AIDS vaccine results questioned

U.S. Army and Thai researchers announced to great fanfare last month that a combination vaccine had produced a statistically significant 31% reduction in new HIV infections in a trial of more than 18,000 people in Thailand -- a modest rate, but the first vaccine results that suggested it may eventually be possible to produce a vaccine against the deadly infection, which has killed more than 25 million people worldwide. In an unusual approach, the researchers decided to make the results public in a news conference rather than wait for formal publication of their findings. The complete results have never been made public.

Now, however, a secondary analysis of the results have suggested that the vaccine was not quite as good as people had believed, reducing infections by only 24%, which was not statistically significant, according to researchers who spoke with Science magazine. The first analysis included all 16,000 people who participated in the trial and produced the promising results. The secondary analysis -- which was part of the protocol and is considered normal for all vaccine trials -- excluded patients who did not follow the experimental regimen. When that was done, the results were less convincing, according to experts who have seen the data.

The study was heavily criticized when it was launched three years ago because each component of the combination vaccine had failed in previous trials. But the National Institute of Allergy and Infectious Diseases, which largely funded the $120-million trial, hoped that it would provide some insight into the kinds of biological responses that must be provoked by a vaccine for it to provide protection. The researchers have subsequently been criticized for how they released the results, but they say that they feared the results would be leaked before they presented them and that they hoped to preempt such an event.

Of course, no one in the public has seen all the data yet. Full details of the trial are expected to be made public Oct. 20 at a meeting in Europe, and the researchers say they are writing a paper to be submitted to the New England Journal of Medicine. Meanwhile, all those who fervently hope for an HIV vaccine can do is wait for further revelations.

-- Thomas H. Maugh II

Comments

[Aaron]

This is why you don't draw conclusions from the results of one single research study. Unfortunately, most news publishers are doing that all the time, and it causes false hopes like this. This just shoes the importance of a little skepticism now and then.

[TBentonM3]

In a clinical study one typically uses an intent-to-treat analysis in which every patient, even those who didn't receive the full treatment course, is included in the analysis. Generally patients who do not receive the full treatment course don't do as well and including them in the analysis usually decreases the observed effect. One may also choose to use a per protocol analysis in which only those patients who comply with the protocol and receive the full course of treatment are included. This results in an analysis of the results from a smaller number of patients than the intent-to-treat protocol, but may be more meaningful in understanding whether the proposed treatment is effective if used as directed. Unexpectedly, in this case, the per protocol analysis of a subset of the intent-to-treat patients seemed to show a decrease in effectiveness, resulting in a result that was not statistically significant. Since this appears to be a directional Phase 2 study in which the hypothesis being tested was whether a combination vaccine could be effective even though its individual components had not been shown to be effective, clinical relevance rather than statistical significance is more important. While the results of this study were modest, both the 24% and 31% improvement results are encouraging to me and suggest that additional trials of HIV vaccines are warranted, especially if the experts agree that this is a clinically relevant response. A better explanation of the unexpectedness of the per protocol analysis results, the clinical importance of a 25-30% reduction of new HIV infections, and whether the results "provided insight into the kinds of biological responses that must be provoked by a vaccine for it to provide protection" would have made this report a better example of journalism rather than just reporting.