The RV144 trial

It was therefore a welcome surprise when the much-criticised RV144 trial reported some degree of efficacy in September 2009, the first vaccine efficacy trial ever to do so.¹ However the result hovered tantalisingly close to the boundary of statistical insignificance, and exactly how the vaccine exerted its effect is still the subject of study.

RV144 compared vaccine to placebo in 16,402 adults, starting in 2003 and ending in 2006. It recruited adults aged 18 to 30 in two provinces of Thailand with high HIV prevalence but did not specifically target people at high risk of HIV infection.

The vaccine combined a ‘prime’ vaccine called ALVAC‐HIV (vCP1521) with a boost of the AIDSVAX gp120 vaccine, an envelope protein segment from HIV subtypes B and E. VaxGen, IDSVAX’s original manufacturers, had dissolved soon after the original AIDSVAX trials but executives from VaxGen went on to form Global Solutions for Infectious Diseases, a non-profit organisation. The presence of AIDSVAX in the vaccination regimen was one reason why many vaccine experts had not expected the trial to show a positive result.

ALVAC(R)-HIV, developed by Sanofi Pasteur, consists of a viral vector containing genetically engineered versions of three HIV genes (env, gag and pro). The ALVAC vector is an inert form of canarypox, a bird virus, which cannot cause disease or replicate in humans; it has been used in trials of cancer vaccines too.

The prime-boost combination of ALVAC(R)-HIV and AIDSVAX(R) B/E lowered the rate of HIV infection by 31.2% compared with placebo. This reduction was statistically significant, meaning that the possibility of the result being due to chance is low, but the confidence intervals for the estimate in the reduction in risk were wide (p=0.039, 95% confidence interval 1.1%-51.1%). This means that there was a 95% chance that the ‘true’ vaccine efficacy lies somewhere between these limits.

Seventy-four placebo recipients became infected with HIV compared to 51 in the vaccine regimen arm. The vaccine regimen had no effect on the amount of virus in the blood of volunteers who became HIV-infected during the study.

The results were surprising to many. In a press conference call after the results were announced, Colonel Nelson Michael, Director of the US Military HIV Research Programme (MHRP), which funded 25% of the $119 million trial, said that the most surprising and probably significant aspect of the result was that the vaccine appeared to produce a mild protective effect while producing no effect on the viral load in those who were infected.

“We are humbled by this finding,” he said. “This may have turned several key assumptions in HIV vaccine research on their heads.” He added that the trial showed that “human experimentation trumps everything we do in animals and test tubes.”

What Michael was referring to is that the RV144 trial consisted of two vaccines with two modes of action. The ‘prime’ vaccine, ALVAC, consisted of HIV genes contained in a canarypox virus vector. This model of vaccine delivery was designed to stimulate a cellular immune response. This does not prevent initial infection, but the scientific model – supported by data from some monkey studies - was that the cytotoxic T-lymphocytes (CTLs or CD8 cells) thereby generated would subsequently reduce HIV proliferation, contain viral load, and slow or stop progression to AIDS.

ALVAC appears to have made no difference to viral load at all, a finding at least in harmony with the failure of the STEP trial. But at the start of RV144 study, CTL vaccines appeared promising.

AIDSVAX, on the other hand, was designed to produce an antibody response against the gp120 protein on HIV’s surface, a concept thought to have been discredited in 2003, just after RV144 started, when it failed its earlier trial. The assumption was that HIV’s genetic hypervariability would mean that any antibody response would be hopelessly specific; at the time HIV research pioneer Dr Robert Gallo said: “This is not a vaccine approach that was based on science.”

Nonetheless what RV144 appears to have done is block infection without containing viral load in those infected, which looks like an antibody response, though Nelson Michael added that a longer-term study, RV152, was following up a proportion of trial participants to see if there were longer-term effects on viral load.

When asked what was causing the vaccine’s efficacy, Anthony Fauci, director of the National Institute of Alliergies and Infectious Diseases (NIAID) commented: “Simply, we don’t know. We did not see broadly neutralising antibody responses, though there is an indication of a small antibody response. We also saw a lympho-proliferative response but didn’t see an increase in CTLs (CD8 cells).

"We may not even have started to measure the correct immune parameters in the body that in fact indicate protection against HIV.”

Fauci said there were a number of further questions left unanswered by the trial. These included:

• Was the vaccine’s effect durable or would further boosters be needed?
• Are there ways of improving this model’s efficacy?
• Would we see similar results in high-risk populations such as men who have sex with men, injecting drug users, and high-risk heterosexual people?

Regarding the last question, MHRP Deputy Director Jerome Kim said that, contrary to some reports, RV144 did not exclude high-risk people. At baseline a small number of participants did turn out to be MSM and sex workers. The trial was designed to vaccinate a “community sample”, in other words a truly representative cross-section of the Thai population, “ranging from people at high risk to people at no risk”.

However, Kim also speculated, one reason for the positive result may have been that “the average intensity of exposure to HIV may have been lower than in high-risk populations…we know a sufficiently large inoculum of virus can overcome any vaccine-generated response.”

In the wake of the trial, some analysts and community groups urged caution, pointing out the extremely wide confidence interval for the results and how perilously close they were to statistical insignificance. The significance of the result hangs on a handful of HIV infections. Was this, in fact, a study with a freak result which reflected no real efficacy?

News had leaked on other data from the trial, which suggested that if analysed in other ways, the results were not statistically significant. In particular, concerns were raised about why the results of the less encouraging per-protocol analysis were not released to the world’s press at the same time as the headline result.

At the Paris AIDS Vaccine 2009 conference, study investigators said statistical analysis showed the results were robust.

Nelson Michael told the conference that three analyses of the study had been planned:

• An intent-to-treat analysis, which included all trial participants randomised, regardless of whether they received the vaccine or not. This analysis found an efficacy of 26.4%, which was not statistically significant (95% confidence interval -4% to 47.9%, p=0.08).
• A modified intent-to-treat analysis, which excluded any randomised participant who was discovered to have evidence of HIV infection through viral load testing prior to vaccination. This analysis includes study participants who may have missed some of the vaccinations, and goes some way to estimating the real-life efficacy of the vaccine. This was the analysis that found an efficacy of 31.2% (95% confidence interval 1.1%-51.1%, p=0.039).
• A per-protocol analysis excluded all participants who missed any doses of the vaccine, or who received vaccinations on days other than scheduled study visits. A per-protocol analysis in a vaccine study is designed to test the efficacy of a particular vaccination schedule. This analysis excluded around 25% of study participants and thus failed to capture around 31% of infections that occurred during the study (just over half of them in the first six months). This showed a reduction in risk of 26.2% but this was not statistically significant (95% confidence interval -13.3% to 51.9%, p=0.16).

In the New England Journal of Medicine report on the trial, the investigators said that the result of the modified intent-to-treat analysis remained statistically significant regardless of the statistical method used to test significance. The investigators used no fewer than six different tests to query the magnitude and robustness of the effect observed, and found that all produced results within the same range, a p value lying somewhere between 0.03 and 0.05.

The researchers concluded that the modified intent-to-treat analysis, which had always been planned as part of the study and which was used throughout the study as the primary analysis to determine if the trial should be stopped on the grounds of futility, gave the most clinically useful information for making future decisions about how to take the vaccine forward, since it most closely reproduced the likely conditions in which a vaccine’s effectiveness would be judged in the field.

Nelson Michael also presented details of sub-group analyses which suggested that the vaccine exerted a greater protective effect in people with fewer sexual partners, although the study was not powered to produce definitive evidence on this question.

Just under half the participants were classified as low-risk, reporting one or no sexual partners in the six months preceding entry to the study and judging themselves to be at low risk. People with partners who were commercial sex workers, injecting drug users, HIV-positive or men who have sex with men were classified as medium or high risk. Individuals who themselves fell into any of these categories were classified as high risk. Only 24% of participants were classified as high risk.

When the vaccine efficacy was compared between the high-risk and the low- and medium-risk groups, there was a suggestion that people in the low- and medium-risk groups experienced a greater reduction in the risk of infection if they received the vaccine. Whereas the vaccine efficacy was 40% in the low-risk group and 46% in the medium-risk group, it was 3.7% in the high-risk group. Nonetheless confidence intervals overlapped for all three estimates, meaning that the results were not statistically significant.

Similarly intriguing was the trend towards a greater protective effect in the first year, indicating that the effect of the vaccine combination may wane over time.

Although strong cell-mediated immune responses were observed against HIV envelope and gag proteins in a modest number of participants, it is still impossible to determine if any immunologic parameters correlate with protection from infection, or to unpick the relative contributions of the two vaccines used in the study.

At the following AIDS Vaccine conference in Atlanta in 2010, Nelson Michael presented some findings on the immune responses induced by the vaccine.

The first finding was that, as suspected, vaccine efficacy waned over time. If the trial had been stopped a year after it started, there would have been 60% fewer HIV infections in vaccine recipients than in people receiving placebo, and this would have been significant. A few antibody responses had been seen that were modestly stronger than ones seen with other vaccines, especially to the regions of the HIV envelope known as the V1, V2 and V4 loops, and to specific peptides (short chains of amino acids) that formed part of those loops.² But exactly how RV144 induced the immune responses it did, and how vaccines can be developed to induce stronger responses, is still the subject of research.

One possibility is that the vaccine, while failing to generate broadly neutralising antibodies, did generate an antibody response consisting of normal, narrowly-neutralising antibodies to gp120, which in in vitro experiments bind themselves to HIV-infected cells that express sections (epitopes) of the gp120 protein on their surface. One possibility being explored by researchers is that while these antibodies are ineffective in generating a broad HIV-specific response themselves, their presence in the body does serve as a signal for cells belonging to other parts of the immune system – both CD8 lymphocytes and natural killer cells – to activate and destroy the cell thus ‘flagged’ by the antibodies.

This is called antibody-dependent cellular cytotoxicity (ADCC) and is not a new concept in immunology or HIV science, having been investigated since the late 1980s. Indeed, in people with chronic HIV infection, it may be part of the process that stimulates a pathogenic inflammatory response that depletes the immune system.³ However in people not yet infected it could be a crucial part of what constitutes an effective immune response to HIV, linking as it does the three main branches of the immune system. Moreover researchers have found that gp120 epitopes do result in the early generation of large amounts of ADCC-mobilising antibodies.⁴ Whether ADCC processes turn out to be a crucial part of what makes an HIV vaccine effective, recognition of the complexity of immune response and the way that parts of the immune system work together has led most researchers to conclude that promising HIV candidate vaccines are most likely to consist of components that stimulate both antibody- and cell-mediated immunity.