This commitment to HIV research and AIDS care has earned McCune the admiration of activists, academicians, and most everyone in between. He has also won their respect with his research, notably the development of the SCID-hu mouse as a model of HIV infection. The work that followed that innovation-what McCune calls "a wild-ass experiment" conceived on the commute from Stanford to San Francisco in 19871-demonstrated that the mouse is a versatile model of HIV pathogenesis and drug therapy.
McCune is now an Associate Investigator at the Gladstone Institute of Virology and Immunology at SFGH and Associate Director of SFGH's General Clinical Research Center. The Journal talked to him about HIV pathogenesis and therapy-and of course the SCID-hu mouse-on October 8, 1995.
McCune: It's hard to tell. To quantitate the kinetics of viral and CD4+ lymphocyte turnover in their studies,2,3 they gave antiretrovirals and showed 2- to 3-log drops from high levels of virus in blood and reciprocal increases in CD4+ counts. Then they measured the slopes of the curves that described those changes and derived a constant for the turnover of virus and CD4 lymphocytes. The turnover calculated is pertinent for patients with that level of virus, that is to say, patients with late-stage disease. It's more difficult to use a similar approach in early-stage disease because viral loads are almost always lower then and CD4+ counts are often normal. If you give an antiretroviral to block replication, the incremental drop in virus will not be as great because there's not as much to start off with. So it would be more difficult to calculate whether the same turnover rates pertain in earlier stages of the disease. In any case, the numbers for viral and cell turnover during early stages of disease have not yet been published.
At this point, people have to make a subjective judgment about whether these viral kinetics apply during the entire course of disease. Some have the impression that those dynamics do occur in early-stage disease. Others are waiting for the experiments to be done.
My reading of the literature suggests that in early-stage disease it's much easier to find evidence of cells that are infected but not actively replicating virus. In a recent review,4 I looked at six or seven years of work by a number of labs showing that "latent" or "restricted replication" states are much more prevalent in early than in late stages of disease. That suggests to me that the dynamics may not be the same throughout the course of infection. The states of restricted replication may in fact even predominate in early stages of disease. Then, as the disease progresses, cells that otherwise would have had latent or restrictively replicating genomes may move into cell cycle and begin to produce virus. And that may be why a high virus load and high cell turnover are observed in late stages of disease. So my own view is that we still can't say whether the findings of Ho and Shaw apply to the entire disease course.
Journal: Michael Nowak has advanced the hypothesis that progression happens when an antigen diversity threshold is crossed.5-8 His theory is based largely on computer modeling. Is this idea feasible? Can it be proved?
McCune: The model says that the immune system can mount a strong response against several HIV variants at the same time. But when the viral population contains too many variants-when it becomes too diverse-the immune system can no longer keep up. Experimentally, it's a testable hypothesis, although it's difficult to test. The paper in Nature that was coauthored with McMichael7 is probably the most concrete work on this hypothesis so far. But it's only two patients. If more patients are studied it's possible that the data would fall out in a way that would or would not be consistent with the hypothesis.
McCune: In vitro there are descriptions of several forms of latency (Figure 1). In one, the HIV genome is completely silent and, theoretically, would be impervious to immunologic or pharmacologic antiviral approaches. In another, the genome exists in a state of restricted replication, producing some but not all viral transcripts and little (if any) virus. The transcripts that are made are generally those for the regulatory genes, the multiply-spliced transcripts that are found early in the course of the viral life cycle. Since these regulatory genes encode proteins, the infected cell could potentially express HIV antigens on the cell surface and therefore could be susceptible to immunologic attack. Additionally, the regulatory genes that are expressed could exert direct pathologic effects. In either case, antivirals that prevent the transcription or action of these gene products may be beneficial.
The problem with any therapy is to balance its efficacy against its side effects. There are precedents in the field of oncology for using toxic drugs in an aggressive manner to eradicate tumors. The paradigm is to treat with doses that are high enough to kill the tumor but not the patient. Much of the regimen is related to keeping the patient alive in the face of these side effects. This is an aggressive approach and is usually justifiable in patients with late-stage metastatic disease. The justification is sometimes less apparent in those with earlier stages of disease, in whom, for instance, surgical removal of the tumor might suffice.
Now, if we're going to talk about treating HIV disease that way, if we're going to treat aggressively with drugs that might have untoward side effects, in the same way that I would stage a cancer patient by determining that patient's load of metastatic cells, in the HIV-infected patient I'd want to know the load of HIV-infected cells-not just the ones that are actively replicating virus, but also those that may not be actively replicating virus. If the drugs tend only to block viral replication, and if infected cells replicating virus constitute a minor subpopulation, then the drugs may not carry as much benefit as when that population is the major population.
There might, for instance, continue to be a sizable population of infected cells that are untouched by the therapy, despite its toxic side effects. Some of those cells may even carry genomes which, when eventually expressed, are resistant to the therapy. If this were to be the case, the toxicities of treatment might outweigh the benefits. On the other hand, let's speculate that latent viral genomes or genomes with restricted replication are abundant in early stages of disease and that you had a way of getting rid of them. You might then give antiretrovirals to cover spread from that small subpopulation that may be actively replicating virus. You would want to keep that replication in check. At the same time, it might make sense to accept the toxicities of treatments that would reduce the size of the infected cell populations that are not producing virus.
How might you do that? At this point, it's completely conjectural. My main point is that we need to find out first what the status is during early-stage disease, and then develop strategies for early treatment. If you knew these latently infected populations were there and you understood more about their biology, it might be more apparent what to do with them. There's been very little done-10 or 20 papers, all of them are referenced in that review4-that address this population of cells in asymptomatic patients. My inclination would be to focus more attention on the biology of infected cells during that stage of disease. I don't disagree with the premise that one should treat early and treat aggressively. I just think that we should plan such treatments in a way that accounts for all major subpopulations of infected cells and monitor people given such treatments with these various subpopulations in mind.
Journal: Luc Perrin, David Ho, and David Cooper are starting trials of pretty aggressive antiretroviral combinations in people during acute infection. Is it known when this latent population takes root? If these aggressive early therapies can control viral replication right in that first month, say, would HIV be able to get into cells in a state of restricted replication?
McCune: I agree with the idea of early and aggressive treatment-if you knew there was an acute infection. If you can stop the virus at that point, it's unlikely that many cells will be infected. Even though there would still be some latent genomes, their total number would probably be smaller. But nobody has really looked at this issue of restricted replication in the setting of acute infection. And the main reason is that acute infection is so hard to catch. We figure even here in San Francisco, where there's a relatively active sexually promiscuous population, the incidence of acute retroviral syndrome is pretty low. To capture those events and study those people would be extraordinarily difficult.
Journal: The figure one often hears for the proportion of HIV-infected people who have noticeable symptoms of acute infection is about one-third.
McCune: That's probably about right. Most people who have any viral infection will have subclinical or just mild clinical symptoms. Then a fraction, usually less than half, will have clinical symptoms that come to medical attention. I think it's fair to say that most people become acutely infected with HIV without knowing it.
Journal: So that figure of one-third is conjectural?
McCune: It's based on taking oral histories from seropositive people and asking them if they remember a contact with someone who was HIV-positive that may have led to their infection, then asking them if they had a flulike syndrome during that time of contact. The viremic phase of HIV disease is similar to the viremic phase of rhinovirus or influenza or other viruses-basically the same symptoms. It's very difficult to say from the oral histories whether the flulike symptoms that these people recall were actually caused by HIV or by some of the more common causes of those symptoms-rhinoviruses, flu, and so on. I'd say they're probably soft numbers. If anything, they probably overestimate the incidence of real clinical symptoms caused by HIV.
McCune: There's evidence for both direct killing and indirect killing, but there are no studies-in humans at least-that quantitate the absolute amount of cell death occurring by each mechanism. So it's really hard to say which is more important. We've shown in the SCID-hu mouse (Figure 2) that when the human thymus is infected with HIV almost all of the cells die within weeks, but only about 10 percent of them are infected.9,10 The rest die by programmed cell death, which is indirectly induced by HIV infection. At least in this mouse model, indirect killing appears to be more important as studied on an absolute basis. I don't think people would argue that indirect death is not happening in humans. I think there's enough evidence now to suggest that it is.
So what are the implications for therapy? The mechanisms of indirect killing may be very different from the mechanisms of direct killing. Both, all would agree, are induced by HIV. So in each case one would give antiretrovirals to block HIV replication. The indirect methods, on the other hand, may rely on mechanisms that can be addressed separately. We're learning more and more, for instance, about mechanisms that induce programmed cell death, and these mechanisms may become pharmacologically approachable. A number of biotech companies are working on such approaches right now. So if indirect death is something that's occurring in vivo, one could find out how important it is by blocking it in humans and asking if such an intervention actually had a significant impact on disease progression. My hunch would be that if you blocked both the indirect and the direct mechanisms, you would probably do better than if you just blocked one alone.
There's another aspect to this. In those instances in which indirect killing mechanisms have been evoked, envelope [env] and tat have been the main mediators. Most of the antiretrovirals out there will block movement of virus from one cell to another. But most of them will not remove cells that are already infected. So, even if you have a potent antiretroviral that blocks spread from one cell to another, envelope and tat may still be made by a significant population of cells. If they are, they may still indirectly induce cell death, which is why giving both antiretrovirals and agents that block indirect mechanisms of cell death may be especially good, at least in the beginning parts of therapy.
McCune: There are two things to consider: If therapy only blocks spread from one cell to another, it's not likely to be as good as therapy that shuts down replication. Shutting down replication almost completely would of course be better than just shutting down spread, because cells that are replicating virus can still cause damage (for example, by indirectly inducing cell death). So the ideal situation would be combinations of antiretrovirals that could shut down replication virtually completely. If so, that's great. That'll likely be a winner. The viral DNA alone, inserted into the cell, should be something that the system can tolerate, particularly because all cells have a finite lifetime. If a patient is infected at day one and then is treated a year later with this dream combination, at that treatment time there's going to be a fixed number of infected cells in the body. Some will be productively infected. Some will be nonproductively infected. We can just forget, for the time being, what fraction of cells exists in each population. If you now give the therapy that blocks replication, the ones that are productively infected are going to die at the end of the given lifespan that Ho and Shaw calculated. And at some point the cells that are nonproductively infected will reach the end of their finite lifetime and be cleared. And if no other cells have been infected in that time period, then you have a host that's cleared of HIV. The key question is, "How long is that time?"
Journal: You sound skeptical that these kinds of combinations are going to be available anytime soon.
McCune: It's complicated because a lot of these drugs have side effects, and it's hard to give drugs that have side effects, especially in combination, to patients who are asymptomatic. I think we're going to learn a lot in the next five years or so about combinations that work well in late-stage patients. And I think in that same period we'll also learn much more about what's happening to the various cell populations in the early stage. Then we're going to approach a time, which may come up pretty quickly in small-scale trials, when combinations of drugs will be given to patients who are asymptomatic. If we have methods in place to analyze efficacy, then we may convince ourselves that these combinations are beneficial. The problem is we don't have many measures of efficacy. The CD4+ count in the asymptomatic person can be normal. The viral load is usually low. Symptoms, if any, are mild. So I fear that it will be difficult to gather clear-cut data on efficacy in the near-term future. The endpoint that everybody will be happy with will be symptom-free disease. Since the interval between infection and the acquisition of symptoms is so long, it becomes hard-almost by definition-to imagine that the data will come out quickly.
There are a few populations in which what I just said is not true. There are some adults who progress quickly. If there are enough asymptomatic adults in this putative trial of the "right combination," the statistical power may demonstrate that you have reduced disease progression in that small cohort that normally does progress quickly. There's also a rather large proportion of kids with HIV infection who progress quickly: 20 to 30 percent. It's possible that in the near-term future such combinations could be used in pediatric HIV disease to show very clearly that they will work to reduce the symptoms and to increase the lifespan. That will beg the question of whether pediatric HIV disease is the same as adult HIV disease, which I think is an open question at this point. But it's probably the population in which this particular hypothesis could be tested most quickly.
McCune: There's definitely a delivery problem. Cytokines are pleotropic; they're expressed in many different cell types. And they usually work by cell-cell interactions. There are very few-erythropoietin is one example-that are made in one place and then circulate to work in another place. If you give pharmacologic doses-high doses-of any cytokine intravenously, you're going to cause multiple side effects.
That being said, you come back to the same equation that I mentioned before: If you can balance side effects with efficacy in a rational way, it may end up making sense. IL-2, for example, has side effects in humans when given continuously at high doses that are less severe when given intermittently at lower doses. Again, it's a drug delivery problem. Many drugs have been found to be toxic when given by one dose or by one route or by one regimen or formulation, and to be very beneficial when given in other ways. In my mind, the first goal is to answer the question, "Is cytokine dysregulation a significant component of disease?" In fact, we're doing a lot of work on that in the lab to try to find out which cytokines are upregulated and what upregulation-or downregulation, for that matter-may mean in the context of the bone marrow or the thymus.
If we find a correlation between upregulation of a given cytokine and a defined aspect of pathology-let's just say, for instance, upregulation of a cytokine causes hematosuppression-that doesn't necessarily mean there's a direct relationship. They might be coincidental. One way to test whether there's a direct relationship is to block the cytokine. We can do that in short-term studies in animal models and in people. In a short-term study, one could give a cytokine antagonist-block the cytokine-and then evaluate whether such inhibition relieves the associated pathology, in this case, hematosuppression. So that's an experiment that helps to validate the hypothesis that this particular cytokine causes a pathologic event.
Now, if I got to the point where I could validate in a person that a given cytokine antagonist was helpful in relieving a given pathologic event, then I would try to figure out how to deliver it safely. In other words, I would start with the mechanism, then work towards the drug delivery problem. Drug delivery is an area unto itself. In the little time I spent in the biotech/pharmaceutical world, I was impressed with how empiric the testing of drug delivery is. You have to evaluate things empirically to figure out what works and what doesn't work. You can't make predictions.
Right now, based on what I know, I'd say yes, cytokine-based therapies are likely to have toxicities because of the biology of the system. But I wouldn't throw in the towel and say therefore they will never be used effectively. My main focus would be to ask "Should it be done?" and "Is there a reason that's logical?" and then move forward from there.
Journal: One of the big questions with IL-2 therapy is whether the additional CD4+ cells it engenders are fully functional-or just more food for HIV. What do you think?
McCune: There are really several questions: Are they functional? Are they representative, that is to say, are they clonal or polyclonal or oligoclonal-is there a full repertoire of T-cell receptors, or just some, or just one? Multiple myeloma is a B-cell neoplasm characterized by lots of B cells in the marrow that are all from the same clone. I don't think that's the case here: Cliff Lane has shown data that there are different V-beta regions on these T cells. But data have not been presented, to my knowledge, about whether these individuals can be immunized, for instance, and if the T cells that are new are responding as they should-just simple data in terms of T-cell function in vitro. I'm sure there will be; they're probably doing these experiments now.
Journal: If you look at it on a simple level, they've been treating some of these people for several years. If the people who reach normal CD4+ cell levels can maintain them for years and don't get opportunistic diseases, would that clinch the case that these T cells are working for them?
McCune: They're going to need a large number of people. Of the patients that have been reported, some but not all showed dramatic rises in T-lymphocyte counts. If the number of patients with such a response becomes large enough, and if they have a lower incidence of OIs and death-that would be pretty convincing.
Journal: Perhaps because the cytokine response is so complex in HIV infection, immunomodulatory therapies are sometimes advanced with lots of confusing fanfare. Cytolin is the latest example. How can clinicians-and especially people with late-stage disease who may be desperate to try anything-figure out which of these new strategies has any merit and which may actually be more dangerous than doing nothing?
McCune: They need to seek information from unbiased sources, and that's hard to do. Obviously it's hard for the makers of these therapies to be completely objective. And many private physicians don't have the objective information that patients need. They just don't know the biology and may not be able to help the patient make a decision. I think the advocacy groups are the way to go. Project Inform here in San Francisco does a wonderful job in communicating current and balanced information. Also in San Francisco, the Community Consortium, which is run by Don Abrams, does a great job in sifting through the various street therapies that are out there-not judging, but providing information about what's known and what's not known. Those two groups in particular have fewer connections with drug makers and governmental bodies that could be seen as damaging ties. With my own patients, I try to give them a balanced viewpoint, and I also tell them to check it out with Project Inform. There are not many groups like that. And Project Inform in accessible nationwide. [Project Inform's national hotline is 800-822-7422.]
McCune: One of the biggest problems with gene therapy right now is, again, a delivery problem: How do you get the gene into the right cells and get it expressed? My own bias has been to put it into progenitor cells, thinking that they would divide and go on to become more mature cells, and the delivered genes would end up in those more mature cells. But introducing genes into progenitor cells has turned out to be even more difficult than introducing genes into more mature cells, which itself is difficult.
That being said, if the anti-HIV gene was inserted into only a small fraction of cells and yet was completely protective, a high turnover rate of unprotected cells could actually facilitate the engraftment of the protected cells: they would move into the spaces left behind. So it could be that the high turnover rate might actually help, not hinder, gene therapy. If there's a high turnover rate in the peripheral lymph nodes, there might be feedback stimulation to the bone marrow that says "Make more cells." If there is, and if there are transduced (or gene-modified) stem cells in the marrow, more gene-modified mature cells might be made. Of course this is all conjectural.
The problem is, if the indirect mechanisms of HIV destruction that we were talking about pertain, then what I just said may not hold. Because even if you had a gene inside the cell and the gene was completely protective against HIV, the cell might still be killed by an indirect mechanism.
The other problem is that the kinetics you're referring to involve only the turnover of CD4+ T cells. But there are lots of other CD4+ cells in the body that are infected and important in HIV disease. Many of them exist in the organs-the lymph nodes, the thymus, the bone marrow-and many are important for differentiation and function of progenitor cells and mature cells. And if those cells are being turned over at the same rate and killed, then there may be no microenvironment left in which the gene-modified cells can differentiate and function. That, I think, is a fundamental question now about gene therapy: In people with HIV infection, are these bone marrow, thymic, and lymph node microenvironments still able to support differentiation and function of gene-modified cells? We don't know yet.
McCune: It's pretty simple. So far, research on HIV disease in humans has focused on infection and destruction of the peripheral lymphoid system. In the mouse model we found that HIV directly and indirectly kills thymocytes and knocks out thymocyte differentiation.9,10 If HIV also infects the thymus in humans and destroys it, it would be working from both sides: HIV would destroy the peripheral lymphoid system and destroy the ability of the thymus to make more cells. Most people would agree that this scenario could be pertinent to pediatric HIV disease, where the thymus is generally considered to be active. So we're looking directly at the structure-the mass-of the thymus, to see if it is indeed destroyed during HIV infection.
Since there are many more adults than kids with HIV disease, we've started with HIV-positive adults, in whom most people think the thymus is not functional at all. In fact, there has been little work on the structure and function of the adult thymus, except in the case of very sick people who have died or a handful of healthy people who have died in accidents. We're doing a more systematic study using noninvasive CT imaging of thymuses in HIV-infected adults at various stages of disease, and we'll be moving shortly to HIV-infected kids as well.
Journal: You developed the SCID-hu mouse model of immune deficiency disease (Figure 2). Are you doing any HIV-related work with the SCID-hu now, and do you think that model has been exploited to its full advantage in HIV infection?
McCune: I think we know much more now about what the SCID-hu model is good for, and in those areas there's a lot more to do. It's proved to be quite useful in the evaluation of central human hematopoiesis, in bone marrow and thymus in particular. We've spent a lot of time looking at effects on thymus, but very little time looking at effects on bone marrow. My lab is using SCID-hu mice now to study the effects of HIV in both organ systems. The studies are much more controlled using the animal model than they would be in humans. We will look for mechanisms of HIV pathology and try to specify them using the animal model. If we identify such mechanisms, we will try to block them pharmacologically to test their relationship to disease progression.
If and when we can formulate hypotheses about mechanisms in the mouse model, we will test whether similar mechanisms are operative in the same organs in humans. It's not easy to get a bone marrow biopsy or aspirate. They hurt. It's even harder to get a thymic biopsy. So unless you have a good hypothesis about a mechanism, it doesn't make much sense to ask volunteers to make donations from those organs. But we're moving into that phase now, where we will be testing whether or not the mechanisms we observed in SCID-hu apply in humans as well.
We've also spent a considerable amount of time testing standard antiretroviral therapies in the SCID-hu model, and that activity has now moved to my lab at Gladstone. We have a system that's been set up over the past four or five years that allows us to evaluate the activity of single agents and combinations of drugs against HIV in vivo.11 Since it seems to be clear that we'll be using combinations in the clinic, and since combinations are difficult to test there, I think one of the future uses of the mouse model that hasn't been exploited as much as it could be is the preclinical evaluation of which combinations work best. There's been very little testing of immune-based therapies in this system, but we'll be doing more of that in the future.
Before long we'll probably have a more logical staging route that begins by testing drugs combinations in small animal models like SCID-hu mice. Then we may be able to move on to drug testing in the SIV-infected Rhesus macaque, especially with SIV-HIV ("SHIV") recombinants. Eventually, a combination of animal models could help us design our trials in humans better.
McCune: At the basic level, we've learned a tremendous amount about the molecular biology of the virus. But we still need to learn more about the biology of the virus in vivo, especially within target organs like the lymph node, thymus, bone marrow, and brain. Then, to translate that into the clinical arena, we need better ways to evaluate whether various therapeutic modalities have the desired effect-whether they stop these pathogenic events. We don't have many tools in the clinic now to do that. And that's a great hindrance to the development of better therapies.
For many people with HIV disease, of course, most of the symptoms come from opportunistic infections that supervene and cause problems. We're doing pretty well against Pneumocystis, but we can do a lot better against many of the others. That's been an area that hasn't commanded much attention and that needs much more emphasis.
In terms of care, at least in San Francisco, we've done pretty well taking care of the male homosexual population. But even here, and certainly elsewhere, there are other groups that are underserved: women, children, injection drug users. This is a big problem that needs to be addressed. Of course, the most underserved groups are those in the underdeveloped nations. Many current and emerging therapies are going to be completely inaccessible in Africa and Asia. Obviously we need to spend more time thinking about how to deal with that. The whole area of prevention is largely overlooked. Vaccines may well work in the third world before they're ever tried here. But here, important aspects of education are almost completely ignored-or if not ignored, they're actively thwarted. Sex education in schools, the distribution of condoms, the distribution of clean needles-those kinds of strategies could have a tremendous impact on slowing the spread of HIV disease in this country. These measures are relatively inexpensive, and they've proven to be effective with other sexually transmitted diseases. Their use should be encouraged, not forbidden.
Mark Mascolini is a freelance writer specializing in HIV infection.
1. Montgomery G. The human mouse. Discover 1989;August:49-55.
2. Ho DD, Neumann AU, Perelson AS, et al. Rapid turnover of plasma virions and CD4 lymphocytes in HIV-1 infection. Nature 1955;373: 123-126.
3. Wei X, Ghosh SK, Taylor ME, et al. Viral dynamics of human immunodeficiency virus type 1 infection. Nature 1995;373:117-122.
4. McCune JM. Viral latency in HIV disease. Cell 1995;82:183-188.
5. Nowak MA, Anderson RM, McLean AR, et al. Antigenic diversity thresholds and the development of AIDS. Science 1991;254:963-969.
6. Nowak MA, McMichael AJ. How HIV defeats the immune system. Sci Am 1995;August:58-65.
7. Nowak MA, May RM, Phillips RE, et al. Antigenic oscillations and shifting immunodominance in HIV-1 infections. Nature 1995;375: 606-611.
8. Coffin JM. Lines drawn in epitope wars. Nature 1995;375:534-535.
9. Su L, Kaneshima H, Bonyhadi M, et al. HIV-1-induced thymocyte depletion is associated with indirect cytopathogenicity and infection of progenitor cells in vivo. Immunity 1995;2:25-36.
10. Bonyhadi ML, Rabin L, Salimi S, et al. HIV induces thymus depletion in vivo. Nature 1993;363:728-732.
11. McCune JM, Namikawa R, Shih CC, et al. Suppression of HIV infection in AZT-treated SCID-hu mice. Science 1990;247:564-566.
12. Shih CC, Kaneshima H, Rabin L, et al. Postexposure prophylaxis with zidovudine suppresses human immunodeficiency virus type 1 infection in SCID-hu mice in a time-dependent manner. J Infect Dis 1991;163:625-627.
13. Kaneshima H, Su L, Bonyhadi ML, et al. Rapid-high, syncytium-inducing isolates of human immunodeficiency virus type 1 induce cytopathicity in the human thymus of the SCID-hu mouse. J Virol 1994;68:8188-8192.