Introduction
Ever since the term auricular fibrillation was coined in 1909,[1 ] the understanding of atrial fibrillation (AF) leading to important developments
for its management has been constantly evolving, with great improvements having been
achieved in the last few decades.[2 ]
[3 ] The more we learn about AF and its interaction with underlying comorbidity, realizing
that it is not just the arrhythmia but also underlying comorbidities that determine
outcome,[4 ] the greater is our ability to characterize the clinical profile of each patient
to provide information relevant for optimal management decisions.
Currently, international AF guidelines and consensus documents list multiple classifications
of AF ([Table 1 ]),[5 ]
[6 ]
[7 ]
[8 ]
[9 ] each classification addressing a single domain of AF (or patient)-related characteristics.
Although these classifications have been gradually evolving toward better precision
and clinical utility,[10 ] they separately address specific features relevant for the arrhythmia, the patient
or the clinical setting in which AF occurs. Lack of integration becomes burdensome
and sometimes the information needed for treatment decisions is incomplete.
Table 1
Classifications of AF in current international AF guidelines
Major international AF guidelines
AF classification
2016 ESC[6 ]
2019 AHA/ACC/HRS[7 ]
2018 (2014) CCS[8 ]
2018 NHFA/CSANZ[9 ]
Temporal pattern
• First diagnosed
• Paroxysmal
• Persistent
• Long-standing persistent
• Permanent
• First diagnosed
• Paroxysmal
• Persistent
• Long-standing persistent
• Permanent
• First diagnosed
• Paroxysmal
• Persistent
• Permanent
• First diagnosed
• Paroxysmal
• Persistent
• Long-standing persistent
• Permanent
Symptom severity
EHRA Symptom severity score:
• I No symptoms
• IIa Mild
• IIb Moderate
• III Severe
• IV Disabling
• Asymptomatic (silent)
• Symptomatic
SAF score:
• 0 (asymptomatic)
• 1 (minimal impact on QoL)
• 2 (minor impact on QoL)
• 3 (moderate impact on QoL)
• 4 (severe impact on QoL)
• Asymptomatic (silent)
• Symptomatic
Underlying comorbidity
• Valvular AF
Replaced with “AF in patients with MS or prosthetic heart valves”
Yes
Yes
Yes
Clinical type reflecting different causes of AF
• Secondary to structural heart disease
• Focal
• Polygenic / Monogenic
• Postoperative
• In athletes
•
No specific list
No specific list
No specific list
Surface ECG appearance
Coarse / Fine
No
No
No
No
Mode of onset
Vagal / Adrenergic
No
No
No
No
Abbreviations: ACC, American College of Cardiology; AF, atrial fibrillation; AHA,
American Heart Association; CCS, Canadian Cardiovascular Society; ECG, electrocardiogram;
EHRA, European Heart Rhythm Association; ESC, European Society of Cardiology; HRS,
Heart Rhythm Society; MS, mitral stenosis; NHFA/CSANZ, National Heart Foundation of
Australia / Cardiac Society of Australia and New Zealand; QoL, quality of life; SAF,
symptom severity in AF.
The most widely used classification of AF based on AF episode duration and temporal
patterns (that is, the 3-P classification to first-diagnosed, paroxysmal, persistent/long-standing
persistent, and permanent AF, depending on the duration of the arrhythmia and its
mode of termination) was proposed by Gallagher and Camm in 1997[11 ]
[12 ] and formally adopted by the European Society of Cardiology Working Group on Cardiac
Arrhythmias in 1998.[13 ] Before this, Lévy et al had redefined paroxysmal AF as lasting no more than 7 days
and terminating spontaneously.[14 ] The temporal pattern-based classification of AF has contributed to a better understanding
of AF prevention and treatment but its limitations and the need for a multidimensional
AF classification are being increasingly recognized.[15 ]
Importantly, the pattern-based classification of AF provided standardization of AF-related
nomenclature and was easily adopted owing to its simplicity. However, despite generally
correlating with the extent of the atrial substrate, remodeling, and AF-related outcomes,
the pattern-based classification of AF lacks precision in differentiating among specific
features relevant for optimal treatment decisions with regards to stroke prevention
and rhythm control strategies such as catheter ablation.[16 ] Indeed, formal recommendations for the management of AF are not based on the pattern
of AF, except for the decision to restore sinus rhythm, but the terms paroxysmal and
persistent AF include a large conglomerate of patients with wide variations in AF
patterns, substrates, and other characteristics and, consequently, different needs
in management. Owing to the availability of continuous rhythm monitoring, we learned
only recently that patients classified to the same clinical AF category may be inherently
heterogeneous in terms of temporal AF persistence and AF burden.[17 ]
[18 ] The differentiation between paroxysmal and persistent AF is often very cumbersome,
both patterns may be observed in the same patient during follow-up, and the heterogeneity
in recurrences and progression of AF poses a challenge to a rhythm-based classification.
Also, it describes only the arrhythmia, whereas other relevant features such as cardiovascular
risk factors and underlying comorbidities or the extent of atrial substrate changes
are not included.[6 ]
In 2010, Lubitz et al proposed a more extensive classification of the arrhythmia,[19 ] and the multidimensional form of classification of AF was presented in more details
by Camm et al in 2012.[15 ] Given the multiplicity of factors relevant for optimal management of AF in clinical
practice, including advances in monitoring of AF and risk assessment tools, and evolving
treatment options apart from the complexity of AF itself, a simple but comprehensive
characterization of AF is urgently needed.
From this viewpoint, we propose a paradigm shift from classification toward a structured
characterization of AF addressing specific domains that have treatment and prognostic implications
to become a standard in clinical practice, thus streamlining the evaluation of AF
patients at all health care level, with the goal to facilitate communication among
physicians, treatment decision-making, and optimal management of AF patients.
Specifically, we propose the 4S-AF structured pathophysiology-based characterization (rather than classification) scheme that includes four AF- and patient-related domains
(Stroke risk; Symptoms; Severity of AF burden; Substrate severity) ([Fig. 1 ]).
Fig. 1 The 4S-AF scheme for characterization of patients with atrial fibrillation. AF, atrial
fibrillation; CT, computed tomography; EHRA, European Heart Rhythm Association; MRI,
magnetic resonance imaging; QoL, quality of life; TOE, transesophageal echocardiogram;
TTE, transthoracic echocardiogram.
The Need for a Structured AF Characterization
Being a multifaceted, complex, and very heterogeneous disease, AF requires structured
patient management and multiple treatment decisions addressing different treatment
domains such as stroke prevention, symptom improvement, and management of concomitant
conditions and risk factors.[6 ]
[20 ] Importantly, these decisions should be regularly reevaluated, owing to dynamic changes
in the patients' individual risk profile.[21 ]
[22 ]
Sometimes the complexity of patient- and/or AF-related features requires multidisciplinary
engagement to facilitate treatment decisions relating to thromboembolic protection,
cardioversion, antiarrhythmic drug therapy, left atrial catheter ablation, or rate
control. In daily clinical work, AF-related communication among practitioners including
expert consultants needs to be rapid but comprehensive, efficiently providing all
the relevant information to facilitate treatment decisions.[10 ]
Modern medicine is characterized by rapidly evolving means of communication of health-related
information among physicians (and patients), including computer- or mobile application-based
decision support tools for clinicians and/or patients. Recently, several such tools
have been developed specifically for AF, and preliminary data suggest a potential
for improving AF management and patient outcomes using these tools.[5 ]
[23 ]
[24 ] Importantly, their output strongly depends on comprehensiveness (and accuracy) of
AF-related and other health information entered for a given patient.
Moreover, the use of electronic medical and health records is increasing worldwide.
These systems provide an opportunity for the rapid creation of large data sets that
can be used for research purposes.[25 ] The use of a uniformly structured characterization of AF patients across various
data sets would further improve the compatibility of data from various sources.
In all these circumstances, including routine clinical practice, the use of a more
structured characterization of AF patients, with well-defined descriptors would facilitate
not only the communication among involved practitioners but also treatment decisions
and overall management of AF patients,[20 ] ultimately improving outcome for the individual patients and health care costs.
Indeed, using such a structured AF patient characterization scheme would help physicians
achieve a good balance of simplicity, practicality, and information-based treatment
decision-making.
The 4S-AF Scheme for Characterization of Patients with Clinically Diagnosed AF
In addition to identification and management of cardiovascular risk factors and underlying
comorbidity, the two key AF-specific treatment decisions are (1) the need for thromboprophylaxis
to prevent stroke or systemic embolic events, most commonly the choice of oral anticoagulant
therapy (OAC), and (2) appropriate choice of rate and/or rhythm control to improve
symptoms and prevent complications of AF such as heart failure.[20 ] In addition, the rhythm control treatment strategy often involves choosing between
long-term antiarrhythmic drug medication or AF ablation therapy (i.e., surgical or
catheter ablation).[6 ]
Whereas patient's age (and other demographic features), cardiovascular risk factors,
and comorbid diseases should be routinely noted in the patient's medical record, a
more detailed characterization of AF-specific features and patient-related characteristics
combined into a structured system for a comprehensive description of AF would facilitate
the choice of optimal treatment (including interventional procedures), not only influencing
the success of interventional procedures but also improving patient outcome.
We, therefore, propose a structured characterization of AF patients in clinical practice
using the 4S-AF scheme that addresses Stroke risk, Symptom severity, Severity of AF
burden, and Substrate for AF to be used in clinical practice ([Fig. 1 ]). This approach is based on the principles generally similar to the most widely
used tumor clinical staging system (i.e., the TNM tumor classification) but unlike
tumors, AF would be characterized rather than staged.
The 4S-AF scheme would provide essential information needed for decision-making on
the use of OAC, choice of rate or rhythm control and between AF ablation or antiarrhythmic
medication, and treatment of underlying cardiovascular comorbidities and risk factors.
At this point, we do not propose this system as a definitive treatment-decision tool,
since data on its association with treatment outcomes are currently lacking, but rather
as a structured descriptive aid in the decision-making process. Nevertheless, we certainly
do not exclude that with the acquisition of data in the future and evaluation of the
proposed characterization scheme, this might become the case.
The 4S-AF Scheme Domains
The Stroke risk (St) domain characterization is currently based on the routinely used and guideline-recommended
clinical risk factor-based CHA2 DS2 -VASc score for stroke risk assessment, and the indication for OAC use is established
as per guideline recommendations from the European Society of Cardiology and other
international bodies ([Fig. 1 ]).[6 ]
Multiple blood biomarkers (e.g., B-type natriuretic peptide, cardiac troponin, biomarkers
of renal function, etc.) and indices of atrial structural and functional remodeling
obtained by various imaging tools have been shown to correlate with individual AF-related
thromboembolic risk, and several biomarker-based stroke risk scores[26 ] have been validated, showing modest but statistically significant improvement in
stroke risk prediction when biomarkers are added to clinical risk factors.[27 ]
Recent evidence suggest that the burden of AF may be associated with thromboembolic
risk and all-cause mortality.[28 ] Whereas the landmark trials of nonvitamin K antagonist oral anticoagulants versus
warfarin for stroke prevention in AF consistently showed that the residual thromboembolic
risk among anticoagulated patients was significantly lower in those with paroxysmal
as compared with persistent AF even after adjustment for baseline characteristics,[29 ] data from earlier trials that included nonanticoagulated controls[30 ] and contemporary AF registries or population-based studies[31 ]
[32 ] are conflicting.
While the clinically assessed burden of AF using only intermittent electrocardiographic
(ECG) monitoring might not significantly impact stroke risk (that is, clinically evident
AF has already crossed the threshold for elevated stroke risk), an increasing body
of evidence derived from cohorts of patients implanted with pacemakers or defibrillators
capable of continuous heart rhythm monitoring or a high-risk cohort of patients with
insertable cardiac monitors or wearable monitoring devices suggests that even the
burden of subclinical AF could impact the risk of stroke.[33 ] However, it remains to be clarified whether stroke risk is a continuum or there
is a specific threshold of AF burden at which the risk significantly increases.
Owing to the rapidly advancing technologies for computer-based decision support tools
and machine learning application in medicine, the description of stroke risk in AF
patients may evolve beyond clinical risk factor-based approach, but the decision to
use OAC will always be a binary yes/no entity.[34 ]
The Symptom severity (Sy) domain addresses the patient-centered, symptom-directed focus of AF management. This component
focuses on the severity of symptoms, currently using the European Heart Rhythm Association
(EHRA) symptom score, and is important for treatment decisions.[20 ]
[6 ] Of note, the EHRA symptom severity score is prognostically relevant for adverse
cardiovascular events.[35 ]
However, the EHRA symptom score is physician-assessed, reflecting how physicians weigh
the symptoms of their AF patients rather than the patients' perception that may differ
substantially.[36 ] In addition, the EHRA score cannot precisely differentiate between AF-related and
concomitant comorbidity-related symptoms. Indeed, in symptomatic patients with underlying
comorbidities, optimal management of concomitant diseases is necessary before proceeding
with a symptom-guided treatment decision for left atrial ablation, and the descriptor(s)
of the symptom severity domain may change in the future to include the assessment
of quality of life, patient-perceived burden of treatment,[37 ] and other features.
The Severity of AF burden (Sb) domain characterizes the proportion of time spent in AF and density of AF episodes in time
(if the arrhythmia is not permanent), including also the mode of termination (spontaneously
terminating or not) as a potential indicator of the propensity toward the development
of chronic arrhythmia.
As mentioned, the clinical adjudication of AF burden using the classification to paroxysmal,
persistent, or permanent AF may be imprecise in distinguishing between paroxysmal
and persistent AF (and, occasionally, even permanent AF), which may influence the
selection of patients suitable for specific AF ablation procedures. As exemplified
in [Fig. 2 ], the outcome of antiarrhythmic drug therapy varies widely even within similar patient
groups with the same type of AF in randomized clinical trials comparing AF ablation
versus antiarrhythmic medication (control arm), which underlines the importance of
finding better tools for characterization of AF features in clinical decision making.
The 4S-AF scheme proposes an empirical, clinical assessment-based semiquantification
of AF burden that reflects elements relevant for treatment decision with regards to
rhythm control, antiarrhythmic drug therapy, and catheter ablation ([Figs. 1 ] and [3 ]).
Fig. 2 Atrial fibrillation recurrence rates in the control arms (antiarrhythmic drugs) in
various randomized trials comparing antiarrhythmic drugs and atrial fibrillation ablation.[50 ]
[51 ]
[52 ]
[53 ]
[54 ]
[55 ]
[56 ]
[57 ]
[58 ]
[59 ]
[60 ], *AF ablation as first-line therapy; N, number of patients in the trial; PAF, paroxysmal
atrial fibrillation; Pers, persistent atrial fibrillation. Note the disparate outcomes
in the control arms in the various trials despite similar AF types in randomized trials
comparing antiarrhythmic drugs and AF ablation. Most studies included mainly paroxysmal
AF patients and evaluated AF ablation as secondary treatment. Percentages in text
denote the dominating AF type. The bars denote the % of freedom from AF at 12 months.
Fig. 3 A hypothetical treatment decision supporting algorithm using the 4S-AF scheme for
characterization of patients with atrial fibrillation in clinical practice. AF, atrial
fibrillation; CV, cardiovascular; LA, left atrium; OAC, oral anticoagulant therapy.
In routine practice, clinicians may roughly assess the duration of symptomatic AF
episodes and history of AF via detailed history taking and intermittent ECG monitoring.
Thus, AF episodes lasting for hours to days can be arbitrarily labeled as short, those
lasting weeks to months would be intermediate, and long AF episodes would be those
lasting 12 or more months. The density of AF episodes can be expressed as the annual
number of episodes, for example, up to 1 to 3 per year (infrequent), > 3 per year
(frequent), or occurring daily to monthly (very frequent). Given that this is only
an arbitrary, not validated stratification, the widely adopted and guideline-recommended
temporal pattern-based classification of AF can be used initially.
The assessment of AF burden is evolving in parallel with rapidly advancing wearable
and insertable technologies for prolonged monitoring of AF that are becoming increasingly
affordable and convenient for long-term use.[18 ] Hence, the description and grading of AF burden will likely change in the near future
as our knowledge on the association of outcomes with AF burden increases (for example,
a good correlation of AF burden with quality of life has been recently reported).[38 ]
The Substrate for AF (Su) domain pertains to the complexity of AF pathophysiology, including simple clinical characteristics
such as patient age, cardiovascular risk factors (e.g., obesity), and underlying comorbidities,
as well as the presence and extent of left atrial enlargement, impaired atrial function,
and fibrosis of the atrial myocardium, all of which have been shown to play a role
in the development and progression of AF.[39 ]
Diagnostic assessment for the presence of cardiovascular risk factors and underlying
comorbidities is not only a routine part of comprehensive clinical evaluation of any
patient suspected of having a heart condition but is also highly relevant for treatment
decisions with regards to thromboprophylaxis and likelihood of successful rhythm control.
Of note, the cardiovascular risk factor burden closely correlates with the lifetime
risk of AF development,[40 ] whereas optimal management of modifiable cardiovascular risk factors and comorbidities
has been associated with a reduction in AF burden.[41 ] Indeed, the structured characterization of AF using the 4S-AF scheme would prompt
practicing physicians to identify and manage these risk factors, whereas the acknowledgment
of multimorbidity that is included in the 4S-AF would influence the arrhythmia-related
treatment decisions.
Atrial structural and functional remodeling predisposing to or resulting from AF is
an important indicator of substrate complexity and correlates well with the outcome
of AF-directed treatment interventions.[42 ] Of note, left atrial size and function generally correlate well with cardiovascular
outcomes including all-cause mortality.[43 ]
[44 ] Transthoracic echocardiography is widely available in routine clinical practice
and provides basic information on the atrial size and function, whereas more sophisticated
assessment including advanced transthoracic echocardiography, transesophageal echocardiography,
cardiac computed tomography, or nuclear magnetic resonance imaging provides additional
indices of atrial dysfunction and structural alterations including fibrosis[42 ] and epicardial fat that have both treatment and prognostic implications and may
inform expert decision-making (e.g., choosing the appropriate ablation strategy).[45 ] The rapidly advancing technologies such as machine learning may also enhance the
characterization of correlates for atrial structure in the risk assessment of AF recurrence.[46 ]
With increasing evidence about the clinical application of the recently proposed concept
of atrial cardiomyopathy,[47 ] the assessment, classification, and staging of the atrial disease may become the
cornerstone of the Su domain in the 4S-AF scheme.
The Use of the 4S-AF Scheme for Characterizing AF Patients—Hypothetical Examples in
Clinical Practice
The 4S-AF characterization of AF patients using the descriptors obtained by routine
diagnostic assessment ([Figs. 1 ] and [3 ]) would provide the basis for treatment decision-making supporting optimal management
of AF by primary care physicians, internal medicine specialists, or general cardiologists,
also facilitating optimal referral for expert consultation where needed ([Fig. 3 ]), whereas further refinement of AF characterization using advanced diagnostic imaging
tools would facilitate expert decision-making.
Importantly, the 4S-AF characterization of AF patients must be accompanied by information
on the patient age and comorbidities and personal preferences.
Example 1 : The 4S-AF characterization is St = 0, Sy = 2, Sb = 3, Su = 0, and the medical report describes a female patient aged 59 years, with no cardiovascular
or other comorbidity besides AF.
It is immediately obvious that this patient does not need long-term OAC (low stroke
risk) but is eligible for rhythm control including consideration for AF ablation as
a first-line treatment owing to severe symptoms, high AF burden, and low substrate
complexity.[6 ]
Warning: this patient would still need OAC before and after AF ablation, as recommended,
and her 4S-AF status should be reassessed regularly in order not to miss dynamic changes
in her risk profile over time.[6 ]
Example 2 : The 4S-AF characterization is St = 1, Sy = 0, Sb = 3, Su = 4. As per the medical report, it is a male patient, 79 years old, with prior stroke,
myocardial infarction, diabetes mellitus, and hypertension, whereas transthoracic
echocardiogram showed preserved left ventricular function and a left atrial anteroposterior
diameter of 48 mm with mild mitral regurgitation.
We can easily appreciate that this patient has a high risk of stroke, mild symptoms,
high AF burden (i.e., long-standing persistent or permanent AF; see [Fig. 3 ]), and significant substrate for AF. Hence, the patient needs lifelong OAC, and rate
control may be the ultimate solution, which should be discussed with the patient weighing
the risk factors for AF recurrence. Again, the 4S status of this patient should be
reviewed regularly. Also, the patient should be routinely monitored for adherence
to treatment, the occurrence of AF-related complications (e.g., stroke, bleeding,
heart failure), and comorbidities.[6 ]
Example 3 : A 63-year-old female patient is referred for expert consultation for further AF
management after electrical cardioversion failure in the local hospital. Her 4S-AF
characterization is St = 1, Sy = 2, Sb = 3, Su = 5. She is obese, has hypertension, bilateral aortic-femoral graft, and chronic
obstructive pulmonary disease, and severely increased left atrium volume (42 mL/m2 ) on transthoracic echocardiographic examination. The 12-lead ECG shows AF with a
ventricular rate of 125 beats per minute.
The patient is highly symptomatic but Su = 5 suggests advanced substrate for AF. Indeed, her symptoms could be attributed
to poor ventricular rate control and possible exacerbation of pulmonary obstruction.
Hence, she would be first assigned to optimization of rate control, treatment of pulmonary
obstruction, and lifestyle and risk factor modifications including weight reduction,
all of which could have been initiated in the local health care center.[48 ] Should she remain highly symptomatic after optimization of her medical condition
and weight control, an advanced substrate evaluation using additional imaging could
be undertaken to reconsider the choice between rate or rhythm control, but it is less
likely that she would be scheduled for left atrial ablation.
