal terms, patients with serious and life threatening diseases have told us repeatedly in recent years that they do not have time to wait as FDA dots every "i" and crosses every "t."

They told us they want to make more of their own choices about the therapeutic risks they are willing to take. To many patients, experimental therapies represent a rare source of hope for conditions that seem to have outsmarted modern science. When a doctor looks a patient in the eye and says there is no cure and no therapy on the market that can interfere with the disease's deadly course, the unproven therapies in FDA's pipeline can take on enormous importance. In the wake of the AIDS epidemic, there have been marked changes in the way we approve drugs. In the last several years, the agency has approved four antiviral AIDS drugs, each in less than seven months: AZT, ddI, ddC and d4t. We have also expanded access to unapproved therapies.

Now some AIDS activists were saying wait...wait a moment.

What I would like to do first today is review some of the dramatic changes we have made in the way we do business when it comes to therapies for serious and life-threatening diseases. And then I will discuss some of the difficult scientific and policy challenges these innovative approaches raise--challenges that underlie the apparent turnabout by the AIDS activists at that meeting.

The FDA is responsible for ensuring that the therapeutic agents used in the United States are safe and effective. An agent must be, according to the statute, "safe for its intended use." This statutory requirement is interpreted to mean that a therapy's known and potential benefits outweigh its known and potential risks, under the proposed conditions of use and considering the risks and benefits of other available treatments.

In addition, "substantial evidence" of effectiveness, based on adequate and well-controlled clinical trials, must be demonstrated. Typically, at least two trials are required.

Approved drugs must also have "adequate directions for use" so that the therapy can be administered properly. These data are gathered through clinical trials divided into several phases.

[slide 1] In phase I, a drug is typically tested in a small number of subjects to determine appropriate dosage and pharmacokenetic information and sometimes to gather other preliminary data.

Phase II involves the first control led trials looking for efficacy in the intended patient population.

Results from phase I and II studies guide phase III studies, which can involve several thousand patients. Sponsors usually submit a new drug application or NDA after Phase III studies demonstrate clinical benefit.

I should add that Phase IV studies, conducted after a product is marketed, can also be an important source of additional information.

This system has served us well for the vast majority of therapies ordinarily coming under FDA review. But the established approach to the latest beta blocker or anti-ulcer medication is not necessarily the best model for potential breakthrough products.

I would argue that in urgent circumstances--when patients are dying and no effective therapy is available--something more is required of the FDA.

And so we have changed the rules. Over the last decade we developed mechanisms that give seriously ill and dying patients access to potential therapies before they are approved for marketing.

And we have established a new policy--accelerated approval-- that allows for the approval and marketing of drugs for serious and life-threatening diseases without the full data set that we would ordinarily require.

I want to emphasize an important point here: In the end, we hold the products approved under accelerated approval regulations to the same standard as traditional approvals. Adequate and well-controlled trials must in the end demonstrate clinical benefit. That's the goal line. That's what we ultimately need to cross. But as we push toward that goal line, there are steps that we can take to provide access to drugs for serious and life-threatening diseases.

It's useful to remember that data on a new drug accumulate gradually. Identifying a new therapy is not like pulling a prize out of a box of Cracker Jacks: One minute you don't know what you will get and the next minute there it is, fully revealed.

Information is cumulative. The decision for FDA, is where on that continuum drugs for serious and life-threatening diseases can reasonably be made available. The next two slides show the mechanisms we have used to expand access to these potential therapies.

[slide 2] A number of informal mechanisms have been in place for several decades. A Group C designation allows cancer patients to gain access to unapproved drugs.

The Emergency IND designation and open-label safety studies are other tools available to desperately ill people. With the emergence of AIDS, we realized that existing approaches to expanded access needed to be formalized and augmented.

[slide 3] In 1987, we officially established the Treatment IND to make drugs available between the time clinical studies are complete, or nearly so, and the time FDA approves the drug.

Sponsors are allowed to charge for therapies that are designated as Treatment INDs. One problem we've encountered, though, is that most insurers don't reimburse for these drugs because they are still considered experimental.

The last two items here--parallel track and accelerated approval--have also been hallmarks of our review and approval of drugs for HIV infection.

Parallel track allows HIV-infected patients who cannot join controlled clinical trials for one reason or another to get promising therapies that are still under study. This has significantly increased access for patients who cannot travel to the site of the clinical trial or who do not meet strict entry criteria, for example.

Many people do not realize that tens of thousands of HIV- infected patients have received unapproved therapies through the parallel track. Industry is not permitted to charge for these experimental drugs, which are usually supplied by private doctors or clinics.

In December of 1992 we issued our accelerated approval regulations for serious and life-threatening diseases. They allow us to approve a drug on a conditional basis without direct evidence that it affects clinical endpoints, such as survival or disease progression.

[Slide 4] You can see from the regulatory language here, that clinical trials must establish the product's effect on a surrogate endpoint that is "reasonably likely" to predict clinical benefit.

A surrogate endpoint or surrogate marker is a laboratory measurement or physical sign that is influenced by a therapy and thought to correlate with clinical outcome. There are various examples from fields such as cancer or cardiovascular disease. Blood pressure is a widely accepted surrogate marker.

In AIDS research, we have tended to rely on CD4 cell counts, a measure of the strength of the immune system. Our use of surrogate markers to accelerate approval comes with certain provisos.

[slide 5] One, the surrogate must have a "reasonable likelihood" of being correlated with actual clinical benefit. Two, approval is conditioned on the completion of postmarketing studies to verify the anticipated clinical benefit. And finally, if the sponsor does not demonstrate clinical benefit in a timely manner, accelerated approval will be revoked. This slide shows how the flow chart for drug approval changes.

[slide 6] Today, drugs that show promise in phase I and II trials are being made available through parallel track or treatment INDs to dying patients prior to approval, and where sufficient data are available, they may qualify for accelerated approval.

[slide 7] Notice the threshold to make a product available at each stage. Drugs to be distributed through the parallel track must be "promising."

For a drug to be made available under a Treatment IND, there must be evidence that it "may be effective."

Both parallel track and treatment IND states do not bestow approval. For a drug to be approved--given full marketing status before clinical trials are complete--there must be an effect on a surrogate endpoint that is "reasonably likely" to predict clinical benefit. Traditional approvals require "substantial evidence" of effectiveness based on adequate and well controlled trials demonstrating clinical benefit.

It is critical to underscore here that drugs receiving accelerated approval must in the end meet this traditional standard. You can go on the market, but your approval is conditioned on your ultimately showing clinical benefit.

These, then, are the policies and regulations we have put in place. But how do they work when they come off the page? I can tell you this: they have not had a chance to gather dust.

Tens of thousands of patients have received experimental drugs through our expanded access programs. And we have approved five drugs through accelerated approval. [slide 8] Three anti- virals for HIV infections--ddI, ddc, and d4t -- Clarithromycin for MAC, an opportunistic infection often seen in AIDS patients, and Beta interferon for multiple sclerosis.

[slide 9] Although each product is going to look a little different, this timeline for ddI offers a good illustration of how the policies can work in practice. DdI actually was approved about a year before the final accelerated approval regulations were issued, but it served as a model for those regulations.

As you look at the timeline, it's useful to recall where we were with AIDS treatment in the late 1980s when ddI was in the pipeline.

There was one approved drug for AIDS--AZT.

It was not a homerun drug. We knew that then; we know it now. Many AIDS patients had stopped receiving any benefit from AZT. Some were never able to tolerate it. There was an urgent need for additional treatments, and what we saw on the horizon was ddI.

In September 1989, ddI was made available on an expanded access basis while trials continued. Although research was not the primary intent of parallel track, we realized it was possible to capture extra data on patient experience through this mechanism. For that reason, we found a way for physicians to secure a lot of safety information with a minimum amount of paperwork. DdI was approved partly because Bristol Myers secured safety data on nearly 8,000 people outside formal trials. In all, 26,000 patients received ddI on parallel track.

In April, 1991, Bristol Myers submitted an NDA for ddI based on phase I studies in 170 adults and 98 pediatric patients.

Usually, we expect to secure only safety information from an uncontrolled phase 1 trial. We generally don't learn much about efficacy. And we certainly don't expect to make an approval decision. Within a year, we knew we would have much more definitive information from ongoing, controlled trials.

But for many, a year would come too late. We sought out whatever data would help fill in the picture that was emerging from the phase I ddI trials. And it came from ACTG 116B/117, one of the controlled studies underway. In that study, patients who had received prolonged prior AZT therapy were randomized to either ddI therapy or continued AZT therapy.

The researchers allowed us to take a highly unorthodox peek at CD4 cell counts of those patients. Although we were not sure of the precise relationship between a drug's effect on CD4 cells and its effect on survival, there was some scientific evidence supporting the link.

Less than six months later, in October, 1991, we approved ddI, primarily because it increased CD4 cell counts slightly.

However, we placed important conditions on our approval. We told the sponsor that it had not yet demonstrated any clinical effect from ddI. We insisted that the company analyze and submit more data when the ongoing trials were complete.

We also said ddI would be withdrawn from the market if the expected results did not materialize. In seven months, in April 1992, the company presented persuasive evidence of ddI's clinical benefit. Along with confirming the earlier approval decision, their data allowed the FDA to expand ddI's indications and to recommend a lower and safer dose.

In many respects, ddI is a success story. An effect on clinical endpoints supported our conditional accelerated approval based on the CD4 counts. But make no mistake: accelerating drug approval can also have risks.

There are tradeoffs. For example, to this day, we don't know when the optimal time is to switch from AZT therapy to ddI, or even if such a time exists. That takes me back to the AIDS meeting I referred to earlier. I started by telling you about that meeting because I think it says better than I could that it is not enough for FDA to get potential therapies to patients sooner.

It's not enough to simply open some regulatory gate so that the products can pass through to patients waiting on the other side. That's the easy part. We can provide access. But we also have to make sure we get the answers--answers to fundamental questions about the value of the product.

The simple fact is: The less time that a drug is in clinical trials, the less we know about it. And that means the more risks we are taking when we make it available.

The challenge is striking the right balance between access and answers. And one day we are going to make a mistake. One day we are going to make a drug available that doesn't work. And we will have to remove it from the market. That's a risk we are taking. But when it comes to getting therapies to dying patients, the riskiest thing we can do is not to be willing to take any risks.

Patients have told us they are willing to take risks, too. They could end up wasting precious time and even be harmed taking an ineffective product.

But they want reassurances, too--I understand that. They want reassurances that the risks they take today will yield answers tomorrow.

We have a responsibility here--not only to the AIDS patient facing imminent death who may feel there is nothing to lose in taking an unproven drug. We have a responsibility to patients today and tomorrow who need the results of studies to guide their therapeutic choices.

The activists at that meeting want early access. "We are not here to drive a spear into the heart of accelerated approval," they told me.

But what I also heard was a passionate plea to know more, sooner, about whether an experimental medicine really works-- and for whom, when, and in what doses.

They did not want the price of early access to be less information, in the long run, about the value of these drugs.

They argued that we may even need to delay drug approvals and perhaps conduct larger studies to get more answers.

Certainly the AIDS community is not in full agreement about how to strike that balance. In subsequent meetings with other activists I heard great concern expressed about any policy changes that might curtail access to, or approval of, therapies.

Let there be no doubt: FDA stands firmly behind accelerated approval of drugs for serious and life-threatening diseases.

But that commitment also carries a responsibility to ask the tough questions along the way, to glean insights from our experience, to stay on the cutting edge of science in this area, and to adjust the policy when that is what is required.

That is what is happening now. And it comes as a new class of antiviral treatments for HIV disease--protease inhibitors-- are likely to be the next candidates for accelerated approval.

We are not where we were in the late 1980s when AZT came before the agency. Then, there was a desperate clamor for anything that might be construed as a therapy. In those days, a drug that could jostle CD4 counts the slightest bit was hailed as an important advance.

I don't hear people arguing today that we made a mistake in approving any of the AIDS drugs now on the market. But there is no question now that some of that early hope has given way to frustration. We need something better.

In September, FDA's advisory committee on anti-viral drugs held a two-day public meeting on expanded access and accelerated approval. More than 200 members of industry, academia, patients with serious or life-threatening diseases and their advocates attended. Nearly fifty came up to the microphone. Some came to tell emotional stories of lives lost waiting for effective treatment. Others made sophisticated presentations proposing new study methodologies or summarizing research.

Virtually all lent support to the innovative policies, but all faced the same question: how can we ensure early access and still get the information we need about the drug?

In the time remaining, I want to discuss a few of the issues that we are grappling with as we try to strike that delicate balance.

First, let me touch on confirmatory studies and innovative study design. Although FDA has a clear idea of the ideal NDA package--the kind of data we would like to have on a new drug, we recognize that often that package is simply not feasible.

And so we have been willing to consider data from trials that were not designed to produce the comprehensive data that we routinely require. We have been receptive to innovative designs that can speed drug approvals.

Although we prefer double blind controlled trials, we recognize that sometimes historical controls can be used, and sometimes blinding is not possible. Sometimes trials have been telescoped, blurring the usual distinctions between phase 1, 2, and 3.

Considerable attention is now focussing on large simple trials as a way to gain information on more patients sooner. The entry criteria for large simple trials are usually not very restrictive, the system of data collection is simple, and endpoints are easy to document. They are designed so that individual community physicians can participate.

In general, the value of a large, simple trial is that it provides a large data base on how a drug is best administered in a clinical setting. However, physicians may be lax about monitoring protocol compliance. Consequently results may be difficult to interpret.

Some people say large simple trials should be a requirement for accelerated approval. Others argue that the design works best for postmarketing studies. Ultimately it depends on the type of questions being asked.

There are also those who argue that until more effective drugs are developed, there is little to be gained from such large trials. When better therapies come out of the research laboratories small trials will be able to demonstrate their benefits just as well.

A variation is to try to gather the kind of data one gets from large simple trials from parallel track. Some say that any drug likely to be presented for accelerated approval should be available first through parallel track students. Critics argue that this may cost too much. They are afraid it might discourage some small companies from participating in the race for new drugs.

A related issue is postmarketing confirmatory studies. I've emphasized that in the end confirmatory studies must demonstrate that a drug approved conditionally on the basis of surrogate markers has a clinical benefit.

I think we at FDA can do a better job of making sure that those confirmatory studies are on their way at the time of the conditional accelerated approval and that we receive the necessary data in a reasonable time period. At the same time, we cannot deny that getting answers can be complicated by the availability of the drug through expanded access.

Patients who feel a new therapy is their only hope are not necessarily willing to be randomized into a trial where they might get standard therapy--or for that matter, an approved drug they don't trust. If the experimental therapy is widely available outside trials, they can get access to it without entering a trial.

The reality is, without a homerun drug, or at least more effective drugs, many HIV patients are designing therapeutic regimens from a variety of proven and unproven products. Many want to retain this therapeutic control and the flexibility to vary their therapeutic approach according to their response.

Participating in a study with a strict protocol may preclude that.

Toward this end, Dr. Ellen Cooper, a former colleague at the FDA who is now with the American Foundation for AIDS Research, proposes large, but not-so-simple trials.

She is advocating trials that would be fluid and flexible, incorporating new pharmaceutical regimens as they become available. While theoretically appealing, the practical difficulties of such a trial could be significant, to say the least.

Another enormously important area of concern is the role and validity of surrogate markers. FDA's policies are only as good as the scientific base they build on. And there is an enormous need to better understand the value and use of specific surrogate markers. Accelerated approval is based on the idea of relying at least in the interim on surrogate markers until data on clinical benefit are available.

As someone recently put it, surrogate markers are not magic markers. We know that they are not always predictors of clinical outcome.

Many of you may recall CAST--the Cardiac Arrhythmia Suppression Trials. That study provided a sobering reminder of their limits. It involved drugs that suppressed extrasystolic ventricular contractions. We know that people with these arrhythmias are at increased risk of death following a heart attack, so intuitively it seemed that a reduction in irregular heartbeats would predict reduced mortality. In fact, the drugs that reduced arrhythmias were associated with a 250% increase in mortality.

One of the problems with surrogate markers is that they give us only a partial picture of the effect of a new therapy. The same therapy that has a positive effect on one component of the disease--could cause countervailing or simply different effects on other components of the disease.

CD4 counts have been the most frequently used surrogate marker for AIDS therapies, but they now appear to be a weaker predictor of clinical outcome than we would like. A drug may alter CD4 counts without having much affect on progression of AIDS or survival. A false positive problem.

Researchers generally believe that the false negative problem--no affect on immune cells but an important effect on clinical endpoints is less likely. But we know this is a risk of surrogate markers as well, and can lead to a drug being abandoned prematurely.

This was illustrated in the case of gamma interferon, approved to treat chronic granulomatous disease--CGD--a rare immune disorder seen in children.

Researchers expected it to affect superoxide production and the destruction of bacteria by blood cells, and thereby alleviate the life-threatening infections that characterize CGD. A few weeks into the trials, it was clear that the drug would not affect these laboratory findings. But the trial continued, and the results were surprising: after a year, gamma interferon had clearly reduced the rate of infection without demonstrating any detectable effect on either superoxide production or the destruction of bacteria as measured in the blood cells.

Possibly, the effects on these surrogate endpoints were smaller than could be measured or maybe the drug worked in unanticipated ways.

There is much more to be done to validate CD4 counts and to understand the role of the immune system in the mechanism of action of AIDS.

Earlier in my remarks I mentioned blood pressure as an accepted surrogate marker. But consider for a moment the difference between it and CD4 counts. A therapy that affects blood pressure hopefully exerts an effect over an extended period of time. But with CD4 counts, the effect may be transitory or it may occur only at certain points in the natural history of the disease.

To complicate matters more, even if we identify how CD4 counts might be most effectively used in drug therapies, the same rules may not apply for vaccines where the mechanism of action is different.

Some researchers are now focussing on viral markers as an alternative or an adjunct to immunologic markers. Viral load markers have been used widely in the early stages of drug development to determine whether what is biologically active in the test tube exerts effects in vivo. But in the last couple of years we are seeing some promising developments in the use of viral markers to monitor disease progress in later stages.

A number of technologies are proving to be sensitive to detecting increases in viral RNA or the amount of circulating virus. These markers show promise as a way of rapidly identifying a drug response and also detecting resistance. If a potent anti-viral is administered, we should be able to see a response in a matter of days. Viral markers offer a more direct measurement than CD4 counts.

We recognize that viral markers also have important limitations. For example, they measure viral load only in serum, but in the early stage of infection the virus is in the lymph nodes where these markers don't reach. Nonetheless, this is a field of enormous activity and importance.

A third concern regards industry incentives. When we put accelerated approval in place, we thought it would lead companies to invest research funds in AIDS development because they could bring drugs to market sooner.

But something else has come into play. Our regulations provide for accelerated approval for drugs to treat patients who have no effective therapies available. Once AZT was on the market, companies tested drugs on populations who were not responsive to AZT to gain accelerated approval.

And with each successive drug, the populations studied have narrowed. Of course, that leaves open the question of whether a drug that was effective in a narrow population in clinical trials would be useful to the many other patients with the disease who don't meet those criteria.

The fear is that while we hoped to be encouraging breakthrough drugs for AIDS, instead we may have inadvertently encouraged companies to go after small improvements in existing therapies that have been shown to benefit only very small portions of the affected population.

Finally, I want to raise one more issue that involves matters of both science and equity: underrepresentation of the poor, minorities, and women in expanded access programs and clinical trials of AIDS therapies.

I think some historical perspective is important here.

When we see that minorities are underrepresented in studies, we worry about inequities in the system. But we hear from many representatives of the African American community that the Tuskegee revelations still have a powerful hold...that there is a widespread reluctance to enter trials for fear that injury and injustice will again be visited upon them. Somehow, we need to bridge that gap, while we also increase opportunities for more representative participation.

But underrepresentation of minorities, women, or children is not just a matter of equity and access. It is of concern to us as scientists as well. There may be genetic differences that translate into different drug affects along gender, age, and ethnic lines. We need to know more about drug interactions within certain subgroups.

I'm going to turn the projector back on to show one more slide because I think it is an important one. [slide 10] It shows that we still have a ways to go in making our trials representative of the patient population.

Women and minorities are also poorly served by the parallel track and Treatment IND mechanisms. Eighty-five percent of the people who received d4t prior to approval were white. Obviously, that figure does not reflect the demographics of HIV infection.

As a practical matter, tens of thousands of people with HIV disease are cut off from expanded access initiatives because they do not have private doctors or other sources of comprehensive care.

On a more encouraging note, recently several AIDS activists from the rural South reminded me how accelerated approval is increasing access for some underserved populations. They said it was a day's journey to the nearest doctor willing to prescribe experimental therapies. Unless they had the time and the resources to make that trek, they had no way to secure promising, but unapproved, drugs.

Those of us from major East Coast and West Coast cities need to remember this reality. AIDS has no geographic boundaries. Access to cutting-edge medical care still does.

I have no doubt that accelerated approval and expanded access is the right policy to pursue. But I am equally sure it cannot give us all the answers we would like about the best treatment strategies for AIDS or other serious and life- threatening diseases.

It is important to appreciate that this requires a commitment to longer term studies and studies of multiple therapeutic regimens that are generally beyond our regulatory reach. But that has been true for most entities. There is a notion in this country that up until the day a drug is approved it is unsafe and ineffective and all of a sudden on the day we approve a drug it becomes safe and effective forever more.

For our part, we are reviewing various options to refine the expanded access and accelerated approval policies. But nothing we do will compromise our commitment to our high standards for drug approvals: all approved drugs, even those for serious and life-threatening diseases, are ultimately required to meet the agency's traditionally rigorous safety and efficacy standards. I can't stress that enough.

Getting drugs to patients quickly is not always as simple as it seems. That was brought home to me recently when I was reading an article in Barrons. In it, a Columbia University professor was quoted as saying: "the person who thought of accelerated approval and saw to its acceptance should be shot."

Nobody ever promised me this job would be easy.

####