Is there a tumour necrosis factor (TNF)-dependent cytokine cascade?

The plethora of proinflammatory cytokines present in rheumatoid arthritis (RA) synovium from patients with active disease, including interleukin (IL)1, IL6, TNF, granulocyte macrophage colony-stimulating factor (GM-CSF), interferon α,1 deterred many groups in the late 1980s/early 1990s from pursuing proinflammatory cytokines as a therapeutic target. This, despite their obvious potential as they are rate-limiting steps closely linked to symptoms and signs. The dilemma was which, if any, might be an effective target. Could blocking one out of a dozen proinflammatory cytokines make a clinically meaningful difference? So most of the groups involved in the cataloguing of cytokine expression dropped out of the “race” to potentially identify new anti-cytokine treatments.

Fionula Brennan performed the key experiment in 1989, blocking with antibodies the spontaneously produced cytokines generated in cultures of rheumatoid synovium.2 This showed that blockade of TNFα downregulated IL1, and subsequently, GM-CSF, IL6, IL83 4 and all other proinflammatory cytokines tested, half of the chemokines and many other active molecules such as matrix metalloproteinases. These results focused our attention on TNFα as a therapeutic target, and therapeutic, post-onset TNF blockade succeeded in mice with collagen-induced arthritis,5 providing a rationale for the subsequent successful trials and clinical development of anti-TNF therapy.6 7 8 9

Studies of the mechanism of action using serum samples from patients in clinical trials showed that after anti-TNF therapy there was a dramatic and rapid diminution of serum IL6 and many other cytokines in the blood and joints,10 establishing that the synovial culture-based concept was replicated in vivo. But what is a cytokine cascade? Cytokines act on receptors on cells, the cells then respond according to their other environmental stimuli, structure and genetic programme. So it is of interest that the response to TNF blockade is so fast in vivo, both at the clinical and serum cytokine level. It confirms the rate-limiting nature of cytokines such as TNF. But is TNF really special? Is it really the pivotal cytokine in RA? Of course it is too early to tell. The rapid kinetics of TNF production, which is released faster than other proinflammatory cytokines and stored preformed on the cell surface and in mast cell granules, suggest that TNF has a different role from cytokines such as IL1β, and especially IL1α, which have great difficulty escaping the producing cell. The rapid onset of symptomatic benefit, relief from tiredness within hours is also in keeping with a pivotal role. Of the other major proinflammatory cytokines, it is not well understood why IL1 blockade, which is so good in mice is not as good clinically, whereas IL6 blockade is as effective clinically,11 but the rate of onset of clinical benefit is much slower.

Perhaps the best descriptive concept is of a cytokine cascade and networks. But how does a TNF-dependent cytokine cascade translate into clinical benefit? The reduction in cytokine levels in blood, and their access to the brain probably explains the reduction in fatigue and the improvement in mood. The reduction in TNF locally normalises the pain thresholds. But probably a dominant integrator is by the reduction of leukocyte trafficking to the joints. This is mediated by reduction in both chemokine expression and adhesion molecules. Reduction in leukocyte trafficking has been documented clinically.12

While we can treat late disease, earlier treatment is more clinically effective

The patients in the first anti-TNF trials, for whom long-term treatment had failed, all responded well, although variably. The variability is a feature but subsequent studies showed that there is not a bimodal distribution of responders and non-responders.13

Initially it was not known how well patients with well-established disease would respond. But they do respond, and a striking feature is that with anti-TNF, especially in the presence of methotrexate, joint protection is excellent. When infliximab or etanercept or adalimumab, is used together with methotrexate, joint protection is almost complete.14 15 In contrast, patients who respond well clinically to methotrexate still continue to progressively damage their joints. Even clinical non-responders to anti-TNF, as judged by failure to reach ACR20, Health Assessment Questionnaire change or C-reactive protein change, nevertheless have protected joints, indicating a poorly understood dissociation.16 Smolen has postulated that this is due to a threshold effect. It is clear that after anti-TNF therapy, levels of matrix metalloproteinases are reduced but levels of osteoprotegerin increase; exactly how the disease process drives joint destruction is not fully understood, and is worthy of further study. There are other aspects of the disease which need further attention. Too few patients reach remission. Patient surveys have shown that pain is still a major clinical problem; there is an increased awareness that both non-steroidal anti-inflammatory drugs and cyclo-oxygenase-2 inhibitors have major toxicities.

New approaches to treatment of RA are needed

There are several drawbacks to anti-TNF therapy. A major one is the cost, about £10 000 a year, as it leads to rationing in most societies. Second, is the degree of efficacy, most patients improve, but few reach remission since treatment is initiated too late. Third, is the risk of side effects, with infection being the most obvious. How might these drawbacks be overcome?

There are many opinions. Might the cost of anti-TNF therapy be reduced with more product competition? It is not yet obvious, but may occur with “biogenerics”, when, or if, these ever appear. Currently, the pharmaceutical industry believes that the public perceives cheaper drugs as less good, and that Adam Smith’s famous bell-shaped curve does not apply.

Orally available drugs may provide an opportunity to reduce the costs while delivering most of the benefit. But the pharmaceutical industry’s biggest research investment, on p38 MAP kinase inhibitors, has failed so far.17

Other promising approaches are a drug which mimics the oxidised tryptophan products of the indoleamine dioxygenase pathway, and inhibitors of Toll-like receptor. Inhibitors of angiogenesis would also be of considerable value as angiogenesis is important in sustaining a chronic inflammatory mass or infiltrate. Dr Paleolog has studied the effects of blocking vascular endothelial growth factor (VEGF) in a variety of ways in animal models of arthritis. The results have been highly consistent and surprisingly effective. It might be thought that blocking angiogenesis would have a relatively slow onset of benefit, as the inflammatory tissue mass cannot expand and shrinks, but VEGF blockade is rapidly effective in mice, suggesting that VEGF has broader effects on inflammation. But it is not just VEGF blockade that is effective, K1-5, an angiostatin derivative, acts directly in endothelium and is also very effective.18 19

It seems from basic principles and has been confirmed by clinical trials that excess blockade of the immune/inflammatory pathways by blocking TNF and IL1 (eg, using anakinra and etanercept),20 or TNF and antigen presenting/immune blockade by CTLA4-Ig,21 is proinfective. However, blocking angiogenesis and TNF should not be synergistically proinfective, and this combination therapy awaits clinical evaluation.

Mellor and Munn reported that the tryptophan pathway is important in immune regulation.22 It has been characterised and the enzyme indoleamine 2, 3-dioxygenase identified as an essential component. While reduced levels of tryptophan, an essential amino acid, have some impact, it is the oxidised products, generically known as kynurenines, that appear to mediate most of the immunoregulatory effect.

Michael Selley, an Australian pharmacologist, predicted that 3-dimethoxycinnamonyl anthranilic acid, also known as tranilast, a drug sold in Japan and Korea for over 20 years for allergy, should be a potent anti-inflammatory agent. This concept was tested by Platten et al in experimental autoimmune encephalitis models23 and by Williams’ group for collagen-induced arthritis.24 The results were consistent, good clinical effects were associated with downregulation of proinflammatory cytokines such as interferon γ, IL12, TNF, as with effects on T- and B-cell proliferation. This drug is entering phase II trials in RA. The results may uncover an effective antirheumatic drug, and will help to evaluate the importance of the indoleamine 2,3-dioxygenase pathway in human immune regulation.

Another approach to cytokine blockade is to target the cells producing the proinflammatory cytokines. Macrophages are the cells considered to be producing the majority of proinflammatory cytokines in rheumatoid synovium, although many other cell types take part, such as T lymphocytes, NK cells, endothelial cells. B lymphocytes may also be important cytokine producers. This possibility is suggested by the observation that before autoantibody levels are much diminished anti-CD20 therapy is efficacious.

The CD200 receptor is an inhibitory receptor expressed on macrophages, and was discovered by Barclay and colleagues. The effect of triggering this receptor by a CD200 Fc fusion protein was evaluated. The results were highly encouraging. The potency of CD200Fc in CIA is similar to etanercept, and the benefit is linked to reducing macrophage-derived proinflammatory cytokines such as TNF, IL1 and IL6.25 This confirms the concept that macrophages are a major source of these cytokines. We are unaware of any clinical development programme for CD200Fc. It may be that the cost–benefit of biological agents for RA at present with many marketed anti-TNF agents, CTLA4-Ig, anti-CD20, anti-IL6 receptor is a commercial challenge, but it would be of interest to compare the inhibitor of a cell type and a cluster of proinflammatory cytokines with inhibition of TNF or IL6.

Might cytokine analysis be used to target unmet medical needs? Investigating postoperative cognitive dysfunction

Cytokines are extracellular molecules, and so can easily be targeted by biological agents (monoclonal antibodies, receptors, etc) which are highly specific. This is an advantage as blocking cytokines with biological agents would have side effects linked to the mechanism of action. This is unlike small organic molecules which typically have both mechanism-linked side effects and unpredictable side effects due to cross reactions. The rate-limiting nature of cytokines is also good for therapeutic targets. So it would be interesting to determine whether cytokine analysis might be used for investigating unmet medical needs. Since cytokines are involved in every key biological process, from inflammation to immunity, migration, fibrosis, angiogenesis, repair and so on, all diseases are likely to have cytokine abnormalities. The postulate is that these may lead the way to new therapeutic approaches.

Maze, Nanchahal, Cibelli and others are examining the clinical problem of postoperative cognitive dysfunction. This is an important common complication (>10%) of all surgery in older (>60%) people, and younger ones with coronary artery bypass graft. Unlike studies in RA or atherosclerosis, it is not possible to analyse the cytokines expressed at relevant sites in humans. So an animal model was developed in which the tibia was operated on and pinned under anaesthesia. With appropriate cognitive testing a transient functional impairment was found even in young mice (Cibelli et al, unpublished data).

Analysis showed modest cytokine upregulation in the blood with a rise in IL1 and IL6, but in the hippocampus, the area of the brain involved in memory, high levels of IL1 were detected, by PCR or by western blotting. This result suggests that IL1 may be of relevance in postoperative cognitive dysfunction, and blockade of IL1 in these mice supports that hypothesis.

Do these data translate into human treatment? Clearly, clinical trials are worth considering, the benefit–cost–risk evaluation appears to be favourable, a short-term relatively cheap intervention may prevent considerable morbidity.

Conclusion

The analysis of cytokine expression in the joints of patients with RA, starting 25 years ago, as the cytokine system was being molecularly defined, has led to progress in science and medicine. We now appreciate the importance of the cytokine system as a powerful and rate limiting signal. Several cytokine pathways have led to production of useful drugs. Based on this concept, more work in this field would be fruitful.