1
Introduction and definition of atrial cardiomyopathy
The atria provide an important contribution to cardiac function [1], [2]. Besides
their impact on ventricular filling, they serve as a volume reservoir, host pacemaker
cells and important parts of the cardiac conduction system (e.g. sinus node, AV node),
and secrete natriuretic peptides like atrial natriuretic peptide (ANP) and brain natriuretic
peptide (BNP) that regulate fluid homoeostasis. Atrial myocardium is affected by many
cardiac and non-cardiac conditions [3] and is, in some respects, more sensitive than
ventricular [4]. The atria are activated, besides the three specialised intermodal
tracts [5], [6], through working cardiomyocytes, so that any architectural or structural
change in the atrial myocardium may cause significant electrophysiological disturbances.
In addition, atrial cells (both cardiomyocytes and non-cardiomyocyte elements like
fibroblasts, endothelial cells, and neurons) react briskly and extensively to pathological
stimuli [3] and are susceptible to a range of genetic influences [7]. Responses include
atrial cardiomyocyte hypertrophy and contractile dysfunction, arrhythmogenic changes
in cardiomyocyte ion-channel and transporter function, atrial fibroblast proliferation,
hyperinnervation, and thrombogenic changes [2]. Thus, atrial pathologies have a substantial
impact on cardiac performance, arrhythmia occurrence, and stroke risk [1], [8].
Ventricular cardiomyopathies have been well classified; however, a definition and
detailed analysis of ‘atrial cardiomyopathy’ is lacking from the literature. The purpose
of the present consensus report, prepared by a working group with representation from
the European Heart Rhythm Association (EHRA), the Heart Rhythm Society (HRS), the
Asian Pacific Heart Rhythm Society (APHRS), and Sociedad Latino Americana de Estimulacion
Cardiaca y Electrofisiologia (SOLAECE), was to define atrial cardiomyopathy, to review
the relevant literature, and to consider the impact of atrial cardiomyopathies on
arrhythmia management and stroke.
1.1
Definition of atrial cardiomyopathy
The working group proposes the following working definition of atrial cardiomyopathy:
‘Any complex of structural, architectural, contractile or electrophysiological changes
affecting the atria with the potential to produce clinically-relevant manifestations’
(Table 1).
Many diseases (like hypertension, heart failure, diabetes, and myocarditis) or conditions
(like ageing and endocrine abnormalities) are known to induce or contribute to an
atrial cardiomyopathy. However, the induced changes are not necessarily disease-specific
and pathological changes often share many similarities [9], [10]. The extent of pathological
changes may vary over time and atrial location, causing substantial intraindividual
and interindividual differences. In addition, while some pathological processes may
affect the atria very selectively (e.g. atrial fibrillation-induced remodelling),
most cardiomyopathies that affect the atria also involve the ventricles to a greater
or lesser extent. There is no presently accepted histopathological classification
of atrial pathologies. Therefore, we have proposed here a working histological/ pathopysiological
classification scheme for atrial cardiomyopathies (Table 1; Fig. 1). We use the acronym
EHRAS (for EHRA/HRS/ APHRS/SOLAECE), defining four classes: (I) principal cardiomyocyte
changes [11], [12], [13], [14], [15]; (II) principally fibrotic changes [10], [14],
[16]; (III) combined cardiomyocyte-pathology/fibrosis [9], [11], [12]; (IV) primarily
non-collagen infiltration (with or without cardiomyocyte changes) [17], [18], [19].
This simple classification may help to convey the primary underlying pathology in
various clinical conditions. The EHRAS class may vary over time and may differ at
atrial sites in certain patients. Thus, this classification is purely descriptive
and in contrast to other classifications (NYHA class, CCS class etc.), there is no
progression in severity from EHRAS class I to EHRAS IV (Table 2). The classification
may be useful to describe pathological changes in biopsies and to correlate pathologies
with results obtained from imaging technologies etc. In the future, this may help
to define a tailored therapeutic approach in atrial fibrillation (AF) (Fig. 1, Fig.
2, Fig. 3).
2
Anatomical considerations and atrial muscular architecture
2.1
Normal atrial structures
2.1.1
Gross morphology
Each atrium has a morphologically characteristic atrial body and appendage (Fig. 4).
In the body, there is a venous component with the orifices of the systemic or pulmonary
veins (PVs) and a vestibular component that surrounds the atrial outlet [20]. The
interatrial septum (IAS) separates the atrial bodies. The venous component of the
left atrium (LA) is located posterosuperiorly and receives the PVs at the four corners,
forming a prominent atrial dome. The LA is situated more posteriorly and superiorly
than the right atrium separated by the obliquity of the plane of the IAS [21].
The LA appendage (LAA) is smaller than the right atrium appendage (RAA). Narrower
and with different shapes has a distinct opening to the atrial body and overlies the
left circumflex coronary artery. Its endocardial aspect is lined by a complex network
of muscular ridges and mem-branes [22], [23]. Different LAA morphologies have been
described, and it appears that LAA morphology correlates with the risk of thrombogenesis
[24].
Bachmann׳s bundle is a broad epicardial muscular band running along the anterior wall
of both atria (Fig. 4). The rightward arms extend superiorly towards the sinus node
and inferiorly towards the right atrioventricular groove, while the leftward arms
blend with deeper myofibres to pass around the neck of the LAA and reunite posteriorly
to join the circumferential vestibule of the LA. The walls of LA are non-uniform in
thickness (1–15 mm) and thicker than the right atrium [25].
2.2
Normal atrial myocardium
2.2.1
Atrial cardiomyocytes
Atrial cardiomyocytes are geometrically complex cylinders that sometimes bifurcate
at their ends where they connect with adjacent fibres via band-like ‘intercalated
discs’. This contractile syncytium is organised in well-defined bands that establish
non-uniform anisotropic propagation of the atrial impulse [9], [11], [26]. The only
clear light-microscopic morphological difference between atrial and ventricular cardiomyocytes
is in size [27]. In paraffin-embedded human specimens, the cardiomyocyte transverse
diameter is ×12 mm in the LAs vs. 20–22 mm in the ventricles [11], [28]. Atrial cardiomyocytes
are mainly mononucleated; a minor fraction possess two or more nuclei. The nucleus
is usually centrally located, with granular and/or condensed chromatin. The nuclear
shape is influenced by fibre contraction, becoming more fusiform with longitudinal
cell stretch [29]. Biochemically, atrial cardiomyocytes have greater lipid content
than ventricular muscle cells [30].
Atrial cardiomyocytes share many characteristics with ventricular in terms of nucleus,
contractile apparatus, cytoskeleton, and organelles [27], [29], [31], [32]. Unlike
ventricular cardiomyocytes, atrial cardiomyocytes do not possess an extensive T-tubule
network but they do have prominent sarcoplasmic reticulum (SR) elements known as Z-tubules
[33]. Therefore, the atrial sarcolemma does not protrude into the cell, and voltage-operated
Ca2+ channels mainly function at the cell periphery [34]. Atrial cardiomyocytes display
specific granules (100–400 nm) situated mainly in the paranuclear area adjacent to
the Golgi apparatus, which contain ANP, the BNP, and related peptides [23], [24].
2.2.2
Atrial interstitium
Atrial interstitium consists of cellular and extracellular components (see Fig. 2,
Fig. 3, Fig. 4, Fig. 5). The cellular elements include fibroblast/myofibroblasts,
adipocytes, undifferentiated mesenchymal cells, and isolated inflammatory cells. The
atrial wall has a significant number of medium-sized blood vessels, especially in
the sub-epicardium. Mature adipose tissue is frequently found in atrial myocardium,
especially the epicardium, and often permeates the layers around intramural coronary
branches. The number of adipocytes is highly variable and increases with age [27].
The extracellular components consist of collagen fibres, which form most of the myocardial
skeleton, proteoglycan particles, lipidic debris, spherical micro-particles, and matrix
vesicles [27].
Collagen fibres, mainly type I, are both normal and essential components (Fig. 1,
Fig. 2, Fig. 3, Fig. 4, Fig. 5). Atrial fibrous tissue may be sub-divided into pure
interstitial and perivascular (or adventitial). Interstitial collagen fibres represent
×5% of the atrial wall volume. The atrial myocardium is also the site of sparse postganglionic
nerve endings (from the ‘intrinsic cardiac nervous system’), mostly within discrete
fat pads but also among cardiomyocytes [35].
3
Atrial-specific physiological and functional considerations
3.1
Atrial-selective electrophysiological properties
The atria have a number of electrophysiological features that distinguish them from
the ventricles and govern their arrhythmia susceptibility.
3.2
Action potential/ion-channel properties
Atrial cardiomyocytes have distinct action potential (AP) properties from ventricular
cardiomyocytes, resulting in a large part from distinct ion-channel properties and
distribution (Fig. 6A) [36], [37]. Atrial background inward-rectifier K+ current (IK1)
is smaller than that of ventricular K+ current, resulting in a less negative resting
potential and more gradual slope of phase-3 repolarization. Atrial cells also have
two K+-currents that are absent in ventricle cells: the ultrarapid delayed rectifier
current (I
Kur) and the acetylcholine-regulated K+-current (I
KACh). In addition, there is evidence that atrial Na+-current has different properties
compared with ventricular current [38]. As well as distinctions between atrial and
ventricular APs, different atrial regions may have discrete AP and ion-channel properties
[37], [39]. These cellular electrophysiological characteristics have implications
for antiarrhythmic drug action and design, and may also affect the responses to atrial
arrhythmias and disease [36], [37].
3.3
Intercellular coupling properties
The atria have a different pattern of cell-to-cell coupling protein (connexin) distribution
compared with ventricular myocardium [36]. Whereas working ventricular cardiomyocytes
express connexin-43 exclusively, atrial cardiomyocytes have significant expression
of connexin-40 (Fig. 6B) [36]. Heterogeneities in connexin-40 distribution are common
in paroxysmal AF and may play a pathophysiological role [40], and gene variants affecting
connexin-40 sequence and/or transcription predispose to AF occurrence [41].
3.4
Atrial structural properties
The atria have a very complex 3D structure (Fig. 6C) not found in the ventricles.
These include interatrial connections limited to Bachmann׳s bundle, the septum, and
the CS; pectinate muscles, the crista terminalis, and fibres surrounding the coronary
sinus in the right atrium; and the PVs with complex fibre orientation around them
in the LA. These structural complexities have important potential implications for
atrial patho-physiology and management of atrial arrhythmias [42]. Extensive recent
work has gone into the realistic mathematical reconstruction of such geometric complexities
[43], and they have been incorporated into analytical approaches designed to implement
patient-specific arrhythmia therapies [44]. Cable-like strands of atrial tissue like
the pectinate muscles and crista terminalis are organised such that conduction within
them is primarily longitudinal, with an ‘anisotropy ratio’ (longitudinal/transverse
conduction velocities) as great as 10, whereas in working ventricular muscle the ratio
is typically more between 2 and 4 [45].
3.5
Autonomic ganglia
There are autonomic ganglia on the surface of the heart that are important way-stations
for cardiac autonomic control [46]. Moreover, alterations in local cardiac innervation
and intracardiac autonomic reflexes play an important role in physiology and arrhythmia
control. Most of the cardiac autonomic ganglia are located on the atria, and in particular
in the region of the PV ostia. Thus, they are well positioned to affect atrial electrical
activity in regions particularly important in AF, and their alteration by therapeutic
manoeuvers like PV ablation may contribute to antiarrhythmic efficacy [42], [46],
[47].
3.6
Left atrium mechanics
The left atrial contribution to overall cardiovascular performance is determined by
unique factors. First, left atrial function critically determines left ventricular
(LV) filling. Second, chamber-specific structural, electrical and ion remodelling
alter left atrial function and arrhythmia susceptibility. Third, atrial function is
highly relevant for the therapeutic responses of AF. Fourth, LA volume is an important
biomarker that integrates the magnitude and duration of LV diastolic dysfunction.
The development of sophisticated, non-invasive indices of LA size, and function might
help to clinically exploit the importance of LA function in prognosis and risk stratification
[1], [48].
Fibre orientation of the two thin muscular layers (the fascicles of which both originate
and terminate at the atrioventricular ring) introduce a complexity that challenges
functional analysis. Ultrastructurally, atrial cardiomyocytes are smaller in diameter,
have fewer T-tubules, and more abundant Golgi apparatus than ventricular. In addition,
rates of contraction and relaxation, conduction velocity, and anisotropy differ, as
does the myosin isoform composition and the expression of ion transporters, channels,
and gap junctional proteins (see relevant sections).
3.7
Functions of the left atrium
The principal role of the LA is to modulate LV filling and cardiovascular performance
by operating as a reservoir for PV return during LV systole, a conduit for PV return
during early LV diastole, and as a booster pump that augments LV filling during LV
diastole. There is a critical interplay between these atrial functions and ventricular
systolic and diastolic performance. Thus, while LA compliance (or its inverse, stiffness),
and, to a lesser extent, LA contractility and relaxation are the major determinants
of reservoir function during LV systole, LV end-systolic volume and descent of the
LV base during systole are important contributors. Conduit function is also governed
by LA compliance and is reciprocally related to reservoir function, but because the
mitral valve is open in diastole, conduit function is also closely related to LV compliance
(of which relaxation is a major determinant). Atrial booster-pump function reflects
the magnitude and timing of atrial contractility, but also depends on venous return
(atrial preload), LV end-diastolic pressures (atrial afterload), and LV systolic reserve.
3.7.1
Left atrium booster-pump function
Left atrium booster-pump function represents the augmented LV-filling resulting from
active atrial contraction (minus retrograde blood-ejection into the PVs) and has been
estimated by measurements of (i) cardiac output with and without effective atrial
systole, (ii) relative LV-filling using spectral Doppler of transmitral, PV, and LA-appendage
flow, (iii) LA-shortening and volumetric analysis, and (iv) tissue Doppler and deformation
analysis (strain and strain-rate imaging) of the LA-body [1]. Booster-pump function
can also be evaluated echocardiographically by estimating the kinetic energy and force
generated by LA contraction. The relative importance of the LA contribution to LV
filling and cardiac output remain controversial. A load-independent index of LA contraction
based on the analysis of instantaneous relation between LA pressure and volume, analogous
to LV end-systolic elastance measurements, has been used as a load-independent measure
of LA pump function, validated ex vivo and in the intact dog (Fig. 7) [49]. While
LA pressure–volume loops can be generated with invasive and semi-invasive means in
humans [50], these methods are cumbersome, time-consuming, and difficult to apply.
Measurement of myocardial strain and strain rate, which represent the magnitude and
rate of myocardial deformation, assessed using either tissue Doppler velocities (tissue
Doppler imaging, TDI) or by 2D echocardiographic (2D speckle-tracking or STE) techniques
(Fig. 8) provide objective, non-invasive measurements of LA myocardial performance
and contractility that overcome these limitations [1], [51].
3.7.2
Left atrium reservoir function
Nearly half of the LV stroke volume and its associated energy are stored in the LA
during LV systole. This energy is subsequently expended during the LV diastole. Reservoir
function is governed largely by atrial compliance during ventricular systole, which
is measured most rigorously by fitting atrial pressures and dimensions, taken either
at the time of mitral valve opening/closure over a range of atrial pressures and volumes
or during ventricular diastole, to an exponential equation [52]. Although this method
requires atrial dimensions and pressures, the relative reservoir function can be estimated
simply with PV Doppler: the proportion of LA inflow during ventricular systole provides
an index of the reservoir capacity of the atrium. Reservoir function can also be estimated
from LA time–volume relations as either the total ejection fraction or distensibility
fraction, calculated as the maximum minus minimum LA volume, normalised to maximal
or minimal LA volume, respectively.
Although largely neglected, the LA–appendage is more compliant than the LA–body [52],
so the contribution of the appendage to overall LA compliance is substantial with
potential negative implications for routine atrial appendectomy/ligation during mitral
valve surgery.
Left atrium strain and strain rates during LV systole predict successful sinus rhythm
restoration following DC cardioversion or AF ablation, and are surrogates of atrial
fibrosis and structural remodelling; coupled with an estimate of atrial pressure (e.g.
transmitral E/E′), strain has the potential to estimate atrial distensibility non-invasively
[1], [53].
3.7.3
Left atrium conduit function
Left atrium conduit function occurs primarily during ventricular diastole and represents
the trasport of blood volume that cannot be attributed to either reservoir or booster-pump
functions, accounting for approximately one-third of atrial flow [54]. A reciprocal
relation exists between LA conduit and reservoir functions; a redistribution between
these functions is an important compensatory mechanism that facilitates LV filling
with myocardial ischaemia, hypertensive heart disease, and mitral stenosis (MS). Conduit
function is estimated by the early diastolic transmitral flow, diastolic PV-flow,
and LA strain and strain rate during early diastole.
3.8
Atrial-selective Ca [21] handling
There are major differences in the expression and function of Ca2+-handling proteins
between atria and ventricles (Fig. 9) [55]. The atria have reduced cardiomyocyte contraction
and relaxation times and shorter Ca2+-transient duration [56], [57], [58]. In atria,
protein levels [57], [59] and activity [57], [59] of the SR Ca2+-ATPase2a (Serca2a)
are two-fold higher, whereas the Serca2a-inhibitor phospholamban (PLB) is less abundant,
vs. ventricles [57], [59]. Atrial, but not ventricular, Serca2a is also regulated
by sarcolipin (SLN) and SLN ablation increases atrial SR Ca2+-uptake and contractility
[60]. L-type Ca2+-current [61] is similar in both chambers, whereas protein levels
of ryanodine receptor type-2, calsequestrin, triadin, junction and Na2+ –Ca2+ exchanger
are lower in atria than in ventricles [59], [62], [63]. In contrast to ventricular
myocardium, T-tubules are less abundant in atrial cardiomyocytes [64]. In addition,
atrial cardiomyocytes possess much more Ca2+-buffering mitochondria than ventricular
cardiomyocytes [56]. As a consequence, the atrial Ca2+ wave starts in the myocyte
periphery and then propagates to the centre of the myocyte, activating additional
Ca2+-releasing sites in the SR [55].
4
Pathology of atrial cardiomyopathies
4.1
Lone atrial fibrillation (atrial fibrillation without concomitant conditions)
‘Lone’ atrial fibrillation (LAF) is diagnosed when no apparent explanation or underlying
comorbidity can be identified [65], [66]. Over the last few years, new epidemiological
associations with AF have emerged and the number of true LAF cases has progressively
decreased [67]. Like AF associated with comorbidities, LAF occurs more frequently
in males than in females with a ratio of 3–4:1 [68]. Recent studies have shown that
true cases of LAF can be diagnosed even in subjects older than 60 years, so that this
age limit seems inappropriately conservative [69]. At the same time, it is unclear
whether cases with left atrial enlargement should be excluded from the LAF category.
In fact, LA enlargement might even be the consequence of the arrhythmia [70].
‘Lone’ atrial fibrillation is at the lower end of the thromboembolic risk spectrum,
with only a 1–2% cumulative 15-year risk of stroke [66]. How-ever, with ageing and/or
the occurrence of cardiovascular comorbidities, the risk of AF-related complications
(including thromboembolic events) increases [71]. Patients originally diagnosed with
LAF may follow different clinical courses based on their left atrial volume: individuals
who retain normal LA size throughout long-term follow-up show a relatively benign
course, while those with LA enlargement experience adverse events like stroke, myocardial
infarction, and heart failure [72]. The majority of LAF patients first present with
paroxysmal episodes and show low progression rates into permanent AF [71], [73].
Atrial fibrillation has clear genetic determinants [7]. These include common gene
variants with low predictive strength and rare gene mutations that have much greater
penetrance [7].
Frustaci et al [14]. explored the histological morphology of right atrial septal biopsies
from patients with lone paroxysmal AF, finding chronic inflammatory infiltrates, foci
of myocyte necrosis, focal replacement fibrosis, and myocyte cytoplasmic vacuoles
consistent with myolysis. Of their 12 patients, 10 showed EHRAS class III changes
and 2 showed EHRAS class II. Stiles et al [74]. found bi-atrial structural change,
conduction abnormalities, and sinus node dysfunction in paroxysmal LAF patients. Skalidis
et al [75]. demonstrated atrial perfusion abnormalities and coronary flow reserve
impairment. Much more recently, morphometric assessment of atrial biopsies from the
LA posterior wall of persistent or long-lasting persistent LAF patients demonstrated
cardiomyocyte hypertrophy, myolytic damage, interstitial fibrosis, and reduced connexin-43
expression vs. controls [76].
4.2
Isolated atrial amyloidosis
The accumulation of insoluble, misfolded proteins is linked to an increasing number
of age-related degenerative diseases [77]. Amyloidosis represent the deposition of
insoluble, fibrillar proteins in a cross b-sheet structure that characteristically
binds dyes such as Congo red. The most common form of age-related or senile amyloidosis
is limited to the atrium, a condition known as isolated atrial amyloidosis (IAA) [17],
[78]. The incidence of atrial amyloidosis increases with age, exceeding 90% in the
ninth decade [79]. Isolated atrial amyloidosis is also linked to structural heart
disease. In atrial biopsies from 167 patients undergoing cardiac surgery, 23 of 26
amyloid-positive specimens were from patients with rheumatic heart disease (RHD),
while the remaining 3 came from patients with atrial septal defects [80]. The overall
incidence of 16% was greater than that was seen in control atrial autopsy specimens
from trauma victims (3%). Histologically, IAA is classified as EHRAS IVa (Fig. 3;
Table 2). Atrial natriuretic peptide is a fibrillogenic protein that forms IAA [81].
Amyloid deposits are immunoreactive for ANP in most patients [17], while transthyretin,
a transport protein implicated in systemic senile amyloidosis, was also identified
in 10% [4] (NT-pro-ANP has been identified in other studies [82]). As with fibrosis,
amyloidosis can cause local conduction block and P-wave duration is increased in IAA.
Atrial amyloid is found more commonly in patients with AF vs. sinus rhythm (Fig. 3).
Both AF and IAA increased with advancing age and female sex, but the relationship
between the two is independent of age and gender [83], [84]. Isolated atrial amyloidosis
is detected in 80% of PV sleeves of elderly patients [84]. For organ-specific amyloidosis
such as Alzheimer׳s disease, there is no detectable correlation between quantity of
fibrillar deposits and disease advancement [85]. Rather, disease phenotype correlates
most closely with accumulation of soluble, prefibrillar protein aggregates [86]. Preamyloid
oligomers (PAOs) are cytotoxic to cardiomyocytes [87]. They do not bind Congo red
and thus are not visible by standard amyloid staining methods. Using a conformation-specific
antibody, PAOs often co-localising with ANP were detected in atrial samples of 74
of 92 patients without AF undergoing cardiac surgery [88]. The preamyloid oligomer
content was independently associated with hypertension. Additional studies are needed
to further confirm this association and whether PAOs are increased in AF.
4.3
NPPA mutations
Atrial natriuretic peptide is released from the atria in response to atrial stretch
or volume expansion, and produces natriuresis, diuresis, and vasodilation [89]. It
also interacts with other endogenous systems, inhibiting the renin–angiotensin-II–aldosterone
and sympathetic nervous systems, and regulates ion currents [90], [91]. Atrial natriuretic
peptide-knockout mice develop cardiac hypertrophy and exaggerated responses to hypertrophic
stress [92]. The gene encoding the precursor protein for ANP, NPPA, encodes prepro-ANP,
a 151 amino acid protein that includes a signal peptide cleaved off to form pro-ANP
[93], which is stored in dense granules in the atria. Released pro-ANP undergoes proteolytic
processing to generate N-terminal pro-ANP and ANP, 98 and 28 amino acids in length,
respectively. N-terminal pro-ANP is cleaved into three hormones with biological activity
similar to ANP: long-acting natriuretic hormone (LANH), vessel dilator peptide, and
kaliuretic hormone.
Genetic studies have linked abnormal ANP production to familial atrial tachyrrhythmias
and atrial cardiomyopathy. In a large family with Holt–Oram syndrome, a missense mutation
in T-box transcription factor 5 (TBx5) resulted in an atypical phenotype with early-onset
AF and the overexpression of multiple genes, including NPPA [94]. In a large family
with multiple members having early-onset LAF, a 2-bp deletion was identified that
abolishes the ANP stop codon, producing a mature protein containing the usual 28 amino
acids plus an anomalous C-terminus of 12 additional residues [95]. The mutant ANP
peptide is present in affected family members at plasma concentrations 5–10 times
higher than wild-type ANP. Studies of the electrophysiological effects of ANP have
been inconsistent [96].
Additional NPPA variants (S64R and A117V) have also been linked to AF [97], [98].
The S64R variant occurs in vessel dilator peptide rather than ANP. A truncated peptide
containing this mutation increased IKs several fold, an effect predicted to shorten
action potential duration (APD) [97], but the variant has also been identified in
unaffected elderly individuals without AF [96], and its functional pathological significance
remains uncertain.
More recently, an autosomal-recessive atrial cardiomyopathy was described in patients
harbouring an NPPA mutation (Arg150Gln) predicted to be damaging to protein structure
[99]. The phenotype is characterised by biatrial enlargement, initially associated
with atrial tachyarrhythmias such as AF and atrial flutter [100]. Biatrial enlargement
progresses to partial and ultimately severe atrial standstill, associated with progressive
decreases in atrial voltage and extensive atrial scarring. Whether atrial structural
changes are primary, or secondary to atrial enlargement, is unknown. Loss of the antihypertrophic
effects of ANP may cause the massive atrial enlargement seen in these patients.
4.4
Hereditary muscular dystrophies
A common finding in many inherited muscular dystrophies is cardiac involvement, related
to myocyte degeneration with fatty or fibrotic replacement (Table 3) [101], [102],
[103]. In some cases, this can be the presenting or predominant clinical manifestation.
Multiple complexes and pathways are involved in the maintenance of myocyte integrity,
and a defective or absent protein component can lead to progressive cell death. The
large dystrophin–glycoprotein complex links the myocyte cytoskeleton to the extracellular
basement membrane. For diseases of dystrophin, sarcoglycans, and other complex-related
proteins, the most prominent manifestation is a dilated cardiomyopathy due to diffuse
myocyte involvement, with arrhythmias and conduction abnormalities secondary to LV
dysfunction [101], [102], [103], [104], [105]. Specific atrial involvement can lead
to sinus node disease and/or atrial arrhythmias with associated thromboembolic events
[106], [107]. Myotonic dystrophy type I is the most common muscular dystrophy presenting
in adults [108]. Up to 15% develop atrial arrhythmias during a 10-year follow-up [109].
The presence of conduction defects and atrial arrhythmias are independent risk factors
for sudden death [103], [110]. In Emery-Dreifuss and Limb-Girdle type IB disease,
widespread atrial fibrosis can lead to atrial standstill [101]. In Emery-Dreifuss,
AF and atrial flutter with slow ventricular responses and asystolic pauses can be
observed, coupled with the occurrence of thromboembolism and stroke [111]. In facioscapulohumeral
muscular dystrophy, arrhythmias are rare, with the most common being supraventricular
tachycardia [112]. Histologically, the tissue composition may vary substantially,
including all EHRAS classes (see Table 2).
4.5
Atrial cardiomyopathy due to congestive heart failure
Congestive heart failure (CHF) is a common cause (contributing condition) of AF [3].
The CHF-induced atrial phenotype is complex. A particularly important component is
atrial fibrosis, which in experimental models occurs earlier in the course of CHF,
and to a much greater extent, than in the ventricles, at least in part because of
atrial-ventricular fibroblast–phenotype differences [4]. Congestive heart failure-related
fibrosis slowly, if at all, and the AF-promoting substrate predominantly tracks fibrosis
rather than other components of atrial remodelling like ion-current or connexin changes.
Unlike the case for AF-induced remodelling, the atrial ion-current changes in CHF
do not abbreviate APD or cause overall conduction slowing [113], [114], so they do
not contribute directly to arrhythmogenesis. On the other hand, CHF atria are prone
to triggered activity due to abnormal Ca2+ handling [115]. The principle underlying
abnormality appears to be increased cellular Ca2+ load. While the underlying mechanisms
are not completely clear, they likely include phospholamban hyperphosphorylation (which
increases SR Ca2+ up-take) and AP prolongation (which increases Ca2+ loading by enhancing
the period during which L-type Ca2+ channels are open). The final phenotypic product
of the CHF-induced Ca2+-handling abnormalities is focal ectopic activity due to aberrant
diastolic Ca2+-release events from the SR, similar to abnormalities seen with paroxysmal
and long-standing persistent AF [116].
Congestive heart failure also causes atrial hypocontractility, despite increased cytosolic
Ca2+ transient, indicating reduced contractile sensitivity to intracellular Ca2+,
possibly because of reduced expression of total and phosphorylated myosin-binding
protein C [115]. This hypocon-tractility may be important in contributing to the increased
likelihood of thromboembolic events in AF patients who also have CHF. Of the atrial
changes that occur in CHF, many are also seen in the ventricle. However, the highly
atrial-selective fibrosis may contribute to atrial cardiomyopathy in the absence of
clear signs of disturbed ventricular function, particularly in patients with prior
CHF events who later become well-compensated under therapy or after resolution of
the underlying cause. Collagen depositions are prominent in CHF, leading most commonly
to EHRAS Class II and III properties. However, EHRAS Class IVi and IVf may also be
found in certain areas of the atria (see Table 2).
4.6
Obstructive sleep apnoea
Obstructive sleep apnoea (OSA) is known to impair cardiac function and predispose
to AF [117], [118], [119]. Obstructive sleep apnoea prolongs atrial conduction times,
slows atrial conduction, reduces atrial-electrogram voltages and increases electrogram
complexity [117], [118]. Signal-averaged P-wave duration is increased by OSA, and
decreases significantly with continuous positive airway pressure treatment [120].
In a rat model, repeated obstructive apnoea over a 4-week period increases AF vulnerability
and slows atrial conduction by altering connexin-43 expression and inducing atrial
fibrosis [121].
4.7
Atrial fibrillation-induced atrial remodelling
Atrial fibrillation itself induces atrial remodelling that contributes to the maintenance,
progression, and stabilisation of AF [41], [116]. The high atrial rate causes cellular
Ca2+ loading. This induces a decrease in ICa,L due to down-regulation of the underlying
Cav1.2 subunits, and an increase in constitutively active IK,Ach
[41], [116], [122], [123] MiR-328 up-regulation with consequent repression of Cav1.2-translation
and Ca2+ -dependent calpain acti-vation, causing proteolytic breakdown of L-type Ca2+
channels [41], [116]. The rate-dependent up-regulation of IK1 results from a Ca2+/calcineurin/NFAT-mediated
down-regulation of the inhibitory miR-26, removing translational–inhibition of Kir2.1
[41], [116]. Increased IK1 stabilizes AF by abbreviating and hyperpolarizing atrial
cardiomyocyte Aps [41]. Small-conductance Ca2+-activated K+ (SK) currents (ISK) also
play a role in AF [41], [116]. Computational modelling shows that increased total
inward-rectifier K+ current in chronic atrial fibrillation (cAF) is the major contributor
to the stabilisation of re-entrant circuits by shortening APD and hyperpolarizing
the resting membrane potential [41], [116].
Atrial tachycardia remodelling reduces Ca2+-transient amplitude by a variety of mechanisms,
contributing to atrial contractile dysfunction [41], [116], [124]. Reduced atrial
contractility causes atrial ‘stunning’ that may be involved in thromboembolic complications.
Long-term atrial tachycardia remodelling causes conduction slowing in several animal
models, at least partly due to INa down-regulaton [122]. Heterogeneously distributed
gap-junction uncoupling due to connexin remodelling likely contributes to atrial conduction
slowing [41], [116]. Heterogeneity in connexin-40 distribution correlates with AF
stability in goats with repetitive burst-pacing-induced AF [125]. Connexin-40 expression
decreases in the PVs of dogs with AF-related remodelling, possibly due to tachycardia-induced
connexin-degradation by calpains [41], [116].
Long-term atrial tachycardia/AF may itself cause atrial fibrosis that contributes
to long-term persistence [126]. Rapid atrial firing promotes fibroblast differentiation
to collagen-secreting myofibroblasts through autocrine and paracrine mechanisms [32].
Atrial tachycardia-induced NFAT-mediated decreases in fibroblast miR-26 may also contribute
to structural remodelling. Atrial fibroblasts have non-selective cation channels of
the transient receptor potential (TRP) family that carry Ca2+ into the cell; the increased
cell-Ca2+ then triggers increased collagen production. Since miR-26 represses TRPC3
gene expression, miR-26 reductions increase TRPC3 expression, promoting fibroblast
Ca2+ entry that causes proliferation/myofibroblast differentiation [127]. TRPM7 may
similarly contribute to fibrotic changes in AF [128].
APD shortening in cAF patients also results from increased inward-rectifier K+ currents
[129], both IK1 and a constitutive form of IK,Ach
[41], [116]. Agonist-activated IK,ACh is decreased in right atrium of AF patients
because of a reduction in underlying Kir3.1 and Kir3.4 subunits [129], whereas agonist-independent
current is increased [41], [116].
Atrial cardiomyocytes from patients with long-standing persistent AF show spontaneous
diastolic SR Ca2+-release events (SCaEs) and delayed after depolarizations (DADs)
[130]. CaMKII-dependent RyR2 hyperphosphorylation underlies the SR Ca2+ leak and SCaEs
[32], [106], [130]. Protein kinase A-dependent RyR2 hyperphosphorylation also occurs
[130], likely promoting the dissociation of the inhibitory FKBP12.6 subunit from the
RyR2 channel. Larger inward NCX current may also contribute to the stronger propensity
for DADs [130].
Although initial work pointed to unchanged INa or mRNA expression of the Nav1.5 a-subunit
in AF patients, recent studies reported reduced peak INa
[41], [116]. There is also evidence for increased INa,late, although its functional
consequences are less clear. Altered mRNA and protein levels of connexin-40/-43 may
also contribute to re-entry-promoting conduction abnormalities in cAF patients. Reduced
connexin-40 expression together with lateralization to the transverse cell membrane
may cause heterogeneous conduction [41], [116].
Overall, ion-channel changes contribute to AF stabilisation and early recurrence after
cardioversion. Ca2+-handling abnormalities are involved in atrial ectopy, and atrial
fibrosis is important in the progression of long-term persistent AF to resistant forms.
Atrial fibrillation-induced atrial myopathy has changes that depend on AF duration.
Very short-term AF produces no ultrastructural alterations, while AF lasting several
weeks causes EHRAS I alterations [13]. Long-term persistent AF produces EHRA III changes
[126].
4.8
Drug-related atrial fibrillation
A large number of drug classes have been associated with the induction of AF either
in patients without heart disease or in individuals with pre-existing cardiac disorders
(Table 4) [131], but drug-induced AF (DIAF) has received less attention than that
it might deserve. The overall incidence of DIAF is still unknown for several reasons:
(a) the evidence associating specific drugs with AF has largely been based on anecdotal
reports, with very few controlled prospective clinical trials, (b) DIAF is often paroxysmal
and documentation may be difficult/poor, (c) while DIAF is easily recognised if it
occurs just after i.v. drug administrations (e.g. adenosine or dobutamine), AF episodes
can be missed if they appear after multiple exposures (e.g. chemotherapy), (d) patients
often receive multiple drugs, making the specific culprit agent difficult to identify,
(e) with non-cardiovascular drugs, DIAF is often diagnosed by non-cardiologists, often
with an imprecise description of the arrhythmic event and clinical history [132].
Multiple mechanisms have been suggested to explain the pathogenesis of DIAF: (a) direct
atrial electrophysiological effects like abbreviated refractoriness, slowed conduction,
or triggered activity due to Ca2+ loading, (b) changes in autonomic tone, (c) myocardial
ischaemia, (d) direct myocardial damage and other mechanisms such as release of pro-inflammatory
cytokines, oxidative stress, hypotension, and electrolyte disturbances [131], [132].
In the majority of cases, DIAF is a benign self-limited disorder. However, DIAF may
be clinically serious in polymedicated patients with underlying comorbidities [132].
Discontinuation of the causative drug(s) usually leads to cardioversion in few minutes
or hours. When AF persists, treatment is similar to that of non-DIAF patients [133],
[134]. Because of the wide range of mechanisms by which drugs cause AF, the histological
changes associated with DIAF may vary substantially from EHRAS class I–IV (see Table
2 for reference). Future studies are warranted to assess specific effects of various
drugs on atrial tissue.
4.9
Myocarditis
Myocarditis refers to an inflammatory disease of the heart, which occurs as a result
of exposure to external triggers (e.g. infectious agents, toxins, or drugs) or internal
ones like autoimmune disorders [135], [136].
The incidence is difficult to ascertain since it depends on the diagnostic criteria.
A likely estimate is 8 to 10 per 100 000 population, representing the third leading
cause of sudden death after hypertrophic cardiomyopathy and coronary artery disease
[137]. In autopsy series, the prevalence of myocarditis varies from 2% to 42% in young
adults with sudden death [138], [139]. Biopsy demonstrates an inflammatory infiltrate
in 9–16% of patients with unexplained non-ischaemic dilated cardiomyopathy [140],
[141].
Myocarditis is defined by the ‘Dallas criteria’ as the presence of a myocardial inflammatory
infiltrate with necrosis and/or degeneration of adjacent cardiomyocytes of non-ischaemic
nature [142]. According to the type of inflammatory cell, myocarditis may be subdivided
into lymphocytic, eosinophilic, polymorphic, giant-cell myocarditis, and cardiac sarcoidosis
[136].
Atrial fibrillation is frequently part of the clinical presentation of myocarditis.
In 245 patients with clinically suspected myocarditis, AF occurred in about 30% [143].
Myocarditis with lone atrial involvement is rarly diagnosed [144], [145], [146]. This
may reflect the fact that atrial myocardium is not methodically sampled either at
autopsy or in routine endomyocardial biopsy. In most such cases, AF dominated the
clinical picture, suggesting a role for architectural remodelling that interferes
with atrial conduction [9], [147]. Giant-cell myocarditis is a distinct – and probably
autoimmune – myocarditis characterised by diffuse infiltration by lymphocytes and
numerous multinucleated giant-cells, frequent eosinophils, cardiomyocyte necrosis
and, ultimately, fibrosis. The natural course is often fulminant and mortality is
high if untreated. An isolated atrial variant of giant-cell myocarditis was first
reported in 1964 [148]. Since then, only a few cases have been described in the English
language literature. The atrial variant appears to have a more favourable course compared
with the classical form [149]. The atrial giant-cell myocarditis may represent a distinct
entity, potentially attributable to atrium-specific auto-antigens [150]. EHRAS Class
IVi is observed in patients with atrial myocarditis. As myocarditis persists and enters
a chronic phase, characteristics may change to EHRAS Class III (see Table 2).
4.10
Atrial cardiomyopathy associated with genetic repolarization disturbances
Atrial standstill, a severe form of atrial cardiomyopathy, is associated with combined
heterozygous mutations of SCN5A and Connexin-40 genes [151]. Gain-of-function mutations
in K+-channel subunits (e.g. KCNQ1, KCNH2, KCND3, and KCNE5) or loss-of-function mutations
in KCN5A have been identified in AF patients [152]. Thus, either gain or loss of K+-channel
function can cause AF, indicating that repolarization requires optimal tuning and
deficits in either direction can be arrhythmogenic. Recently, early repolarization
or J-wave syndrome has been associated with AF although, in middle-aged subjects,
early repolarization in inferior leads did not predict AF [153]. A gain-of-function
mutation in KCNJ8, encoding the cardiac Kir 6.1 (KATP) channel, is associated with
both increased AF susceptibility and early repolarization [154]. There is an established
association between atrial arrhythmias and primary ventricular arrhythmia syndromes,
which was first reported among conditions that manifest with obvious structural abnormalities
[155]. Atrial fibrillation is relatively common in hypertrophic cardiomyopathy (prevalence×20%)
[156]. In arrhythmogenic right ventricular cardiomyopathy, an even higher proportion
(up to 40%) of patients may manifest AF [157]. The association with AF also extends
to primary arrhythmia syndromes without obvious structural heart disease. Supraventricular
tachycardias, primarily AF/AFl, have been reported in Brugada syndrome [158], [159].
Among long QT syndrome (LQTS) patients, prolongation of action potentials leading
to atrial fibrillation has been suggested to be an atrial form of ‘torsades de pointes’
[152]. A subtle form of ‘cardiomyopathy’ that includes increased left atrial volumes
occurs in ×12% of LQTS patients [160]. The reports available mostly implicate genetic
variants in Na+-channel genes [161]. Patients with early-onset lone AF have a high
prevalence of LQTS-associated SCN5A variants [162]. A mouse model of LQT3 is prone
to atrial arrhythmias due to EADs [163]. There are sporadic reports of atrial arrhythmias
in patients with CPVT [164]. Taken together, the associations between AF and sudden
death syndromes likely reflect common mechanisms between atrial and ventricular arrhythmogenesis.
4.11
Ageing
In elderly dogs, premature impulses show markedly slowed conduction, associated with
a doubling of fibrous-tissue content APD prolongation and spatial heterogeneity in
repolarization [165], [166]. Clinical mapping studies have also demonstrated similar
findings of conduction abnormalities, prolonged refractoriness, reduced myocardial
voltage, and a greater number of double potentials and fractionated electrograms [167],
[168]. Perhaps as a result of these atrial changes, alteration of wavefront propagation
velocities has been described with an inverse correlation to age [169]. Histologically,
fibrotic changes are the most obvious alteration (EHRAS Class II; see Table 2).
4.12
Hypertension
Hypertension accounts for at least one in five incident AF cases [170]. In hypertensive
subjects, both left atrial enlargement and P-wave changes are predictive of AF occurrence
[171], [172].
In small animal models, mimicking hypertension by partial aortic clamping induces
LA hypertrophy, fibrosis, connexin-43 down-regulation and slow/inhomogeneous conduction
[173]. Prenatal corticosteroid exposure-induced hypertension in sheep causes atrial
conduction abnormalities, wavelength shortening, and increased AF [174]. Lau et al.
utilised a one-kidney one-clip model to investigate the impact of short- and long-term
hypertension on the evolution of an atrial cardiomyopathy [175], [176]. Utilisation
of this model intrinsically is more reflective of a disordered renin–angiotensin axis.
Short-term hypertension progressively enlarged the LA, reduced LA emptying fraction,
prolonged atrial refractoriness, slowed conduction, and caused LA interstitial fibrosis
and inflammatory cell infiltration [175], [176]. In patients with established hypertension
and LV hypertrophy, there is global and regional conduction slowing associated with
fractionated electrograms and double potentials along the crista terminalis, along
with an increase in low-voltage areas [177].
Importantly, population studies show increased AF risk even with ‘pre-hypertension’
(systolic blood pressure 130–139 mmHg) [178]. The ab-normal atrial substrate is reversible,
with studies demonstrating improved electrical and structural parameters and reduced
AF burden following treatment with renin–angiotensin–aldosterone system blockers [179],
[180], [181]. In patients with resistant hypertension and improved blood pressure
following renal denervation, there was a global improvement in atrial conduction and
reduced complex fractionated activity. Histologically, pressure overload induces hypertrophy
of atrial myocytes (EHRAS Class I). Collagen deposition may also occur (EHRAS II–III)
with more severe hypertension causing LV hypertrophy and diastolic dysfunction (see
Table 2).
4.13
Obesity
Several population-based studies have demonstrated a robust relationship between obesity
and AF [182], [183], [184]. A recent meta-analysis estimates a 3.5–5.3% excess risk
of AF for every one unit of body mass index increase [185].
Left atrium dilation and dysfunction are known consequences of the cardiomyopathy
due to obesity [186]. In a sheep model of obesity, progressive weight gain over 8
months was associated with increased atrial volume, pressure, and pericardial fat
volume along with atrial interstitial fibrosis, inflammation, and myocardial lipidosis
[187]. This was associated with decreased conduction velocity, increased heterogeneity
of conduction and a greater inducibility of atrial fibrillation. With more sustained
obesity, animals not only demonstrate progressive atrial changes but also in areas
adjacent to pericardial fat there is infiltration of the atrial myocardium by fat
cells [188].
Obese patients have higher left atrial volume and pressure with lower left atrial
strain associated with shorter refractoriness in the LA and the PVs [189]. A detailed
evaluation of atrial changes associated with human obesity showed an increase in the
left atrial epicardial fat, a global reduction in atrial conduction velocity, increased
fractionation, and preserved overall voltage but greater low-voltage areas [190].
The low-voltage areas were observed in regions adjacent to epicardial fat depots.
Pericardial fat volume has been shown to be associated with AF incidence, severity,
and adversely effects ablation outcome [191], [192]. Epicardial adiposity is associated
with altered 3D atrial architecture, adipocyte infiltration into the myocardium, and
atrial fibrosis that may contribute to conduction heterogeneity that promotes AF [193],
[194], [195].
In the ovine model of chronic obesity, weight reduction is associated with reduction
in total body fat, atrial dilatation, and interstitial fibrosis together with improved
hemodynamics, atrial connexin-43 expression and conduction properties that result
in reduced vulnerability to AF [196]. In humans, aggressive management of weight and
associated risk factors is associated with favourable changes in pericardial fat volume,
atrial size, myocardial mass as well as electrophysiological and electroanatomical
changes along with reduced AF inducibility and burden [197]. Furthermore, weight loss
in morbidly obese subjects is associated with reduced epicardial fat [198]. Weight
reduction in obese individuals can result in regression of LV hypertrophy, reduction
in left atrial size and reduction in AF burden/severity [199], [200], [201]. Histologically,
fatty infiltrates (EHRAS Class IVf) as well as collagen depositions are present (EHRAS
III; see Table 2).
4.14
Diabetes mellitus
Diabetes is an independent risk factor for development and progression of AF [202].
In a rat model of diabetes mellitus, atrial tissue fibrosis deposit is associated
with decreased conduction velocity and greater AF inducibility [203]. Patients with
abnormal glucose metabolism have larger left atrial size, lower left atrial voltage,
and longer left atrial activation time compared with controls [204]. Insulin resistance
is associated with increased left atrial size and structural heterogeneity [205],
[206].
Mitochondrial function is impaired, leading to oxidative stress, in diabetic atria
[207]. Oxidative stress and activation of the advanced glycation end-product (AGE)-AGE-receptor
(RAGE) system mediates atrial interstitial fibrosis up-regulation of circulating tissue
growth factors and pro-inflammatory responses [207], [208]. In addition, prolonged
hyperglycaemic stress leads to accumulation of AGE-RAGE and nitric oxide inactivation,
leading to endothelial dysfunction and myocardial inflammation [209].
Hyperglycaemia and AGE-RAGE ligand interactions lead to decreased phosphorylation
of connexin-43, potentially impairing intercellular coupling [210]. Advanced glycation
is also related to alterations in myocardial calcium handling and hence contractility
[211]. These findings could explain the electrophysiological alterations that serve
as a central mechanism of the vulnerability to AF in diabetes [212].
Aggressive treatment of diabetes and adequate glycemic control may prevent or delay
the occurrence of AF, despite little direct evidence of the effects of anti-diabetic
drugs on AF. Peroxisome proliferator-activated gamma receptor agonists may offer protection
against AF beyond glycemic control, due to their anti-inflammatory, antioxidant, and
anti-fibrotic effects [213]. However, caution should be taken in extrapolating these
experimental findings to patients with diabetic cardiomyopathy. Histologically, changes
in the atrial myocytes are the initial findings without significant fibrosis (EHRAS
I). Later on the disease tissue appearance may change to EHRAS Class III and EHRAS
Class IV (see Table 2).
4.15
Atrial cardiomyopathy due to valvular heart disease
Mitral valve disease (MVD) and aortic stenosis (AS) have been associated with atrial
structural remodelling and a propensity for AF. Although secondary atrial cardiomyopathy
is most often associated with age, hypertension, and heart failure in developed countries,
RHD is responsible for over 40% of AF in the developing world [214].
4.15.1
Mitral stenosis
In atria from 24 patients with isolated MS and normal sinus rhythm undergoing mitral
valvuloplasty, John et al [215]. reported unchanged or an increased effective refractory
period (ERP), widespread and site-specific conduction delay, myocyte loss and patchy
electrical scar, suggesting that structural changes and their electrophysiological
consequences precede the development of AF. Factors associated with these structural
changes include direct myocardial effects (pathognomonic inflammatory Ashoff bodies),
ultrastructural changes, atrial fibrosis, immunoactive cytokines, and matrix metalloproteinase
remodelling (decreased MMP-1 and MMP-3) [215], [216], [217]. Reverse atrial remodelling
(an immediate reduction in LA pressure and volume and an improvement in biatrial voltage;
and further increases in RA voltage 6 months later) was demonstrated in 21 patients
with isolated MS undergoing commissurotomy [218]. In contrast, atrial remodelling
did not reverse in patients with lone AF undergoing successful AF ablation; indeed,
substrate abnormalities progressed (decreased voltage and increased regional refractoriness)
over the subsequent 6–14 months [219].
Atrial enlargement and fibrosis are important determinants for the development and
maintenance of AF. Increases in collagen I and collagen III (the latter which increase
in cultured fibroblasts exposed to mechanical stretch) [220] were seen in patients
with AF and MVD, but only type I was seen in patients with lone AF [221]. Cellular
decoupling and myocyte isolation, tissue anisotropy, and conduction inhomogeneities
were considered the substrate for local re-entry and arrhythmia.
4.15.2
Mitral regurgitation
Verheule et al [222]. found changes in atrial tissue structure and ultrastructure
1 month after creating severe mitral regurgitation (MR) by partial mitral valve avulsion.
Effective refractory periods were increased homogeneously and sustained AF (.1 h)
was inducible in 10 of 19 MR dogs; in this model, there were no differences in either
atrial conduction pattern or velocities. Interstitial fibrosis, chronic inflammation,
and cellular glycogen accumulation were noted in the dilated left atria, but myocyte
hypertrophy, myolysis, and necrosis were absent. In contrast, myocyte hypertrophy,
dedifferentiation, and degeneration and fibrosis are described in pigs with surgically
created chronic MR [223] and patients with MR [12], [224].
High-density oligonucleotide microarrays, enrichment analysis, and a differential
proteomics approach were used to characterize the molecular regulatory mechanisms
and biological processes involved in the atrial myopathy that is seen in pigs with
moderate to severe chronic (6 and 12 months) MR [225]. Renin-angiotensin-system and
peroxisome proliferator-activated receptor signalling pathways and genes involved
in the regulation of apoptosis, autophagy, oxidative stress, cell growth, and carbohydrate
metabolism were differentially regulated [225]. MLC2V (a marker of cardiac hypertrophy
and important in the regulation of myocyte contractility) had the highest fold change
in the MR pigs. Increased activity of a membrane-bound containing NADPH oxidase in
atrial myocytes, which correlated with the degree of cellular hypertrophy and myolysis,
was demonstrated in patients with isolated severe MR. The authors suggest that atrial
stretch-induced NADPH oxidase activation and intracellular oxidative stress contributes
to apoptosis, atrial contractile dysfunction, and atrial dilatation [226].
Correction of MR reverses many features of atrial remodelling and corrects functional
abnormalities. Early LA reverse remodelling (45% reduction of mean LA maximal volume)
and increased active atrial emptying was found in the early (30 day) postoperative
period in 43 patients undergoing mitral valve surgery (successful repair or replacement)
for chronic organic MR [227] and a similar improvement at 6 months was reported by
Dardas et al [228]. Histologically, EHRAS Class III is the most prominent finding
in MVD, although the histological appearance of the tissue may vary substantially
over time and interindividually and, therefore, all EHRAS classes may be found in
the tissue (see Fig. 1; Table 2).
4.15.3
Aortic stenosis
Although AS is associated with chronic AF [229], animal models of AS and atrial remodelling
are lacking. Kim et al [173]. studied atrial electrical re-modelling in excised perfused
hearts in a rat model of increased afterload simulating AS (ascending aortic banding),
which produced LVH without systemic hypertension, heart failure, or neurohormonal
activation. Banded hearts showed marked LA hypertrophy and fibrosis at 14 and 20 weeks
post-operatively. The incidence and duration of pacing-induced AF was increased at
20 weeks and was associated with decreased mean vectorial conduction velocity and
inhomogeneity of conduction, decreased expression of connexin-43, but without changes
in ERP. Importantly, atrial remodelling was not present at 8 weeks, when the greatest
degree of LVH was present [173].
Left atrium volumes are higher in patients with AS compared with controls and decrease
significantly after valvuloplasty [230]. Plasma natriuretic peptide (ANP) levels are
higher in symptomatic than asymptomatic patients with AS [231] and N-ANP levels predict
atrial remodelling and late (2 month) post-operative AF after surgery for AS [232].
Taken together, these data support the notion that substrate-based AF is a consequence
of the abnormal haemodynamics and atrial remodelling that accompany valvular heart
disease. In this instance, atrial remodelling is the consequence of multiple biological
processes that create structural and ultrastructural abnormalities and a change in
conduction (as opposed to refractoriness) that favours the development and maintenance
of AF. Histologically, EHRAS Class III is the most prominent finding, although the
histological appearance of the tissue may vary substantially over time and interindividually
(see Fig. 1, Fig. 2, Fig. 3; Table 2). Atrial pathology often also affects specialised
conduction system tissues like the sinus and AV nodes. However, these changes are
beyond the scope of the present consensus report, which focuses on atrial cardiomyocytes
and tissue.
5
Impact of atrial cardiomyopathies on occurrence of atrial fibrillation and atrial
arrhythmia
Controversy about the mechanism of AF has been alive for over 100 years, yet given
the continued increase in worldwide burden of AF [233], ongoing investigation will
drive improved treatment and prevention. Currently, there are two opposing sides in
the debate about re-entrant mechanisms in AF. On one side are those who promote variants
of the original idea of Gordon Moe that fibrillation, whether atrial or ventricular,
results from the continued random propagation of multiple independent electric waves
that move independently throughout the atria [234], [235]. On the other side are those
who adhere to the theory that fibrillation is a consequence of the continued activity
of a few vortices (rotors) that spin at high frequencies, generating ‘fibrillatory
conduction’ [236], [237]. In either case, arrhythmia maintenance is favoured by abbreviated
APD/refractory period [13], [238], [239]. Another pre-requisite of the multiple wavelet
hypothesis is that there should be slow conduction, which is not the case for rotors.
According to rotor theory, slowing of conduction is established dynamically by the
curvature of the rotating wave front, which is steepest near the rotation centre,
at which refractory period is briefest and conduction velocity is slowest [240]. Which
of the above two mechanisms prevails in human AF has not been fully established, yet
[241].
Regardless of the mechanism that maintains it, AF leads to high-frequency atrial excitation,
which if sustained, results in ion-channel remodelling that further abbreviates the
APD and refractory period to boost its stabilisation. Such AF-induced electrical remodelling
is reversible in the short term (minutes, hours, or days), but less so when lasting
months or years. For a detailed discussion of AF-induced remodelling, see chapter
3. How these changes contribute to AF perpetuation in the long term has not been fully
determined.
In a recent study using a sheep model of persistent AF induced by intermittent atrial
tachypacing there was a progressive spontaneous increase in the dominant frequency
(DF) of AF activation after the first detected AF episode [240], [242]. The results
suggested that, unlike the tachypacing induced electrical remodelling that can occur
over minutes or hours, there existed a protracted, slowly progressing electrical and
structural remodelling secondary to AF that sustains for days or weeks [240], [242].
In addition, a consistent left-vs.-right atrial DF difference correlated with the
presence of rotors, DF gradients, and outward propagation from the posterior LA during
sustained AF in the explanted, Langendorff-perfused sheep hearts [242], and an underlying
basis is seen in humans [243]. The DF of non-sustained AF increases progressively
at a rate (dDF/dt) that accurately predicts the transition from episodic, non-sustained
AF to persistent, long-lasting AF [126]. Although fibrosis developed progressively
[126], it is unknown what role if any fibrosis played in rotor acceleration or stabilisation.
Other studies using different animal models have also demonstrated that long-term
atrial tachypacing results in atrial fibrosis [244], with concomitant release of cytokines
that are known to modify atrial electrical function [245]. In the sheep model, atrial
structural changes leading to PLA enlargement likely made rotors less likely to collide
with anatomic boundaries, thus contributing to their stabilisation and AF persistence
[242], [246].
Distinct stresses of the atrial myocardium could contribute to the transformation
of atrial cardiomyopathy into an arrhythmogenic substrate for AF. For instance, mechanical
stress is a major regulator of cardiac electrical properties. The two atria are particularly
sensitive to changes in mechanical coupling due to their ‘reservoir’ position and
their function of ‘pressure sensor’ with a specific endocrine role, i.e. the secretion
of natriuretic peptides. Many mechanosensors are expressed in the atrial myocardium
and contribute to the interplay between membrane electrical properties, mechanical
stresses, and myocardial wall deformation [247]. Recently, it has been reported that
shear stress of atrial cardiomyocytes regulates the surface expression of voltage-gated
potassium channels via the stimulation of the integrins that link myocytes to the
extracellular matrix [248], [249]. During atrial haemodynamic overload, the mechano-sensor
signalling pathways, are constitutively activated, such that myocytes are no longer
able to respond to shear stress. This process results in the acceleration of atrial
repolarization and could contribute to AF vulnerability [249].
Oxidative stress is also thought to be important in AF-induced atrial remodelling
leading to cardiomyopathy and AF perpetuation [250]. How-ever, the manner in which
reactive oxygen species (ROS) mediate atrial ionic remodelling is inadequately understood.
NOX2/4 activity increases in fibrillating atria and is a potential source of ROS in
AF. Mitochondrial ROS is potentially another important source of oxidative stress;
mitochondrial dysfunction has been demonstrated in AF. It remains to be determined
whether atrial oxidative stress directly affects atrial APD and refractoriness and
thus contributes to rotor acceleration and stability in AF. Several sarcolemmal ionic
currents are directly or indirectly modulated by ROS [251], but the relevance of these
mechanisms to human AF has not been demonstrated.
Sustained AF activates the release of pro-inflammatory cytokines and hormones related
to cardiovascular disease and tissue injury, including angiotensin-II (Ang-II), tumour
necrosis factor (TNF)-a, interleukin (IL)-6, and IL-8 [252]. Pro-inflammatory stimuli
such as NOX-derived ROS, growth factors, and other hormones has been demonstrated
to have a role in Ang-II function [253]. However, the precise molecular modifications
of the putative signalling targets of ROS after Ang-II stimulation are yet to be identified.
Knowing which NOXs are activated by Ang-II in the normal atria may help generate better
interventions aimed at preventing AF associated with Ang-II activation. Ang-II is
a well-known trigger of fibroblast activation and differentiation into myofibroblasts,
which are key factors in the generation of fibrosis. Pro-inflammatory cytokines also
promote ion-channel dysfunction, which together with myocyte apoptosis and extracellular
matrix remodelling predisposes patients to AF.
Recently, atrial adipose tissue has emerged as a potential player in the pathophysiology
of AF [3], [254]. In addition to its paracrine effects [192], adi-pose tissue can
infiltrate the subepicardium of the atrial myocardium and become fibrotic [255] contributing
to the functional dissociation of electrical activity between epicardial layer and
the endocardial bundle network, favouring wavebreak, and rotor formation. Lone AF
or rapid atrial pacing promotes adipogenesis through the regulation of genes specific
to metabolic adaptation. Therefore, it is possible that the accumulation and infiltration
of adipose tissue reflects metabolic stress secondary to excessive work of the atrial
myocardium [191]. Furthermore, adipose tissue can induce fibrosis and alter gene-expression
patterns [195], [256].
6
Atrial cardiomyopathies, systemic biomarkers, and atrial thrombogenesis
6.1
Atrial cardiomyopathies and systemic biomarkers
6.1.1
Atrial inflammation and inflammatory biomarkers
Infiltration of neutrophils, macrophages, and lymphocytes accompanies surgical injury
or pericarditis, promoting the development of atrial fibrosis, resulting in heterogeneous
and slowed conduction, a risk factor for re-entrant arrhythmia [257], [258], [259],
[260], [261]. This provides a mechanistic link between inflammatory activation and
atrial arrhythmogenesis. Anti-inflammatory interventions such as prednisone are effective
in preventing neutrophil infiltration in sterile pericarditis and in suppressing pacing-inducible
atrial flutter [262], and steroid pre-treatment has been found to reduce the incidence
of postoperative AF in an appropriately powered randomized, clinical trial [263].
An ongoing trial studies the effect of colchicine (NCT 001128427).
In a mouse model of persistent hypertension, Ang-II infusion promotes increased atrial
abundance of myeloperoxidase (MPO, a neutrophil and macrophage oxidant-generating
enzyme) and promotes atrial fibrosis [261]. In MPO knockout mice, the profibrotic
response to A-II infusion was eliminated. Angiotensin II and endothelin-1 are linked
to inflammatory and proarrhythmogenic atrial remodelling [2], [264], [265], [266].
This evidence suggests that inflammatory cell infiltration has an important role in
promoting the creation of a substrate for AF, as a result of conduction heterogeneity
and slowing, both in the setting of cardiac surgery and beyond.
6.1.2
Systemic inflammatory activation in atrial fibrillation
In addition to haemodynamic stress-induced cellular inflammation of the atria, a cross-sectional
study demonstrated that AF was associated with higher plasma levels of C-reactive
protein (CRP), a sensitive but non-specific biomarker of systemic inflammation produced
by the liver [267]. A follow-up secondary analysis of the participants Cardiovascular
Health Study participants further revealed that elevated CRP predicted incident AF
[268].
Subsequent studies have demonstrated relationships between several different serologic
markers of inflammation and AF, including IL-6 [269], TNF-a [270], aldosterone [271]
and simple white blood cell counts [272]. Analyses of multiple inflammatory biomarkers
within the same study have suggested that IL-6 and osteoprotegerin [273] may be especially
important. The relationship between IL-6 and AF may be mediated by left atrial enlargement
[269].
While evidence that inflammatory markers presage the development of AF has been replicated
[268], [274], there are also multiple studies to dem-onstrate that atrial arrhythmias
likely contribute to inflammation: specifically, cardioversion of AF [275] as well
as ablation of either AF [276] or atrial flutter [277] has resulted in a decrease
in inflammation. Indeed, Marcus et al. demonstrated that the rhythm at the time of
the blood draw (AF vs. sinus) was an important determinant in detecting an elevated
CRP or IL-6 level [278]. Taken together, these data suggest that the relationship
between inflammation and AF may be bidirectional and progressive.
6.1.3
Intra-atrial sampling studies
As the enhanced risk of stroke in the setting of AF has been attributed to status
of blood flow and in particular thromboemboli originating in the left atrial appendage,
there has been an interest in determining whether peripheral blood can adequately
reflect the hypercoagulability that may be present locally within the atria (see Fig.
10) [279]. The first intra-atrial sampling study failed to identify evidence of statistically
significant differences between several markers of hypercoagulability in right and
left atrial vs. femoral vein and arterial samples among persistent AF patients with
MS [280]; of note, the same markers revealed statistically significant differences
when compared with normal controls without AF [279]. In contrast, a subsequent study
demonstrated that platelet activation acutely increased in coronary sinus blood in
AF, while systemic platelet activation (obtained from the femoral vein) revealed no
such change [281].
A similar approach to multi-site sampling has also been applied to better understand
the relationship between inflammation and AF. Liuba et al. found higher levels of
IL-8 in the femoral vein, right atrium, and coronary sinus than the left and right
upper PVs among eight permanent AF patients (without any such differences 10 paroxysmal
AF patients or 10 controls) [280].
6.1.4
Practical implications and use of systemic biomarkers
Systemic biomarkers have been used to predict development of AF and/or its complications
(Table 5). Various studies have examined the role of inflammatory indices, natriuretic
peptides, injury markers, etc. in predicting incident AF, especially in the post-surgery
setting. Many of these bio-markers are non-specific, and high levels may reflect infection
or sepsis, an acute phase reaction, etc [282], [283], [284].
Adding BNP and CRP to a prediction score derived from CHARGE-AF (which included data
from the Atherosclereosis Risk in Communities Study (ARIC), Cardiovascular Health
Study (CHS), the Framingham Heart Study, the Age, Gene/Environment Susceptibility
Reykjavik Study (AGES), and the Rotterdam Study) and utilising age, race, height,
weight, systolic and diastolic blood pressure, current smoking, use of antihypertensive
medication, diabetes, history of myocardial infarction and history of heart failure
[285] improved the statistical model [286]. Once again, the addition of CRP did not
meaningfully improve the model.
In another study evaluating the relationship of extracellular matrix modulators (matrix
metalloproteinases, MMPs, and their tissue inhibitors, TIMPs) and AF risk, only elevated
MMP9 levels were significantly associated with AF risk [287]. Proteases having desintegrin
and metalloprotease activities (ADAM) are related to atrial dilatation and thereby
influence mechanical performance of the atria [288].
The clinical benefit of considering biomarkers associated with AF is questionable
unless there is clear evidence of a direct benefit in AF risk prediction and management-
this has not been achieved to date.
6.2
Prothrombotic indices – coagulation, platelets
Over 150 years ago, Virchow proposed a triad of abnormalities that contributed to
thrombus formation (thrombogenesis), that is, abnormalities of vessel wall, abnormal
blood flow and abnormal blood constituents (Fig. 10). In the setting of AF, abnormalities
of vessel walls are evident by the association of thromboembolism with structural
heart disease (eg. mitral valve stenosis) and complex aortic plaque, as well as endothelial
damage/dysfunction, whether recognised by biomarkers (eg. von Willebrand factor (vWF),
tissue plasminogen activator, tPA), immunohistochemistry studies of the left atrial
wall, electron microscopy, or by functional studies (eg. flow mediated dilatation)
[289]. Abnormal blood flow in AF can be visualised by spontaneous echocontrast in
the LA, as well as low left atrial appendage Doppler velocities. Abnormal blood constituents
in AF are evident from abnormalities of coagulation, platelets, fibrinolysis, inflammation,
extracellular matrix turnover, etc. that are all directly or indirectly associated
with thrombogenesis, or a predisposition to the latter. While abnormalities of platelets
are often evident in AF, they may be more reflective of associated vascular disease
or comorbidities than of AF per se [290], [291]. Indeed, thrombus obtained in AF is
largely fibrin-rich (‘red clot’) compared with arterial thrombus, which is largely
platelet-rich (‘white clot’), providing a mechanistic explanation for the role of
anticoagulation therapy, rather than antiplatelet therapy for AF-related thromboembolism
[291], [292].
The concept of AF being a prothrombotic or hypercoagulable state was first proposed
in 1995 [293]. Many prothrombotic indices in AF have been related to subsequent stroke
and thromboembolism, whether in non-anticoagulated or anticoagulated subjects (Fig.
10). Initial studies showed that coagulation-related factors, such as fibrin D-dimer
(an index of fibrin turnover and thrombogenesis) were related to stroke risk strata
as well as an adverse prognosis from thromboembolism, whether or not patients were
anticoagulated [294], [295], [296], [297]. In contrast, there was no prognostic advantage
of platelet indices [295], [298], [299].
6.2.1
Prediction of thrombogenesis
Addition of vWf refines clinical risk stratification in AF, first shown in the non-anticoagulated
or suboptimally anticoagulated patients from the SPAF study [300]. More recently,
vWf has been related to thromboembolism as well as bleeding risks in anticoagulated
AF patients [301]. Ancillary studies from large Phase 3 anticoagulation trials have
reported prognostic implications for increased levels of D-dimer, troponin, natriuretic
peptides, and novel biomarkers (e.g. GDF15) [302], [303], [304]. Many of these studies
have been performed in selected clinical trial cohorts, and the prognostic role in
risk stratification requires prospective testing in unselected large ‘real-world’
cohorts with a broad range of stroke risk and renal function. As in the case of AF
prediction, evidence for the additive value of biomarkers for stroke risk prediction
from large prospective non-anticoagulated ‘real-world’ cohorts is limited [305]. Endocardial
thrombogenic alterations in diseased atria, which appear to be related to oxidative
stress, appear to contribute to clot formation, particularly in the left atrial appendage
[306], [307], [308], [309], [310]. Thus, the impact and the relation between EHRAS
Classses and the extend of endocardial thrombogenic alterations have to be assessed
in future studies. Interestingly, duration of AF does not correlate with the extent
of abserved endocardial changes [309].
7
Imaging techniques to detect atrial cardiomyopathies mapping and ablation in atrial
cardiomyopathies
It is well established that an enlarged LA is associated with adverse cardiovascular
outcomes [311], [312], [313], [314], [315], [316]. In the absence of MVD, an increase
in LA size most commonly reflects increased wall tension as a result of increased
LA pressure [317], [318], [319], [320], as well as impairment in LA function secondary
to atrial myopathy [321], [322]. A clear relationship exists between an enlarged LA
and the incidence of atrial fibrillation and stroke [323], [324], [325], [326], [327],
[328], [329], [330], [331], [332], risk for overall mortality after myocardial infarction
[321], [322], [333], [334], risk for death and hospitalisation in patients with dilated
cardiomyopathy [335], [336], [337], [338], [339], [340], [341], [342], [343], [344],
and major cardiac events or death in patients with diabetes mellitus [345]. left atrium
enlargement is a marker of both the severity and chronicity of diastolic dysfunction
and magnitude of LA pressure elevation [317], [318], [319], [320]. A recent consensus
report on multi-modality imaging for AF patients summarizes the current status of
atrial imaging in more detail [346].
7.1
Echocardiography
Echocardiography is the imaging modality of choice for screening and serially following
patients with diseases involving the LA morphology and function [347].
For assessment of atrial size, most widely reported is the linear dimension in the
parasternal long-axis view using M-mode or 2 delayed enhancement (DE) [324], [325],
[326], [327], [328], [329], [330], [331], [332], [333], [334], [335], [336], [337],
[338], [339], [345], [347], [348], [349]. However, due to the complex 3D nature of
the atrium and the non-uniform nature of atrial remodelling, this measurement frequently
does not provide an accurate picture of LA size [350], [351], [352], [353], [354].
Thus, when assessing LA size and remodelling, the measurement of LA volume is a more
powerful prognostic indicator in a variety of cardiac disease states [329], [331],
[333], [334], [335], [336], [337], [338], [339], [345], [347], [348], [349], [350],
[351], [352], [353], [354], [355], [356], [357], [358], [359], [360]. Two-dimensional
echocardiographic LA volumes are typically smaller than those reported from computed
tomography or cardiac magnetic resonance imaging (CMR) [361], [362], [363], [364],
[365]. Left atrium volume from 2D images is best measured using the disk summation
algorithm because it includes fewer geometric assumptions [366], [367]. The advent
of 3-D ECHO has improved the accuracy of ECHO volume measurements which correlate
well with cardiac computed tomography [368], [369] and magnetic resonance imaging
[370], [371]. Compared with 2D assessment of LA volume, 3DE also has superior prognostic
prediction [372], [373].
The recommended upper normal indexed LA volume is 34 mL/m2 for both genders which
fits well with a risk-based approach for determination of cut-off between a normal
and an enlarged LA [323], [357], [358], [359].
7.2
Left atrial function by Doppler echocardiography
Left atrium function can be assessed by pulsed-wave Doppler measurements of late (mitral
A) diastolic filling. Multiple studies have used this parameter as an index of LA
function assessment, but it is affected by age and loading conditions [317], [374],
[375], [376], [377], [378], [379], [380], [381], [382]. The PV atrial reversal velocity
has also been used as a measurement of LA function [317], [377], [379], [380], [381],
[382]. In the presence of reduced LV compliance and elevated filling pressures, atrial
contraction results in significant flow reversal into the PVs [80], [81]. Studies
have also demonstrated that Doppler tissue imaging can be used as an accurate marker
of atrial function [383], [384].
7.2.1
New echocardiographic techniques
Two-dimensional speckle-tracking echo has been used as a more sensitive marker to
detect early functional remodelling before anatomical alterations occur [385], [386],
[387], [388], [389], [390], [391], [392], [393], [394], [395], [396], [397], [398],
[399], [400].
Strain (S) and strain rate (SR) imaging provide data on myocardial deformation by
estimating spatial gradients in myocardial velocities [385], [388], [392], [393],
[401], [402], [403], [404], [405]. This technique has been used as a surrogate of
LA structural remodelling and fibrosis [388], [389], [390], [391], [392], [393]. Interestingly,
LA dysfunction with changes in strain and strain rate has been observed in patients
with amyloidosis in the absence of other echocardiographic features of cardiac involvement
[402]. Abnormalities in atrial strain have been observed in diverse conditions, including
AF, valvular pathology, heart failure, hypertension, diabetes, and cardiomyopathies
[388], [389], [396], [397], [398], [399], [400]. Population-based studies have demonstrated
the prognostic value of LA strain analysis for long-term outcome [388], [394].
Less research and fewer clinical outcomes data are available on the quantification
of RA size. Right atrial volumes are also underestimated with 2D echocardiographic
techniques compared with 3DE [343], [406], [407].
7.3
Cardiac computed tomography
Cardiac CT may be used for accurate assessment of atrial volumes. Volumetric data
from cardiac computed tomography (CCT) are comparable to data generated by CMR and
3D echocardiographic imaging and is superior to 2D echocardiography [371]. The LA
volume prior to catheter ablation and the presence of asymmetry of chamber geometry
predicts the likelihood of maintaining sinus rhythm post-procedure [408]. As the LA
enlarges, the shape of the LA roof initially becomes flat and then becomes coved,
and this progression may correlate with development of non-PV substrate in patients
undergoing AF ablation [409].
CCT may also be used to screen for thrombus prior to AF ablation. The diagnostic accuracy
of CT has been studied by multiple groups, with a systematic review of 19 studies
and 2955 patients reporting a sensitivity and specificity of 96 and 92%, respectively,
translating to a positive predictive value of 41% and a negative predictive value
of 99% [410]. Diagnostic accuracy increased to 99%, with 100% specificity, when delayed
imaging was performed. An advantage of using CT imaging to exclude thrombus is that
CCT is frequently performed prior to AF ablation for integration into the electroanatomic
mapping systems routinely used during AF ablation procedures. CCT can also provide
accurate information about PV anatomy and variants and correlates well with CMR in
that regard [411].
7.4
Magnetic resonance imaging of the atrium
Over recent years CMR has been used in clinical and research settings to provide gold
standard volumetric assessments of chamber structure and function. Drawbacks are that
CMR is expensive and has more limited availability than echocardiography. Recently,
contrast-enhanced CMR with gadolinium has been used as a technique to detect atrial
fibrosis [412]. Although these methods are still in relatively early stages and have
not been extensively reproduced, the ability to identify early degrees of atrial structural
change would no doubt enhance our ability to detect varying degrees of remodelling
that may not be as clear from volumetric or functional assessment. In addition to
late-gadolinium-enhanced (LGE) CMR to detect replacement fibrosis, post-contrast T1
mapping [413], [414] has been used to quantify diffuse interstitial fibrosis. Both
techniques have been correlated with bipolar voltage measured during invasive mapping
[412]. However, these techniques require specialised post-imaging processing. While
they are commonly used for ventricular imaging, they have not been widely employed
for atrial imaging because of the technical challenges in achieving adequate image
resolution in the thin-walled atrium [415].
Using a systematic scoring system for the extent of delayed enhancement, a recently-published
multicentre study has related the extent of LGE CMR detected fibrosis to the outcome
of AF ablation [416]. The risk of recurrent AF increased from 15% for stage I fibrosis
(,10% of the atrial wall) to 69% for stage IV fibrosis (≥30% of the atrial wall).
The authors suggested that CMR quantification of fibrosis may play a role in the appropriate
selection of patients most likely to benefit from AF ablation. Late-gadolinium-enhanced
CMR has also been used to predict development of sinus node dysfunction [417], stroke
risk [418], and progression of atrial fibrillation from paroxysmal to persistent [419].
However, various studies have highlighted the need to further improve the methods
of accurately identifying replacement fibrosis and to improve reproducibility of data
analysis before LGE CMR can be considered a routine clinical tool [420], [421].
Recently, a number of studies have used CMR DE late gadolinium enhancement (LGE) in
order to non-invasively characterize the extent and distribution of scarring present
following AF ablation [422], [423], [424]. Several studies observed that patients
with more extensive scar at 3 months (or greater percentage scar around the PV circumference)
had a lower AF recurrence rate [423], [425]. Another study showed a correlation between
measured contact force at the time of ablation, and the extent of CMR determined scar
development [426]. Other studies have shown a concordance between scar around the
PVs and low-voltage regions on invasive electroanatomic mapping (EAM) [427], [428].
Isolation of PVs at repeat procedures could be achieved guided by the imported MR
image to identify the gaps [427], [428]. However, other studies found no association
between CMR scar gaps and mapped PV reconnection sites. A study in 50 paroxysmal AF
patients undergoing either wide area or ostial ablation found that the proportion
of patients in whom CMR could correctly identify the distribution of ablation lesions
varied from as low as 28% to 54% depending on the technique used [429]. These authors
concluded that LGE imaging of atrial scar was not yet sufficiently accurate to reliably
identify ablation lesions or to determine their distribution. Whether CMR will have
the resolution to detect such focal regions where scar is incomplete remains uncertain.
Of note, Harrison et al. used an animal model to correlate lesion size on CMR with
lesion volume at pathology. The correlation depended critically on the definition
of pixel intensity used to define scar with small changes in definition leading to
large changes in estimated scar volume [415].
7.5
Imaging with electroanatomic mapping
Electroanatomic mapping systems have become the standard for invasive substrate characterisation
of atrial cardiomyopathies. Using various technologies, these systems allow for rapid
characterisation and reproduction of atrial anatomy with 3-D display rendering. Anatomic
variations in PV anatomy, including common ostium or additional veins, may be identified.
Visualisation software allows for accurate measurements of atrial distances [430]
and gross volumetric data but assessment of venous diameter may be suboptimal owning
to venous susceptibility to distortion. Anatomic imaging of the atria may be enhanced
with the co-registration of DICOM images from previously acquired cardiac MRI or CT
or with the use of real-time contrast angiography or intracardiac echocardiogram.
While EAM allows for anatomic reproduction of the atria, it also enables the assessment
of the atrial substrate through the geographic display of unipolar and bipolar signal
amplitude data, as well as other signal characteristics, on rendered atrial surfaces.
Regions of low-voltage, electrical silence, fractionation, or double potentials are
reputed to correlate with underlying atrial fibrosis, surgical patches, or scar. In
the same way, electrical activation of the atrium may be imaged allowing for assessment
of regional changes in conduction velocity [431] that may be proarrhythmic and support
the perpetuation of atrial fibrillation. The use of EAM for activation mapping of
atrial arrhythmia will be discussed in the subsequent section on ablation techniques.
Electroanatomic mapping has been used to image the electroanatomic substrate of atrial
cardiomyopathy associated with sinus node disease [432], rheumatic MS [215], atrial
septal defect [218], [431], CHF [433], obstructive sleep apnoea [117], and ageing
[167]. It has been a powerful research tool that has enhanced our understanding of
the atrial substrate in patients with paroxysmal and persistent atrial fibrillation
and [74], [434] those who have failed initial PV antrum isolation [435].
Unlike cardiac MR, CT, or echocardiography, EAM requires invasive catheterization
and mapping. However, despite recent advances in MRI techniques that allow for imaging
atrial scar, EAM imaging arguably has a great clinical feasibility and superior ability
to image and to define the atrial substrate that leads to the development of atrial
fibrillation. A recent consensus report on multi-modality imaging for AF patients
is a useful detailed reference [346].
7.6
Ablation of atrial tachyarrhythmia
Numerous single-centre, randomized studies and larger multicentre observational registries
have demonstrated the superiority of AF ablation over drug therapy for maintenance
of sinus rhythm. However, late recurrences are common and associated with more advanced
atrial substrate associated with structural heart disease [436], [437], [438], [439],
[440], [441], [442], [443], [444], [445], [446].
It is in this context that it is important to consider the various types of underlying
atrial cardiomyopathy and how they may affect ablation outcomes. This is timely, as
it has recently been observed that lone AF is a rapidly disappearing entity as we
recognise conditions such as sleep apnoea, obesity, endurance exercise etc. previously
not suspected of being causally associated with atrial fibrillation [447]. In addition,
emerging data suggest that treating these underlying causes may be central to improving
long-term ablation outcomes [199], [200], [448], [449]. In addition, LA ablation procedures
may alter atrial size, structure, and mechanical atrial function. Catheter ablation
may thus influence ongoing pathologies and atrial thrombogenesis [450], [451].
Mapping studies have demonstrated a common electrophysiological endpoint for a range
of such conditions affecting the atrium either primarily or secondarily, many of which
have been shown to be associated with atrial remodelling characterised by conduction
slowing and myocardial voltage reduction suggesting fibrosis [117], [167], [177],
[433], [452], [453]. Magnetic resonance imaging techniques attempting to characterize
the extent of myocardial fibrosis have demonstrated that this appears to be the strongest
independent predictor of AF recurrence after ablation [416], [454]. Whether the EHRAS
classification has value for informing catheter ablation in human atria remains to
be determined.
7.7
Age and atrial fibrillation ablation
Increasing age has been shown to be associated with increasing atrial fibrosis in
both basic and clinical studies [167], [455]. Numerous studies have evaluated ablation
outcomes in ageing patients (variously defined as .65 through to .80) [444], [445],
[456], [457], [458], [459], [460], [461], [462]. Observational studies have consistently
reported high multiple procedure success rates at 12 months of up to 80% in older
patients. Conflicting data exist regarding outcomes in comparative studies with one
study demonstrating a reduced success rate in patients over 65 years while another
study showed similar efficacy in patients over the age of 80 years to the younger
cohort [461], [463].
7.8
Hypertension
Hypertension is another well-recognised risk factor for development of atrial fibrillation.
Mapping studies have demonstrated the presence of a more advanced atrial substrate
in hypertensive patients compared with controls [177], [464]. Hypertension has been
shown to be a risk factor for recurrence of AF after AF ablation in numerous studies
on univariate analysis, but it is less clear whether this is independent of factors
such as atrial size. Recent preliminary studies have suggested that aggressive treatment
of hypertension improves post-ablation outcomes [200], [464], [465].
7.9
Heart failure and atrial fibrillation ablation
Contractile dysfunction has similarly been associated with advanced atrial remodelling
and predisposition to atrial fibrillation both in basic and in clinical studies [113],
[433]. Numerous studies have evaluated the efficacy of catheter ablation of both paroxysmal
and persistent atrial fibrillation with significant impairment of systolic function
[437], [466], [467], [468], [469], [470], [471], [472], [473]. The weight of evidence
is that sinus rhythm can be successfully achieved in 50–80% of patients although repeat
procedures are common and follow-up periods are usually not more than 12 months. Successful
ablation has been associated with significant improvements in ejection fraction and
reduction in atrial size in the majority of studies [470], [474].
7.10
Metabolic syndrome and obesity
A number of studies have evaluated the impact of the metabolic syndrome on catheter
ablation outcomes in atrial fibrillation patients [475], [476], [477], [478], [479],
[480]. Although the data are mixed, the weight of studies and a systematic review
[477] suggest a higher risk of AF recurrence. In the ARREST AF study, patients with
BMI over 27 undergoing AF ablation had a much lower risk of recurrence if weight loss
was achieved and maintained [200]. Observational studies have demonstrated a significantly
lower risk of recurrent AF in patients with treated compared with untreated OSA [481].
7.11
Impact of diabetes on ablation outcomes
Several studies have documented an increased recurrence rate of atrial fibrillation
after an ablation procedure in patients with diabetes mellitus [204], [475], [482].
An abnormal atrial substrate and non-PV triggers have been shown to underlie this
worse outcome.
7.12
Role of myocarditis
Markers of inflammation such as CRP and IL-6 have been linked to risk of AF [267],
[483], [484], [485]. Recently, giant-cell myocarditis involving only the atria has
been shown to result in atrial fibrillation with enlarged atria [149]. Patients with
apparently lone atrial fibrillation frequently demonstrate histological findings consistent
with an atrial myocarditis [486]; and those with past myocarditis may have atrial
electrical scar, conduction abnormalities, or atrial standstill [146], [487], [488],
[489]. Baseline CRP levels have been associated with the risk of recurrent AF after
catheter ablation [278]. Recently, colchicine has been used to prevent atrial fibrillation
recurrence after PV isolation [490]. It is also possible that AF in itself can result
in inflammation and the development of an ‘atrial myocarditis’ [491].
7.13
Impact of atrial fibrillation duration on atrial myopathy and atrial fibrillation
ablation outcomes
Longitudinal studies in AF patients have demonstrated clinical progression of AF over
time in a significant proportion with risk strongly associated with drivers such as
increasing age, structural heart disease, and hypertension [492]. Chronic AF results
in structural change with a recent study showing that in proportion to AF burden,
atrial remodelling may progress significantly even over a time period as short as
1 year.
Numerous studies have demonstrated that atrial size and occasionally mechanical function
may improve following ablation [493], but at least one invasive study showed no improvement
in atrial electrophysiology 6 months after successful ablation [219]. Overwhelmingly,
studies evaluating long-term outcomes after ablation of persistent atrial fibrillation
have demonstrated lower rates of procedural reversion to sinus rhythm and higher late
recurrence rates reflecting more advanced atrial substrate.
7.14
Impact of ongoing atrial fibrillation on electrical and structural remodelling
It is now well known that in the presence of an appropriate heterogenous AF substrate,
a focal trigger can result in sustained high-frequency re-entrant AF drivers, named
rotors. The waves that emerge from these rotors undergo spatially distributed fragmentation
and so give rise to fibrillatory conduction. When high-frequency atrial activation
is maintained for at least 24 h, ion-channel remodelling changes the electrophysiologic
substrate, promoting perpetuation of re-entry and increasing the activity of triggers,
further contributing to AF permanence [494]. Atrial fibrillation itself leads to remodelling,
causing electrophysiological (electrical), contractile, and structural changes [495],
[496]. Although AF can typically be reversed in its early stages, it becomes more
difficult to eliminate over time due to such remodelling [238], [497]. Dominant-frequency
analysis points to an evolution of mechanisms in AF patients, with PV sources becoming
less predominant as AF becomes more persistent and atrial remodelling progresses [498].
The data suggest that in patients with long-standing persistent AF, atrial remodelling
augments the number of AF drivers and shifts their location away from the PV/ostial
region.
7.15
Impact of catheter ablation on atrial pathology
Several studies have examined LA size before and after catheter ablation and have
demonstrated a 10–20% decrease in the dimensions of the LA after catheter ablation
of AF [499], [500]. Although the precise mechanism of this decrease in size is not
known, it appears consistent with reverse remodelling. It has been suggested that
earlier aggressive intervention to maintain sinus rhythm, including AF ablation if
needed, may aid to prevent ‘chronicization’ of AF and improve long-term outcomes [501].
A large-scale multicentre trial is presently testing this idea [502].
The true impact of atrial cardiomyopathies on the success of catheter ablation has
not been elucidated. Nevertheless, it is very likely that atrial pathology affects
energy delivery to tissue and specific forms of cardiomyopathy may differentially
affect ablation procedures. However, the true impact and interaction of various energy
sources with different atrial pathologies need to be studied.
8
Conclusion
Atrial cardiomyopathies as defined in this consensus paper have a significant impact
on atrial function and arrhythmogenesis. The EHRAS classification (EHRAS Class I–IV)
is a first attempt to characterize atrial pathologies into discrete cohorts. Because
disease-related histological changes in atrial tissue are often poorly characterised,
not necessarily specific and vary considerably over time their classification is challenging.
Further studies are needed to implement and validate the EHRAS classification and
to assess its value in guiding clinical understanding and management of AF. Nevertheless,
a more precise, defined classification of atrial pathologies may contribute to establishing
an individualised approach to AF therapy, which might improve therapeutic outcomes.
Supplementary material
Supplementary material is available at Europace online.
Conflict of interest
A detailed list of disclosures of financial relations is provided as Supplementary
material online.