As reported yesterday by AVAC's Advocates Network, NIAID released a statement this past Monday reporting that a total of 11 infections occurred in the South African study of Merck's HIV vaccine, dubbed the Phambili trial or HVTN 503. Immunizations in the Phambili were stopped before it was fully enrolled due to the results from the preceding international STEP/HVTN 502 trial, which showed no efficacy of the vaccine and evidence of enhanced susceptibility in a subset of immunized volunteers. Of the 11 infections in Phambili, seven occurred in vaccine recipients (although most had not received all three immunizations) and four in placebo recipients. Among the vaccine recipients, six infections occurred in people with titers of anti-Ad5 antibodies over 1:18 compared to three infections among people with similar titers in the placebo group. In participants with no detectable antibodies to Ad5 at enrollment, there was one case of HIV infection among vaccine recipients and one case of HIV infection among placebo recipients. Given the small numbers, the three additional infections in the vaccine group cannot be interpreted as evidence that the vaccine enhanced susceptibility in this population, although it is of course possible that such evidence would have emerged, had the trial not been stopped. The apparent difference in infections based on anti-Ad5 antibody titer (6 vs. 3 infections compared to 1 vs. 1) also involves far too few events for any inferences to be drawn, although it may still prompt speculation that a similar scenario to the STEP trial was developing. Notably, all but one of the cases of HIV infection in the Phambili trial were among women, who made up 360 of the 801 participants at the time immunizations in the study were halted.
NIAID have also released more materials regarding the multivariate analyses of the STEP data that have recently been conducted. The analyses find that receipt of the vaccine was associated with enhanced susceptibility but that this effect was strongest in uncircumcised men with antibody titers against the Ad5 vector of over 1:200. Circumcised men with undetectable Ad5 antibodies did not show any evidence of vaccine-related enhanced susceptibility. However, the picture is less clear for men with either risk factor (uncircumcised or Ad5 antibody titer over 1:200), as both groups showed some evidence of vaccine-related enhanced susceptibility in the multivariate analyses.
In terms of the association between anti-Ad5 antibody titers and risk of HIV infection seen in the placebo arm, the NIAID document states that: "Although the study was stratified on Ad5 titers, the Ad5 seropositive population differed substantially from the Ad5 seronegative population. When we looked in the multivariate analyses for association of Ad5 serostatus with infection risk independent of vaccination, Ad5 serostatus was not independently associated with HIV infection. We did find, though, that being North American was associated with an increased risk of acquisition. It may be that the reason for the higher rates of infection in the placebo arm among those who were Ad5 seronegative was because that subgroup was primarily in the North American cohorts, which reported riskier sexual practices and substance use. We are exploring this in other observational cohort studies."
So in this formulation, people in the placebo group with anti-Ad5 antibody titers below 1:18 were more likely to be North American, people with antibody titers between 1:18 and 1:1000 were a bit less likely to be North American and people with antibody titers over 1,000 were much less likely to be North American. Individuals with antibody titers over 1:200 were also less likely to be circumcised, so while circumcision appeared to have a protective effect in vaccine recipients, no such effect was seen in the placebo arm. People in the vaccine arm with higher anti-Ad5 antibody titers should presumably also have been at less risk because they were less likely to be North American, but the vaccine increased their risk somehow. To provide some context, this is the overall incidence table from STEP, with the confidence intervals included to give a sense of the uncertainty associated with the findings.
|Anti-Ad5 antibody titer||HIV incidence vaccine (95% CI)||HIV incidence placebo (95% CI)||Relative incidence V:P|
|less than 1:18||4% (2.5-6.3)||4% (2.5-6.2)||1.0|
|1:18-1:200||4.4% (1.9-8.8)||2.2% (0.6-5.5)||2.0|
|1:200-1:1000||6.1% (3.3-10.2)||3% (1.2-6.2)||2.0|
|over 1:1000||4.4% (1.2-6.2)||1.2% (0.2-4.5)||3.5|
The NIH is holding a summit on March 25 to discuss the implications of the STEP results for the future of HIV vaccine research, meeting and registration information are available online.
Update 2/29/08: Thanks to Sarah Alexander from the HVTN for pointing out several significant errors in the original posting:
1) the sample size for Phambili was slated to be 3,000, the same as for STEP, so it was not intended to be a smaller trial
2) women did not make up the majority of the Phambili enrollees (of the 801 participants, 360 were women and 441 men)
3) follow-up is continuing in both the STEP and Phambili trials so while immunizations and enrollment have ceased, the trials continue.
Apologies for the mistakes, which have also been corrected in the body of the post