Major FDA Meeting on Drug Approval Standards, February 13 and 14

Crucial issues on how to prove efficacy for new-drug approval will be considered at a meeting of the FDA's Antiviral Drug Products Advisory Committee on February 13 and 14, near Washington, DC. The meeting is open to the public.

The Wednesday, February 13, session concerns endpoints in AIDS trials, and the role of CD4 and other laboratory markers as indicators of clinical benefit. While this meeting is not scheduled to consider ddC and ddI, those drugs clearly provide its immediate context and rationale.

On Thursday, February 14, the Committee plans to focus on AZT, especially followup on the trials which supported expanded indications for treatment (T-helper counts to 500).

While there has been little public attention to this meeting, there has been much activity behind the scenes, especially preparations by scientists. Since AIDS treatment activists from around the country will be in Washington on those days, there may also be activist meetings before and after the antiviral committee meets.

The Antiviral Drug Products Advisory Committee is scheduled to meet at the Holiday Inn Crown Plaza, Rockville, MD, at 8:30 a.m. on both days.

Comment

Here is a scenario which we find useful for organizing our thoughts for this meeting. Suppose that in the near future an AIDS or HIV treatment is found which works very well -- either a single drug, or more likely, a combination. Suppose that almost everyone who uses the treatment becomes healthy again (except for any irreparable damage which might have been caused by severe illness): weight and energy return, all blood work becomes normal, and there are no opportunistic infections or new malignancies, as long as patients keep taking the treatment. Assume that the toxicities are small or manageable.

The question is, "What would happen then" -- and what should happen?

The present system

Under the current system it does seems clear what would happen. First, if the treatment were a combination of new drugs, it would not be tested at all, because the FDA almost never allows a trial of more than one unapproved drug at the same time. All but one of the drugs would have to go through the entire approval process separately, proving itself alone compared to AZT. Then, years later, the combination could be tried.

Suppose the new treatment were a single drug, so that the above problem was not an issue. The first bottleneck- -organizing corporate commitment and funding -- would already have occurred. Waiting for the patent office could have delayed the project for a year or more. Then, if luck were bad, additional years could have been spent in litigation.

We have not investigated the detailed requirements for animal testing. But we have heard from persons familiar with this area that they are often grossly excessive and irrational.

Next comes the process of getting approval for phase I (dosage and toxicity) human trials. Here the biggest problem is juggling busy peoples' schedules to get the required people and papers into the same room at the same time. Perhaps the IRB or other required body does not finish its other business in time, so the drug is postponed for weeks or months until the next meeting. Then some of the people cannot attend, and the trial is postponed again.

Finally the phase I trials actually start. They often take more than a year, because one group must take the drug for an extended time at a very low dose, before the next group starts at the next dose, etc. for a number of different doses. Traditionally, these phase I trials were planned to run until toxicity was found, even if it became clear that the drug showed efficacy at a non-toxic dose. The higher doses that would never be used still had to be tested. (ddI was substantially delayed for this reason.)

Next comes the design of phase II comparative efficacy trials. Traditionally this is done by research physicians, with no input from the "front line" physicians experienced in treating patients. Therefore the resulting trials can be difficult to administer, and often have trouble recruiting patients.

The early steps in drug development, before phase II, are outside the scope of the Antiviral Drug Products Advisory Committee. Now we come to the later steps in development, where the public pays attention, creating pressure for reform, and therefore meetings like the one in February. (The earlier steps do not generate such pressure, because they are usually secret until human trials begin. The public does not see the inefficiency, and until recently, was simply told that good science takes time, and that the only issue was whether or not to weaken the scientific process.)

Today there is much confusion about what standards to use to judge efficacy of AIDS drugs. The traditional ones, which still prevail, are illustrated by the large ongoing trials of ddI and ddC. The original idea was that each new drug would be compared with a placebo -- and the trial would have to prove superiority of the drug, by accumulating a statistically significant number of deaths or major opportunistic infections in the placebo group. When it became unethical to use a placebo in a trial designed to end in death or major illness, AZT was simply substituted for placebo, and the goal of drug equivalence was more or less substituted for superiority. No one thought through the effects of these changes on the trial design.

With or without these complications, it was necessary to wait for deaths or OIs in the volunteers not receiving the treatment being tested. Therefore, even if the new treatment cured everybody instantly, it would still take many months or years to meet the efficacy standards set for these trials.

Because deaths in the AZT group occur slowly, these studies need to be very large and/or run for a long time (for example, 18 or 24 months) to expect to attain statistical significance. Hundreds of volunteers are required. Each trial needs to run identically at a number of major medical centers in order to find enough qualifying volunteers, creating major delays for administration as well as for recruitment. (For example, the IRB for each site must meet to approve the study -- and each IRB meets on its own schedule. Each IRB must either accept or reject the study exactly as given -- it cannot change any of the science, or the data will not be compatible. Since IRBs do not like to serve as rubber stamps, they exercise themselves by making insignificant changes in the informed consent form which the volunteers will sign -- the only changes they are allowed to make. The study then has a separate consent form for each site -- a minor, almost humorous problem, compared to the others.)

When the trial finally begins, a group of experts called a Data Safety Monitoring Board meets periodically during the trial and secretly unblinds the data, to stop the study for ethical reasons if extreme differences between the groups are found. But in practice, the standards for stopping the trial early are extremely severe. Therefore the Data Safety Monitoring Board serves more as a public-relations cover than as a real protection for patients; but because its deliberations are secret, the public does not realize this fact.

An additional problem now is the lack of clarity about what will be considered evidence of drug efficacy. Proof that a new drug is superior to AZT, in preventing death and major disease progression, is the most conservative possible standard. The problem with this standard is that for reasons cited above it will take a very long time to meet this level of proof, even if the drug is in fact better. In addition, a drug which is only equally effective as AZT -- or even less effective -- could save many lives, if it has different toxicities, is effective in different patients, or has unique value in combination with AZT or other drugs. What is happening today, however, is that companies are confused about what standards the FDA will use; therefore they are reluctant to submit their data, and the FDA does not see it.

These operational problems explain why, years after AZT approval, no new AIDS antiviral has been properly tested and approved, although attractive candidate drugs have long been available. The problems continue because no one is in charge of the national response to the epidemic, so no one has the authority to correct them.

What Should Be Done?

The hope is that the February 13 meeting will reduce the confusion by recommending a viable efficacy standard. What might such a standard be?

There is considerable movement toward a consensus of accepting improvement in T-helper cell count (or percent), together with at least some measurement of clinical improvement, such as regained lost body weight, as proof of efficacy for an antiviral (unless, of course, there is some good reason to reject the evidence; we are not suggesting that approval would be automatic). Part of the rationale is that there seems to be no group of persons with HIV whose average T-helper counts will increase over time without treatment; the direction is always down. If a drug increases T-helper count by means of its antiviral effect -- a result which can be seen within eight to 12 weeks, about a tenth the time required by a trial which looks for death or major disease progression -- and there is no reason to believe that the drug would increase T-helper cells by any mechanism other than inhibiting the virus which causes AIDS -- then it is hard to believe that the drug is not having an effect on the disease. Requiring some measure of direct patient benefit
is an additional precaution; it should cause little objection, as few physicians or patients would want to use a drug if there was no direct evidence that patients using it got better.

If a drug (or combination of drugs) can pass this test, then it should be considered to have met the efficacy standard for approval -- at least in the current emergency. Otherwise, the practical consequence of maintaining the current system is that all new AIDS antivirals will be delayed for years, the new generation of drugs such as protease inhibitors will not be developed expeditiously, and tens of thousands of deaths will be guaranteed.

We have heard little opposition to changing the efficacy standard for AIDS antivirals along the lines of the developing consensus described above. The issue is not whether there should be a change, or even what the change should be -- although academic arguments will likely be raised concerning any particular proposal. The real issue is whether anything will get done, given the lack of coordination and high-level leadership on Federal AIDS policy.