August 6, 2008
Listen to Audio (90 min.)
Please note: These files can be quite large. Allow some time for them to download.
Plenary moderator: Good morning, and first, congratulations to the award recipient, and it is my pleasure to introduce the first speaker for the plenary session this morning. Dr. Robert Siliciano is a member of the Howard Hughes Medical Institute and Professor of Medicine and Molecular Biology in Genetics at the Johns Hopkins University School of Medicine. He completed his undergraduate work at Princeton University and received his M.D. and Ph.D. degrees from Johns Hopkins. After a post doctoral fellowship at Harvard Medical School, he joined the faculty at Johns Hopkins.
He is the recipient of a Distinguished Clinical Scientist award from the Doris Duke Charitable Foundation, as well as two NIH [National Institutes of Health] merit awards. Dr. Siliciano's research focuses on the dynamics of HIV replication in vivo, and mechanisms of HIV persistence in patients on highly active antiretroviral therapy.
He will speak today on "HIV Persistence on Patients on HAART: Reevaluating Prospects for Eradication." Dr. Siliciano.
It has been 11 years since the introduction of highly active antiretroviral therapy, or HAART. And the dramatic reductions in viremia experienced by patients on HAART led initially to hopes that the infection could be cured with two to three years of continuous treatment. However, to date, not a single patient has been cured, and the initial hopes for eradication have given way to a pessimism that is so profound that eradication has almost become a taboo subject.
What I would like to do today in my talk is to try and take an objective look at the steps that we need to take in order to find a cure for HIV infection, and review the progress that has been made.
So, in my opinion, there are three steps that we need to take. First, we have to stop the virus from replicating. This is the critical, initial step. This takes away HIV's main advantage, its ability to mutate and evolve, and it is very hard to see how we are going to cure the infection unless we can carry out this initial step. And, of course, the approach is to use antiretroviral drugs to block specific steps in viral replication.
Now the next step is to identify all of the stable reservoirs where non-replicating forms of the virus persist in the body. And, finally, we need to identify ways in which we can eliminate each of those reservoirs.
So I am going to review the progress towards each of these goals, concentrating on the first critical step. And I will begin with some general background. So when patients start on a standard three-drug HAART regimen, the level of viremia falls from the pre-therapy steady state all the way down to the limited detection of current clinical assays, which is about 50 molecules of HIV RNA per mL of plasma. And this rapid decay reflects the fact that the infected cells that produce most of the plasma virus are very short-lived, and so if you stop new cells from becoming infected with HAART, the previously infected cells die off quickly and viremia decays.
Now it was originally hoped that the decay would continue until the infection was eradicated. What we see in a patient who is doing well on HAART is this, simply measurements that come back below the limit of detection and so you do not know whether the decay is continuing or not.
Well, it turns out that it is not. And the reason has to do with the unique mechanism of viral persistence that exploits the fundamental physiology of CD4 positive T cells. Now most of the CD4 cells in the body are in a profoundly quiescent state, and these resting cells include naive T cells that have not yet responded to any foreign antigen, as well as memory T cells that have previously participated in an immune response. And these cells circulate throughout the tissues, essentially awaiting encounter with an antigen that they can recognize. And when one of the cells encounters a foreign antigen that it can recognize, it becomes activated and divides and generates lots of activated effector cells of the same specificity.
Now at the conclusion of the immune response, most of these cells die, but some of them survive and go back to a resting state as long-lived memory T cells. And these memory cells survive for very long periods of time -- decades, in fact -- allowing future responses to the same antigen.
Now what happens in HIV infection is that the virus preferentially replicates in the activated cells, and it tends to kill them. It does not really replicate in resting cells, but what can happen, rarely, is that one of these activated cells can become infected as it is in the process of reverting back to a resting state. And this gives you a stably integrated form of the viral genome in a long-lived memory T cell. And what is particularly interesting is that as the cell makes this profound transition from an activated state back to a resting state, HIV gene expression is turned off. And that is due, in part, to the fact that HIV gene expression depends on a host transcription factor, NF-kappa B, which is excluded from the nucleus in these resting cells.
So as the cells make this transition, HIV gene expression is almost automatically extinguished, and the end result of this is a stably integrated, but silent or latent, form of the viral genome in a long-lived memory T cell.
Now this is almost a perfect recipe for persistence because it allows the virus essentially to persist just as information, and in this form, it is unaffected by immune responses and antiretroviral drugs. And, of course, if the cell becomes activated again in the future, it can begin to produce virus.
So in 1995, we developed assays to detect these cells in patients and we found that they were present in everybody with HIV infection, but only at low frequency. Only about one in a million resting CD4 cells harbors this form of latent HIV. The problem is that these cells persist in patients on HAART, and this was originally shown by Tony Fauci and Doug Richman and by our group in 1997. And we measured the decay rate of this pool of latently affected cells in patients who were doing well on HAART, and the decay rate is extremely slow, as you can see. Note that the time scale here is years, not days. And here are the data representing measurements in large numbers of patients who had suppression of viremia to below the limit of detection for as long as seven years. And at this rate of decay, it would take over 70 years to clear this reservoir.
So we know that HAART can reduce the level of viremia, the level of free virus particles, in the blood to below the limit of detection. And we also know that HIV can persist in these resting memory T cells.
And the next important point is that everybody on HAART is actually viremic. That is, with especially sensitive assays, it is possible to detect free virus particles in the blood of patients on HAART who have a clinically undetectable viral load. And we call this residual viremia. This was first demonstrated by Roger Pomeranz, and confirmed very elegantly by Sarah Palmer, who has developed an assay that can see a single virus particle in the plasma.
So what this means is that what HAART actually does is to reduce the level of viremia to a new plateau that is just a little bit below the limit of detection of current, clinical assays. Now the explanation that is usually given for this residual viremia is that it reflects ongoing cycles of viral replication that are occurring, but just at a lower level due to the presence of the drugs.
Now this is a very disturbing scenario because ongoing replication in the presence of the drugs will inevitably lead to the evolution of resistance and treatment failure. Fortunately, there is an alternative hypothesis. We believe that in the optimal situation, HAART actually stops all ongoing viral replication and that the residual viremia that patients experience is simply a reflection of the release of the virus from stable reservoirs, cells that were infected prior to the initiation of therapy, for example, those long lived memory T cells that I mentioned. And you might imagine that every day some of those cells become activated and they begin to produce virus, and that virus cannot go on and infect other cells because of the suppressive effects of the drugs. However, it can be detected as HIV RNA in the plasma.
Now this is actually a much more optimistic hypothesis because it says that HAART actually stops viral replication. In other words, we have accomplished the first step that I mentioned. So this hypothesis makes a number of predictions that we can test. First it predicts that the viruses in the plasma, the free virus particles, should be genetically similar to viruses in the latent reservoir.
So here is a phylogenetic tree of viral sequences from a single patient. Here are the sequences in the latent reservoir, and now I am going to add the sequences detected in these free virus particles in the plasma using an extremely sensitive sequencing technique. And, as you can see, the two populations of viruses are phylogenetically intermingled, and in some cases, identical. And this is consistent with the idea that at least some of the residual viremia is derived from this latent reservoir.
Now a second prediction is that the residual viremia should continue without viral evolution. In other words, we should not see progressive, evolutionary change in the residual viremia, or the appearance of new drug resistance mutations because the complete cycles of replication that are necessary for evolutionary change are not actually occurring.
So if you look at this phylogenetic tree, you can see that the free viruses in the plasma, the colored triangles, have not diverged significantly from the other viruses in this patient, and furthermore, sequence analysis does not reveal the presence of any new drug resistance mutations. And this absence of new resistance mutations is also characteristic of the free viruses in the plasma of patients during blips, these transient elevations in viremia that many patients on HAART experience. In fact, blips are likely just a reflection of the fact that everybody on HAART is viremic due to the release of virus from stable reservoirs, and that there will inevitably be some biological and statistical fluctuation in the level of residual viremia that is occasionally captured as a blip.
Now the third and most important prediction of this hypothesis is that intensification of HAART should not further decrease residual viremia. That is, if we add a fourth drug to a potent HAART regimen, we should not see a further decrease in the level of residual viremia. And that is because there are really no ongoing cycles of replication for the drug to inhibit.
So we carried out an intensification study in collaboration with Frank Maldarelli and John Coffin at the NCI [National Cancer Institute], and John Mellors at the University of Pittsburgh. And I would like to present results from a subset of patients who are on an optimal, efavarinz-based regimen and had measurable levels of HIV RNA in the plasma. And for these patients a boosted form of the potent protease inhibitor atazanavir was added to their regimen for eight weeks.
So here are the mean levels of HIV RNA prior to intensification. And you can see that they are below the limit of detection of current clinical assays. But they are readily measurable with Sarah Palmer's single copy assay. Now what happens when you add atazanavir for eight weeks? What happens is that nothing happens. The level of viremia does not change. And here are the levels post-intensification. We have seen this same result in all 15 patients studied with three different intensification drugs. The result is that intensification of HAART has no effect on residual viremia. This result strongly suggests that the residual viremia is derived from the release of virus from stable reservoirs of long lived cells infected prior to the initiation of therapy.
Now we cannot exclude a small contribution from ongoing replication, but it is clear at this point that the major problem is the release of virus from stable reservoirs. So before I turn to how we find and eliminate those reservoirs, I would like to present some new research into the basic pharmacology of HIV drugs that helps us understand how it is that HAART can actually stop all ongoing viral replication.
So this is a dose response curve for a hypothetical antiretroviral drug. And what is plotted is the fraction of infection events that remain uninhibited as a function of the drug concentration. And this reaction decreases, obviously, as you add more drug.
Now we can describe this dose response curve with a simple mathematical formula, and I realize it is a little early in the morning for mathematical formulas, but this one simply says that the amount of inhibition is a function of three things. The drug concentration D, the IC50, which is a standard pharmacologic measure of drug potency, and the exponent M, which describes the slope, or the steepness, of the dose response curve. And this slope parameter is analogous to the Hill coefficient, it is a measure of cooperativity in the binding of multiple drug molecules to a single target.
Now interestingly, the HIV protease and reverse transcriptase inhibitors that we use to treat HIV infection, bind to a single site on the relevant target enzyme, and so this slope parameter has largely been ignored, or has been considered to be equal to one, in which case it drops out of the equation. And if the value is greater than one, the curves simply become a little bit steeper.
Now we tend to use these antiretroviral drugs at concentrations well above the IC50 in this pink shaded range. And it does not appear that in this range, the slope parameter makes much difference. But that is only because the conventional way of plotting these dose response curves does not adequately describe the degree of inhibition produced by antiretroviral drugs. In fact, because viruses replicate exponentially, it does not make any sense to plot the inhibition of viral replication on a linear 1:100 scale as is conventionally done. Rather, we should use a logorhythmic scale so that we can see the fraction of events going down from 100-percent to 10-percent, one percent, 0.1, and even lower.
So if you do this and you take the same dose response curves from the last slide and simply replot them with a logorhythmic Y axis, the result is truly shocking. What happens is that the dose response curves for drugs with different values of the slope parameter diverge dramatically in the clinically relevant concentration range such that drugs with a higher value of the slope parameter cause much, much more inhibition of viral replication by orders and orders of magnitude.
And therefore we decided that we better measure the slope parameter for current HIV drugs, and determine whether it was different than the expected value of one, and we did this with a very sensitive assay that can see a single infection event.
So for the AZT-like drugs, the nucleoside analog RT inhibitors, the slope values were all exactly one. For the non-nucleoside RT inhibitors, however, the values were around 1.7. For the protease inhibitors, they ranged all the way up to 4.5, for the T20-like drugs infusion inhibitors around 1.7. And for five structurally diverse inhibitors of HIV integrase, the values were all one.
Now this begins to make a bit of mechanistic sense if you consider the fact that the two drug classes that have a slope value of one, the nucleoside analog RT inhibitors and the integrase inhibitors both target an enzyme nucleic acid complex for which there is only a single relevant copy per virus.
In contrast, multiple copies of HIV protease are simultaneously involved in the maturation of each virus particle and this, we believe, allows for a form of intermolecular cooperativity that gives rise to the steep dose response curves.
Well, what difference does all this make in terms of how well the drug actually works in patients? To address this question, Lin Shen, a graduate student in our lab, developed a very simple index called the instantaneous inhibitory potential. And what this is, is simply the number of logs, or the number of tenfold reductions in the amount of infection that you get from a drug at some clinically relevant concentration, for example, the CMAC, the peak plasma concentration. So, for example, here are the dose response curves for two drugs that would be judged to be equally potent based on the conventional IC50 measure, but which have different values of the slope parameter.
For a drug with a slope of one, you are getting about a two log reduction, a 100-fold reduction at the peak plasma concentration, so the IIP value is two. For a drug with a slope value of three, however, you are actually getting about a million-fold reduction, a six log reduction, and so the IIP value is six. And so Lynn has gone on to measure the IIP values for current antiretroviral drugs because this index captures the importance of a slope parameter in a number that is fairly easy to understand and it is just simply the number of logs by which you knock down viral replication.
So for the nucleoside analogs, the values for the IIP range from one up to four; for the non-nucleoside RT inhibitors, from three up to six; for the protease inhibitors, from two all the way up to ten, with lower values being seen for the fusion and integrase inhibitors.
So the most striking result here is that some of the HIV protease inhibitors can cause a ten log reduction in just a single round of infection. This is a ten billion-fold reduction, and at this level of inhibition I think it is easy to understand how it is that HAART can actually stop all ongoing replication. In fact, it would be surprising if any replication were occurring.
And these results also help us understand, at long last, why it is that the HAART regimens that have consistently been shown to be the most effective in multiple clinical trials always contain a non-nucleoside RT inhibitor, or protease inhibitor. These are the drug classes that have this extraordinary potential to inhibit viral replication.
So now that we have seen how it is that HAART can actually stop ongoing replication, let us turn to the issue of viral reservoirs. How do we find them and eliminate them?
Our approach to identifying viral reservoirs has been to focus on the free virus in the plasma, the residual viremia, which we now know is derived from release of virus from stable reservoirs, and we hope that by studying the free virus in the plasma, we can deduce the nature of the cells that are continuing to produce virus in patients on HAART.
So, as I mentioned before, at least some of the residual viremia appears to be derived from the latent reservoir. However, in a number of patients, the story is a bit more complicated. For example, in this patient, I have only shown you part of the phylogenetic tree. Here is the rest of the phylogenetic tree. And what is particularly interesting is that in this patient, almost all of the viruses in the plasma are identical to one another representing a single virus species that has been captured in the plasma in multiple, independent blood samples over a three year period. And we have seen this same sort of phenomena in about half of the patients studied, where the residual viremia is dominated by a small number of viral clones. And these clones do not show evolution over time, and most interestingly, we cannot find these clones in resting CD4 positive T cells, suggesting that they are coming from another source. And I should say that Tobin et al, the only other group really that has done sequence analysis of the residual viremia, has seen exactly the same phenomena.
And our current hypothesis is that this represents the rare infection of a stem cell or a progenitor cell, for example, in the monocyte macrophage lineage, and that this cell is able to divide after infection. And so it copies the viral genome without error into multiple progeny cells, and then these cells go off and produce virus particles for their normal life span.
So the idea that there is a second major source of residual viremia, perhaps in a cell that has some capacity for self renewal, is clearly a disturbing one and will require much further research.
Now what about eliminating viral reservoirs? For the latent reservoir in resting CD4 positive T cells, the approach that several groups have taken is to use agents that induce global T cell activation and in the course of doing so, turn on latent HIV. Now the problem with this approach is that you also activate all of the uninfected resting CD4 cells, and those cells go on to produce cytokines in such large amounts that unacceptable toxicity results. It would clearly be preferable to find agents that can turn on latent HIV without inducing a global T cell activation. And the viruses that are produced in the process would not go on and infect other cells because of the suppressive effects of the drugs and one would hope that these cells would go on to die, either from virocytopathic effects or from host immune attack.
Now, unfortunately, we do not have such an agent right now, but we can take hope from the fact that several laboratories have developed ways to generate these latently infected cells in the test tube from primary T cells. And this gives us, at long last, a realistic in vitro model for HIV latency so that we do not have to depend on the transformed cell lines that have been used in previous research. And the hope is that with these really more realistic model systems, it will be possible to identify agents that are able to target the latent reservoir.
And let me just describe how we do this in our laboratory. What we do is to take primary T cells from a normal donor, and transduce them with a gene that promotes survival, the BCL2 gene. This allows the cells to survive for a long period of time in the test tube, and then we can activate the cells and infect them with a form of HIV that is relatively non-cytopathic, it does not kill the cells as readily, and then culture them, and they actually survive long enough in vitro to go back to a resting state so that the virus becomes latent. And this, essentially, recapitulates the process by which a latency is generated in vivo.
And the hope is that with models like this, we will be able to produce enough latently infected cells in the test tube that we can use them to screen libraries of drugs, large drug libraries, to identify novel compounds that are able to activate latent HIV, hopefully without inducing global T cell activation or unacceptable toxicity.
So let me just sort of summarize what I have said. As a result of the efforts of literally thousands of people in universities and in the pharmaceutical and biotech industry, people who have participated in the development of HAART. I think that we can now say that the first step has essentially been accomplished.
Now it would be a mistake to say that we are one-third of the way there to finding a cure for HIV infection, because the second and third steps may turn out to be much more difficult, although some progress has been made. But the fact that the current HAART regimens can stop ongoing viral replication is really of enormous significance because what that does is to take away HIV's main weapon against us, and that is its ability to evolve. And the fact that current HAART regimens can stop ongoing replication poses a unique challenge for everybody in this room.
What it means is that treatment failure is not inevitable. Even with the drugs we have today or with forms of drugs that are very similar to what we have today, if we could develop forms of these drugs that could be taken for life without unacceptable toxicity, then it is, in principle, possible to offer everyone who is currently living with HIV infection the chance for a normal life. And this is the long term challenge that all of us face.
So, as I conclude, let me just thank the people in my laboratory that did the work that I mentioned. The longitudinal studies on the latent reservoir were done by my wife Janet and Diana Finzi, a former graduate student, and this work would not have been possible without the help of some fantastic physicians in our HIV service, Tom Quinn, Dick Chaisson, and Joel Gallant. The work on the intensification of HAART was done by Scott Kim and Jason Dinoso, along with our collaborators, Frank Maldarelli, Sarah Palmer, and John Coffin at the NCI, and John Mellors at the University of Pittsburgh. The work on the slow parameter was done by Lin Shen, along with Moira McMahon, Susie Peterson, and Haili Zhang, Ben Jilek, and Marc Callender. The work on the predominant plasma cone was done by Tim Brennan, Justin Bailey, and Ahmad Sedaghat, and Rick Nettles and Tara Kieffer. And finally the development of an in vitro model for HIV latency was done by Andrew Yang and Yen Zhou. Thank you very much.
|Please note: Knowledge about HIV changes rapidly. Note the date of this summary's publication, and before treating patients or employing any therapies described in these materials, verify all information independently. If you are a patient, please consult a doctor or other medical professional before acting on any of the information presented in this summary. For a complete listing of our most recent conference coverage, click here.|
The content on this page is free of advertiser influence and was produced by our editorial team. See our content and advertising policies.