Symptom-Reduction Trials at FDA Clinical Trials Workshop

At the FDA workshop on clinical trials (September 6 and 7, with associated meetings on September 8; see discussion in AIDS TREATMENT NEWS #228 and #229), our major goal was to put our proposal for symptom-reduction trials onto the table for discussion. We were impressed with the openness of people to considering this kind of trial design -- not as something they necessarily agreed with, but as a possibility worth serious examination. (For an earlier draft of our proposal, see AIDS TREATMENT NEWS #229.)

"Symptom reduction" trials is a name we used for a kind of trial in which everyone entering must have a certain, carefully chosen, "indicator symptom" -- a symptom or condition which is believed to be caused by immune deficiency or otherwise by HIV disease, and which is unlikely to resolve quickly on its own, or with conventional treatment. The trial then tests an antiviral drug (or an immune modulator, or other treatment for HIV disease) to see how well it works to make the symptom go away. The major advantage of this kind of trial is that it tests for a CLINICAL BENEFIT of the proposed treatment (a concrete benefit to patients, not only a blood-test number), and yet it produces results RAPIDLY, since anecdotal reports suggest that many HIV-related symptoms can respond in days or weeks to effective treatment for HIV -- almost as fast as viral load responds.

Conventional AIDS/HIV clinical-benefit trials take the other approach, of measuring disease progression by waiting for new symptoms or conditions to appear. Since HIV disease develops slowly, and most patients will never contribute to the study result, this approach requires large, expensive trials that usually take years to organize and run.

Recently we learned that Abbott Laboratories has been exploring a symptom-reduction kind of trial design, independently of us, in an effort to find a faster way to get its protease inhibitor approved. And at the clinical-trials workshop we learned that the idea of using symptom reduction to tests antivirals has occurred to AIDS researchers over they years. But we do not know of any AIDS trial yet which screens for a specific symptom as an entry criterion -- a symptom NOT directly responsive to the drug being tested -- so that the symptom can indirectly indicate whether or not there is immune-system recovery due to the experimental treatment.

At the clinical-trials workshop, others suggested three important improvements to our proposal:

(1) We had divided clinical trials between those whose primary outcome is surrogate markers (such as viral load, or CD4 count), and those whose primary outcome is clinical. A better view is to distinguish three kinds of trials: surrogate marker, short-term clinical benefit, and long-term studies (which examine benefits vs. side effects of drugs over a long time span, usually many years). Symptom-reduction trials produce clinical evidence, but not long-term results. (There is increasing agreement that long-term protocols are problematic -- that AIDS trials certainly need better long-term FOLLOWUP, but perhaps without requiring patients to stay on the same treatment or protocol for years.)

(2) A number of people proposed different indicator symptoms which might work better in drug trials than the ones we had suggested. Pediatricians thought that correcting growth failure in children would be an appropriate test for an anti-HIV treatment. Others suggested improvement in sinusitis.

(3) Statistician Andrew M. Hill, from Glaxo Research and Development Limited in the UK, suggested using viral load in addition to symptom reduction, and then doing a dose-response analysis -- seeing if patients with the biggest change in the "dose" of HIV also had the biggest reduction in symptoms. A related analysis might help to validate viral load as a proven "surrogate marker" of clinical improvement -- in a time frame of months, instead of the years which would be required for a trial to validate the marker as a surrogate for progression to AIDS.