Role of tumor necrosis factor alpha in the pathogenesis of atrial fibrillation: A novel potential therapeutic target?

Abstract

Atrial fibrillation (AF) is the most common sustained arrhythmia in clinical practice and a major cause of morbidity and mortality. Although the fundamental mechanisms underlying AF remain incompletely understood, atrial remodeling, including structural, electrical, contractile, and autonomic remodeling, has been demonstrated to contribute to the substrate for AF maintenance. Accumulating evidence shows that tumor necrosis factor alpha (TNF-α) plays exceedingly important roles in atrial remodeling. This article reviews recent advances in the roles of TNF-α in the pathogenesis of AF, elucidates the related mechanisms, and exploits its potential usefulness as a novel therapeutic target.

Key words::

• Atrial fibrillation
• atrial fibrosis
• remodeling
• therapy
• tumor necrosis factor alpha

Key messages

• Accumulating evidence supports the notion that tumor necrosis factor alpha (TNF-α) plays paramount roles in atrial remodeling, including structural, electrical, contractile, and autonomic remodeling.

• TNF-α may be a good prognostic biomarker and novel potential therapeutic target for atrial fibrillation.

Introduction

Atrial fibrillation (AF) is a supraventricular tachyarrhythmia with uncoordinated atrial activation and consequently ineffective atrial contraction (1). AF increases in prevalence with advancing age. The prevalence of AF is 2.3% in people older than 40 years and 5.9% in those older than 65 years. Eighty-four percent of persons with AF are older than 65 years, and 32% are older than 80 years. Seventy percent of people with AF are between 65 and 85 years of age (2). AF is associated with a 4–5-fold increase in the risk of stroke, a doubling of risk for dementia, tripling of the risk of heart failure (HF), and a 40%–90% increased risk of mortality (3). The fundamental mechanisms underlying AF have long been debated, but atrial remodeling, including structural, electrical, contractile and autonomic remodeling, has been demonstrated to contribute to the AF substrate for AF maintenance in animal models and humans (4). Substantial evidence has confirmed that inflammation may participate in the structural and electrical remodeling during the initiation and perpetuation of AF (5).

As a proinflammatory cytokine, tumor necrosis factor alpha (TNF-α) plays a pivotal role in various immune and inflammatory processes, including cellular activation, survival, and proliferation, as well as cell death by necrosis and apoptosis (6). In the cardiovascular system, TNF-α is produced in cardiac myocytes, cardiac fibroblasts, smooth muscle cells, and endothelial cells in response to endotoxin independent of the presence of inflammatory cells in ex vivo and in vivo cardiac studies (7–10). Accumulating evidence has demonstrated that TNF-α was closely associated with a variety of cardiovascular diseases such as HF (11–13), cardiac hypertrophy (14,15), atherosclerosis (16,17), acute myocardial infarction (AMI) (18), unstable angina pectoris (19), myocarditis (20,21), pulmonary artery hypertension (22), and peripheral artery disease (23). Recent experimental and clinical studies have shown that TNF-α was involved in the atrial structural, electrical, contractile, and autonomic remodeling (Table I) and therefore can contribute to the progression of AF (24–32). The purpose of this article is to review the investigated data supporting the role of TNF-α in the pathogenesis of AF, to elucidate the related mechanisms, and to explore its potential usefulness as a novel therapeutic target.

Table I. The role of TNF-α in atrial remodeling.

Roles	Study subjects	Major findings	References
Structural remodeling	Cardiac fibroblasts of neonatal rats	↑AT1R density and IP production	(54)
	Mice deficient in TNFR1 or TNFR2	↑Ang-II-dependent cardiac fibrosis	(57)
	Human atrial myofibroblasts	Proliferation, invasion, and MMP-9 secretion by activating TNFR1 and P38 MAPK	(36)
	HL-1 cell line	↑Apoptosis of atrial myocytes	(58)
	TNF1.6 transgenic mice	Atrial hypertrophy and dilation	(24)
	Mice	Atrial fibrosis; altered connexin 40 expression via TGF-β/Smad pathway; activation of myofibroblasts; ↑MMPs	(62)
	Diabetic rabbits	Atrial hypertrophy and fibrosis	(63)
	Atrial tissue from AF patients	Atrial fibrosis	(26,27,64)
Electrical remodeling	Atrial myocytes of mice	↑Abnormalities in action potential propagation and Ca^2+ handling	(24)
	Pulmonary vein cardiomyocytes	↑Amplitude of DADs; ↓ICaL; ↓[Ca^2+]i; ↑Transient inward currents; ↑NCX currents; ↓Sarcoplasmic reticulum calcium currents; ↓Diastolic [Ca^2+]i; ↓SERCA2a expression; ↑Decay of [Ca^2+]i	(68)
	HL-1 cell line	↓SERCA2a expression via hypermethylation SERCA2a promoter; ↓Ca^2+ transient	(69)
	Rat atrial myocardium	↑APD via NO-dependent mechanism	(70)
	Transgenic mice	↓Connexin 40; ↑atrial arrhythmias	(25,62)
Contractile remodeling	Hamster trabeculae	↓Contractile function	(78)
	Human atrial myocardium	↓Contractile function via sphingosine pathway	(30)
	Atria of TNF1.6 mice	↓Contractile function	(24)
	Atria of rats	↓Contractile function	(32)
	Human atrial myocardium	↓Contractile function	(31)
Autonomic remodeling	Isolated mouse atria	↓Release of [^3H]-noradrenaline	(28)
	Human atrial appendages	↓Release of [^3H]-noradrenaline	(29)

APD = action potential duration; AT1R = angiotensin II type 1 receptor; [Ca²⁺]i = intracellular calcium transient; DADs = delayed afterdepolarizations; I_CaL = L-type Ca²⁺ current; IP = inositol phosphates; MAPK = mitogen-activated protein kinase; MMP = matrix metalloproteinase; NCX = Na⁺-Ca²⁺ exchanger; SERCA2a = sarcoplasmic reticulum ATPase; TNFR1 = TNF type 1 receptor; TNFR2 = TNF type 2 receptor.

Basic biology and function of TNF-α

In 1975, Carswell et al. (33) first identified TNF-α, an endotoxin-induced serum factor that caused necrosis of tumors. The human TNF-α gene is located on chromosome 6. TNF-α is synthesized as a 212-amino acid-long type II transmembrane protein arranged in stable homotrimers of approximately 26 kDa (mTNF-α), of which a soluble 17 kDa fragment (sTNF-α) is proteolytically cleaved from the membrane-integrated form by the metalloprotease TNF-α-converting enzyme (TACE). Soluble TNF-α is generally regarded to be an endogenous mediator of inflammation modulating a wide spectrum of cellular responses including activation of genes involved in inflammatory and immunoregulatory responses, cell proliferation, antiviral responses, growth inhibition, and cell death (34).

The biological effects of TNF-α are mediated by two distinct surface receptors, TNFR1 (p55) and TNFR2 (p75) with molecular weights of 55–60 kDa and 70–80 kDa, respectively. Expression of the two receptors has been found in the human and rat myocardium on cardiomyocytes (35), cardiac myofibroblasts (36), vascular smooth muscle cells (37,38), and vascular endothelial cells (38). Both receptors are type 1 transmembrane proteins and share homology in their extracellular domains, but differ considerably within their intracellular regions, suggesting that each receptor has distinct modes of signaling and cellular functions (39,40). TNFR1 signaling is fully activated by both the membrane-bound and soluble trimeric forms of TNF-α, while TNFR2 signaling can only be efficiently activated by the membrane-bound form of the TNF homotrimer. TNFR1 is the main receptor subtype expressed in most tissues including myocardium and its downstream signaling system has been extensively investigated, whereas TNFR2 is mainly expressed on lymphocytes as well as endoepithelial cells and its signal transduction remains less well characterized (40). The binding of TNF-α to TNFR1 implies the consecutive formation of two different TNF receptor signaling complexes separated both temporally and spatially. The first of them (Complex I) activates the expression of the transcriptional factors such as activating protein 1 (AP-1) and nuclear factor kappa-B (NF-κB) that prevent the triggering of cell death processes, whereas the second complex [Complex II or death-inducing signaling complex (DISC)] triggers cell death processes after the internalization of the receptor (41). Depending on the cellular context and the microenvironment conditions, TNFR1 activation can lead to the induction of proliferation processes, apoptosis, or necroptosis (40). Similarly, after the binding of mTNF-α to TNFR2, TNF receptor adaptor factor (TRAF) 2 interacts directly with the receptor, thus leading to the generation of a signaling complex that also includes TRAF3, cellular inhibitor of apoptosis (cIAP) 1, and cIAP2. This complex induces the activation of the transcription factors AP-1 and NF-κB related to cell proliferation and survival (40). The signals triggered by TNFR2 may not only play a protective role in some disorders (42–45) but also have a detrimental effect on several autoimmune diseases such as familial rheumatoid arthritis (46), systemic lupus erythematosus (47), Crohn's disease (48), and ulcerative colitis (49).

Role of TNF-α in atrial structural remodeling

Atrial structural remodeling promoting the initiation and maintenance of AF

Atrial structural remodeling involves cell death, including necrosis, apoptosis, and autophagy, myocyte hypertrophy, fibroblast proliferation, and excess extracellular matrix (ECM) production, causing fibrosis. Reactive fibrosis results in interstitial expansion between bundles of myocytes, whereas coexisting reparative fibrosis replaces dead atrial cardiomyocytes, interferes with electric continuity, and slows conduction (50,51). In a mathematical model of the two-dimensional canine atrium, increased atrial mass can accommodate a greater number of circulating wavelets, setting up multiple circuit re-entry (52). Atrial fibrosis not only creates a vulnerable tissue substrate with slow conduction and susceptibility to unidirectional conduction block but also promotes the emergence of triggers that, on encountering this vulnerable substrate, have a high probability of initiating re-entry (53). Accordingly, fibrosis causes AF progression to permanent forms. Additionally, AF itself may promote structural remodeling, creating a long-term positive feedback loop that contributes to the development of permanent forms.

Differentiation and proliferation of cardiac fibroblasts and ECM production

In neonatal rat cardiac fibroblasts, TNF-α increased angiotensin II (Ang-II) type 1 receptor (AT1) mRNA levels in a time- and dose-dependent fashion, which was associated with an increase in membrane receptor density and enhanced production of Ang-II-stimulated second messenger inositol phosphates within the cell (54). Ang-II acting on cardiac fibroblasts (CFs) elicited pro-fibrotic effects on the heart through multiple mechanisms, including increased ECM protein synthesis, decreased matrix metalloproteinase (MMP) activity, and increased tissue inhibitor of metalloproteinase (TIMP) activity. Specifically, Ang-II stimulated collagen I, collagen III, and fibronectin synthesis through AT1R activation in adult and human CFs (55). Ang-II can also induce secretion of several other important bioactive molecules from CFs, thereby regulating cellular function in an autocrine/paracrine manner. For example, Ang-II induced expression and secretion of TNF-α (8,9) and transforming growth factor-beta (TGF-β) (56) predominantly via the AT1R. Recently, an elegant series of experiments in mice deficient in either TNFR1 or TNFR2 has further demonstrated that TNF-α induced Ang-II-dependent cardiac fibrosis by signaling through TNFR1, but not via TNFR2, which enhanced the generation of monocytic fibroblast precursors in the heart (57). Additionally, a previous study corroborated that TNF-α increased human atrial myofibroblast proliferation, invasion, and MMP-9 secretion predominantly by binding to the TNFR1 receptor and activating p38 MAPK pathway (36).

Apoptosis and autophagy of atrial myocytes

A recent study on the murine atrial cardiomyocytes HL-1 cell line has revealed that enhanced expression of TNF-α promoted cardiomyocytes apoptosis by initiating the extrinsic (caspase-8) and intrinsic apoptotic pathways (bax and caspase-9) and then activating caspase-3 (58). However, it is unclear whether TNF-α plays a role in structural remodeling attributable to autophagy of atrial cardiomyocytes. In this regard, Jia et al. (59) demonstrated that TNF-α can induce mRNA expression of autophagic gene microtubule-associated protein 1 light chain 3 through JNK and protein kinase B pathways, enhance Beclin-1 protein expression via the JNK pathway, and promote autophagy of human atherosclerotic vascular smooth muscle cells. On the other hand, a recent report by Garcia et al. (60) has provided novel evidence that ultrastructural atrial remodeling characterized by an impaired cardiac autophagy was present in patients developing postoperative AF after coronary artery bypass surgery. Similarly, another more recent study has further proved that autophagy dependent on adenosine monophosphate-activated protein kinase (AMPK) occurred in atrial tissues both from rapid atrial pacing-induce AF canines and from patients with chronic AF, which may be a novel mechanistic contributor to AF (61). Therefore, the roles and mechanisms of TNF-α underlying autophagy of atrial myocytes in atrial structural remodeling warrant further investigation in the future.

Atrial hypertrophy and fibrosis in animal models

In a TNF1.6 transgenic mouse model of cardiomyopathy, Saba et al. (24) demonstrated that cardiac-specific overexpression of TNF-α was associated with atrial hypertrophy and dilation, although how TNF-α can promote atrial cardiomyocyte hypertrophy remains to be further clarified. The average weight of isolated left atrium (after surgical removal of an organized blood thrombus, typically observed in TNF1.6 atria) was 3.6-fold larger in 6–9-month-old female TNF1.6 versus control mice. Additionally, marked collagen deposition was found in TNF1.6 versus control atria. In another mouse model in which the effects of direct delivery of TNF-α on atrial fibrosis were investigated, Liew et al. (62) proved that TNF-α was implicated in the pathogenesis of atrial fibrosis and altered connexin-40 expression in mice through the TGF-β/Smad signaling pathway, activation of myofibroblasts, and increased secretion of MMPs, suggesting that TNF-α may contribute to the arrhythmogenic substrate and development of AF. With similarity to the findings, the investigation on the diabetic rabbits indicated that increasing expression of TNF-α can lead to hypertrophy of atrial tissue and atrial fibrosis which may serve as important substrates for the development of AF (63).

Atrial fibrosis in humans

Clinically, Chen et al. (63) confirmed that the plasma levels of TNF-α were markedly higher in patients with permanent AF than in those with sinus rhythm (SR). The finding was further verified by Qu et al. (26) that patients with valvular disease with AF exhibited higher NF-κB activity, higher concentration of TNF-α, severe lymphomonocyte infiltration, and more fibrosis in right atrial tissues than those with valvular disease with SR. In addition, NF-κB activity correlated positively with TNF-α levels and collagen volume fraction. Similarly, Deng et al. (27) also revealed that TNF-α elevated remarkably in the plasma and left atrial tissue and had a positive correlation with left atrial diameter in patients of chronic AF resulting from rheumatic heart disease. Collectively, these results showed that TNF-α was strongly correlated with atrial fibrosis in AF patients.

Taken together, the aforementioned data from cell experiments, animal models, and humans suggest that TNF-α plays extremely important roles in atrial structural remodeling contributing to the pathogenesis of AF under different pathophysiological conditions (Table I, Figure 1).

Figure 1. Schematic showing the mechanisms of atrial structural remodeling resulting from TNF-α. On the one hand, TNF-α can enhance the density of AT1R and then promote the profibrotic roles of angiotensin II. Additionally, TNF-α exerts the profibrotic effects on atrial fibroblasts by binding to TNFR1 and activating the P38 MAPK signaling pathway. On the other hand, TNF-α can promote atrial myocyte apoptosis by activating the extrinsic (caspase-8) and intrinsic apoptotic pathways (bax and caspase-9), leading to caspase-3 activation, although it is still not fully understood how TNF-α can induce autophagy as well as hypertrophy of atrial myocytes and thereby cause atrial fibrosis and hypertrophy. Ang-II = angiotensin II; AT1R = angiotensin II type 1 receptor; ECM = extracellular matrix; FADD = FAS-associated death domain protein; MAPK = mitogen-activated protein kinase; MMPs = matrix metalloproteinases; TIMPs = tissue inhibitor of matrix metalloproteinases.

Role of TNF-α in atrial electrical remodeling

Atrial electrical remodeling identified to date consists primarily of ion channel remodeling and gap remodeling. The former includes the downregulation of L-type Ca²⁺ current (I_CaL) as well as upregulation of rectifier background K⁺ current (I_K1) and constitutive acetylcholine-regulated K⁺ current (I_KACh). The last-mentioned comprises abnormal expression and distribution of the gap junction connexin hemichannels that connect cardiomyocytes electrically. In addition, the intracellular Ca²⁺ overload and abnormal Ca²⁺ handling may enhance the development of triggered activities, hence promoting the development and maintenance of AF (64).

Ion channel remodeling

Reduced I_CaL decreases the inward Ca²⁺ current, maintaining the AP plateau, shortening the AP duration (APD) as well as atrial refractoriness, and thereby promoting re-entry. Inward rectifier currents such as I_K1 upregulated in AF are a particularly important determinant of AF maintaining re-entry (65). Another important inward rectifier current, I_KACh, mediates the effects of acetylcholine and underlies the marked ability of vagal activation to promote AF by causing spatially heterogeneous increases in inward rectifier current and reductions in APD (66).

Currently, growing evidence indicates that TNF-α is involved in atrial electrical remodeling (Table I). In atrial myocytes isolated from failing heart overexpressing TNF-α, abnormalities in action potential propagation and Ca²⁺ handling may contribute to the initiation and maintenance of re-entrant atrial arrhythmias such as atrial fibrillation and flutter (24). More importantly, an elegant series of experiments by Lee et al. (67) confirmed that, compared with the control pulmonary vein (PV) cardiomyocytes, the TNF-α-treated PV cardiomyocytes had a significantly larger amplitude of the delayed afterdepolarizations (DADs), smaller I_CaL, larger transient inward currents, larger Na⁺–Ca²⁺ exchanger currents, a smaller intracellular calcium transient, smaller sarcoplasmic reticulum calcium content, larger diastolic intracellular calcium, a longer decay portion of the calcium transient, and a decreased sarcoplasmic reticulum ATPase (SERCA2a) expression. The findings suggested that TNF-α can increase the PV arrhythmogenic activity and impair the calcium regulation, thereby causing inflammation-related AF. Additionally, Kao et al. (68) described the molecular mechanism behind the downregulation of SERCA2a by TNF-α in a HL-1 cardiac muscle cell line. They confirmed that TNF-α repressed SERCA2a gene expression through hypermethylation of the SERCA2a promoter, resulting in decreased SERCA2a protein levels and Ca²⁺ transients in HL-1 cells independently of NF-κB activation or oxidative stress. Furthermore, a recent study on rat atrial myocardium has revealed that TNF-α can provoke a marked prolongation of action potential, subsequently transforming into lump-like depolarizations and finally leading to occurrence of arrhythmias via a NO-dependent mechanism (69).

Gap remodeling

Gap junctions contain transmembrane ion-channel protein called connexins, of which connexins 40 and 43 are the most important in atrial tissue and mediated cardiomyocytes to one another or to fibroblast electrical coupling. The relationship between connexin 40 level and atrial conduction is inconsistent with high levels correlated with slow and complex conduction in some studies, whereas reduction in connexin 40 was associated with increased conduction velocity in another study (70,71). Furthermore, myofibroblast–myocyte coupling, if it occurs under pathophysiological conditions, could be a significant additional cellular factor synergistically promoting the emergence of triggers due to diastolic depolarization, DADs, or early afterdepolarizations (EADs), especially when combined with the structural source-sink effects of fibrosis at the tissue level (53).

Similarly, in a transgenic mouse model with overexpression of TNF, Sawaya et al. (25) corroborated that TNF played a contributory role in downregulation of connexin 40 that was associated with an increased prevalence of atrial arrhythmias, which was consistent with the finding by Liew et al. (61).

Collectively, all of the basic experiments raise the possibility that TNF-α plays important roles in atrial electrical remodeling.

Role of TNF-α in atrial contractile remodeling

Experimental and clinical studies have shown that atrial contractile dysfunction correlated with the duration of AF (72–75). The loss of atrial function is thought to be triggered by Ca²⁺ overload during short episodes (several minutes to hours) of AF and might be mediated by a decrease in the release of Ca²⁺ from the sarcoplasmic reticulum during post-tachycardia contractile dysfunction (76). However, atrial contractile remodeling during several days of AF goes hand in hand with electrical remodeling and might be caused by the I_CaL downregulation (75).

Several lines of evidence have been found to support the notion that TNF-α also plays an important part in atrial contractile remodeling (Table I). After Finkel and colleagues (77) reported that TNF-α decreased contractile function in isolated hamster trabeculae, Cain and co-authors (30) corroborated that TNF-α and IL-1 beta separately and synergistically depressed human atrial myocardial function through the sphingosine signaling pathway. Likewise, atrial enlargement in TNF1.6 mice not only failed to improve global atrial contractile function but also was associated with a marked reduction in force generated per unit of atrial mass. In addition, increasing the stimulation rate from 1 to 6 Hz was associated with a more significant reduction of developed force in TNF1.6 atria than in control, and an increase in basal tension in the transgenic atria. Thus the impaired contractile function of TNF1.6 atria may result partially from the structural changes (24), which was in accordance with the observations that neonatal hypoxia-ischemia resulted in long-term atrial contractile dysfunction and ultrastructural degenerative alternations in the atria of rats, and that anti-TNF-α treatment with etanercept can noticeably ameliorate the aforementioned detrimental process (32). Moreover, TNF-α can cause an immediate negative inotropic effect and increase specific oxygen demand in human right-atrial myocardium resulting from an impaired economy of contraction as well (31).

Role of TNF-α in atrial autonomic remodeling

Autonomic remodeling results in some electrophysiological changes, thereby contributing to the initiation, perpetuation, ventricular response rate, and termination of AF. Vagal discharge can enhance I_KACh, reduce APD, and stabilize re-entrant rotors (66), whereas β-adrenoceptor activation is able to hyperphosphorylate type 2 ryanodine receptors (RYR2s), to increase diastolic Ca²⁺ leak, and to promote DAD-related ectopic firing (78). Some studies (79,80) suggested that the imbalance between the activity of the sympathetic and parasympathetic nerves acted as an important contributor to the initiation and perpetuation of AF. Additionally, other reports (81–83) corroborated that the occurrence of AF was followed by autonomic remodeling, including alternations in sympathetic and parasympathetic nerve density and spatial distribution, which then facilitated the maintenance of AF.

In this regard, TNF-α can inhibit the release of [³H]-noradrenaline from both isolated mouse atria (28) and human atrial appendages (29), suggesting that TNF-α may modulate the activity of the sympathetic nervous system in atria (Table I). However, further investigations are required to determine whether TNF-α can modulate the activity of the parasympathetic nervous system in atria, whether TNF-α is capable of exerting some effects on ganglionated plexi, or whether TNF-α is correlated with AF as a consequence of autonomic remodeling.

The Role of TNF-α in AF complicated by HF, especially HF with preserved ejection fraction (HFPEF)

Inflammation plays a crucial role in the pathogenesis of HF (84). TNF-α has been implicated in HF progression as a mediator of myocardial dysfunction and adverse remodeling (85). The Framingham Heart Study has demonstrated that there was a 68% increase in risk of incidence of HF per tertile increment in TNF-α (mg/dL) among participants with no previous history of myocardial infarction or HF (86). The other clinical studies to date have also shown that elevated circulating levels of TNF-α are associated with increased mortality in HF patients regardless of ejection fraction (EF) (12,87,88). Furthermore, TNF-α can improve risk prediction over established indicators in HFPEF and HF with reduced EF (HFREF) (12). More importantly, a recent report has indicated that circulating TNFR1 levels are significant predictors of incident HF, particularly for HFPEF versus HFREF (89). In a small-sample study, TNF-α was highest in HFREF compared to HFPEF and similar between HFPEF and controls, whereas sTNFR1 and sTNFR2 levels in HFPEF were significantly higher compared with controls (90). In another larger sample study, plasma levels of TNF-α and TNFR1 were significantly elevated in HFPEF relative to controls, while levels of sTNFR2 were significantly higher in HFPEF than both controls and HFREF (91). In addition, patients with HFPEF had a higher prevalence of AF than those with HFREF, although AF was prevalent in both HF phenotypes (91–93).

Why can TNF-α be closely related to AF complicated by HFPEF? One possible explanation is the new paradigm for HFPEF development recently proposed by Paulus et al. (94). In the new paradigm, a high prevalence of comorbidities such as obesity, diabetes mellitus, chronic obstructive pulmonary disease, and salt-sensitive hypertension induce a systemic proinflammatory state (e.g. elevated TNF-α levels), which results in coronary microvascular endothelial inflammation, further reducing nitric oxide bioavailability, cyclic guanosine monophosphate content, as well as protein kinase G (PKG) activity in adjacent cardiomyocytes, and then favors the development of hypertrophy and the increase of resting tension because of hypophosphorylation of titin. Finally, both stiff cardiomyocytes and interstitial fibrosis contribute to high diastolic left ventricular stiffness and heart failure development.

Altogether, all of the above studies demonstrated that inflammation mediated through TNF-α and its receptors TNFR1 as well as TNFR2 may represent an important component of a comorbidity-induced inflammatory response that partially drives the pathophysiology of HFPEF and promotes the occurrence of AF.

TNF-α: a good prognostic biomarker for AF?

Recently, several clinical studies have demonstrated that circulating TNF-α can predict the degree of left ventricular reverse remodeling after cardiac resynchronization therapy (95), cardiovascular events in patients with ST-segment elevation myocardial infarction (96), as well as the progression of severity and late cardiac death in patients with congestive heart failure (97). Furthermore, elevated levels of soluble CD40 ligand and receptor activator of nuclear factor-κB ligand (RANKL), as members of the TNF superfamily, are independent predictors for postoperative AF (98,99), except that a high soluble RANKL level can predict AF recurrence after primary ablation procedure in lone AF patients (100). More importantly, a growing body of evidence has shown that TNF-α was strongly associated with AF (26,27,63,101,102). Therefore, although TNF-α as a good prognostic biomarker for AF is established in theory, further investigations on the deduction via large-scale randomized controlled trials are warranted.

TNF-α-targeted therapy: is it suitable for AF?

As discussed earlier, TNF-α as a proinflammatory cytokine plays paramount roles in atrial remodeling consisting of structural, electrical, contractile, and autonomic remodeling. Therefore, anti-TNF-α therapeutic strategy for AF is theoretically appropriate.

Types and functions of anti-TNF-α agents

Since the end of the 1990s, several new biotechnological agents with the capability to inhibit TNF-α have been available for initial use in the treatment of rheumatoid arthritis (RA). Etanercept is a recombinant fusion protein that consists of the soluble TNF receptor (p75) linked to the Fc portion of human IgG1 (TNFR: Fc). Infliximab, adalimumab, and golimumab are monoclonal antibodies against TNF-α. Certolizumab pegol (CZP) is a human anti-TNF-α antibody Fab fragment that is chemically linked to polyethylene glycol to increase half-life. It neutralizes membrane-associated and soluble TNF-α (103,104). All the above-mentioned drugs bind and inhibit sTNF-α with comparable efficiency (105), but they bind with different kinetics to mTNF-α (106,107), which can influence the relative efficacy of the therapy (108). Recently, two different strategies have been proposed to target specifically the pathological sTNF–TNFR1 signaling: First, specific inhibition of sTNF-α by using the dominant negative sTNF-α mutant, XPro 1595 (109), so that the mTNF-α–TNFR1 and mTNF-α–TNFR2 interactions remain intact; second, inhibitors of TNFR1, including monoclonal antibodies [Atrosab (110)] and antagonistic TNF mutants with specificity for TNFR1 [R1antTNF (111)], have been proposed to leave TNFR2 signaling fully functional. In the past two decades or more, these drugs have gained a pivotal role in the treatment of many immune-mediated diseases, including RA, psoriatic arthritis, juvenile idiopathic arthritis, ankylosing spondylitis, psoriasis, multiple sclerosis, and inflammatory bowel diseases.

Incidence of AF in chronic systemic inflammatory diseases (CSID)

Recently, increasing data have indicated that inflammation plays a key role in the pathogenesis of AF (112), and large prospective cohort studies found an association between systemic inflammation and incident AF, also after adjustment for traditional cardiovascular disease risk factors (113,114). More importantly, chronic systemic inflammatory diseases (CSID), including RA, psoriasis, systemic sclerosis, celiac disease, and inflammatory bowel disease, were associated with an increased incidence of AF and stroke (115–120), which raises the possibility that AF is potentially linked to CSID, especially RA.

Although the incidence of AF is significantly higher in CSID than in non-CSID subjects, especially reaching 40% in RA patients (117,119), it remains unknown whether such an increased risk 1) is inherently associated with RA via inflammatory mechanisms, 2) is only an indirect consequence of the disease through other intermediate factors, such as ischemic heart disease (IHD), HF, and the use of certain medications, or 3) results from a combination of both. Further investigations are warranted to clarify the topic fully.

Effects of anti-TNF-α therapy on atrial remodeling

As illustrated previously, TNF-α as a proinflammatory cytokine is widely implicated in the pathogenesis of AF, which underlies the anti-TNF-α therapeutic strategy. However, there is little evidence showing the role of TNF-α inhibitor in attenuating atrial structural remodeling (121,122). Recently, in the heart of rats, it has been proved that neonatal hypoxia-ischemia caused long-term cardiac dysfunction and ultrastructural degenerative changes. Etanercept administration soon after hypoxia-ischemia may have heart-protective effects (32). Similarly, in an experimental rat model of diabetic cardiomyopathy, TNF-α-monoclonal antibody treatment can ameliorate left ventricular function through abrogating intramyocardial inflammation and cardiac fibrosis (121). Furthermore, in another clinical study, infliximab therapy for a period of three months can improve echocardiographic left atrial global strain and volume index parameters, which are typical in AF, in RA patients with preserved LV function (122). Apart from the above monotherapy, combination of TNF-α ablation and MMP inhibition can prevent HF after pressure overload in TIMP-3 knock-out mice (123), which provides new insights into anti-TNF-α and MMP inhibitor combination therapy in AF accompanied by HF. Actually, no real investigations on animal models and humans to date have been launched into exploring the roles and mechanisms of TNF-α inhibitors in AF, especially in AF complicated by CSID.

On the contrary, there has been a case report of new-onset AF in a 57-year-old man with RA who received a combination of etanercept and methotrexate for five months, illustrating the potential risk of new-onset AF associated with anti-TNF-α therapy (124). Nevertheless, several reasons responsible for this phenomenon should be taken into consideration: 1) the new-onset AF probably resulted from either etanercept or methotrexate, or a combination of them, because methotrexate therapy was able to cause atrial flutter (125); 2) the emerging AF may be attributed to other potential diseases such as IHD and HF in the patient; and 3) TNF-α polymorphism may account for the side effect of anti-TNF-α treatment because divergent genotypes contributed to different responses to TNF-α inhibitors (126–129). For example, patients with the TNF-α -308G/G genotype are better infliximab responders than those with A/A or A/G genotypes (126,128). Therefore, it is plausible that the best individualized treatment aiming at divergent TNF-α genotypes would produce better outcomes, and meanwhile mitigate the side effects to a minimum in patients with AF complicated by CSID.

Taken together, current evidence has shown that TNF-α has a close association with AF, and that the incidence of AF in CSID is significantly higher than that in non-CSID. Therefore, anti-TNF-α therapy is feasible in theory. However, there are few studies that focus on this topic and therefore almost no available experience in the effects of TNF-α antagonist on AF, especially AF in RA subjects. Hence, there is a long way to go before anti-TNF-α therapy for AF can be implemented in clinical practice.

Conclusion and future perspectives

The last 10–15 years have seen great progress in the role of TNF-α in the pathogenesis of AF. The available evidence to date supports the notion that TNF-α is extensively involved in atrial remodeling, including structural, electrical, contractile, and autonomic remodeling, which contributes synergistically to the initiation and maintenance of AF. Additionally, TNF-α as a prognostic predictor for AF is theoretically established, so it is as a novel potential therapeutic target for AF in the future.

These results, on the one hand, provide some insights into future studies on other emerging fields. For example, TNF-like weak inducer of apoptosis (TWEAK), a typical member of the TNF ligand family, has been recently demonstrated by us to promote the proliferation and collagen synthesis of rat CFs via the NF-кB pathway (130). Furthermore, TWEAK was closely correlated with cardiac remodeling (131). Thus, given the current evidence, it is tempting to speculate that TWEAK is also strongly associated with cardiac fibrosis, contributing to the pathogenesis of AF.

On the other hand, these data on TNF-α lay a solid foundation for TNF-α inhibitors prescribed clinically for the treatment of AF, particularly in AF complicated by CSID. However, hitherto some specific issues that remain unanswered should be prioritized in future studies.

First, in vitro experiments should be designed to determine 1) how TNF-α induces apoptosis and hypertrophy of atrial myocytes, 2) whether autophagy of atrial cardiomyocytes are mediated by TNF-α, and 3) what net effects TNF-α exerts on atrial myocytes during autonomic remodeling.

Second, in vivo studies using various animal models of AF need to evaluate the efficacy and safety of anti-TNF-α treatment. More recently, a novel and highly specific small-molecule TNF-α inhibitor C87 has been demonstrated to be of potential use in treating TNF-α-mediated inflammatory diseases (132). Nevertheless, it remains unclear whether C87 has a better effectiveness on anti-TNF-α treatment for RA, especially for AF accompanied by RA, than the conventional TNF-α inhibitors.

Finally, some large-scale, randomized controlled clinical trials need to be conducted to ascertain the possibility of TNF-α as a prognostic predictor for AF, to determine further the efficacy and safety of anti-TNF-α therapy in patients with AF and RA, and to exploit the crucial roles of TNF-α polymorphism in individualized treatment in these subjects as well.

In the future, as the aforementioned issues—particularly the efficacy and safety of anti-TNF-α therapy in patients with AF and RA—are completely addressed, it appears promising that TNF-α inhibitors would be applied to the treatment of AF first in RA patients as first-line medications, then in other CSID subjects, and finally in the general population.

Funding: This study was sponsored by the Natural Science Foundation of China (81270238), the Scientific Research Foundation for the Doctoral Degree, State Education Ministry of China (20100131110059), and supported by the Scientific Development Plan of Shandong Province of China (2012G0021850).

Declaration of interest: The authors report no conflicts of interest.