VaxGen: Are there hints of race-based effects?

In February 2003, AIDS vaccines hit the headlines when VaxGen, a California-based company, announced the final results from the world's first Phase III trial of an AIDS vaccine. The study tested a candidate called AIDSVAX® B/B, which contains part of the outer coat (envelope) protein of HIV. The vaccine was tested in the United States, Canada and Europe in roughly 5,500 volunteers. 95% of the volunteers were men who have sex with men, and the remainder were high-risk women.

The trial was designed to find out whether AIDSVAX® could either prevent HIV infection or reduce the severity of disease in people who were vaccinated and went on to acquire HIV through sexual exposure. VaxGen scientists reported that, overall, AIDSVAX® did not provide either type of protection: Volunteers who received the vaccine were just as likely to become infected as those who received a placebo. And vaccinated participants who later became infected had similar CD4 cell counts and viral load levels as infected volunteers in the placebo group.

But VaxGen also made the startling and highly controversial claim that, when they subdivided the volunteers by race, the vaccine protected 2/3 of the African-American, Asian and mixedrace volunteers.

This finding was instantly challenged when statisticians pointed out a key flaw in the data analysis. The flaw hinged on the fact that, like most clinical studies, the AIDSVAX® trial was designed to answer a scientific question based on data from all the volunteers. (A certain number of volunteers is needed to be sure that an observed effect is real and not just a coincidence.) When statisticians single out specific subgroups, they adjust their analyses to make it more likely that they will be able to identify a real finding. VaxGen apparently did not make this adjustment. The small number of minority volunteers (less than 500 total) also made it difficult to draw firm conclusions, since there were only a handful of infections in the entire group.

Scientists and advocacy groups have also questioned the company's justification for the pooling of data from different non-white racial groups, since these groups are not known to share common genetic features or immune markers. VaxGen did not look for this type of data initially. However, the company does have a repository of stored blood samples from volunteers which could be studied further.

Since the initial announcement there has been a great deal of debate and discussion about these results. At the same time, VaxGen has attempted to find biological data that support or explain its claim of race-based differences in protection. Simply put, at this point there is no good evidence for the claim, nor is there enough data to be absolutely sure that it can be dismissed.

This dilemma prompted the US National Institutes of Health to give VaxGen technical and financial support for further data analysis. VaxGen had said that it did not have enough money to fund extensive analyses of the data.

At press time, the follow-up analysis was focused on two areas: the levels of antibodies induced by the vaccine in volunteers of different races, and the precise strains of HIV seen in infected volunteers of different races. Another open question is whether there were gender-based trends in protection. The data suggested a possible increase in rates of protection among women, compared to men. However, with only 309 women enrolled, there were not enough data to yield answers by traditional statistical analysis.

More data on AIDSVAX® will come from a second Phase III trial in 2,500 intravenous drug users in Thailand that will be completed in late 2003.

In the meantime, vaccine developers are emphasizing the need for future Phase III trials of other vaccines to enroll sufficient numbers of different ethnic groups, and both genders, so that if trends towards race- or gender-based effects appear for other candidates, they will be easier to detect. When Phase III trials of other candidates take place in Africa and Asia, they will provide important information about whether there are differences in vaccine-induced responses within broad racial categories.