Article Text

Download PDFPDF

Therapeutic inhibition of tumour necrosis factor α in patients with heart failure: cooling an inflamed heart
  1. P A Henriksen,
  2. D E Newby
  1. Department of Cardiology, University of Edinburgh, Royal Infirmary, Lauriston Place, Edinburgh, UK
  1. Correspondence to:
    Dr Peter Henriksen, Department of Cardiology, Royal Infirmary, Lauriston Place, Edinburgh EH3 9YW, UK;


Heart failure is a systemic disorder characterised by tissue hypoxia and secondary organ dysfunction which occurs in response to various myocardial insults that include ischaemia, viral infections, and toxins. In addition to maladaptive neurohumoral activation, heart failure is associated with an inflammatory state that appears to have a detrimental effect on cardiac function and prognosis. This has led to the suggestion that anti-inflammatory interventions may have therapeutic potential in the symptomatic and prognostic treatment of patients with heart failure. This review considers the role of inhibition of the cytokine tumour necrosis factor α in the treatment of heart failure.

  • tumour necrosis factor α
  • heart failure
  • ATTACH, anti-TNF alpha therapy against chronic heart failure
  • IL-1β, interleukin 1β
  • iNOS, inducible nitric oxide synthase
  • NF-κB, nuclear factor κB
  • RENAISSANCE, ramdomized etanercept North American strategy to study antagonism of cytokines
  • RECOVER, research into etanercept antagonism in ventricular dysfunction
  • TNFα, tumour necrosis factor α
  • TNFR, tumour necrosis factor α receptor

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Tumour necrosis factor α (TNFα) is a 157 amino acid cytokine that is produced in response to inflammatory or infectious stimuli by macrophages, lymphocytes, neutrophils, and structural cells including fibroblasts and smooth muscle cells. It is one of many immunological mediators involved in critical aspects of cell function including the response to injury, inflammation, and death through necrosis or apoptosis. We describe the rationale for, and results of, recent clinical trials of TNFα blockade in patients with chronic heart failure.


TNFα acts through two transmembrane receptors, TNFα receptor (TNFR) types 1 and 2. When stimulated, these receptors interact with a variety of cell signal pathways increasing expression of proinflammatory cytokines, such as interleukin 6, and regulating cell survival, growth, and apoptotic death. However, the response to TNFα is non-uniform and critically depends upon the cell type, the state of cellular activation, and the local concentrations of the cytokine (table 1). For example, TNFα may promote cell survival and proinflammatory gene expression, and yet in the presence of transcription factor blockade the same stimulus may drive the cell towards apoptosis and death. This contextual dependence of TNFα action is not only restricted to in vitro observations. Infusion of TNFα in healthy humans produces vasoconstriction1 and procoagulant effects,2 while in volunteers injected with endotoxin, high TNFα concentrations are associated with hypotension, increased vascular permeability, and myocardial depression.3 Thus the environment within which the TNFα signal is received is crucial and may explain some of the conflicting evidence regarding its actions on cardiomyocyte function and survival.

Table 1

Contrasting actions of tumour necrosis factor α in patients with heart failure


Cytokines have been implicated in the syndrome of chronic heart failure for over a decade.4 Levine and colleagues5 first described raised plasma TNFα concentrations in non-septic patients with chronic heart failure. Subsequent studies showed comparable elevations of two other cytokines, interleukin 6 and interleukin 1β (IL-1β),6 and a direct correlation between plasma TNFα concentrations and the degree of heart failure, as assessed by functional heart failure classification7 (fig 1). The increase in plasma TNFα concentrations appears to be most striking in those patients with generalised body wasting or cardiac cachexia—a clinical feature that is particularly associated with a poor prognosis.8,9 However, plasma TNFα concentrations are not necessarily representative of the functional activity of the cytokine. Localised myocardial production of TNFα has been demonstrated within the failing heart, suggesting that there may be significant additional autocrine or paracrine actions.10

Figure 1

Relative (percentage) plasma tumour necrosis factor α concentrations according to New York Heart Association (NYHA) classification of symptoms (left panel; data from Torre-Amione and colleagues10 ) and the presence of cachexia (right panel; data from Anker and colleagues9 ) in patients with heart failure.

What drives TNFα production in patients with heart failure?

Inflammation contributes to the development and progression of heart failure,11 and markers of inflammation, such as C reactive protein, are independent predictors of the presence of heart failure.12 TNFα production represents a phylogenetically conserved response to cellular stress, inflammation, and injury. The specific stimuli responsible for increased TNFα production in patients with heart failure have not been elucidated but there are several potential candidates.

Endotoxin, a component of the outer coat of Gram negative bacteria, is a potent stimulus for macrophage TNFα production and it is intriguing that non-septic patients with decompensated failure and oedema are more likely to have subclinical endotoxaemia.13 Indeed, resolution of oedema with diuretic treatment is associated with the disappearance of endotoxin, leading to a proposal that oedematous bowel is more permeable to bacteria.

Plasma markers of oxidative stress are increased in patients with heart failure.14,15 The tissue origins of these markers have not been elucidated, although poorly perfused peripheral muscles and the failing myocardium are likely candidates. TNFα and angiotensin II stimulate free radical production in a variety of cell types16 and contribute to oxidative stress in patients with heart failure. The reactive oxygen species produced may exert their effect through the nuclear factor κB (NF-κB) system, a ubiquitous transcription factor responsive to oxidative stress.17 Active NF-κB is present within failing cardiomyocytes, and its activity correlates inversely with the improvement associated with a left ventricular assist device.18 It interacts with the regulatory elements of many genes concerned with response to injury, including TNFα itself and inducible nitric oxide synthase (iNOS). The NF-κB transcription factor system may be one point of molecular convergence for diverse pathologies—including ischaemia, haemodynamic overload,19 and viral infection—that result in increased TNFα expression in patients with cardiac failure.


Myocyte contractility

TNFα impairs myocyte contractility. It has been shown to have negative inotropic activity measured both by left ventricular pressure–volume curves and isolated cardiac myocyte fractional shortening.20 This results from a decrease in the voltage dependent calcium current responsible for depolarisation of the myocyte and the release of sarcoplasmic reticular calcium stores. Although uncoupling of β adrenoceptors from their intracellular signalling pathways may have a role,21 this effect appears to be mediated, at least in part, through mechanisms that involve increased nitric oxide synthesis.

Nitric oxide has negative inotropic effects that include the inhibition of creatine kinase22 and the downregulation of mitochondrial respiration.23 TNFα increases the expression and synthesis of iNOS that has been described in the myocardium of patients with heart failure.24,25 Nitric oxide synthase inhibition is able to reverse the TNFα mediated reductions in isolated ventricular papillary muscle tension.26 However, these findings are not universal, and Mann’s group failed to demonstrate reversal of myocardial depression with a nitric oxide synthase inhibitor following TNFα infusions in rats.27

Myocardial fibrosis and apoptosis

Transgenic mice which selectively overexpress TNFα in cardiomyocytes have a phenotype of myocarditis with interstitial infiltrates, four chamber dilatation, and compensatory myocyte hypertrophy.28 There are striking parallels between these transgenic murine TNFα hearts and the myocardium of patients with chronic heart failure. At the cellular level, transgenic hearts have patterns of gene expression that are particular to failing myocardium. These include reactivation of the fetal gene programme within the ventricle (for example, atrial natriuretic factor) and an increase in metalloprotease activity that facilitates extracellular matrix remodelling. Interstitial cell (and to a lesser extent myocyte) apoptosis is also a consistent feature of the transgenic myocardium.29

The transgenic models may represent one extreme of sustained or excessive production of TNFα. The generation of TNFα is initially compensatory and sufficient to promote the hypertrophic growth of cardiac myocytes and to induce the expression of cytoprotective antiapoptotic factors in a variety of cell types.30 This protective role is borne out by the observation that TNFR-1/TNFR-2 deficient mice have an increased susceptibility to cardiomyocyte apoptosis following myocardial infarction.31 These studies illustrate the potentially important and contrasting roles of TNFα as a negative inotrope following prolonged expression in the failing myocardium, while paradoxically acting as a protective agent limiting myocyte death during acute ischaemia and infarction.



If cytokine production represents a maladaptive response to cellular injury leading to further organ dysfunction (compare the neurohormonal activation occurring in heart failure), then inhibition of TNFα may be beneficial. Several lines of evidence suggest this may be the case.

Animal models

Anti-TNFα treatments were first assessed in animal models of septicaemia. Infusion of recombinant TNFR-2 fragments protected rabbits from bacterial endotoxin induced septic shock. This was associated with an improvement in cardiac output and arterial pressure, indicating that TNFα blockade was effective in reversing the depressed myocardial contractility and vascular tone associated with this condition.32 Subsequent studies in the TNFα transgenic mouse have shown that neutralising fragments of the TNFα receptor attenuate certain features of the heart failure phenotype. Interstitial infiltrates were reduced and matrix metalloprotease concentrations returned to normal, although hypertrophy and fibrosis persisted.33 Both these septic and transgenic animal models are characterised by myocardial depression in the context of extreme TNFα production. Little work has been done on TNFα blockade in heart failure models more relevant to human disease, although one group has shown an improvement in cardiac function following ischaemia-reperfusion injury in rabbit hearts treated with neutralising TNFα antibody.34

Clinical observational studies

A progressive increase in plasma TNFα with increasing severity of decompensated heart failure and a fall with therapeutic intervention provides indirect evidence that strategies to reduce TNFα signalling may be beneficial. Two further observations lend support to this view. High intracellular concentrations of adenosine would be expected to reduce TNFα production, and a mutation in the adenosine monophosphate deaminase-1 gene, which slows adenosine metabolism, is associated with improved survival following symptom onset in patients with heart failure.35 Second, phosphodiesterase inhibitors can also serve as inhibitors of TNFα production, and the phosphodiesterase inhibitor pentoxyfilline significantly improved symptoms and left ventricular ejection fraction in patients with idiopathic dilated cardiomyopathy over a period of six months.36 However, there are several potential alternative explanations for these clinical observations and it is unclear whether any of these effects are mediated through TNFα. Moreover, all previous clinical experience with the prolonged use of phosphodiesterase inhibitors, such as milrinone, suggests that their positive inotropic activity worsens long term survival.

Therapeutic agents

The favourable results of anticytokine treatment in other inflammatory diseases37 have paved the way for trials of TNFα blockade in patients with heart failure. Two antagonists that bind and neutralise circulating TNFα are currently licensed and have been successfully used in the treatment of rheumatoid arthritis and inflammatory bowel disease. Etanercept is a fusion protein that employs a TNFR-2 fragment to recognise circulating TNFα. It is effective in reducing markers of acute inflammation, improving functional scores, and, importantly, slowing progression of joint damage in patients with severe rheumatoid arthritis in comparison with methotrexate.38 The treatment remains effective and well tolerated during long term administration and follow up.

Another strategy to inhibit cytokine signalling is the use of neutralising monoclonal antibodies, although potential drawbacks include direct activation of the complement system leading to cell lysis and loss of efficacy with the development of host antibodies. Analogous to abciximab, infliximab is a monoclonal antibody that has been raised in mice with neutralising activity against human TNFα. In order to reduce the immunogenicity of this mouse protein, a chimeric construct has been produced with segments of the protein replaced with human counterparts. This treatment has proven efficacy in patients with rheumatoid arthritis39 and inflammatory bowel disease,37 and early clinical trials suggest it may have a therapeutic role in patients with severe psoriasis.40

As the use of etanercept and infliximab has increased, so have reports of undesirable side effects relating to immunosuppression, including invasive pulmonary aspergillosis,41 aseptic meningitis,42 and in particular the reactivation of tuberculosis with infliximab. Patients should be screened for latent tuberculosis infection before receiving infliximab, and the occurrence of rare but serious side effects relating to immunosuppression may limit the widespread applicability of both these treatments to heart failure populations.

Early clinical experience in heart failure

Early trials of TNFα blockade in patients with heart failure were encouraging. A pilot double blind randomised controlled trial of etanercept, given as a single intravenous infusion, led to significant increases in quality of life scores, six minute walk distance, and ejection fraction in 18 patients with moderately severe heart failure.43 Positive effects were only seen in higher dose groups and a follow up study examined the effects of prolonged antagonism with twice weekly subcutaneous doses given for three months. The treatment was well tolerated and again significant dose dependent improvements in functional status and ventricular function were reported, although the numbers were small (n = 47).44 Etanercept also appears to correct the impaired endothelium dependent vasodilatation in patients with heart failure, an effect that was proportional to the plasma TNFα concentrations.45 However, this initial promise was not borne out by two larger trials (RECOVER and RENAISSANCE) that randomised over 1500 patients with NYHA grade II–IV heart failure in Europe and North America, respectively. Last year, both trials were stopped early because of the apparent futility of the treatment on an interim analysis, with little likelihood of any benefit in terms of symptoms, hospital admissions, or deaths. The complete findings of these studies have yet to be published (table 2).

Table 2

Recent trials of anti-tumour necrosis factor α treatment in patients with heart failure

There has been only one small scale phase II trial of infliximab in patients with heart failure. The ATTACH trial recruited 150 patients with NYHA grade III–IV heart failure who were randomised to placebo, low dose infliximab, and high dose infliximab. This trial was also terminated early owing to an apparent excess of hospital admissions and deaths, particularly in patients receiving the higher dose.


The evidence implicating TNFα and other cytokines in the pathogenesis of heart failure is compelling. Clinical trials may have been considered premature, given that the mechanisms through which TNFα blockade could be beneficial are unknown. The potential adverse effects of this novel treatment are of concern, and the cytoprotective properties of TNFα may be more important to certain categories of patients with heart failure who may deteriorate following TNFα blockade. The cost of long term administration and the possibility of side effects from immunosuppression and hypersensitivity to these agents are further problems that need to be addressed.

Clinical trials of TNFα blockade have failed to produce the expected benefits in a large non-selective population of patients with stable heart failure. Selected patient populations with certain TNFα promoter polymorphisms may be predisposed to TNFα induced ventricular dysfunction. Future anti-TNFα trials may be able to target patients with susceptible genotypes, and a recent study demonstrated overrepresentation of the TNFα2 allele—a promoter polymorphism associated with increased inducible transcription of TNFα—in patients with non-ischaemic left ventricular dysfunction.46

Generation of inflammatory mediators is a secondary event in most cases of heart failure, and the balance of maladaptive versus adaptive effects may change following disease progression or during acute exacerbations. It may be that differences in the aetiology of heart failure are important, and future research should focus on dissecting out the critical pathways mediating ventricular dysfunction and determining which patients may benefit most. It should also be remembered that the inflammatory response to heart failure is associated with the release of many cytokines and not just TNFα. For example, IL-1β is a cytokine with similar signalling properties to TNFα, is produced by cardiomyocytes, and is also increased in heart failure. The deleterious actions of IL-1β will not be ameliorated by anti-TNFα treatment and this may contribute to treatment failure.47 Alternative anti-inflammatory or anticytokine treatments, used alone or in combination, now need to be explored.


PAH is supported by a Wellcome Trust clinical training fellowship at the University of Edinburgh. DEN is conducting work in this area supported by the British Heart Foundation (PG/2001/068).


Linked Articles

  • Miscellanea
    BMJ Publishing Group Ltd and British Cardiovascular Society