ddI Approval Recommendation

On July 19 the Antiviral Drug Products Advisory Committee, a panel of non-government experts chosen by the U. S. Food and Drug Administration (FDA), recommended that the FDA approve the AIDS antiviral dideoxyinosine (ddI, brand name VIDEX) for adults and children who cannot tolerate or are not adequately responding to AZT. The Committee's vote was 5-2. This action is only advisory; the FDA must make the final decision, and it is expected to approve ddI later this summer or fall. Approval will make ddI the second AIDS antiviral regularly available, after five years of AZT alone.

The advisory committee meeting was widely reported in the press, which understood its importance in setting precedents for future evaluation of new drugs to treat life-threatening conditions. We will not repeat the news here, but look at some of these precedents and how they might affect drug approval in the future.

Ideal data on the use of ddI is not available, and as a result the Committee's task differed from business as usual in two ways. First, it needed to weigh the use of "surrogate marker" data, such as T-helper counts and measures of viral load, because it did not have the efficacy data required in the past -- death or major disease progression in controlled trials. Second, it needed to decide if "phase I" data, obtained from a number of small studies which were not designed to test drug efficacy, could substitute for data from the efficacy trials. Our interpretation of the Committee's actions is that surrogate markers basically were accepted (although not unanimously) as a major contributor to drug approval, at least for nucleoside analogs -- especially since the markers are getting better.

But phase-I data was basically rejected as a sole basis for drug approval. After the first day of the two-day meeting, there was much gloom among observers; many thought that the Committee would vote to reject ddI -- that although the drug did work, the evidence available did not prove it worked. On that first day, Bristol-Myers Squibb had presented the phase I data which was the basis for its new drug application (NDA); later that day, the FDA, using sophisticated statistical analysis, raised serious doubts about that evidence. Only on the second day, when preliminary data was brought in from the ongoing controlled trial ACTG 117 (comparing ddI to AZT for persons who had already received extensive treatment with AZT), and this data tended to confirm the phase I results presented the day before, was a majority of the Committee convinced that, on balance, it was better to recommend that the drug be approved than not approved. It is almost unprecedented to release data early from an ongoing trial.

An important element in the Committee's decision is that there is much more safety data on ddI than on most drugs when they are approved -- due to Bristol-Myers Squibb's expanded access program, which provided ddI to about 20,000 people and collected information about side effects. Also, the fact that the drug clearly shows biologic activity (indicated by reduction in p24 antigen and other viral measures) did impress the panel.

Still, the Committee was reluctant to recommend approval, for several reasons. First, the controlled-trial evidence showed only a small benefit. Average T-helper counts were only 10 to 15 higher in the group which switched to ddI as compared to those who stayed on AZT (opportunistic infection rates in the ongoing study were not released). Second, there is serious toxicity with ddI, especially neuropathy and pancreatitis. And third, there was concern that approving the drug now might end the controlled trials, if persons who think they are getting AZT in the trials drop out to purchase ddI instead. Loss of the trials would be tragic, because their data is needed not only to confirm that the decision to approve ddI was the right one, but also to give physicians guidance on the best ways to use the drug. Also, these trials will refine scientists' understanding of surrogate markers, and therefore help in the development of future drugs.

Comment

The likely approval of ddI, while important for providing a second treatment option, should be kept in perspective. ddI could have been tested and approved several years ago, at about the same time as AZT.

It has taken years for government and academia to realize that the chances of developing drugs rapidly enough to meet the needs of the epidemic are near zero so long as statistical differences in death or major disease progression are required for drug approval. AZT was approved by a fluke, probably depending in large part on luck and/or experimental error. In any case, only one drug can go through that door, as new agents cannot ethically be compared to placebos in similar trials, but must be compared with AZT, requiring trials which are large, expensive, slow, and hard to interpret. It is better to accept the risk that a drug will be approved in error, based on blood tests and other rapidly-measurable indicators of improvement, than to face the near certainty of years of delay during an epidemic.

The drug-approval evolution now seen with ddI is very important and welcome, but it applies only to the last and most public stage of development, when a drug ready for use is awaiting final FDA action. Early problems -- such as patent delays, excessive and unworkable animal-test requirements, and poorly designed phase I trials, have as much impact on the ultimate date of new-drug availability. But these problems are much less visible, and therefore harder to influence through public opinion. The drug-development system is like an iceberg, in that only a fraction is easily visible; only that tip is now being reformed.

Should Activists Reconsider Placebo Trials?

The most important single implication for the future to come from the Committee's meeting on ddI is that, in certain cases, improvements in laboratory test results (such as T-helper counts) can have a major role in drug approval -- but they must be obtained in well-conducted, controlled trials, and there must be some supporting evidence of clinical improvement (such as weight gain).

For many new drugs, this means that the fastest way to get them approved will be short (perhaps three months) relatively small trials comparing the new drug to no treatment, using immunological or virological "surrogate markers" as major endpoints of the trial (instead of waiting for death or disease progression, as in earlier trials). Volunteers entering these studies can be relatively healthy (with T-helper counts around 300, for example), so that a few weeks of no treatment would be unlikely to do serious harm. Also, these trials can be designed so that anyone found to be doing badly can quickly be taken out of the trial and given appropriate treatment, or switched to the active drug if they had been receiving the placebo.

During the first day of the Committee's consideration of ddI, when it looked like the vote would be to reject the drug, such a study would have made all the difference in building a consensus for approval. Fortunately, the AIDS community was rescued this time by the data from ACTG 117 which was brought in on the second day. But ACTG 117 is slow, cumbersome, and expensive -- because it was designed to look primarily for death or disease progression, not for surrogate markers -- and in most future cases of needed drugs, no such study will be available.

The reason AIDS activists have strongly opposed placebos in the past is that trials were designed to look for death or major opportunistic infections; many people had to die on the placebos, or suffer major health problems, in order to prove the drug. It is widely believed that important treatment or prophylaxis had sometimes been forbidden to study participants so that people would get sick faster, making it easier to get statistical proof that the drug worked. But surrogate-marker trials do not need to count people getting sick in order to prove a drug is effective. Therefore they can be ethical in their treatment of volunteers, and at the same time greatly accelerate the availability of proven new treatments for everyone.

The conventional wisdom among many scientists and government or drug-company officials is that AIDS activists will not tolerate trials in which some volunteers receive only a placebo. But in our discussions with activists we have found widespread agreement, or at least openness to the possibility, that there can be situations where properly-designed placebo trials are appropriate. We urge further discussion and consideration of this issue.