Discussion

Russell: You administered a high dose of Mycobacterium vaccae to mice, and your interpretations were based on antigen-specific T cell responses. However, a large dose ofM. vaccae will contain lipopolysaccharide, as well as bacterial cell wall lipids, and this is a pro-inflammatory environment which will likely contain a high level of interleukin (IL)-6. This environment will favour a T helper (Th) 2-type response, irrespective of the antigen given.

Rook: First, these preparations are suitable for human use and they do not contain endotoxin which, in any case, is not a component of mycobacteria. Second, the dose is not that large — it is 60 p.g of material given two months before the infection. Therefore, your suggestion is unlikely.

Russell: You could administer an irrelevant antigen and look at the proportion of CD4+ cells produced.

Rook: Yes, it would be interesting to administer ovalbumin in similar amounts and see the effect of the Th2 response.

Russell: Barry Bloom and colleagues have recently argued that mycobacteria escape into the cytosol (McDonough et al 1993), which is why a CD8 response is observed, and I would argue that there's no need to invoke that kind of route, because latex beads coupled with protein and inoculated into mice results in a response that includes CD8+ T cells. Therefore, there is an intersection of the class I presentation pathway with the endocytic pathway, and this is not fully understood but it does generate a CD8+ cytotoxic T cell response.

Rook: There are many examples where this occurs, but the point I was making was that killed Bacillus Calmette—Guerin (BCG) and killed Mycobacterium tuberculosis do not generate this type of response, whereas killed M. vaccae does (Skinner et al 1997, M. Skinner, S. Yuan, R. Prestidge, J. D. Watson & P. L. J.

Watson, unpublished paper, Vaccines beyond 2000, Queensland, Australia, 6—9 July 1997). There may be other environmental saprophytes which have not been studied that produce this response.

Russell: Leishmania also produces a CD8 response.

Orme: There is a general consensus that when M. vaccae is injected into mice interferon is produced, so how can the production of CD8+ cytotoxic T cells be explained?

Rook: These pathways don't obey the current dogma, but the data are strong (Skinner et al 1997).

Blackwell: When you heat kill this bacteria are you also destroying all the DNA? Could you effectively be vaccinating them with DNA? Because this may lead to a CD8 response.

Rook: We're not destroying the DNA. The best way to extract DNA from this particular organism is to heat kill it. It is indeed possible to do PCR with the autoclaved material.

Jlmer: It would be highly unlikely that the DNA caused a CD8 response, unless some of this mycobacterial DNA was expressed by the eukaryotic cell. However, DNA itself may have some immunostimulatory effects. There are motifs in DNA of bacterial origin that do have direct effects on lymphocytes, although these tend to stimulate a Th1 response.

Kaplan: Can you clarify your hypothesis concerning the age-related effects of the immune response? Are you suggesting that young children (< 5 years) are more susceptible to tuberculosis but children aged 5—10 years are less so because of age-related differences in the immune response?

Rook: There are many models in which an exact balance of glucocorticoids and anti-glucocorticoids seem to be essential. This is what is suggested in Table 1 ofmy chapter.

Kaplan: But whatever this balance is that determines whether cytokines are or are not made, which cytokines will give rise to differential susceptibility to tuberculosis in that population?

Rook: Mice that have high levels of corticosterone and are not given any dehydroepiandrosterone (DHEA) die because of a massive proliferation of M. tuberculosis. I suspect that the macrophages are further deactivated and the response is further pushed towards a Th2-type response (Hernandez-Pando et al 1996).

Kaplan: If cytokine production by macrophages is shut down why isn't cytokine production of T cells also shut down?

Rook: In an environment of mixed Th1 and Th2 cytokines, T cells become resistant to glucocorticoids because of a fall in the affinity of their glucocorticoid receptors (Kam et al 1993, Chrousos et al 1996). Moreover, glucocorticoids drive newly recruited T cells towards a Th2 response (Ramirez et al 1996).

I need to clarify our recent cytokine data from tuberculous mice subjected to endocrine manipulations, and the hypothetical relationship between these data and age-related changes in the pathogenesis and incidence of tuberculosis in children. This relationship is summarized in Table 1 of my chapter. Whereas glucocorticoid alone (corticosterone) accelerates death of tuberculous mice by causing bacterial overgrowth, a critical balance of corticosterone to 'anti-glucocorticoid' (DHEA or androstenediol, AED) increases survival and Th1 production (R. Hernandez-Pando & G. A. W. Rook, unpublished results 1997). However, when too much DHEA or AED is present, massive necrosis appears and the animals die rapidly. I am suggesting that these results correspond to three distinct age-related changes in childhood susceptibility to tuberculosis (Donald et al 1996). From birth until the onset of adrenarche at about 5 years, children are susceptible but die without the necrosis and cavitation seen in adult disease. During the onset of adrenarche DHEA levels start to rise, resulting in an intermediate ratio of glucocorticoid to DHEA. Such children, aged 5—10 years, are resistant to the infection. Finally, at puberty DHEA levels become high, and children become susceptible to the adult necrotic and cavitory forms of tuberculosis.

I suspect that this last effect is due to excessive production of tumour necrosis a (TNF-a) in an environment where there is insufficient glucocorticoid to down-regulate TNF-a production or to inhibit tissue damage. It is clear that TNF-a becomes particularly toxic in sites where there is a mixture of Th1 and Th2 cytokines, whether in models of tuberculosis (Hernandez-Pando et al 1997), schistosomiasis (Wynn et al 1995) or leishmaniasis.

Kaplan: What are the levels of cytokines in the final phase, where the incidence of tuberculosis is the highest?

Rook: There are high levels of Th1 cytokines and pro-inflammatory cytokines.

Kaplan: Are you suggesting, therefore, that in the middle phase there is a predominant Th1 response, rather than Th2, and that this is why there is less susceptibility to M. tuberculosis in this age group?

Rook: Yes, but the real question is: what is the actual Th1-dependent effector mechanism that is so successful at this point? And we don't yet know the answer to that.

Kaplan: How do you explain the susceptibility observed in the first phase?

Rook: In young children aged < 5 years with low DHEA levels I suspect that the Th1 response is not able to function sufficiently. We know that there is a deficit of Th1 responsiveness in neonates (Shu et al 1994), but we know little about children between this stage and those aged less than 5 years.

Kaplan: So, in the first phase there's not enough of something, in the middle phase there's a balance between two things, and the good one overrides, and in the final phase there's too much of everything.

Rook: That is not how I would express the hypothesis. Glucocorticoid levels (cortisol in humans) are more or less constant throughout life, but DHEA levels change markedly. Therefore, our working hypothesis is that as far as the effector mechanisms of immunity to tuberculosis are concerned (whatever they may be), there is too little DHEA relative to cortisol in the first 5 years of human life, the optimal balance during the next five years, but too much DHEA relative to cortisol from puberty onwards. This high DHEA/cortisol ratio causes massive production of Th1 cytokines and of TNF and IL-1.

'Kaplan: But these cytokines aren't necessarily regulated together.

Rook: They're not necessarily regulated together, but in this situation they are because 3^-17^-AED increases both Th1 cytokine production, and the proinflammatory cytokines TNF and IL-1 (R. Hernandez-Pando & G. A. W. Rook, unpublished results 1997).

Kaplan: The terminology is being used a little too loosely. When we talk about the Th1 response, I think we all agree we're talking about y-interferon and IL-2. However, when you talk about the Th2 response, are you talking about inflammatory cytokines (i.e. IL-1 and TNF-a) or cytokines that drive antibody production (i.e. IL-4 and IL-5)?

Rook: We are looking at the presence of IL-4-producing cells, and we find them in the peripheral blood of tuberculosis patients. However, we are not suggesting that the IL-4 itself is the problem. We are saying that something which correlates with the presence of IL-4-secreting cells severely compromises Th1-mediated immunity to tuberculosis.

Orme: So why do IL-4 knockouts develop chronic disease just like normal animals? By your thinking one would imagine that mice without the Th2 component would be less susceptible.

Rook: All I can say is that cytokine knockouts often don't do what you expect them to do because the developing immune system adapts. You would need to inactivate the gene in the adult.

Kaplan: There is no evidence that in tuberculosis patients IL-4 is a major component of the in vivo response during infection.

Rook: You have demonstrated the presence of IL-4 mRNA in the peripheral blood T cells of tuberculosis patients (Schauf et al 1993). Tuberculosis patients also have activated Th2 cells, which can be made to make IL-4 in vitro by the addition of TPA (12-O-tetradecanoylphorbol 13-acetate) and calcium ionophore. This response is not observed in normal individuals. Tuberculosis patients also have IgE antibodies for M. tuberculosis (Yong et al 1989). It is a minor Th2 component, but in the mouse a minor Th2 component dramatically affects the ability of the mouse to cope with infection. We don't know the nature of the final effector pathway, but we do know that it is sensitive to the presence ofeven a minor Th2 component (Hernandez-Pando et al 1997, Lindblad et al 1997).

Kaplan: I still don't understand how the inflammatory cytokines fit into your model.

Rook: There are two different systems running in parallel: the Th1/Th2 shift and the regulation of the inflammatory cytokines, which are down-regulated by glucocorticoids. In a Th1 background TNF is good, whereas in the presence of a Th2 component TNF is toxic. I'm not saying that the Th2 response causes the release of the inflammatory cytokines. I'm saying that the presence of the Th2 response causes the change in the role of the inflammatory cytokines. The same situation occurs in the schistosomiasis model, where there's a mixture of Th1/ Th2 cells and TNF. If the Th2 component is removed (Wynn et al 1995) then you no longer observe permanent tissue damage, suggesting that the role of TNF changes in the presence of Th2 cells, although we don't know the nature of that change.

Orme: Let's go back to the classical immunology experiments of 30 years or so ago. Lefford & McGregor (1974) injected a large dose of M. tuberculosis into mice, and waited a few months before administering isoniazid. Isoniazid results in bacterial death and bacterial presentation. We now know that this is accompanied by increased IL-4 levels and decreased y-interferon levels. If the animal is then re-challenged with another dose of M. tuberculosis, it is highly immune. However, it seems to me that if the Th2 response was depressing the Th1 response, then the animals would not be resistant. How can you explain those classical experiments using your model?

Rook: It is not reasonable retrospectively to guess the Th1 or Th2 balance of those mice at the time of challenge.

Blackwell: In leishmanial diseases the condition under which mucocutaneous leishmaniasis is observed is a mixed Th1/Th2 response in the presence of high levels of TNF. We don't know the precise mechanism of that either, but the parallels are interesting.

Fine: I had the impression from your presentation that there is a male/female difference in plasma DHEA levels with age, such that after the 'golden age' there are higher levels in females.

Rook: There have been studies of this (de Peretti & Forest 1978), but I wouldn't take the male/female difference too seriously because DHEA levels are similar in males and females.

Fine: There are two interesting age-dependent sex differences in tuberculosis. One of these is that in most populations the prevalence of tuberculin sensitivity becomes higher among males from adolescence onwards, but paradoxically the risk of tuberculosis disease is higher among females than males at young adult ages. This has always struck me as a bizarre observation, and I wondered if you could tease this out.

Rook: Possibly, but we haven't yet tried.

Orme: Didn't Ray Daynes show that DHEA increases resistance to infection?

Rook: Ray Daynes has not used DHEA in a model of tuberculosis. We are the only group to have done so (see Rook et al 1997). It's a question of the balance with glucocorticoid. You could think of it as being like giving glucagon and insulin together, i.e. the blood sugar levels remain the same but you now have a completely new physiological state. When we give glucocorticoid and AED simultaneously to those animals, we observe a Th1-switching effect of the AED but not the pro-inflammatory effect, and we observe a anti-inflammatory effect of the glucocorticoid but not the Th2-switching effect.

Kaplan: What happens in old age, when individuals become even more susceptible to tuberculosis infection?

Rook: The cortisol levels remain constant and the DHEA levels decrease.

Orme: Do people with tuberculosis have small adrenals?

Rook: That is not well documented. Anecdotally, post mortem examination of patients with severe tuberculosis shows that some have small adrenals.

Bateman: What do the adrenals of mice that die of tuberculosis look like at post mortem?

Rook: Their adrenals increase in size during the early Th1-dominated phase of the disease, but then atrophy to about 50% of their normal weight during the late Th1+Th2 phase (Hernandez-Pando et al 1995).

Bateman: Do we take home the message that your hypothesis doesn't address this in the golden years?

Rook: No. It suggests that during the tuberculosis-resistant years (~ 5—10 years) there is exactly the right ratio of glucocorticoid to AED, which is probably the relevant derivative. The deaths either side of what you call the 'golden years' are for different reasons. In older children and adults it is immunopathology associated with the turning on of everything, and in infants prior to adrenarche it's an immunopathology associated with insufficient Th1 response. The situation in the middle is balanced.

Orme: The Th1/Th2 classification of CD4+ T cell responses is nice in those terms, but they represent two ends of a wide spectrum, and all the rest of the T cell clones are mixtures of both of them.

Rook: I agree.

Kaplan: You are one of the few people in the field who feel that the Th2 hypothesis is still relevant in tuberculosis. Most immunologists in the field of tuberculosis believe that a Th2 response does not determine the outcome in M. tuberculosis infection.

Beyers: Phil Bardin (Department of Internal Medicine, University of Stellenbosch) has done some immunocytochemistry on lung tissue from tuberculosis patients and found not only IL-2, IL-12 and y-interferon, but also IL-4 and IL-10 (personal communication 1997).

Rook: Similarly, Bergeron et al (1997) have shown Th0 cells in tuberculous but not sarcoid lymph nodes. One should take note that much of the work denying a role of a Th2 response in tuberculosis has been performed on either pleural effusion or lymph node tuberculosis, which are high resistance forms of the disease (Barnes et al 1993). Others use peripheral blood and look for cytokine production after adding antigen. In such experiments Th2 cytokine production is swamped by the Th1 response, which down-regulates much of the Th2 activity. It's not difficult to show Th2 cells in the peripheral blood of tuberculosis patients. If you block cytokine secretion completely with monensin or brefeldin-A and then stimulate the cells non-specifically, you find cells that you would not find in normal individuals which make Th2 cytokines.

'Kaplan: But that's because the normal individual doesn't have an ongoing immune response. Infected individuals with active disease have ongoing immune responses — their leukocytes are activated and they express the cytokines and the surface markers that are associated with immune activation. We know antibody is made in tuberculosis patients, but does the Th2 response regulate outcome? And is there a Th1/Th2 balance that will determine whether the patient is going to do well or not? Many immunologists feel that there is no concrete evidence to date to prove that there is a Th2 determinant which regulates outcome. There was a period when the Th1/Th2 balance idea was popular, but those papers have not been confirmed. Indeed, a Th2 response does not appear to be the regulatory determinant for outcome in studies that have been done recently in animals and in humans (Lin et al 1996, Orme & Collins 1994, Kaplan & Freedman 1996). Therefore, is the Th1/ Th2 balance the correct hypothesis on which to model our thinking in terms of what's happening during the immune response to tuberculosis?

Rook: I disagree. In our mouse model the Th2 response is crucial, and as I have already pointed out, if you look in the right way and in the right place, a Th2 component is clearly present in humans with tuberculosis.

Young: I would like to bring some genetics into this discussion. Would genetic polymorphisms of candidate genes associated with the endocrine axis be useful?

Bellamy: Possibly. What would be useful is the identification of polymorphisms that resulted in individuals developing a Th1 or a Th2 predominant response following infection. We have looked at polymorphisms in several cytokine genes believed to be important in the activation ofa Th1 or a Th2 immune response, and we have not found any associations with tuberculosis. However, this does not exclude the possibility that these cytokine genes are important in determining which individuals will succumb to infection.

Rook: Hennebold & Daynes (1997) claim to have identified a polymorphism in the enzyme responsible for inactivating corticosterone to dehydrocorticosterone in the stromal cells of murine lymphoid tissue. They claim that the difference between BALB/c and C57BL/6 mice in their tendencies to drive a Th2 response is attributable to a major difference in the activity of this enzyme. The hypothalamopituitary-adrenal axis is rich in polymorphisms, and it could be a fruitful hunting ground.

Fourie: I would like to turn attention towards multidrug-resistant tuberculosis. Why would there be clinical efficacy in multidrug-resistant patients and not in ordinary tuberculosis patients?

Rook: Because we would not have to compete with an effective therapy. The World Health Organization and the Food and Drug Administration are going to have to look at how to test new therapies if they have to be superimposed upon a therapy which in hospitalized patients is already fully effective. Can one expect the immune system to contribute anything when it is superimposed upon daily therapy with fully effective drugs? It is not the same as trying to treat certain cancers or allergies, for example, where nothing else works. The problem with multidrug-resistant tuberculosis is that one has to tinker with the treatments for each patient, so it's going to be extremely difficult to design a study.

Fourie: Are you proposing immunotherapy with Mycobacterium vaccae as an alternative rather than an adjunct to chemotherapy?

Rook: No. We've been proposing immunotherapy ultimately as an adjunct, but in order to prove clinical efficacy one's going to have to turn to multidrug-resistant disease. In the field tuberculosis is rarely treated as effectively as it is under directly observed therapy short-course (DOTS) and I don't believe that DOTS is ever going to prove to be universally applicable.

Fourie: I expected that your basis for making that statement would be on study design rather than on other observations.

Rook: We do have anecdotal data on about 274 multidrug-resistant patients, although I am not the right person to assess those data. We need to do a proper study.

Kaplan: In principle, this is going to be the case for any so-called weak therapeutic intervention. By weak, I mean compared to the three most effective drugs that clear sputum in about two weeks and sterilize M. tuberculosis infection in about three to four weeks. Any weak intervention is going to be masked by something extremely powerful. The problem is how do you demonstrate that it is contributing to the response? You take a situation where the three most effective drugs don't work, i.e. multidrug-resistant tuberculosis patients, and you look at whether the interventions have an effect. It is not ethical to test any weak interventions in the absence of these powerful drugs, although this can be done in animal models.

Duncan: This is a critical issue, and I will be covering some of it in my presentation. When you start to think about any new therapy, i.e. one that differs from today's bactericidal agents, you have to face up to the ethical issues, otherwise these therapies will never go into clinical trials. We have to get over the hurdle of being able to design clinical studies in a way that will show efficacy. If you do these studies in multidrug-resistant patients, you have to face other issues; for instance, the patients may have severe side-effects from the second-line drugs they are taking, thus masking the benefit of a new agent.

Kaplan: In the general population there will always be individuals who for some reason do not respond, and those are the individuals who need the additional interventions. The problem is that this is a difficult situation to test because of inherent heterogeneity, i.e. it is not possible to test everyone with the same drugs under the same conditions because each patient has his/her own individual history ofresistance and sensitivity to the drugs. This has to be accommodated in the study design.

Duncan: This is why these experiments should not be performed in patients with multidrug-resistant infections, rather against purely sensitive organisms.

Kaplan: But you can't ethically do this if you already have the best drugs available.

Duncan: Why not? It's possible as long as you have strong animal data which suggest that your therapy is going to work.

Kaplan: But your therapy will never be as good as the best drugs already available.

Duncan: So there's no point in even testing it.

Kaplan: Yes, there is, because there may come a time when the best available drugs can no longer be used; for example, if the patients become multidrug resistant.

Rook: I can't accept that one would not want to consider a weaker therapy under field conditions, even in patients taking effective chemotherapy. The actual cure rate of tuberculosis in South Africa at the moment is only about 50%, which is way below what we ought to be able to achieve. We need additional therapies that provide cover for patients who default or who have drug-resistant disease.

Colston: This isn't a problem that's unique to tuberculosis. For example, leprosy drugs are highly effective and the prevalence rates have decreased dramatically over the last 10 years, and yet a completely new regimen has just been introduced using none of those existing drugs. Therefore, it is possible to introduce a new regimen that does not use any of the first-line drugs. The way this is done, is that you first have to have strong experimental evidence of an effect, and then you have to design a short-term trial, which may not be therapeutically relevant but which will show an effect.

Kaplan: But in leprosy disease progression is so slow that it takes a month for the final diagnosis before drug therapy is initiated. This gives you a month during which new interventions can be tested. You can't wait that long in patients with tuberculosis; you have to use the best therapy available as soon as you suspect tuberculosis.

Fine: With reference to M. vaccae, the evidence from animal models is not, in my opinion, as strong as we would like. However, I just wanted to point out that a much larger formal randomized control trial of M. vaccae immunotherapy is now being undertaken by ourselves in northern Malawi, in collaboration with Alwyn Muringa at the University Teaching Hospital in Lusaka, Zambia. This trial involves more than 1200 patients, and is being conducted in areas where compliance is low.

Bateman: A study of M. vaccae immunotherapy has just been completed in Natal. To date, only sputum conversion at eight weeks has been analysed, and further analyses are awaited. If, however, this study were to show that in those who are non-compliant with chemotherapy, immunotherapy was associated with a better outcome, this would be another form of efficacy. The trouble with the multidrug-resistant group, which in the Western Cape amounts to over 4000 cases, is that the early conversion rate in the compliant patients is in excess of 80% (although sustained cure is much lower). Therefore, even in these cases it will be difficult to show the contribution, if any, of M. vaccae. In addition, one would have to demonstrate that M. vaccae, or any other immunotherapy, has an effect during the follow-up; that is, the first two years after the initial six-month treatment period. There is no substitute for doing classic intervention trials involving single and multiple doses for dose ranging, add-on trials and then substitution studies. A major problem when testing treatments for tuberculosis is that the endpoints are longer and must include sputum conversion, consolidation and relapse.

'Duncan: Yes, and that makes them totally unrealistic to do. If each trial takes two years, then that's a total of six years, and no company will support this.

Anderson: Why are the ethics different from the HIV trials? These trials are running beyond two years, it's a chronic disease and the outcome is uncertain. There are three drugs that work well, and yet trials are going on at the moment to replace one of those drugs with new protease inhibitors as new products are developed.

Duncan: There are a number of issues. One is that in contrast to tuberculosis, AIDS is a disease for which there is a clear, high value market. The HIV trial you describe is a replacement trial, where new protease inhibitors are being tested that have better properties than those already available. I don't have a problem with a clinical trial that, for example, tests an isoniazid replacement, but the approach to testing something that may stimulate the immune system is completely novel and difficult because you can't compete with the best therapy.

Kaplan: There is another difference in that the HIV drugs do not cure HIV, whereas under optimal conditions the tuberculosis drugs cure tuberculosis. At the moment the three HIV drugs work together effectively, but sooner or later drug resistance is likely to develop, so we are going to need replacements. This gives us the ethical justification to design and test replacements.

"Donald: I would like to give my support to the study of multidrug-resistant patients. They are a valuable resource for evaluating new therapies, and the outcomes can be more easily measured because they have a higher mortality and the conversion to sputum culture negativity is slower. What has never been accurately assessed is whether effective therapy will give earlier culture negativity and whether the speed with which culture negativity is achieved will correlate with ultimate sterilization. In addition, the early bactericidal activity of the drugs can be easily measured within a period of 48 to 72 h, so it will tell you quickly how fast your drug is killing the bacteria, although this doesn't help your problem with immunotherapy. Therefore, there is a fairly clear path towards evaluating new drugs.

Duncan: There's a clear path to evaluating new isoniazid replacement drugs, but anything else poses more of a problem. It comes back to the point of taking too long to conduct these trials.

Donald: Yes, I acknowledge the commercial imperative in these trials.

Was this article helpful?

0 0

Post a comment