29 Atrial Fibrillation

Atrial fibrillation (AF) is such a common arrhythmia that it is often wrongly regarded as an acceptable alternative to normal sinus rhythm. Its first onset may present with rapid and uncomfortable palpitations, breathlessness, dyspnoea, chest pain, and anxiety. Often it is entirely asymptomatic and discovered quite by chance. Paroxysmal and persistent recurrences may eventually lapse into permanent AF.

The causes of AF are legion and should be identified since many can be corrected and the management of the arrhythmia can be simplified. The consequences of AF may be dire: heart failure, stroke, sudden death, markedly reduced exercise capacity, and degraded quality of life. Appropriate thromboprophylaxis, adequate rate control, and sucessful rhythm control in suitable patients are essential.

Stroke risk can be diminished by appropriate thromboprophylaxis with aspirin or vitamin K antagonists as indicated by systematic risk stratification; heart failure can be improved by competent management of underlying comorbid disease and proper rate control. Rhythm control with antiarrhythmic drugs, including beta-blockers, and left atrial ablation techniques may be needed to ameliorate symptoms and alleviate anxiety.

AF can be recognized on the electrocardiogram (ECG) by an irregular ventricular rhythm without consistent P waves ( Fig. 29.1) [1]. Epidemiological studies using short-term (often one single) ECG recording and symptoms have estimated the prevalence of AF in the population at 0.5–1% [2–5]. Continuous ECG monitoring of AF patients suggests that more than half of AF episodes are not detected by standard ECG recordings, even in symptomatic AF patients [6–8]. Hence, the ‘true’ prevalence of AF may be closer to 2% of the population [8]. As the population ages, it is expected that the number of patients with AF will double or triple in the next few decades ( Fig. 29.2).

The prevalence of AF increases with age. Less than 0.5% of 40–50-year-olds suffer from AF, while an estimated 5–15% of the population present with AF at the age of 80 years [3, 9–11] ( Fig. 29.3). Men are more often affected than women, and the age of diagnosis is later in women, but women with AF appear to be more prone to AF-related complications [4, 12]. The recurrent, initially often self-terminating nature of the arrhythmia [13] translates to a lifetime risk of AF around 25% in current 40-year-olds [14, 15]. Furthermore, in the Western world the age-adjusted prevalence of AF has increased over the past decades [2, 3, 10].

Similar to AF prevalence, its incidence appears to increase with age, is higher in men than in women [2], and has risen in the last two decades [10], varying from 1.6–1.0% per annum [11] to 0.54 cases per 1000 person years [2]. A high proportion of undiagnosed AF suggests a large potential overlap between newly detected ‘prevalent’ and truly ‘incident’ AF in the context of scheduled surveys using ECG recordings.

The prevalence and incidence of AF in non-white populations are less well studied, but lower values were reported for Asians and African Americans [16]. In the health survey of 664,754 US veterans, white men were significantly more likely to have AF compared to all races but Pacific Islanders (odds ratios vs. blacks, 1.84; vs. Hispanics, 1.77; vs. Asians, 1.41; vs. Native Americans, 1.15; p < 0.001) [17]. Whites were more likely to have valvular heart disease, coronary artery disease, and congestive heart failure; blacks had the highest hypertension prevalence; Hispanics had the highest diabetes prevalence associated with AF. Racial differences remained after adjustment for age, body mass index, and other comorbidities.

Clinically, four types of AF are distinguished based on the presentation and duration of the arrhythmia: first diagnosed AF, paroxysmal AF, persistent AF, and permanent (accepted) AF ( Fig. 29.4).

This classification is useful for clinical management of AF patients, especially when AF-related symptoms are also considered. A symptoms score (‘EHRA score’; Table 29.1) [8] provides a simple clinical tool to assess symptoms during AF. Combined with a stroke risk estimation, the symptom score and AF classification guide clinical decisions in AF patients.

In the majority of patients, AF is a progressive arrhythmia. An AF patient may experience asymptomatic, self-terminating episodes of the arrhythmia before AF is first diagnosed. The rate of recurrent AF is 10% in the first year after the initial diagnosis, and approximately 5% per annum thereafter. Comorbidities and age significantly influence the progression and complications of AF [4]. In a small group of highly selected patients with ‘lone AF’ without concomitant conditions that contribute to progression of the arrhythmia, AF remains paroxysmal over several decades [18].

When persistent AF is cardioverted, the recurrence rate is between 30–70% in the first month. Most recurrences occur in the first few weeks after termination of AF [6, 19]. The frequency and duration of arrhythmia recurrence tends to increase over time, and the distribution and duration of arrhythmia episodes is not random [20], but clustered ( Fig. 29.5) [8, 21, 22]. Hence, ‘total AF burden’ can vary markedly over months or even years in individual patients [22, 23]. Therefore reliable and valid assessment of AF recurrences is difficult in clinical practice, and especially in clinical trials. Asymptomatic AF is common even in symptomatic patients, irrespective of whether the initial presentation was persistent or paroxysmal AF, and has important implications for therapy (dis)continuation.

AF is associated with a variety of cardiovascular conditions that may perpetuate the arrhythmia [4, 12, 24, 25]. These conditions may also be markers for global cardiovascular risk [26] and/or cardiac damage rather than causative factors.

Hypertension ( Chapter 13) is the most common underlying condition in AF patients, found in approximately 2/3 of all AF patients in different surveys ( Fig. 29.6) [4]. Inadequate control of blood pressure is a risk factor for incident (first diagnosed) AF in hypertensive patients and for AF-related complications such as stroke and systemic thromboembolism, even in anticoagulated populations [27–29].

Antihypertensive treatment with angiotensin-converting enzyme (ACE) inhibitors or angiotensin receptor blockers (ARBs) is more effective in preventing incident AF than therapy with beta-receptor blockers or calcium antagonists in patients with left ventricular (LV) dysfunction or LV hypertrophy, despite similar reductions in blood pressure [30, 31] but less so in patients with well-controlled hypertension and without structural heart disease [32]. This is because the former may prevent atrial remodelling (cellular hypertrophy and atrial fibrosis) in patients prone to such changes (see ‘Upstream’ therapy, p.1113).

Heart failure (Chapter 23), defined as heart failure with dyspnoea on exertion (New York Heart Association (NYHA) classes II–IV) is found in 30% of AF patients [4, 12], and AF is found in 5–50% of heart failure patients [33]. The prevalence of AF clearly increases with the clinical severity of heart failure, with an AF prevalence of almost 50% in patients with NYHA IV heart failure ( Fig. 29.7) [34].

Heart failure can be either a consequence of AF (tachycardiomyopathy or decompensation in acute-onset AF) or a cause for the arrhythmia (by atrial pressure and volume overload, or chronic neurohumoral stimulation). In most patients, both conditions sustain each other to different degrees. Importantly, new-onset AF appears to be an independent predictor of death and prolonged intensive care unit stay in non-selected heart failure patients [35], and exercise performance is markedly reduced in patients with heart failure and permanent AF [36].

Of note, many pharmacological treatments that prevent death in heart failure trials (ACE inhibitors, beta-blockers) also prevent new-onset AF (see ‘Upstream’ Therapy, p.1113) [34]. The common myocardial damage mechanisms that precipitate AF and heart failure (see Chapter 23) may suggest that AF is an ‘atrial cardiomyopathy’. Diastolic LV dysfunction is also strongly correlated with the incidence of AF.

Tachycardiomyopathy is a form of LV dysfunction caused by AF: rapid irregular rate and loss of atrial contractile function results in severe LV dysfunction in selected patients with AF in the absence of structural heart disease. Similar effects have been described for other incessant tachycardias, including atrial and atrioventricular (AV) nodal tachycardia [37], and right ventricular outflow tract tachycardia [38]. Tachycardiomyopathy should be suspected when there is a fast ventricular rate and LV dysfunction but no signs for structural heart disease. Tachycardiomyopathy is confirmed when LV function is restored with adequate rate control or maintenance of sinus rhythm ( Fig. 29.8). Recovery of LV function may require several weeks. In some patients, only repeated measurements of LV function over time can distinguish between mild forms of ventricular cardiomyopathy complicated by AF and tachycardiomyopathy.

Valvular heart disease ( Chapter 21), especially mitral valve stenosis and regurgitation, induces left atrial (LA) pressure or volume overload that can provoke AF. Secondary to the improved earlier treatment of valvular heart disease and the prevention of rheumatic heart disease, severe mitral valve disease has become less common in Europe. The contribution of mild valvular heart disease to the development of AF is less clear, but some degree of valvular heart disease is found in approximately 30% of AF patients [4, 12].

Cardiomyopathies ( Chapter 18) including the primary electrical cardiac diseases ( Chapter 9 and 30) [39] carry a high individual risk for AF. In fact, sudden death and AF are clearly associated in epidemiological studies, suggesting the possibility of a common mechanism for both arrhythmias. The genetically determined cardiomyopathies comprise hypertrophic, dilated, and right ventricular cardiomyopathy. In addition ‘electrical cardiomyopathies’ such as long QT syndrome, short QT syndrome, Brugada syndrome, and catecholaminergic VT should be considered.

The high individual risk for AF at younger ages [40–43] probably explains why relatively rare cardiomyopathies are found in 10% of AF patients [4, 12]. Genetic studies suggest that a small proportion of patients with ‘lone AF’ are actually mutation carriers for ‘primary electrical’ cardiomyopathies [44].

With the possible exception of catecholaminergic ventricular tachycardia (VT), all genetically determined cardiomyopathies carry a risk for AF [41, 45]. In the long QT syndromes, ‘AF’ appears to be caused by after-depolarizations and prolongation [46, 47], rather than shortening of the action potential [48] (atrial ‘torsade’). Occasionally, ‘twisting’ of the P waves during atrial arrhythmias has been reported in long QT syndrome (29.1). In short QT syndrome, AF appears to be even more prevalent than ventricular arrhythmias [49, 50].

There is a high prevalence AF in Brugada syndrome ( Chapter 9) [42] but the mechanism is not delineated. It may be linked in some patients to mutations in the sodium channel (SCN5A) gene, and conduction slowing. In hypertrophic cardiomyopathy and arrhythmogenic right ventricular cardiomyopathy (ARVC) ( Chapter 18), a mixture of altered myocardial structure, changes in haemodynamic function, and primary electrical changes appear to cause AF [51, 52]. The reported association between ventricular pre-excitation or Wolff–Parkinson–White syndrome and AF (29.2) may be either due to the more severe symptoms of AF in patients with rapid conduction of atrial activity via the bypass tract, or due to a common genetic cause ( Table 29.4).

Atrial septal defects ( Chapter 10) are associated with AF in 10–15% of patients according to earlier surveys [40]. This association has important clinical implications in the antithrombotic management of stroke or transient ischaemic attack (TIA) survivors with an atrial septal defect. In patients with large shunt volumes and/or a high atrial load due to the septal defect, AF may be secondary to pressure and volume overload. In others, the association between AF and septal defects points to a common abnormality (e.g. a genetic predisposition) that predisposes to both ASD and AF.

Other congenital heart defects ( Chapter 10) are often associated with AF. Patients with single ventricles, atrial repair (Mustard operation) of a transposition of the great arteries, or after a Fontan procedure are at especially high risk for AF. Altered atrial anatomy, either due to the primary defect or secondary to repair surgery, and the ensuing altered atrial haemodynamics, may be one of the main predisposing factors for AF. AF can be highly symptomatic in these patients, and is often difficult to treat.

Coronary artery disease (CAD) ( Chapters 16 and 17) is present in 20% of the AF population. Studies that recruited patients in hospital usually have higher prevalence of coronary heart disease than when outpatients are enrolled, most likely secondary to the relatively high likelihood of in-hospital treatment of coronary heart disease [53]. Whether CAD per se predisposes to AF is not known, given that the association between AF and CAD may be mediated via heart failure in survivors of a myocardial infarction, or by shared risk factors, such as hypertension. AF is rare in patients with stable CAD and preserved LV function, but indicates those with a poor prognosis after a myocardial infarction [54].

Thyroid dysfunction is an important cause of AF. In recent surveys, only a relatively small percentage of the AF population (10% in Germany, a country with a high prevalence of goitre due to iodine deficiency) presents with apparent hyper- or hypothyroidism [4, 12]. In a population-based study of 5860 subjects aged 65 years and older, the biochemical finding of subclinical hyperthyroidism is associated with AF on the resting ECG, and even in euthyroid subjects with normal serum thyroid-stimulating hormone (TSH) levels, serum free T4 concentration is independently associated with AF [55].

Obesity is found in 25% of AF patients [4]. In meta-analysis of the population-based cohort studies, obese individuals have an associated 49% increased risk of developing AF compared to non-obese individuals [56]. Obesity is associated with hypertension, sleep apnoea, chronic obstructive airways disease, and diabetes mellitus, among others, and most likely is a marker for cardiovascular risk that associates, among other events, with AF.

A proportion of patients with metabolic syndrome (see Chapter 15) develop AF. An apparent correlation between the presence of metabolic syndrome and increased susceptibility to AF during a mean follow-up of 4.5 years has recently been demonstrated in a large (>28,000 subjects) community-based cohort in Japan [57]. Again, this association may reflect that metabolic syndrome summarizes several of the known conditions that predispose to AF.

Sleep apnoea is associated with an atrial pressure rise and dilatation and may predispose to AF [58]. The retrospective cohort study of 3542 Olmsted County adults who were referred for a diagnostic polysomnogram, has found that obesity and the magnitude of nocturnal oxygen desaturation, which is an important pathophysiological consequence of sleep apnoea, were independent risk factors for new incident AF in individuals <65 years of age [59].

Tall stature is associated with AF [60]. The mechanisms are unclear, but mechanical distension of pulmonary veins (PVs) and LA dilatation are possible.

Diabetes mellitus, when defined as a medical condition that requires treatment, is found in 20% of AF patients. It is conceivable that badly controlled diabetes mellitus may contribute to atrial cell death and ‘structural remodelling’, and thereby contribute to the perpetuation of AF [61].

Chronic obstructive pulmonary disease is found in 10–15% of AF patients [62]. In some patients, pulmonary disease may rather be a marker for cardiovascular risk in general than a specific predisposing factor for AF. Cancer of the bronchus may also present with AF.

Chronic renal disease (see Chapter 15), measured as renal dysfunction, is present in 10–15% of AF patients. Renal failure appears to increase the risk for AF-related complications, especially cardiovascular complications. Renal failure, diabetes mellitus, and chronic obstructive pulmonary disease are more prevalent in patients with permanent AF.

During AF, regulation of heart rate by the sinus node is lost, and the ventricular rate is regulated by the conduction properties of the AV node (‘arrhythmia absoluta’). The reduction of cardiac output is aggravated by loss of atrial contractile function. The latter leads to abnormal atrial stasis which, in association with abnormal blood constituents and endothelial alterations, fulfils Virchow’s triad for thrombogenesis.

Through these main mechanisms, AF has important effects on health by causing or associating with death, stroke and other thromboembolic events, reduced quality of life and exercise capacity, and LV dysfunction. Due to these severe complications and its high prevalence, AF causes substantial cardiovascular morbidity on a population level. Table 29.2 lists rates of relevant outcome events reported in recent large AF trials.

Death rates are doubled in association with AF in epidemiological studies. This is independent of other, known predictors of death ( Table 29.3) [63–66]. Hence, the prevention of AF-related death is important, although difficult. The biological processes that cause AF-related deaths are not fully understood. So far, only antithrombotic therapy and treatment with dronedarone (see Dronedarone, p.1107) has been shown to affect AF-related deaths and reduce thromboembolic strokes [67]. Patients without structural heart disease (lone AF), particularly who were <60 years at the time of AF diagnosis, appear to have a risk of death similar to that in a general population after 20–30 years [18].

Stroke is the most devastating consequence of AF ( Fig. 29.9) [68–72]. Approximately every fifth to sixth stroke is due to AF [71, 73]. Strokes in AF patients are markedly more severe, more often result in death or permanent disability than strokes of other origin, and recur more commonly [68, 74, 75]. Paroxysmal AF carries the same stroke risk as permanent or persistent AF [76, 77]. Because much AF is asymptomatic, a number of so-called ‘cryptogenic strokes’ may be due to undetected (often paroxysmal) AF [78]. Cerebral micro-emboli may cause cognitive dysfunction in patients with AF in the absence of clinically overt stroke [79] and may contribute to the degraded quality of life in AF patients.

Quality of life and exercise capacity are impaired by AF in patients with AF-related symptoms. Patients with AF have significantly poorer quality of life than healthy controls, the general population, or patients with CAD in sinus rhythm [80]. Although many patients present with ‘accidentally diagnosed’ and/or asymptomatic AF, AF is one of the most common reasons for hospitalizations [81]. Even patients with otherwise asymptomatic AF report lower quality of life compared with subjects in sinus rhythm [82]. Adequate control of ventricular rate and successful maintenance of sinus rhythm can improve quality of life and exercise capacity in AF patients [83, 84].

LV function is often impaired by the irregular, fast ventricular rate and by loss of atrial contractile function and reduced end-diastolic LV filling in patients with AF [8, 36, 85–87]. Either adequate rate-control therapy or maintenance of sinus rhythm [87–90] can improve LV function in AF patients. This effect is most visible in patients with a first episode or recent-onset AF. Drugs that prevent progression of LV dysfunction and heart failure can prevent AF (see ‘Upstream’ therapy, p.1113) [34].

The genesis of AF is complex and multifactorial, unlike many other supraventricular arrhythmias, and is comparable only to the genesis of ventricular fibrillation and sudden death.

The diagnosis of AF is based on irregular ventricular rate and the loss of discernible P waves in the surface ECG ( Fig. 29.1) [91]. This apparently simple yet often missed ECG-based diagnosis [92] lumps together arrhythmias with completely or partially different mechanisms of initiation and/or maintenance. The ECG pattern of AF can be created by multiple, rapid repetitive atrial foci, single, rapid atrial foci with fibrillatory conduction [93], or re-entrant activity maintained via several simultaneous electrical wave fronts [94].

The simultaneous existence of multiple foci or a single wandering re-entrant electrical wave (‘mother rotor’) is possible, although less likely in clinical and experimental studies ( Fig. 29.10). It has been assumed that multiple re-entrant wave fronts maintain AF in the majority of patients [95], but several ‘drivers’ may be present. Multiple wavelet re-entry requires several (at least two, but in most measurements four to eight) simultaneous meandering wavelets which maintain each other by creating areas of functional conduction block and slow conduction.

One condition that facilitates functional re-entry is a short ‘wavelength’ of the electrical activation waves, i.e. a short product of its atrial conduction velocity and refractory period. In normal atria activated during sinus rhythm, the wavelength is longer than the size of the atria, thereby precluding re-entrant activation. During AF, the conduction velocity of activation waves is markedly slowed and the action potential duration and refractory period are shortened. These factors shorten atrial wavelength to an extent that allows multiple areas of functional re-entry.

In concordance with the ‘wavelength’ concept, factors that either shorten the atrial refractory period or slow conduction (i.e. parameters that reduce atrial wavelength) contribute to the maintenance of AF. In long-standing AF or in atria with marked structural changes and conduction inhomogeneities, the number of simultaneous wavelets may be higher and their three-dimensional distribution can be complex [96], possibly secondary to increased ‘electrical isolation’ between cardiomyocytes and separation of atrial cardiac tissue by fibrotic tissue. Although not proven so far, the available experimental data and many aspects in the epidemiology and treatment of AF can be readily explained by the multiple wavelet hypothesis ( Fig. 29.10).

The aetiology of AF is complex, and several different factors contribute to changes that promote focal atrial activity or one of the different forms of re-entry in the atria. Several experimental models have helped to dissect AF-induced pathophysiological changes that contribute to the perpetuation of AF. Fig. 29.11 summarizes several of the known vicious circles that contribute to the initiation and perpetuation of AF. In a given patient (or a given experimental model), different pathophysiological changes may have more or less impact. These vicious circles are important because their interruption by therapeutic interventions (drugs, lifestyle changes, or catheter ablation) is probably the main determinant of treatment success for rhythm-control therapies.

Focal ectopic activity in the PVs was first described in patients with AF [93]. Especially in the early phases of AF and in patients with so-called ‘lone AF’ (i.e. AF without concomitant conditions that might explain the arrhythmia), AF episodes are often initiated by rapid focal electrical activity originating from within the PV, the junction between PVs and posterior LA, and to a lesser extent in other regions of the venous parts of the atria [93]. Abnormal calcium release may be driving these focal discharges [97], although re-entry within the PV myocardium may also contribute to the initiation of ‘focal activity’ in the PVs [98]. Ablation therapy of AF which targets these focal sources has proven effective in its prevention (see Catheter ablation strategies, p.1121) [99].

The process of AF-induced ‘electrical remodelling’ was first described in goats with pacing-induced AF ( Fig. 29.12) [100]. In essence, the atrial myocardium responds to the high atrial rate in the fibrillating atria by shortening of atrial refractoriness and atrial action potential duration. This is most likely a ‘cellular survival mechanism’ that helps to extrude excess intracellular calcium from the rapidly activated atrial myocardial cells [101]. On the molecular level, an altered expression and regulation of proteins involved in potassium and calcium channels reduces calcium currents and increases potassium currents ( Fig. 29.13) [102, 103].

In addition to acceleration of repolarization (action-potential shortening), the resting membrane potential is more positive in chronic AF, most likely secondary to a constitutive activation of I_KACh [104]. However, shortening of the atrial action potential not only preserves viability during AF, but also renders the atria prone to recurrent AF. The atrial action potential normalizes in the first few weeks, and possibly even in the first few days, after restoration of sinus rhythm.

The concept of electrical remodelling provides a pathophysiological rationale for the use of drugs that prolong the atrial action potential to prevent recurrences of AF after cardioversion, i.e. in atria that underwent the process of electrical remodelling [105–108]. Time-dependent recovery from electrical remodelling in sinus rhythm suggests that therapy with ion channel-blocking agents may be stopped once electrical remodelling has been reversed [108].

Continuous electrical reactivation of atrial cardiomyocytes results in reduced duration of electrical diastole, increased calcium influx through the sarcolemmal membrane,

increased calcium release from the sarcoplasmic reticulum, and consequently a rapid intracellular calcium overload in the atrial myocardium [109]. The resulting altered expression and function of calcium-handling proteins and the ensuing altered regulation of the contractile apparatus reduces atrial contractile function and increases atrial size [110]. These changes persist for several weeks after restoration of sinus rhythm (atrial stunning)[111]. Contractile dysfunction suggests an important role for intracellular calcium overload in the maintenance of AF [112] and has also been regarded as a reason for anticoagulation in the first weeks after cardioversion [1].

The relatively high AF recurrence rate in the active-treatment arms of ion channel-blocking drug trials already suggests that electrical remodelling is not the only factor that maintains AF and/or provokes its recurrence. AF induces an increased expression of extracellular matrix proteins and increased atrial fibrosis ( Fig. 29.14) [113]. These changes result in conduction slowing in the atria, and in electrical isolation of atrial cardiomyocytes, even prior to development of AF [114]. The molecular signalling pathways that mediate these structural changes are not fully understood, but the angiotensin system is activated, sympathetic stimulation is increased, a ‘fetal’ gene expression programme is activated, and thrombogenic molecules are expressed in the endothelium in AF [115]. Inhibition of either the ACE or the angiotensin II receptor in the heart can prevent some of these AF-induced changes [116].

The lack of intracellular ATP donators, possibly due to mitochondrial dysfunction, may contribute to extracellular matrix formation, cellular dysfunction, and potentially to cell death, and de-differentiation of atrial myocardium [115, 117]. This is one cause of major metabolomic changes in atrial tissue from AF patients [118] and animal models of AF [119]. The antifibrillatory effect of ACE inhibitors, ARBs, and statins is most likely explained by the effects of these drugs on AF-induced structural changes. Whether inflammation of the atrial myocardium [120, 121], identifiable by increased content of inflammatory cells in atrial tissue, and in part by increased C-reactive protein (CRP) levels in the blood of patients with AF [122, 123], is an independent process or not, is unknown.

The prevalence of AF is clearly age-dependent (see Prevalence of atrial fibrillation, p.1070). Ageing is associated with an increased isolation of cardiomyocytes through altered expression of connexins and increased development of fibrous septa between atrial muscle fibres [124]. Denatured atrial natriuretic peptide creates amyloid-like deposits that also lead to fibre separation. AF-induced mitochondrial DNA deletions which are associated with altered mitochondrial energy production are comparable to the DNA alterations found in ageing myocardium [125]. In this regard, ageing is associated with some of the changes that are found as a structural consequence of AF. Conversely, the structural changes found in AF may be a form of accelerated ‘AF-induced’ ageing of the atrial myocardium. A mild form of genetically conferred accelerated ageing (progeria) is also associated with AF [126].

Intracellular calcium overload due to the loss of electrical diastole occurs even after short episodes of AF. Calcium has several major roles in the cardiomyocyte, including initiation and regulation of cellular contraction by binding to troponins, regulation of cellular proteins, protein kinases, and subsequent regulation of intracellular signalling pathways, and regulation of sarcolemmal and sarcoplasmic reticulum electrical function. Intracellular calcium overload may be the initial event that triggers some of the known AF-induced electrical [127], contractile [128], structural [129], metabolomic, and inflammatory changes [130].

In some patients, especially in patients without concomitant conditions, focal sources of electrical activity (often in the PVs) are the most likely cause of the first AF episode. Some of these patients, however, never develop sustained AF [18], while others present with sustained AF from the start. The likelihood of sustained forms of AF is almost linearly related to the number of predisposing conditions in the majority of (often elderly) patients with AF. Many of these factors, e.g. hypertension, diabetes and obesity, or heart failure, are present prior to the initial diagnosis of AF. The factors that precede AF may provide attractive therapeutic targets for the preventative treatment of the arrhythmias [131].

Some of the pathophysiological responses to ‘atrial stressors’ that predispose to AF have been delineated in experimental studies: dogs in which ventricular pacing induces heart failure are prone to conduction slowing and inducible AF [132, 133]. In a sheep model of early-onset arterial hypertension, inducible AF was associated with electrical isolation of atrial cardiomyocytes, increased formation of extracellular matrix, and conduction disturbances without evidence for action-potential shortening [114]. Chronic dilatation of the atria secondary to chronic pressure and/or volume overload of the atria may induce these structural changes that appear to precede AF, possibly related to AF in the setting of severe valvular heart disease or heart failure [12, 134, 135].

High-level endurance sports may be sufficient to predispose to AF, most likely via chronic atrial dilatation [58, 136]. Atrial ischaemia may also contribute to the initiation of AF. In another study in dogs with ‘spontaneous’ sustained AF, action-potential shortening was found prior to AF, suggesting that these electrical changes can also form prior to AF [109]. It is conceivable that many of the vicious circles mentioned earlier ( Fig. 29.11) are activated prior to AF and contribute to its initiation. Retrospective analyses from randomized studies [34] suggest that structural changes that precede AF may be a reasonable target for primary prevention of AF.

AF, especially early-onset AF, can be clustered in families, and associations between familial AF and the first gene loci were described over a decade ago [137]. The past few years have brought about many reports of different genetic alterations in atrial fibrillation ( Table 29.4).

Sudden death and ventricular fibrillation in the young are often caused by inherited cardiomyopathies [138]. Many of the proteins that are mutated in these diseases are also expressed in the atria, and the same electrical abnormalities which are arrhythmogenic in the ventricles can cause AF. Patients with short QT syndrome, i.e. a genetic cause for short cardiac action potentials and a QT interval <0.33s, are at high risk for AF, e.g. detected by inappropriate defibrillator discharges [49, 50]. The first identified mutation found in a family with AF affects a gene that is also altered in the short QT syndrome (KCNQ1) [48]. Short QT syndrome can probably be considered a ‘genetic’ cause of atrial action potential shortening (see Electrical remodelling, p.1079).

More recently, several groups have found an increased risk of AF in patients with long QT syndrome [41, 43, 46], polymorphisms associated with QT prolongation co-associate with AF [139], and mutations in long QT syndrome-causing genes are found in approximately 5% of patients with ‘idiopathic’ AF [44]. The limited available data suggest that prolongation of the atrial action potential and after-depolarizations cause AF in these patients [46] (29.1). Similar mechanisms may be present in patients with a nonsense mutation in the Kv1.5 gene responsible for the ‘atrial-specific’ potassium current I_Kur [135].

The pathophysiological factors that cause AF and other supraventricular arrhythmias in Brugada syndrome are less well established, but the prevalence of AF is high [42]. It is conceivable that the combination of a ‘Brugada-type’ ECG with AF is indicative of patients with a sodium channel mutation [44, 140], and that either conduction slowing in the atria [141–143] and/or prolongation of the atrial action potential and after-depolarizations (‘atrial torsade de pointes’) [46, 144] cause AF in such patients.

Sodium channel mutations have also been identified in patients with ‘lone’ AF [44, 145] and in a large family suffering from heart failure and atrial fibrillation [146]. Mutations in the cardiac ryanodine receptor, the calcium release channel in the sarcoplasmic reticulum, have been associated with AF and catecholaminergic VT [147]. Whether AF in hypertrophic cardiomyopathy is mediated via electrical changes, e.g. secondary to altered calcium handling by the mutated sarcomeric proteins, or a consequence of the increased atrial pressure and volume load in obstructive forms of the disease, has not been convincingly studied. A small group of patients suffer from a genetic abnormality in the PRKAG protein. Clinically, these patients carry a combination of AF, ventricular pre-excitation, and atypical LV hypertrophy [148, 149].

Somatic, i.e. ‘heart-restricted’ mutations in connexin 40 have been identified in a small series of patients in whom atrial tissue could be genetically studied [150], and a frameshift mutation in the gene coding for the atrial natriuretic peptide (ANP) [151] has recently been identified in patients with ‘lone’ AF. Hence, either genetically conferred conduction disturbances or genetically-conferred altered ANP signalling and/or hypertension may cause AF.

In addition to these genetic alterations with more or less known functional consequences, a population-wide study has associated a common variant in the PITX2 gene with AF [152]. This finding was replicated in other cohorts. PITX2 is an important regulator of cardiac development including formation of the LA and pulmonary tissue, and a part of the PVs [153, 154]. Another study found an association with TBX5, another regulator of cardiac development, and AF in patients with the Holt–Oram syndrome [155], and a transgenic model with enhanced atrial activity of TGFβ1 predisposes to AF, atrial fibrosis and conduction slowing [156–159].

An irregular pulse should always raise suspicion of AF. The ECG diagnosis of AF follows simple criteria: AF is characterized by complete irregularity of RR intervals (‘arrhythmia absoluta’) and loss of discernible P waves ( Fig. 29.1). As the anterior right atrium and the right atrial appendage often show organized atrial activation during AF, small atrial waves may be present in the right precordial leads in AF. Any arrhythmia that has the ECG characteristics of AF and lasts for 30s or longer should be considered AF [8], and episodes of at least 5-min duration are associated with an increased mortality in retrospective analyses [160]. The risk of AF-related thromboembolic complications (e.g. stroke) is not different between short AF episodes and sustained forms of the arrhythmia [76].

Several supraventricular arrhythmias, most notably atrial tachycardias and atrial flutter, but also forms of frequent atrial ectopy, may conduct to the ventricles producing rapid irregular RR intervals, thereby mimicking AF ( Fig. 29.15A–C). Usually, AF has an atrial cycle length of <200ms [46, 161, 162], while most atrial tachycardias and flutters are slower. This may be helpful for the differentiation of AF from atrial flutter or atrial tachycardia. Atrial cycle length during AF may be longer in patients receiving conduction-slowing or action potential-prolonging drugs, e.g. sodium channel blockers, sotalol, or amiodarone.

The differential diagnosis of slow AF includes severe sinus node dysfunction with changing ectopic pacemakers and unusual forms of higher degree AV block. AF may coexist with complete heart block, particularly in elderly patients, when the ventricular escape rhythm is often regular ( Fig. 29.15D). These conditions can usually be discerned from the 12-lead ECG by a careful search for occasional, but definite, P waves which excludes the presence of AF.

To differentiate the common diagnosis of AF from the rare other rhythms with irregular RR intervals, a 12-lead ECG during the arrhythmia is usually needed. Therefore, any episode of suspected AF should be recorded by a 12-lead ECG of sufficient duration and quality to evaluate P waves. Occasionally, especially when the ventricular rate is high, AV nodal blocking techniques such as a Valsalva manoeuvre, carotid massage, or intravenous adenosine administration during the ECG recording can help to unmask P waves.

Silent AF may be found incidentally during routine physical examinations, pre-operative assessments, occupation assessments, or population surveys. The prevalence of sustained silent AF in epidemiology studies is believed to be about 25–30% of AF patients [163].

In some cases, silent AF is diagnosed only after a complication such as stroke or heart failure has occurred. Thus, in the Framingham Study database, AF was found incidentally in about 18% of admissions for stroke and subsequently diagnosed within 2 weeks in another 4.4% [164]. Therefore, a thorough screening for AF is needed in patients with such complications. Given that detection of AF has therapeutic consequences (e.g. anticoagulation), prolonged Holter ECG monitoring is recommended in such patients with suspected asymptomatic AF.

Wider use of implantable rhythm-control devices, such as pacemakers and cardioverter defibrillators, has shown that a significantly greater proportion of patients has silent AF than was previously thought. About 50–60% of patients may have unsuspected episodes of the arrhythmia, almost half of which last >48 hours [165]. The use of implanted monitors, ECG garments, or patient-operated ECG systems may allow expansion of the monitoring period for suspected AF [8].

The acute management of AF patients aims to ameliorate symptoms and to estimate AF-associated risk. This can be remembered by the acronym SHS (Symptoms, Heart rate, Stroke risk assessment). Asymptomatic or mildly symptomatic AF does not require urgent diagnostic or therapeutic steps, but should result in a thorough risk evaluation. The EHRA symptom score and CHADS₂ stroke risk score should be estimated ( Table 29.1). A thorough clinical evaluation is recommended to exclude other causes of acute dyspnoea. Symptomatic patients may require urgent rate control or cardioversion.

A complete diagnosis of AF includes analysis of associated medical conditions (see Predisposing clinical conditions, p.1073) and stroke risk factors. This will guide arrhythmia therapy as well as antithrombotic treatment. In addition, the clinician should assess bleeding risk if antithrombotic treatment is indicated. In symptomatic patients, vagal, adrenergic, and other potential trigger mechanisms should be considered. AF should be classified as first onset, paroxysmal, persistent, or permanent ( Fig. 29.4). AF may be symptomatic and asymptomatic in the same patient at different imes [163]. Symptoms may relate to the arrhythmia itself, or to the complications of AF, or the associated medical condition.

Vagal AF [166] is a paroxysmal arrhythmia that usually occurs in the evening, at night, or during weekends, particularly after heavy meals and possibly alcohol consumption. Usually sinus bradycardia or sinus pauses precede the development of the AF and during the AF the heart rate is relatively slow.

Adrenergic AF, on the other hand, is less common, occurs during the day and is provoked by physical or mental stress, and may sometimes be triggered by a stress test. An acceleration of sinus rhythm usually anticipates the arrhythmia and the ventricular rate during the AF tends to be rapid. Some patients have both autonomic forms of AF and many paroxysms are not classically one or the other.

AF presents with an irregular pulse, irregular jugular venous pulsations, and variations in the loudness of the first heart sound and in systolic blood pressure. There may be a deficit between number of heart beats and number of peripheral pulsations, especially when the ventricular rate is rapid. Signs of valvular heart disease, ventricular dilatation, and heart failure may be found.

The 12-lead ECG ( Chapter 2) should, in addition to the detection of AF, also be screened for signs of acute myocardial infarction and other signs for structural heart disease (e.g. prior myocardial infarction, LV hypertrophy, bundle branch block or ventricular pre-excitation, signs of cardiomyopathy or ischaemia). In sinus rhythm LA conduction delay is often present, and rare cardiac abnormalities may occasionally be found (29.4).

The ECG may also show other abnormal sinus node function, atrial arrhythmias ( Fig. 29.15A–C) that may trigger AF, or ventricular arrhythmias that may be a sign of heart disease. It is always important to distinguish aberrant conduction from VTs, especially when using antiarrhythmic drugs ( Fig. 29.16).

The chest radiograph may help to reveal cardiac enlargement ( Fig. 29.8) but is most helpful in detecting intrinsic pulmonary disease and abnormalities of the pulmonary vasculature as in heart failure and pulmonary hypertension.

Echocardiography should be performed at least once in all AF patients. It is indispensable for the management of AF-associated medical conditions. In addition, markers of thrombosis (spontaneous echo contrast) or even a thrombus may be found. Risk markers for stroke include LV hypertrophy or dysfunction, LA enlargement, and low-flow velocities in the LA appendage (LAA). Also, aortic plaques are associated with a high stroke risk. New technical developments have improved the performance of transthoracic echocardiography but for detection of thrombi (e.g. for echo-guided cardioversion) transoesophageal echocardiography is the current standard [1] ( Fig. 29.17; 29.5).

Laboratory evaluation may be limited to thyroid function tests, serum electrolytes, haemoglobin levels, a serum creatinine measurement, analysis for proteinuria, and a test for diabetes mellitus (usually a fasting glucose measurement). Markers of heart failure (BNP) or inflammation (CRP) or infection may be useful. Thyroid and hepatic function should be assessed when treating patients with amiodarone.

Holter monitoring or event recorders may be used in order to establish the diagnosis of AF. In patients with an implanted device (pacemaker, implantable cardioverter defibrillator, or implantable loop recorder) the logging functions in the device may be very helpful. If patients remain symptomatic with palpitations or reduced exercise tolerance despite normal resting heart rate, the adequacy of rate control may be checked with Holter (or exercise testing in fit patients). AF may start during sinus bradycardia or tachycardia, with single or bouts of atrial ectopic beats, while the start of AF with a supraventricular tachycardia or transitions between AF and atrial flutter may also be seen. Identification of these initiating mechanisms may guide treatment, e.g. vagolytic or beta-blocking agents, atrial pacing or catheter ablation of focal AF or other initiating arrhythmias (see Pacemaker and defibrillator therapy, p.1115). Although the AF burden (total duration or percentage of time in AF) can be measured, its clinical relevance is uncertain.

Exercise testing is useful in patients with permanent AF who remain symptomatic despite adequate rate control at rest. The test may reveal an excessive heart rate rise during the lower stages of exercise, thereby limiting exercise tolerance due to dyspnoea, fatigue, or palpitations. In these cases, rate-control drugs may be targeted to the heart rate during a lower level of exercise. Exercise testing may be useful in detecting ischaemic heart disease, which has consequences for antiarrhythmic treatment of AF. Finally, exercise testing may be used to evaluate the safety of antiarrhythmic drug treatment, e.g. for detecting excessive QRS widening on class IC drugs.

Electrophysiological evaluation may be needed in selected cases, especially in patients in whom other arrhythmias, sick sinus syndrome, or a focal origin is suspected. Many of these patients may receive catheter ablation or a pacemaker.

The risk of stroke- and thromboembolism-related AF has long been recognized. Independent studies, performed before the time when anticoagulation became recommended for these patients, verify a relative risk increase of 2.3–6.9 for patients in AF without signs of rheumatic mitral valve disease (so-called non-valvular AF) compared with arrhythmia-free controls [167]. In valvular AF, for example when AF is related to rheumatic mitral valvular disease, this risk of thromboembolism is increased 17-fold [168]. The risks of thromboembolism in non-valvular AF are not homogeneous, being related to the presence of other clinical risk factors.

Cohort data from one epidemiological study (Framingham) and non-warfarin arms of clinical trials have identified clinical and echocardiographic risk factors that can be related to an increased risk of stroke [167]. However, these risk factors represent what was prospectively documented in these studies, as (for example) peripheral artery disease was not systematically documented in the clinical trials.

The systematic review conducted as part of the United Kingdom National Institute for Health and Clinical Excellence (NICE) national clinical guidelines for AF management, identified a history of stroke or TIA (or thromboembolism), older age, hypertension, and structural heart disease (LV dysfunction or hypertrophy) as good predictors of stroke risk in AF patients [169]. In this overview, the evidence regarding diabetes mellitus, female gender, and other characteristics were less consistent in the AF population per se, although diabetes is regarded as an important risk for stroke generally.

In a systematic review of stroke risk factors in AF by the Stroke Risk in Atrial Fibrillation Working Group, prior stroke/TIA/thromboembolism (relative risk (RR) 2.5, averaging 10% per year), increasing age (RR 1.5 per decade), a history of hypertension (RR 2.0), and diabetes mellitus (RR 1.7) were found to be the most consistent independent risk factors [170]. Again, female sex was inconsistently associated with stroke risk, whereas the relevance of heart failure or CAD was considered ‘inconclusive’.

Patients with paroxysmal AF should be regarded as having a similar stroke risk compared to persistent and permanent AF, in the presence of risk factors [171]. Patients with ‘lone AF’, that is, those with non-valvular AF aged <60 years and who have no clinical history or echocardiographic evidence of cardiovascular disease carry a very low cumulative stroke risk, estimated to be 1.3% over 15 years [18, 172]. In this group, cerebrovascular events occurred at the same rate in patients with a paroxysmal form of AF and in patients who progressed to permanent AF. All patients with lone AF who had a cerebrovascular event had developed at least 1 risk factor for thromboembolism (hypertension, heart failure, or diabetes) and the majority were not taking antiplatelet agents or anticoagulants at the time of stroke. Thus, the probability of stroke in young patients with lone AF appears to increase only after many years of disease (at least 25 years), with advancing age or development of hypertension. These observations emphasize the importance of re-assessment of risk factors for stroke over time.

The presence of moderate–severe LV systolic dysfunction on transthoracic echocardiography is the only independent echocardiographic risk factor for stroke on multivariate analysis [173]. On transoesophageal echocardiography, the presence of LA thrombus ( Fig. 29.17), complex aortic plaques, spontaneous echo-contrast, and low LAA velocity on transoesophageal echocardiography have been suggested as predictors of stroke and thromboembolism [168].

Table 29.5 presents the risk categories for stroke or systemic embolism for the patient with non-valvular AF and additional risk factors.

‘Definitive’ risk factors (previously referred to as ‘high risk’ risk factors) are those that have been associated with an increased risk of stroke and thromboembolism, such as previous stroke, or TIA, or thromboembolism, the elderly (aged ≥75), or valvular heart disease (mitral stenosis or prosthetic heart valves).

‘Combination’ risk factors (previously referred to as ‘moderate risk’ risk factors) are heart failure (especially moderate–severe LV dysfunction, defined arbitrarily as ejection fraction ≤40%), hypertension, or diabetes. Note that risk factors are cumulative, and the simultaneous presence of two or more ‘combination’ risk factors would justify a high enough stroke risk to require anticoagulation.

Less validated ‘combination’ risk factors (previously referred to as ‘less validated risk factors’) have a less robust evidence-base link for stroke and thromboembolism risk, and include female gender, age 65–74 years and vascular disease (specifically, myocardial infarction, as well as complex aortic plaque and peripheral artery disease). The available evidence is controversial as to whether thyrotoxic AF is an independent risk factor for stroke compared to other causes of AF. Thus, antithrombotic therapies should be chosen based on the presence of validated stroke risk factors.

The identification of stroke clinical risk factors has led to the publication of various stroke risk schemes. The simplest and most validated is the CHADS₂ (Cardiac failure, Hypertension, Age, Diabetes, Stroke doubled) score, as shown in Table 29.1. The CHADS₂ risk index is based on a point system in which 2 points are assigned for a history of stroke or TIA and 1 point each is assigned for age >75 years, a history of hypertension, diabetes, or recent cardiac failure. As shown in Table 29.6, there is a clear relation between CHADS₂ score and stroke rate [174]. The original validation of this schema classified a CHADS₂ score of 0 as low risk, 1–2 as moderate risk, and >2 as high risk.

The Stroke in AF Working Group [175] performed a comparison of 12 published risk stratification schemes to predict stroke in patients with non-valvular AF and concluded that there were substantial, clinically relevant differences among published schemes designed to stratify stroke risk in patients with AF. Most had modest predictive value in predicting stroke (c-statistic of approximately 0.6).

Similarly, it has been reported that the published stroke risk schemes had only fair discriminating ability, with c-statistics ranging from 0.56–0.62 [176]; also, the proportion of patients assigned to individual risk categories varied widely across the schemes, where the proportion considered high risk ranged from 16.4–80.4%. The CHADS₂ score categorized most subjects as ‘moderate risk’ and had a c-statistic of 0.58 to predict stroke in the whole cohort.

Nonetheless, especially given the observed rates of inadequate oral anticoagulation in AF patients, the CHADS₂ score is currently the most validated, simple system to give some initial estimate of stroke risk in AF patients. Other, as of now, less validated, risk factors for stroke, and estimators for bleeding risk, need to be applied with clinical judgement to decide optimal antithrombotic therapy in patients at ‘intermediate risk’ for stroke.

The risk of stroke or systemic embolism in patients with non-valvular AF is linked to a number of underlying pathophysiological mechanisms [177].

‘Flow abnormalities’ in AF are evident by stasis within the LA, with reduced LAA flow velocities and visualized as spontaneous echo-contrast on transoesophageal echocardiography. The LAA is a blind pocket, and inside, it is markedly trabeculated ( Fig. 29.18A). The more the outflow velocity from the LAA decreases, the higher the risk for development of thrombi within its cavity [178]. The LAA volume has been measured by a cast technique in a necropsy study and found to vary between individuals, ranging between 0.7–19.2ml ( Fig. 29.18B) [179]. There is also marked variability in the size of the LAA orifice (5–27mm) and the maximal diameter (10–40mm). The LAA of subjects with verified AF had generally larger dimensions than in those known to have been free from the arrhythmia. Persistence and permanence of AF is followed by structural remodelling [180], further depressing the function of the LAA.

‘Vessel wall abnormalities’—essentially, anatomical and structural abnormalities—in AF include progressive atrial dilatation, endocardial denudation, and oedematous/fibroelastic infiltration of the extracellular matrix. Indeed, the LAA is the dominant source of embolism in AF patients [181] being the embolic source in 91% when AF is of non-valvular origin. Fig. 29.17 illustrates a transoesophageal view of an LAA hosting a thrombus.

Finally, ‘abnormalities of blood constituents’ are well described in AF, and include haemostatic and platelet activation, as well as inflammation and growth factor abnormalities [177]. The presence of this ‘triad’ of abnormalities of blood flow, vessel wall abnormalities, and abnormalities of blood constituents, results in the fulfilment of Virchow’s triad for thrombogenesis, and are in keeping with a prothrombotic or hypercoagulable state in AF, as first proposed in 1995 [182].

An embolus originating from a fibrillating LA may follow the bloodstream to any part of the body. However, the incidence of stroke and of a clinically evident systemic embolism in non-valvular AF differs markedly from the proportion of blood flowing to the brain and to the rest of the body. Thus, the stroke rate is ten times higher than the rate of systemic embolism in patients with non-valvular AF who have not received any anti-thrombotic treatment. In one study of patients with non-valvular AF with incident peripheral embolism, small emboli are more likely to lodge in the cerebral circulation as a result of hydrodynamic, anatomic, and physical factors related to AF [183]. However, advanced age, atrial enlargement and other comorbidities may result in the formation of ‘larger thrombi’ per se which may bypass the carotid orifice merely as a function of size. Of note, ‘silent’ embolism is likely to be more common in the systemic circulation, although this may also occur in cerebral vessels [79, 184].

Numerous clinical trials have provided an extensive evidence base for the use of antithrombotic therapy in AF [168, 185].

The CHADS₂ score should be used as an initial, rapid, and easily memorable means of assessing stroke risk, particularly suited to primary care and non-specialists. In patients with a CHADS₂ score of ≥2, chronic oral anticoagulant therapy with a vitamin K antagonist (VKA) is recommended in a dose adjusted to achieve the target intensity INR of 2–3, unless contraindicated.

For a more detailed stroke risk assessment, a comprehensive risk factor-based approach is recommended. The presence of one ‘definitive’ risk factor merits anticoagulation with an oral VKA (to a target INR 2–3) ( Table 29.7). Patients with two or more ‘combination’ risk factors should all be considered for anticoagulation. Where less validated ‘combination’ risk factors are being included, the decision should be individualized, after discussion of the pros and cons with the patient.

Patients with one ‘combination’ risk factor should be managed with antithrombotic therapy, either oral VKA or aspirin 75–325mg daily. Where possible, such patients at intermediate risk should be considered for a VKA rather than aspirin ( Table 29.7). Where there is one less

validated ‘combination’ risk factor patients should similarly be considered for antithrombotic therapy, either VKA or aspirin 75–325mg daily.

Full discussion with the patient would enable agreement to use VKA instead of aspirin to allow greater protection against ischaemic stroke, especially if these patients value stroke prevention much more than the (theoretical) lower risk of haemorrhage with aspirin and the inconvenience of anticoagulation monitoring. Of note, the Birmingham Atrial Fibrillation Treatment of the Aged Study (BAFTA) found no difference in major bleeding between warfarin (INR 2–3) and aspirin 75mg in an elderly AF population in primary care [186].

Patients with no risk factors are at low risk (essentially patients aged <65 years with lone AF, with none of the risk factors, whether definitive or combination risk factors can be managed with aspirin 75–325mg daily or no antithrombotic therapy, given the limited data on the benefits of aspirin in this patient group (that is, lone AF) and the potential for adverse effects [187].

Table 29.7 summarizes the guidelines for antithrombotic therapy in AF.

The possible benefit of prophylactic VKA therapy in non-valvular AF has been explored in several randomized controlled studies. The first to be published was the Danish Atrial Fibrillation, ASpirin, AntiKoagulation (AFASAK) study, illustrating a 54% relative risk reduction of stroke associated with a VKA regimen [188]. Since then, other randomized controlled studies exploring the role of VKA as stroke prevention in non-valvular AF have been published.

When pooled together in a meta-analysis, the relative risk reduction was highly significant and amounted to 64%, corresponding to an absolute annual risk reduction in all strokes of 2.7% [189]. When only ischaemic strokes were considered, adjusted-dose warfarin was associated with a 67% relative risk reduction ( Fig. 29.19). Notably, this was the risk reduction in those patients who were intended to take oral anticoagulants in the trials. The relative risk reduction for the patients who did indeed use the medication was strikingly high at 85%. All-cause mortality was substantially reduced (26%) by adjusted-dose warfarin versus control. The risk of intracranial haemorrhage was small.

It is important to note that these studies generally excluded patients with low risk of embolism or markedly increased bleeding risk as well as those with the highest risk of embolism. The latter category, in which VKA treatment is strongly advised (although its benefit has not been illustrated in randomized clinical trials), includes patients who in addition to AF have a prosthetic valve, or suffer from rheumatic mitral valve disease or hypertrophic cardiomyopathy. In addition, AF in the setting of hyperthyroidism is often placed in this category, although the evidence for high stroke risk is weak.

Supported by the results of the trials cited above, VKA treatment is strongly recommended for patients with AF with ‘definitive’ or two or more ‘combination’ stroke risk indicators provided there are no contraindications.

Eight independent randomized controlled studies, together including 4876 patients, have explored the prophylactic effects of antiplatelet therapy, most commonly acetylsalicylic acid (ASA), compared with placebo on the risk of thromboembolism in patients with AF [189]. When aspirin alone was compared to placebo or no treatment in seven trials, meta-analysis showed that aspirin was associated with a non-significant 19% (CI, -1% to 35%) reduced incidence of stroke. There was an absolute risk reduction of 0.8% per year for primary prevention trials and 2.5% per year for secondary prevention [189]. When all randomized data from all comparisons of antiplatelet agents and placebo or control groups were included in the meta-analysis, antiplatelet therapy reduced stroke by 22%.

Of note, the dose of aspirin differed markedly between the studies, ranging from 50–1300mg daily. Furthermore, much of the beneficial effect of aspirin was driven by the results of the SPAF-I clinical trial, which suggested a 42% stroke risk reduction with aspirin versus placebo [190]. In this trial, there were inconsistencies for the aspirin effect in this trial between the results for the warfarin-eligible (relative risk reduction, 94%) and warfarin-ineligible (relative risk reduction, 8%) arms. Also, aspirin has less effect in people older than 75 years and did not prevent severe or recurrent strokes. The magnitude of stroke reduction of aspirin versus placebo is comparable to that seen if aspirin were given for vascular disease subjects—given that AF commonly coexists with vascular disease, the modest benefit seen for aspirin in AF is likely to be related to the effects on vascular disease.

Direct comparison between the effects of VKA and ASA has been undertaken in nine studies, illustrating that VKA was significantly superior with a relative risk reduction of 39% [189]. The latter relative risk reduction has largely been driven by the BAFTA trial, which showed that warfarin (target INR 2–3) was superior to aspirin 75mg in reducing the primary endpoint of fatal or disabling stroke (ischaemic or haemorrhagic), intracranial haemorrhage, or clinically significant arterial embolism by 52%, with no difference in risk of major haemorrhage between warfarin and aspirin ( Fig. 29.20) [186].

When the trials prior to BAFTA were considered, the risk for intracranial haemorrhage was doubled with adjusted-dose warfarin compared with aspirin, although the absolute risk increase was small (0.2% per year) [189].

Several attempts have been made to combine drugs with different antithromboembolic mechanisms, mostly low-dose VKA treatment with an antiplatelet drug. No study has shown any superior outcome for such a drug combination compared with dose-adjusted VKA only in preventing stroke or embolism in non-valvular AF.

Antiplatelet therapy has also been combined with therapeutic anticoagulation in AF cohorts, and these have shown a high rate of bleeding in the combination treatment arm [191, 192]. An ancillary analysis from the SPORTIF (Stroke Prevention using an Oral Thrombin Inhibitor in Atrial Fibrillation) trials, which compared aspirin users with non-users [192] found no additive effect of taking aspirin for stroke prevention or a reduction in vascular events (including death or myocardial infarction) in patients who were given anticoagulation (warfarin or ximelagatran); instead, aspirin use resulted in a substantial increase in risk of bleeding.

There is no role for the routine addition of aspirin to anticoagulation therapy in AF patients with stable vascular disease, given the lack of benefit in reducing vascular events, and the significant increase in bleeding risks [193].

Clopidogrel, in combination with aspirin, has been compared against warfarin in the Atrial fibrillation Clopidogrel Trial with Irbesartan for prevention of Vascular Events (ACTIVE W) trial. This trial was stopped early due to the inferiority of aspirin–clopidogrel combination therapy against warfarin for stroke prevention, with no difference in bleeding events between the treatment arms ( Fig. 29.21) [194]. An ancillary analysis of AF subjects participating in the Clopidogrel for High Atherothrombotic Risk and Ischemic Stabilization, Management, and Avoidance (CHARISMA) trial of patients with stable cardiovascular disease or multiple cardiovascular risk factors, which compared combination aspirin–clopidogrel therapy to aspirin alone, found no benefit of combination therapy but an excess of severe or fatal extra-cranial haemorrhage with combination therapy [195].The recent ACTIVE-A trial tested the hypothesis that the addition of clopidogrel 75mg to aspirin would reduce the risk of vascular events in 7554 patients with AF [196]. At a median of 3.6 years of follow-up, major vascular events were reduced in patients receiving aspirin–clopidogrel (6.8%/year vs. aspirin alone, 7.6%/year: RR 0.89; 95% CI 0.81–0.98; P = 0.01). This difference was primarily due to a reduction in the rate of stroke with combination therapy (2.4%/year vs. 3.3%/year: RR 0.72; 95% CI, 0.62–0.83; P <0.001). However, major bleeding increased with combination therapy (2.0%/year vs. 1.3%/year. RR 1.57; 95% CI, 1.29–1.92; P <0.001). Thus, in AF patients for whom VKA therapy was unsuitable, the addition of aspirin–clopidogrel reduced the risk of major vascular events, especially stroke, but increased the risk of major haemorrhage. Nonetheless, 50% of subjects entered the trial due to 'physician perception of being unsuitable for VKA therapy' and 23% had a relative risk factor for bleeding at trial entry. However, of patients perceived not to be candidates for VKA therapy, only a small proportion still have the same relative contraindication to VKA a year later. Hence aspirin–clopidogrel therapy could be considered as an interim measure pending use of more effective thromboproplylaxis with VKA. Other antiplatelet agents such as indobufen and triflusal have been investigated in AF, with the suggestion of some benefit, but more data are required before definitive conclusions can be made [167].

Several new anticoagulant drugs—broadly in two classes, the oral direct thrombin inhibitors (DTI) and the oral factor Xa inhibitors (FXaI)—have been developed as possible alternatives to VKA ( Fig. 29.22) [197]. The first oral DTI, ximelagatran was tested in two large-scale clinical studies, and was found to be as effective as adjusted-dose warfarin in the prevention of ischaemic strokes or systemic emboli (RR 1.04; 95% CI: 0.77–1.40) ( Fig. 29.19) with less risk of major bleeding (RR, 0.74; 95% CI, 0.56–0.96) [198]. Further development of ximelagatran has been discontinued due to liver toxicity.

Other DTIs (e.g. dabigatran and AZD0837), as well as Factor Xa inhibitors (rivaroxaban, apixaban, edoxaban, YM150, etc.) and non-warfarin vitamin K antagonists (ATI-5923) are being developed for stroke prevention in AF [199]. The possible benefit of these drugs awaits the outcomes of clinical studies.

One indirect FXaI, idraparinux, which requires once-weekly administration, was tested against warfarin in one trial [200]. This found that idraparinux was non-inferior to warfarin for stroke prevention, but there was a marked excess of bleeding with idraparinux compared to warfarin. A biotinylated version of idraparinux is undergoing a clinical trial in AF, with the potential of anticoagulant reversal should bleeding occur.

Currently, the level of anticoagulation in a serum sample is expressed as the INR, which is the ratio between the actual prothrombin time and that of a standardized control serum. Although arguments have been raised against the accuracy of this comparative technique [201], it has become widely accepted.

Based on the pooled information from achieving a balance between stroke risk with low INRs and an increasing bleeding risk with high INRs [202, 203], a consensus has been reached that an INR of 2.0–3.0 is the likely optimal range for prevention of stroke and systemic embolism in patients with non-valvular AF. A mortality risk analysis in a VKA-treated population comprising >40,000 patients suggests that the target INR window should be narrower, with a target value close to 2.2–2.3 [204]. Fig. 29.23 illustrates the risk of ischaemic stroke and intracranial bleeding relative to the INR level as well as the 1-month mortality risk in relation to INR in patients with AF.

Keeping within a target INR of 2–3 is difficult. One of the many problems with anticoagulation with VKA is the high inter- and intra-individual variation in INRs. VKAs also have significant drug, food, and alcohol interactions. Patients stay within target INR range (2–3) for 60–65% of the time [205], but many ‘real life’ studies even suggest that this figure may be <50%.

The elderly may pose a particular problem. One hospital-based cohort suggested a high bleeding rate in elderly subjects (>80 years old) initiated on warfarin, especially in the first 3 months following the start of warfarin [206]. In this study, increasing stroke risk was associated with an increasing risk of bleeding, which in turn led to more discontinuations. Whilst a lower target INR range (1.8–2.5) has been proposed for the elderly, this is not based on any trial evidence, and cohort studies even suggest a twofold increase in stroke risk at INR 1.5–2.0 [207], and is therefore not recommended. Also, the BAFTA trial shows the benefit of conventional dose warfarin over aspirin in the elderly (aged ≥75 years), irrespective of age categories [186].

In addition to the fear of drug-induced bleeding complications, VKA treatment is associated with several other reasons that contribute to its underuse in patients with non-valvular AF [208]. In short, the barriers to the use of VKA treatment in non-valvular AF patients may relate to the doctor as well as the patient and the healthcare system. In addition, many patients with AF have limited understanding of the disease process as well as the need for anticoagulation therapy to prevent thromboembolism [209].

The maintenance, safety, and effectiveness of INR within range can be influenced by pharmacogenetics of VKA therapy, particularly the cytochrome P450 2C9 gene (CYP2C9) and the vitamin K epoxide reductase complex 1 gene (VKORC1). CYP2C9 and VKORC1 genotypes can influence warfarin dose requirements, whilst CYP2C9 variant genotypes are associated with bleeding events [210]. The ability to determine mutations in the genes coding for these two proteins could influence future VKA dosing patterns (i.e. genotype-guided therapy), but just how much would these would really add to regular, conscientious monitoring of the INR and dose adjustment is undetermined. Ongoing trials are addressing this question.

The need for regular monitoring of oral anticoagulation therapy has led to a substantial increased use of ‘point of care’ or ‘near patient’ testing schemes, as well as patient self-monitoring approaches ( Fig. 29.24).

Thus, patients may have anticoagulation-monitoring testing at a local clinic (e.g. general practitioner clinic using an INR testing machine, or blood sample taken at a local clinic which is batched and sent to a central laboratory) and the INR result phoned to the patient, with VKA dose change, if necessary. Alternatively home monitoring can be performed using a personal INR testing machine allowing patients to alter the warfarin dose themselves [211].

Self-monitoring could be considered if the patient is physically and cognitively able to perform the self-monitoring test [212, 213] or a designated carer may be used. Appropriate training by a competent healthcare professional is needed and the patient must remain in contact with a named clinician.

The stroke and thromboembolic risk in paroxysmal AF is less well defined, and such patients have represented the minority (usually <30%) in the clinical trials of thromboprophylaxis. A recent large trial programme did not identify a difference in stroke risk between paroxysmal or chronic forms of AF [77].

A retrospective hospital record study followed >400 individuals with paroxysmal AF for >25 years [214]. In these non-VKA taking patients, there was a clustering of thromboembolism at the time of onset of paroxysmal AF, namely 6.8% during 1 month. Later, the annual embolic rate varied from 0.6–2.6%. Following transition to permanent AF, which occurred in every third patient, the rate of embolism rose to a significantly higher level. Some of the observed differences in thromboembolic rates are clearly attributable to the presence or absence of stroke risk factors.

Another study, exploring the embolic risk factors in >700 patients with paroxysmal AF, verified a 2.2% annual rate of embolism, typically occurring in males above the age of 65 [77]. Importantly, individuals without any underlying disorder had a low embolic event rate (0.7% per year). Data from the non-VKA arms of clinical trials show that stroke risk in paroxysmal AF was comparable to that seen in permanent AF, and is dependent upon the presence of stroke risk factors [171, 215]. Thus, current guidelines suggest that prevention of thromboembolism is appropriate in patients with paroxysmal AF [1].

Patients with AF who are anticoagulated will require temporary interruption of a VKA before surgery or an invasive procedure. Many surgeons require an INR of <1.5 or even INR normalization before undertaking surgery. Thus, VKAs should be stopped approximately 5 days before surgery, to allow the INR to fall appropriately. VKA should be resumed at the ‘usual’ maintenance dose (without a loading dose) on the evening of (or the next morning) after surgery assuming there is adequate haemostasis. If surgery is urgent but the INR is still elevated (>1.5), the administration of low-dose oral vitamin K (1–2mg) to normalize the INR may be considered.

In patients with a mechanical heart valve or AF at high risk for thromboembolism, management can be problematic. Such patients should be considered for ‘bridging’ anticoagulation with therapeutic doses of heparin (either low-molecular-weight heparin (LMWH) or unfractionated heparin (UFH)) during the temporary interruption of VKA therapy [216].

Increasing numbers of patients with AF may present with an acute coronary syndrome (ACS). Given that AF is also associated with CAD, some AF patients would require percutaneous coronary intervention, with the possibility of stent implantation. Current guidelines for ACS and/or PCI recommend the use of aspirin–clopidogrel combination therapy after ACS (9–12 months), and after a stent (4 weeks for a bare metal stent, 6 or more months for a drug-eluting stent).

However, there are limited data to guide the optimal antithrombotic regimen to use in anticoagulated patients with AF, given that bleeding with triple therapy (VKA, aspirin, and clopidogrel) may be substantial. The largest published cohort [217] suggests that non-VKA use was associated with an increase in mortality and major adverse cardiac events, with no significant difference in bleeding rates between VKA and non-VKA users.

Current consensus guidelines suggest that drug-eluting stents should be avoided and triple therapy (VKA, aspirin, clopidogrel) used in the short term, followed by more long-term therapy with VKA plus a single antiplatelet drug [218, 219].

Acute stroke is a common first presentation of a patient with AF, given that the arrhythmia often develops asymptomatically. There are limited trial data to guide our management, and there is concern that patients within the first 2 weeks of having had a cardioembolic stroke are at greatest risk of having stroke recurrence because of further thromboembolism. However, anticoagulation in the acute phase may result in intracranial haemorrhage or haemorrhagic transformation of the infarct. In most stroke survivors with AF it is reasonable to initiate anticoagulation at approximately 2 weeks after the stroke in patients with minor strokes or TIA, anticoagulation could be initiated earlier, after cerebral imaging as excluded any intracranial bleeding.

Large-scale prospective observational or interventional studies on the stroke rate in atrial flutter per se are lacking. However, the risk of stroke linked to atrial flutter has been studied retrospectively in a large number of older patients, and was similar to that seen in AF [219]. Thus, thromboprophylaxis in a patient with atrial flutter should follow the same guidelines as if the patient had AF [185].

On occasion, AF may occur in pregnant women, especially in relation to valvular heart disease, prosthetic heart valves, or venous thromboembolism. A thorough discussion of the potential risks and benefits of any anticoagulation regimen is needed with the patient, and a close liaison on management between the cardiologist and obstetrician is mandatory.

VKAs can be teratogenic and in many cases should be substituted with UFH or LMWH for the first trimester of the pregnancy [221]. Pregnant patients with AF and mechanical prosthetic valves who elect to stop warfarin between weeks 6 and 12 of gestation should receive continuous intravenous UFH, dose-adjusted UFH, or dose-adjusted subcutaneous LMWH [222] and may start VKA in the second trimester at an only slightly elevated teratogenic risk. For pregnant women with acute venous thromboembolism, subcutaneous LMWH or UFH should be continued throughout pregnancy and anticoagulant prophylaxis continued for at least 6 weeks postpartum [221].

The increased risk of embolism following cardioversion is well recognized. Therefore, full antithromboembolic treatment is considered mandatory before elective cardioversion in cases where AF duration exceeds 48 hours [185]. Despite these precautions, embolism has been reported to occur in up to 2% of all electrical cardioversions [223]. These occur typically during the initial few days following reappearance of sinus rhythm ( Fig. 29.25).

The mechanism of post-cardioversion embolism is complex. Pre-existing thrombi may dislodge from the endocardial wall when the atria resume a slower and regular rhythm. However, following conversion from AF to sinus rhythm, the mechanical function of the atrial myocardium is not immediately restored (‘atrial stunning’), presenting

an opportunity for thrombus formation. Furthermore, there is activation of coagulation and platelets, leading to a propensity to thrombogenesis in the immediate post-cardioversion period.

The importance of an adequate anticoagulation level at the time of cardioversion must be stressed. In a study of >2500 elective direct-current cardioversion attempts in almost 2000 patients, no post-cardioversion embolism could be confirmed when the INR exceeded 2.4 on the day of the procedure [224]. In contrast, embolism was increasingly more common at a lower INR and appeared to increase with decreasing INR.

Current guidelines state that VKA treatment (INR 2–3) should be given for at least 3 weeks before cardioversion of a patient in whom AF has been maintained for >48 hours or which is of unknown duration [185]. Thromboprophylaxis is recommended irrespective of whether the cardioversion is performed using a pharmacological or an electrical method. Anticoagulation treatment should be continued for a minimum of 4 weeks post-cardioversion. However, in patients with stroke risk factors, or those at high risk of AF recurrence, long-term VKA treatment is usually required.

The 3–4-week period of adequate anticoagulation prior to cardioversion can be shortened with the use of transoesophageal echocardiography. This technique may not only show thrombi within the LAA or the LA proper but also identify indicators for thrombus formation, such as spontaneous echo-contrast or complex aortic plaque. The safety and applicability of transoesophageal echocardiography-guided cardioversion has been repeatedly verified [225, 226]. Following exclusion of any thrombus, anticoagulation can commence with LMWH [226, 227], the cardioversion performed, and anticoagulation post-cardioversion continued, as highlighted earlier. Oral anticoagulation is started and LMWH discontinued when the INR is therapeutic.

A history of previous stroke or TIA is the most powerful risk factor for a recurrent cerebrovascular event. As stroke in patients with AF is primarily embolic, the detection of asymptomatic cerebral emboli may identify high-risk patients. The prevalence of silent cerebral infarct on computer tomography images of the brain in AF patients without neurological deficit ranged from 15–26% in the SPINAF and SPAF (Stroke Prevention Atrial Fibrillation) studies [228, 229]. These silent strokes may contribute to impaired cognitive function in AF patients.

Transcranial Doppler ultrasound may identify asymptomatic patients with an active embolic source or patients with prior stroke who are at high risk of recurrent stroke. During 1-hour bilateral Doppler monitoring from the middle cerebral arteries, the frequency of embolization was 13% in patients with previous AF-related stroke or TIA and 16% in those individuals without a history of a cerebrovascular event [230]. The incidence of embolic signals was higher in untreated patients compared with those receiving warfarin (11.9% vs. 1.5%) [231].

Since the majority of all LA thrombi form in the LAA [181], different techniques have been developed that aim to eliminate the LAA as a possible source of thromboembolism. The result of LAA resection in patients who undergo cardiac surgery for other reasons has been tested in a series of >400 patients [231]. No strokes were reported following surgery and it has been suggested that the LAA should be resected ‘whenever the chest is open’.

Other non-pharmacological techniques for elimination of the LAA as a possible thrombotic location are currently undergoing clinical evaluation, including thoracoscopic obliteration of the LAA as well as endocardial occlusion of the LAA with different devices [199, 232]. Fig. 29.26 illustrates two different devices that can be inserted in the LAA to prevent thrombus formation and subsequent stroke. At the American College of Cardiology meeting in 2009, the PROTECT AF Trial, a Randomized Prospective Trial of Percutaneous LAA Closure vs Warfarin for Stroke Prevention in AF found that all cause stroke and all cause mortality risk using the WATCHMAN device was non-inferior to warfarin, with a lower haemorrhagic stroke risk. However, there were early safety events, specifically pericardial effusion.

The fundamental principles of pharmacological therapy for AF, apart from anticoagulation and treatment of underlying conditions associated with AF, include specific termination of the arrhythmia and maintenance of sinus rhythm, prevention of AF recurrence (secondary prevention), and control of ventricular rates during AF. Antiarrhythmic drugs can be used for rhythm control as sole agents or as an adjunct to catheter ablation or electrical cardioversion. Prevention and effective treatment of conditions that are commonly associated with the development of AF, such as hypertension and congestive heart failure, may prevent the occurrence of new AF (primary prevention) or reduce the recurrence rate, and delay the progression to permanent AF (secondary prevention) [233]. Treatment of the underlying heart disease to prevent atrial remodelling and formation of the substrate for AF is often termed ‘upstream’ therapy [34].

Symptomatic AF can be managed by therapy aimed either at restoring and maintaining sinus rhythm (‘rhythm control’) or therapy that controls ventricular rate (‘rate control’). Sinus rhythm offers physiological control of heart rate and regularity, normal atrial activation and contraction, the correct sequence of AV activation, and normal valve function. Rhythm control might be an ideal approach for both stroke prevention and symptom alleviation and it might improve survival. However, the long-term maintenance of sinus rhythm has proven difficult to achieve, the recurrence rate is high, and rhythm-control management is time consuming, expensive (due to the cost of the antiarrhythmic drugs and the increased need for hospitalization, e.g. for cardioversion), and is not completely free from complications. A well-appreciated limitation of current rhythm-control management is poor long-term efficacy and the adverse effects of antiarrhythmic drugs (e.g. proarrhythmia and organ toxicity) that may negate the inherent advantage of sinus rhythm over AF [234].

The advantages of rate control are simplicity, availability, and low cost, although in some patients, adequate rate control may be difficult to achieve and may not result in symptom relief, or may be associated with adverse effects (e.g. bradycardia or worsening heart failure) and increased mortality [234, 235].

Several randomized studies directly and prospectively compared the effect of rate and rhythm control on patient outcomes ( Table 29.8). The major studies were the Atrial Fibrillation Follow up Investigation of Rhythm Management (AFFIRM) trial [237], the RAte Control versus Electrical Cardioversion (RACE) trial [238], and the Atrial Fibrillation Congestive Heart Failure (AF-CHF) trial [239]. There were also a series of pilot studies performed, including the Pharmacological Intervention in Atrial Fibrillation (PIAF) [240], Strategies of Treatment of Atrial Fibrillation (STAF) [241], and How to Treat Chronic Atrial Fibrillation (HOT CAFÉ) [242] among others. Virtually all studies have shown that primary rate control is not inferior to rhythm control. Meta-analysis has demonstrated no significant excess or reduced mortality with either strategy [243].

The AF-CHF trial compared rate- and rhythm-control strategies specifically in patients with an ejection fraction of 35% or less and NYHA II–IV heart failure ( Fig. 29.27) [239]. The study showed no benefit from rhythm control (mainly with amiodarone) on top of optimal medical therapy on cardiovascular death (the primary outcome) as well as pre-specified secondary outcomes (total mortality, worsening heart failure, stroke, and hospitalization). Cardiovascular death occurred in 26.7% of the patients in the rhythm-control group compared with 25.2% in the rate-control arm.

Rate control as a primary strategy is now accepted in older, sedentary, and asymptomatic (or only mildly symptomatic) patients (EHRA I–II) who have had their arrhythmia for many years, without significant impairment of ventricular function and exercise tolerance. Following the AF-CHF study, rate control is a legitimate primary treatment option for patients with heart failure who can tolerate AF without worsening NYHA functional class. Rhythm control with antiarrhythmic drugs, cardioversion, or ablation remains treatment of choice in patients who are symptomatic, in patients with recent onset AF, or in young and active individuals.

On-treatment analysis of outcomes in the AFFIRM trial has shown a 47% reduction in the risk of all-cause death if sinus rhythm was maintained irrespective of the treatment strategy [244]. If safer and more effective antiarrhythmic agents were available, sinus rhythm might confer a favourable outcome and many new antiarrhythmic agents are under investigation ( Fig. 29.28) [107].

The choice of an antiarrhythmic drug for cardioversion as well as for long-term management of AF depends on underlying heart disease ( Fig. 29.29) [1]. Class IC antiarrhythmic agents (propafenone and flecainide), and class III agents (sotalol and ibutilide) are recommended for cardioversion of AF in patients with moderate structural heart disease or hypertension without LV hypertrophy. These agents are not recommended in patients with a history of heart failure, myocardial infarction with LV dysfunction, and significant LV hypertrophy. Amiodarone and dofetilide (dofetilide is not available outside the USA) can be used in patients presenting with symptoms of heart failure and known advanced heart disease. Although oral quinidine and oral or intravenous procainamide (class IA antiarrhythmic agents) are still available, there has been a significant decrease in their use worldwide. Quinidine as a fixed combination with verapamil has a limited use.

Where to initiate antiarrhythmic drug therapy must take into account risk and practicality. Patients at anticipated high risk of developing adverse effects (e.g. patients with an inherently prolonged QT interval, or patients at risk of tachycardia–bradycardia syndrome upon termination of AF), should not be prescribed antiarrhythmic drugs outside the hospital setting. For some antiarrhythmic agents, e.g. dofetilide, there is formal requirement for in-hospital initiation.

In the absence of proarrhythmia concerns and formal labelling, convenience and cost effectiveness favour out-of-hospital initiation, e.g. oral propafenone and flecainide (usually in combination with AV blocking drugs to prevent fast ventricular rates if atrial flutter occurs) in patients with lone AF or AF associated with hypertension without significant LV hypertrophy. Amiodarone can be safely started on an outpatient basis, given its long elimination half-life period and low probability of developing torsade de pointes. An ECG control and/or trans-telephonic ECG monitoring should be arranged to provide surveillance of heart rate, PR and QT interval durations (sotalol, amiodarone, dofetilide), QRS width (flecainide, propafenone), and assessment of the efficacy of treatment. When sotalol therapy is initiated outside hospital, the initial dose should be low and up-titration should depend on the heart rate and QT interval.

Cardioversion with antiarrhythmic drugs is usually effective if initiated early, i.e. within a week, probably even 3 days, after onset of AF. Within 24–72 hours, about 45% of patients with AF may convert spontaneously [245] and drug-assisted conversion will occur in approximately 70% [245–247].

Systematic analysis of placebo-controlled studies of pharmacological cardioversion for AF has shown that among patients with AF of <24 hours, 66% spontaneously converted to sinus rhythm compared with only 17% of those with AF of longer duration (odds ratio 1.8) [246].

In patients with no or minimal structural heart disease and relatively infrequent (less than monthly), symptomatic paroxysms of AF of distinct onset which do not cause significant haemodynamic compromise (e.g. hypotension), a single loading dose of propafenone or flecainide can be used for expedient cardioversion [248]. In a proof of concept study, patients with paroxysmal AF, who had been successfully treated in hospital with either oral flecainide or propafenone, were instructed to take a single oral dose of the relevant drug within 5min of noticing palpitations. This ‘pill in the pocket’ strategy resulted in the reduction in the number of visits and hospitalizations, despite the same frequency of arrhythmia episodes [249].

The experience with this approach is limited and neither drug is licensed for patients to use for self-treating single attacks. As there is a danger of developing atrial flutter with 1:1 AV conduction, QRS widening, and rarely LV dysfunction, it is mandatory that the efficacy and safety of this strategy is first tested in-hospital. Furthermore, it is advisable to combine these agents with an AV-nodal slowing drug, e.g. a beta-receptor blocker, verapamil, or diltiazem. Atrial flutter was reported in 5–7% of patients who received oral loading doses of propafenone or flecainide for conversion of AF [250].

Intravenous propafenone and flecainide are very effective in cardioversion of AF of <72 hours’ duration, with conversion rates as high as 80–90% within an hour after the start of infusion [251–253]. When taken orally as a loading dose (450–600mg for propafenone, 200–300mg for flecainide), restoration of sinus rhythm is more delayed, but conversion rates (70–80% at 8 hours) are comparable to those observed after intravenous administration [250, 254–259]. The drugs are less effective for termination of AF lasting >7 days.

Class IC drugs are ineffective for conversion of atrial flutter. They slow conduction within the re-entrant circuit and prolong the atrial flutter cycle length, but rarely interrupt the circuit. The efficacy rates have been reported to be as low as 13–40% [260].

Ibutilide is moderately effective for cardioversion of AF and is more effective for cardioversion of atrial flutter (31–44% vs. 56–70%) [261, 262]. The drug is available only as an intravenous formulation and is usually administered as a 1-mg bolus over 10min. Higher doses of ibutilide such as a single bolus of 2mg or two successive infusions of 1mg may be required for cardioversion of AF [261, 263, 264]. The antiarrhythmic effect of ibutilide decreases if the arrhythmia had persisted for >7 days.

Ibutilide prolongs the QT interval and may induce ventricular proarrhythmia. In the ibutilide trials, the incidence of polymorphic VT or torsade de pointes requiring electrical cardioversion was 0.5–1.7% and the incidence of self-terminating polymorphic VT was 2.6–6.7% [262, 265]. There are insufficient data to support the use of ibutilide in patients with significant structural heart disease as many controlled studies of ibutilide have only enrolled patients with mild or moderate underlying heart disease.

Oral dofetilide has been studied extensively and is considered safe and relatively effective for pharmacological cardioversion of AF including arrhythmia duration of >7 days. Two medium-size prospective studies, SAFIRE-D (Symptomatic Atrial Fibrillation Investigative Research on Dofetilide) and EMERALD (European and Australian Multicenter Evaluative Research on Dofetilide) reported a modest 30% cardioversion rate of persistent AF with high-dose (1000mcg twice daily) oral dofetilide compared with 1.2% of spontaneous conversion on placebo [266] and 5% on sotalol [267].

In the pooled analysis from the DIAMOND (Danish Investigations of Arrhythmia and Mortality ON Dofetilide) studies in patients with symptomatic heart failure (DIAMOND-CHF) or myocardial infarction with LV dysfunction (DIAMOND-MI), oral dofetilide had a neutral effect on mortality and also demonstrated a greater rate of conversion to sinus rhythm (44% vs. 14%) [236].

Dofetilide causes QT interval prolongation and a non-negligent risk of torsade de pointes. The effect on the QT interval is dose-related and proarrhythmia typically occurs within the first 2–3 days after initiation of therapy. Therefore, it is mandatory that dofetilide be initiated in-hospital and that patients should be monitored for at least 3 days. In addition, the dose of dofetilide requires adjustment to creatinine clearance ( Table 29.9).

Intravenous sotalol (1–1.5mg/kg) is ineffective for acute pharmacological cardioversion of AF: the conversion rates at 10–20% were not different from placebo [261, 262, 268, 269]. The antifibrillatory effect of sotalol is limited by reverse use dependency of its effect on atrial refractoriness: sotalol prolongs the atrial effective refractory period at normal and slow atrial rates, but not during rapid AF.

There is evidence that oral sotalol may offer a modest benefit of facilitating conversion to sinus rhythm and, in addition, can ensure ventricular rate control in patients awaiting electrical cardioversion. In the Sotalol Amiodarone Atrial Fibrillation Efficacy Trial (SAFE-T), 24.2% of patients with persistent AF treated with sotalol converted to sinus rhythm within 28 days, compared with 27.1% on amiodarone and only 0.8% on placebo [19].

The adverse effects of sotalol include hypotension, bradycardia, QT interval prolongation, and associated ventricular proarrhythmia (torsade de pointes). Bradycardia and hypotension were the commonest with intravenous sotalol [19]. The risk of proarrhythmia is increased in the presence of LV hypertrophy and in renal failure and the drug is contraindicated in these conditions.

Amiodarone is considered a relatively safe drug for acute pharmacological cardioversion and is the drug of choice in patients with advanced underlying heart disease. Amiodarone does not have any negative inotropic effect, controls the ventricular rate, and is associated with a low incidence of torsade de pointes. In meta-analysis of 13 randomized controlled studies in 1174 patients, the placebo-subtracted efficacy of amiodarone was 44%, but its effect was delayed up to 24 hours [270]. Intravenous amiodarone followed by an oral maintenance dose increases the likelihood of conversion to sinus rhythm [271].

The mortality and morbidity study CHF-STAT (Congestive Heart Failure Survival Trial of Antiarrhythmic Therapy) in patients with a mean ejection fraction of 25% has shown that long-term treatment with oral amiodarone 400mg daily for the first year and 300mg daily for the remainder of the 4.5-year trial was associated with greater rates of conversion to sinus rhythm compared with placebo (31% vs. 7.7%) [272].

Unlike propafenone and flecainide, amiodarone preserves its efficacy in patients with long-standing AF. Thus, in patients with a mean AF of almost 2 years, amiodarone 600mg daily for 4 weeks restored sinus rhythm in 34% of patients compared with 0% in the placebo group [273].

Amiodarone prolongs the QT interval, but, unlike pure class III agents, exhibits a low arrhythmogenic potential (<1%) [274]. The most common adverse effects of intravenous amiodarone are hypotension and relative bradycardia.

Class IA drugs procainamide (oral and intravenous) and quinidine (oral) have been widely used for cardioversion of AF. In direct comparisons, the efficacy of intravenous procainamide 1000–1200mg was comparable to that of propafenone (69.5% vs. 48.7%) [275], flecainide (65% vs. 92%) [276], or amiodarone (68.5% vs. 89.1%) [277] for conversion of AF of <48 hours. Quinidine given orally in a cumulative dose of up to 1200–2400mg over 24 hours has been shown to cardiovert 60–80% of recent onset AF [278, 279].

Because of the non-target effects (vasodilatation, hypotension, anticholinergic action, AV node blockade, worsening heart failure, and in addition to these, gastrointestinal discomfort and a 6% increased risk of torsade associated with quinidine), the drugs are less commonly used for cardioversion of AF.

There is limited evidence for the efficacy of intravenous disopyramide for acute pharmacological cardioversion in patients with AF. In one study, disopyramide administered as a bolus of 2mg/kg over 5min restored sinus rhythm in (71%) patients with self-limiting lone AF and three of seven (43%) patients with atrial flutter [280]. There are concerns, however, that the adverse effects such as proarrhythmia, hypotension, asystole, and non-target effects resulting from anticholinergic activity of disopyramide, may offset its modest antiarrhythmic potential.

Vernakalant is a new antiarrhythmic drug agent [281] with an affinity to ion channels specifically involved in the repolarization processes in atrial tissue, in particular, the ultrarapid potassium repolarization current I_Kur, but has little impact on major currents responsible for ventricular repolarization. It does, however, inhibit the inward sodium current (I_Na) and slows myocardial conduction, especially at high rates.

In the randomized, double-blind, placebo controlled Atrial arrhythmia Conversion Trials (ACT), vernakalant administered as a 10-min infusion of 3mg/kg (followed by a second infusion of 2mg/kg if AF persisted after 15min) was significantly more effective than placebo in converting AF of <7 days (52% compared with 3.6–4%, respectively) [282, 283]. The highest efficacy was observed for AF of up to 72 hours (70–80%). Vernakalant was ineffective in converting AF of >7 days duration and did not convert atrial flutter.

The drug was well tolerated; the most common (>5%) side effects of vernakalant were dysgeusia, sneezing, and nausea. Minor QTc prolongation was reported but there has been little or no proarrhythmia [282, 284].

Magnesium sulphate Meta-analysis of eight studies in 476 patients showed that magnesium sulphate, administered intravenously at an initial dose of 1200–5000mg over 1–30min (in some studies followed by the second dose or continuous infusion for 2–6 hours), was superior to placebo or the active comparator in cardioverting AF with an odds ratio of 1.6 (95% CI, 1.07–2.39) [285]. The most common side effects were sensation of warmth and flushing. Magnesium prolongs the atrial and AV node refractory periods and therefore in addition to its antifibrillatory effect, it may slow the ventricular rate. Magnesium is not routinely used for pharmacological cardioversion of AF, but it may potentiate the effect of other antiarrhythmic agents.

Digitalis, beta-blockers, and calcium antagonists usually are ineffective for acute conversion of AF [286, 287]. Digoxin may even be profibrillatory due to its cholinergic effects which may cause a non-uniform reduction in conduction velocity and effective refractory periods of the atria [288]. Short-acting intravenous beta-blockers (e.g. esmolol) and non-dihydropyridine calcium antagonists (verapamil and diltiazem) are more commonly used for rate control than for restoration of sinus rhythm.

Prophylactic antiarrhythmic drug therapy is recommended for patients with paroxysmal AF when paroxysms occur frequently and are associated with significant symptoms or lead to worsening LV function, and for patients with persistent AF when the likelihood of maintenance of sinus rhythm is uncertain, especially in patients with structural heart disease and a remodelled LA. After cardioversion, approximately 25–50% patients will have the recurrence of persistent AF within the first 1–2 months (early recurrence). Thereafter, the recurrence rate is about 10% per year ( Fig. 29.30).

A systematic review of 44 studies in 11,322 patients has shown that antiarrhythmic drugs significantly reduced AF recurrence after cardioversion; the number of patients needed to treat to prevent a recurrence ranged from two to nine, depending on the agent used ( Fig. 29.31) [289]. The majority of large-size, high-quality studies were conducted in patients with persistent AF, mainly because the recurrence of a persistent episode is more ‘predictable’, is likely to occur during the first year, and is easier to recognize and document, especially when the time to first symptomatic recurrence is used as an outcome parameter. Hence, the efficacy of an antiarrhythmic drug to control paroxysmal AF is usually derived from the results of studies in persistent AF.

Beta-blockers are modestly effective in preventing AF and are mainly used for rate control. In anecdotal reports, the annual recurrence rate after cardioversion for persistent AF was slightly lower on beta-blockers (metoprolol 100mg or bisoprolol 5mg) than on placebo (48% vs. 60%) [290] or comparable to that on sotalol (42% vs. 41%) [291]. There is no evidence of superiority of one type of beta-blocker over the other for prevention of AF [292]. However, because of their safety and the effect on AV node conduction during rapid AF, beta-blockers are often used as initial therapy in patients with new onset AF.

In addition, beta-blockers may contribute to upstream therapy in AF associated with congestive heart failure: a meta-analysis of seven studies in 11,952 patients has shown that therapy with beta-blockers was associated with a statistically significant reduction in the incidence of new onset AF by 27% during mean follow-up of 1.35 years [295]. Beta-blockers are first-line therapy in patients with thyrotoxicosis or, rarely, adrenergically-mediated AF [293, 294].

Propafenone and flecainide are recommended as first-line therapy for AF in patients without significant structural heart disease, i.e. patients without congestive heart failure, LV dysfunction, marked hypertrophy, previous myocardial infarction, or CAD. Both propafenone (300–900mg daily in 2–3 divided doses) and flecainide (50–150mg twice daily) reduced the recurrence rate by approximately two-thirds [254, 289, 296–300], with no advantage of one drug over the other. In a meta-analysis of propafenone, the incidence of recurrent AF at 1 year was 56.8% (52.3–61.3%) [254]. All-cause mortality associated with propafenone was 0.3%. Several placebo-controlled and comparator trials of flecainide at 200–300mg daily have consistently reported a 60–70% likelihood of maintaining sinus rhythm after 1 year with an acceptable risk:benefit ratio [296, 299, 300].

Propafenone is available as a sustained-release (SR) formulation at 225, 325, or 425mg twice daily. In the North American Recurrence of Atrial Fibrillation Trial (RAFT) [301] and its European equivalent, ERAFT [302], propafenone SR was superior to placebo in prolonging time to first symptomatic recurrence of paroxysmal AF in patients with minor structural heart disease. Flecainide is also available as a long-acting formulation in some parts of Europe, but no formal studies of its safety and efficacy have been reported.

Drugs with class IC mode of action, other than flecainide and propafenone, are available in some countries, e.g. pilsicainide is available in Japan and cibenzoline is used in Japan and France. Both drugs are modestly effective in cardioversion (45% for pilsicainide) and/or prevention of AF (maximum 41% reported for pilsicainide and 50% for cibenzoline at 1 year) and exhibit a similar adverse event profile to other class IC agents [303, 304].

Quinidine has been used for treatment of AF since the discovery of its antiarrhythmic properties in the early 1920s. In a meta-analysis of six randomized controlled trials in 808 patients, published in 1990, 50% of patients treated with quinidine were in sinus rhythm after 1 year compared with 25% among controls [305]. The antiarrhythmic effect of quinidine was offset by high all-cause mortality and sudden death in the quinidine-treated patients compared with controls (2.9% vs. 0.8%; odds ratio 2.98) [305]. In a 2006 analysis which included two recent large-scale studies of quinidine, PAFAC (Prevention of Atrial Fibrillation After Cardioversion) and SOPAT (Suppression Of Paroxysmal Atrial Tachyarrhythmias), quinidine reduced recurrent AF by 49% [289]. In PAFAC [6] and SOPAT [7], quinidine was not associated with increased risk of death, probably because it was used at lower doses (320–480mg as opposed to 1000–1800mg daily in the previous trials [305]), in a fixed combination with verapamil (single combination tablet), and in patients with overall less significant structural heart disease. However, the foremost safety issue for quinidine remains its propensity to cause ventricular proarrhythmia including torsade de pointes even at low or sub-therapeutic doses [234].

Disopyramide is rarely used for treatment of AF because of its negative inotropic effect, the torsadogenic potential, and poor tolerance due to antimuscarinic properties. However, the use of disopyramide is advocated in patients with lone, vagally-mediated AF. Data on the efficacy of disopyramide in AF are sparse [305, 306].

Sotalol can prevent recurrent AF in the absence of heart failure, myocardial infarction, or hypertension with significant LV hypertrophy [308–310]. Because of its beta-blocking effect, sotalol offers the additional benefit of ventricular rate slowing during recurrences. The usual dose for AF is 80–160mg BID. In a meta-analysis of nine randomized controlled studies in 1538 patients, sotalol reducedthe recurrence rate by 47% [289]. In the Canadian Trial of Atrial Fibrillation (CTAF), sotalol was inferior to amiodarone for the long-term maintenance of sinus rhythm ( Fig. 29.32) [311]. In the SAFE-T study, sotalol 160mg twice daily was superior to placebo, but less effective than amiodarone in prevention of AF recurrence after electrical cardioversion [19]. At 2 years, approximately 30% of patients treated with sotalol remained in sinus rhythm compared with 60% of patients on amiodarone and 10% of patients on placebo. The efficacy of sotalol was similar to that of class I antiarrhythmic drugs and inferior to that of amiodarone in the AFFIRM sub-study (48%, 45%, and 66%, respectively) [312].

Hypotension and bradycardia were the most common cardiovascular adverse effects of sotalol with an incidence of 6–10%, while ventricular proarrhythmias associated with prolongation of the QT interval were reported in 1–4% of patients and were dose related [308, 310]. Ventricular proarrhythmia is a relevant concern and is often related to QT prolongation.

Dofetilide is relatively safe to use in patients with previous myocardial infarction and/or congestive heart failure. In the DIAMOND AF sub-study of DIAMOND-CHF and DIAMOND-MI trials, 506 patients with AF at baseline were more likely to remain in sinus rhythm on treatment with dofetilide 500mcg twice daily compared with placebo (79% vs. 42%) [236]. Patients treated with dofetilide had a lower incidence of new onset AF than those on placebo (1.23% vs. 3.78%) [311, 312]; this effect was more pronounced in patients with NYHA class III and IV heart failure enrolled in DIAMOND-CHF (1.98% vs. 6.55%) [313]. Dofetilide was almost twice as effective for the long-term prevention of atrial flutter than of AF (73% vs. 40%) [266].

The major safety concern about dofetilide is its torsadogenic potential which is dose related. For instance, the incidence of torsade de pointes with dofetilide ranges from almost ‘zero’ at doses below 250mcg BID to 10% or greater at doses higher than 500mcg BID, with more than three-quarters of episodes occurring during the first 3–4 days of drug initiation [234]. Dofetilide is excreted predominantly via kidneys and its dose should be adjusted for creatinine clearance (see Table 29.9); the drug should not be prescribed in patients with significantly impaired renal function (creatinine clearance <20), hypokalaemia, hypomagnaesemia, or a QT interval of >500ms. If the QT interval is prolonged to >500ms or by >15% versus baseline, the dose should be reduced.

Amiodarone is the best available antiarrhythmic drug for maintenance of sinus rhythm in patients with advanced heart disease or in patients in whom class IC agents or sotalol were ineffective. Because of its neutral effect on all-cause mortality [274, 315, 316], amiodarone is considered a drug of choice for management of AF in patients with congestive heart failure, hypertrophic cardiomyopathy, and hypertension with significant LV hypertrophy.

In the CTAF trial of 403 patients with paroxysmal and persistent AF, amiodarone administered at 10mg/kg for 2 weeks, followed by 300mg per day for 4 weeks and a maintenance dose of 200mg reduced the incidence of recurrent AF by 57% compared with sotalol 160–320mg per day or propafenone 450–600mg ( Fig. 29.32) [309]. Patients who received amiodarone at a maintenance dose of 300mg per day in the CHF-STAT study had fewer recurrences, and were half as less likely to develop new AF compared with placebo [272].

Despite prolonging cardiac repolarization, amiodarone has a low (<1%) torsadogenic potential [234, 274]. The residual risk of torsade de pointes with amiodarone occurs mainly in patients with other risk factors, such as bradycardia or hypokalaemia. The reason for the low propensity of amiodarone to cause torsade de pointes is not clear, but is presumably related to its complex mode of action which involves class I, II, and IV effects alongside its class III properties and a low propensity to increase the heterogeneity of refractoriness across the myocardial layers.

A significant downside of amiodarone is multiple non-target effects, which range from transient and relatively trivial (e.g. gastrointestinal disturbances), to partially preventable (e.g. skin toxicity) and medically correctable (e.g. underactive thyroid), to serious such as pulmonary toxicity, liver damage, hyperthyroidism, bradycardia, significant or irreversible neurological symptoms (e.g. peripheral neuropathy), and visual disturbances (e.g. optic neuritis). Amiodarone is therefore not recommended as first-line therapy in patients with little or no structural heart disease for whom therapy with class IC drugs or sotalol is more appropriate. As many serious adverse effects of amiodarone develop after prolonged therapy (years), amiodarone therapy is less appropriate for younger patients.

Dronedarone is a structural analogue of amiodarone which is devoid of iodine atoms and is believed to have a better side-effect profile with lower risk of pulmonary fibrosis, ocular adverse effects, and skin photosensitivity. Dronedarone 400mg BID is moderately effective in preventing AF recurrences after electrical cardioversion [317, 318]. In the EURIDIS (EURopean trial In atrial fibrillation or flutter patients receiving Dronedarone for the maintenance of Sinus rhythm), the median time to the recurrence of AF was 41 days in the placebo group and 96 days in the dronedarone group ( Fig. 29.33) [318]. In the American–Australian–African equivalent of the European trial (ADONIS), the median time to recurrence was 59 on placebo and 158 days on dronedarone.

The post hoc analysis of the EURIDIS and ADONIS studies has shown that patients treated with dronedarone had a 27% reduction in relative risk of hospitalization for cardiovascular causes and death [318]. The beneficial effect of dronedarone on survival has been confirmed in a further study called ATHENA (A placebo controlled, double blind Trial to assess the efficacy of dronedarone for the prevention of cardiovascular Hospitalization or death from any cause in patiENts with Atrial fibrillation and flutter) in >4000 high-risk patients with AF ( Fig. 29.33) [319]. In ATHENA, dronedarone reduced cardiovascular hospitalizations or all-cause death by 24% compared with placebo. This effect was driven by the reduction in cardiovascular hospitalizations (by 25%), particularly hospitalizations for AF (by 37%). The beneficial effect of dronedarone on cardiovascular hospitalizations and death was consistent across all subgroups of patients, including those who remained in AF throughout the study. In addition, dronedarone reduced the ventricular rate response during AF by 10–15bpm. All-cause mortality was similar in the dronedarone and placebo groups (5% and 6%, respectively); however, dronedarone significantly reduced deaths from cardiovascular causes.

Unlike the trials which reported a beneficial effect of dronedarone on cardiovascular mortality, the ANDROMEDA (ANtiarrhythmic trial with DROnedarone in Moderate to severe heart failure Evaluating morbidity DecreAse) study specifically enrolled patients with NYHA functional class III or IV heart failure and recent heart failure decompensation. The trial was stopped prematurely after 627 patients out of the 1000 planned were enrolled, because an interim safety analysis revealed an excess of deaths in the dronedarone arm compared with placebo (8% vs. 13.8%; hazard ratio 2.13) [320]. The risk of death was the greatest in patients with severely depressed ventricular function and there were more hospitalizations for heart failure in the dronedarone arm.

The adverse outcome in ANDROMEDA is in part thought to be associated with more frequent discontinuation of ACE inhibitors or angiotensin receptor blockers in patients who received dronedarone because of increases in creatinine levels which were misinterpreted as progressive renal failure. These increases were secondary to the now known inhibitory effect of dronedarone on renal tubular excretion of creatinine. Consequently, excess mortality in the dronedarone group was related to worsening heart failure. In the subsequent analysis of a small proportion of patients (n = 200) with NYHA class III heart failure, many of who had an ejection fraction of <35%, therapy with dronedarone was, in fact, associated with a lower likelihood of hospitalizations or death from cardiovascular causes as well as lower all-cause mortality compared with placebo. However, withdrawal of potentially life-saving therapy does not explain all fatalities in ANDROMEDA, and dronedarone is therefore contraindicated in patients with NYHA class IV heart failure.

A raft of other amiodarone analogues (e.g. celivarone) and amiodarone-derivative with modified bonds within the molecule is currently at various stages of development ( Fig. 29.28) [107]. An oral formulation of vernakalant 600mg BID has been reported to be useful for prevention of AF recurrence after electrical cardioversion, with a modest superiority to placebo (51% vs. 37%) [107]. There is evidence that an antianginal drug, ranolazine, may also produce an antiarrhythmic effect due to multiple channel blockade, particularly late sodium current blockade.

Antiarrhythmic drugs can be used to facilitate electrical cardioversion and to prevent immediate or early recurrence of AF. Synergistic action of antiarrhythmic drugs may be due to prolongation of atrial refractoriness, conversion of AF to a more organized atrial rhythm (e.g. flutter) which may be cardioverted with less energy, the suppression of atrial premature beats that may re-initiate AF, and prevention of atrial remodelling. The disadvantages are increased risk of ventricular proarrhythmia and bradycardia [321].

Pre-treatment with intravenous ibutilide, flecainide, or sotalol lowered the energy requirement by around 30% and improved the success rate of cardioversion, including previously failed electrical cardioversion [321–324]. Slightly higher rates of restoration and maintenance of sinus rhythm were reported in patients pre-treated with oral amiodarone, propafenone, verapamil, and diltiazem, but the evidence is inconsistent [325–327].

Risk of recurrence is increased (25–50%) during the first 1–2 months after electrical cardioversion. It is therefore important to continue antiarrhythmic drug therapy if risk of recurrence is deemed to be high (e.g. in patients who had previously reverted to AF). The CONVERT (CONtinuous Vs Episodic pRophylactic Treatment with amiodarone) study reported that patients who stopped amiodarone after 1 month of sinus rhythm following cardioversion had a higher incidence of recurrent AF (80% vs. 54%), as well as all-cause mortality and cardiovascular hospitalizations (53% and 35%) during a median follow-up of 2.1 years compared with those who continued amiodarone treatment [328].

After LA ablation therapies, the incidence of AF or atrial tachycardia has been reported to be 45% during the first 3 months despite antiarrhythmic drugs [99]. In many cases, early recurrence of AF after ablation is transient and is thought to be secondary to inflammation after radiofrequency injury, nerve ending damage and resulting imbalance of the cardiac autonomic nervous system, and a delayed effect of ablation associated with ‘maturation’ of lesions.

Early AF often subsides after 3 months upon the resolution of inflammation and restoration of autonomic regulation, without the need for re-ablation. Therefore antiarrhythmic drug therapy to suppress early recurrence of AF is commonly employed for the first 1–3 months after ablation or if the arrhythmia recurs after discontinuation of the antiarrhythmic drug. These are usually the same agents that have been previously ineffective, but may now be efficacious because of a synergistic effect with ablation. Amiodarone has been reported to be most commonly prescribed because of its antifibrillatory as well as AV conduction slowing properties. The synergistic effect of propafenone, flecainide, or sotalol has been demonstrated in the 5A (AntiArrhythmics After Ablation of Atrial fibrillation) study. The use of antiarrhythmic drugs increases the likelihood of staying in sinus rhythm by about 30% [329].

The target heart rate for acute rate control is 80–100 beats per minute (bpm) during AF. In patients with rapid ventricular rate, intravenous verapamil, diltiazem, and beta-blockers (usually metoprolol or the rapidly-eliminated esmolol) are commonly used for rapid ventricular rate control ( Table 29.10) [330]. All drugs are equally effective in reducing the ventricular response rate by approximately 20–30% in 20–30min and have a similar risk of adverse effects (usually hypotension and bradycardia; although LV dysfunction and high-degree heart block may occur). Beta-blockers are preferable in patients with a history of myocardial infarction or if thyrotoxicosis is suspected as a cause of the arrhythmia, whereas verapamil and diltiazem are preferred in patients with acutely exacerbated chronic obstructive airways disease.

Intravenous digoxin is no longer the treatment of choice for acute rate control because of delayed onset of its therapeutic effect (>60min). However, because digoxin has a positive inotropic effect, it is a reasonable adjunct to beta-blockers in patients with impaired LV function and moderately fast ventricular rates.

Agents with the primary effect on the AV node should not be used for rate control in patients with known or suspected pre-excitation syndrome: these agents will only affect AV nodal conduction and will have no effect on rapid conduction via the accessory pathway. In patients with accessory pathways, sodium-channel blockers (e.g. intravenous ajmaline or flecainide) can slow anterograde conduction via the accessory pathway.

Intravenous amiodarone can be used for acute ventricular rate control when other agents have no effect on ventricular response or are contraindicated, for example, in haemodynamically unstable AF refractory to electrical cardioversion [331]. The AV blocking effect of amiodarone is complemented by its antifibrillatory action. The disadvantages of intravenous amiodarone are slow onset of action and increased risk of phlebitis. Sotalol can also slow down the ventricular rate due to its beta-blocking properties [332], but its negative inotropic effect in patients with LV dysfunction and risk of torsades due to QT interval prolongation reduce its value as a rate controlling agent.

There is limited evidence of the use of clonidine (due to its central sympatholytic activity) [333] and magnesium [285] for ventricular rate control. In a meta-analysis, intravenous magnesium reduced the ventricular rate to <90–100 within 5–15 min of infusion. Adenosine derivatives with a high specificity for adenosine A1 receptors and longer half-life periods (e.g. tecadenoson, selodenoson, and capadenoson) are currently under investigation [107, 334].

Rate control is an essential constituent of management of AF and is pertinent to all types of the arrhythmia. In paroxysmal and persistent AF, rate control is important during the recurrence or prior to electrical cardioversion. Verapamil, diltiazem, and beta-blockers (metoprolol, atenolol, bisoprolol, carvedilol, and nadolol) are drugs of choice. They are also commonly prescribed in combination with the class IC agents, propafenone and flecainide, to prevent fast ventricular rates due to 1:1 AV conduction when recurrent AF evolves to flutter.

Digoxin has been shown to reduce the ventricular rate during symptomatic paroxysms of AF by approximately 15bpm and has probably rendered some episodes asymptomatic [335], but because of its profibrillatory effect, digoxin usually should be avoided if the arrhythmia is self-terminating or electrical cardioversion is planned and rhythm control is pursued.

Some antiarrhythmic drugs, such as sotalol, amiodarone, and dronedarone, slow AV conduction and thus offer an additional benefit of ventricular rate control during recurrences of AF without significantly affecting the overall exercise capacity [19, 272, 336–338]. However, antiarrhythmic drugs are not likely to be routinely employed for long-term rate control, e.g. in permanent or accepted AF, because of potential proarrhythmic and non-target side effects.

Current guidelines define adequate rate control as maintenance of the ventricular rate response between 60–80bpm at rest and 90–115bpm during moderate exercise, but few systematic studies explored the effect of rate slowing drugs on chronotropic competence in AF or defined upper limits of the appropriate ventricular rate response during exercise [339] and no controlled clinical trials have validated these values with regard to their effects on morbidity or mortality. In post hoc pooled analysis of the AFFIRM and RACE studies, patients with mean ventricular rates during AF within the AFFIRM (≤80bpm) or RACE (<100bpm) criteria had a better outcome than patients with ventricular rates ≥100bpm (hazard ratio 0.69 and 0.58, respectively for ≤80 and <100 compared with ≥100bpm) [340].

Although rapid ventricular rates can be detrimental, too slow heart rate can be problematic, particularly in patients with impaired diastolic filling, hypertension, and LV hypertrophy when loss of atrial contraction may cause a marked decrease in cardiac output. Furthermore, rhythm irregularity per se may contribute to ventricular dysfunction [341].

Ventricular rate control at rest does not always translate into effective control during exercise. Ambulatory 24-hour ECG monitoring is considered sufficient for assessment of rate control in elderly and sedentary individuals, whereas in younger individuals, an exercise stress test may be necessary ( Figs. 29.34 and 29.35). RACE II was instigated to assess whether maintaining strict rate control, i.e. mean resting heart rate <80bpm and heart rate during mild exercise <110bpm, can offer any additional benefit to standard practice [342].

Drugs for rate control. Digoxin, beta-blockers, verapamil, and diltiazem, are commonly employed ( Table 29.11). Digoxin is effective in controlling ventricular rates at rest as it prolongs AV node conduction and refractoriness through vagal stimulation, by direct effects on the AV node, and by increasing the amount of concealed conduction. However, the effect of digoxin is negated during exercise when most vagal tone is lost and AV conduction is further enhanced by the increased sympathetic tone. Digoxin alone was effective in only 58% of patients [343]. Therefore, digoxin as monotherapy can be used in older, sedentary patients, but a combination with beta-blockers or calcium antagonists is often necessary to achieve rate control in the majority of patients ( Fig. 29.35).

Non-dihydropyridine calcium antagonists and beta-blockers are effective as primary pharmacological therapy for rate control, but multiple adjustments of drug type and dosage may be needed to achieve the desired effect [343–345]. For example, in the AFFIRM trial, only 58% of patients in the rate-control group achieved adequate rate control with the first drug or combination; drug switches occurred in 37% of the patients and drug combinations were commonly used [343]. Overall, adequate rate control was ascertained in 80% of patients at 5 years (most frequently on a beta-blocker with or without digoxin).

Patients with congestive heart failure are particularly prone to the adverse effects of antiarrhythmic drugs. Electrical cardioversion may be considered in younger patients with short arrhythmia duration who have compensated heart failure. Amiodarone or dofetilide are the drugs of choice to prevent recurrences as they have been shown to be safe and effective in this context [236, 272, 313–315]. Neither drug is associated with deterioration of LV function nor predisposes to proarrhythmic effects as long as they have been initiated and followed carefully. ECG monitoring of the QT interval is of special importance in this setting.

All patients with LV dysfunction should also be treated with beta-blockers and ACE inhibitors or angiotensin receptor blockers because, in addition to their proven beneficial effect on survival, they may delay atrial remodelling and deter onset of new AF and possibly prevent recurrent AF (see ‘Upstream’ therapy, p.1113) [34]. The magnitude of their effect may be modest but if applied to a very large population, the outcome could be significant.

Postoperative AF occurs predominantly during the first 4 days. More than 90% of patients present with a paroxysmal or first-onset form of the arrhythmia. Atrial flutter and atrial tachycardias, including multifocal atrial tachycardia, are also common. Electrical or pharmacological restoration of sinus rhythm with subsequent prophylactic antiarrhythmic therapy should be considered in haemodynamically unstable or highly symptomatic patients with postoperative AF. If AF is well tolerated, rate control may be sufficient as AF after isolated coronary bypass surgery is often self-limiting and there is a high likelihood of spontaneous conversion to sinus rhythm, usually within 6 weeks [346].

Beta-blockers should be considered the first-line treatment because of their beneficial effects on the hyper-adrenergic postoperative state. The best evidence of the efficacy in prevention of postoperative AF has been accumulated for beta-blockers, sotalol, and amiodarone ( Fig. 29.36) [346]. Two earlier meta-analyses of randomized controlled studies have shown that treatment with beta-blockers may reduce the incidence of postoperative AF by approximately 50% [347, 348]. In the recent meta-analysis of 27 trials in 3840 patients, therapy with beta-blockers reduced risk of postoperative AF by 61%; there was also a trend towards shorter length of stay [346]. The downside of beta-blockers is that they may increase risk of bradycardia (5–10%) and risk of longer ventilation (1–2%).

Magnesium (20 studies, 2490 patients), amiodarone (19 studies, 3295 patients), and sotalol (eight studies, 1294 patients) reduced the incidence of postoperative AF by 46%, 50%, and 65%, respectively, compared with placebo [346]. In the Prevention of Arrhythmias that Begin Early After Revascularization (PAPABEAR) trial of 601 patients undergoing isolated coronary artery bypass surgery, treatment with amiodarone 10mg/kg per day starting 7 days before surgery, was associated with a significant decrease in the incidence of postoperative AF compared with placebo (30% vs. 16%) reflecting a 52% reduction in relative risk [349]. The use of sotalol risks bradycardia and torsade de pointes, especially in patients with electrolyte disturbances. Amiodarone may cause hypotension and bradycardia necessitating inotropic and chronotropic support or pacing.

The proarrhythmic potential and the negative inotropic effect of class I agents, ibutilide, and dofetilide offset their modest efficacy in prevention and/or conversion of AF after cardiac surgery.

Several randomized controlled studies have suggested the beneficial role of agents with anti-inflammatory properties, such as statins, or specific anti-inflammatory drugs such as corticosteroids [350, 351]. In the randomized prospective, double-blind, placebo-controlled study, ARMYDA-3 (Atorvastatin for Reduction of MYocardial Dysrhythmia After cardiac surgery), in 200 patients undergoing elective coronary bypass grafting surgery pre-treatment with atorvastatin starting 7 days before surgery was associated with a 61% reduction in the incidence of in-hospital AF [350]. Pre-treatment of patients with poly-unsaturated fatty acids for 5 days before coronary bypass grafting surgery reduced the occurrence of postoperative AF by 65% [352].

It has been recently appreciated that primary and secondary prevention of AF with ‘upstream’ therapy and risk factor modification is likely to produce a larger effect in the general population than will specific interventions [34, 233]. Angiotensin II, whose local synthesis is increased in AF, has been recognized as a key element in atrial remodelling. Angiotensin II produces a variety of direct and indirect effects on atrial structure and electrophysiology by stimulating atrial fibrosis and hypertrophy, promoting inflammation, modifying ion channels, uncoupling gap junctions, and disrupting calcium handling. Aldosterone, which can also be produced locally in the heart, may stimulate mediators of inflammation, activate fibroblasts and matrix metalloproteinases (MMPs), and probably, has direct electrophysiological effects.

There is accumulating evidence that, beyond its therapeutic effects on underlying heart disease, such as hypertension and heart failure, inhibition of the renin–angiotensin–aldosterone system may offer some protection against atrial structural and possibly electrical remodelling associated with AF ( Fig. 29.37) [34, 233].

Meta-analysis of several retrospective reports as well as prospective studies in patients with congestive heart failure and hypertension has reported that therapy with ACE inhibitors and angiotensin receptor blockers reduced risk of new-onset AF by approximately 20–30% [116]. Furthermore, the results of small prospective studies of secondary prevention of AF have uniformly demonstrated a significant reduction in AF recurrence associated with ACE inhibitor or angiotensin receptor blocker plus amiodarone therapy compared with amiodarone alone before and after electrical cardioversion [34, 233]. However, the results of the GISSI-AF study which enrolled 1442 AF patients with typical risk factors (heart failure, hypertension, diabetes, CAD or peripheral artery disease) as well as patients with lone AF but with risk factors for AF such as a dilated LA, were disappointing [352]. Patients, the majority of whom had hypertension as a primary diagnosis, were randomized to receive valsartan 320mg daily or warfarin. There was no difference in the primary endpoint of time to first AF recurrence of AF (hazard ratio, 099, 95% CI, 0.85–1.15; p = 0.84) as well as no difference in secondary point of all-cause hospitalizations and hospitalizations for cardiovascular causes in the overall patient population or pre-specified groups. Several large prospective trials will assess the antiarrhythmic value of renin–angiotensin–aldosterone blockade in the absence of formal indications, e.g. myocardial infarction, heart failure, or severe hypertension.

Increased levels of CRP and pro-inflammatory cytokines (e.g. interleukin-1b, interleukin-6, and tumour necrosis factor-α) in AF have been reported in epidemiological and observational studies [35]. Higher CRP levels have been shown to be associated with a greater incidence of AF in the general population [123]. Recent meta-analysis of seven prospective observational studies has demonstrated that baseline CRP levels determine freedom from AF recurrence after electrical cardioversion [122].

Statins have strong anti-inflammatory and antioxidant effect and may, therefore, abolish AF mediated through these mechanisms [353]. By reducing oxidative stress and oxidized low-density lipoproteins which can up-regulate angiotensin II type 1 receptors, statins may counteract the arrhythmogenic effects of angiotensin II. In addition, statins may change ion channel conductance by altering the lipid content of the membrane. By increasing the synthesis of nitric oxide in the endothelium, statins protect atrial myocytes during ischaemia associated with rapid atrial rates and they may regulate the variety of MMPs which play a role in extracellular matrix remodelling in AF.

Meta-analysis of six randomized controlled studies in 3557 patients has shown that the use of statins was associated with a 61% reduction in the incidence of recurrent AF, but the effect was driven by the decrease in postoperative AF [355]. Several well-designed prospective trials were initiated to assess the antiarrhythmic value of statins. The role of polyunsaturated fatty acids in AF is controversial, but prospective studies are under way.

There is insufficient evidence to warrant a recommendation to expand the indications in wider patient populations at risk of AF.

Direct-current cardioversion of AF was first reported by Lown in 1963 [354]. The term ‘direct-current cardioversion’ implies delivery of an electrical shock synchronized with the intrinsic activity of the heart (QRS complex) to avoid electrical discharge during the ventricular vulnerable period, when there is a risk of inadvertent induction of ventricular fibrillation. Usually, the R wave of the surface ECG is chosen for this synchronization since it can be easily sensed by the defibrillator.

External electrical cardioversion is achieved by means of cutaneous electrode paddles positioned directly on the chest wall. Cardioversion is performed with the patient having fasted and under adequate general anaesthesia (or sedation) in order to avoid pain related to the delivery of the electrical shock. Success of cardioversion is determined by density of the current that traverses the muscle of the chamber to be defibrillated. The intensity of current that flows depends on the output waveform, selected energy level, and the transthoracic impedance. The higher the impedance, the lower the current delivered. The determinants of transthoracic impedance mainly include body habitus, the interface between the electrode and skin, and the size and position of the electrode paddles. For successful cardioversion, a critical mass of atrial muscle has to be encompassed by the electrical field. This is the rationale for using the anteroposterior paddle position rather than the anteroapical paddle position. Some randomized studies have shown greater success with anteroposterior configuration [357–359]. Since the anteroposterior paddle position has never been shown to be less effective than the anteroapical, this configuration is the first choice in clinical practice (29.6). Because the optimum paddle configuration for a given patient is not known before cardioversion, the clinician should consider an alternative arrangement if the initial position is unsuccessful.

Most devices used for external cardioversion deliver current with a monophasic waveform ( Fig. 29.38) with a maximum energy output limited to 360J. The success rate for cardioversion of persistent AF is usually around 80% and is related to several clinical parameters such as long arrhythmia duration, patient age, LA enlargement, the presence of underlying heart disease, cardiomegaly, and obesity [360, 361].

The most modern type of external defibrillator delivers current with a biphasic waveform ( Fig. 29.38). The maximum energy output is limited to 200J. Randomized studies have shown that biphasic defibrillators have a greater efficacy, require fewer shocks and lower delivered energy, and result in less skin burns than monophasic defibrillators [362–364]. The efficacy of transthoracic cardioversion was >90% with a biphasic shock waveform. Starting with higher energies may reduce the number of shocks (and thus total energy) delivered.

After one or two failed cardioversion attempts with maximum energy output with both paddle positions, antiarrhythmic drugs before further shock delivery, double shocks (delivery of energy with the use of two defibrillators) or internal cardioversion may be considered. Internal cardioversion of AF using high-energy (200–300J) direct current delivered between a catheter positioned in the right atrium and a backplate has been shown in a randomized trial to achieve higher conversion rates using a monophasic defibrillator for external cardioversion in the control group [365]. This can be useful in obese patients and patients with chronic obstructive lung disease, particularly if an electrophysiology study is planned for other purposes.

Other techniques for internal cardioversion apply low-energy (<20J) shocks via a large-surface electrode catheter (cathode) in the right atrium and another catheter (anode) positioned in the coronary sinus or the left pulmonary artery [366, 367]. Transoesophageal cardioversion has also been studied as an alternative approach for external cardioversion. With this technique, intermediate-level energy (20–50J) is delivered between oesophageal electrodes and a mid-sternum patch. It has been proved to be safe and efficacious [368] and can be combined with transoesophageal echocardiography to ensure that no atrial thrombus is present prior to cardioversion.

Transient ST-segment elevation may appear on the ECG after cardioversion and serum levels of creatine kinase may be increased whereas troponin-T and troponin-I levels are not. Myocardial damage related to electrical cardioversion is not observed at a microscopic level.

Cardioversion is contraindicated in patients with digitalis toxicity because a malignant ventricular arrhythmia may be triggered by the direct-current shock. However, at therapeutical levels, digoxin is not associated with the induction of malignant ventricular arrhythmias during cardioversion. Because hypokalaemia may precipitate a malignant ventricular arrhythmia after cardioversion, serum potassium levels should be in the normal range before direct-current shock delivery. Appropriate anticoagulation prior to cardioversion is mandatory.

Device-based therapy has been evaluated to treat AF with regards to two major concepts: the first concept seeks to avoid AF initiation by alleviation of bradycardia-induced dispersion of atrial activation and repolarization, and atrial overdrive suppression of premature atrial beats (‘preventive pacing’). The second concept is to terminate AF episodes by high-rate pacing once the arrhythmia has started (‘anti-tachycardia’ pacing).

The potential to maintain AV synchrony and prevent the development of mitral regurgitation and/or ventriculoatrial conduction that could cause stretch-induced changes in atrial repolarization may lessen the chance of AF recurrence.

Atrial or dual-chamber pacing has shown some benefit over ventricular pacing alone in decreasing episodes of AF [369–372]. Meta-analysis of five major pacing mode selection trials, which included a total of 35,000 patient-years of follow-up, has shown a statistically significant 20% reduction in AF with atrial-based pacing and a 19% decrease in stroke that was of borderline significance ( Fig. 29.40) [373]. Patients with sinus node dysfunction and preserved AV conduction seem to benefit most from atrial or dual-chamber pacing. However, atrial-based pacing modes had no effect on mortality or the development of heart failure.

It has been hypothesized that excessive right ventricular stimulation during dual-chamber pacing may worsen LV function and thus may negate the physiological advantage of AV synchrony and preservation of sinus node control over heart rate [374]. Thus, in patients with sinus syndrome and preserved native AV conduction enrolled in the MOST (Mode Selection Trial), the incidence of AF at 33.1 months (≈2.7 years) was 26.7% in the group assigned to ventricular-based pacing and 15.2% in the group assigned to physiological pacing [371].

The use of specific algorithms to minimize unnecessary ventricular stimulation in dual-chamber pacemakers has had no significant beneficial effect on mortality or heart failure, but has further reduced the risk of AF, in particular persistent AF, by 40% compared with conventional dual-chamber pacing [375]. In the SAVE PACe (Search AV Extension and Managed Ventricular Pacing for Promoting Atrioventricular Conduction) trial, which enrolled patients similar to those in the MOST study, the incidence of AF after 1.7 years was 12.7% in the group treated with conventional dual-chamber pacing and 7.9% in the group who treated with dual chamber-pacing plus an algorithm reducing unnecessary ventricular pacing [375].

Continuous pacing at selected sites (alternative, dual or bi-atrial) may change atrial activation patterns, increases homogeneity of left and right atrial electrical properties in conduction and refractoriness, reduce dispersion of refractoriness that occurs with premature atrial contractions (PACs) or with abrupt changes in atrial rate and thus prevent the development of atrial re-entry.

Alternative pacing sites: experimental and clinical studies have demonstrated that septal pacing, dual-site, or bi-atrial-site pacing shortens total atrial activation time [376, 377]. A number of clinical trials have evaluated the effects of selective atrial pacing sites on the prevention of AF in patients that required a pacemaker for other indications than AF, mainly with bradycardia as an indication (Bachmann bundle, inter-atrial septum) [377]. However, the largest randomized trial of right atrial pacing versus septal pacing ASPECT (Atrial Septal Pacing Efficacy Clinical Trial) failed to demonstrate a reduction in AF frequency and burden over a short follow-up [378].

Several techniques for bi-atrial stimulation have been proposed (right atrial and coronary sinus pacing) in patients with sinus node dysfunction. In the DAPPAF (Dual site Atrial Pacing to Prevent Atrial Fibrillation) trial in antiarrhythmic drug-treated patients, dual right atrial pacing increased symptomatic AF-free survival compared with support pacing or high right atrial pacing, supporting the use of a hybrid approach to rhythm management [379].

Preventative pacing algorithms: the Atrial Fibrillation Therapy (AFT) trial, which investigated modes of onset of AF to determine potential arrhythmogenic triggers, has shown that AF is most commonly caused by premature atrial complexes (PACs) (48%) and sinus bradycardia (33%), whereas sudden onset was less common (17%) ( Fig. 29.41) [380].

Specific algorithms are designed to prevent bradycardia and to avoid significant atrial rate variations associated with PACs. These include rate-adaptive pacing that monitors the underlying intrinsic rate to pace just above it, elevation of the pacing rate after spontaneous PACs, transient overdrive pacing after mode switch episodes, and increased post-exercise pacing to prevent an abrupt drop in heart rate.

The results of randomized clinical trials completed to date are not conclusive with respect to the beneficial effects of atrial pacing for AF and to the definition of an AF population (with except probably of patients with sinus node disease) that would obtain the most advantage from pacing strategies and which of these strategies are the best [380–384].

Atrial pacing prevention algorithms have modest to minimal incremental benefit for the prevention of AF and are not warranted in the absence of bradycardia indications for pacing [385]. At present, AF alone should not be an indication for preventative pacing, except for documented vagally-mediated AF.

In some episodes, AF can be controlled with antitachycardia devices through the use of antitachycardia pacing or high-frequency atrial burst pacing. This approach is based on the concept that even AF may be sufficiently organized at its onset to allow pacing intervention, tissue capture, and arrhythmia termination [380]. Pace termination of atrial tachycardias or atrial flutter may prevent the development of AF and reduce the overall AF burden. However, when added to a pacing prevention algorithm, no further reduction of AF burden was demonstrated in a prospective randomized trial [381, 386]. In addition, there is so far no trial that demonstrates any long-term efficacy of device algorithms for both prevention or termination of AF.

Initial short-term clinical experience with the ‘stand-alone’ atrial defibrillator suggested that atrial defibrillation was safe and effective. The first generation of atrial defibrillator was followed by a commercially available dual-chamber AV defibrillator. However, in a long-term analysis of 106 patients, only 39 were still actively using their device after a median of 40 months. In half, the device had been turned off or even explanted and in 14 patients it was solely used to monitor AF episodes [387]. The major drawback of internal cardioversion devices is that even a relatively low energy shock (<1J) is intolerably painful for the patient, making repetitive shocks in patients with frequent AF episodes a very unattractive treatment strategy [388].

For very selected patients who already have indications for implantable devices, device-based atrial defibrillation may be an option as a ‘backup’ option for managing AF when preventive pharmacological/electrical measures fail ( Fig. 29.42). There is limited evidence that the combination of antitachycardia pacing, atrial shock, and preventative pacing is effective in reducing AF burden [389, 390].

Hybrid therapy involves the use of different types of treatment which together provide some form of synergism [389]. The three available rhythm management therapies are antiarrhythmic drugs, ablation, and devices. Antiarrhythmic drugs are often used in combination with ablation or device therapy. Increasingly, ablation and device therapy is combined. Although efficacy may be improved, side effects from both therapies may occur. Hybrid therapy does not refer, however, to the concomitant use of anticoagulation or upstream therapies. This remains largely a theoretical concept that has only been strictly evaluated in a small number of studies. Principles of hybrid therapy are listed in Table 29.12.

Drug-induced atrial flutter: although atrial flutter and AF have different electrophysiological mechanisms, the two arrhythmias often coexist. Atrial flutter may initiate the first re-entrant circuit and may transform into AF as multiple wavelets develop [392]. The usual flutter pathway

may continue to perpetuate AF whenever engaged. Typical atrial flutter is a relatively common finding in patients with AF treated with class IC and IA drugs and amiodarone ( Fig. 29.43). These agents may modify wavefront activation by creating lines of functional block that interrupt multiple wavelets and promote conduction preferentially through a large single re-entrant circuit such as the isthmus-dependent flutter circuit. In these patients, ablation of the cavo-tricuspid isthmus can be considered first-line therapy before proceeding to LA ablation.

A less appreciated example of hybrid therapy is pacing for drug-induced bradycardia which allows dose up-titration to achieve the best efficacy of antiarrhythmic drugs, although the intentional use of such combination therapy is limited [393].

Other examples of hybrid therapy include pacing-induced reduction of PACs and antiarrhythmic drug modification of substrate; device-based monitoring of rate/rhythm control in AF treated by pacing or antiarrhythmic drugs; biventricular pacing and ablation of the AV node in patients with AF and heart failure.

AF patients may benefit from control of both the ventricular rate and its regularity. Therefore, ventricular pacing after AV junction ablation to induce heart block (’ablate and pace’) is a useful strategy for patients with permanent AF in whom rate control is difficult to accomplish by drugs ( Fig. 29.44). The benefits of this procedure have been demonstrated in many controlled and non-controlled trials [394]. AV nodal ablation is especially useful when an excessive ventricular rate induces a tachycardia-mediated decline in ventricular systolic function despite appropriate medical therapy.

The ‘ablate and pace’ strategy improves exercise tolerance, cardiac function, healthcare utilization, and quality of life. Its safety has been established, and in a meta-analysis the risk of sudden death and total mortality was 2% and 6% respectively at 1 year [395], figures similar to the mortality observed with medical therapy of AF [396]. After total AV nodal block, ventricular pacing is performed from the apex of the right ventricle. This ‘non-physiologic’ activation from apex to base enhances cardiac dyssynchrony and can result in worsening of ventricular function, an effect which explains the deterioration of some patients [397].

By only modifying the AV node conduction properties to decrease the ventricular rate during AF (‘AV node modulation or modification’), complete AV block and lifetime pacemaker therapy could be avoided. However, this technique is less widely performed because of lower success rates, the risk of inadvertent AV block, and the persistence of symptoms due to irregular beats during AF [398, 399].

In the presence of congestive heart failure and AF, a significant benefit has been demonstrated for improvement of LV ejection fraction when a cardiac resynchronization therapy (CRT) device is implanted. Meta-analysis of five studies which compared the effect of CRT on outcomes in 1164 patients with sinus rhythm or AF, has shown that patients with AF have significant improvement after CRT, with similar improvement in ejection fraction as patients in sinus rhythm [400]. However, patients with AF seemed to gain smaller benefits in functional outcomes (NYHA functional class, 6-min walk, and quality of life).

The effect of CRT on outcome is significantly enhanced if the AV node is ablated [401, 402]. Recently, a large registry suggested that AF patients who received a CRT device (‘biventricular pacemaker’) may have a better outcome with an ‘ablate and pace’ therapy than with pharmacological rate control [403].

On the other hand, AV node ablation and pacing pre-empts the later use of potentially newer and more effective non-pharmacological or pharmacological treatments, and should generally be reserved for patients refractory to other therapies. A specific atrial preventative pacing algorithm has been tested as adjunct to CRT, but although proven safe, it has had no effect on the 1-year incidence of AF [404].

In principle, the maintenance of sinus rhythm is of significant benefit for patients with AF [244]. The means to achieve stable sinus rhythm by antiarrhythmic medication however, are limited in their efficacy and have potential adverse side effects that can offset the benefits. A better means to achieve sinus rhythm without the side effects of antiarrhythmic medication would be advantageous and should be superior to medical treatment.

The first reports on catheter ablation of AF date back to 1994 [405]. Since then, multiple different strategies and technologies have been investigated ( Table 29.13) with variable success. These developments have moved AF ablation from an experimental option in highly selected patients, to a reproducibly effective treatment option (PV isolation) for symptomatic patients with paroxysmal AF. However, ablation of persistent AF is still a challenge, with inconsistent results and unresolved acute ablation endpoints.

Clinical AF results from the complex interaction of triggers with the perpetuators and substrate that then maintain fibrillation [406]. While theoretically all myocardial cells could act as a ‘self-depolarizing’ source of AF triggers, the PVs are clearly the dominant source, in up to 60–94% of paroxysms of AF ( Fig. 29.45) [93, 407, 408]. Anatomical studies have demonstrated extension of a sleeve of atrial musculature into the PVs, thus creating the milieu for preferential conduction, unidirectional conduction block, and re-entry [409, 410]. Clinical studies have demonstrated that the PVs and the adjacent antrum of patients with AF have distinctive electrophysiological properties characterized by shorter refractory periods and greater anisotropy compared with patients without AF [411, 412]. These properties are capable of sustaining high-frequency activity [413, 414], especially when exposed to pulsatile stretch with every atrial contraction. Ablation of these arrhythmogenic structures forms an essential part of the ablation strategy for AF.

The initial aim of AF ablation procedures was to eliminate focal triggers from within the PVs by direct localized ablation [93]. However, a trigger could only be localized when it was actually ‘firing’, a condition that was not necessary present during the procedure and was not easily provoked (pacing, pharmacological stress, etc.) [415]. In addition, multiple sites within a PV and multiple PVs could be arrhythmogenic and intra-PV ablation led to scarring of the ablated tissue and PV stenosis and/or occlusion.

The initial experience led to the strategy of electrically isolating all PVs to prevent any interaction of these triggers with the atrial substrate [416, 417]. This approach was facilitated by circumferential mapping catheters that were positioned within the PV ostia to guide ablation targeting the ‘connecting’ fibres by ‘segmental’ ablation [416]. Ablation lesions were deployed relatively close to the PV ostia or just within ( Fig. 29.46), risking ostial stenosis and/or occlusion. In addition, high AF recurrence rates were reported mostly due to electrical re-conduction of the PVs, but also to some extend due to ‘ostial’ triggers in the presence of more distally isolated PVs [418–421].

In order to facilitate the ablation procedure and to reduce the risk of PV stenosis, the ablation sites have moved more into the atrium forming a long linear lesion around one PV or even both ipsilateral PVs together ( Fig. 29.46) [414, 422, 423], the latter having the advantage that the small space between the PV ostia (ranging from as little a several mm to cm) is spared, reducing further risk of complications [424, 425] (29.7).

There is now strong evidence suggesting that the veins and the antrum are in fact critical for maintenance of AF, rendering the classification of ‘trigger’ versus ‘substrate modification’ inadequate to fully explain the role to the PVs. In patients with paroxysmal AF, who are undergoing PVI during AF, the AF cycle length slows during ablation and AF may finally terminate in up to 75% of cases [426]. Following PV isolation of all veins, about half of patients can no longer sustain AF, suggesting that in significant proportions of patients with paroxysmal AF the PVs form the substrate maintaining AF [427].

This is a purely anatomical approach that does not require the endpoint of electrical disconnection of the encircled area ( Fig. 29.46) [428]. Since no simultaneous mapping within the PVs is performed, only a single trans-septal puncture is required. In addition, no waiting time after successful isolation is required, thereby shortening the procedure time dramatically. Using this technique up to 45% of PVs are not isolated, with persistent PV–LA conduction, which is potentially arrhythmogenic [429]. Reports from groups that have advocated the use of a purely anatomical approach have reported a significant incidence of organized arrhythmias occurring after such ablation. Indeed, a recent study reports that incomplete encircling lesions (‘gaps’) were the most predictive factor for the subsequent development of organized arrhythmias [430]. This finding argues further in favour of achieving complete lesions, with complete lesions being crucial for the prevention of macro re-entry; conversely, incomplete lesions promote the occurrence of atrial arrhythmias.

A recent expert consensus has recommended that complete electrical PV isolation should be achieved when treating patients with paroxysmal AF [99]. However, results from prospective trials investigating the need of permanently isolated ablation lines are still pending (e.g. GAP-AF trial).

Further evidence of the need for PVI is provided by studies that have evaluated the recurrence of AF after ablation. These studies have observed that the vast majority of patients with recurrence of AF demonstrate PV re-conduction. Repeat PVI in these patients has been associated with the elimination of AF in 90% of patients [424, 431]. In fact, patients with complete PV isolation fared best (SR in the absence of antiarrhythmic medication). Patients with re-conducting PVs with long conduction times were similar to patients with complete isolation, while patients with rapid conduction times required antiarrhythmic therapy and experienced AF recurrences [432]. These data implicate residual PV–LA connections in the development of clinical arrhythmia and serve to further accentuate the importance of achieving complete isolation. The rate of re-conduction of PVs in patients without symptomatic AF recurrences, however, is not known.

Despite exclusion of triggers, most patients with persistent or longstanding persistent AF may need additional substrate modification. The conceptual basis for substrate modification by compartmentalization of the atria is based on the multiple-wavelet hypothesis, which suggested that AF is maintained by multiple re-entrant wavelets propagating simultaneously in the atria and that a minimum mass of electrically continuous myocardium must be present to sustain the wavelets on re-entry.

Linear ablation is performed between anatomical or functional electrical obstacles in order to transect these regions and thereby preventing re-entry. A variety of different linear configurations has been investigated; however, prediction of which line is more suitable in a given patient has been elusive [433–436]. Typical applied lines are a roof line (connecting the upper PVs) or a so-called ‘mitral’ or ‘left isthmus’ line (connecting the inferior left PV to the mitral annulus (MA) ( Fig. 29.46). Ablation of the LA isthmus offers several theoretical benefits over the other LA lesions [436]. This line is short but creates a contiguous long lesion with the ablated PVs and does not interfere with normal sinus activation of the atria. In addition, its proximity to the coronary sinus allows positioning of catheters on either side of the line to evaluate the integrity of the line. However, despite these features, ablation at this site requires 20–25min of radiofrequency energy application, with 68% requiring ablation within the coronary sinus. Due to the close proximity to the circumflex artery, substantial harm may result including acute coronary stenosis requiring immediate intervention [437].

An anterior line (from the roof line to the MA) would need to cross the LA insertion of Bachmann’s bundle. This thick muscular connection between the right and LA is in fact the fastest connection between the atria and conduction block is difficult to achieve (due to the thickness of the tissue). However, complete anterior line deployment would result in a splitting of the P wave in sinus rhythm and in the presence of both a complete roof and LA isthmus line a completely isolated LAA [438].

A posterior line (between the septal and lateral PVs across the posterior wall of the LA) has been performed utilizing high power as part of the anatomical ablation strategy, leading in a number of patients to developed atrio-oesophageal fistulae [439].

Critical to the concept of linear lesions is that ablation has to be coalescent and transmural in order to achieve complete lesions. In some cases this can be challenging, with an increased procedural risk (particularly of tamponade, stroke and fistula formation). Incomplete linear lesions however may be pro-arrhythmic, resulting in rapid AV conduction of gap-related atrial tachycardia ( Fig. 29.47) [440–442]. The completeness of the line should be validated in order to avoid iatrogenic complications even though an initially ‘complete’ linear lesion may exhibit conduction gaps in the long term. In addition, thoughtful placement of lesions can help to prevent formation of macro-reentrant circuits which will result in LA flutters. Their incidence varies widely between studies (and centres), in part due to slightly different anatomical positions of ablation lines and lesions.

After PV isolation alone, approximately 12% of patients have inducible macro re-entry and 5% present with spontaneous macro re-entry. Mapping and ablation have demonstrated that the vast majority of these use the ablated zone as a central obstacle, resulting in either peri-mitral or peri-PV re-entry being more prevalent in patients with larger atria [429].

When an anatomical approach of PV encircling has been utilized or no line validation has been performed, the reported incidence of re-entrant arrhythmias has been much higher, with some groups reporting spontaneous arrhythmias in up to 27%, presumably due to the greater but incomplete atrial ablation [441, 443, 444].

More recently, other techniques of substrate modification have been evaluated. The ablation of complex fractionated electrograms, without any isolation attempt of PVs has been proposed [445]. It is not clear why ablation of these points should be helpful, but reports from single centres are favourable. Interestingly, arrhythmia recurrences after such procedures seem to be dominated by arrhythmias arising from the PVs [446–450].

Several groups have described the results from ablating ganglionic plexuses (at atrial sites where a vagal response is observed after local stimulation), in addition to PV isolation [451–453]. However, long-term results of a procedure limited to these ganglia are not yet available.

Lesions performed to isolate the PVs should be placed within the atria or as proximally as possible rather than within the distal PVs [454]. This strategy not only reduces the incidence of PV stenosis, but may also improve efficacy by excluding proximal foci of activity from the arrhythmogenic substrate. When ablating along the anterior aspects of the left PVs, the catheter needs to be positioned along the ‘ridge’ between the PVs and the LAA. Catheter stability and lesion formation can be difficult, but avoidance of intra-PV ablation is key.

The use of a circular mapping catheter does not imply that the ablation lesions are being performed at the PV ostia. A distance >1 or 1.5cm between both catheters is commonly observed when ablation is performed atrially and careful identification of the respective PV ostia (e.g. by angiography, intracardiac echocardiography, etc.) is strongly recommended [99, 422]. The use of a circular mapping catheter also provides a very clear endpoint of complete isolation, which is achieved by wide encircling lesions around the PVs from within the atrium itself.

The use of pre-acquired three-dimensional imaging facilitates the understanding of the individual anatomy (which may vary substantially; Fig. 29.48; 29.9), however careful registration is necessary to implement the three-dimensional volume of the LA correctly, so that the operator is not misled by an wrongly positioned image. Accuracy of image fusion techniques vary and are reported to reach about 2mm, depending on the registration method [453–455].

While cardiac magnetic resonance (CMR) imaging does not add any radiation exposure to the patient, three-dimensional imaging by cardiac computer tomography (CT) exposes the patient to substantial radiation (about 20mSv) [458, 459]. Image quality, with properly applied imaging sequences are definitely comparable, therefore CMR is preferred whenever possible [460]. Intraprocedural rotational angiography of the LA or true three-dimensional intracardiac ultrasound may supplement these imaging modalities.

Despite significant improvements, catheter ablation of AF is still associated with major complications ( Fig. 29.49; Table 29.14) [99, 329].