We think it is going to take a significant amount of innovation and risk to address the obstacles of the HIV vaccine.
(Click on image for a larger version)
Since the early days of the epidemic, we have now over 60 million HIV infections, a significant number of deaths as a result of AIDS and a significant number of orphans – and the epidemic is still in its early stages. As we all know, the emphasis and the scourge of the epidemic is in the developing world and, although there is a series of antivirals available, when the HIV gets in, it stays in. There hasn’t been a single case of the cure of HIV.
(Click on image for a larger version)
I want to address three issues. First I will look at the obstacles in the path of an HIV vaccine. Then I will give a brief update on the current state of the field. And I will finish with some of the opportunities, some of the ideas outside of the box, and the way that we are looking at the HIV vaccine field now.
(Click on image for a larger version)
The initial obstacle, really, is the extensive amount of variability of HIV. As shown in the slide, if you look at the extent of the global influenza epidemic, you see that the amount of variability in the individual is comparable to that in a single HIV infected individual. If we pan out on the dendrogram and look at the cohort – as an example, the Amsterdam cohort – we see it is expansive. And then if we look at a single country, again there is an incredible amount of diversity. This is because of the error nature of the RT (reverse transcriptase) enzyme of the virus, the rapid rate of the replication of the virus, and the ability of combination and recombination of the virus.
(Click on image for a larger version)
If we look at the epidemic at the present stage, we see a variety of subtypes of HIV. There is a variety of sub-subtypes of HIV. And then there are the circulating recombinant forms. This is a changing, moving target and it is a challenging target.
(Click on image for a larger version)
However, in the last couple of years we have learned something that is extremely interesting. That is, it appears that at the point of the transfer of the virus from the ‘donor’ there is a bottleneck. That is, although the individual has a range of HIV isolates, it is a single isolate in the majority of times that is going to be the infecting virus and is the ‘founder’ virus. So we may have a window of opportunity there.
(Click on image for a larger version)
In some of the recent data out of Shaw’s lab, in a series of analyses of about 100 individuals with HIV, we have seen now in about 75 per cent of those individuals the evidence of this single virus, of the founder effect, if you will. And then in a comprehensive analysis of the sequencing and a phenotypic analysis of the envelopes of those viruses, there are not a lot of surprises.
At the bottom line on the neutralising antibody, on the issue of the ability of the founder viruses to be easily neutralised or moderately neutralised, is that there isn’t any difference, really, between what are the founder viruses and what is seen in some of the other HIVs.
(Click on image for a larger version)
That leads us into our second challenge – we don’t know how to make neutralising antibody against this virus. The reason that we don’t is shown in this slide. You see here a computer model of the outer protein of HIV, which is covered with a mask of carbohydrate in the little spokes that are coming out. It has a series of variable loops that are dominant loops and serve as decoys to the types of antibody that we are looking for. Of the four or five antibodies that have been identified as the broadly neutralising antibodies, these are against a series of the regions of the virus that are not easily accessible. We’ll talk about that a little later.
(Click on image for a larger version)
Another challenge is the brief window of opportunity. As shown in a series of elegant studies out of the lab of Ashley Haase and a variety of collaborators, HIV can infect as an infected cell or as a free HIV, as a particle, and within a week or so this virus has taken hold in the body. And it is in that initial week that we see the window of opportunity.
If you are HIV infected through any route, within a week this virus amplifies in the gut associated lymphoid tissue (GALT). So the action really is occurring in the GALT, and we believe that an effective HIV vaccine is going to have to attack in that early stage, between the time we have the founder virus and the seeding of the persistent HIV infection. And that is really asking a lot of a vaccine.
(Click on image for a larger version)
Which HIV Antigens to Include in the Vaccine to Control HIV (Cell Mediated Immunity)]
Another challenge is that it was assumed it would be easier to control HIV than it would be to solve the neutralising antibody problem. In reality it may not be easier, and in fact at this point in time we don’t really understand what antigens we have to have in an HIV vaccine to ensure that we can elicit a control of the virus – that is, to be able to maintain a low amount of the viral load.
So we just have started an immunogen design program based on information that is coming out of the series of the cohorts of the elite controllers, a subset of individuals that in the absence of the antiviral therapy have controlled HIV to a lower or undetectable level.
(Click on image for a larger version)
This example of data from the Walker laboratory was published about a year ago. It shows that as you increase the amount of the CTL (cytotoxic T-lymphocyte) epitopes against one of the internal antigens known as gag, there is a lowering or a suppression of the viral load. And the inverse is seen with the increased amounts of CTL epitopes against the envelope antigen.
So it is really beginning to now open up a series of questions as to what are the antigens that we have to have in the vaccine on the control of HIV. I will come back to that in the latter part of the talk.
(Click on image for a larger version)
Lastly, there is not an ideal animal model. There is a very good animal model, the SIV or the simian virus – although SIV is not HIV and the macaque of course are not people. And particularly on the antigen design issue it becomes a challenge as to how best to identify the antigens that are needed on the actual control of HIV in the animal model. So we use the animal model for a lot of things but it is unlikely for the antigen design issue.
Similarly, in our phase I trials, we have done 15 phase I trials of a series of the candidates, and we can line up the candidates based on the ELISPOT assays and flow-based assays and such. However, there isn’t a correlate of the protection at the present time, and, as I will show in the Merck data, the ELISPOT assay is not a correlate. So we have just recently introduced a concept that was published about a year ago, of doing a series of small screening test-of concept-trials, to get a preliminary assessment of the potential efficacy of a candidate, and to gain guidance on the basis of that for decisions as to what should advance into efficacy trials.
(Click on image for a larger version)
I want to move on to where we are in the state of the field.
(Click on image for a larger version)
There have been only a couple of candidates that have gone as far as the efficacy trials. Each of these candidates has failed. One of the antibody vaccines – with the monomeric outer protein of HIV – had no effect on the acquisition of HIV and had no effect on the viral load. Then we had the recent data on the first of the CMI (cell-mediated immunity)-based vaccines, and I will show that data.
We have a third vaccine that is in the efficacy trials now and is eliciting a combination of CD4 help and the neutralising antibody against the laboratory strains of HIV. We expect to have the data in about a year; I wouldn’t bet the farm on that vaccine either. And at the bottom line we don’t have in the pipeline at the present time any candidate that has effective neutralising antibody. We don’t have any candidate at the present time that is really targeting the GALT, the gut associated lymphoid tissue; and in the monkey model we don’t have any candidates that are as good as a live attenuated SIV vaccine.
(Click on image for a larger version)
I want to talk about the adenovector, just for a minute. This is the Merck Ad5 vector. It had three of the internal antigens of HIV, the gag, pol and nef. It was just run in a clinical trial in which about half of the individuals had low or no pre-existing Ad5 antibody and the other half had pre-existing antibody.
In the monkeys it was extremely effective in suppression of viral load if one had a SHIV challenge, an SHIV virus, as a challenge. It didn’t have any effect at all if the challenge was the SIV virus. We’ve learned now and are shifting emphasis towards the SIV model a lot more than the old SHIVs, in which almost every vaccine has suppressed viral load.
And in the Phase I data, about 60–70 per cent of individuals on the ELISPOT showed up as positive. There was a blunting of this effect with individuals that had pre-existing Ad5 immunity.
(Click on image for a larger version)
So at the bottom line the vaccine didn’t have any effect on viral load at all.
(Click on image for a larger version)
And when they had looked at the acquisition in individuals who had low or no Ad5 antibody, again they found there was not any effect. In fact, there was an equal amount of the infection in each of the groups.
(Click on image for a larger version)
In contrast, though – and this is the issue on the table at the moment – in the individuals that had Ad5, had pre-existing Ad5 immunity, the data has shown 29 infections in the vaccine group and 13 infections in the placebo. This has raised a lot of questions and we don’t have any of the answers at the present time. Is it simply that it is weak as a vector, this vaccine is a weak vaccine? Are the wrong antigens in the vaccine? Is the CMI as a concept flawed for HIV? We don’t have any of those answers.
And on this acquisition issue, is it really the result of the Ad5 antibody? A large number of these individuals who had the highest titre of Ad5 have a variety of the characteristics that are different in the individuals, and in the low-Ad5 group there are a lot of cases of individuals that are uncircumcised. We just now are looking at the HSV data and the other STD data as the co-factors in these individuals, and so we don’t have the answers.
But it certainly has put a chill over the HIV vaccine field. There is a lot of discussion, there is a lot of ongoing work, on the potential mechanism as to what is going on. If Ad5 is the cause, is there an expansion of CD4 cells as a result of the pre-existing vector immunity? Has that increased the ability to infect? And a series of other hypotheses are on the table.
As you can imagine, this data is new. There is going to be an extensive amount of work on the concept and, ideally, understanding what is going on here.
(Click on image for a larger version)
In the immunogenicity of the vaccine there was no difference between individuals who became infected and individuals who were not infected. There is again a blunting of the response in those individuals that have the highest titres, over a couple of hundred, of the Ad5 pre-existing antibody.
(Click on image for a larger version)
To look now at the overall pipeline: we have a large number of DNA vectors and a large number of the viral vectors in the pipeline, all focused on the CMI concept. There is also a smaller number of the protein-based vaccines and of the peptide-based vaccines, but in the general context we are at a point where we are asking: if the Merck vaccine, which had been the most immunogenic of all the vaccines in the pipeline, has done as poorly as it has done, how does that play out for some of the other candidates in the field? If you pick up any recent issue of Science or Nature, you find another report on what to do next and why.
(Click on image for a larger version)
At the bottom line, we don’t have a candidate in the pipeline that elicits effective neutralising antibody, and if you look at the live attenuated SIV model that I will show you, you will see that there is not a candidate that is as good as what has been observed with live attenuated SIV.
(Click on image for a larger version)
I want to wrap up with where I think the field is going, with a couple of the assumptions that we have made at IAVI, and leave you with a few of those ideas.
(Click on image for a larger version)
Our initial assumption is that there is not going to be an AIDS vaccine if we don’t solve the neutralising antibody problem. As shown in this slide, we know a lot about the envelope proteins of HIV. A few of these broadly neutralising monoclonal antibodies have been identified; we know exactly where on the protein they are linking to and binding. We have made a number of immunogens as mimics of these binding sites; we have not succeeded at the present time in mimicking the breadth of the broadly neutralising antibodies.
(Click on image for a larger version)
In vitro what is shown here is we have about 100 viruses that we have looked at in a standard panel of HIV; we have looked at a breadth of monoclonal antibodies. The blue shows where there is not any neutralising antibody at all. What is shown here is the four of the broadly neutralising antibodies. The antibody in the third column from the right, the 4E10 antibody, has neutralised all of the viruses, although it is not a potent antibody.
(Click on image for a larger version)
And we know from a series of recent studies that we have done in the Neutralising Antibody Consortium that in vivo now, if in a passive study you infuse the antibodies into monkeys and give a reasonably high titre of these antibodies, either you are able to have a total blocking of HIV infection – in this case the model is a SHIV virus – or in a few of the animals you see a little blip of viral load and then it goes away.
(Click on image for a larger version)
So we have set up a concept known as the Neutralizing Antibody Consortium, which is basically linked to a series of labs all over the world, with the following idea. Since we have only four of the monoclonal antibodies, can we increase the number of these, can we get even better antibodies, can we identify where on HIV the antibodies are going to target, and can we initiate a high-throughput immunogen design and screening program?
(Click on image for a larger version)
We have done a lot of that. Shown here are a couple of examples. The 2G12 antibody, for example, is binding on the carbohydrate of the HIV outer protein, with a series of mannose receptors. We have mimicked those, put those on a Qβ carrier, and we now have those in rabbits. Similarly, the structure is known of the CD4 binding site antibody, and through a series of computational models we have put these into scaffolds, and these are in the animals.
Unfortunately, though, all of these attempts have thus far failed to elicit anything that is broadly neutralising.
(Click on image for a larger version)
So we get back to the issue of innovation and risk. What else are we able to do, outside of the standard retro-vaccinality, if you will, where you have the antibody and it binds onto the protein and we try to mimic it? I am going to show just a couple of examples here.
(Click on image for a larger version)
The first example goes back about 100 years. We ask: can we increase the potency of the antigen and the breadth of neutralising antibodies, if the immunogen now is given as an immunogen-antibody complex?
(Click on image for a larger version)
The answer is yes. In the pilot studies we just took a monomeric gp120, the outer protein of HIV, with a couple of antibodies. We have used one that is binding on the CD4 binding site, the other on the V3 loop. We immunised these into a bunch of rabbits. The ELISA titres are shown at the bottom of this slide, showing a significant increase in the titres.
(Click on image for a larger version)
To look at the neutralising antibodies (again this is just a pilot study): with a low dose of the antigen alone you don’t see anything. If you use the antigen with an adjuvant, Adjuplex in this case, you can have a couple of isolates neutralised. And yet the complexes are neutralising six out of 10 of the isolates. So our program is moving ahead, looking at the potential of antigen-antibody complexes as vaccine candidates.
(Click on image for a larger version)
Another concept – this is the work of Phil Johnson – is to say: if we can’t make an immunogen, is it possible to have gene delivery of the broadly neutralising antibodies? What Phil has done in a pilot study is to take the fB frame of one of the broadly neutralising monoclonal antibodies, put that into AAV (adeno-associated virus) as a vector, inject that into the muscle of the animal and ask: can you elicit a broadly neutralising antibody?
(Click on image for a larger version)
And the answer is yes. His pilot studies in mice are shown in this slide, with one of the four broadly neutralising antibodies against HIV. And you see if you use a large, large dose of AAV as a vector single shot, you get a reasonable titre of the broadly neutralising antibody into the mice.
(Click on image for a larger version)
He has now gone ahead and taken a broadly neutralising SIV monoclonal, and he has put that into three different kinds of AAV vectors. I don’t have the time at this meeting to go into the details on that; however, the bottom line is that he has protected six out of nine of the monkeys, and in one series of the vectors had three out of three.
So if we had a candidate HIV vaccine that elicited high titres of broadly neutralising antibodies, we would have this in the clinic as fast as we could. What Phil has looked at is whether, in the absence of the antigen, of the immunogen, you can do this with a gene transfer approach. His pilot studies are pointing in the right direction, so we have a dialogue that is going on with the regulatory agencies now to see what it would take to look at the B12 antibody as a candidate in a phase I clinical trial.
(Click on image for a larger version)
I want to try and wrap up now on the second assumption that we have made. That is that even if we make neutralising antibody it is likely that we will also need to control HIV. We will not be able to have a block against every one of the viruses.
So here I show you data from a paper that I wrote a couple of years ago, in which we looked at hundreds of animals and we kept the bar at a 3 log suppression of viral load. It was black-and-white: live attenuated SIV is extremely effective; all of the other vectors – pick whatever vector you like, and this is true for the protein-based vaccines, the peptide-based vaccines, the VLPs and the killed vaccines – are not as good as a live attenuated SIV in this system.
(Click on image for a larger version)
The type of data that we see on a routine basis is shown in a recent study here that we have done as a collaboration of a consortium of labs in the United States and Amsterdam working on this problem. This study in particular is the Watkins study, and he is showing here that a live attenuated 239 virus (in green) gives an infection of the monkeys. Then we come back in and we give a challenge, and the animals that have not received the vaccine consistently have seven or eight logs of virus, and the animals that received the vaccine have a five or six logs reduction of the viral load.
(Click on image for a larger version)
Suppose we expand now and we ask the question: in real life, because of the variability of the virus, one is not going to see a virus that is the exact same match as HIV. In the SIV model the only other challenge that we have is the E660. If we look at the CTL epitopes between E660 and the 239 virus, we see that about half of the epitopes are not seen in E660 that are seen in 239.
(Click on image for a larger version)
So now if we come back and we give a challenge with E660 instead of 239 as the challenge, we see a completely different picture. Only four out of 10 of the animals have the ability of control of the SIV infection. The others have basically overcome an initial blunting of viral load, and that is likely as a result of the escape around CTL.
(Click on image for a larger version)
Continuing on the live attenuated model, we think it is important to understand why it is that an adenovector can only give a log of suppression of viral load when a live attenuated SIV can give six logs of suppression. Similarly, with a single cycle, if you cripple SIV, you get only a log or a log and a half of suppression of viral load. So it isn’t simply that we have the wrong antigens in the vaccine.
(Click on image for a larger version)
So, IAVI is moving in the direction of replicating vectors. In blue at the bottom of this slide is a series of the candidates that have been evaluated in the animal models but have failed to give a significant suppression of viral load in the SIV model. We know that a number of our best vaccines in the paediatric world are live attenuated vaccines. We don’t think there is going to be a live attenuated HIV vaccine, because of the ability of this virus to potentially have the mutations return to a wild type.
(Click on image for a larger version)
So we are looking at a series of vectors. We will have data on CMV (cytomegalovirus) at the end of the year; there has just recently been a rescue of the potential of Rio virus as a vector in the targeting into the GALT; we have a couple of the chimeras and the paramyxoviruses; and a Sendai vector is coming along in the pipeline.
(Click on image for a larger version)
I want to close by saying that HIV really offers incredible challenges. A series of the preliminary proof of concept studies, with the broadly neutralising antibodies and live attenuated SIV, we think are pointing us in the right direction. And we think it is going to take innovation and risk to achieve success.
(Click on image for a larger version)
I want to thank all of the players who were involved in the development of the data.
(Click on image for a larger version)
And I want to thank our donors too.
Discussion
Question: Thanks very much for providing such an insight into the difficulty and the problems. You mentioned that broadly neutralising antibodies to HIV do exist. But do you know what proportion of HIV-infected subjects are capable of developing broadly neutralising antibodies?
Wayne Koff: We are looking at that now. We have just screened about 1800 people in the developing world. Of those, we see that on average about 120 have what is, on the basis of our platform, broadly neutralising antibodies. We have prioritised those individuals now on the basis of the breadth and of the potency, and have gone to about 15 of those individuals, isolated out the PVMCs (proliferating vascular mural cells) from those individuals, and are looking for a next round of broadly neutralising antibodies. We have a couple of novel monoclonal antibodies that have come out of that program. But it is probably of the order of between one in 10 and one in 20 where we see that.
We are just beginning to look at some of the HLA considerations. What is it that is different about these individuals? Is it the virus that is different in these individuals than some of the others? It is a comprehensive study.
Question: Thank you for a beautiful summary of a complex area. On the question of the, I think, quite clear-cut results of a bad effect of pre-existing Ad5 antibodies in the vaccine, we have two prior examples of antibody-dependent enhancement – namely, the failed Wyeth respiratory syncytial virus vaccine and the much worse example of dengue haemorrhagic fever shock syndrome.
Would you accept the possibility that what is happening here is that the Ad5-positive people see the vaccine coming, develop immune complexes – admittedly, how many of those immune complexes work their way into the vagina is doubtful – and there stir up things? If there is anything an AIDS virus loves, it is damaged epithelium, it is some kind of inflammatory site, hence of course the well-known effect of intercurrent sexually transmitted diseases heightening the risk of catching it. Do you think that is a viable hypothesis?
Wayne Koff: Well, it certainly is one that is on the list. If you tweak the system in vitro – you can show in a series of macrophage cultures, for example, that it has a potential for the antibody immunogen enhancement – there are a lot of people who think that the action is in the GALT, where this virus is really amplifying. If you get infected by any route, this virus almost immediately is going into the GALT. It is seeding, it is getting a small type of a founder effect, and then it’s blossoming out.
There is a lot that we don’t understand about the GALT, there is a lot that we don’t understand about CMI in the mucosal areas, and the data is new. So I think we will have to wait a couple of years to see how some of the testing occurs there.
Question: Wayne, a great talk. One of the paradoxes is that as you move towards ever more immunogenic live viral vectors to try and deliver an immune response, the immune response that you get is directed more and more against the vector that you are trying to use as the carrier. I guess, particularly when you come back for a second or a third time to try and boost immunity, this is going to be a significant problem.
Is perhaps the strategy not rather that we should be moving to less immunogenic viral vectors? Certainly we want effective vectors to deliver the vaccine payload, but if we move to more immunogenic ones we may get into a problem with limited numbers of vectors, first of all, for all the different diseases, and then secondly the problem that we might not be able to boost.
Wayne Koff: That is a point that has come up, as you can imagine, in the discussions. At the bottom line, we don’t know, and we are going to try and find out. We will have data on the CMV vector by the end of the year in the SIV model, and that has all of the SIV antigens in that vector. So if a replicating vector is not going to give any benefit – and we don’t know the data, because it could hit a home run too – then we are going to really have to rethink what to do.
On the question of running out of vectors and stuff, people have talked about the potential of the heterologous and the prime-boost vectors, and in some of the monkey studies the data is better if you use a combination of the vectors, because of this pre-existence issue.
Of course, in a practical sense – and you know well the practical issues of vaccine development – we would be a lot better off if we could just get away with a single vector. So that’s really where we are in the field right now.Visit these websites
The Sir Mark Oliphant Conference Series is supported by the Australian Government under the International Science Linkages Program.
© 2008 Sir Mark Oliphant Conferences.
Website design acidgreen
