(How) Can NHP Models Accelerate Vaccine Development?

It is a question of “how” not “if” non-human primate studies should be used to elucidate mechanisms of protection against HIV.

By Alan Schultz

The world needs and wants an AIDS vaccine, and definitive HIV vaccine efficacy can only come from human trials. There is no argument about that. But there are different opinions on the best path to reach the goal.

Some believe that advancing more candidate vaccines into clinical trials will accelerate the process of finding one that is efficacious. But the results of the five efficacy trials mounted since the discovery of HIV in 1984 underscore the difficulty of the problem. Three trials showed no evidence of efficacy, and another suggested the vaccine candidate possibly enhanced risk of HIV infection. Only one trial, RV144, provided a modest 31 percent efficacy.

It is important to recognize just how difficult a pathogen HIV is and how steep the hill is that researchers must climb to reach the goal of developing an AIDS vaccine. By comparison, consider the situation with Ebola. In response to the recent outbreak, there was an immediate call for development of a vaccine to protect against the deadly virus. As researchers set about this goal they could take comfort from the fact that up to 50 percent of Ebola-infected individuals recover from this terrible infection on their own. This is because 20 million evolutionary years of T- and B-cell immunity gave the human immune system the tools to contain and eliminate this new virus, as well as many others. The immune responses mounted against the Ebola virus in 50 percent of cases happened spontaneously and fast enough to effectively eliminate the pathogen. This means that further accelerating that type of an immune response through prior vaccination should be effective at reducing death and morbidity and, in fact, some Ebola vaccine candidates already look very promising in clinical trials.

This is very different from the situation with HIV. Although the progression to death is much slower with AIDS (10 years on average, compared to three weeks for Ebola), there are essentially no survivors of AIDS; there is not a single documented case of an HIV-infected individual clearing the infection on their own†. The timetable may be slow, but without drug treatment, the uniform outcome of AIDS is death. Not only does this depressing fact deprive us of learning what the immune system could “do” to block HIV, it also implies that the human immune system is essentially ill-prepared to “handle” HIV. Though vigorous T- and B-cell responses ensue after infection, HIV easily mutates and stays ahead of them all. A successful HIV vaccine likely will need to direct human immune responses down a path different from the one they normally take.

That the number of HIV efficacy trials is meager is a simple consequence of this lack of survivors. What do we need the vaccine to “do”? What are the response(s) that are worth accelerating to combat HIV? We don’t know. Without that knowledge, immunogenicity measurements in Phase I clinical trials tell us little about whether vaccine A will be more efficacious than vaccine B. This is why simply advancing more candidates into clinical testing is unlikely to provide a quicker path to an eventual vaccine.

Given this, how can we be confident that a vaccine is good enough to incur the enormous resources required to conduct a large efficacy trial? Could correlates of protection that have been proposed from the modest efficacy of prime-boost regimen tested in the RV144 trial become the Phase I standard? Unfortunately, a prospective test of this hypothesis won’t occur until completion of the upcoming efficacy trial of clade C ALVAC/gp120 vaccine candidates in South Africa, which is years away.
Developing vaccine candidates that can induce neutralizing antibodies against HIV is one clear pathway to an efficacy trial. Today’s neutralization assays are high-throughput, reliable, and show there are antibodies that can neutralize HIV, but breadth of neutralization is very narrow. Though rare monoclonal antibodies that have impressive breadth of neutralization can be isolated, the track record of inducing them is poor. Some excellent scientists are vigorously testing a rational approach to this problem by developing various vaccine immunogens designed to guide the immune system to develop these highly specialized antibodies, but progress is agonizingly slow. We need a contingency plan to learn if something else can protect.

Modeling HIV infection in animal models and reversing the process is another alternative. Instead of optimizing the protective response—knowledge of which we lack—in human trials, multiple vaccine concepts can be tested by immunizing and then challenging the animals. Protection observed in NHPs has been used as evidence that a concept may be valid and deserves further development, but what is also needed from such experiments is a strong correlate that could be useful in Phase I clinical trials.

Unfortunately, animal models of HIV/AIDS are far from ideal. The only non-human species in which HIV replicates is the chimpanzee, but experimentation in chimpanzees was abandoned for several reasons: no disease endpoint existed, experiments in chimpanzees are extremely expensive, there is an insufficient number of available animals, and most recently, ethical concerns were raised about experimentation. Mice reconstructed to contain human immune cells, so-called humanized mice, are at present incomplete models. The best compromise is simian immunodeficiency virus (SIV) and its pathogenic sequelae in rhesus macaques, which replicates many important features of HIV infection. Chimeric SIV/HIV strains known as SHIVs, which substitute HIV’s Env protein into SIV, allow for testing of HIV Env vaccines in non-human primates (NHPs). But as analog models, their relevance to predicting the outcome of HIV vaccine candidates in humans is unclear. Additionally, the first 20 years of SIV vaccine trials in macaques were largely unsuccessful, leading some to conclude that a vaccine simply couldn’t be made and others to condemn NHP models as unproductive, unsuitable, certainly unvalidated, and therefore irrelevant to HIV vaccine development.

Three major developments in the last 10 years have changed the debate considerably. First, the vaccine concepts being tested actually have improved, and second, NHP models improved as well. New repeat-exposure mucosal challenge models use lower doses of virus capable of transmitting approximately one infecting particle, which is a much better model of human sexual transmission. Encouragingly, there are now several vaccine candidates that either significantly reduce per-exposure risk of infection or prevent establishment of infection in a large proportion of vaccinated animals. Thirdly, the RV144 trial was transformational. Though efficacy was short-term and too low to merit licensure, it established in humans for the first time that vaccination could prevent acquisition of HIV, and did so in the absence of neutralizing antibodies. Finally, vaccine correlates questions have become broader and more refined. Do antibodies need to neutralize to contribute to protection? Do innate responses affect the development of adaptive immune responses, and can they be modified by vaccination?

To improve these partially protective vaccines we must move beyond empiricism and begin rationally guiding vaccine development by thoroughly investigating and comparing these concepts in protected and unprotected NHPs. Invasive analysis of tissues available from NHP studies is ideally suited to uncovering the mechanism of protection, not merely imputing a “correlate,” which may or may not be causative. What (and where) did the immune system do to intercept virus in these protected animals? While we wait for vaccines capable of inducing broadly neutralizing antibodies to be developed and for evaluation of the RV144 correlates from the follow-up trial in South Africa, redoubling efforts to analyze these partially protective vaccines in NHPs should be paramount. Despite their imperfections, NHP models remain the best way to make progress and intelligently guide HIV vaccine design.

† The tiny number of long-term survivors do not appear to mount strong immune responses against the virus but seem instead to be beneficiaries of unusual genetic resistances that provide no clues for vaccine development.


Alan Schultz is a preclinical team leader at the Vaccine Research Program of the Division of AIDS at the National Institute of Allergy and Infectious Diseases (NIAID) in Bethesda, Maryland. This article represents his personal views and not that of the Division of AIDS at NIAID.