Getting It Right Early

The STEP trial offers lessons for future preclinical/clinical AIDS vaccine development

By Simon Noble

The reverberations are still being assessed, the data is undergoing analysis, and researchers are still discussing and debating next steps, but after the failure of Merck's adenovirus-based vaccine candidate (MRKAd5) some familiar questions are already being rehashed as to what can be done differently to develop the next generation of vaccine candidates.

Animal model

Like virtually all animal models of human disease, there are differing opinions as to which of the various animal models most faithfully recapitulates HIV infection in humans. "Animal models are animal models. The key word is model," says Emilio Emini, head of vaccine research and development at Wyeth. "All models are approximates," says Stanley Plotkin, executive advisor at Sanofi Pasteur, "it is not unusual in vaccine development to find that a vaccine doesn't work as well in humans as it does in animals."

One of the many issues that AIDS vaccine researchers have wrestled with in recent years concerns the fine specificities of the various nonhuman primate (NHP) and simian immunodeficiency virus (SIV) models. Viruses like the pathogenic SIVmac239 have been used extensively in challenge studies to test vaccine concepts, but this has been viewed as a particularly stringent test because it replicates to high peak viral loads and more rapidly causes disease in rhesus macaques than HIV does in humans. Some years ago hybrid simian-human immunodeficiency viruses (SHIVs) were constructed to try to provide a more reasonable challenge for a vaccine concept to protect against. The most commonly used is SHIV-89.6P, which contains tatrev, and env genes from HIV. But the pathogenesis of this hybrid virus is markedly distinct—it causes rapid and almost complete loss of CD4+ T cells in macaques, elicits neutralizing antibodies, and uses CXCR4 rather than CCR5 as a coreceptor when infecting cells. Paradoxically, SHIV-89.6P challenge has proven relatively easy to protect against with many vaccine approaches, resulting in preserved CD4+ T cells and greatly reduced viremia.

Since the halting of immunizations in the STEP trial, the fact that MRKAd5 was most notably successful against SHIV-89.6P challenge in NHPs (Nature 415, 331, 2002) has been a point of discussion. A similar vaccine candidate—the Merck Ad5 backbone encoding only Gag, rather than the Gag, Pol, and Nef included in MRKAd5—was ineffective against an SIVmac239 challenge when administered alone, and only marginally effective when accompanied by a DNA prime vaccination (J. Virol. 79, 15547, 2005).

Strong opinions on the relative veracity of the SIV or SHIV challenge models have been expressed in the past. "The field as a whole has preferred the SIV model," says Gary Nabel of the Vaccine Research Center at the National Institute of Allergy and Infectious Diseases (NIAID). "We have always felt that SIV is probably a better approximation of human disease." Indeed, John Moore of Cornell University and Mark Feinberg of Merck warned some time ago that "SHIV-89.6P may be a sheep in wolf's clothing, with the potential to lead the AIDS vaccine development effort down the wrong path" (Nature Medicine 8, 207, 2002).

The results of the STEP trial have only reinforced this position for some researchers. "We've already learned some pretty important things from this trial. For one, we have learned that the SHIV model does not have very good predictive value," says Tony Fauci of NIAID. When more data from the STEP trial was released at the HIV Vaccine Trials Network (HVTN) meeting in Seattle on November 7, the discussion again turned to the predictive value of the NHP model. Jeff Lifson of the National Cancer Institute in Maryland said, "The NHP data strongly correlated with these results."

David Watkins of the University of Wisconsin, Madison says, "This shows that the monkey model is highly relevant, but only with a rigorous SIV challenge, not a SHIV-89.6P challenge. We had the fond hope that it might be easier to protect humans than it was to protect monkeys against rigorous SIV challenge, but that proved not to be the case."

Assessing assays

Another question raised by the STEP trial is the validity of the current assays used to gauge immunogenicity. There has been increasing opinion in recent years that the interferon (IFN)-g ELISPOT assay is not really up to the job of assessing the relative efficacy of vaccine candidates. The problem is that ELISPOT simply measures a T cell's ability to secrete IFN-g (or other soluble factor) and this does not reliably correlate with any particular biological function of that cell. "An ELISPOT is an easy first check of immunogenicity. I think no one is satisfied that by itself it is the only assay one should do," says Peggy Johnston of the Division of AIDS, part of NIAID.

In the STEP trial, high levels of immune responses were induced by MRKAd5 to the HIV antigens and these responses were similar overall to those reported in Phase I and II trials (see A STEP back?). In the group with low levels of pre-existing immunity to Ad5, 79% of participants generated immune responses at a level greater than 55 spot-forming cells per million peripheral blood mononuclear cells (PBMCs), as measured by IFN-g ELISPOT assay, to at least one of the HIV antigens included in the vaccine. In the 1,500 volunteers with high titers of antibody against Ad5, 62% generated responses to at least one antigen. Danny Casimiro of Merck presented data at the HVTN meeting showing that even though the vaccine was not effective, it did induce high levels of T cells secreting IFN-g and interleukin (IL)-2. Watkins says this trial once again indicates that "immunogenicity is not efficacy."

Researchers have begun to develop improved assays that accurately reflect biological function-such as in vitrolysis of HIV—infected target cells and consequent virus inhibition, for instance—to better assess vaccine candidates in preclinical and clinical development. "The ELISPOT really only scratches the surface," says Emini, "we know the science has moved well beyond the ELISPOT." Casimiro says these assays will be conducted with the samples from the STEP trial to look at the cytolytic potential of the T cells induced by MRKAd5 and their ability to neutralize virus.

Watkins draws a parallel with neutralizing antibodies. "You might have binding antibody, but if you don't have neutralizing antibody it will be very difficult to protect. The same is probably true with CTLs [cytotoxic T lymphocytes]; ELISPOTs are the equivalent of measuring antibody binding to envelope, but we really don't have an assay for CTL efficacy. We desperately need the equivalent of a neutralizing antibody assay."

Some researchers have balked at the prospect of abandoning the well-established and validated ELISPOT assay, but Watkins's opinion is that "just because it's difficult doesn't mean it's not the right thing to do. We've got to find better assays, and production of cytokines is not the way to go—we need functional assays."

Rick Koup of the VRC thinks ELISPOT assays are still critical if "you want to know if your vaccine is stimulating a T-cell response," and says it is impossible to say definitively whether or not viral suppression assays will better evaluate efficacy until there is a candidate that is shown to be effective and suppresses virus in vitro but doesn't give a strong ELISPOT response.

"We really can't say at this point," says Fauci, "maybe we need better ELISPOTs, but right now the relationship between ELISPOT intensity and outcome is not known. We need a comparison, this is only one trial." José Esparza of the Bill & Melinda Gates Foundation concurs that researchers are in a Catch-22 situation, saying that "the only way to know which lab assay correlates with protection is after conducting a successful efficacy trial."

Future strategies

Beyond the finer points of animal models and immunological assays, some researchers have expressed opinions on what the STEP trial might mean for future preclinical and early clinical vaccine development strategies. Moore says that researchers will "have to demonstrate that their favorite vaccine is better than what has gone before; we can argue about what 'better' means, but there is no point in advancing immunogens or vaccines that don't do any better than the ones that have failed."

"We have to be a lot more honest and rigorous about the preclinical assessment of vaccine candidates before they go into expensive and lengthy Phase IIb or III trials," says Watkins. He says the STEP trial "tells us that if you're not protecting against SIVmac239 with autologous challenge, then don't even bother going into the clinic." He also thinks novel approaches will be paramount. "We need creative, radical new ideas-to use a football metaphor, running up the middle is not going to work, we need to be creative in our playmaking. Those discoveries are going to have to come from basic research labs; the big science approach is not going to be effective, or at least it will be more difficult."