Talk about a hard call.

Investors may have been quick to attach their unmistakable meaning to last month's disclosure by VaxGen Inc. of disappointing Phase III results for its HIV vaccine - the company's stock value nosedived by 47.3 percent - but the scientific world was far from convinced a disaster had taken place.

The mainstream media scrambled to get a handle on the story, as reporters tried not to embarrass themselves in efforts to be both accurate and politically correct. Gone was the old problem of writing about HIV when it was wrongly considered a "gay disease"; now, the challenge (for journalists and, in a different way, for VaxGen) was race.

In the trial, which enrolled more than 5,000 subjects, only 3.8 percent (p=0.76) of those given at least three injections of the AidsVax B/B vaccine showed reduced infection, which was the primary endpoint of the trial. That's an overall failure, since the trial was designed to demonstrate the vaccine worked at least 30 percent better than placebo in stopping HIV infection. VaxGen didn't deny the failure. Wall Street exacted its punishment.

In blacks and Asians, however, trial results proved far more encouraging. Among 498 volunteers classified as minorities other than Hispanics, there were 67 percent fewer HIV infections in recipients of vaccine compared to placebo (p<0.01). In blacks specifically, there were 78 percent fewer HIV infections among the 314 volunteers who received vaccine compared to placebo (p<0.02).

An intent-to-treat analysis showed the reduction in infection in individuals who received at least one dose of vaccine or placebo was similar, but the real focus was on the benefit seen in blacks and Asians, who apparently produced higher levels of antibodies against HIV after they were vaccinated. Analysis of the data is ongoing.

Of course, pointing to favorable results in patient subsets long has been a strategy of companies trying to put a good face on trial blow-ups. Corporate apologists are apt to say things like, "Our arthritis compound didn't relieve joint swelling or pain, but a number of patients lost weight. We may have an obesity drug."

Sharon Seiler, an analyst with Punk, Ziegel & Co., acknowledged that companies tend to "slice and dice the data any way they can to show the drug works."

But in the case of VaxGen's product - which is aimed at an incurably fatal disease that has taken 28 million lives worldwide, and which seems to work in a subset that was racially defined - the situation is hardly typical, and "this wasn't just data dredging," Seiler said.

"This was a pre-specified subgroup," she said, for one thing. And statistically, the data hardly can be ignored. "In many cases with results like these you would say, Well, forget it.' But we're talking about a disease that's 100 percent fatal."

Some noted that, although the percentage response in blacks was strong, still only 13 black trial participants became infected.

"You can really only measure events that happen," Seiler said. "The question is, are the data real or is it a statistical artifact?"

Approval Based On Racial Subset Unlikely

Several days after disclosing the Phase III data, the company issued a statement "in response to media inquiries" that noted the trial design "followed a statistical analysis plan that was agreed on in advance" with the FDA and "included analyses of various subgroups, including racial backgrounds."

Even more to the point, the company said "the number of required adjustments for VaxGen's subgroup analyses is subject to interpretation and there are a variety of methods to calculate those adjustments," adding that it "cannot predict the impact these adjustments may have on the findings since that determination will ultimately rest with regulatory authorities."

In other words, Seiler said, the all-important "p" values could change.

"It would have been better if [VaxGen] had got it right the first time," she said, although the company noted its analysis was correct as done under the guidelines used so far.

"The issue of the statistical penalties to be taken for the subgroup analyses tends to decrease one's confidence that the efficacy in the subgroups is real, but nothing definitive can be said until the rest of the clinical and laboratory data are analyzed," she told BioWorld Financial Watch.

AidsVax is recombinant form of the gp120 protein on the surface of HIV, and is produced in mammalian cell culture to include two HIV subtype B antigens, called MN and GNE8. The bivalent vaccine contains non-infectious, genetically engineered proteins that mimic proteins on the surface of two strains of HIV subtype B, prevalent in North America, Europe, Australia, Japan and Puerto Rico.

The trial, which enrolled mainly men who have sex with men but also included more than 300 at-risk women, was conducted in the U.S., Canada, Puerto Rico and the Netherlands. Volunteers, believed to be HIV-negative when they joined the trial, were administered injections on the first day, followed by injections after one month, six months, a year, 18 months, two years and 30 months.

VaxGen said it will evaluate the "potential utility" of AidsVax "based on the full range of observations including primary and secondary endpoints, correlates of protection and other aspects of the immune response." More results from the trial will be reported at a scientific meeting called "HIV Vaccine Development: Immunological and Biological Challenges," to be held March 29 through April 4 in Banff, Alberta.

Meanwhile, a smaller trial in Thailand is nearing its conclusion. That study is designed to test the ability of AidsVax B/E (a separate strain) in protecting against blood-borne HIV. Enrolled are 2,500 volunteers with a history of injection drug use.

VaxGen's studies, which began in 1998, are the first-ever, large-scale efforts to learn about the efficacy of a prophylactic vaccine for HIV, so the stakes are high - for the company and especially, on the basis of data known so far, for blacks and Asians.

Some activists didn't like media characterizations of the trial as a "failure" or "disappointment," since it apparently was neither for blacks or Asians. But the bottom-line question is whether the FDA might grant approval of AidsVax B/B for a specific racial population. This has never happened.

Nor is anyone saying it should happen yet, based on what's been discovered. The data must be examined further to rule out statistical flukes. The agency could demand more trials to prove efficacy in blacks and Asians.

But if it did happen, VaxGen could find itself with a U.S. market in the billions of dollars, since there are 7.6 million black Americans, Seiler said.

Data Analysis At Start Of Long Path'

"Everybody's a little squeamish about the idea that it's skin color [at issue in the results]," she said. "But there's going to be some underlying molecular mechanism, some component of the immune response that's different."

VaxGen will need to show that anyone with that marker will benefit from the vaccine, she added. "It may be that the marker is more common in blacks than in whites," but focusing on a marker rather than skin color may make people more comfortable, Seiler said. Also, "from a public health perspective, in the U.S., blacks are over-represented in new infections," she said.

Pat Fast, medical director for the International AIDS Vaccine Initiative, a not-for-profit organization that works in advocacy as well as clinical research, said the VaxGen trial showed "also a [favorable] trend in women, which is not being discussed very much. But the bottom line is, you can't put a whole lot of credence in a subgroup analysis if you don't have overall efficacy."

As an analogy, she mentioned herpes vaccine trials last year by GlaxoSmithKline plc in which it appeared the drug worked against sexually transmitted herpes in women, but only in those who did not have cold sores indicating infection by a related herpes virus.

"That was a pretty weird result, but it turned up twice, not once, and the upshot was to run another trial," Fast told BioWorld Financial Watch. "You can't completely dismiss these kinds of results. On the other hand, you have to take it with a big grain of salt."

Fast said about 30 trials of AIDS vaccines are ongoing, in various stages.

"The first concept people try to use was the one VaxGen just tested," she said. "If we just make a recombinant version of the [HIV] envelope, will we be able to get good neutralizing antibodies? I think we now know that's not the panacea."

Another approach is to try to induce cytotoxic T cells that release substances to block viral replication and actually kill HIV-infected cells. These vaccines, "if they don't eradicate HIV, they might minimize the impact of an infection," Fast said.

Her organization has several vaccine candidates in the latter category, one currently being tested in Phase I and Phase IIa trials. IAVI also has partnerships, and "some are biotech companies with a lot of their own capabilities, but we provide money and advice and clinical trial support."

One such arrangement is with Targeted Genetics Inc., which is developing a recombinant adeno-associated virus for a potential HIV vaccine. "We're getting ready to submit regulatory applications," Fast said, adding that a third partner in the deal is the Children's Research Institute of Ohio.

IAVI also does "a lot of work in developing countries," she said, with trials under way in Kenya and Uganda, early work in India and China, and trials expected to begin in South Africa.

Investors, though, are keeping their eyes less on developing countries and world health than on money. Seiler, whose firm co-managed VaxGen's initial public offering in 1999, said the hit taken by the stock on the most recent news was severe, but could have been worse since shares were not trading "at the kind of premium you'd expect" for a company with a product in Phase III from which success was expected.

Apart from this, "for reasons that are not completely clear, [VaxGen] seems to be a company and management people love to hate," Seiler said. "People were hoping for them to fail."

Whether the vaccine decisively failed or not has yet to be sorted out.

"We are just at the beginning of the process of analyzing the data," said Phillip Berman, VaxGen's senior vice president of research and development and the vaccine's inventor, during a conference call. The company will take a dual approach, examining the immune response shown in those vaccinated and checking the genetic sequences of the viruses that infected trial participants.

"It is a long path," Berman said.