FDA Meeting on Approving Immune Therapies: Background and Comment.

On October 16 the FDA's Antiviral Drugs Advisory Committee met to explore issues around what advice the Agency should give to industry on development and approval of immune-based treatments (see announcement in AIDS Treatment News #351, September 22, 2000). It seemed widely agreed that the meeting went well, and was important and successful in that it happened at all. But clearly it did not crack the critical question it had been called to address: How might the FDA encourage or at least not impede research into immune-based treatments for HJV, given current scientific understanding and commercial constraints?

In the end we sensed that while the meeting illuminated many of the issues, it could not approach a solution to the central problem, because within the constraints of prevailing thinking there is in fact no solution. If this is true, hen at least in AIDS, the path forward on immune-based therapies--one of the most critically important frontiers of medicine--will have to come from unexpected developments, not through public policy.

But policy can create conditions that encourage or at least allow unexpected, outside-the-system advances--or conditions that make them difficult or impossible. So policy can be judged on this basis. The meeting did not address this issue--nor fully acknowledge that for the central problem it had been asked to explore, no solution appears possible within the current terms of discourse. This situation and its critical implications were implicitly left to be handled some time in the distant and hazy future.

This article presents our analysis. See below for references to the FDA's Web site for a transcript and other information about this meeting.

Background: The Problem

Clearly the current antiretroviral combination treatments for HIV are not enough. And there are many reasons to believe that immune-based therapies will be the next major area of treatment advance. But for years research has been seriously impeded because there is no accepted path for such a treatment to be approved for marketing. Companies will not invest if they do not see any way they could profit. And lack of money in a field dampens the interest of academic and government researchers as well.

Since immune-based therapy will clearly be important not only for HIV and other infectious diseases but for cancer, autoimmune diseases, and probably a considerable number of other conditions as well, it would seem to make sense for a large company to conduct research and development in order to build position and leadership in the larger area, even if there is no visible path of return for the investment. But U.S. business is judged on short-term not long-term results--and the so-called "innovator" pharmaceutical companies are primarily driven by a marketing not a research culture, so long-term "blue sky" funding by industry is scarce.

In HIV, a new antiretroviral can be initially approved for market based on safety data and improvements in viral load. But there is no similar test for approval of immune therapies. There are dozens of "biological markers" of immune function--but no accepted "surrogate marker" which can be used for drug approval. While many of the biological markers can have abnormal levels in persons with AIDS, and drugs might be developed to move these values toward normal, there is no proof yet that doing so will necessarily benefit the patient. And without that proof, the FDA is understandably unwilling to approve the drug.

What is a "surrogate" marker? Before viral load and CD4 count were accepted (within limits) for drug approval, any new drug for treating HIV was expected to show benefit through reduction of "clinical endpoints"--generally interpreted to mean death or a major AIDS-related complication (one serious enough to end the clinical trial for that patient, who was then considered a treatment failure for research purposes). A "surrogate" marker is a blood or other laboratory test which is accepted as being able to substitute for death or other clinical endpoint for the approval of a new drug.

What about using viral load as a surrogate marker for immune-based treatments as well as for antivirals? This is a serious possibility, and was discussed extensively at the meeting. But current policy is that a surrogate marker is specific to a class of drugs-so viral load is accepted only for approval of new antiretrovirals, but not for immune-based therapies (unless it is separately established as a surrogate for them). And another obstacle is that while one would probably expect a good immune-based therapy to substantially reduce viral load, that is a lot to ask in the early stages of development of these treatments.

Another possibility being discussed is that an immune-based treatment might in theory prove itself by extending the average time before viral rebound during a treatment interruption.

Without an accepted surrogate marker of some sort, immune treatments have been required to show that they can reduce clinical endpoints. (The large trial of Remune, the HIV immunogen developed by the late Dr. Jonas Salk, was an attempt to do just this. The large trials of IL-2 which are now recruiting are also an attempt to prove an immune-based treatment through clinical endpoints--although it is far from clear that these trials, even if successful, would validate CD4 count, the candidate marker which led to the trials, as a general-purpose surrogate marker for other immune-based therapies in the future, since an increase in the CD4 count can mean many different things).

As these trials illustrate, in the United States today it is almost impossible to meet the clinical-endpoint burden of proof, even if the treatment could in fact greatly reduce clinical endpoints. Due to the success of antiretrovirals there are fewer deaths and AIDS-related infections today-- and of course all participants must be offered the best antiretroviral therapy available. A trial might be put together of patients who have failed all antiretrovirals and decided to stop taking them, so that clinical endpoints are more likely--but such a trial would include highly diverse, advanced patients who had proved difficult to treat for various reasons, making it hard to prove a new drug in such a trial.

Immune-Based Surrogate Marker?

Why, then, can't we solve this problem by developing an immune-based surrogate marker, like viral load for antiretrovirals? There are a number of reasons.

One is that there is more diversity among immune-based treatments than among antiretrovirals. All antiretrovirals share a common target--HIV. But immune-based treatments can be fundamentally different from each other, and a marker appropriate for one might not be for others.

Another serious problem is that a surrogate marker, as understood today, can only be established in a highly successful trial of a drug of the same class. So a surrogate marker is never available for the first drug of a new kind--only for later ones. Since there is no immune-based HIV treatment today that has been proven to reduce death or other clinical endpoints, there is by definition no surrogate marker for immune-based HIV treatments. And if there is no viable opportunity to approve an immune-based drug, there will not be a marker. (Even worse, it turns out to be much more difficult for data to establish that a marker works, than to establish that a drug works--making it even less likely that surrogate markers can be validated for approval of immune-based drugs.)

The bottom line is that today there is not even a substantial beginning of a professional consensus on how an immune-based treatment for HIV could be approved. Therefore, research and funding in the area are neglected--completing the vicious circle, as little progress is made toward developing scientific consensus for the future. So the FDA called a meeting about this problem, and on October 16 the FDA's own experts and outside experts talked for a day, putting lots of information and issues on the table but with no real movement toward resolution. If the process continues as it is going now, the best prediction is that in a few years there will be a little more consensus, but no approved HIV immune-based drugs or markers. Then the FDA will call another meeting, and the experts will gather to talk again.

What Should Be Done?

Could anything change this outcome to one that is less grim? Yes, but something unexpected has to happen, something which changes the system from without. Otherwise there cannot be an immune-based surrogate marker until at least one immune-based treatment has been proven by clinical endpoints--and it will take a large trial and many years to get such proof, at least for any treatment tested in rich countries where antiretrovirals are available. (And the trial which works, which proves an immune-based treatment by clinical endpoints, must also happen to include a candidate surrogate marker which also works and is otherwise satisfactory, in order to establish that marker for more rapid approvals of similar treatments in the future. Otherwise the drug would be proven, but it would be necessary to start over on the marker.)

To see how a different outcome is possible, consider the history of viral load, the clearly successful surrogate marker for antiretrovirals. Today viral load is well established technically as a surrogate marker for antiretroviral treatments-- an outcome that would not have been possible without the great success of modern HAART treatments, which provided the successful trials through which this surrogate marker could be validated. But historically, viral load was used in drug approval ahead of full, rigorous statistical proof that it was a surrogate marker, because the mechanism made sense, and the alternatives were so grim. Today there is more skepticism about new treatments (mainly due to the side effects of antiretrovirals), so progress in immune-based markers might not follow the same path.

An engineering analogy might help to illustrate the choice before us. Assume that we have a digital video as a computer file, which must be transmitted online to anyone who requests a copy, so that they can view the video on their computer. Digital video files are likely to be very large and therefore difficult for some computer users to receive--but they can be mathematically compressed to allow faster transmission. And here we have the choice. If the file must be received perfectly--guaranteed not even one pixel changed--then it cannot be compressed very much. But if we allow the possibility of minor errors that will probably never be noticed, the file can be made many times smaller.

The analogy with surrogate markers is that if we demand perfect proof, we will probably block drug development and prevent progress in immune-based therapies. But if we settle for less than perfection, and use our best judgment while accepting a small chance of being wrong, we could move much more rapidly in drug development, fueling basic research in immunology as well. Our concern is that this decision may not be made rationally, because many people like the idea of being right, and are attracted to the case for taking the time to do it right--no matter what research opportunities will be sacrificed for the assurance of certainty.

Here are some specific suggestions for unblocking research in this field:

* The FDA could consider drugs for approval for the indication that they improve abnormal immune parameters--letting the medical community decide whether or not they believe the product has value to patients. Such a policy would open the door today to developing drugs for increasing HIV-specific cellular immunity, for example--because a company which could produce such a drug would have a clear path to marketing approval, and might go ahead based on its prediction that after the several years it would take to complete development, the medical community would probably be interested in the treatment. Today, in contrast, there is no clear way to approve such a drug, so its development could not be financially justified within the company.

Even if the FDA can make a better medical decision than the "market" of HIV-treating physicians--which could be debated--that is not the right comparison. The issue is whether to let the medical community make that decision several years down the road, vs. not having the choice because the research and development were never done. If the FDA opens the door now, doctors will make their decisions in several years, with the benefit of the additional knowledge.

Today this change is unlikely, because the FDA's institutional dynamic--like that of almost any active organization--is to expand its role, not contract it. But we are facing an emergency here. Professional consensus could make a difference, and we need more discussion of this issue.

* In trial design, we should not tie proof of clinical improvement so tightly to the idea of clinical endpoints (death or major new AIDS-related conditions). Statistically significant reduction in clinical endpoints is only one way to prove clinical benefit--and it has lots of disadvantages, including the fact that it takes so long to accumulate the statistical proof (and hundreds if not thousands of volunteers, further slowing results) even when a strong benefit exists. Clinical endpoints developed historically from cancer trials, where five-year survival was the gold standard for proving that a treatment worked.

But in immune-based treatments for HIV, other kinds of proof of clinical benefit could lead to much faster results. For example, the treatment could be tested in patients who need continuous therapy to control thrush or other AIDS-related infection, to see if HIV treatment reduced the need for the other drugs, or otherwise benefit patients in measurable ways. Such trials could be much smaller than clinical-endpoint trials, and would produce results much, faster, because they do not need to wait for thankfully rare events in order to get the data that counts.

* Better understanding of mechanisms of immune damage in HIV infection would be very helpful. A drug that blocked such damage might lead to improvements in health.

* Also critically important are quality control and standardization of tests for immune markers, so that researchers can know their data is accurate, and comparable across labs. This work is already being done. We need to be sure it is adequately funded, and without spot shortages of funding or other resources which could seriously impede progress, but could easily be corrected.

What might most speed development of immune-based treatments and markers would be examples of clear success in the area. This week's announcement of a successful vaccine test in animals (see "Vaccine Advance: Monkeys Still Infected, but Protected from Illness," in this issue) is one. The research establishment must be able to move quickly to take advantage of unexpected success.

For More Information

A transcript should be available two weeks after the meeting--around the end of October. Go the FDA Docket page at http://www.fda.gov/ohrms/dockets/ac/acmenu.htm click on the year 2000, scroll down to the Center for Drug Evaluation and Research (CDER) meetings, and click on the name of the committee, the Antiviral Drugs Advisory Committee. There is also background information on this site.