The Immune-Based Therapies and Preventive Technologies Pipeline
as 52.1%. But given the poor track record of the candidates involved, the glass-half-empty view inevitably must consider that a large swathe of the numerical territory between 1.1% and 52.1% is indistinguishable from the efficacy of the vaccines being zero. Beyond the borderline nature of the findings, there was some controversy at the time of the initial announcement by the trial’s sponsors—the U.S. Military HIV Research Program, the Thai Ministry of Health and the U.S. National Institutes of Allergy and Infectious Diseases (NIAID). Rumors abounded that the statistically significant result was undermined by other unreported analyses. The publication of the data in the New England Journal of Medicine a few weeks later set these concerns to rest, as the issue turned out to be related to the strictest “intent-to-treat” (ITT) approach, under which the five individuals found to be HIV infected at baseline had to be included. Using this method the result no longer reached statistically significance, but given that exclusion of people who were infected prior to receipt of any vaccine is both logical and standard in these trials there was no reason for this to be controversial. The fact that so few infection endpoints could affect significance in this way does, however, highlight the statistical fragility of the main result (which is described as the “modified ITT” analysis). But even if the Thai trial had produced a statistical trend toward efficacy rather than achieving significance, it would have been important for the vaccine field to follow up, and this is indeed what is occurring. Data from participants will be mined in the hope of revealing vaccine-induced immune responses associated with reduced risk of acquiring HIV, known as correlates of immunity. While discovering correlates would be a huge advance, the marginal efficacy and limited numbers of samples may stymie the effort. Currently, the main theories point to some type of antibody-mediated effect, either relating to binding antibodies or antibody-dependent cellular cytotoxicity (ADCC) wherein nonneutralizing antibody responses promote killing of virus-infected cells. The reasoning behind these theories is partly based on evidence of a short-term protective effect in the trial; after the first year, infection rates in the vaccine and placebo groups were very similar, and this observation tracks with binding antibody responses, which peaked in magnitude after the last vaccine booster at six months and declined precipitously thereafter (Michael 2010). Evidence of ADCC was observed in the majority of recipients of the vaccine regimen in earlier trials (Karnasuta 2005). Less likely candidates for immune correlates include HIV-specific CD4 T-cell responses, detected in around 50–90% of vaccinees (as measured by lymphoproliferation to p24 or gp120 antigens) and HIV-specific CD8 T-cell responses that, while detected in around 20% of recipients in prior studies (Nitayaphan 2004), were essentially undetectable in the analyses reported in the New England Journal of Medicine paper (Rerks-Ngarm 2009). The lack of CD8 T cell responses is consistent with the failure of vaccination to measurably affect post-infection viral load among the trial participants who became infected.
49