CARNAL NATION

"For First Time, AIDS Vaccine Shows Some Success," read the September 25, 2009, headline in the New York Times, announcing preliminary findings from a clinical trial in Thailand suggesting that an experimental combination vaccine known as ALVAC/AIDSVAX reduced the risk of HIV infection by about 30%.

"AIDS Vaccine Protects People, Shocks Researchers," wrote Reuters, while the Sydney Morning Herald trumpeted, "HIV Vaccine Breakthrough"—ironically, the same headline used a decade earlier by the BBC to describe yet another in the long string of promising vaccine candidates that later proved disappointing.

The trial results announced in September were hailed as "proof of concept" that a vaccine has the ability to prevent HIV infection in humans.

In 1997, President Bill Clinton predicted that an HIV vaccine would be available within ten years. In hindsight, he can be forgiven for underestimating the difficulty of the task. Even a partially effective HIV vaccine must overcome multiple obstacles, including the natural variability and rapid mutation of the virus, its ability to hide long-term in inactive immune cells, and the fact that it attacks the very cells responsible for an effective immune response.

Within a few days of last month's announcement, the latest "breakthrough" also lost its luster, as additional data showed that the results were not as good as the U.S. Military HIV Research Program—which issued the initial September press release—had led us to believe.

Thai Trial Findings

The trial, known as RV144, was the largest HIV vaccine study conducted so far in humans. It included more than 16,000 HIV-negative men and women in Thailand who were randomly assigned to receive either the vaccine combination consisting of four doses of the ALVAC primer vaccine followed by two AIDSVAX booster shots—or else inactive placebo injections. Study volunteers were tested for HIV every six months and counseled about disease prevention and safer sex.

The two genetically engineered vaccines contain genes from HIV strains common in Southeast Asia. Both ALVAC (HIV genes carried by a canarypox "vector") and AIDSVAX (made up of proteins from HIV's outer envelope) failed to protect people against HIV infection when used separately in earlier trials, so many experts were skeptical from the outset that they would work well in combination.

The trial results announced in September were hailed as "proof of concept" that a vaccine has the ability to prevent HIV infection in humans, which some scientists and activists had come to doubt after years of disappointing findings.

"Until now, we’ve had evidence of feasibility for an AIDS vaccine in animal models," said Seth Berkeley, president of the International AIDS Vaccine Initiative. "Now, we’ve got a vaccine candidate that appears to show a protective effect in humans, albeit partially."

Interpretation—or Misinterpretation?

A couple of issues come into play when discussing the interpretation—or misinterpretation—of the trial results. First is the difference between absolute and relative risk, which can easily trip up people who are not well-versed in statistics (and, all too often, even those who are).

As an example, if 600 out of 10,000 study volunteers (6%) become infected while using a vaccine compared with 900 out of 10,000 (9%) without the vaccine, that 30% reduction in risk means 300 lives saved—quite an impressive benefit. But if the figures are six people (0.06%) infected using the vaccine versus nine (0.09%) without, the same 30% risk reduction doesn't look so dramatic, and could easily be due to chance given the small numbers.

In this case, however, the problem wasn't just misinterpretation—it was missing information. As first reported by Jon Cohen in the ScienceInsider blog, the results released in September didn't tell the whole story.

Those results, from a so-called "modified intention-to-treat" analysis involving 16,395 trial participants, indicated that 51* people who received the vaccine combination and 74* who received placebo injections became infected with HIV, showing that the vaccine was 31.2% effective. The p-value of 0.04 indicated that the difference just reached the usual cut-off for statistical significance, meaning the results had a 96% likelihood of not being due to chance alone. However, the confidence interval—another standard statistical measure—was very wide, indicating considerable uncertainty.

Full data from the trial were presented last week at the AIDS Vaccine 2009 conference in Paris and published simultaneously in the October 20 issue of the New England Journal of Medicine.