Credit: Russell Kightley/Photo Researchers, Inc.

Despite the existence of antiretroviral drugs that can combat HIV, it is thought that the development of an effective vaccine will be required to control the spread of this deadly virus. In 2009, the results from a phase 3 trial (RV144) of a vaccine consisting of a combination of a canarypox vector expressing HIV-1 immunogens and a recombinant HIV-1 envelope protein offered some hope, as the vaccine showed modest efficacy in preventing HIV-1 infection (31.2%) in a Thai population. Barton Haynes et al.1 have now performed a case-control analysis to investigate the immune correlates of protection provided by this vaccine. They found that IgG antibody binding of the variable regions 1 and 2 (V1 and V2) of the HIV-1 envelope (Env) glycoprotein 120 inversely correlated with the rate of HIV-1 infection, and plasma IgA antibody binding to Env directly correlated with the rate of infection. We asked three experts to comment on how these results affect our understanding of immune-mediated protection from HIV and how they might be used to design future vaccine strategies against this virus.

Stanley A Plotkin

The trial showed that protection against viruses is not always mediated by neutralizing antibodies, or for that matter by cellular immune responses.

In view of the initial opposition by distinguished scientists to the trial, RV144 was a triumph of hope over skepticism, if I may paraphrase Samuel Johnson. Not only did it show some level of efficacy against acquisition of HIV, but also it gave us clues regarding the mechanistic correlates of protection against HIV2. As the correlate identified was a non-neutralizing antibody1, the trial showed that protection against viruses is not always mediated by neutralizing antibodies, or for that matter by cellular immune responses, despite the ability of the latter to control replication after acquisition of a virus. Undoubtedly, the entry of viruses into cells is often complex, involving multiple receptors3, and the identified utility of antibodies against the V1–V2 loops of HIV glycoprotein does not exclude the possible additional value in designing future vaccine strategies that elicit broadly neutralizing antibodies against other parts of the glycoprotein that mediate attachment and entry or that can stimulate effector CD8+ T cells capable of killing the earliest infected cells4,5.

Host protection against the acquisition of a virus can thus be influenced by multiple factors, some of which were illustrated in RV144. For example, efficacy seemed to be lower in higher-risk vaccinees, illustrating the importance of challenge dose and frequency of exposure, and efficacy waned with time, emphasizing the need for durable effector functions in a vaccine that is designed to protect over years of viral exposure. The as yet unexplained interference by plasma IgA antibodies in the efficacy of the vaccine tested in RV144 (ref. 1), whether due to these antibodies themselves or whether they act as a surrogate for some other immune marker, illustrates the complexity of protection. Notably, vaccination with the HIV envelope glycoprotein gp120 alone in the past did not elicit protective responses, and it is becoming imperative to define the mechanisms by which heterologous prime-boost regimens are more immunogenic than homologous boosts. Nevertheless, the way is now open to improve antibody responses to multiple portions of this HIV glycoprotein through new constructs, additional vaccine doses or better adjuvants.

Emeritus Professor of Pediatrics, University of Pennsylvania, Philadelphia, Pennsylvania, USA, and Vaxconsult, Doylestown, Pennsylvania, USA.

Harriet L Robinson

Efficacy trial results have generated seismic shifts on the path to an HIV vaccine. In 2003, antibody elicited by gp120, which mediates binding of HIV to cells, failed to protect against viral acquisition. In 2007, cytotoxic T cells, which recognize and kill infected cells, also failed to protect. And then, in 2009, in the RV144 trial conducted in Thailand, the first protection, albeit modest, was achieved by a vaccine that had been scorned for its potential to elicit protective immune responses6,7.

Now, a collaborative report involving 43 authors identifies two vaccine-elicited correlates of risk for RV144 participants1. The first, IgG directed against a scaffolded V1–V2 region of the HIV envelope (Env), correlated inversely with risk of infection; whereas the second, monomeric Env-specific IgA, correlated directly with risk of infection. The V1–V2-targeting IgG did not have neutralizing activity and thus implied that non-neutralizing IgG was mediating prevention of acquisition, a new and fresh concept for HIV vaccine development.

Interestingly, the IgA correlate identified by these authors corroborates the IgG correlate if one hypothesizes that monomeric serum IgA was competing for binding with non-neutralizing, yet protective, IgG. IgG and IgA have distinct Fc regions that bind to different Fc receptors and immune molecules, such as complement proteins, to mediate the recognition of opsonized particles, lysis of cells and degranulation of effector immune cells. Thus, the correlation with reduced risk of infection for binding of IgG, but not monomeric IgA, suggests that protective non-neutralizing activities were being mediated by interactions of the Fcg region of bound IgG, but not the Fca region of bound IgA, with physiological effectors.

Further identification of targets, coupled with passive transfers of mixtures of monoclonal antibodies that bind these targets, are now needed.

Not all targets for the binding of non-neutralizing IgG are protective. Currently, two targets have shown some protective activity against HIV: the scaffolded V1–V2 region in the RV144 vaccine trial and the conserved immunodominant region of the gp41 subunit of Env8. A trend for the protective potential of the latter was shown in preclinical passive-transfer studies. Further identification of targets, coupled with passive transfers of mixtures of monoclonal antibodies that bind these targets9, are now needed to test whether non-neutralizing activities of IgG can be moved from a correlate of risk to a correlate for protection for HIV vaccines.

Chief Scientific Officer, GeoVax, Smyrna, Georgia, USA.

Miles P Davenport

A follow-up study of the RV144 HIV vaccine trial has suggested that antibody binding to the virus may protect against infection1. As the authors note, the statistical approach was designed to generate hypotheses, not rigorously demonstrate association, and this study is more about finding a correlate of risk of infection rather than a clear cause of protection. The results imply that HIV-binding antibodies may reduce the risk of HIV infection, perhaps through new mechanisms such as antibody-dependent cell-mediated cytotoxicity. Although serum concentrations of HIV-binding IgG antibodies were associated with decreased infection by the virus, HIV-specific serum IgA antibodies (usually associated with mucosal immunity) seemed to negate this protection. This observation is reminiscent of the results of the original trial of an HIV gp120 immunogen (one of the components of this vaccine), where high levels of antibody may have been protective but the impact of this on overall protection was negated by the fact that low levels of antibody may have increased infection in immunized individuals10.

Skeptics of large HIV vaccine trials may remain unconvinced, as doubts remain over the original trial's marginally significant (and modest) protection.

Skeptics of large HIV vaccine trials may remain unconvinced, as doubts remain over the original trial's marginally significant (and modest) protection (31% efficacy, P = 0.04), which was only seen with a 'modified intention-to-treat' statistical analysis11. However, advocates of HIV vaccine trials may suggest that this study provides robust support for the partial efficacy observed in the RV144 trial and that the next step is to find HIV vaccines that induce high IgG and low IgA antibody responses. It is important to remember that the immunized individuals did not receive the alternative of either an IgA-inducing vaccine or an IgG-inducing vaccine. They all received the same vaccine and responded in different ways. Therefore, the IgA effect might result from IgA 'blocking' of IgG binding, or, alternatively, a propensity to produce IgA-biased responses may be a marker of mucosal immune activation. Understanding the antiviral mechanisms of IgG-binding antibodies and the host factors that led to high IgA levels may be a helpful first step in both validating the results of this analysis and moving the HIV vaccine field forward.

Professor of Medicine, Centre for Vascular Research, Faculty of Medicine, University of New South Wales, Sydney, New South Wales, Australia.