Article Text


Clinical implications of tumour necrosis factor α antagonism in patients with congestive heart failure
  1. Guillermo Torre-Amione,
  2. Sonny S Stetson,
  3. John A Farmer
  1. Department of Medicine, Section of Cardiology, Winters Center for Heart Failure Research, USA
  1. Dr G Torre-Amione, Baylor College of Medicine, Section of Cardiology, 6550 Fannin, MS: SM1901, Houston, TX 77030, USA.

Statistics from

Tumour necrosis factor (TNF) α is a pro-inflammatory cytokine that is produced by the heart under certain forms of stress. Patients with advanced heart failure have increased levels of circulating TNFα1 but more importantly, TNFα can produce left ventricular remodelling, pulmonary oedema and cardiomyopathy in human subjects.2 3 Therefore, increased production of TNFα may play a contributory part in the pathogenesis of heart failure.

The evidence supporting the importance of TNFα in heart failure stems from the following observations: Firstly, increased levels of circulating TNFα occur on patients with Class III-IV heart failure.4 5 Secondly, in a sub-study of the Studies On Left Ventricular Dysfunction (SOLVD) we showed that patients who became symptomatic with respect to heart failure had progressively higher circulating TNFα levels.5 Thirdly, TNFα is elaborated in the failing heart in response to pressure or volume overload, physiological stimuli for TNFα production in heart failure patients.6 And finally, strong experimental support comes from the observations that transgenic mice over-expressing TNFα in cardiac tissue develop dilated cardiomyopathy and die prematurely. In this review, we will discuss the human and experimental evidence that supports a causative role of TNFα in heart failure and discuss the clinical implications of TNFα antagonism in patients with congestive heart failure.

TNFα in cardiac function

Two important clinical observations led to the initial investigations of the role of TNFα in cardiac function. Firstly, as TNFα was a mediator of cachexia and patients with advanced heart failure manifest profound metabolic abnormalities and the syndrome of cachexia, it was reasonable to suggest that TNFα may be increased in patients with advanced heart failure. Indeed, the initial report on patients with advanced heart failure demonstrated that approximately 30–40% of patients with advanced heart failure have increased circulating TNFα levels.1 Secondly, it has long been recognised that patients with advanced sepsis, a condition characterised by increased cytokine production and in particular TNFα, may develop reversible cardiac dysfunction.7Accordingly, the hypothesis was put forth that TNFα may contribute to contractile abnormalities of failing myocardium.

After those initial clinical observations, a number of experimental reports demonstrated that TNFα is produced by myocytes8 and that it is capable of inducing negative inotropic effects in vivo and in vitro.9-11 But more interestingly, transgenic mice that over-express TNFα develop hypertrophy, cardiomyopathy and die prematurely consistent with the hypothesis that increased cardiac concentrations of TNFα are sufficient to induce the failing phenotype. Clearly, the experimental observations of the effects of TNFα on cardiac function indicate that TNFα may be an important mediator of cardiac injury in heart failure patients.

Regulation of cardiac TNFα production


The initial experimental reports demonstrated that TNFα can be produced by isolated myocytes or the intact heart when perfused with lipopolysacharide (LPS), a well known stimuli for TNFα production12 The ability of LPS to induce TNFα production in the heart is important because it establishes the ability of myocardium to produce TNFα; however, if TNFα is clinically important haemodynamic stimuli like pressure or volume overload should be capable of stimulating TNFα production. Indeed, isolated feline hearts perfused in a Langendorff apparatus produced TNFα in response to either pressure or volume overload.6 These data, taken together demonstrated that adult myocardium can produce TNFα in vivo and provided the basis to support the concept that haemodynamic loading conditions may lead to production of myocardial TNFα in heart failure patients.


Perhaps the most important observation that supports the hypothesis that over-production of TNFα is detrimental to the heart was the demonstration that TNFα is present in failing but not in non-failing myocardium. Figure 1 shows intracardaic TNF levels in normal (n=4) and failing human heart (n=4). As clearly shown, TNFα expression is absent in the normal but abundant in the failing heart. The importance of demonstrating increased intracardiac TNFα levels in failing human myocardium is also emphasised by the recognition that TNFα found in the heart was not the result of passive transfer of TNFα from the serum but rather local production. This statement is supported by the two following observations: Firstly, in patients with advanced heart failure serum TNFα levels do not correlate with myocardial TNFα levels and patients with high myocardial TNFα content may have low peripheral TNFα levels.13 Secondly, in transgenic mice that overexpress TNFα, there is no correlation between the failing phenotype and serum TNFα levels. In fact, some transgenic mice that over-express myocardial TNFα develop a cardiomyopathy and have undetectable peripheral TNFα.14It follows from these observations that it is the presence of TNFα within the heart that induces cardiac dysfunction and not TNFα that circulates in the periphery. Indeed, it is quite possible that peripheral TNFα found in heart failure patients is not produced within the myocardium but represents extracardiac production. It thus seems; that intracardiac TNFα in failing myocardium is produced in response to haemodyanmic loading conditions and that it is myocardial and not peripheral TNFα that is responsible for cardiac dysfunction.

Figure 1

Expression of TNFα in failing myocardium. Cardiac TNFα concentrations were determined by measuring TNFα immunostaining in normal and failing myocardium. Data are presented in arbitrary pixel units as mean (SEM).

TNF receptors in normal and failing myocardium

Despite the extensive clinical observations and associations between TNFα levels in patients with congestive heart failure (CHF), it was not known whether TNF receptors were present in human myocardium. To further support a causal role for TNFα in heart failure, it was imperative to demonstrate that myocardial cells express TNFα receptors because most biological effects of TNFα are mediated by protein receptor interaction.

There are two forms of TNF specific receptors: TNF-R1 and TNF-R2. Both receptors are present in equal proportions in normal myocardium however; the biological effects of each receptor are distinct. In fact, while TNFα binds with similar affinity to both receptors, the negative inotropic effects of TNFα are mediated by its interaction with TNF-R1 but not with TNF-R2.15

The expression of myocardial TNF receptors in failing myocardium, like β adrenergic receptors, is significantly decreased in comparison with non-failing myocardium.13 In contrast, serum levels of the soluble form of TNF-R1 and TNF-R2 are increased in patients with moderate to severe heart failure.16 It is possible that “shedding” of myocardial membrane bound TNF receptors contributes in part to the increased peripheral levels. In this manner, soluble TNF receptors would neutralise the biological effects of circulating TNFα.17 18 Furthermore, the decreased expression of myocardial TNF receptors will prevent the biological effects of TNFα in the local microenvironment.

The presence of myocardial TNFα along with decreased expression of TNF receptors in failing myocardium suggests activation of the TNFα-TNF receptor axis in the heart failure “milieu” and it is consistent with a pathogenic role of TNF in heart failure.

Mechanisms of TNF induced cardiac injury

Experimentally, TNFα induces reversible contractile dysfunction. The negative inotropic effects of TNFα can be reversed by either washing TNFα off the culture media, by pharmacological removal of TNFα with anti-TNFα antibody or by using soluble TNF receptor proteins.18 The mechanisms by which TNFα induces contractile abnormalities may be in part mediated by preventing the rise in intracellular calcium concentration9 or by stimulating nitric oxide production that exerts negative contractile dysfunction.10 Thus, one of the mechanisms by which TNFα overexpression in heart failure may contribute to its pathogenesis is its ability to induce left ventricular dysfunction in vivo.

Indeed, under acute or short-term experimental conditions, TNFα exerts reversible inotropic effects in vivo, however, whether TNFα chronically present in human myocardium exerts negative inotropic effects is not known. Two lines of evidence suggest that contractile function can be maintained in the presence of high intra-cardiac TNFα levels. Firstly, transgenic mice that overexpress TNFα in the myocardium, have normal contractile function before the development of a dilated cardiomyopathy, suggesting that TNFα induced cardiac injury may result from mechanisms other than direct depression of systolic function. Secondly, heart transplant recipients with no evidence of cellular rejection and normal systolic function have high intracardiac TNFα levels.19

The recognition of increased intracardiac TNFα levels in cardiac allografts is important because in this setting cardiac function is not “normal”. In fact, cardiac allografts are characterixed by left ventricular hypertrophy and diastolic abnormalities. More importantly, over time, heart transplant recipients develop a cardiomyopathy even in the absence of obstructive epicardial coronary artery disease. Thus, it is reasonable to assume that in the transplant setting, the chronic exposure of human myocardium to TNFα may not directly result in systolic dysfunction, but create cellular abnormalities that over time produce a cardiomyopathic state, similar to the cardiomyopathy seen in transgenic mice overexpressing TNFα.

Germane to this discussion are the recent observations that demonstrate that programmed cell death or apoptosis may be a major mechanism for progressive cardiac dysfunction in the failing human heart,20 a condition characterised by overexpression of TNFα. In this context, TNFα, a known stimuli for the activation of programmed cell death, may be a mediator of myocardial apoptosis. Whether, the effects of TNFα in cardiac function are mediated by inducing contractile abnormalities or by promoting disease progression via apoptosis will not be easily dissected, but it is more probable that both mechanisms operate in conjunction to depress cardiac function.

Normalisation of cardiac function is associated with decreased expression of intracardiac TNFα

End stage heart failure may be treated with placement of a left ventricular assist device (LVAD). This form of artificial heart is indicated for patients with dilated cardiomyopathies and refractory heart failure awaiting heart transplantation. The LVAD is implanted in the left ventricular apex and blood enters the pump via a conduit and it is subsequently pumped out through a graft into the ascending aorta. The effect of the LVAD on the heart is to maintain the ventricle empty or chronically unloaded, “resting”. The device occasionally may fail or predispose to chronic infections that required surgical removal. In some patients who underwent removal of the device it was found that cardiac function improved and this observation led to the hypothesis that a chronically unloaded ventricle may recover cardiac function even if had appeared to be terminal. If one postulates that TNFα is important in the pathogensis of heart failure, it would be reasonable to predict that recovery of cardiac function would be associated with elimination of intracardiac TNFα content. Indeed, in a patient that underwent LVAD placement we obtained myocardial tissue at the time of LVAD implantation (fig 2A) and when the device was removed (fig 2B) for the purposes of transplantation. As clearly shown, there was a significant reduction in TNFα expression after LVAD support.21 These findings support the concept that haemodynamic loading conditions regulate the expression of myocardial TNFα in humans and indicate that recovery of cardiac function is associated with elimination of intracardiac TNFα.

Figure 2

Effect of chronic mechanical support on cardiac TNFα content. Myocardial tissue at the time of LVAD implantation (A) and when the device was removed (B) for the purposes of transplantation and stained for TNFα. As clearly shown, there was a significant reduction in TNFα expression after LVAD support.

TNFα antagonism in heart failure patients

There are various pharmacological agents to block the biological effects of TNFα. However, only two have been used in patients with heart failure: pentoxyfilline (a non-specific agent that suppresses TNFα production) and etanercept (a recombinant human TNF receptor fusion protein that specifically blocks the effects of TNFα).

Pentoxyfilline was used in a double blind placebo controlled study in 28 symptomatic heart failure patients. The study patients were treated for six months with either 400 mg of pentoxyfilline three times (n=14) per day or matching placebo (n=14). At the end of the study there were four deaths, all of which occurred in the placebo treated group. There was an increase in ejection fraction (38.7 v26.8, p=0.04) and more patients were free of symptoms in the treated group than in the placebo group. In addition, at the end of the study the TNFα concentration were lower in the pentoxyfilline treated group than in the placebo (2.1 v 6.5 pcg/ml, p<0.001). Given the fact that the number of patients treated in this study is small and that the effect of pentoxyfilline is non-specific, it is difficult to conclusively demonstrate that the beneficial effects of this treatment were secondary to TNFα suppression. However, this clinical study is the first to be completed in an attempt to demonstrate a clinical effect of TNFα antagonism in heart failure patients.22

Compared with pentoxyfilline, etanercept is a specific antagonist of the biological effects of TNFα. Recombinant human etanercept is a dimer of two molecules of the extracellular portion of the p75 TNF receptor (or TNFRII) consisting of 235 amino acids. The two receptors are fused to the Fc portion of human IgG1 consisting of 232 amino acids. This drug binds to biologically active TNFα and prevents its interaction with membrane bound TNF receptors (fig 3). Etanercept has been approved by the FDA for its use on patients with rheumatoid arthritis and two placebo controlled trials in heart failure patients have been completed. The first etanercept study in patients with heart failure, consisted of a single intravenous infusion ascending dose, placebo controlled, safety study in patients with Class III-IV CHF who had serum TNF levels of > 3 pg/ml. No cardiac side effects related to etanercept were observed and several patients had significant improvements in symptoms and in the six minute walk test with a decrease in biologically active TNFα levels by 85%.23

Figure 3

Etanercept (Enbrel): mechanism of action. Etanercept is a specific antagonist of the biological effects of TNFα. Recombinant human etanercept is a dimer of two molecules of the extracellular portion of the p75 TNF receptor (or TNFRII) consisting of 235 amino acids. The two receptors are fused to the Fc portion of human IgG1 consisting of 232 amino acids. This drug binds to biologically active TNFα and prevents its interaction with membrane bound TNF receptors.

The second etanercept study was a double blind randomised placebo control study of two doses of etanercept (5 and 12 mg/m2) on 47 patients with class III and IV CHF and LVEF <35%. The study showed that etanercept given for three months was well tolerated and resulted in a trend toward overall improvements in NYHA classification and quality of life. The changes appeared more consistent in the 12 mg/m2 dose group.24 On the basis of the two previous etanercept trials, a large multicentre, double blind randomised, placebo controlled, phase II/III study is currently underway. The primary objective of that study is to evaluate the efficacy of two dosing regimens of etanercept and placebo on patients with class II-IV CHF.


The experimental and clinical evidence that demonstrates the effect of TNFα in heart failure patients continues to accumulate. It is well established that increased concentrations of TNFα appear in the circulation of heart failure patients and that the levels may have prognostic significance. Also, increased TNFα levels may be responsible for the decreased expression of myocardial TNF receptors observed in failing myocardium. Along with these clinical data, it has been clearly demonstrated that increased levels of TNFα lead to cardiomyopathy and eventually death in experimental animals; therefore, it is reasonable to assume that increased concentrations of TNFα in heart failure patients may be detrimental to cardiac function. In support of this concept we found that cardiac TNFα concentrations were significantly reduced in heart failure patients treated with LVADs; a form of mechanical support that leads to improved cardiac function.

The hypothesis that TNFα contributes to the pathogenesis of heart failure has recently been tested at the clinical level. The results of specific TNFα antagonism in symptomatic heart failure patients demonstrate that anti-TNFα treatment is safe and it may be effective. This hypothesis is currently being tested in a large randomised multicentre study that is expected to be complete in the next couple of years. Perhaps the most important aspect of the ongoing research on the role of cytokines in heart failure is that the recognition of ongoing activation of inflammatory mediators provides new targets for therapeutic intervention.


View Abstract

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.