OUP user menu

Early and comprehensive management of atrial fibrillation: Proceedings from the 2nd AFNET/EHRA consensus conference on atrial fibrillation entitled ‘research perspectives in atrial fibrillation’

Paulus Kirchhof, Jeroen Bax, Carina Blomstrom-Lundquist, Hugh Calkins, A. John Camm, Ricardo Cappato, Francisco Cosio, Harry Crijns, Hans-Christian Diener, Andreas Goette, Carsten W. Israel, Karl-Heinz Kuck, Gregory Y.H. Lip, Stanley Nattel, Richard L. Page, Ursula Ravens, Ulrich Schotten, Gerhard Steinbeck, Panos Vardas, Albert Waldo, Karl Wegscheider, Stephan Willems, Günter Breithardt
DOI: http://dx.doi.org/10.1093/europace/eup124 860-885 First published online: 16 June 2009


Atrial fibrillation (AF) is already an endemic disease, and its prevalence is soaring, due to both an increasing incidence of the arrhythmia and an age-related increase in its prevalence. Indeed, 1–2% of the population suffer from AF at present, and the number of affected individuals is expected to double or triple within the next two to three decades both in Europe and in the USA.14 Although epidemiological data for other parts of the world are less robust, a similar increase in AF in the community can be assumed in other countries.

Atrial fibrillation causes marked morbidity and mortality on a population basis. Epidemiological observations suggest that AF is still associated with a doubling of mortality, even after adjustment for confounders.2,5 This observation from the last millennium appears to continue into current randomized trials in AF patients. Also, AF is the single most important risk factor for ischaemic stroke. Furthermore, strokes associated with AF result more often in death or permanent disability than strokes that occur as a result of other aetiologies.69 The presence of AF is also associated with a marked reduction in everyday functioning and quality of life.1013

The harm associated with AF and the perceived detrimental effects of the arrhythmia on general health contrast with the outcome of six trials that compared a ‘rate control’ therapy strategy, aiming at accepting AF and controlling the ventricular rate, with an antiarrrhythmic drug-based ‘rhythm control’ therapy strategy, aiming at maintenance of the ‘natural’ sinus rhythm. Apart from a slight improvement in 6 min walk test in a small trial14 and post hoc analyses,15 the outcome of patients randomized to rhythm control therapy was not better than patients randomized to rate control therapy,14,1620 and antiarrhythmic drug therapy was even associated with adverse outcome in the AFFIRM trial.21 One possible explanation of these results is that AF per se is less of a problem than we thought. However, taken in the context of epidemiological observations and the well-established complications of AF, these trials suggest that the lack of benefit from sinus rhythm maintenance reflects the inadequacy of presently available methods to manage AF and prevent its consequences. The recently published effect of dronedarone on cardiovascular hospitalizations, cardiovascular death, and possibly also other relevant outcomes in AF patients22 is a first controlled suggestion that maintenance of sinus rhythm, if done well, can prevent AF-related complications. This lack of therapeutic efficacy is paralleled by a lack of understanding of unknown, but potentially relevant aspects of the pathophysiology of AF. The medical need for improving the therapy of AF is reflected by the large number of clinical trials examining AF-suppressing interventions that are currently registered (www.controlled-trials.com: 86 trials; www.clinicaltrials.gov: 385 trials; accessed on 11 January 2009).

With this in mind, the German Atrial Fibrillation competence NETwork (AFNET, www.kompetenznetz-vorhofflimmern.de) and the European Heart Rhythm Association (EHRA, http://www.escardio.org/communities/EHRA/Pages/welcome.aspx) recently defined relevant outcome variables in AF trials during the first joint AFNET/EHRA consensus conference.23,24 Thereafter, we organized the 2nd AFNET/EHRA consensus conference on ‘research perspectives in atrial fibrillation’. Over 70 expert participants from academia and industry convened for a 2-day meeting in the European Heart House in Sophia Antipolis, France, on 27 and 28 October 2008. A list of the attendees, many of whom contributed relevant thoughts expressed in this document, is found in Appendix 1. This paper summarizes the conclusions of the conference.

Areas in need of research

In preparing for the conference, seven relevant research areas and unanswered questions were identified. During the conference, these were organized in three research areas (Table 1). This paper aims to define the most important and clinically relevant gaps in scientific knowledge regarding AF. The research areas have been judged with respect to their relevance for the main clinical outcomes associated with AF (Table 2). We hope that this publication will stimulate research that will contribute to a better understanding of AF mechanisms, help to improve the management of patients with AF, and eventually contribute to reducing the burden of AF in the community.

View this table:
Table 1

Research questions identified during the preparation of the conference (left column) and convergence of these questions into three research areas (right column)

Unanswered research questionsConvergence of research questions into three research areas to improve management of AF patients
1. Indications for and duration of anticoagulation1. Understanding the mechanisms of AF (partially covers questions 2–4 and 7)
2. Timing and duration of antiarrhythmic drug treatment2. Improving rhythm control monitoring and management (partially covers questions 2–4, 6, and 7)
3. Timing and extent of AF ablation procedures3. Comprehensive cardiovascular risk management in AF patients (partially covers questions 1–5 and 7)
4. Relevance and intensity of ECG monitoring in clinical practice
5. Relevance of risk factors for AF progression
6. Novel therapeutic goals
7. What causes the first AF episode and how can pathophysiological parameters guide treatment?
View this table:
Table 2

Clinical variables advocated for outcomes atrial fibrillation trials, modified from Kirchhof et al.23

2Stroke and cerebral bleeds
3Quality of life
5LV function
6Health economics
7Additional important outcome variables: cardiovascular complications, e.g. acute coronary syndrome or decompensated heart failure

1. Understanding the mechanisms of atrial fibrillation

Relevance of pathophysiological understanding for better atrial fibrillation treatment

A simple paradigm has successfully guided interventional treatment of most supraventricular arrhythmias: once the mechanisms of an arrhythmia are thoroughly understood, clinical development of a successful, anatomically well defined, mechanism-based treatment is within reach.25 Such treatments successfully terminate and often ‘cure’ most supraventricular arrhythmias and a variety of ventricular arrhythmias at present. Along those lines, it is tempting to speculate that we would be able to treat AF better if we understood its causes well enough.26

New knowledge about focal triggers in the pulmonary veins; electrical, and structural ‘remodelling’; abnormalities in intracellular Ca2+ handling; and genetic causes of AF are all examples of the enormous progress made in identifying mechanisms of AF that may provide a basis for new treatment approaches. Nevertheless, tools with which to identify the mechanism of the arrhythmia and predictable response to therapy in individual patients are still largely lacking. A better understanding of the mechanisms of AF in an individual patient may allow novel therapeutic approaches or ‘targeted’ therapy. Such research will probably require an understanding of the pathophysiological changes that precede the first episode of AF, as well as a continued effort to understand perpetuation of the arrhythmia.

Different pathophysiological mechanisms can cause and maintain atrial fibrillation

Atrial fibrillation is defined by a characteristic ECG pattern, which may relate to different electrical abnormalities in the atria in different patients. A better understanding of these factors in an individual patient may help to guide therapy. We therefore propose to define different forms of AF that might require different types of treatment (Table 3). The following mechanisms of AF have been described:

View this table:
Table 3

Suggested classification of different types of atrial fibrillation

Type of AFClinical examplesClinical or ECG finding that may identify this type of AF
Focal AF• ‘Lone AF’Frequent atrial ectopy (spontaneous, multiple P-wave morphologies usually present)
• AF mainly caused by focal triggers, often in the pulmonary veins
• Typical ablation patients
Inheritable AF• AF in patients with inherited cardiomyopathiesECG signs of cardiomyopathies (QT interval changes, abnormal hypertrophy, etc.)
• Requires inheritability
• The cause of AF may vary depending on the genetic defect (Table 4)
• AF can be the only clinical sign of cardiomyopathy
‘Multiple wavelet AF’
 AF due to shortening of AERP• Tachycardia-induced AFInvasive assessment or fast Fourier transformation of signal-averaged ECG or echocardiographic analysis of atrial rate during AF
• Early AF recurrences after cardioversion
 AF due to abnormal conduction patternsMarkedly hypertensive patients without marked enlargement of the atriaProlonged P-wave duration on ECG or signal-averaged ECG or by other techniques
 AF due to enlargement and ‘electrical fragmentation’ of the atria• Severe mitral valve disease and enlarged atria; multiple wavelets may be due to increased atrial size more than other factorsIncreased left atrial size, P sinistroatriale (double-notched P-wave)
AF due to one or a few re-entrant driver or drivers‘Focal’ AF or re-entrant ‘left atrial’ flutters after pulmonary vein isolation‘Coarse’ AF, AF with distinguishable F-waves in the ECG. This AF type probably requires invasive diagnosis at present
AF in the elderlyUltrastructural and structural alterations are a prominent causeAdvanced age
  • The disparate AF pathophysiologies suggest that the different forms of AF may require different treatment (‘graded therapy’). In patients, several of these types of AF may be present. AF, atrial fibrillation; AERP, atrial effective refractory period; ECG, electrocardiogram.

Focal triggers of atrial fibrillation

Focal activity in the pulmonary veins initiates AF in many patients with paroxysmal, often ‘lone’ AF.27,28 Although some experimental studies have attempted to identify arrhythmogenic mechanisms in the pulmonary veins and adjacent myocardium,2932 others have failed to document clear mechanisms and the prominent role of the pulmonary veins for the initiation of ‘lone’ and paroxysmal AF remains poorly understood in experimental models. Examples of seminal questions that remain unanswered are the following.

  • –If muscle sleeves are present in the pulmonary veins in everyone, why do some develop AF and others do not?

  • –Does a ‘natural’ functional electrical block between the pulmonary veins and the left atrial myocardium exist, and would this protect against AF?

  • –Why does ‘focal’ AF develop at age 30 in one patient and age 70 in another?

  • –Why do periods of frequent AF paroxysms alternate in unpredictable patterns with periods of sinus rhythm in most patients?23

  • –Why do some patients have numerous paroxysms of AF without ever-developing persistent forms, while other progress to sustained forms of AF within a short time?

Understanding the mechanisms that cause pulmonary vein firing—both in atrial health and disease—is of utmost relevance, as catheter-based electrical isolation of the pulmonary veins appears very effective to maintain sinus rhythm in paroxysmal AF patients27,28,33,34 but less so in persistent AF patients or in patients with marked structural heart disease. Ablation procedures to date remain time-consuming and costly; and there is a relevant, albeit small, risk of complications.28,35,36 A better understanding of the mechanisms initiating AF will help to develop ablation strategies beyond isolation of the pulmonary veins (see below) as well as other therapeutic modalities.

Electrical remodelling in atrial fibrillation

Atrial remodelling refers to the changes in atrial properties and function that promote AF.26 Rapid atrial activation provokes both a shortening of the atrial action potential and refractory period, as well as an impaired rate adaptation with reduced wave length, thereby enhancing the risk for functional re-entry.37,38 This is a cellular survival mechanism that prevents cell death due to intracellular calcium overload. Although this mechanism is probably beneficial at the cellular level, APD shortening abbreviates atrial wave length and facilitates re-entry. Furthermore, APD shortening is further facilitated by up-regulation of the inward rectifier current IK1 which also hyperpolarizes atrial myocytes, thereby removing voltage-dependent INa inactivation, and potentially accelerating re-entrant rotors and stabilizing AF.37 ‘Reversal’ of this electrical remodelling is a main effect of ion-channel blocking drugs. Such ‘antiarrhythmic drugs’ usually prolong the atrial refractory period and can terminate persistent AF,3941 facilitate cardioversion,42,43 or prevent recurrent AF after cardioversion.44 Although these clinical observations suggest that ‘reversing’ electrical remodelling by prolonging the atrial action potential and refractory period can prevent AF in part, conditions that are known to prolong the atrial refractory period such as the long QT syndromes or systolic left ventricular (LV) dysfunction are clearly associated with AF.

Altered intracellular calcium handling in atrial fibrillation

Cumulative evidence points to a role of abnormal intracellular Ca2+ handling, not only for electrical remodelling, but also for triggered activity, focal drivers, and multiple re-entrant mechanisms of AF that contribute to its initiation and perpetuation.45 High atrial rate enhances overall Ca2+ influx. On a short-term scale, cellular Ca2+ load is counterbalanced by enhanced inactivation of voltage-dependent L-type Ca2+ channels. Over time, however, adaptive mechanisms alter the functional state of multiple Ca2+ handling proteins, and profoundly alter gene transcription, protein expression, and protein regulation, e.g. through phosphorylation. For instance, diastolic Ca2+ leak from hyperphosphorylated ryanodine receptors may play a prominent role in maintaining triggered activity by threshold-reaching delayed afterdepolarization due to activation of Na+/Ca2+ exchanger. Differential activation of protein kinases, e.g. protein kinase C or calmodulin-activated kinase II, and altered function of protein phosphatases, e.g. PP1 or PP2, can greatly extend the effects of altered calcium handling by regulating an avalanche of cardiac proteins in a calcium-dependent fashion.46,47 We propose that a sound understanding of the molecular mechanisms of abnormal Ca2+ handling in AF may open new strategies for treatment. Another important challenge will be to provide evidence that triggered activity and/or abnormal Ca handling actually initiate or perpetuate AF.

Structural changes

Increased atrial pressure and volume48 related to structural heart disease,49 arterial hypertension, or ageing50,51 and certain genetic alterations46,52 result in a steady process of ultrastructural changes in the heart. This process takes place both in the ventricles and the atria, but may be more pronounced in the atria.53 It occurs independently of AF,50,51 but is also accelerated by the arrhythmia54 and may predispose to AF.55 Activation of fibroblasts, enhanced collagen deposition, and fibrosis are hallmarks of the structural remodelling process, resulting in electrical dissociation between muscle bundles and local conduction heterogeneities that facilitate the initiation and perpetuation of AF. One can assume that these processes contribute to the increased incidence of AF with increasing age. This electro-anatomical substrate is characterized by electrical isolation of individual myocytes, increased non-uniform anisotropy of activation, local conduction heterogeneities, and even macroscopic slowing of conduction. Electrical dissociation between muscle bundles promotes the occurrence of more complex conduction patterns of fibrillation waves, facilitating persistence of AF.5659 In a model of hypertension, inducible AF develops due to conduction heterogeneities without electrical remodelling.55,60 Understanding this ultrastructural substrate of AF may help to suggest biomarkers to identify AF-prone patients as well as new molecules for upstream therapy and targeted ion channel blocker therapy.

Atrial fibrillation in inherited cardiomyopathies

One way to identify new factors that can cause AF are investigations in patients with inherited cardiomyopathies.61 Many patients with inherited cardiomyopathies develop AF, and genetic abnormalities have been identified in patient cohorts with AF. The genetic abnormalities in these patients range from defects in ion channels (mainly sodium and potassium channels) that shorten or lengthen the atrial action potential, altered formation of myocardium through altered regulatory proteins (PRKAG and atrial natriuretic peptide mutations), or even polymorphisms in genes involved in early cardiac development (PITX2, Table 4). Fortunately, models for many of these diseases are available. Factors that precipitate AF in these conditions warrant further research. In addition to understanding AF in patients with inherited cardiomyopathies, subclinical cardiomyopathic changes may contribute to a relevant portion of unexplained (‘lone’) AF.

View this table:
Table 4

Genetic abnormalities associated with atrial fibrillation identified in patients with inherited cardiomyopathies carrying a high risk for atrial fibrillation (upper part) and genetic defects found in association with atrial fibrillation (lower part)

Cardiac abnormalityGenetic defectAF prevalence (estimate)
Inherited cardiomyopathies61 associated with AF
Brugada syndromeLoss-of-function SCN5A mutations (10–15% of patients)10–20%249
Long QT syndromeLate gain-of-function SCN5A and loss-of-function K channel mutations, among others5–10%63,250–252
Short QT syndromeGain-of-function K channel mutations70%253,254
Catecholaminergic VTLoss-of-function ryanodine receptor mutationRare families255
Hypertrophic cardiomyopathySarcomeric proteins5–15%224–256
Wolff–Parkinson–White syndrome and abnormal left ventricular hypertrophyPRKAG mutationsRare familial forms257,258
Holt–Oram syndrome with AFTBX5 mutations (regulatory gene)Family clusters259
Gene defects associated with AF
Type of AFGenetic defectAssociated with AF in
‘Lone’ AFLoss-of-function SCN5A mutations5% of ‘lone’ AF patients70,260
AF and heart failureSCN5A mutationRare forms of AF261
‘Lone’ AFGain-of-function K channel mutationsRare families with AF and short QT interval262
‘Lone’ AFLoss-of-function K channel polymorphismsRare families, associated with long QT syndrome263
‘Lone’ AFLoss-of-function KV1.5 mutation (IKUr)Rare patients263
‘Lone’ AFSomatic connexin 40 mutations‘Lone AF’ patients264 (requires atrial tissue for testing)
‘Lone’ AFFrameshift (loss-of-function) ANP mutationLarge families265
All types of AFPITX2 polymorphism (involved in pulmonary and cardiac development)Populations in Iceland266 and elsewhere

Atrial fibrillation requires a translational research approach

The available data already suggest that multiple initiating and perpetuating mechanisms interact to generate AF in different scenarios. A potential interplay of these mechanisms is depicted in Figure 1. This complex nature of the initiation and perpetuation of AF urges for translational and interdisciplinary research to allow progress in understanding AF. Such research should combine molecular, cellular, organ, and in vivo experiments and measurements in patients to translate research concepts into clinically applicable diagnostic tools and therapeutic options. Translational research can be a win–win scenario: clinicians feel the strong need to interpret their observations in daily practice based on current pathophysiological knowledge, and, vice versa, basic research would appreciate feedback on the potential clinical significance of their mechanistic findings. Another potentially relevant component of translational research in AF may be the testing of potentially relevant pathophysiological changes ‘in silico’, i.e. by integrated functional computer modelling.

Figure 1

Interdependence of mechanisms that contribute to the initiation and maintenance of atrial fibrillation (AF). Each circle represents a relevant factor that may initiate or perpetuate AF. The pie chart within each circle gives educated guesses as to how often this pathophysiological mechanism will be due to AF itself (black pie piece), genetic predispositions (light blue), a response of the atria to stressors such as hypertension, diabetes, or valvular heart disease (grey), and ageing (light green). It has to be emphasized that these proportions are educated guesses to illustrate the interdependence of different causes, and not based on real data. Some of the main pathophysiological mechanisms also correspond to a ‘type’ of AF (compare Table 3). In an individual patient (but also in a specific experimental model), AF will be due to a ‘blend’ of these different factors as indicated by the blended overlap between the circles. There may be additional mechanisms, and their interaction will be different in different patients.

Improving non-invasive diagnostic tools to assess substrates for atrial fibrillation

Epicardial or endocardial activation mapping would allow to assess mechanisms initiating and maintaining AF in patients. Apart from relatively complicated and costly studies in patients undergoing AF ablation or open-heart surgery, this is usually not feasible in patients. To allow a better ‘translation’ of experimentally delineated AF mechanisms to patients, we would like to suggest clinical diagnostic tools that may help to differentiate the various types of AF (Table 3). The clinical use of these proposed diagnostic tools is not yet validated, neither for diagnosis of a specific ‘AF type’ nor for therapeutic guidance. Furthermore, given the fact that the pathophysiological factors described above can usually only be validated by analysing the beating heart invasively or cardiac tissue ex vivo, a validation of the techniques mentioned below will be difficult.

The group felt, however, that a better characterization of ‘treatable’ causes of AF may help to individualize therapy in AF patients, e.g. by using these tools. The methodological limitations notwithstanding, validation of these proposed diagnostic techniques is needed.

The ECG is the main tool used to diagnose AF. Detailed analysis of information available in the ECG signal may help to non-invasively diagnose triggered activity and conduction slowing in the atria.

Focal events may be associated with frequent atrial ectopy, atrial bigeminy, or a ‘P on T phenomenon’ on the surface ECG.

Electrical remodelling and prolongation of the atrial action potential are measurable by recording monophasic action potentials invasively in patients,29,6264 but difficult to assess non-invasively at the atrial level. The QT interval may provide a rough ventricular surrogate for ubiquitously expressed (e.g. genetically determined) abnormalities. Unfortunately, transient or localized action potential changes in the atria are not well measurable by ECG. Signal-averaged ECG or body surface recordings, e.g. during induced AV block, may allow measurement of atrial repolarization non-invasively.

Conduction abnormalities that predispose to AF perpetuation underlie the complexity of the conduction pattern of fibrillation waves65 and may also be reflected by slow atrial activation during normal rhythms.48,66 Prolonged duration of the P wave, ideally measured in the signal-averaged ECG, could be a reasonable marker for atrial conduction disturbances,67 particularly in the left atrium, and has been related to the presence of paroxysmal AF in prior studies.68,69 Prolonged P wave duration may also identify patients with genetically determined conduction abnormalities associated with AF.70 Prolongation of the total atrial activation time, reflected by the electro-echocardiographic interval between the onset of the P wave and the local Doppler tissue imaging signal in the left lateral wall of the atrium, may precede the first diagnosis of AF.71 These initial observations require confirmation in clinical trials. For example, the prolonged P wave duration or atrial activation time as measured by Doppler is associated with new-onset AF, complex activation patterns during AF, or longer AF episodes. If such studies confirm the diagnostic value of such ECG-based parameters, they may become useful to select patients for upstream therapy.

Atrial fibrillation cycle length closely correlates with the frequency of f-waves measured by frequency analysis.72 In the future, body surface mapping may allow analysis of single atrial fibrillation waves (Yoram Rudy, personal communication, 2008). Electrical dissociation during AF may, after further refinement of such techniques, also be accurately identifiable by the Doppler-based measurement of active atrial tissue motion to assess local AF cycle lengths.73

Cardiac imaging is already an important tool in the evaluation of patients with AF. Enlarged left atrial diameters are associated with AF recurrence after electrical cardioversion74,75 and may be useful to estimate the efficacy of antiarrhythmic drugs, although the value of left atrial size for the prediction of AF recurrences is less evident in other trials. Real-time three-dimensional (3D) echocardiography and magnetic resonance imaging (MRI) may allow accurate measurement of left atrial volume, both in systole and in diastole,76 or to guide ablation.77,78 In addition, left atrial enlargement and/or scarring may identify patients with an ‘ultrastructural substrate’ for AF.

Graded or targeted therapy for different substrates of atrial fibrillation?

One possible reason for the limited success of pharmacological therapy of AF may be the complexity of underlying mechanisms and the different types of electrophysiological changes in individual patients (‘AF is a symptom and not a disease’). This concept suggests that ‘multimodal’ therapy of the different factors that contribute to AF, e.g. focal triggers, electrical remodelling, altered calcium homeostasis, and formation of an ultrastructural substrate, may be more effective than current approaches that often rely on a single therapeutic modality. We, therefore, suggest evaluating of ‘multimodal’ and ‘graded’ therapies that target AF-causing factors: in the absence of a significant substrate of AF and without detectable triggered activity, upstream therapy might be the only recommendation for preventing new-onset AF. Marked triggered activity without pronounced ultrastructural changes may suggest that sodium channel blockers and catheter ablation are appropriate. A combination of both factors may require multimodal treatment, while in the presence of a very severe substrate, rhythm control may be futile unless other treatable causes of AF are identified. Table 3 gives an overview of the suggested invasive and non-invasive markers of electrophysiological changes in patients with, or prone to, AF. Whether these (or other) diagnostic markers will enable a reliable determination and/or grading of different types of AF, and whether these can be used to ‘tailor’ therapy, needs to be evaluated.

Basis for research:

  • Among the known factors that contribute to AF are focal triggers, AF-induced electrical remodelling, localized re-entrant drivers, and ultrastructural ‘remodelling’ in the atria, altered intracellular Ca2+ handling, as well as an inherited predisposition.

  • It is likely that a variable combination of the above-mentioned factors causes AF in a given patient.

  • In many patients, several of the above-mentioned mechanisms are active before AF occurs.

  • Many AF-promoting processes are caused by atrial damage unrelated to AF per se and are then aggravated by AF itself.

  • Successful therapy of AF likely requires identification of all AF-causing factors and ‘graded’ therapy of all relevant pathological processes.

Research perspectives. The following research perspectives appear relevant in the near future:

  • What causes the first episode of AF?

  • What causes progression of AF in the majority, but not in all patients?

  • How do genetic factors and interactions between the heart and the autonomic nervous system predispose to AF?

  • Can subclinical dysfunction of the structures involved in genetic alterations, subtle ultrastructural changes, and/or minimal alterations in Ca2+ handling and electrical atrial function contribute to the initiation of AF?

  • Can an early and ‘graded’ therapy prevent AF more effectively in specific patients if clinical tools were available to assess the different factors predisposing to AF (see also next section)?

2. Improving rhythm control management

2a. Rhythm control monitoring

Does atrial fibrillation duration relate to outcomes?

At present, any arrhythmia that has the ECG criteria of AF and lasts for 30 s or longer is considered as AF.4,23,24 This clinical definition is also used in clinical trials, for example, to monitor success of rhythm control interventions.24,28

Most available outcome studies used short-term ECG recordings and related outcomes such as stroke, death, or heart failure to diagnosis of AF in single short-term ECG recordings.24,28 The available data indicate that paroxysmal AF carries the same stroke risk as persistent or permanent AF,24,79,80 albeit within the context of a limited ability to measure AF burden (Figure 2). Hence, the definition used for AF is very reasonable for clinical practice and as a starting point for further research (Table 5). Given the fact that AF episodes are often asymptomatic,23,24,81,82 it is likely that patients with short episodes of AF on standard ECG will also experience undiagnosed longer episodes of AF.

Figure 2

Efficacy of detecting paroxysmal atrial fibrillation (AF) and of assessing AF burden using a standard 24 h Holter ECG, two 7 day Holter ECGs, telemetric short-term ECG, and continuous (e.g. implantable) ECG monitoring devices in a 1-year period. The black bar on the bottom shows the biological sequence of periods of AF (black) and periods of sinus rhythm (grey). Monitored times are shown in red in each line. Longer ECG monitoring intervals will result in higher AF detection rates. pAF: paroxysmal AF, no AF: no atrial fibrillation.

View this table:
Table 5

Relevance of ECG monitoring in different clinical settings

Populations at risk for cardiovascular eventsPotential indication for ECG screening for AF
Patients with palpitationsDemonstration of AF, differential diagnosis of other arrhythmias, relation of symptoms and arrhythmia
Patients with syncopeEvidence for arrhythmogenic syncope (e.g. tachycardia-bradycardia syndrome)
Patients on rhythm control therapyRelation of symptoms and recurrent AF; assessment of AF recurrence (in trials); pro-arrhythmia monitoring
Patients on rate control therapyMonitoring or rate control in daily life and during standardized exercise tests

The clinical relevance of AF of 30 s duration (as opposed, for example, to 60 s or 5 min) has never been systematically studied and was chosen arbitrarily based on available recording technology and in concert with the definition of sustained ventricular tachyarrhythmias.82 Modern, long-term, ECG monitoring tools, such as digital Holter ECG recorders, ECG garments, analysis algorithms of implantable pacemakers and defibrillators, or dedicated implantable ECG monitoring devices, increase the sensitivity of detecting very short episodes of AF.24,81 Whether short AF episodes detected by such technology have the same prognostic relevance as AF detected by standard techniques is not clear. Some retrospective, preliminary data suggest that AF episodes detected by long-term ECG monitoring may require longer durations to have prognostic relevance: atrial high-rate episodes of ≥5 min duration detected by pacemakers were associated with premature death in an MOST subanalysis.83 Similar information appears to emerge from more recent studies.8486 This may reflect the fact that very short AF paroxysms (a few minutes duration) carry a lower risk for symptoms and complications, but may also relate to the fact that longer episodes can be analysed more accurately by automated algorithms.

Further research is needed to better delineate the relevance of very short AF episodes and to relate quantitative measures of AF (total AF duration, number of episodes, or other parameters), i.e. AF burden, with clinical outcome in patients without any previous documentation or clinical history of AF with or without additional risk factors. Such research may be difficult, given the irregular and seemingly unpredictable distribution of AF recurrences.23,24

Screening for asymptomatic AF in high-risk populations may allow identification of patients at risk for AF-related complications such as stroke. Whether such screening would warrant therapy (e.g. antithrombotic therapy) requires prospective study. Such patients may also become candidates for early interventions to prevent recurrent AF to reduce or even eliminate the risk of AF-related complications.87 These are important research areas that need clinical exploration.

2b. Rhythm control management

Antiarrhythmic drug therapy

Antiarrhythmic drug therapy is an important part of any rhythm control strategy. The use of antiarrhythmic therapy could be improved by aiming at more mechanistic use of compounds (e.g. by ‘graded therapy’), safe and timely use of existing and novel agents, and potentially strategic combination therapy with other rhythm control interventions.

Novel antiarrhythmic drugs

Antiarrhythmic ion-channel blocking drugs can prevent AF recurrences, possibly through countering electrical remodelling and through the prevention of electrical triggers, although ion-channel blocking therapy is not sufficient to maintain sinus rhythm in all patients. Pro-arrhythmic events are, however, a rare but important adverse effect of ion-channel blocking drug therapy. Several novel antiarrhythmic agents are in late stages of clinical development and may become available in 2009 or 2010.22,8890 These novel agents should be monitored closely using prospective scientific registries for their efficacy and, most importantly, for their safety. The careful selection and classification of specific causes of AF will be of great importance, as some drugs may be efficient in some patients, and not in others. Other types of antiarrhythmic agents may become available based on novel targets such as connexins or new ion channels. Among the newer antiarrhythmic drugs that appear close to introduction into clinical medicine, dronedarone is worth a short note: this multiple ion-channel blocker is chemically related to amiodarone and prevented not only recurrent AF, but also, at least in a composite outcome, cardiovascular deaths and cardiovascular hospitalizations, in the recently published ATHENA trial.22

In whom and when can antiarrhythmic drug therapy be discontinued?

Currently, antiarrhythmic drug therapy is applied as long-term—conceptually life-long—therapy. Pathophysiological considerations such as the reversal of electrical remodelling after several weeks of sinus rhythm suggest the possibility that antiarrhythmic drug therapy can be confined to periods ‘at high risk for AF recurrence’, e.g. after cardioversion or post-operatively.91,92 Other existing concepts use drugs for immediate conversion of AF (e.g. pill-in-the-pocket91). This can markedly reduce exposure to antiarrhythmic drugs, possibly without altering efficacy,93 and may help to reduce drug-related adverse effects. In some patients, however, cessation of drug therapy may have pro-arrhythmic or other unwanted effects as shown with amiodarone.94 Antiarrhythmic drug discontinuation may also be reasonable in patients in whom concomitant conditions have been successfully treated (e.g. mitral valve repair, significant weight loss, long-term hypertension control) or in whom prominent triggers have been eliminated (e.g. after pulmonary vein isolation). Discontinuation of antiarrhythmic drug therapy would offer the benefit of decreased adverse effects, and the clinical potential to re-initiate the same therapy upon AF recurrence. Controlled trials on the effects of antiarrhythmic drug discontinuation could help to determine the clinical applicability of these concepts.

Ablation in the left atrium for atrial fibrillation

After only a decade since the first publication of catheter-based ablation of focal sources of AF in the pulmonary veins,27 ablative therapy with the aim of isolating the pulmonary veins has almost become a standard procedure. It appears highly effective in patients with paroxysmal AF without marked concomitant cardiac disease (70–85% sinus rhythm rates in most trials in paroxysmal AF patients) and relatively safe (3–5% overall complication rate; 1–2% severe events).28,35,36,95 Initially, it was expected that ablation would ‘cure’ AF; this notion has largely been abandoned as it becomes evident that there is a need for an individual approach to different types of AF. Nonetheless, pulmonary vein isolation appears the most effective intervention to maintain sinus rhythm in patients with paroxysmal AF in whom focal triggers of AF are an important driver of the arrhythmia.

Improving feasibility and safety of pulmonary vein isolation

Despite the effectiveness of pulmonary vein isolation,28 it is notable that so far no ablation approach or technology has been developed, which allows consistent complete and durable pulmonary vein isolation. Although current ablation technologies lead to acute PV isolation in nearly all patients,28 early and late recurrences of PV conduction are common.9698 Recovery of pulmonary vein conduction may occur in the first hour after ablation in many pulmonary veins.99,100 Of note, recovery of conduction through previously complete linear lesions is also common.101 An ongoing controlled trial in Germany (GAP-AF, NCT00293943, http://www.kompetenznetz-vorhofflimmern.de/mediziner/projekte/bereich_b/b4/index.php) currently investigates whether complete isolation of the pulmonary veins is necessary for the clinical outcome ‘maintenance of sinus rhythm’ when compared with incomplete circumferential PV ablation in patients with paroxysmal AF.

In parallel with improved efficacy and durability of pulmonary vein isolation, the safety of the procedure requires improvement35,36,102121 (Table 6). Such developments will be incremental, and an effect on rare complications is difficult to assess in a controlled trial. Safety concerns are among the main limitations to widespread clinical applicability of AF ablation. Improvements may stem from technical advances such as better catheter steerability, better controlled energy sources, improved energy delivery and tissue contact, and/or avoidance of critical structures such as the oesophagus. In addition, it may be helpful to visualize ablation lesions and left atrial structural changes during the procedure. The use and accuracy of MRI122127 or intracardiac echocardiography128130 for imaging atrial scar tissue is not yet known, and intraprocedural visualization of scars is not yet feasible. Enhancing durability and safety can probably in a first attempt best be accomplished by standardization and simplification of the ablation procedure.

View this table:
Table 6

Safety issues in atrial fibrillation ablation that may require attention for the reduction of procedure-related complications

ComplicationDescription and suggested precautions and interventions
Pulmonary vein stenosisIs rare when ablation is performed at the PV–LA antrum
Cardiac tamponadeDepends on training and technique of transseptal puncture in most cases. Can be managed in the cath lab. May improve with more standardized puncture tools and training for techniques
Stroke and TIAIs still relevant but has declined most likely related to intensive periprocedural monitored anticoagulation, preprocedural transoesophageal echo, flushing of sheaths, irrigated-tip catheters, and adequately low temperature and energy settings. Can probably be managed by immediate cerebral re-perfusion therapy. No controlled data available
Local char formationHas been reduced by adequate temperature and energy settings and probably by irrigated-tip catheters. May be monitored by intracardiac echocardiography
Local haematomaMay be reduced depending on the periprocedural antithrombotic regime, access site and size of sheaths placed, and on experience of the operator. Controlled trials should be done
Systemic bleedingIs likely to depend on periprocedural antithrombotic regime
Left atrial flutterAppears to be related to the exact placement of PV isolation lesions (and additional lesions)
Phrenic nerve injuryIs relatively rare and at times reversible; most often due to ablation in the right pulmonary veins
Oesophageal damage, atrio-oesophageal fistulaIs rare, difficult to avoid completely, and most likely associated with linear lesions more than pulmonary vein isolation. Temperature monitoring in the oesophagus and intraprocedural imaging (intracardiac echocardiography) have been suggested and debated. The oesophagus is a mobile structure

In this context, it is of utmost importance that all rhythm control trials and registries in AF patients systematically report safety outcomes including deaths, procedure- or treatment-related complications, and major cardiovascular events on an intention-to-treat and on-treatment basis. For AF ablation, registries with patient enrolment prior to the ablation procedure are needed to address true complication rates of AF ablation in ‘real life’, which may be higher than those reported in voluntary registries. Such scientific, prospective registries are in the interest of patients and health-care providers. Ideally, an impartial, for example, scientific organization would organize and steer such a registry that would be financed by the health-care providers and/or industry as a part of their quality control programs. It would be desirable to initiate such a registry at the European level.

Advancing ablation for atrial fibrillation beyond pulmonary vein isolation

Pulmonary vein isolation has almost become a standard procedure in patients with paroxysmal AF.4,28 Different additional ablation techniques, such as linear lesions,131134 ablation of fractionated electrograms,77,135 ablation of areas with short cycle lengths during AF,78 or targeted ablation of ganglionated plexus and fat pads,136138 large diameter encircling of the pulmonary veins to ablate parts of the posterior left atrium, disseminated ablation of the posterior left atrium,34,139141 or a combination of these techniques,142144 have been advocated by some groups to increase sinus rhythm maintenance rates in patients in whom ablation at the pulmonary veins alone is not sufficient to maintain sinus rhythm. Although small studies in experienced ablation centres suggest that such techniques may result in higher sinus rhythm rates in patients with AF due to structural remodelling,33,141,145,146 none is presently well-established,28 and no technique has been tested in large trials. There is an urgent need to teach such techniques to a wider community of interventional electrophysiologists and to evaluate the safety and efficacy of different approaches on a broader basis in controlled, multicentre settings.

Do we understand what we ablate in the left atrium?

Experimental studies suggest that focal drivers of AF or stable or unstable re-entry circuits may reside in different critical regions of the atria,75,147,148 including the pulmonary veins.149 In one animal model, targeted ablation at Bachmann's bundle prevented further induction of atrial fibrillation,150 whereas another animal study suggested that ablation in the inferior septal right atrium (Triangle of Koch) may prevent inducible AF.151 The original surgical MAZE procedure successfully eliminated AF putatively by preventing potential multiple re-entrant wavelets in patients without intentionally isolating pulmonary vein foci.152 For successful AF ablation, other patients require elimination of electrical activity within the ligament of Marshall, a structure with electrical connections to different parts of the left atrium as well as the parasympathetic system.153156 Even ‘standard’ antral pulmonary vein isolation may result in ablation of parasympathetic ganglia.157 Ablation of continuous fractionated electrograms may guide ablation to different regions in the posterior left atrium, even when analysis is automated.158,159 Extrapolating from these examples, it is possible that the lesions used to isolate the pulmonary veins might also be eliminating regions critical to other AF-maintaining mechanisms. Identification of individual mechanisms and critical regions for AF may therefore be important for improvement of AF ablation as currently exercised in patients who present with left atrial flutters or focal left atrial tachycardias after pulmonary vein isolation for AF.107,113,160,161 The group proposes large, multicentre trials to determine mechanisms of recurrent AF after pulmonary vein isolation and to compare different ablation strategies in patients with recurrent AF after pulmonary vein isolation.

‘Upstream therapy’ interventions for primary prevention of atrial fibrillation

From a mechanistic perspective, primary prevention of AF should target all treatable factors that initiate the arrhythmia. ‘Substrate modification’ by inhibition of the renin-angiotensin system can prevent AF in patients with structural heart disease,60,162,163 including patients at risk of AF with preserved systolic LV function,164 and even (secondary prevention) in patients with AF undergoing cardioversion.165 On the other hand, in the GISSI-AF trial,166 an angiotensin receptor blocker (valsartan) did not prevent recurrent AF in patients with controlled hypertension in the absence of overt structural heart disease.166 Other trials like Active-I167 or ANTIPAF168 will also address the effectiveness of angiotensin receptor blockers for prevention of recurrent ‘lone’ AF. Similarly, the use of anti-inflammatory agents, statins, and N-3 polyunsaturated fatty acids has been proposed, but not sufficiently evaluated with the possible exception of AF prevention after cardiac surgery. Potential reversibility of fibrosis by aldosterone antagonists might be another added benefit that has not been explored beyond experimental data.169 In light of the different factors that may cause AF in an individual, there are probably specific patient groups in whom ‘upstream therapy’ can prevent AF, whereas the other patient groups (e.g. patients with predominant trigger mechanisms) may not benefit from such treatments. Many of these questions may be evaluated by detailed ECG monitoring of patients enrolled in cardiovascular prevention trials (see below).

Early initiation of rhythm control therapy (after the first episode)

Current practice guidelines do not recommend antiarrhythmic drug treatment at the time of first diagnosis of AF,4,170 but only in patients with recurrent AF. This concept has the advantage of delaying potentially harmful treatments such as antiarrhythmic drugs until a second episode occurs. This recommendation is, however, not in line with the observation that AF is a chronically progressing arrhythmia in the vast majority of patients2,5,23,171 and that only a highly selected subgroup of patients may remain with short, infrequent episodes of paroxysmal AF over several decades.172 Given the fact that many of the processes that finally predispose to AF develop prior to the first AF episode, and even more so prior to the first diagnosis of AF, such delayed treatment may in fact lose time for effective, early prevention of AF recurrences.87 Atrial fibrillation recurrence rates in the rhythm control arm of previous ‘rate vs. rhythm’ studies were high (20–30% on short-term ECG monitoring); while sinus rhythm rates were equally high (20–30%) in the ‘rate control’ arms of these studies. This suggests that rhythm control interventions were not sufficiently effective.21

Our group feels that an early initiation of rhythm control therapy using drugs and/or ablation, i.e. in patients with a first documented AF episode, should be tested in controlled trials. In fact, current understanding of AF pathophysiology would suggest that such ‘early treatment’ may be more successful and potentially less harmful than current practice. Very long-term follow-up would be necessary for such studies to detect a delay in AF progression, including the long-term effects on LA mechanical function and the incidence of very late recurrence.

Combining interventional and pharmacological treatments to improve rhythm control therapy

Catheter ablation of AF has so far been evaluated in its effects on symptoms, the main indication for rhythm control therapy at present, and for maintenance of sinus rhythm.33,173 Small controlled trials in a single or a few centres found a positive effect on LV systolic function after catheter ablation for AF.141,145,174 This is contrasted by the incomplete recovery of left atrial function after AF ablation procedures involving more than pulmonary vein isolation.175,176 One retrospective single-centre analysis suggested even a prevention of death and cardiovascular events associated with catheter ablation.177 It is time to perform large, prospective, multicentre trials of a comprehensive rhythm control strategy compared with standard care for AF patients. Several trials mainly using catheter ablation as the study intervention in different settings are either in their planning phase or already under way, e.g. CABANA (NCT00578617), AMICA (NCT00652522), RAAFT (NCT00392054), or CASTLE-AF (NCT00643188), among others. Such trials should apply long-term follow-up for relevant outcomes for AF (see Table 2 and Kirchhof et al.23,24).

Given the fact that AF ablation will not target all mechanisms that initiate or maintain AF or prevent its complications, and that AF is initiated and maintained by multiple mechanisms, a ‘multimodal’ antiarrhythmic therapy strategy appears reasonable that makes adequate use of catheter ablation, antiarrhythmic drugs, and upstream therapy, all of which may have synergistic benefits for patients. Such a strategy could be particularly relevant in patients in whom an intervention on pathogenetic factors of AF is successful (mitral valve repair, significant weight loss, long-term hypertension control) or in patients with recently diagnosed (‘early’) AF. A successful rhythm control therapy strategy will most likely require adequate management of concomitant conditions such as vascular, myocardial, and valvular heart disease, thyroid dysfunction, diabetes mellitus, and heart failure.

Starting points for research:

  • Any arrhythmia that fulfils conventional criteria of AF on an ECG and lasts >30 s is defined as AF, which is reasonable when applied to standard ECG monitoring tools (short-term ECG), as it carries prognostic information in large outcome studies and surveys.

  • Improvement of symptoms and quality of life is at present the only accepted indication for maintenance of sinus rhythm in AF patients.

  • Catheter-based isolation of the pulmonary veins is an effective technique to prevent recurrent AF in patients with paroxysmal AF. Pulmonary vein isolation should be attempted during the first ablation procedure.

  • Solid scientific data to perform more than pulmonary vein isolation routinely during the first ablation procedure is lacking. The optimal type of additional ablation is also not known.

  • That maintenance of sinus rhythm can positively affect cardiovascular events in AF patients is supported by the recent report of the ATHENA trial and by retrospective and post hoc analyses. This is also a persistent clinical perception despite the negative outcome of six trials using antiarrhythmic drugs to maintain sinus rhythm (PIAF, AFFIRM, RACE, STAF, HOT-CAFÉ, AF-CHF).

Research perspectives:

  • What are the distribution, duration, and frequency of episodes and the patterns of recurrence in AF patients? What is the impact of AF patterns on AF progression and outcome when standardized long-term ECG monitoring is applied?

  • What is the minimal AF duration that has a prognostic impact using modern, long-term, ECG monitoring tools (pacemakers, ECG garments, or implanted devices)?

  • Does ‘AF burden’, measured as the number or duration of AF episodes, relate to complications of AF?

  • Can antiarrhythmic drug therapy be improved by developing safer and more effective antiarrhythmic agents and drug regimens that encompass defined therapy durations?

  • Can AF ablation targeting the pulmonary veins be made safer and more standardized so that a widespread use of this intervention is possible in patients with paroxysmal AF? This is one of the most important tasks at present and needs to be evaluated in large, prospective trials.

  • What are the mechanisms of recurrent AF after pulmonary vein isolation? Which therapy is adequate to treat recurrent AF after ablation?

  • Do extensive ablation strategies involving either linear lesions, ablation of continuous fractionated electrograms, ablation of ganglionated plexus, or wide antrum pulmonary vein isolation encompassing larger parts of the left atrium result in different outcomes than repeat pulmonary vein isolation alone? This should be addressed in controlled, multicentre trials.

  • Can cardiovascular outcomes be prevented by a combined treatment of concomitant diseases and targeted rhythm control interventions when compared with conventional care, or will such a therapy cause harm?

  • Could an early initiation of rhythm control therapy (drugs, ablation, and ‘upstream therapy’) result in a slower progression of AF and can, in the long term, AF-related complications be prevented by such a therapy?

3. Preventing atrial fibrillation-related complications

3a. Improving stroke prevention

How do we improve and define existing stroke risk stratification in atrial fibrillation?

AF is the most prevalent risk factor for cardioembolic stroke. Furthermore, AF-related strokes are more severe than other cerebrovascular events.178 Current stroke risk estimation schemes have been developed by analysing data from non-vitamin-K-antagonist arms of trials and one cohort study (Framingham). Specifically, the CHADS2 score combined the risk factors driving event rates in the SPAF-1 trial and the AF Investigators trials and studied these risk factors in the National Registry of Atrial Fibrillation cohort.179 The CHADS2 score and related stroke risk assessment schemes are valuable and accepted in identifying patients at high risk for stroke, i.e. patients carrying at least two of the risk factors, and for the identification of a group with a decidedly low risk, i.e. patients without any of the risk factors.4,180,181 Overall, however, the scheme only modestly predicts stroke and thrombo-embolism (c-statistics of ∼0.6182). This may be due to the fact that many patients (up to 60%) are classified as ‘intermediate risk’ for stroke following the available scores, e.g. ‘CHADS2 = 1’ patients. In these patients, the decision to anticoagulate is based on individual factors, not on robust data.

The existing classifications can be refined: heart failure can be defined as a clinical syndrome or as the presence of LV dysfunction, and hypertension may be adequately controlled or not well treated.183 Other risk factors may predict stroke and bleeding events in such patients: coronary artery disease (e.g. manifest as a prior myocardial infarction) or peripheral arterial disease associate with strokes. Cerebral small vessel disease as seen by MRI may be a predictor of both subsequent lacunar (non-cardioembolic) stroke and intracerebral bleeding risk.184 Large left atrial volume associates with cardiovascular events185 and with incident AF.186 Complex aortic plaques on transoesophageal echocardiography or serum markers of a prothrombotic or pro-inflammatory state187 may also relate to stroke risk in AF patients. Some of these parameters are investigated in ongoing trials, others may be evaluated in future trials. Other clinical parameters such as AF burden, various parameters of LA mechanical function, and echocardiographic parameters (e.g. smoke, LAA emptying velocities) should also be evaluated in trials.

Stroke events have been declining recently in AF populations.188 Indeed, medical treatment has improved in recent years, with more proactive hypertension and heart failure management, as well as improved secondary prevention of stroke, and early managed care of transitoric ischaemic attacks, among others.

Risk schemes need to recognize the impact of asymptomatic AF on thrombo-embolism.23 This indicates a need for trials investigating the value of ECG screening in populations at high cardiovascular risk to initiate antithrombotic therapy upon diagnosis of ‘silent’ AF. Also, better techniques for diagnosing asymptomatic AF in patients presenting with stroke and thrombo-embolism are needed.

Current information on stroke risk can be obtained from mergers of warfarin arms of ongoing trials, e.g. the Rocket AF, RE-LY, and other antithrombotic treatment trials in AF patients. Such analyses may provide a solid basis for a better definition of AF patients in need for oral anticoagulation who are currently described as ‘intermediate risk’. Such insights should be validated in data sets from registries/cohorts such as the AFNET registry or the EuroHeart Survey follow-up data set.

How to integrate bleeding risk assessment into thromboprophylaxis recommendations?

Any decision for an antithrombotic therapy must weigh the potential benefit (mainly prevention of thrombo-embolic complications) against the risk associated with therapy (most prominent, the risk of major bleeding events). In the vast majority of AF patients, the risk–benefit ratio is in favour of antithrombotic therapy. Modelling of clinical events suggests that persons taking warfarin must fall about 300 times per year for warfarin to not be the optimal therapy.189 This has recently been re-emphasized by the benefit of anticoagulation in the BAFTA trial in elderly patients with AF treated in a primary care setting, many of whom are judged as ‘high bleeding risk’ patients.190 Unfortunately, current bleeding risk prediction schema are largely based on ‘general’ anticoagulated patients.191 One exception is the HEMORR2HAGES score for AF patients.192 A recent systematic review concluded that no existing scheme could be recommended for widespread use in practice at present.193 Furthermore, the best available bleeding risk predictor, the outpatient bleeding risk indicator (OBRI), actually identifies the same patients who are also at risk for stroke: OBRI recognizes age ≥65 years, history of stroke (1–2 points), history of gastrointestinal (GI) bleeding, serious comorbidity (renal insufficiency, recent myocardial infarction, severe anaemia), and AF. With the exception of a history of GI bleeding, these factors overlap with stroke risk factors (Table 7). Furthermore, registry data suggest that bleeding risk may actually increase with age to a similar extent both in anticoagulated and in non-anticoagulated patients, the latter usually managed with aspirin.194

View this table:
Table 7

Clinical factors associated with an increased risk for stroke and an increased risk of severe bleeding in AF patients

Risk factors for thrombo-embolismBleeding risk factors
Previous stroke, transient ischaemic attack, or embolismCerebrovascular disease
Age ≥75 years
(Age 65–74 year)
Advanced age (>75 years)
Age >85 years
Heart failure or moderate–severe LV dysfunction on echocardiography (e.g. ejection fraction ≤40%)
(Vascular disease)
History of myocardial infarction or ischaemic heart disease
HypertensionUncontrolled hypertension
Diabetes mellitus?
(Female gender)?
Mitral stenosis
Prosthetic heart valve
[Renal dysfunction (stages III–V)][Renal dysfunction (stages III–V)]
History of bleeding
Concomitant use of other antithrombotic substances such as antiplatelet agents
  • Note that most factors pose patients at risk for both types of events. In AF patients in general, thrombo-embolic events (strokes) are approximately one magnitude more likely than severe bleeds. Modified from Hughes and Lip.267 Less validated factors given in parentheses. ‘Vascular disease’ refers to myocardial infarction, complex aortic plaque, carotid disease, peripheral artery disease, and similar manifestations of arteriosclerosis.

Unstable INR values are a risk factor for bleeds. In some patients, low-level substitution of vitamin K may stabilize the effect of vitamin K antagonists. In others, a genetic predisposition,195 drug–drug interactions (e.g. with amiodarone, propafenone potentially with statins, and a very long list of others),196 and altered drug excretion and metabolism are potentially under recognized causes for unstable INR values. These considerations warrant evaluation in controlled settings.

To better understand factors associated with bleeding in AF patients, bleeding events should prospectively be collected and counted. We suggest the International Society for Thrombosis and Haemostasis (ISTH) definition of major, clinically relevant non-major, and minor bleeds,197 possibly with modifications as to the amount of haemoglobin loss that should counted as ‘major bleeding’. As stated before, the outcome ‘stroke’ should include both ischaemic and haemorrhagic stroke.23 This requires differentiation of extracranial and intracranial bleeds. Intracranial bleeds are separated into intracerebral and intracranial extracerebral (subdural, subarachnoid haemorrhage). Prospective validation of bleeding risk scores could then be performed in longitudinal AF cohorts from controlled trials or registries, perhaps in combination with stroke risk stratification schema, to allow ‘net clinical benefit’ decision-making. Last, cerebral MRI for white matter lesions and microbleeds should be explored in AF cohorts to identify patients at risk for microbleeds. In addition, cerebral microbleeds could be predictors of major bleeds.

What is the best antithrombotic therapy in specific settings in atrial fibrillation patients?

The high prevalence of AF and the high likelihood of concomitant cardiovascular disease steers many AF patients into situations that urge for dual or ‘triple’ antithrombotic therapy. The optimal antithrombotic regime in these patients is not well-defined at present, and often poses clinical problems that are difficult to resolve due to lack of systematic data.

Atrial fibrillation patients with concomitant coronary artery disease

Further research is certainly warranted in patients that require a combination of anticoagulant therapy and platelet inhibition, e.g. anticoagulated patients undergoing percutaneous interventions. Better prediction of severe bleeding events would be needed to incorporate a bleeding risk scheme into clinical practice.198 For example, AF patients may present with an acute coronary syndrome and/or require percutaneous coronary interventions, often including stenting. Published case series in AF patients undergoing stenting support the use of triple antithrombotic therapy,199,200 but the optimal duration of such treatment—given the need to balance recurrent cardiac ischaemia and/or stent thrombosis against stroke prevention and bleeding risk—remains uncertain, especially with the use of drug eluting stents. Similarly, many anticoagulated AF patients have stable coronary and/or peripheral artery disease, and common practice is to treat such patients with warfarin plus one antiplatelet drug, usually aspirin. However, the available data suggest that adding aspirin to warfarin does not reduce stroke or vascular events (including myocardial infarction), but substantially increases bleeding events.201204 It was noted that the benefits from aspirin post-myocardial infarction occur early on,205 and compared with aspirin, oral anticoagulation has similar effects on the prevention of coronary events in patients with known coronary artery disease.206,207 Published consensus recommendations suggest triple therapy in short term, and anticoagulation monotherapy thereafter.199 An ongoing European Society of Cardiology Working Group on Thrombosis Position Statement endorsed by EHRA and by the European Association for Percutaneous Coronary Revascularization can be expected in 2009. In the next years, trial and registry data on this growing clinical issue will become available.

‘Bridging’ of antithrombotic therapy in atrial fibrillation patients undergoing interventions

There is also some uncertainty over best ‘bridging’ antithrombotic therapy strategy in patients in whom a therapeutic intervention with a risk for bleeding is planned, e.g. an operation. Current AF guidelines suggest that anticoagulation can be stopped for 1 week,4 whereas other recommendations suggest a ‘half-dose’ bridging therapy with low-molecular heparin.208 Similar uncertainty exists for the duration of anticoagulation in low-risk patients after AF ablation: the three published recommendations on post-ablation anticoagulation28,209,210 have some subtle differences, and there still remains a need for prospective randomized trials. In addition, whether such therapeutic schemes should be modified based on the supposed absence of AF recurrences is also not known.

Antithrombotic therapy in atrial fibrillation patients with an acute stroke

In the setting of AF and acute stroke, immediate anticoagulation (largely heparin) in AF patients suffering from an acute stroke did not show benefit over aspirin211—and benefit (if any) was offset by bleeding. Currently, anticoagulation is usually initiated in the first days after a stroke, often depending on the cerebral imaging results, ‘perceived bleeding risk’, and severity of stroke. Often, patients with transient ischaemic attacks are at high risk of subsequent stroke. The optimal care of these patients is currently studied, and some observations suggest that intensive and hospital-based care immediately after a transient ischaemic attack can prevent subsequent strokes.212 In this context, it will be worth testing whether intensified ECG monitoring is useful in patient with a ‘cryptogenic’ stroke or transient ischaemic attack to identify undiagnosed intermittent AF.23,24 Other controlled (e.g. cluster-randomized) trials may evaluate anticoagulation strategies in the early phase after a stroke.

When is discontinuation of anticoagulant therapy appropriate?

Given the fact that AF is a chronically progressing arrhythmia, anticoagulation is usually, at least conceptually, life-long.170 Nonetheless, there is a perceived need to test whether ‘apparent’ sinus rhythm maintenance would allow intermittent discontinuation of anticoagulation in selected patient groups. At present, however, it appears likely that discontinuation of antithrombotic therapy in patients with AF and a risk for stroke will be confined to small, highly selected patient groups.

Improving safety for and compliance with thromboprophylactic therapy

Despite the proved usefulness of continuous anticoagulation for many AF patients, discontinuation of anticoagulant therapy often happens in real life due, for example, to dementia, intercurrent illness, and other causes and is, in part, explainable by poor risk–benefit appreciation. But even in controlled trials, the target INR range (23) is achieved in only 60–65% of the time.213 In real life, this percentage is likely to be even smaller and may be too small for the expected clinical benefit.213 This shortcoming of vitamin K antagonists may be overcome with newer, fixed-dose anticoagulants (direct thrombin inhibitors, oral Factor Xa Inhibitors). Nonetheless, the first clinically tested new antithrombotic agent, ximelagatran, was not superior in preventing strokes in AF patients when compared with warfarin.214,215 This area requires further research in settings that are close to ‘real life’, e.g. large registries or trials with low inclusion threshold.

3b. Cardiovascular risk management in atrial fibrillation patients

Most patients with AF present with other concomitant diseases and conditions that are not only known to promote AF-related complications, but also can contribute to AF itself by causing ultrastructural changes and atrial load (Table 8). Understanding the interplay of these factors with occurrence and recurrence of AF and its complications is an urgent issue for the management of AF patients.

View this table:
Table 8

Risk factors associated with first manifestation of atrial fibrillation, progression of atrial fibrillation, and atrial fibrillation-related complications

ConditionFirst manifestationProgressionComplications
Severe mitral valve diseaseXXX
Severe aortic valve disease?X??
Age (linearly per decade)XXX
Hypertrophic cardiomyopathyXXX
Other genetic defects associated with inherited cardiomyopathiesX??
Arterial hypertensionX?X
LV hypertrophyX??
Diastolic dysfunction of the left ventricle/filling dysfunctionX?X
Large left atrial volume??X
Left atrial scar tissue???
Reduced LV function??X
Clinical heart failureX?X
Coronary artery disease??X
Peripheral vascular disease??X
Diabetes mellitusX?X
Sleep apnoeaXX?
Manifest hyperthyroidismX?X
Subclinical thyroid dysfunctionX/???
High statureX??
Intensive endurance training or sports activityaX??
Marked sedentary lifestyleaX??
Alcohol consumption acute/chronicXXX
  • aThere appears to be a ‘U-shaped’ relation between physical activity and incidence of AF.

Relevance of risk factors for atrial fibrillation progression

Retrospective analyses suggest that angiotensin receptor inhibitors and angiotensin receptor blockers, agents with antifibrotic effects and antihypertensive action, can reduce the likelihood of developing AF in patients with heart failure or LV hypertrophy.162164 Similar effects have been attributed to statin therapy, albeit based on weaker data,216 and possibly to positive pressure ventilation or intensified diet which are known to improve diastolic ventricular function.217,218 Most of these studies were limited not only by their retrospective nature, but also by infrequent assessment of cardiac rhythm. One of the most pressing issues in this field is to systematically and prospectively obtain AF outcomes in intervention trials targeting cardiovascular risk. The group proposes ECG substudies in some of the ongoing large hypertension and/or heart failure trials. This is recommended for large controlled trials in diabetes, hypertension, heart failure, sleep apnoea, obesity, coronary artery disease, and even in high-risk populations without overt cardiovascular disease. This should be done by ECG, 24 h (Holter) ECG, and at times by implantable ECG recorders. Such an approach has been advocated by the European Society of Cardiology for some time. The group suggests the development of a ‘bolt-on package’ for such trials that should include echocardiography, ECG and Holter ECG, and possibly serum markers for genetic and potentially inflammatory markers.

There is a need for additional epidemiological information on AF risk factors and interactions of the above-mentioned factors. Alcohol consumption, physical activity, thyroid stimulating hormone, inflammatory markers, neurohormones, and genetic risk should be measured as additional risk factors as well as their interaction with left atrial size and age. Available data suggest that there is a ‘U-shaped effect’ of physical activity on AF development. There should be a systematic study of cardiovascular indices (blood pressure, LV function, and left atrial size and function) and rhythm in a high-training population (cyclists, soccer players) with long-term follow-up. An alternative study design would be a case–control design, as presently under way.219 An additional question would be whether interventions are warranted to prevent AF in high exercise-level athletes. A longitudinal study may be of interest to other cardiology communities and to the ‘sports industry’.

Comprehensive cardiovascular risk management in patients with atrial fibrillation

There is increasing awareness of the importance of concomitant conditions that predispose to AF (Table 8). In some patients, AF may be related to specific triggering events (such as electrocution, alcohol consumption, vagal reaction) or reversible diseases such as hyperthyroidism. Having a first episode under these circumstances may indicate risk for later appearance of AF. Large registries and/or prospective studies are needed to define clinical factors related to a high risk of recurrence at the time of the initial diagnosis of AF. A population thus defined would be appropriate for interventional studies at an early stage of the disease.

There may be more to the first episode of AF than early therapy of the arrhythmia itself. Many patients carry relevant concomitant diseases and treatable risk factors for AF-related complications. In such individuals, AF may actually represent an additional risk factor suggesting a need for more intensive treatment of preventable predictors of cardiovascular complications. Severe mitral valve disease,220223 clinically apparent coronary artery disease,221,223 hypertrophic cardiomyopathy,224226 reduced LV function, and clinically detectable structural heart disease227 are concomitant conditions that indicate a high risk of AF recurrence and of cardiovascular complications. Arterial hypertension, obesity, sleep apnoea, diabetes mellitus, chronic renal dysfunction, and other vascular disease are also ‘risk factors’ found in AF patients.171,223,228230 Presence of these factors often renders AF more persistent or permanent,228,231 providing clinical evidence that the ‘ultrastructural substrate’ suspected in patients with such concomitant conditions is relevant for perpetuation of AF. Left ventricular perfusion is impaired by AF, especially in patients suffering from coronary artery disease, and atrial ischaemia promotes AF.232234 Atrial fibrillation patients should receive intensified treatment and interventions to reduce their risk of cardiovascular complications,235238 and sinus rhythm patients with AF-related stroke risk factors (e.g. CHADS2 scores ≥2) may be candidates for ECG screening.

Valve therapy intervention populations may provide important opportunities for AF research, e.g. a controlled trial of ‘upstream therapy’ in mitral valve surgery patients to promote sinus rhythm or a trial of mitral valve surgery with and without left atrial appendage excision and/or with and without concomitant intraoperative AF ablation. Although it may seem reasonable to propose early mitral valve surgery to prevent AF (a surgical form of ‘upstream therapy’), the operation is highly invasive, and this concept would require testing in a controlled trial.

3c. Novel therapeutic goals

Accepted outcomes in AF patients are provided in Table 2.23,24 Using these outcomes, studies have been disappointing in terms of demonstrating benefit of rhythm control. The reason for this lack of demonstrated benefit is likely in part due to insufficient rhythm control (see above). In addition, it may relate to failure to identify the appropriate outcome (calling for novel clinical outcome measures) or incorrect measurement of the established outcome (calling for improved measurement techniques or parameters). There has recently been a shift of concern from symptoms and rhythm to cardiac structure and function, other end organ damage (e.g. in the brain), quality of life, and negative effects of the disease on global functioning and well-being.

During the conference, the group prioritized potential novel therapeutic goals for evaluating AF therapy. In setting priorities, we considered the following criteria: potential impact (relevance to the patient and potential change in clinical practice); feasibility (likelihood of a meaningful result, availability of measurement tools, time-line, cost); and novelty or unresolved nature of the issue. The following therapeutic goals may be considered in the future (Table 9).

View this table:
Table 9

Novel therapeutic goals

8Silent stroke (including asymptomatic intracerebral bleeds) and cognitive function
9Social functioning and disease-related quality of life
10Progression to more sustained forms of AF
11LV function
12Left atrial function
  • This table adds potentially useful novel metrics to the list of previously advocated relevant trial outcomes in AF patients (Table 2) that may help to quantify a subtle, but in the long-term potentially relevant, effect of AF therapy.

Silent stroke, cognitive decline, and hippocampal atrophy are associated with AF,239 and silent stroke is associated with cognitive decline239 and a two-fold higher chance for developing dementia.240 Even with anatomic brain findings that are ‘silent’, there may be subtle cognitive effects that would be evident if appropriately investigated. If AF increased the frequency of silent stroke or cognitive decline, this would suggest that more aggressive rhythm control would be warranted even in patients with minimal or no symptoms of their arrhythmia. As such, we recommend assessing silent stroke and cognitive decline associated with AF in ongoing rhythm control trials.

Quality of life is markedly impaired in AF patients, and AF is one of the most common reasons for hospitalizations.11 Even patients with otherwise asymptomatic AF report lower quality of life compared with subjects in sinus rhythm.11 Quality of life has been evaluated prospectively and can be improved both by rate and by rhythm control therapy, often without a difference between the two strategies.10 Although many patients present with ‘accidentally diagnosed’ and/or asymptomatic AF, adequate control of ventricular rate and successful maintenance of sinus rhythm can improve quality of life and exercise capacity in AF patients.12,13,241 This finding is consistent with the general assumption that sinus rhythm maintenance is associated with improved quality of life in symptomatic patients and suggests that trial design and/or measurement techniques have been inadequate. We recommend that AF-specific instruments to assess quality of life should be developed and validated, and then applied prospectively, in combination with established methodology. The assessment should include anxiety, depression, and feelings of resignation or futility. In addition, any trial should consider unique features in the AF population. For example, patients with paroxysmal symptoms may under-emphasize the effect of AF on quality of life if queried at an intercurrent asymptomatic interval. Furthermore, quality of life assessment might include interview of family or spouse to assess their perspective on the patient symptoms as well as effect of illness on the family unit. It is noted that alternate instruments are already developed, such as SAGE (standard assessment of global activities in the elderly) which could be used, possibly with slight adaptations for younger populations, to assess global social functioning over time in patients with AF.242244

Avoidance of AF progression is important in view of the increased complexity of arrhythmia management during later stages of AF, along with a potentially higher risk of complications as the episodes become longer or permanent. We therefore recommend a prospective trial in patients with newly diagnosed or recent-onset AF and an aggressive rhythm control intervention strategy (including cardioversion, antiarrhythmic drugs, and ablation) compared with standard care, with the outcome measure of progression to persistent or permanent forms of AF.

Left ventricular function is an accepted surrogate outcome for cardiac complications.23,24,141,145 Traditionally, LV volumes and ejection fraction are used to assess LV function. Both can reliably be measured by 2D echocardiography in the normal heart without regional dysfunction. Using 3D echocardiography, superior assessment of LV volumes and ejection fraction has been demonstrated, with good correlation with MRI or multislice computed tomography. In addition to the use of 3D echocardiography to assess LV function in AF patients, novel echocardiographic indices may provide more subtle information on LV function, e.g. measurement of myocardial longitudinal velocities (using tissue Doppler imaging) and strain (providing information on active deformation of the myocardium). The assessment of global LV strain provides a particularly interesting marker of systolic performance and permits detection of subtle abnormalities even before LV ejection fraction diminishes. Finally, it is possible to calculate (based on apical and basal LV rotation) the torsion of the left ventricle adding another dimension to LV function assessment.245 Diastolic function may also be important for prognosis, but its measurement is limited by biological variations in RR interval timing and LV filling during AF. Large studies employing novel technology for systolic and diastolic LV function assessment in patients with AF are needed.

Left atrial function may be a therapeutic goal in itself based on the assumption that preserved left atrial function may prevent thrombus formation in the atria and contribute to global cardiac performance. Two-dimensional echocardiography can assess left atrial dimensions and estimate left atrial volumes, but 3D echo will be more accurate.246 Magnetic resonance imaging and multislice computed tomography provide similar information on left atrial volumes, albeit using image averaging over several cardiac cycles. In addition, these techniques permit precise assessment of pulmonary vein anatomy.246 Atrial volumes and pulmonary vein anatomy change with each cardiac cycle. Therefore, it is required to define the time in the cardiac cycle (diastole, systole) during which atrial volumes are measured, as well as the heart rate during the measurement. Left atrial function has so far mainly been evaluated using pulsed-wave Doppler derived mitral valve inflow parameters, including the E (early inflow, passive inflow) and A (late inflow, atrial contribution) waves.247 Left atrial function can be derived from left atrial volumes and expressed using the following parameters: total atrial emptying fraction, active atrial emptying fraction (active contraction), passive atrial emptying fraction (conduit function), and atrial expansion index (reservoir function).76 In addition, left atrial velocities can be derived from tissue Doppler, and left atrial strain permits direct assessment of left atrial active deformation.247,248 These techniques may in the future help to identify an effect of therapy on left atrial function in AF patients.

A call for coordinated research

One of the great opportunities and, at the same time, challenges of a comprehensive research program is the chance to perform studies that are linked, coordinated, and harmonized from the onset. Several advantages result: the coordinated introduction of registries, embedded cohort studies, and randomized controlled trials facilitates patient recruitment, permits long-term follow-up, and allows health-economic evaluations. If the research program is modularized, the modules may be easily transferred to other settings or added to ongoing studies as bolt-on packages.

Coordinated studies allow to integrate the individual patient data into huge databases. Pre-planned individual patient data meta-analyses (IPDMA) can thus be performed that are far more powerful than using isolated studies and have the potential to overcome many shortcomings of traditional literature-based meta-analyses. Individual patient data meta-analyses should be performed using general or generalized mixed or multilevel linear models that allow to cope with cluster effects and other correlation structures that are unavoidable in multicentre research or longitudinal studies. In particular, these models allow to study not only the impact of interventions on the patient level, but also on the centre level which may be of particular importance if different ablation techniques are compared that can only be introduced centrewise.

Randomized controlled trials can be performed using adaptive designs that give a greater flexibility in the case of new insights from other studies. However, such gains in flexibility require a more complicated and challenging trial design. Complicated trials may be less convincing and trustworthy than simple trials. Blinded interim analyses that can be performed with minimal loss of power may help to improve performance and data quality. Unblinded interim analyses require specification in advance, but contribute substantially to fast access to crucial information.

Knowing that in many ‘real-life’ settings, it will be difficult to coordinate research initiatives a priori (e.g. due to competing sponsors), we emphasize the mutual advantages of coordinated research activities for all participants. If formal coordination cannot be achieved, a clear and unified definition of outcome parameters23,24 and a unified scheme for patient characterization and assessment of relevant outcomes should be a minimum requirement for large controlled trials in AF.

‘Translation’ of therapeutic concepts into clinical practice

We have called for a ‘translational’ research approach from bench to bedside before. Research monitoring the ‘translation’ of new diagnostic and therapeutic concepts into daily life clinical practice is also needed. Efforts to achieve such translation effectively are needed. In this regard, the organizing bodies of the 2nd AFNET/EHRA consensus conference have an important role and obligation. The effectiveness of such translation of new findings and concepts requires careful evaluation to identify optimal communication tools and incentives to make good therapy available to everyone.

Starting points for research:

  • Preventing strokes is one of the most important clinical issues in AF patients. The CHADS2 score and similar risk stratification schemes are adequate for the identification of patients at high and at low risk for stroke. Unfortunately, a large number of patients are currently classified as ‘intermediate risk’ for stroke, leaving physicians and patients without clear guidance on antithrombotic management.

  • Prevention of strokes in patients with silent AF, whose first clinical manifestation of the arrhythmia is a stroke or a transient ischaemic attack, is an unresolved clinical problem.

  • Clinical risk factors for bleeding and stroke risk overlap to a relevant extent in AF patients.

  • An increasing number of AF patients at risk for stroke experience acute coronary syndromes or stent implantation that, in principle, require combined anticoagulant and antiplatelet therapies. The optimal antithrombotic therapy in such patients is not known.

  • Despite its proved net benefit on death and stroke, anticoagulation is underused, and when used, anticoagulants often result in patients being inadequately anticoagulated or over-anticoagulated in a relevant proportion of the time.

  • In addition to established outcomes in AF patients, novel therapeutic goals may help to identify subtle effects of rhythm control therapy.

Research perspectives:

  • Are there any new risk factors for stroke in ‘intermediate risk’ patients (clinical parameters, biomarkers, or imaging markers) that can be identified in analysis of databases from ongoing antithrombotic therapy trials in AF patients and validated in appropriate registries?

  • Can existing databases be used to identify new clinical markers for bleeding risk?

  • What is the value of ECG screening in patients who would be candidates for anticoagulant therapy if AF was known?

  • What is the value of subsequent initiation of adequate AF management?

  • Can intensified INR monitoring, genetic testing prior to initiation of vitamin K antagonist therapy, and low-dose supplementation of vitamin K improve compliance with and increase in efficacy and safety of oral anticoagulant therapy?

  • Can structured delivery of anticoagulant therapy and implementation of anticoagulation recommendations in managed health-care settings improve the situation in comparison with standard care?

  • Can anticoagulants that are under clinical development improve anticoagulant therapy in AF patients? Thorough and prospective surveillance of safety and efficacy after assumed approval is important for each of these substances.

  • Is dual or ‘triple’ antithrombotic therapy needed in AF patients undergoing special situations such as stent implantation? If so, for how long? This should be tested in controlled trials. Is antithrombotic ‘bridging therapy’ management needed in AF patients undergoing elective operations?

  • Is AF a ‘cardiovascular risk factor’ in addition to established cardiovascular risk factors? This should be explored in registries. It is likely that the presence of AF warrants more intensive risk factor management. Future cardiovascular risk management trials may consider AF as a risk factor for cardiovascular events and death.

  • Does sinus rhythm carry benefits beyond the ‘classical’ outcome parameters death, stroke, quality of life, and LV ejection fraction? New therapeutic goals that could measure effects of sinus rhythm maintenance may comprise, among others:

    • –cognitive dysfunction as assessed by functional tests

    • –silent strokes as assessed by MRI

    • –quality of life as assessed by social functioning

    • –left atrial function and size as assessed by novel echocardiographic parameters.

4. Outlook: from ‘rate versus rhythm’ to comprehensive management of atrial fibrillation

Therapy of AF patients is inadequate at present, with a high morbidity and mortality still assigned to AF on a population scale. Many aspects of the current care for AF patients differentiate between treatment ‘strategies’ which may, in many patients, rather need to be combined to achieve an optimal result. This idea applies foremost to the ‘old’ debate of ‘rate versus rhythm control’ since rhythm control is generally added to underlying (continued) rate control therapy, but also applies to therapy of conditions that predispose to AF and contribute to cardiovascular complications including antithrombotic therapy to prevent strokes and prevention of heart failure and acute coronary syndromes. Almost all working groups that discussed different facets of AF management suggested ‘comprehensive AF management’, ‘combination of upstream and classical therapies’, or ‘therapy of the different factors that cause AF and its complications’ as an answer to the current therapeutic dilemma for AF patients. The overall final conclusions of this conference are that:

  • AF is a rising epidemic and, in almost all patients, is a slowly progressing, chronic disease. Owing to its consequences and complications, AF represents an unresolved population-based clinical problem in the 21st century.

  • The causes of AF and its consequences, including complications, are multifaceted.

  • Understanding the different pathophysiological processes that cause AF and its complications will help to devise mechanism-based therapies.

  • Adequate therapy of AF will need to simultaneously address management of underlying and concomitant conditions, early and comprehensive rhythm control therapy, adequate control of ventricular rate and cardiac function, and continuous therapy to prevent AF-associated complications.


The 2nd AFNET/EHRA consensus conference was funded by AFNET and EHRA. Industry participants paid an attendance fee.

Appendix 1

The 2nd AFNET/EHRA consensus conference was a group exercise. Many of the concepts, observations, and hypotheses were aired by participants of the conference. The authors of this paper are members of a ‘writing group’ that compiled the main findings of the conference in a style suitable for publication. The organizers of the conference and the members of this ‘writing group’ would like to explicitly acknowledge the contributions of many other participants of the conference. Therefore, a list of all participants of the conference in alphabetical order is published here: Maurits Allessie; Dietrich Andresen; Jeroen Bax; Carina Blomstrom-Lundqvist; Martin Borggrefe; Gianluca Botto; Günter Breithardt; Michele Brignole; Martina Brückmann; Hugh Calkins; John Camm; Riccardo Cappato; Francisco G. Cosio; Harry J. Crijns; Hans-Christoph Diener; Dobromir Dobrev; Nils Edvardsson; Michael Ezekowitz; Thomas Fetsch; Robert Hatala; Karl Georg Häusler; Hein Heidbüchel; Andreas Heppel; Gerd Hindricks; Alexander Huemmer; Carsten Israel; Warren M. Jackman; Lars Joensson; Stefan Kääb; Otto Kamp; Lukas Kappenberger; In-Ha Kim; Paulus Kirchhof; Stefan Knecht; Karl-Heinz Kuck; Karl-Heinz Ladwig; Angelika Leute; Thorsten Lewalter; Gregory Y.H. Lip; João Melo; Jay O. Millerhagen; Lluís Mont; Stanley Nattel; Seah Nisam; Michael Oeff; Dieter Paar; Richard L. Page; Ursula Ravens; Ludger Rosin; Patrick Schauerte; Ulrich Schotten; Anna Schülke; Dipen Shah; Gerhard Steinbeck; Christoph Stoeppler; Ruth H. Strasser; Natalie Taylor; Jan G.P. Tijssen; András Treszl; IsabelIe C. Van Gelder; Panagiotis E. Vardas; Albert Waldo; Karl Wegscheider; Thomas Weiß; Karl Werdan; Stephan Willems; Stefan N. Willich.

Appendix 2

Conflict of interest declaration of the writing group

NameConsulting Fees/HonorariaSpeaker's BureauOwnership/Partner-ship/EmployeeResearch GrantsSalary
Paulus Kirchhof3M Medica
Boehringer Ingelheim
MEDA Pharma
Bayer AG
NoneNone3M Medica/MEDA Pharma
Cardiovascular Therapeutics
German Federal Ministry for Education and Research (BMBF)
Fondation LeDucq
German Resarch Foundation (DFG)
St Jude Medical
Jeroen BaxNoneNoneNoneBiotronik
BMS medical imaging
Edwards Lifesciences
GE Healthcare
St Jude Medical
Carina Blomstrom-LundquistSanofi-Aventis
St Jude
St Jude
Hugh CalkinsAblation Frontiers
Biosense Webster
Sanofi Adventis
Boston Scientific
I Rhythm
NoneNoneBiosense Webster
Boston Scientific
St Jude Medical
John CammSanofi-Aventis
Bristol-Myer Squibb
Bristol-Myer Squibb
Riccardo CappatoBiosense Webster
St Jude Medical
Boston Scientific
Cameron Health
Biosense Webster
St Jude Medical
Boston Scientific
CameronBiosense Webster
St Jude Medical
Boston Scientific
Harry CrijnsSanofi Aventis
Sanofi Aventis
St Jude Medical
St Jude Medical
Sanofi Aventis
Hans-Christoph DienerAbbott
Bayer Vital
Boehringer Ingelheim
Janssen Cilag
Bertelsmann Foundation
Boehringer Ingelheim
European Union
German Federal Ministry for Education and Research (BMBF)
German Resarch Foundation (DFG)
Heinz-Nixdorf Foundation
Andreas Goette3M Medica
NoneNoneGerman Federal Ministry for Education and Research (AFNET; NBL3)
3M Medica/MEDAPharma
Carsten W IsraelMedtronic Inc
St Jude Medical
St Jude Medical
NoneMedtronic Inc
Sanofi St Jude Medical
Karl-Heinz KuckBiosense-Webster
St Jude
St Jude Medical
Gregory YH LipAstellas
Daiichi Sankyo
Stanley NattelAstraZeneca
Pierre Fabre
NoneNonePierre FabreNone
Richard L. PageAstellas
Ursula RavensCardiome
MEDA Pharma
NoneNoneGerman Federal Ministry for Education and Research (BMBF)
Fondation Leducq
German Research Foundation (DFG)
Ulrich Schotten3M Medica
NoneNoneDutch Heart Foundation (NHS)
European Union
Center of Translational Molecular Medicine (CTMM)
Dutch Research Organization (NWO)
German Federal Ministry for Education and Research (BMBF)
Fondation Leducq
Gerhard SteinbeckAstra Zeneca
Bayer Vital
Boehringer Ingelheim
Boston Scientific
St Jude Medical
St Jude
Panos VardasBoston Scientific
St Jude Medical
NoneNoneHellenic Cardiological SocietyNone
Albert WaldoAstraZeneca
Biosense Webster
NoneBoehringer Ingelheim
Bristol-Myers Squibb
CV Therapeutics
Karl WegscheiderAstraZeneca
Bayer Health Care
Boston Scientific
Brahms Diagnostics
Stephan WillemsSt Jude MedicalNoneNoneGerman Federal Ministry for Education and Research (BMBF)None