Clearly, this is not a time for celebration. That the progress made to date has been insufficient is plainly evident in most of our hospitals every day. The acquired immunodeficiency syndrome continues to take an unrelenting toll on our patients and our society; it is now the leading cause of death in the United States for men between the ages of 25 and 44 years, and will almost certainly achieve that unwelcome distinction for women of the same age by the end of the decade [2].

However, it is worth taking stock of where we were in 1984 to put into perspective what has been achieved and what still needs to be done. Patients with AIDS live longer in 1994 than they did in 1984—almost three times as long—and although the cause of this increased survival is not totally apparent, it is largely due to improved therapy and prevention of opportunistic processes [3]. In 1984, no antiretroviral drugs were available. In 1994, we have agents that, when used as monotherapy, slow the progress of disease; when used in patients with advanced infection, they have a marginal effect on survival and disease progression. The most dramatic demonstration of their effectiveness was the recent trial indicating the success of zidovudine in preventing the transmission of HIV from mother to infant [4]. We have developed sophisticated virologic technology and have learned, not unexpectedly, that the failure of the nucleoside analogs can be explained in part by the emergence of resistant strains of HIV.

Yet, all advances have been overshadowed by the failure of the clinical research effort to provide a curative therapy. Disappointment may have clouded our ability to understand what has been achieved and may lead us to misguided reform that will throw the baby out with the bathwater. The failure to deliver a cure should not be an argument to abandon clinical research. It ought to be clear from the gaps in our knowledge that we need not less, but better, clinical investigation, and more of it. Infection with HIV is a human disease for which there is no suitable animal model, and it is unlikely that more basic bench research will increase our ability to understand the complex host-parasite relationship. Instead, we should focus much of our clinical research effort on increasing our understanding of HIV pathogenesis while improving our therapies, however small these increases and improvements may initially be.

In this issue, Lederman [5] outlines some of the approaches to therapy directed at the host response. This area of research is clearly in its infancy, and studies need to be strictly designed and formulated. The progress to this point, for example, in the vaccination of infected patients has not been promising. A shotgun approach, using drugs that are possible immunomodulators without a clear rationale based on current understanding of pathogenesis, is no more likely to be successful than a similar approach to antiviral therapy has been in the last decade. Furthermore, current assays used to measure the immune response are crude; we need to invest in research to develop and standardize new assays.

Our clinical trials should be built on scientific rationale but should also be designed to answer questions other than those about the effectiveness of a therapeutic strategy. For example, we ought to use trials of immune-based therapy to test hypotheses about the immunopathogenesis of HIV. A similar approach can be taken with virologic research. For the past decade, we have used indirect markers of immune function—a quantitative measurement of CD4 lymphocytes and the development of opportunistic events—as measures of the therapeutic effect of antiviral drugs. We are increasingly aware that some of these surrogates are not useful in predicting the effectiveness of treatment. With the emergence of new virologic technologies, we can begin to ask direct questions about our therapies. Do antiviral drugs decrease the amount of virus in blood or tissues? Do changes in viral burden that result from antiviral drugs correlate with clinical outcome; for example, does a decrease in the amount of virus a patient has translate into a clinical benefit for that patient over 1 or 2 or more years? We can expand on early studies of resistance to determine whether resistance completely accounts for drug failure. We need to learn whether virologic markers (virus load, viral phenotype, or the emergence of resistance) can be used by clinicians to determine when to start, stop, or change therapy. Similar questions can and should be asked about immunologic markers and drugs that affect the immune system.

We can start addressing these and similar problems by linking scientifically based clinical trials with state-of-the-art retrovirology, immunology, and pharmacology laboratories and by actively encouraging an exchange between basic scientists and the clinic. However, clinical researchers need to be frank, especially with the public, about what they can and cannot hope to achieve. It is always possible that next year we will find the magic bullet, the penicillin for HIV, but that is unlikely. It is more likely that we will continue to make slow but steady progress with agents that have modest efficacy; this is unspectacular research, perhaps, but research that we hope will affect the standard of care and patients' lives. We need to be honest, too, in our interpretation of our results, not promising what we have not achieved and not deducing what we have not shown. For example, although I may believe that combination therapy administered early in the course of HIV infection probably results in long-lasting remissions for infected patients, this has not been shown in any clinical trial and so remains merely my opinion, a hypothesis to be proved or rejected by well-done clinical research. Honesty is also needed in our analysis of flaws in the design and execution of previous studies so that mistakes are not repeated. We know, for example, that long-term studies (those lasting longer than 18 to 24 months) are extremely difficult because of patient withdrawal, yet long-term trials of new therapies are almost certainly needed until validated surrogate markers are identified (particularly to verify that clinical benefit is sustained in a disease that has an average progression time of 10 years). Consequently, we need innovative approaches to study design and analysis that will be acceptable to clinicians and to patients.

The problems in AIDS research reflect many of the problems in our health care system as a whole—disjointed administration, unclear mandates, and unrealistic expectations. As a remedy, some have called for a Manhattan Project on AIDS, but the real analogy from World War II is more likely to be the Pacific Islands campaign—a long, grueling war of incremental steps carried out by many unsung heroes. We can only hope that the outcome will be successful.