It appears that this workshop (see FDA Workshop on Clinical Trial Design, September 6-7 -- Registration by Aug. 18) has been requested to address profoundly wrong issues. It is being asked to design trials to meet the requirements of the accelerated approval regulations. Worse, the official announcement of the workshop includes the statement, "In HIV, the clinical endpoints that have been considered meaningful are survival and disease progression as manifested by the development of AIDS-defining opportunistic infections." As we will show below, this statement reflects the fundamental error which has made AIDS clinical trials unproductive. If it becomes the focus or agenda of the workshop, then nothing valuable will emerge.

We believe that the real reason confirmatory trials are proving so difficult is that the kind of confirmatory trials which are now being designed and run should not be done at all. The accelerated approval regulations, when Kessler introduced them several years ago, were a tremendous advance over what existed before. We and many others have strongly supported them. But now it is time to move on. We may not even need to change the accelerated approval regulations -- let alone abandon them. But we do need to interpret them in a way which is productive for today. We need confirmatory trial designs which are rational and cost-effective in the light of current knowledge.

Today -- even though it has not yet been officially "proven" in a formal, academic sense -- there is in fact an ever- growing, unmistakable if not overwhelming practical sense that modern surrogate markers do mean something. In case after case, when viral load goes down, good things happen clinically -- for example, symptoms of many different kinds go away, and HIV infected mothers with low viral load do not transmit the virus to their babies. Long-term survivors are usually people who have low viral load naturally. Researchers may not have prospectively "proven" that patients who maintain a viral load under 10,000 copies due to drug treatment live longer than if they let their viral load stay at 500,000 or a million, but the reason for the lack of such proof is that no one has done the study. Who would volunteer to be left with the high viral load until something happened to them? And even if volunteers could be found, it would take some time to complete such a trial, because it would take time for enough people to develop AIDS-defining infections or die. We do not have that time. But quite a number of bright people are now designing just this kind of study -- the stylish "strategy" trial -- with the expectation of finishing it before "confirmatory" trials, as currently understood, are allowed to go away.

What needs to be done now is to break the irrational mind- set, which started early in the AIDS epidemic and has continued as if on autopilot ever since, that clinical proof of a drug (that is, proving that the drug really helps patients, not just improves blood tests) requires clinical endpoints (understood as death or AIDS-defining opportunistic infections). In other words, instead of starting with mostly healthy people and waiting until they get AIDS-defining illnesses or die, start with people who are clinically ill, and whose symptoms are easily measurable and not intolerable, and see if the antiviral treatment in question can get them better -- can make their symptoms go away. This is clinical proof of drug efficacy -- proof that the drug does help real patients, not just blood-test numbers. And at the same time, this kind of trial uses the easily measured symptoms as a "marker" for more profound activity of the drug.

Admittedly this kind of trial would not prove survival benefit. We can afford to do without this proof because, as time moves on: (1) we are becoming more confident that the survival benefit is in fact there, even though it would be very difficult and ethically questionable to prove it for sure; (2) the cost of trying to prove survival is continually increasing, and the feasibility of doing so is becoming less, as patients have more treatment options open to them; (3) if we are wrong about survival benefit -- if a particular treatment lowers viral load, and raises CD4 count, and also makes clinical symptoms go away, and yet does not affect survival -- doctors would soon recognize this incongruity, and such a very surprising finding would provide invaluable leads in pathogenesis research; and (4) the drugs could be justified even in the unlikely case that they did not prolong survival, as they would have been formally proven to improve quality of life.

The one truly bad consequence of the unlikely possibility of being wrong about survival benefit (when viral load, CD4, and disappearance of clinical symptoms all do show treatment benefit) is that thousands of patients may choose a treatment which turns out not to keep them alive longer, when otherwise they might have chosen one which did help keep them alive. But this worst possible scenario is no worse, and may be less bad, in the system we are proposing than in what exists now. For in the current system, where confirmatory trials have to prove survival (or AIDS progression) benefit, the drug is already approved; thousands of people are using it before its failure to improve lifespan is known. In our proposed system, the long, slow survival-type confirmatory trial is replaced by a rapid, symptom-reduction confirmatory trial. Since this trial takes weeks instead of years, doctors get the symptom- reduction information almost immediately. If a treatment makes people healthier in every measurable way, including clinical symptom reduction, but still they die as fast with the drug as without, then physicians will probably realize that something is wrong, due to the unexpected deaths, and shift to other treatments, even before any survival trial results could be ready.

The kind of confirmatory trial we propose -- seeing if an antiviral treatment relieves existing AIDS-related symptoms, instead of seeing if it delays progression to AIDS or death -- not only can be done in weeks instead of years, but also needs a few dozen or fewer volunteers, not hundreds. And this kind of trial can ethically use a placebo control -- scientifically the best control -- provided that the symptoms are tolerable, and that the volunteers freely choose to risk enduring them for a few additional weeks, in order to contribute to science. No one we have talked to, including some hard-line skeptics, has been able to find any fundamental problem with this approach to confirmatory trials. But still the trials do not happen, the idea does not translate into reality, into a far better system for doing confirmatory studies, because inertia is strong and no institutional machinery exists to oppose it.

If it really is so easy, why wasn't this done years ago? We do not know. Perhaps the wrong ways of doing things were enshrined by necessity early in the epidemic, and then became a religion for many.

This writer will probably attend the September 6-8 meetings. We are reluctant, fearing yet another waste of time, because for years we have addressed meeting after meeting with little result -- little or no input into the ongoing national conversation about clinical trial design, about AIDS research, about why AIDS research has been unproductive, about how it could be made to work better. People are set in their ways. Ongoing political battles, often addressing obsolete issues, have a long history, and the combatants are now dug in.

If you might also attend the meetings, or send a statement, and will help me to get the above message heard, let me hear from you.

www.aegis.org