Adenovirus Vectors: Promise and Possible Pitfalls

New studies show the promise of alternate serotype adenovirus vector-based HIV vaccine candidates, but also raise questions about their ability to overcome pre-existing immunity issues

By Andreas von Bubnoff

More than four years ago, the HIV vaccine field was shocked and disappointed when a Phase IIb trial known as STEP was halted ahead of schedule because there was no evidence that the vaccine candidate protected against infection or controlled the virus in those who became infected. The candidate, developed by Merck and known as MRKAd5, used an adenovirus serotype 5 (Ad5) vector to deliver the HIV antigens Gag, Pol, and Nef, and was widely considered to be one of the most promising in clinical trials at the time.

Even worse, subsequent analyses of the STEP trial indicated that MRKAd5 actually led to an increased risk of HIV acquisition in a subgroup of vaccinees who were uncircumcised and had pre-existing Ad5 antibody immunity due to natural exposure to this serotype of the common cold virus. In addition, pre-existing Ad5 seropositivity was found to dampen HIV-specific cellular immune responses to MRKAd5.

To circumvent these issues of pre-existing immunity, a few research groups have been exploring vaccine regimens that use alternate serotype Ad vectors that are less common worldwide, including the serotypes Ad26 and Ad35. These vectors also elicit different immune responses than Ad5, according to Dan Barouch, professor of medicine at Harvard Medical School and chief of vaccine research at Beth Israel Deaconness Medical Center (BIDMC), who led a recent study that provided promising results with vaccine regimens containing these vectors.

The study found that among several prime-boost regimens that could partially protect rhesus macaques from challenge with simian immunodeficiency virus (SIV) mac251, considered one of the tougher challenge viruses in pre-clinical studies, the two that were best overall were Ad26 combined with modified vaccinia Ankara (MVA) or with Ad35 (1). Based on these results, researchers are planning to test an Ad26/MVA regimen in Phase I clinical trials.

However, there is some concern that alternate serotype Ad vectors may not entirely sidestep the issue of pre-existing immunity. Another recent study by Nicole Frahm, associate director for laboratory science at the HIV Vaccine Trials Network (HVTN), and her colleagues has shown that pre-existing T-cell responses to Ad5 can also dampen cellular immune responses to the HIV inserts of MRKAd5 (2). These T-cell responses are cross-reactive, targeting conserved sites that are shared by multiple Ad serotypes. Such a dampening effect, therefore, might also be an issue for vaccines that use rare serotype Ad viruses as vectors. What that would mean for vaccine efficacy is still unclear, but it suggests that Ad-specific T-cell responses should at least be monitored when using rare serotype vectors in human trials.

The promise of protection

Before Barouch’s recent study, vaccinating rhesus macaques with live-attenuated SIV was the only approach capable of protecting them from the difficult to neutralize and highly pathogenic SIVmac251 challenge. However, this approach is thought to be too dangerous to test in humans because later studies found that the attenuated viruses used as a vaccine could regain their infectiousness and become pathogenic again.

Last year, Norman Letvin of Harvard Medical School and colleagues reported that a DNA/Ad5 prime-boost vaccine regimen that delivered the SIVmac239 Gag, Pol, and Env immunogens could partially protect rhesus macaques from repeat low-dose rectal challenge with SIVsmE660, but not from challenge with SIVmac251, which is more difficult to neutralize than SIVsmE660 (see Research Briefs, IAVI Report, May-June 2011).

The SIVmac251 challenge virus used in Barouch’s study also set a high bar for protection because it contained different viral sequences than those in the vaccine candidates, a so-called heterologous challenge. The vaccine carried the gag, pol, and env genes from SIVsmE543, an SIV strain closely related to E660. The Env protein in the vaccine candidates was 18% different in its amino acids from the Env in the SIVmac251 challenge virus—a difference that is greater than the difference between HIV strains in an infected person, but smaller than the difference between HIV clades. Barouch says the choice of a tough challenge was intentional. “We wanted to see whether vaccines made from a heterologous strain could offer an acquisition effect as well as a virologic control effect against a particularly hot virus challenge.”

To make the challenge even more vicious, the researchers used a dose of challenge virus that was about 100 times more infectious than HIV is in humans, so that one exposure to the challenge virus infected at least half the control animals. They compared how well four different prime-boost vaccination regimens (DNA/MVA, MVA/MVA, Ad26/MVA, and MVA/Ad26) carrying the 543 Gag, Pol, and Env immunogens protected groups of eight animals from the 251 challenge. They vaccinated the animals with a prime, followed by a boost six months later, and then gave them six weekly rectal challenges starting at month 12. All primes were done once except the DNA prime in the DNA/MVA regimen, which consisted of three monthly vaccinations.

It took just one challenge to infect at least half, and three challenges to infect all of the control and the MVA/MVA vaccinated animals. By contrast, it took two challenges to infect half the animals vaccinated with DNA/MVA, and three challenges to infect half the animals vaccinated with the Ad26/MVA and MVA/Ad26 regimens. Seven of the eight animals in all three heterologous prime-boost regimen groups became infected after all six challenges. This means that all three heterologous prime-boost regimens reduced the per-exposure risk of infection by about 80%.

In a separate experiment, researchers vaccinated 16 macaques with an Ad35/Ad26 vaccine regimen and challenged them with SIVmac251 in a similar way. This too resulted in about 80% protection per exposure. However, macaques that received the same vaccine regimen without the env gene were not protected, suggesting Env was required for protection. “It continues to add credence to why Envelope is going to be important to include [in vaccines],” says Nelson Michael, director of the US Military HIV Research Program and one of the senior authors of the recent study led by Barouch.

The researchers also found that the Ad26/MVA and Ad35/Ad26 heterologous prime-boost regimens not only provided about 80% protection per exposure, but also resulted in at least a two log or 100-fold lower set-point viral load 84 days after the last challenge. The Ad26/MVA regimen reduced set-point viral load by 2.32 logs, while Ad35/Ad26 lowered it by 2.18 logs. “For the combined endpoint of acquisition and virologic control, the two optimal regimens are the Ad26/MVA regimen and the Ad35/Ad26 regimen,” Barouch says. “[This] shows that optimal vaccine regimens can indeed offer at least a partial acquisition effect as well as a virological control effect against a heterologous challenge with a difficult to neutralize virus.”

For Michael, the “clear winner” of the two was the Ad26/MVA regimen, because the set-point viral load was undetectable in three of the seven Ad26/MVA vaccinated animals that got infected, but in only one of the 13 Ad35/Ad26 vaccinated animals that got infected.

Because some, but not all, animals vaccinated with the different vaccine regimens were protected or showed viral load control, researchers were also able to determine immunological correlates of protection and of viral load control. For protection, antibodies binding to Env were most important, although neutralizing antibodies to easy-to-neutralize viruses were also relevant, according to Barouch. For viral load control, nine different cellular and humoral immune responses were important, including Gag-specific cellular immune responses. This is consistent with previous observations that viral load control is associated with Gag-specific cellular responses, Michael says. “The view from 35,000 feet is antibodies for infection and cells for virologic control,” he says. In addition, the best viral load control, which was seen in the Ad26/MVA vaccinated animals, correlated with an unusually balanced immune response between central and effector memory CD4+ and CD8+ T cells compared with the DNA/MVA vaccinated animals, which showed no viral load control, Michael adds.

The observation that binding antibodies, as well as neutralizing antibodies, were the most important correlates for protection from SIVmac251 suggests, Barouch says, that antibody effector functions other than those measured in traditional neutralization assays may also be relevant for protection. “There is more research that’s needed to understand the full spectrum of antibody effector function,” he adds.

The finding that binding antibody to Env was an important correlate of protection in the most recent study suggests that SIVmac251 challenge might be a good animal model for what happened in RV144, the first vaccine trial to show modest protection from HIV infection, says Louis Picker, a professor of pathology at the Oregon Health & Science University who was not involved in the Barouch study (see A Bangkok Surprise, IAVI Report, Sep.-Oct. 2011). In RV144, the protective effect seemed to be due to Env binding and not neutralizing antibodies. “[SIVmac251] is actually a very good model to ferret out the correlates of this kind of protection,” Picker says. “[Barouch and colleagues] were able in the monkey model to recapitulate that weak acquisition protection with a non-neutralizing antibody correlate. This model will help dissect the results of the Thai trial and provide a path forward to improving that kind of efficacy.”

While there was 80% protection per challenge, Picker says, almost all vaccinated animals were infected after all six challenges, and it took just two additional challenges to infect half the vaccinated animals than it took to infect half the unvaccinated control animals. “This is significant but it’s still weak,” he says, adding that, “I think there should be more work in the monkey model to isolate and improve the effect before you bring it into humans.”

Still, Barouch, Michael, and others are now planning to test the Ad26/MVA regimen in Phase I clinical trials in humans. The vaccine will carry mosaic antigens that were developed by Barouch and Bette Korber of the Los Alamos National Laboratory and are computationally designed to achieve optimal coverage of the many different versions of HIV circulating globally.

Alternative adenovirus vectors, including Ad26, have already been tested in the clinic and have been shown to be safe and immunogenic, Barouch says. Other HIV candidate vaccines with such vectors are also being tested. For example, Barouch is collaborating with the HVTN, IAVI, and the Ragon Institute to conduct IAVI B003/IPCAVD-004, a Phase I trial of different prime-boost combinations of Ad35 and Ad26 vector-based vaccine candidates (seeVaccine Briefs, IAVI Report, Sep.-Oct. 2010).

Reassessing pre-existing immunity

Despite the promising results from Barouch’s study, there is concern that alternate Ad vectors may not be able to overcome some of the issues of pre-existing immunity. Until now, researchers have measured Ad5-specific antibodies in blood to characterize pre-existing immunity to Ad5 in STEP trial participants. Now, Frahm and colleagues looked for the first time at cellular immune responses to Ad5 in placebo recipients from the STEP trial a month after the time of the final vaccination, and found that about three quarters of the people with Ad5-specific antibodies also had Ad5-specific CD4+ T-cell responses. To their surprise, more than half of the placebo recipients without Ad5-specific antibodies also had Ad5-specific CD4+ T-cell responses. They used placebo recipients for the analysis because samples were not collected from STEP trial volunteers before vaccinations started.

This was unexpected, Frahm says, because it argues against the hypothesis that Ad5 seropositive STEP trial vaccinees were more susceptible to HIV infection because they had more Ad5-specific CD4+ T cells that were activated by the vaccination, resulting in an increased pool of HIV target cells. Frahm says that if this hypothesis were true, few, if any, Ad5 seronegative people should have Ad5-specific CD4+ T-cell responses compared with seropositive people, which is not what she and her colleagues found. In fact, they found that 93% of the seronegative vaccinees, but only 78% of the seropositive vaccinees, had such responses. “That’s exactly opposite of what we were expecting,” Frahm says.

However, her measurements were in blood and therefore don’t shed much light on what happened in mucosal tissues, which is where the infection most likely occurred. Previous studies did not find any hints that mucosal CD4+ T cells in STEP trial vaccinees were easier to infect by HIV than in placebo recipients in terms of CCR5 expression, activation markers, or homing markers, Frahm says. However, these CD4+ T cells might have been taken too late after vaccination and from too few vaccinees to see any effects, she adds. To better address this question, the HVTN is conducting HVTN 076, a trial in Ad5 seronegative people from whom mucosal samples are taken right after the vaccinations with a DNA/Ad5 vaccine regimen to see if mucosal Ad5-specific CD4+ T-cell responses are higher in vaccinees than in placebo recipients.

Frahm and colleagues also analyzed T-cell responses in volunteers from a trial called HVTN 071, in which volunteers received the same MRKAd5 vaccine as was used in the STEP trial, but which was stopped after two of three vaccinations because the results of the STEP trial had been announced. In these volunteers, the researchers found for the first time that pre-existing Ad5-specific CD4+ T-cell responses before vaccination were associated with dampened CD4+ T-cell responses to the HIV inserts and with lower breadth of the HIV-specific CD8+ T-cell responses to the vaccine. In addition, the Ad5-specific CD4+ T cells recognized epitopes shared by many different Ad vectors, in addition to Ad5. “We are the first to really show what epitopes are targeted by these T cells and that these epitopes are really identical across the different adenoviruses,” Frahm says. This cross reactivity suggests that the dampening effect on cellular immune responses to the HIV vaccine inserts might even affect vaccines that use rare serotype Ad viruses such as Ad26 or Ad35 as a vector.

Just how much this dampening effect would affect the efficacy of vaccines that use Ad viruses as a vector is unclear, Frahm says, because the magnitude of insert-specific responses needed for an HIV vaccine to be protective is unknown. Michael, who was not involved in Frahm’s study but wrote a commentary on it in the issue of the Journal of Clinical Investigation where the study appeared, agrees. “What that means in terms of predicting efficacy no one knows,” he says.

Also, the immune responses induced by Ad vectors are so high that the dampening effect may be negligible, Frahm says. “Just because [Ad vectors] are so good at inducing immune responses, they can overcome the pre-existing immunity and are still going to give a relatively decent immune response that may still be better than an immune response by a weak vector that may not have any pre-existing [immunity].” For now, she thinks these findings don’t necessarily mean that alternative Ad vectors shouldn’t be used in clinical trials or that people with Ad-specific T-cell responses need to be excluded from trials that use alternative Ad-vectored vaccine candidates, at least not until further studies suggest any deleterious effects on vaccine efficacy. But once a trial that uses alternative Ad-vectors shows efficacy, these effects are something to keep in mind when interpreting the results, she says. “We will definitely have to look if these immune responses at baseline have anything to do with efficacy.”

“While Frahm raises concerns for cross reactivity between pre-existing T-cell responses to Ad5 and rarer serotype Ads, the impact of this observation needs to be directly tested in humans vaccinated with vectors such as Ad26,” Michael says. “We are moving forward with testing Ad26 with MVA.”

Barouch says he has data that show that most people have Ad26 and Ad35 T-cell responses (3), consistent with Frahm’s observation that Ad-specific T-cell responses are cross-reactive, and is currently investigating whether these Ad-specific cellular responses affect the immunogenicity of alternative Ad vectors in humans.

Meanwhile, Gary Nabel at the National Institute of Allergy and Infectious Diseases’ Vaccine Research Center is collaborating with the biopharmaceutical company Okairos to explore the use of chimpanzee adenoviruses as vaccine vectors. Nabel says that in humans, seropositivity against chimp adenoviruses is much lower than against human Ad5 and Ad26, and generally also lower than seropositivity against human Ad35. In addition, chimp Ads seem to have a similar ability to Ad5 to stimulate immunity, he says.

1. Nature 482, 89, 2012
2. J. Clin. Invest. 122, 359, 2012
3. Nat. Med. 15, 873, 2009