One-Two Combination

Heterologous prime-boost is currently the regimen of choice for many AIDS vaccine approaches, even though exactly how it works is still far from clear

By Andreas von Bubnoff

We all use things every day even though we have little appreciation of how they work. Cars, cell phones, and computers come to mind. Still, we know that someone somewhere knows how they work so we can ask them how to fix them when they don't work as they should.

Yet, that's not always the case with AIDS vaccine candidates that are in development. One vaccination regimen, for example, dubbed prime-boost, is currently used in many AIDS vaccine trials. But surprisingly little is known about precisely how it works.

Intuitively, the rationale is simple: The first administration—the prime—generates a pool of memory T or B lymphocytes that allows for a quicker and stronger immune response to the second administration—the boost. And it works. "Essentially, most vaccination strategies are primes and boosts," says Larry Corey who leads the HIV Vaccine Trials Network (HVTN) in Seattle. Giving a second dose of the same vaccine is common practice for many vaccines, such as those against chickenpox and measles.

Now heterologous prime-boost—using different vaccines for the prime and boost—is the regimen of choice for many AIDS vaccine candidates that are in clinical trials. The hope is that the regimen will lead to increased immunogenicity and could also result in a broader immune response because each vaccine component might stimulate a different type of immune response. In addition, using a different vector for the boost also ensures that there is no antivector immunity generated by the priming immunization that interferes with the boost.

Indeed, Hildegund Ertl of the Wistar Institute in Philadelphia thinks a successful AIDS vaccine is likely going to be a heterologous prime-boost regimen. "That's the approach where I'd put my money right now," she says. Tomas Hanke of the University of Oxford even goes as far as to suggest that, for now, it's better to try combinations of the existing vaccines and vectors in clinical trials than develop new ones. "I don't think we need more vaccines, more vectors," he says.

Typically, the regimens combine DNA and replication-deficient or attenuated viral vectors such as adenovirus (Ad) or modified vaccinia Ankara (MVA) to deliver HIV proteins to antigen-presenting cells. Indeed, many different combinations have been tried and more are planned, and about half of the 30 or so ongoing AIDS vaccine trials use such approaches. Even so, there is surprisingly little data that resolves the immunological mechanism behind the relative robustness of prime-boost, and just why some combinations work better than others remains mysterious.

Trial and error

That's why finding the right combination often boils down to trial and error. In 1998, for example, Hanke, then in Andrew McMichael's group at the University of Oxford, published the first study that used DNA as prime and MVA as boost for an AIDS vaccine candidate (Vaccine 16, 439, 1998). Hanke says they chose DNA and MVA because they were readily available for use in humans at the time. DNA is also attractive because it is itself inert and doesn't elicit an immune response to anything other than the HIV immunogens it encodes.

However, the choice of which to use as the prime and which for the boost was very much empirical, Hanke says. "We wanted to try their combinations, first without really thinking why we should use this one first as opposed to the other," he says. "We tried all combinations of DNA/DNA, MVA/MVA, DNA/MVA and MVA/DNA, and found that DNA/MVA was by far the best." The mouse study showed that the DNA/MVA combination resulted in the strongest CD8+ T-cell immunogenicity. The work was done in collaboration with Adrian Hill's lab, who showed around the same time that the approach also worked for an experimental malaria vaccine (Nature Medicine 4, 397, 1998).

Even today, deciding which vectors to use for prime and boost still involves a lot of "educated empiricism," Corey says. Peggy Johnston of the Division of AIDS at the National Institute of Allergy and Infectious Diseases (NIAID) prefers to call the approach "thoughtful empiricism." "Try it and see what happens, but there is some thought behind it," she says. Empiricism sometimes does have a particular advantage: the outcome can come as a surprise. In primates, for example, two groups have reported that Ad serotype 35 (Ad35) followed by Ad5 gives a better overall response than vice versa, Corey says.

Safety is an overriding factor when it comes to choosing vectors and is one reason that many trials use MVA, which is a modified version of the smallpox vaccine that has been used for hundreds of years. Similarly, one reason that Ad vectors are used is that their safety profile is well known, Johnston says.

Lost in translation

Demonstrating that a prime-boost regimen works in animal models doesn't mean that it's going to work in humans. Hanke and McMichael's DNA/MVA regimen, for example, also worked well in monkeys, but not so well in humans; a Phase I/IIa trial showed that fewer than 15% of the vaccinees had an immune response detectable by an interferon (IFN)g ELISPOT assay. A smaller trial later showed a response in half of the vaccinees, but, according to McMichael, it was a weak response compared with other vectors such as Ad.

Hanke thinks that one reason for that was the limited DNA delivery attainable in humans; it's simply not possible to get the same high doses of DNA per weight into humans as into animals. To give humans an equivalent dose to that used in animal experiments, one would have to use something like 70 mg. "We use a gene gun for mice that covers half their body. You don't use a cannon to go into humans," Ertl says. "Imagine injecting a pint of material," adds McMichael. Both Hanke and McMichael have now turned away from using DNA as a prime.

Other researchers still see value in DNA. David Ho of the Aaron Diamond AIDS Research Center in New York is using a DNA/MVA prime-boost regimen with five HIV clade C genes. In mice and rabbits the combination works about 10 times better than each component by itself, and Phase I trials have shown that each alone is safe and immunogenic. Ho will check the DNA/MVA combination in Phase II trials next and anticipates good results because of technical improvements; for example, he is adapting the DNA vaccine's GC content in the HIV gene inserts so that the mammalian host can better express them.

Ho is also working with a company called Ichor to develop electroporation as an improved DNA delivery method, which uses electric current to temporarily poke holes into cell membranes that enables the DNA to pass through. Data in mice and rabbits suggest that electroporation improves humoral immune responses 100-fold and cellular responses 10-fold, and there is no drop in delivery efficiency between rodents and monkeys, suggesting that delivery in humans will be efficient as well. Ho plans to test electroporation in a Phase I safety trial later this year.

To circumvent the problem of DNA delivery, other groups are working on heterologous prime-boost regimens that combine different viral vectors. Dan Barouch at Harvard University is testing different Ad serotypes in rhesus macaques and says that a heterologous Ad26/Ad5 approach proved 8-fold more immunogenic than Ad5/Ad5, presumably at least partly because the heterologous boost avoids anti-vector immunity generated by the priming immunization.

One potential issue with using Ad5 as a vector is that a majority of some developing-country populations, particularly in sub-Saharan Africa, have preexisting immunity to the wild type virus that can dampen the vaccine-induced HIV-specific immune response.

To circumvent this problem, Barouch has made a chimeric Ad5 vector, Ad5HVR48, that has the surface loops of its major capsid protein replaced with ones from a less-prevalent serotype, Ad48. These loops are dominant targets of Ad5-specific neutralizing antibodies. Barouch says one particularly potent combination in animal experiments is Ad26 prime followed by a boost with either the regular Ad5 or with Ad5HVR48. He plans to bring Ad26 and Ad5HVR48 into Phase I clinical trials to test their safety and immunogenicity.

Early results in humans

Most of the evidence that heterologous prime-boost works better than homologous comes from animal experiments, but some regimens have already proven safe in humans and appear to show some immunogenicity in Phase I trials. Giuseppe Pantaleo of the University Hospital in Lausanne, Switzerland, is one of the coordinators of a European Phase I trial that uses a DNA/NYVAC poxvirus combination that resulted in cellular immunity in 90% of vaccinees, whereas NYVAC alone induced responses in only 40% of vaccinees. What's more, DNA/NYVAC showed a 5-fold greater and more durable response than NYVAC/NYVAC. A Phase II trial has already started enrollment.

Another regimen that appears to work in humans uses DNA as a prime and Ad5 as a boost. In monkeys, the DNA/Ad5 candidates developed by NIAID's Vaccine Research Center (VRC) are more immunogenic than Ad5/Ad5 or either vector alone. According to Corey, early returns in Phase II studies with the VRC DNA/Ad5 vaccine suggest that more than 70% of the recipients develop HIV-specific T-cell responses. The regimen will be tested soon in a Phase IIb efficacy trial called PAVE 100.

Corey says that additional human trial results with heterologous prime-boost regimens will be presented at the upcoming AIDS Vaccine 2007 conference in Seattle in August; for example, a small Phase I trial has indicated that a combination of MVA and fowlpox vectors results in better immunogenicity than repeated doses of MVA.

These preliminary human results look encouraging. But not everyone has observed that heterologous prime-boost regimens work better in humans than using the same vaccine repeatedly. "Heterologous prime-boost has not worked in people in our hands in fairly large Phase I studies, " says John Shiver of Merck Research Laboratories. "We haven't found anything that shows that prime-boost adds a synergistic effect in people, and we have tested probably more things than anybody else. I know of some that work in monkeys but not any that work in people. " For example, he says, Ad5 prime and canarypox boost does not work better than Ad5/Ad5. What's more, Merck did not see any difference between DNA/Ad5 and Ad5/Ad5 regimens in humans. This is why the company is currently conducting Phase IIb trials that use repeated injections of Ad5. "The immune responses we get with the adeno vector [alone] are actually pretty good, " Shiver says.


Given that, in many cases, heterologous prime-boost appears to induce stronger immune responses, the question is how. The approach makes sense because it circumvents antivector immunity, but there also appears to be a degree of synergy because heterologous vectors induce more robust responses, and this may include a distinct type of response. But there is a dearth of experimental data to demonstrate how this happens. "I don't think people know," Johnston says. "It could be that each vector will potentially enter different cell types."

"Exactly why it's better, I don't think anybody knows," says Rockefeller University's Sarah Schlesinger who collaborates with Ho on the DNA/MVA trials. She thinks that one problem is that it's difficult to directly measure priming.

As for why the DNA/NYVAC regimen works better than NYVAC alone, Pantaleo thinks that what DNA does from an immunological standpoint is very different from what poxvirus does. "DNA goes into different cells and the way it is presented is different," he says, adding that very little is known about the mechanism. He thinks that one problem is that it's hard to study in vitro what happens in humans in vivo. He has his own hunch as to why priming with DNA results in a longer lasting and more powerful T-cell response, and he suggests there could be large deposits of DNA in muscle cells at the site of injection. "But I don't have experimental evidence for that," Pantaleo says.

There is, however, some evidence that using heterologous prime-boost results in a more diversified immune response than using the same vaccine repeatedly. Several groups have observed that different vaccines induce a different type of cellular response when used as a prime or boost; both Hanke and Pantaleo say that DNA appears to cause a better CD4+ T-cell response than viral vectors.

A mouse study comparing a DNA/Ad5 regimen with either DNA or Ad5 alone showed that DNA induced more of a balanced CD4+ and CD8+ T-cell response (J. Virol. 79, 8024, 2005), whereas Ad5 induced more of a CD8-biased response, according to Gary Nabel, the director of the VRC who led the study. What's more, DNA/Ad5 induced a broader CD4+ T-cell response and a stronger CD8+ T-cell response than either DNA or Ad alone. "It's a quantitative increase in the CD8s, it's both a quantitative and a qualitative increase in the CD4s," Nabel says, adding that there aren't many analyses that do this kind of comparison. "I can't say I know of any other papers that do this kind of [analysis]," he says. "We don't know a lot about the mechanism of how DNA is immunogenic."

Together, these observations may help explain why some but not all heterologous prime-boost combinations work. Using DNA as prime followed by Ad boost works very well, says Rick Koup of the VRC. "But if you do it the other way around, with Ad first and then with DNA, it doesn't seem to work." Similarly, Merck's Shiver says that in monkeys, Ad prime followed by poxvirus boost works well, but not the other way around. "We don't know [why]," he says. Koup thinks that in most combinations that work well, the prime generates more of a CD4+ response and the boost more of a CD8+ response. "It may be that having a good CD4+ T-helper cell response really helps the boost," Koup says. Still, he adds, "that's a theory."

But just why it is that some vaccines target more of a CD4 response, and others more of a CD8 response, one can only speculate, Koup says. He believes the difference may have to do with how different vectors affect the antigen-presenting cells, for example the specific profile of chemokines or cytokines that they induce the antigen-presenting cells to secrete.

Apples and oranges

One obstacle to understanding which regimens work better than others is that it's often difficult to compare the results of different studies since they use slightly different vectors and HIV gene inserts. "I think there is a false assumption that a DNA is a DNA and an MVA is an MVA," Johnston says. "That's just not true."

This could in part account for different results of studies that use similar prime-boost regimens, such as DNA/Ad5 or DNA/MVA. Hanke says the use of different HIV gene inserts could explain why his DNA/MVA regimen showed immunogenicity in fewer people than the similar DNA/NYVAC trial. DNA/NYVAC uses HIV genes including env, whereas the DNA/MVA regimen used only gag. Hanke thinks that immunogenicity to Env, however, may not be very helpful. "Some data suggest that immune responses to Env do not correlate with better control to the virus," Hanke says (Nature Medicine 13, 46, 2007). What's more, he suggests the Env-specific responses are immunodominant and could suppress other relevant responses, for example against Gag. Different inserts can also differ in expression levels of HIV immunogens and in their ability to cross prime—that is, to make non-dendritic cells express antigen that is then taken up and presented by dendritic cells.

Such differences are the reason why Nabel and others have initiated studies looking at a standardized insert—such as the env gene from HIV clade A—as an immunogen in the context of different vectors. "We are trying to organize efforts to look at a common insert," Nabel says; a Phase I trial (HVTN 072) that just began uses this insert in DNA, Ad5, and Ad35 in various combinations. "I think that's going to be a very worthwhile set of studies," he says. Barouch's group will also use this insert in upcoming clinical trials.

Meanwhile, Pantaleo says a lack of understanding as to why some prime-boost regimens work better than others shouldn't delay clinical trials. "If you are trying to develop a vaccine, you need to go fast," he says.

Still, none of this may be relevant if cellular immunity shows no degree of protection in the first place. Merck's Ad5/Ad5 trial is widely anticipated to demonstrate whether it does or not. "If it doesn't work, there is no prime-boost or other modality of making a T-cell response that's going to work," says Shiver.

Even if a heterologous prime-boost regimen works one day, the cost and logistics of administering it will likely be higher than using just a single vaccine component. There is currently no licensed vaccine that's a heterologous prime-boost regimen, Schlesinger says. "Ideally you would have a single product," she says. "The only reason we are doing [heterologous] prime-boost is that we don't."