American researchers have encountered an important setback in their attempts to develop a vaccine against AIDS.

Initial attempts to use experimental vaccines to protect chimpanzees against the AIDS virus have apparently failed, despite some promising signs, according to researchers familiar with the studies.

One key experiment, led by researchers at the National Cancer Institute (NCI) and the Duke University Medical School, at first seemed a success, as it appeared to prevent infection of chimps for weeks after they were inoculated with the AIDS virus. But all six chimps in the experiment have since become infected, according to researchers connected with the experiment.

"If you had called me two weeks ago, I would have been very excited," said Dr. Robert Gallo, a discoverer of the AIDS virus and head of NCI's Laboratory of Tumor Cell Biology. "Now, I would say the data is disappointing but not anything to be startled by."

Vaccines are considered the best hope to stop the spread of acquired immune deficiency syndrome. Drug treatments have begun to show promise in alleviating symptoms of the disease but not in preventing it. While setbacks are expected in vaccine work, the failures in chimps have been disappointing -- underscoring the prediction of federal officials that a useful vaccine will not be available within five years.

"It was our first shot," said Dani Bolognesi of Duke University, a leader of the experiment. "It looks like a failure now. But it is not over yet," he said. There is some chance that the chimps' immune systems will defeat the viruses. He added that it is a hopeful sign that any protection was achieved.

In addition to the NCI experiments, word has been spreading among the research community that other laboratories have failed to protect chimpanzees in vaccine experiments, even though some positive effects were shown, researchers said.

The testing of vaccines in chimpanzees marks a new stage in the effort to make a vaccine against AIDS. A number of prototype vaccines have been developed and and are being tested in the animals. Experiments with human volunteers in the United States are expected this year.

"There are quite a few groups poised, ready to begin work {in humans}. But no one knows if any of them will be successful," said Dr. Allan Goldstein of George Washington University, who has developed a vaccine with the Institute for Immunological Disorders in Houston.

Yesterday, here at the Third International Conference on AIDS, Daniel Zagury of the Pierre and Marie Curie University in Paris reported early, positive results in the first work on human volunteers. In France, laws on drug experimentation with humans are much less restrictive.

Results showed that the immune systems of some of the volunteers were mobilized by Zagury's vaccines, raising antibodies to the virus and showing other signs of immune response. Zagury was one of those to get the experimental vaccine and to have his immune system respond. But the vaccines he tried are not ready for wider tests in people, he said.

The advantage of working with chimps is that, unlike humans, they can be inoculated with live virus to determine whether the vaccine will prevent infection.

Until a short time ago, researchers worked only with animals such as mice, goats and monkeys in trying to devise a vaccine that could block infection with the human immunodeficiency virus that causes AIDS. Chimpanzees, the closest relative to humans, are reserved for the most promising experiments that emerge after other animal tests because so few chimpanzees are available for medical research.

Though chimpanzees can become infected with the virus, they have few or no symptoms and do not develop AIDS.

Because chimpanzees are otherwise similar to humans, researchers have hoped that work with them will give some information about whether the vaccines are safe and can protect against infections.

Work with chimps and humans will proceed simultaneously. Several groups are believed to have carried out work on chimpanzees. At least two -- Bristol Myers Co. of New York and the Institute for Immunological Disorders in Houston -- have applied for permission to test vaccines on humans.

The principle of vaccination is to prime the body's immune defense system against an invader such as a virus or bacterium. The body makes antibodies tailored to attach themselves to one type of virus or bacterium. If enough of the right antibodies can be produced with sufficient speed by the body, they can find attacking microbes, attach to them and thus disable them.

A second type of immune reaction, called cell-mediated response, alerts the body's "killer" cells that an invader has infected cells in the body. The killer cells find and attack the infected cells.

For a time in the NCI study, researchers became excited because it appeared for more than two weeks that the chimps were immune to infection.

But finally, said Duke's Bolognesi, the intense excitement gave way to a letdown as the six chimps in the experiment became infected by the third week of the test.

"But remember, this is very, very early in the vaccine work . . . . Sometimes in an experiment like this you can get viral replication, then a vigorous immune response that knocks down the virus again," Bolognesi said.

"The news was not so nice," said Kai Krohn of the NCI. "It means more time, more money, more experiments." But he said that parts of the AIDS virus that are key to creating a strong immune response have been identified, so it may be possible for the next vaccine candidate tested to have the right chemistry to induce a better response.

The NCI experiment began a month ago at the Frederick Cancer Research Facility. The chimps were given a candidate vaccine that consisted of a piece of the AIDS virus, the proteins of its outer shell.

After the chimps were vaccinated, one set was given 100 infectious units of the virus, the other was given 1,000 infectious units. One infectious unit is considered enough virus to cause an infection.

Researchers said that even if the chimps had been protected permanently against the infection, there would be numerous problems to overcome before a vaccine would be ready for widespread use.

There are 10 or more strains of the AIDS virus. Different strains exist in different geographical regions, and patients may even carry more than one at a time. Early vaccine candidates are not designed to stop more than one strain.

Influenza vaccines, for example, developed over the years commonly contain many viral strains.

There are several different strategies to making an AIDS vaccine. One was pioneered by Dr. Bernard Moss of the National Institute of Allergy and Infectious Diseases. It is the same approach used by Zagury in his work on human volunteers and in the work of Bristol Myers on chimps.

This vaccine candidate consists of a live vaccinia (cowpox) virus with a small piece of the AIDS virus inserted in it. After innoculation with the vaccinia, the body mounts a defense against both the vaccinia and the AIDS viruses.

The work reported by Zagury and his colleagues in Zaire yesterday included two experiments.

In 10 patients infected with the AIDS virus, researchers drew white blood cells from each and killed the infected cells, then reinjected them in an attempt to provoke a cell-mediated immune response that would block further dissemination of the virus in the patients.

A group of 12 healthy, uninfected volunteers, including Zagury and a French colleague working in Zaire, were studied in tests of a candidate vaccine. The volunteers were given a vaccinia-carried AIDS protein plus a "booster." The booster consisted of cells from the volunteer that have been infected with vaccinia, then reinjected, to help bring out an immune reaction.

Using a vaccine plus a booster of one's own infected cells is far too expensive and time-consuming to be practical as a widespread vaccination procedure. Zagury is searching for a substitute booster to use in conjunction with the vaccinia.

In the Bristol Myers experiments, chimps received a vaccinia-based vaccine but became infected quickly after being injected with high doses of AIDS virus. Robert Nowinski, president of Genetic Systems, a subsidiary of Bristol Myers, said the vaccine reduced the amount of virus in those chimps by tenfold to 100-fold compared with chimps not vaccinated.