Why is it so hard to make a vaccine against HIV?

When the then US Health Secretary Margaret Heckler made her 1984 forecast that a vaccine would soon be developed, the experts listening knew better than to expect a vaccine in a couple of years. Several scientists seated in the packed auditorium "blanched visibly" at Heckler’s declaration, according to Randy Shilts’ history of the early epidemic, And the Band Played On.¹

They were right to be cautious. It had taken 105 years after the discovery of the typhoid bacterium to develop a vaccine for typhoid. For whooping cough (pertussis) it had taken 89 years; for polio and measles 47 and 42 years respectively. But the time lag was getting shorter. It had only taken 16 years from the discovery of the hepatitis B virus to the development of a vaccine.

Almost thirty years after the discovery of HIV, however, a truly preventive HIV vaccine is clearly still many years ahead. Researchers previously optimistic about a vaccine have tempered their optimism in the last decade. Nonetheless the development of a vaccine for such a novel and difficult pathogen, with only a quarter-century of knowledge to work with, is not necessarily ‘behind schedule’. Researchers have had to temper undue optimism.

At the Barcelona International AIDS Conference in 2002, for instance, Jose Esparza, who was working at the time for the World Health Organization Vaccine Initiative, predicted that at least one Phase III trial of a workable vaccine would be underway within the following three years - and that there should be at least one effective HIV vaccine available by 2009.

Two years later at the Bangkok International AIDS Conference he was much less optimistic.² Esparza pointed out that the search for a first generation of HIV vaccines – ones that used simple viral proteins to elicit an antibody response – was started soon after HIV was discovered in 1984 and appeared to have finally run into the ground 20 years later in February 2003 when the first-ever Phase III efficacy trial of an HIV vaccine ended with failure (though AIDSVAX later secured for itself an unexpected afterlife, as we will see below).

He said that the search for the second generation – vaccines that elicit a cellular immune response and stimulate anti-HIV CD8 cells – started about 1990. He said it might take at least two decades to find out if this approach yields an effective vaccine. Indeed, the STEP trial results came along two years after this, showing that cellular vaccine development would not be straightforward.

Esparza therefore advocated for the development of a third generation of HIV vaccines, ones that include both a CD8-stimulating component but which also stimulate elusive broadly neutralising antibodies that would act against fleetingly exposed ‘conserved’ parts of HIV that cannot evade immune control.

Given the time taken to show proof-of-concept (or lack of it) for the previous two generations of vaccine candidates, therefore, Esparza forecast (in 2004) that an effective HIV vaccine might now not be available till 2017-2021.

HIV vaccine researchers tend to divide into ‘incrementalists’ and ‘serendipidists’. The former emphasise that no vaccine trial is a wasted one, as it teaches us more about immune responses. The latter point out that many effective vaccines have been discovered by chance by researchers testing entirely new hypotheses. They say that we need more basic research before we can develop a wide enough variety of plausible candidates to mount a vaccine trial programme in humans that is logical, feasible, and economical.

At the 2008 Conference on Retroviruses and Opportunistic Infections (CROI), in Boston, Ron Desrosiers of the New England Primate Research Centre at Harvard University³ mounted a critique of current HIV vaccine theory, saying that vaccine research had to go back to basic science and also change its philosophy of generating ‘pipeline products’ to be tested in large human trials.

Desrosiers said that the enormous genetic diversity of HIV, its ability “to replicate unrelentingly despite everything the immune system can throw at it”, the fact that the immune system cannot protect against superinfection, and the fact that we do not currently know what constitutes an immune response to HIV, all persuaded him that “the discoveries that are going to lead to a successful vaccine have not been made yet”.

Although some animal studies had demonstrated the efficacy of cellular CD8 vaccines in reducing viral load in subsequently infected animals, he said, an analysis of the trials in which such vaccines had been used against a broader range of viruses revealed much less efficacy, with none producing reductions of more than 0.8 to 1.5 logs (7- to 30-fold).⁴ The vaccine used in the STEP trial had contained only one single genetic sequence of each of its three HIV antigens, compared with the thousands of variants HIV could throw up.

He urged a return to basic discovery research.

Desrosiers was not the first scientist to warn of the Sisyphean nature of HIV vaccine research. In a paper in The Lancet in November 2005, pioneering HIV scientist Robert Gallo summarised the barriers to developing a vaccine and made recommendations as to future directions for research.⁵

He said HIV vaccine development was difficult because of the following factors:

• An HIV vaccine cannot consist of attenuated, actively replicating (live) HIV, due to possible reactivation.
• Killed whole virus (like the polio vaccine) might also be dangerous because one could not be sure one has killed all viral particles. A killed whole virus vaccine had also worked poorly in animal studies.
• There were successful vaccines that use subunits of viruses such as individual proteins: an example was the hepatitis B vaccine. However medical science was less experienced with them.
• There isno truly useful small animal model for studying HIV infectionand vaccines had to be developed in monkeys using SIV or the artificial monkey/human virus SHIV in monkeys, which did not have exactly the same immune effects as HIV.
• We do not know with certainty which immune response will provide protection; this is a major problem. Pre-efficacy studies of vaccines in monkeys and humans used correlates of immunogenicity such as CD8 cell response, but we do not know whether this immune response is in fact a correlate of efficacy.
• We have never before attempted to develop a vaccine against a retrovirus like HIV. Retroviruses, by integrating their genome into ours, are able to hide completely from immune surveillance within quiescent lymphocytes. This means that any vaccine has a small window of opportunity in which to prevent infection and would have to be extremely effective, repelling all attempts by HIV to attach to and infect host cells.
• HIV produces viral proteins such as tat and vif that actively interfere with what would otherwise be a potent anti-HIV response in both infected and uninfected cells.
• CD8 cellular vaccines do not block infection because they act at too late a stage, so, if they worked, would do so by reducing the viral load in chronic infection. They would not necessarily reduce the peak level of viraemia in the early burst of viral reproduction and might not control infections transmitted by people in acute infection.
• HIV is capable of developing immunity to the CD8 cellular response.

Gallo concluded by saying that instead of focusing on some elusive correlate of effective immunity, obtaining or approaching sterilising immunity should be the goal; both conceptually and experimentally we know of only one practical way to accomplish this, namely by eliciting neutralising antibodies that are broadly reactive against various HIV strains and that are expressed for long periods.

Otto Yang of the University of California, Los Angeles explained why he thought CD8/CTL (cytotoxic T-lymphocyte) responses appeared insufficient to contain HIV infection in a presentation to the HIV Vaccine Trials Network Conference in November 2007.⁶

It was unquestionable that CTLs could contain HIV in vitro and in some people in vivo, he said, if CTL responses were tuned exactly right and the cells recognised the viruses circulating in the body. However there were several reasons why CTL responses in vivo rarely did manage this feat.

• The pace of HIV replication is such that the body’s population of CTL cells would have about 24 hours after the first CD4 cell was infected by HIV in which to destroy enough infected cells to contain viral production to undetectable levels.
• A large part of HIV’s genome is devoted to the production of viral proteins that defeat natural immune responses: the tat protein induces a huge burst of proliferation in HIV-infected CD4 cells, and forces lymphocytes to mature into CD4 cells. The nef protein causes infected cells to downregulate – pull within themselves – the HLA molecules, whose job it is to display epitopes and warn the immune system that they have been invaded by foreign pathogens. Thus, despite this burst of viral replication, the immune system fails to ‘see’ the cells responsible for producing it.
• HIV has evolved a way of tricking the immune system into mounting the strongest and speediest response to the parts of HIV that are most variable. This creates a situation in which CD8 immune responses remain narrowly tuned to the infecting viral strain and do not respond to later variants – a situation Yang called ‘original antigenic sin’. The only way a CTL-inducing vaccine could be effective would be to do intensive research into those parts of the HIV envelope that are most strongly conserved across all clades.