It turns out that the compound Matthews identified doesn't work as an AIDS vaccine, but it may still make a very good AIDS drug. It belongs to a family of molecules known as entry inhibitors that, as the name suggests, prevent HIV's entry into healthy immune cells. While none are yet available in pharmacies, they are probably the most promising new class of anti-HIV drugs under review. Bolognesi's company, Trimeris, based in Durham, N.C., collaborates with Hoffman-LaRoche and is already in the final stages of human testing with one compound and in the earliest phases of testing with a second. Other biotech firms, including Progenics Pharmaceuticals in Tarrytown, N.Y., are right behind it. Progenics currently has two compounds in human trials. Together, these drug candidates represent a sophisticated new generation of antiviral compound - drugs born of a better understanding of how HIV works, at the molecular level, inside the body.

Entry inhibitors, for example, include at least three families of potential drugs designed to block key steps in HIV's entry into a cell. Scientists know that in order for HIV to establish an infection, it first needs to make contact with the right cells. They also know that HIV binds very quickly to a particular section on the surface of a type of immune cell known as CD4. You can think of these points of entry as windows or doors into the cell; in order to get inside without destroying the cell (which has machinery the virus needs to reproduce itself), HIV has to find and pick the lock. To do this, it links simultaneously to an adjacent section on CD4 known as a cytokine receptor. Once the virus has bound to both CD4 and the cytokine, chemical changes take place that alter the structure of the cell's membrane, and the lock is "tripped." Fusion between the virus and the cell occurs, and HIV is free to spill its genome into the cell and begin replication.

Trimeris' compound, called T-20, blocks the final structural contortion from taking place. For this reason it and a second candidate, T-1249, are known as fusion inhibitors. Progenics has been testing a different type of entry inhibitor, a molecular decoy for CD4 whose job is to find, bind and lure HIV away from the real CD4 cells.

Regardless of how they work, what all entry inhibitors share is an ability to thwart the virus outside the cell, before it has an opportunity to infect healthy immune cells. And that, some experts believe, may give them a better chance of dispatching HIV than the currently available antivirals, all of which work inside cells that have already been infected. "Working outside the cell gives them in theory a major advantage, says Dr. David Ho, director of the Aaron Diamond AIDS Research Center in New York City, "because cell membranes can present barriers to some drugs, and some have molecules that pump out drugs that manage to get inside." This suggests that the new drugs might be effective at lower doses, making fusion inhibitors safer for the patient in the long run. Trimeris' studies support this; so far, neither of its compounds seems to cause any of the serious toxic side effects associated with today's AIDS drugs, such as nausea, vomiting and abnormalities in fat metabolism.

If the early promise of the fusion inhibitors bears out, they will be a welcome and badly needed addition to the HIV drug arsenal. For while the available antivirals have had a dramatic impact in reducing the death rate from AIDS, they are not enough. As more patients take these cocktails of powerful drugs for more and more years, the dual problems of resistance and toxicity are beginning to tarnish the promise they once held.

About 30% of patients who begin therapy with a combination of antiviral drugs have to stop, either because their bodies cannot tolerate the toxic side effects or because they cannot keep up with the grueling regimen of strictly scheduled pill popping. An additional 30% to 50% are currently in salvage therapy, which is what AIDS specialists call the last-ditch potions of drug cocktails given to patients who have become resistant, one by one, to every class of antiviral on the market. "We have, by treating lots of individuals relatively successfully for varying periods of time, accumulated a new target of patients now in desperate need of new options and new drugs," says Bolognesi.

So far, Trimeris has concentrated its testing of T-20 on these very patients. The company's latest data, on 70 patients with advanced AIDS who are no longer responding to at least 10 currently available drugs, are promising. After almost a year of taking T-20 in combination with other antivirals, 56% of the patients showed sharp declines in the amount of HIV in their blood - at least tenfold below their starting levels and in some cases to levels undetectable by current tests. The only disadvantage to T-20 therapy at the moment is that it's an injection that needs to be given twice a day. But Trimeris is working on making the molecule easier to take, possibly as a skin patch.