VaxGen’s AIDSVAX trial

March 2003

In mid-1998, with considerable media hoopla, VaxGen announced the initiation of a phase-3 study of its recombinant gp120 AIDS vaccine, AIDSVAX. Even Infectious Disease News seemed to participate in the celebratory mood; the sub-headline of the story about this said, “FDA approval of VaxGen’s clinical trials is another sign that science is closing in on a workable AIDS vaccine.” In reality, it meant no such thing.

A long road to trial

This vaccine had a curious history well before the start of the trial. Initially developed and endorsed by the late Jonas Salk, it was subsequently taken over and further developed by Genentech. Subsequently, however, Genentech put the vaccine on the shelf after the NIH AIDS Vaccine Advisory Committee recommended against phase-3 trials of gp120 vaccines because antibody responses were believed to be non-protective, failed to neutralize HIV infection in tissue-culture assays and failed to produce a consistent cytotoxic T-cell response. In 1996, the product was spun out of Genentech and into VaxGen, a start-up company co-founded by Dr. Donald Francis, who previously worked for Genentech. Dr. Francis is well known to be something of a “firebrand” when it comes to AIDS vaccine development. When it became apparent that NIH was not about to fund a phase-3 trial of this vaccine with their money, Dr. Francis went about raising the necessary $20 to 25 million to support the trial from private sponsors, and the trial got underway.

A substantial majority of the AIDS scientific community were opposed to or at least failed to support this trial, since they had already concluded that the product was unlikely to be effective. The fact that the trial occurred at all was seen as more a political event than a scientific event.

Well, now — almost five years later — the trial has been concluded, and behold — it did not work – unless you believe strongly in subgroup analyses. The trial had more than 5,000 participants, and there was a 3.8% reduction in HIV infection in the vaccines as compared to controls, an insignificant reduction. This major finding may have been disappointing, but it surely came as no surprise, given the scientific skepticism that surrounded this trial from the start.

Subgroup analyses

In a number of subgroup analyses, however, VaxGen is claiming highly significant protective effects among 314 black volunteers and also significant protection among Asians. These minority subgroups had a 67% lower rate of HIV infection than those in the control arm. This represents, according to VaxGen, only a 1% to 2% likelihood that these results could be due to chance alone.

Interestingly, the Hispanic minority arm evidently failed to show a protective effect. Media reports have suggested that there appeared to be a difference between the magnitude of antibody responses in the white majority population that was not protected, and the black and Asian minority populations that appeared to gain some protection. Nonetheless, the complete data set has not been fully analyzed yet, and has not yet appeared in the scientific literature, so firm and substantive conclusions simply cannot yet be made.

Nonetheless, it is well known that subgroup analyses can lead to all kinds of mischief. If one does 20 subgroup analyses in a body of data that shows no difference between two arms of a trial, there is an approximate 50% likelihood that one significant difference will be found. Both numerators and denominators get progressively smaller in subgroup analyses, and the likelihood of spurious findings increases. That may account for the findings in the VaxGen trial, since the numbers of infections among the black and Asian minorities were quite small and the confidence intervals quite broad. That possibility noted, it is worth emphasizing again that any conclusions are premature at this point.

Is a failed trial better than no trial at all? I asked that question but failed to answer it five years ago in my editorial comments; nor can I answer it now. As noted in the news account in this issue, however, there is concern within the Gay Men’s Health Crisis in New York about the “spin” of VaxGen’s news release, and how it might impact the participation of U.S. minority populations in future vaccine trials. There is also concern that false hopes may be raised in Thailand, where a very similar VaxGen vaccine is close to finishing a phase-3 trial. Indeed, Southeast Asia would be a major market for any AIDS vaccine. If there is no protection found in the Thai trial, that would strengthen the concern that the Asian subgroup analysis results in the U.S. trial were spurious.

Two review articles on AIDS vaccines that have appeared in the last six to eight months were of interest, and which I would commend to readers who do not spend their lives with AIDS vaccine issues but who would like a glimpse of where the field is heading. These are, Letvin NL. Strategies for an HIV vaccine. J Clinical Invest. 2002;110:15-20; and Smith KA. The HIV Vaccine Saga. Med Immunol. 2003;2. The former is a current scientific review, the latter more of a historical review.

On a (somewhat) related issue, two and a half years ago in JAMA, Dr. Lawrence Gostin proposed a revision of the now 13-year-old CDC policy on preventing transmission of bloodborne pathogens within the health care setting, specifically, the section of the policy dealing with the HIV-positive health care worker. Dr. Gostin is a well-known and articulate legal figure in the HIV public policy arena and a champion of individual civil liberties and the rights of HIV-positive people. He suggested five specific ways in which the policy should be revised. Space does not permit me to detail his suggestions, but I suggested at the time that the ball was now in the CDC’s court.

No revision of that policy document has been forthcoming, and it appears increasingly likely that the CDC prefers to keep that bottle tightly corked — hoping to keep the genies, so widespread when that policy was conceived — well contained. That policy has been aptly described as “a sleeping dog,” and as we all know, sometimes it is best to let sleeping dogs lie.