On July 17, a high-ranking official at the U.S. National Institutes of Health (NIH) pulled the plug on a hotly anticipated clinical trial for a government-funded vaccine to combat human immunodeficiency virus (HIV), the bug that causes full-blown AIDS. The announcement by Anthony Fauci, director of the NIH's National Institute of Allergy and Infectious Diseases (NIAID), was the latest in a series of setbacks in the search for a vaccine the world has been anxiously awaiting for more than two decades.

Fauci's reason for canceling the trial: There is not enough evidence that it's effective to justify a wide-reaching trial.

Vaccines come in two forms: protective and therapeutic. A protective vaccine typically consists of a weakened form of a virus that, when injected, alerts the immune system (antigens or proteins on the virus' surface), which in turn generates antibodies (other proteins) that clobber the invaders and remain on high alert should the bug ever try to attack again. If the virus hits again, the immune system will be armed and ready, with the weapons it needs to launch a preemptive strike to wipe it out before it can infect any cells.

Therapeutic vaccines are designed to battle illnesses already in the body by helping disease-killing T cells recognize and target them. In the case of HIV, candidate vaccines typically consist of parts of the virus that furtively slipped into the body and is successfully eluding the immune system. Whereas the virus entered the body undetected, the protein does not: The T cells will view it as a dangerous invader and attack it everyplace in the body, including on the viruses, reducing and perhaps even clearing the infection completely.

Fauci said that he nixed the trial, in part, because of the failure of Merck & Company's 3,000-person STEP vaccine trials in September. In lab tests, the Merck vaccine showed that immune system cells produced signaling proteins called cytokines when they came in contact with the vaccine. Researchers believed these so-called "correlates of immunity"—essentially signs that immune system was responding to the vaccine— indicated that it would fight infection.

They were wrong: The Merck vaccine proved ineffective at preventing infection or reducing levels of the virus in an infected person's body. The government-funded vaccine, known as PAVE (Partnership for AIDS Vaccine Evaluation), is formulated similarly to Merck's vaccine, consisting of three genes found in HIV attached to a weakened form of the common cold designed to draw the attention of the immune system. Originally set to be tested on 8,500 people, PAVE's trial was downsized to 2,400 soon after the failure of the Merck vaccine.

After a NIAID-sponsored summit in March revealed that there were too many unanswered questions, Fauci decided that priorities needed to be revised.

Toward that end, has co-authored an article in Science last week that calls for more basic research and smaller studies (to prove a candidate vaccine's effectiveness) before conducting large-scale human trials. Fauci spoke with ScientificAmerican.com to discuss this new tack—and whether there's still a chance of an HIV vaccine.


There seem to be a lot of moving goal posts with respect to the HIV vaccine. It was first promised in the late '80s. Then Bill Clinton set a goal of developing one by 2008. Here we are in 2008, and the posts are being pulled down altogether and the mission is being redefined. Is this just evidence of how complex an adversary we're dealing with in HIV?
I think all of those predictions of time frames really reflect an innocent but unintentional lack of appreciation for how unique and different HIV is. Margaret Heckler [former congresswoman (R–Mass.)], who way back when the virus was discovered, in a press conference, said we should be well into a vaccine trial within two years and maybe have a vaccine soon thereafter; and President Clinton saying the goal would be to have a vaccine in 10 years—all this was really predicated on how we look at vaccines in the standard, classical way.

But, HIV can't be tamed by classical methods?
This is the critical issue. If people understand this, they'll really get it. That is: How extraordinarily unique and different HIV is.

When you're setting out to develop a vaccine, you generally use natural infection as your model, or, put a different way, as your experiment of nature. Regardless of what microbe or virus you're dealing with—even the deadly ones like smallpox and the maiming ones like measles and polio—the majority of people, and sometimes the overwhelmingly vast majority of people, may get sick and then ultimately get better. They don't die and they don't have any lasting remnants of the infection. What happens is that the body adequately and appropriately ultimately eliminates the virus and provides you with lasting protection against subsequent challenges [from that virus].

So, if you're going to go and try to develop a vaccine for microbe X, you only need to…look at what the body's natural response is and vaccinate the person with either a modified form of the virus or a subcomponent of the virus that will elicit that same antibody response or that same cell-mediated immune response, which you are certain is associated with protection because you have many many examples of the natural infection to show that. So, you already have your road map of what you need to get to get a vaccine.

But, the body's natural response is different with HIV?
Astoundingly, of the tens of millions of people who have been infected, there's not a single documented case of someone who has established infection and ultimately eradicates the virus from the body. There are a number of people who are what we call "long-term nonprogressors"—for one reason or another, perhaps their genetic makeup, they seem to handle the virus reasonably well for a long period of time. For the overwhelming majority of people, the virus ultimately overcomes their immune system's attempts to curtail the virus. So, we are dealing with a situation where we don't even know if the body is capable of eliciting a protective immune response. And if it can, we know it's very difficult, because when you look at infected people, it is so unusual to see people with very good, broadly reacting neutralizing antibodies. So, when you go after developing a vaccine for HIV, you're in an entirely different ballpark than you are when you're trying to develop a vaccine for influenza or smallpox or polio or measles.

We as a field didn't fully appreciate that early on, as a matter of fact. That's the reason why it would be an understandable statement, though now retrospectively clearly premature, to say, "Okay, we have the virus in our hand, we are growing it, we know what it is. It should be a snap to develop a vaccine." Now we know that this just is not the case, so our challenge for the future is to do much better than what natural infection does because natural infection clearly is not successful in inducing an ultimately protective response.

So, where do you go without relying on natural infection?
What we have to do is develop immunogens and find out the fundamental, basic questions: Why is it so difficult for the body to develop a neutralizing antibody response? And given that, what can we do to present an immunogen to the body to allow it to develop a neutralizing antibody response readily—not in a very difficult situation, but readily in the vast majority, if not all, of the people that get vaccinated? Those are fundamental questions, the answers to which we don't have.

Do you think we got this far without very basic questions being answered because the HIV vaccine is such a desperately sought after goal—that the world is truly hungry for one to exist?

Obviously the world is very hungry, as you put it, for an HIV vaccine. But following the classical vaccine paradigm, what we were doing was not unusual. It was only when we had the full realization that, "Wait a minute. There's something very very different about this virus, so therefore the classic paradigms might not hold so strictly." So, I don't think the only motivation to move ahead was that we were "hungry for a vaccine." Our motivation was that that approach had been imminently successful with other viruses.

But, these moments of realization, have they come in nuggets over the past 20 years or are they big signs like the failure of Merck's STEP vaccine last year?
There were suggestions and hints of it very early on when we found out that antibodies that were elicited in people generally neutralized lab-adapted strains [of HIV], but did not neutralize the wild type strains from the body. This indicated that antibodies that were elicited weren't particularly effective against the kinds of viruses that were circulating in the community. That was a hint, but most of the people felt, "Well, let's just give it an empiric shot and vaccinate people with the envelope [the antigen, or protein on HIV's surface] and see if we can elicit a really good neutralizing antibody response.

It was only as the science moved along in parallel with those empiric attempts that we realized that the part of the envelope that neutralizing antibodies bind to is very cryptic in its conformation and doesn't reveal itself to the immune system very well or for a significantly long period of time. It's hidden. It has a confirmational component to it that doesn't allow the immune system to see it very well, much less elicit an antibody. We didn't really know that until we'd already launched some trials way back years ago using the empiric approach of what we would do with any other virus: get the outer coating, that is, the envelope, and use it as an immunogen. We didn't appreciate that the part of that immunogen that you would need to show to the immune system didn't readily reveal itself.

So, how does the STEP trial factor in? Its failure seems to be one of the primary triggers for the HIV vaccine reassessment.
The STEP trial was another interesting thing because as the antibody component failed, and we needed to go back to the basics to figure out how to elicit an antibody response, there were studies in an animal model that suggested that perhaps you couldn't block acquisition of infection. But, perhaps you could get the lowering of the viral set point so low that the person who was vaccinated would benefit because their disease would not progress. They would almost be like a long-term nonprogressor, and the other herd immunity-type benefit would be that if their viral load was low, then perhaps they would not readily transmit the virus to someone with whom they come into contact.

So, that's when the STEP trial approach of a T cell immune response was done. But, the correlates of immunity that were measured in the early part of the trial and in the animal were assumed to be correlates that would be therapeutic. As it turned out, when you vaccinated people in the STEP trial, it neither blocked acquisition—which we didn't expect it to do anyway—but it had virtually no effect on the viral set point. So, the question was, "Well maybe we can look at the immunological response and get some information from a subset of people in the trial who seem to have responded well?" That's a fallacious approach. If the vaccine doesn't work, doing immunological correlates, in my mind, and the minds of many many of my colleagues, is not worth doing.

That's really the fundamental reason why I elected to not accept the proposal for a moderately sized—or big-sized—PAVE 100 trial. I said, "I reject that." The reason I reject that is we really need to know that if the product is interesting and different enough from the STEP product. It's a DNA–DNA–DNA followed by an adeno and it also contains envelope, in addition to the other viral genes. I think it's worth pursuing, but not in such a large trial that it will allow you to get all the immunological correlates you want.

I first want to find out if it has a beneficial effect. If it does, then we'd be willing to invest more resources, more people, more specimens. If it doesn't succeed in lowering the viral set point, then I don't think it's worth going and pursuing these elusive immunological correlates. You don't even know what they are correlated with.

Since you're adding the envelope protein gene into the PAVE vaccine, is there a chance that vaccine could be both preventative and therapeutic?
It likely would be some antibody effect, but unlikely would be neutralizing antibodies, because the envelope is presented in that conformational form that doesn't reveal very well its component that would induce that neutralizing antibody. So, yes, there would be some antibody response—and hopefully it will add to the benefit. But, it is more likely that it would elicit a T cell response rather than an antibody response.

To your decision to reject this vaccine trial, there must have been some point where, in smaller trials, the PAVE vaccine proved to be safe and at least somewhat effective. It would have had to in order to get to a large trial, right?
No efficacy at all, because it wasn't done in a large enough trial or among at-risk people, who would allow enough infection to occur to see if it would work. All it showed was safety to some degree—I mean, we didn't do a lot patients—but safety in those we looked at. And the fact that it induces an immune response, which one would argue was somewhat better than what you saw in the STEP trial.

There was some argument about that at several meetings. "Somewhat better" is what? A lot better? I didn't think it was a lot better. That gets back to the point I made at the beginning of our conversation: So it elicits a similar immune response [to STEP]. Well, we know that immune response was not associated with protection [from the virus] in the STEP trial, so I'm not even sure that those immunological correlates are the relevant ones. So it really is sort of like we're swimming in the dark because we don't really know what the immunological correlates are. So, that's why I said I don't accept the original proposal for the PAVE trial. It's powered large enough to give you a whole bunch of immunological correlates, which, to me, are irrelevant if the vaccine doesn't work.

First show me a lean, mean trial with as little risk as possible to the people, as few samples as you possibly can to statistically give me a "yes" or "no" as to whether it lowers the viral load. If the answer is yes, then I definitely want to pursue it by getting many more people to see if we can get a good correlate.

So, that brings us to your piece in Science, outlining the way forward for vaccine research...
That was based fundamentally on the summit that we had in March of this year, in which we looked at the basic fundamental role of discovery versus development, animal models, clinical research. We addressed each of those. So, we outline some of the critical issues: Why don't we get a neutralizing antibody response? How will we be able to elicit it? What is a correlate of immunity? What about nonhuman primate models?

Nonhuman primate models can be very valuable, but right now we've got to link them closer to what goes on in the human. For example, the challenges in the nonhuman primate model are homologous challenges, which means you inject them with the same virus that you vaccinate them with. That's not the way it works in the real world. If you vaccinate somebody with a particular strain or a particular component of a virus, it is overwhelmingly likely that they are going to get exposed to something that's a little bit different. So, if you can't protect against a heterologous challenge, we've got to perfect the monkey model a little more, so that we use heterologous challenges.

Also, we need to totally reexamine what we're calling "potential correlates of immunity."

So, there are some basic definitions of success that need to be agreed upon before moving forward?
We need to have answers to some concepts first before we embark on large, empirically based clinical trials. Clinical trials, as we admit in Science, can be an important part of discovery research. You can get discovery doing clinical research. But, we are going to set a much higher bar for going into a very large empiric-type of efficacy trial. We need to have some really good reason to think that there's a reasonable chance that [a vaccine] will work.

But given the track record, you probably wouldn't want to predict a time frame for when we could realistically see a vaccine, right?
I really think it's folly to predict a time frame when you're dealing with discovery. You just never know when you're going to discover what it is that's going to really put you on the pathway. When you're dealing with development where you know exactly what you're aiming at, as we've done with other viruses, for which we have made successful vaccines, then it's pretty easy for me to give you a time frame.

When you envision a final product, is it good enough—given all the difficulties with HIV's envelope protein—to settle for a vaccine that encourages T cells to reduce a person's viral load?
I think the T cell vaccine could have something to do with blocking acquisition. Clearly the gold standard of blocking acquisition of infection is a good neutralizing antibody response. If we don't have a good neutralizing antibody response, we will not—I can predict for you—successfully get a vaccine that prevents initial infection.

So, T cell, yes—it might influence the viral load. But, I'm more interested in T cell responses to synergize with the neutralizing antibody response to help block initial infection. So, that neutralizing antibody gets rid of a lot of the virus as it comes into the body, but the cells that do escape and get infected, the T cell immune response could eliminate those cells. I think of the T cell response and the antibody response as being synergistic very early on—not just antibody blocks infection, and if that fails then T cells later on keep you in a less progressive state of disease. That could be, but none of us are giving up on the possibility that a vaccine might actually prevent initial infection.

Is there a component to this decision to redraw the way ahead in vaccine research that's driven by budgetary concerns—becausethe NIH budget is due to remain flat?
Of course. If you want to pursue new pathways and develop new programs, have new initiatives, and you have a very flat budget, it's much more difficult to maneuver that. You have to reprioritize and reassign resources from one project to another. It's much more difficult to do that—because we have a lot of worthy projects—if you have a flat budget. If you have an increase, you can be sure that we're going to selectively give that increase to vaccine discovery. But, in the face of a flat budget, it certainly constricts and constrains the flexibility that you have to do some of the things that we're talking about.

So, if a new administration came in and increased the NIH budget, the HIV vaccine program would be at the top of the list to receive new funds—even though global spending on finding a vaccine has quadrupled to near $800 million in the last decade?
The HIV vaccine is a very high priority for NIH, for NIAID, so whenever we get new resources—and hopefully that will be soon—there certainly would be a preferential targeting of those resources to vaccine discovery and development. That doesn't mean that anything that's new will automatically go to AIDS vaccine. We have a lot of other problems we need to deal with. There's malaria, there's tuberculosis, there's a whole bunch of things. But, given the seriousness of the situation, we will selectively favor that—not to the exclusion of other things, but we certainly would selectively favor that.

In terms of the priorities you outlined in the Science article, what's the top one on that list that you would want to achieve to moving the field forward?
The highest priority is a neutralizing antibody response. That brings in everything from getting the [virus's] crystallographic structure and the confirmation of the envelope to understanding how you could scaffold the epitope [antigen surface eliciting an immune response] of that particular binding site into something that's immunogenic. That, to me, is really the highest priority.

So, you want it all?