Atrial fibrillation (AF) alters the function of the heart beyond mere changes in cardiac rhythm. It has been associated with untoward clinical outcomes in patients, such as the development of heart failure and stroke. Recent years have brought attention to several aspects of AF-associated alterations termed ‘remodelling’. This article will provide an overview of pertinent aspects of this subject with a view to therapeutic implications. Here, aspects related to electrical and contractile properties, alterations in atrial structure and aspects of inflammatory changes and endothelial remodelling are covered. Much of the detailed knowledge and understanding of AF remodelling stems from animal models. However, many of the changes observed in animals could be similarly well-documented in human disease, and there is often overlap with respect to the potential therapeutic implications of the findings.
Electrical and Contractile Remodelling
Since the early 20th century, the simultaneous co-existence of multiple re-entrant circuits was assumed to underlie AF.1 Short refractoriness is a strong determinant and prerequisite of re-entrant activity.2 Short atrial refractory periods as well as lack of action potential duration (APD) rate adaptation have been firmly linked to AF in humans.3
The cellular abnormalities underlying short APD – the single most important cellular determinant of refractoriness – and specific ionic changes were first demonstrated in a canine AF model using rapid atrial pacing.4 They were subsequently described for human atria (see Figure 1).5 The key changes involve a prominent downregulation of the L-type calcium current (ICa,L). This is initially (within minutes to hours) achieved through biophysical channel inactivation as a short-term response to rapid activation.6 With AF persisting over hours to days, changes in messenger RNA (mRNA) transcription and ion channel synthesis occur, leading to further reduction in L-type channels. While this adaptation represents a primarily successful strategy to protect atrial cells from Ca2+ overload, it enhances vulnerability to AF and promotes AF persistence.7 The AF-related increase in the protease calpain might also contribute to reduced ICa,L.8 Additionally, the activity of protein phosphatases is enhanced in AF and channel dephosphorylation by these enzymes reduces ICa,L.9
Upregulation of inward rectifier currents (inward-rectifying K+ current [IK1] and constitutively active acetylcholine-dependent activation of a cardiac potassium channel or [IKACh]) also contribute to the abbreviation of atrial cellular APD.10 The regulation of the atrial ultra-rapid delayed rectifier current (IKur) in AF has been a matter for considerable debate. This current is called ‘ultra-rapid’ because of its fast activation kinetics. It is present in human atrial but not ventricular cells.11–13 One early study demonstrated reduced IKur with human AF,14 which could limit the efficacy of IKur inhibition, but this result was not confirmed by subsequent investigations.5,15 The effects of IKur inhibition on atrial repolarisation depend strongly on action potential morphology. The brief, triangular action potentials (see Figure 1) during AF are particularly susceptible to prolongation by IKur inhibition.16
The association of AF with reduced atrial contractility has been known for a long time.17 Contractile atrial dysfunction develops over a similar time-course to electrical remodelling and APD abbreviation, with the subsequently reduced Ca2+ entry appearing to play a role.18 Recently, important changes in intracellular Ca2+ handling has moved into the focus of AF research.19 AF is associated with spontaneous Ca2+ release from the sarcoplasmic reticulum (SR), causing after-depolarisations.20 Calcium is released from the SR into the cytoplasm by ryanodine receptor type-2 (RyR2). This has a higher single-channel open probability in AF, leading to a greater proportion of Ca2+ being extruded.21 RYR2 phosphorylating calmodulin-dependent protein kinase II (CaMKII) is increased in AF,22 as is phosphorylation of RYR2, further enhancing Ca2+ exit from the SR.23 Besides changes in SR calcium handling, the Na+-Ca2+ exchanger (NCX) is upregulated in AF. Mediating a Ca2+ efflux and Na+ influx (in forward mode), NCX further depolarises the cell with a net inward movement of charge. Accordingly, it might trigger arrhythmogenic after-depolarisations.24
Recent advances in drug development have brought about new directions for antiarrhythmic therapy. While conventional antiarrhythmic drug therapy is hampered by potentially significant side effects (such as ventricular pro-arrhythmia), novel substances with atrial specificity have moved into the focus of development.25
Furthermore, the newly approved antiarrhythmic substance dronedarone (a congener of amiodarone – not an atrial-specific substance) has been associated with a reduction in cardiovascular morbidity and mortality in a large clinical trial, setting new standards for the development of antiarrhythmic drugs.26
Pharmacological Inhibition of IKur
Several substances have been designed to specifically inhibit atrial ionic channels (having been described first, IKur remains the prototypical drug target of this type). While concerns have been raised that drugs that affect IKur may in fact promote AF if applied to ‘unremodelled’ atria,27 these substances prolong atrial refractoriness without ventricular side effects and effectively convert AF in animal models.28 Respective compounds that affect Kv1.5 channels (underlying IKur) are effective in the channel-open state.29In silico analyses suggest maintained efficacy at rapid activation rates.30 As IKur (and another rapidly activating K+ current, transient outward current [Ito]) contribute to early atrial repolarisation, drugs inhibiting these targets have been termed ‘early’ class III compounds.31
IKur/Ito blockade (by the experimental substance AVE0118) markedly prolonged atrial refractoriness without affecting QT duration in studies using a goat model of pacing-induced AF.31 While the class III effect of dofetilide (an IKr blocker) was reduced after 48 hours of AF, the action of AVE0118 was even enhanced after atrial electrical remodelling. AVE0118 suppressed the inducibility of AF, whereas dofetilide failed to prevent induction of AF paroxysms.31 During persistent AF, AVE0118 prolonged the atrial wavelength in a dose-dependent manner and persistent AF was cardioverted by drug administration.31 Besides effective AF conversion, application of this ‘early’ class III drug represents an elegant means of selectively restoring atrial contractility in AF, providing a novel potential option for the reduction of atrial clot formation and subsequent embolism in AF.32
Despite the fact that IKur has historically been described as a purely atrially expressed ion channel, there are no clinically applicable drugs that exclusively inhibit IKur. Vernakalant is the most prominent among the clinically effective substances that have inhibitory effects on IKur.33 Vernakalant is a novel atrial-specific antiarrhythmic compound with multiple ion channel-blocking actions (see below). It has proved to be safe and effective in cardioverting AF of recent onset (up to three days) in phase III clinical trials.45 It is accordingly awaiting the approval of regulatory institutions for use in AF cardioversion.
Pharmacological Inhibition of INa
Shortened APD allows for smaller re-entry circuits and stabilisation of re-entrant rotors, helping to perpetuate the arrhythmia. Rotors are excitation waves with excitable but unexcited cores that can act as generators for the maintenance of re-entrant arrhythmia.34
Class I antiarrhythmic drugs are Na+-channel blockers that decrease conduction velocity. Accordingly, their use should in fact promote re-entry, according to classic leading circle theory.35,36 The paradoxical benefit in AF is explained by mechanisms leading to the extinction of rotors.37 The centres of rotors enlarge with decreased INa availability. Enlarged rotors are less stable and have a greater likelihood of extinguishing at anatomical or functional boundaries. Finally, there is a reduction in the number of daughter rotors that can help maintain the arrhythmia. Class I drugs also suppress ectopic activity from pulmonary veins, representing another important additional antiarrhythmic effect.38 In line with these findings, the clinical use of class I antiarrhythmic compounds (flecainide and propafenone) for conversion of short-lasting AF (<48 hours) is well-established and highly effective.39,40 The lack of class I antiarrhythmic drug efficacy in long-standing AF may relate to progressive remodelling involving structural changes in the later stages of AF.41
Besides the above-mentioned conventional class I antiarrhythmic drugs that are widely used for converting AF of recent onset, novel substances also inhibit INa and have accordingly been found to be effective in treating AF.33,42 It is of note that the biophysical properties of atrial and ventricular sodium channels differ, providing ways to selectively affect atrial INa.43 Such atrial-specific action on sodium channels (exploiting these biophysical differences) has been shown for the anti-anginal substance ranolazine.
While ranolazine does affect ventricular ion channels, it inhibits peak and late INa, leading to successful conversion of AF in animal models without significant ventricular side effects.43 In fact, there are clinical indicators that the substance – which is generally well tolerated – may work to clinically suppress AF in addition to reducing the incidence of ventricular arrhythmias in patients with acute coronary syndromes.44 Respective controlled trials of ranolazine for AF termination and relapse prevention are under way.
Vernakalant (briefly mentioned above) is an effective, rapidly acting substance for the conversion of recent-onset AF. Ito and IKACh are among other pharmacological targets of this substance besides IKur and INa. This mixture of ion channel inhibition makes it particularly interesting as it is the first atrial-specific compound that has significantly progressed through clinical trials. Intravenous vernakalant rapidly terminated recent onset AF. In particular, patients with short (three to seven days) duration of AF randomised to vernakalant converted rapidly to sinus-rhythm compared with patients receiving placebo (51 versus 4%, p<0.001).45 There were no incidences of ventricular pro-arrhythmia within 24 hours following cardioversion, despite mild QT prolongation. One incidence of Torsade de pointes tachycardia was observed 32 hours after the drug was administered. Conversion rates were limited in long-duration AF (seven to 45 days).
Pharmacological Inhibition of IK1 and IKACh
Upregulation of IK1 (caused by increases in underlying Kir2.1 ion channel expression) is another hallmark of AF-related electrical remodelling. These changes facilitate the occurrence and persistence of AF. Increased IK1 helps to stabilise re-entrant rotors, and ionic current inhibition has accordingly been suggested as an antiarrhythmic approach for AF therapy. However, to date there is no specific inhibitor of IK1 and, given the important role for setting the resting membrane potential of cardiomyocytes, channel inhibition might have detrimental side effects. Kir3-based muscarinic currents (IKACh) are functionally upregulated during AF owing to increased constitutive activity (passing current in the absence of an agonist). Upregulated IKACh supports AF sustenance in a similar manner to IK1. Like IKur, IKACh is only expressed in the atria, making inhibition of this current a potentially interesting antiarrhythmic target devoid of ventricular side effects. Experimental studies of specific inhibitors indicate efficacy in various disease models, but pure IKACh inhibitors are not available for clinical use.46 Some antiarrhythmic compounds have IKACh-inhibiting properties (e.g. dronedarone), but the contribution of this component to overall efficacy is impossible to determine.47
The importance of atrial structural changes for the domestication of AF is increasingly being recognised. Atrial biopsies from patients with lone AF have consistently demonstrated increased fibrosis besides hypertrophy and inflammation.48 In fact, fibrosis is a hallmark of arrhythmogenic structural remodelling, leading to anisotropic impulse conduction with subsequently increased AF vulnerability.49 Tissue fibrosis results from an accumulation of fibrillar collagen and other extracellular matrix (ECM) proteins. Importantly, downstream effects of the renin–angiotensin–aldosterone system (RAAS) and its mediator transforming growth factor-β (TGF-β) are involved in the genesis of fibrosis. Angiotensin-II and TGF-β are peptide hormones that act together, stimulating fibroblasts to synthesise and secrete respective ECM proteins (see Figure 2). The RAAS is involved in myocardial fibrosis in AF-prone conditions such as hypertensive heart disease and congestive heart failure.
Attenuation and reversal of structural remodelling have increasingly moved into the focus of therapeutic innovation. The specific antifibrotic agent pirfenidone has been associated with greatly reduced AF vulnerability in an animal model of heart failure.50 Moreover, several angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs) that reduce fibrosis have been associated with improved AF indices in experimental models and clinical studies of patients with an indication for RAAS blockade treatment.51,52
Results from several small prospective clinical trials have indicated favourable effects of RAAS inhibition on the incidence of AF. There was a trend towards better sinus rhythm maintenance after electrical cardioversion with ACE inhibition compared with placebo in a trial of patients with AF in heart failure.53 Madrid et al. prospectively studied the effects of irbesartan in addition to amiodarone for sinus rhythm maintenance after electrical cardioversion.54 In this trial, AF relapses were reduced in patients receiving irbesartan. Another comparably designed trial consistently showed that the addition of an ACE inhibitor to amiodarone reduced the number of immediate and mid-term AF relapses after cardioversion.55 Interesting evidence from a small prospective study in patients with a first episode of lone AF indicated that a greater proportion of patients randomised to receive ramipril (without any indication for the substance) remained free from AF relapses during a follow-up of three years.56
By contrast, no such benefit was documented in a large cohort of AF ‘all-comer’ patients. The Italian Group for the Study of the Survival of Myocardial Infarction – Atrial Fibrillation (GISSI-AF) investigators failed to document any beneficial effect of valsartan treatment among 1,442 patients in sinus rhythm with previous AF. In this trial, the follow-up period was one year and rhythm was monitored with weekly trans-telephonic transmission. AF recurred in roughly half of the patients randomised to valsartan and a similar proportion of patients receiving placebo.57 Results of the large Atrial fibrillation Clopidogrel Trail with Irbesartan for prevention of Vascular Events (ACTIVE-I) study were recently presented at a scientific meeting. They showed that in a still larger cohort of patients with AF randomised to irbesartan (n=4,518) or placebo (n=4,498), irbesartan had no effect on the incidence of the composite end-point of vascular death, myocardial infarction or stroke.58 During a mean follow-up of four years, 5.4% of patients in both groups experienced such an event. Both GISSI-AF and ACTIVE-I may nevertheless under-represent the potential effects of RAAS inhibitory treatment, as a substantial proportion of patients was pre-treated with an ACE inhibitor. For instance, in GISSI-AF 57% of patients received such medication at baseline. However, subgroup analysis comparing patients receiving ACE inhibitors and those without did not reveal any difference in the primary outcome. Accordingly, the value of RAAS blockade in conditions associated with substantial pre-existing fibrotic changes is questionable and treatment should be initiated early in the course of underlying heart disease.
Inflammatory Changes, Endothelial Remodelling and Potential Therapeutic Intervention
Inflammatory changes in the atria of patients with lone AF have been documented by histological examination of atrial samples.48 Initiation of AF has been linked to systemic inflammation, and levels of circulating markers of inflammatory condition (such as C-reactive protein [CRP]) increase with longer persistence of the arrhythmia.59 Systemic inflammation and endothelial activation are tightly linked to each other in conditions such as coronary artery disease. Apparently, such an association also exists in AF patients. Serum markers of inflammation and endothelial activation were studied in detail in a small population of patients with AF and compared with matched control patients.60 In this study, levels of monocyte chemoattractant protein-1 (MCP-1), high-sensitivity CRP, intercellular adhesion molecule-1 and vascular adhesion molecule-1 (VCAM-1) levels were elevated during AF. The highest levels were documented in patients with atrial clot formation.
Among the factors studied, VCAM-1 and MCP-1 were independent predictors of the presence of atrial thrombi. The levels of circulating adhesion molecules and inflammatory markers remained elevated for weeks after successful cardioversion, which may imply a persistently increased risk of atrial thrombus formation.60 In a larger observational study, 278 patients were followed for a median of 32 months, and elevated levels of soluble VCAM-1 and metallo-matrix protease-2 at study entry were associated with untoward outcomes in terms of the combined end-point of all-cause mortality, myocardial infarction or embolisation.61 Anti-inflammatory approaches have proved effective in experimental animal and clinical trials. For instance, in a study of 104 patients after cardioversion of a first AF episode who were maintained on propafenone for relapse prevention, the addition of methyl-prednisolone was effective in reducing AF relapses over or placebo treatment during a median follow-up of ~24 months.62 In another study in 200 patients without previous AF undergoing bypass surgery for coronary artery disease, randomised treatment with 40mg/day atorvastatin (initiated one week prior to surgery) was associated with a significant reduction in the incidence of post-operative AF.63 While statin therapy might represent an adequate preventative measure for post-operative AF, treatment with steroids (with their inherent risk of side effects) does not represent a realistic treatment option for arrhythmia.
In line with these findings, experimental studies in pigs illustrated a prominent downregulation of endothelial nitric oxide synthase (eNOS), aggravating endothelial function and leading to upregulation of surface proteins associated with further endothelial damage.64,65 Treatment with the angiotensin receptor antagonist irbesartan attenuated cellular upregulation of adhesion molecules. While changes in eNOS were most prominent in the left atrium (providing a potential explanation for a predeliction to clot formation), alterations of coronary endothelium were also documented in pigs subjected to AF induced by rapid atrial pacing.66 Animals showed a significant reduction in coronary flow reserve as a marker of endothelial damage. Impaired coronary flow reserve was again reversible by treatment with irbesartan.66
Further work using this experimental model identified a potentially relevant effect of dronedarone. Dronedarone was similarly effective in reversing tachy-pacing-induced reductions in coronary flow reserve.67 This is of particular interest in terms of mechanistic explanation as acute coronary syndromes (ACS) were reduced by dronedarone treatment in A Placebo-Controlled, Double-Blind, Parallel Arm Trial to Assess the Efficacy of Dronedarone 400mg/bid for the Prevention of Cardiovascular Hospitalization or Death from Any Cause in Patients with Atrial Fibrillation/Atrial Flutter (ATHENA).26 The finding of reduced ACS has been thought-provoking in terms of causal explanations. Beyond the above-mentioned experimental mechanistic explanation, dronedarone was also associated with a significant reduction in arterial blood pressure, potentially further reducing the incidence of ACS.
With the clinical introduction of more novel antiarrhythmic drugs that are currently in development by several companies, the armamentarium for the treatment of AF is likely to be enhanced. This is highly necessary given the expectation that up to 30 million people will be affected by the disease in North America and Europe in the coming decades. While catheter ablation of AF will represent another important therapeutic avenue for selected patients, medical management is likely to remain the mainstay of therapy. Important new incentives for medical management are to be expected from other atrial-selective drugs and the establishment of dronedarone as a pharmacological means of reducing cardiovascular events in elderly patients with AF.