J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8

ª 2 0 1 8 B Y T H E AM E R I C A N C O L L E G E O F C A R D I O L O G Y F O UN DA T I O N

P U B L I S H E D B Y E L S E V I E R

I S S N 2 4 0 5 - 5 0 0 X / $ 3 6 . 0 0

h t t p : / / d x . d o i . o r g / 1 0 . 1 0 1 6 / j . j a c e p . 2 0 1 7 . 0 8 . 0 1 2

Biatrial Electrical and StructuralAtrial Changes in Heart FailureElectroanatomic Mapping in Persistent Atrial Fibrillationin Humans

Sandeep Prabhu, MBBS,a,b,c,d Aleksandr Voskoboinik, MBBS,a,b,c,d Alex J.A. McLellan, MBBS, PHD,a,b,c,d

Kah Y. Peck, MBBS,a Bhupesh Pathik, MBBS,c,d Chrishan J. Nalliah, MBBS,c,d Geoff R. Wong, MBBS,c,d

Sonia M. Azzopardi, CC BC RN,a,b Geoffrey Lee, MBCHB, PHD,c Justin Mariani, MBBS, PHD,a,b

Liang-Han Ling, MBBS, PHD,a,b,c,d Andrew J. Taylor, MBBS, PHD,a,b Jonathan M. Kalman, MBBS, PHD,c,d

Peter M. Kistler, MBBS, PHDa,b,d

ABSTRACT

Fro

Au

He

Go

fun

Au

Au

of

au

Ed

All

ins

vis

Ma

OBJECTIVES This study sought to characterize the biatrial substrate in heart failure (HF) and persistent atrial

fibrillation (PeAF).

BACKGROUND Atrial fibrillation (AF) and HF frequently coexist; however, the contribution of HF to the biatrial

substrate in PeAF is unclear.

METHODS Consecutive patients with PeAF and normal left ventricular (NLV) systolic function (left ventricular ejection

fraction [LVEF] >55%) or idiopathic cardiomyopathy (LVEF #45%) undergoing AF ablation were enrolled. In AF,

pulmonary vein (PV) cycle length (PVCL) was recorded via a multipolar catheter in each PV and in the left atrial

appendage for 100 consecutive cycles. After electrical cardioversion, biatrial electroanatomic mapping was performed.

Complex electrograms, voltage, scarring, and conduction velocity were assessed.

RESULTS Forty patients, 20 patients with HF (mean age: 62� 8.9 years; AF duration: 15� 11 months; LVEF: 33 � 8.4%)

and 20 with NLV (mean age: 59 � 6.7 years; AF duration: 14 � 9.1 months; p ¼ 0.69; mean LVEF: 61 � 3.6%; p < 0.001),

were enrolled. HF reduced biatrial tissue voltage (p < 0.001) with greater voltage heterogeneity (p < 0.001). HF was

associatedwith significantlymore biatrial fractionation (left atrium [LA]: 30%vs. 9%; p<0.001; right atrium [RA]: 28%vs.

11%; p < 0.001), low voltage (<0.5 mV) (LA: 23% vs. 6%; p ¼ 0.002; RA: 20% vs 11%; p ¼ 0.006), and scarring

(<0.05 mV) in the LA (p ¼ 0.005). HF was associated with a slower average PVCL (185 vs. 164 ms; p ¼ 0.016), which

correlated significantly with PV antral bipolar voltage (R ¼ �0.62; p < 0.001) and fractionation (R ¼ 0.46; p ¼ 0.001).

CONCLUSIONS HF is associated with significantly reduced biatrial tissue voltage, fractionation, and prolongation of

PVCL. Advanced biatrial remodeling may have implications for invasive and noninvasive rhythm control strategies in pa-

tients with AF and HF. (J Am Coll Cardiol EP 2018;4:87–96) © 2018 by the American College of Cardiology Foundation.

m the aDepartment of Cardiology, Alfred Hospital, Victoria, Australia; bBaker IDI Heart and Diabetes Institute, Victoria,

stralia; cCardiology Department, Royal Melbourne Hospital, Victoria, Australia; and the dFaculty of Medicine, Dentistry, and

alth Sciences, University of Melbourne, Victoria, Australia. This research has been supported in part by the Victorian

vernment’s Operational Infrastructure Funding. Drs. Prabhu, Ling, McLellan, Voskoboinik, Nalliah, and Pathik have received

ding from the Australian National Health and Medical Research Council (NHMRC) and/or the National Heart Foundation of

stralia. Drs. Prabhu and McLellan have also received funding from the Baker Heart and Diabetes Research Institute (Melbourne,

stralia). Drs. Kalman, Lee, and Kistler have been supported in part by the NHMRC. Dr. Kalman has served on the advisory board

Biosense Webster; and has received research and fellowship support from Medtronic, Abbott, and Biosense Webster. All other

thors have reported that they have no relationships relevant to this paper to disclose. Katia Zeppenfeld, MD, served as Guest

itor for this article.

authors attest they are in compliance with human studies committees and animal welfare regulations of the authors’

titutions and Food and Drug Administration guidelines, including patient consent where appropriate. For more information,

it the JACC: Clinical Electrophysiology author instructions page.

nuscript received March 22, 2017; revised manuscript received August 24, 2017, accepted August 28, 2017.

ABBR EV I A T I ON S

AND ACRONYMS

AF = atrial fibrillation

CFE = complex fractionated

electrogram

CMR = cardiac magnetic

resonance

CV = conduction velocity

HF = heart failure

LA = left atrium

LAA = left atrial appendage

LV = left ventricle

LVEF = left ventricular ejection

fraction

NLV = normal left ventricle

PV = pulmonary vein

PVCL = pulmonary vein cycle

length

PVFPVAverage = average

pulmonary vein cycle length of

the fastest pulmonary vein

PVI = pulmonary vein isolation

RA = right atrium

TCMP = tachycardia- or

arrhythmia-mediated

cardiomyopathy

Prabhu et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8

Atrial Remodeling in Heart Failure and AF J A N U A R Y 2 0 1 8 : 8 7 – 9 6

88

A trial fibrillation (AF) and heart fail-ure (HF) are burgeoning epidemicscontributing to significant morbidity

and mortality and an increasing burden onhealth care systems (1,2). AF and HF coexistin a significant proportion of patients due tosignificant overlap in the pathophysiologicalprocesses driving both conditions (3).Recently, catheter ablation for AF has beenpurported as an effective treatment forselected patients with AF and HF, with im-provements in left ventricular (LV) functionand functional class (4–6). However, the suc-cess of ablation is inferior in patients with HFcompared to those with normal left ventricu-lar (NLV) function (7,8). AF and HF are bothassociated with atrial structural remodeling,which adversely affects long-term outcomesafter catheter ablation. However, within thepersistent AF population, the specific contri-bution of LV systolic dysfunction to atrialremodeling, over and above the contributionof persistent AF itself, has not been deter-mined. We sought to characterize structuralremodeling in patients with persistent AF,with and without LV systolic dysfunction,using comprehensive detailed biatrial elec-troanatomic mapping.

SEE PAGE 97

METHODS

PATIENT SELECTION. Consecutive patients from 2Australian centers undergoing catheter ablation forpersistent AF were prospectively screened. Patientswere included in the study if they were >18 years, hadsymptomatic persistent AF of at least 3 months’duration, and had systolic left ventricular ejectionfraction (LVEF)#45% (HF group) or$55% (NLV group)on pre-procedural cardiac imaging (either echocardi-ography or cardiac magnetic resonance [CMR]).Patients were excluded if they had paroxysmal AF or aknown structural cause of HF such as previousmyocardial infarction or severe valvular disease, hadLVEF between 45% and 55%, or were unable tomaintain sinus rhythm after electrical cardioversionto facilitate mapping. Persistent AF for this study wasdefined as AF duration >3 months to allow theassessment of pulmonary vein (PV) cycle length(PVCL) without the risks of spontaneous reversion.Long-standing persistent AF was defined as contin-uous AF duration for >1 year. All patients wereadequately rate controlled for a minimum of 4 weeks

before mapping and ablation. The study was approvedby the human ethics committee at each study location.MAPPING TECHNIQUE. Antiarrhythmic drugs werediscontinued 5 half-lives before the procedure. Anti-coagulation was stopped 2 to 5 days before ablation orcontinued (in the case of vitamin K antagonists) at theoperator’s discretion, with intravenous heparin usedintraprocedurally aiming for activated clotting time>350 s. After transesophageal echocardiography toexclude intracardiac thrombus, double transseptalaccess was performed. Left atrial (LA) geometry wasconstructed using a 20-pole lasso catheter (BiosenseWebster Inc., Diamond Bar, California) and registeredwith the pre-procedural computed tomography orCMR. In AF, PVCL was recorded in each PV and in theleft atrial appendage (LAA) as described in the“Pulmonary Vein Cycle Length” paragraph. Electricalcardioversion to sinus rhythm was then performed.Mapping was performed before ablation using theCARTO3 electroanatomic mapping system (BiosenseWebster Inc.) during pacing from the coronary sinusat 600 ms with a 3.5-mm irrigated ablation catheter(Biosense Webster Inc.) with contact force capability,aiming for an even distribution of at least 150 pointsacross all atrial regions, up to and including the PVostia but excluding points within the PV. Only pointswith contact force >10 g were included for analysis.Mapping was performed in an identical fashion in theright atrium (RA) after LA ablation but before any RAablation (e.g., cavotricuspid isthmus ablation).ELECTROGRAM ANALYSIS. Electrogram analysis wasperformed off-line after the procedure. Points wereanalyzed at 200 mm/s sweep speed for fractionation,voltage, and scar. Complex fractionated electrograms(CFEs) were defined as electrograms with $3 de-flections and duration $50 ms. Unipolar and bipolarvoltages were recorded for each point as themaximum voltage difference between the highest andlowest amplitude deflections. Points with bipolarvoltage <1.5 mV were defined as reducedvoltage, <0.5 mV as low voltage, and <0.05 mV asscar. For the purposes of regional analysis, the atriawere divided into 4 segments for the RA and 5 seg-ments for the LA (Figure 1). The PV antrum wasdefined as the LA region within a circumferential ring2 cm proximal to the PV ostia, approximating a typicalwide encirclement ablation line (marked in green inFigure 1). Percentages represent the proportion oftotal points across the entire atria (for global analysis)or within each region (for regional analysis) meetingthe defined criteria for CFE, low voltage, or scar.Global and regional voltages were determined by theaverage voltage across the entire atria for globalanalysis or within each region for regional analysis.

FIGURE 1 Segmentation of the Left Atrium Into Regions

Division of the left atrium into anterior, posterior, inferior, septal, and lateral regions for regional analysis. The antral region was defined at

the circumferential region 2 cm proximal to the pulmonary vein/left atrial junction. It encompasses a typical wide antral circumferential

ablation to achieve pulmonary vein isolation. The posterior region was identified as the posterior area between the right and left antral region,

as shown in red. LAA ¼ left atrial appendage; LIPV ¼ left inferior pulmonary vein; LSPV ¼ left superior pulmonary vein; RIPV ¼ right

inferior pulmonary vein; RSPV ¼ right superior pulmonary vein.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8 Prabhu et al.J A N U A R Y 2 0 1 8 : 8 7 – 9 6 Atrial Remodeling in Heart Failure and AF

89

CONDUCTION VELOCITY. Conduction velocity (CV)in each region was determined using a methodologypreviously described (9). In brief, pairs of points wereselected in each region perpendicular to isochronesmeasured at 5 isochronal steps in areas of leastisochronal crowding. CV was determined by dividingthe measured shortest surface distance betweenpoint pairs by the difference in local activation time.Regional CV was determined as the average CVmeasured from 5 different point pairs in each region.Global CV consisted of the average of each regionalCV within each atrium.

PULMONARY VEIN CYCLE LENGTH. Pulmonary veincycle length (PVCL) was measured using the meth-odology described by Pascale et al. (10). In brief,before ablation, the multipolar catheter was placed ineach PV for 60 s. Average PVCL for each vein wasdetermined as the average of 100 consecutive PV cy-cle lengths. Average PVCL for each patient was theaverage of each measured PV. Average LAA CL wasused as a surrogate for LA AF cycle length and wasmeasured in an identical manner.

STATISTICAL ANALYSIS. Data are expressed asmean � SD unless otherwise indicated. After assess-ment of normal distribution with the Kolmogorov-Smirnov test, 2-group comparisons were made using

the Student’s t-test for continuous variables or thechi-square test for categorical variables. Theindependent-samples Mann-Whitney U test was usedfor non-normally distributed variables. Correlationwas performed using a Pearson correlation test.A 2-tailed p < 0.05 was considered significant.Analyses were conducted using SPSS software, version24 (IBM, Chicago, Illinois).

RESULTS

STUDY POPULATION. Forty patients with persistentAF underwent biatrial mapping (20 in the HF groupand 20 in the NLV group). Baseline characteristics arelisted in Table 1. The HF group had an average LVEFof 33 � 8.4% compared to 61 � 3.6% in the NLV group(p < 0.001). Both groups were well matched withrespect to age, gender, body mass index, comorbid-ities, biatrial dimensions, and AF burden, althoughexpectedly the HF group had higher average CHADS2(1.38 � 0.86 vs. 0.79 � 0.71; p ¼ 0.011) and CHA2DS2-VASc scores (2.19 � 1.21 vs. 1.00 � 0.82; p ¼ 0.001)compared to the NLV group.

There were expected differences in medicationswith anti-HF therapy, with beta-blocker use (100%vs. 45%; p < 0.001), spironolactone use (45% vs. 0%;p < 0.001), angiotensin-converting enzyme inhibitor

TABLE 1 Baseline Characteristics of the Study Population

HF Group(n ¼ 20)

NLV Group(n ¼ 20)

pValue

Age (yrs) 62 � 8.9 59 � 6.7 0.24

Male 95 90 0.54

LVEF 33 � 8.4 61 � 3.6 <0.001

Long-standing PeAF 85 70 0.26

LA area (cm2) 30.3 � 6.1 27.6 � 4.9 0.13

RA area (cm2) 26.0 � 7.2 22.7 � 3.3 0.10

Continuous AF pre-ablation (months) 14.8 � 11.3 13.6 � 9.1 0.69

Previous electrical cardioversion 90 90 1.00

Hypertension 40 50 0.52

T2DM 20 20 1.00

Average CHA2DS2-VASc score 2.19 � 1.21 1.00 � 0.82 <0.001

Average CHADS2 score 1.38 � 0.86 0.79 � 0.71 0.024

Obstructive sleep apnea 15 10 0.63

Average BMI (kg/m2) 28.8 � 4.6 29.9 � 3.2 0.46

ACE inhibitor or ARB 95 30 0.011

Beta-blocker 100 45 <0.001

Spironolactone 45 0 0.018

Diuretic 55 0 <0.001

Any antiarrhythmic drug 55 65 0.52

AF and HF codiagnosed 65

AF preceded HF 25

HF preceded AF 15

Tachycardia-related cardiomyopathy* 55

Values are mean � SD or %. *LVEF improved to $50% after ablation.

ACE ¼ angiotensin-converting enzyme; AF ¼ atrial fibrillation; ARB ¼ angiotensin receptorblocker; BMI ¼ body mass index; CHA2DS2-VASc ¼ congestive heart failure, hypertension,age $75 years, diabetes mellitus, prior stroke, transient ischemic attack, or thromboembolism,vascular disease, age 65-74 years, sex category (female); HF ¼ heart failure; LA ¼ left atrium;LVEF ¼ left ventricular ejection fraction; NLV ¼ normal left ventricle; PeAF ¼ persistent atrialfibrillation; RA ¼ right atrium; T2DM ¼ type 2 diabetes mellitus.

Prabhu et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8

Atrial Remodeling in Heart Failure and AF J A N U A R Y 2 0 1 8 : 8 7 – 9 6

90

or angiotensin receptor blocker therapy (95% vs. 65%;p ¼ 0.018), and diuretic therapy (50% vs. 0%;p < 0.001) being significantly more frequent in the HFgroup. The use of any antiarrhythmic drug wasequivalent between groups (55% vs. 65%; p ¼ 0.52).Fifty percent of patients (9 of 18) in the HF group hadCMR-detected late gadolinium enhancement. In theNLV group, the average E/e0 was 7.5 � 1.3, with nopatients having E/e0 $9.2, suggesting significant HFwith preserved ejection fraction was unlikely (11,12).

ELECTROANATOMIC MAPPING. There was no dif-ference between groups with regard to the number ofmapping points in both the LA and RA (HF vs. NLV:LA: 221 � 79 vs. 210 � 59; p ¼ 0.62; RA: 200 � 36 vs.224 � 59; p ¼ 0.14) (Table 2, Figure 2).Tissue vol tage . In the LA, global unipolar and bi-polar voltages were significantly lower in the HFgroup (unipolar: 2.36 � 0.92 mV; bipolar: 1.47 � 0.60mV) compared to the NLV group (unipolar: 3.58 � 1.07mV; p < 0.001; bipolar: 2.28 � 0.69 mV; p < 0.001).Voltage heterogeneity, as measured by covariance(SD/mean), was significantly increased in the HF

compared to the NLV group (unipolar: 0.60 � 0.12 vs.0.47 � 0.07; p < 0.001; bipolar: 0.75 � 0.15 vs. 0.61 �0.07; p < 0.001).

In the RA, global unipolar and bipolar voltageswere significantly lower in the HF group (unipolar:1.88 � 0.44 mV; bipolar: 1.45 � 0.36 mV) compared tothe NLV group (unipolar: 2.67 � 0.84 mV; p ¼ 0.001;bipolar: 2.13 � 0.75 mV; p ¼ 0.001). Voltage hetero-geneity, as measured by covariance (SD/mean), wasalso increased in the HF compared to the NLV group(bipolar: 0.80 � 0.11 vs. 0.70 � 0.11; p ¼ 0.006).

HF in both atria was associated with a significantincrease in reduced-voltage (#1.5 mV) points (LA:59.9 � 19.9% vs. 35.6 � 20.2%; p < 0.001; RA: 64.2 �12.9% vs. 44.8 � 20.0%; p ¼ 0.001) and low-voltage(#0.5 mV) points (LA: 23 � 17% in HF vs. 6.3 � 5.9%in NLV; p < 0.001; 20 � 11% vs. 11 � 7.9%; p ¼ 0.006)compared to the NLV group.

Scar points (bipolar voltage#0.05) were foundmorefrequently in the HF group in the LA (1.4 � 1.5% vs. 0.2�0.9%; p¼0.005) and RA (1.7� 3.9% vs. 0%; p¼0.09).However, in both atria, a significantly higher propor-tion of patients in the HF group had scar present(LA: 75%vs. 10%; p<0.001; RA: 50% vs. 0%; p<0.001).

Within the HF group, atrial tissue voltage wassignificantly higher in patients with tachycardia- orarrhythmia-mediated cardiomyopathy (TCMP), asdefined by an improvement in LVEF to $50% aftercatheter ablation (n ¼ 11), compared to those who didnot demonstrate LV recovery (non-TCMP; n ¼ 9).Bipolar voltage was 1.72 � 0.64 in TCMP versus 1.20 �0.44 mV in non-TCMP (p ¼ 0.045); unipolar voltagewas 2.67 � 0.95 in TCMP versus 1.92 � 0.60 mV innon-TCMP (p ¼ 0.048); and low voltage (<0.5 mV)was 17 � 16% in non-TCMP versus 31 � 14% in non-TCMP (p ¼ 0.05).Conduct ion veloc i ty . In the HF group, global CV wassignificantly slower in the RA (0.91 � 0.17 m/s vs. 1.03� 0.08 m/s; p ¼ 0.026), with a nonsignificant differ-ence in the LA (0.98 � 0.21 m/s vs. 1.06 � 0.15 m/s;p ¼ 0.22) (Table 3).

Complex fract ionated e lect rograms . The HFgroup was associated with a significant increase inCFEs in both atria compared to the NLV group (LA:31 � 17% vs. 9.1 � 8.5%; p < 0.001; RA: 28 � 14% vs.11 � 8.5%; p < 0.001).

REGIONAL ASSESSMENT OF ATRIAL SUBSTRATE.

PV antrum. Bipolar voltage was significantlyreduced in the antrum in the HF group comparedto the NLV group (1.18 � 0.51 mV vs. 2.00 � 0.68 mV;p < 0.001). In the HF group, antral bipolar voltage wassignificantly lower in the antral compared to thenonantral region (1.18 � 0.51 mV vs. 1.76 � 0.80 mV;

TABLE 2 Biatrial Electrophysiological Parameters in HF and NLV Function Groups

HF Group(n ¼ 20)

NLV Group(n ¼ 20)

Standard Difference(95% CI) p Value

Electroanatomic mapping of left atrium

Average total no. of points 221 � 79 210 � 59 0.62

Bipolar voltage (mV) 1.47 � 0.60 2.28 � 0.69 0.81 (0.40 to 1.22) 0.0003

Covariance of bipolar voltage 0.75 � 0.15 0.61 � 0.07 0.14 (0.07 to 0.22)

Unipolar voltage (mV) 2.36 � 0.92 3.58 � 1.07 1.26 (0.64 to 1.88) <0.001

Covariance of unipolar voltage 0.60 � 0.12 0.47 � 0.07 0.12 (0.19 to 0.07)

Reduced voltage 59.9 � 19.9 35.6 � 20.2 24.3 (11.5 to 37.1) <0.001

Low voltage 23.4 � 16.5 6.3 � 5.9 17.3 (9.3 to 25.2) <0.001

Scarring 1.4 � 1.5 0.2 � 0.9 �1.2 (�2.0 to �0.38)

Presence of any scar (% of patients) 75 10 <0.001

Complex electrograms 31.1 � 11.8 9.1 � 8.5 22.0 (15.3 to 28.6) <0.001

CV (m/s) 0.98 � 0.21 1.05 � 0.15 0.10 (�0.01 to 0.22) 0.21

Covariance for CV (SD/mean) 0.33 � 0.07 0.31 � 0.07 �0.15 (�0.06 to 0.03)

Electroanatomic mapping of right atrium

Average total no. of points 200 � 36 224 � 59 0.14

Bipolar voltage (mV) 1.45 � 0.37 2.13 � 0.75 0.68 (0.29 to 1.07) 0.001

Covariance of bipolar voltage 0.79 � 0.11 0.70 � 0.11 0.11 (0.03 to 0.18)

Global unipolar voltage (mV) 1.93 � 0.50 2.67 � 0.84 0.79 (0.34 to 1.22) 0.003

Covariance of unipolar voltage 0.63 � 0.10 0.57 � 0.11 �0.06 (�0.13 to 0.1)

Reduced voltage (<1.5 mV) 64.2 � 12.9 44.8 � 20.0 19.4 (8.21 to 30.6) 0.001

Low voltage (<0.5 mV) (%) 19.8 � 10.7 10.5 � 7.9 10.1 (3.6 to 16.6) 0.006

Scarring (<0.05 mV) (%) 1.7 � 3.9 0 �1.7 (�3.7 to 0.28)

Presence of any scar (% of patients) 53 0 <0.001

Complex electrograms (%) 28.2 � 13.9 11.2 � 8.5 17.1 (9.4 to 24.7) <0.001

CV (m/s) 0.91 � 0.17 1.03 � 0.08 0.12 (0.02 to 0.22) 0.021

Covariance for CV (SD/mean) 0.27 � 0.07 0.34 � 0.07 0.07 (0.02 to 0.12)

Values are mean � SD or %.

CI ¼ confidence interval; CV ¼ conduction velocity; other abbreviations as in Table 1.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8 Prabhu et al.J A N U A R Y 2 0 1 8 : 8 7 – 9 6 Atrial Remodeling in Heart Failure and AF

91

p ¼ 0.016). In the NLV group, no significant differencewas seen between antral and nonantral voltage(2.01 � 0.68 mV vs. 2.39 � 0.75 mV; p ¼ 0.10). CFEswere significantly increased in the antrum in the HFgroup compared to the NLV group (40 � 15% vs. 15 �14%; p < 0.001). Again within the HF group, CFEswere most pronounced in the antrum versus thenonantral region (26 � 13%; p ¼ 0.022). Within theNLV group, no significant difference was seen be-tween antral and nonantral regions (8.2 � 7.1%;p¼ 0.10). HF was associated with a significant increasein low voltage (25� 25% vs. 9.5� 11% in the NLV group;p < 0.016) within the PV antrum (Tables 3 and 4).Other reg iona l LA analys i s . Tissue voltage wasreduced in the HF group in all subregions (Figure 1) ofthe LA (Table 3, Figures 2A and 2B). In the posteriorLA, HF was associated with a significant reduction involtage (bipolar: 1.34 � 0.57 vs. 2.18 � 0.77; p < 0.001;unipolar: 2.23 � 0.89 vs. 3.59 � 1.1 � 9; p < 0.001)and regional low voltage (p < 0.001). Atrial scarwas more frequent in proportion (2.4 � 3.3% vs.0.1 � 0.4%; p ¼ 0.005) and number of patients withscarring in HF (50% vs. 10% in NLV). In the posterior

LA in the HF group, CFEs were significantly greater(32 � 19% vs. 8.8 � 8.3% in NLV; p < 0.001). Signifi-cantly reduced tissue voltage in the HF group wasalso pronounced in the septum.PVCL assessment . In AF, the average PVCL wassignificantly longer in the HF group compared to theNLV group (185 � 27 ms vs. 165 � 19 ms; p ¼ 0.016)(Figure 2B). The average PVCL of the fastest PV(PVFPVAverage) was also significantly slower in the HFgroup (172 � 24 ms vs. 155 � 17 ms; p ¼ 0.013). Theaverage of each parameter relative to the LAA cyclelength was also significantly higher in the HF versusthe NLV group (PV4PVAverage/LAA: 1.06 � 0.09 vs. 0.99� 0.10; p ¼ 0.028; PVFPVAverage/LAA: 1.03 � 0.12 vs.0.93 � 0.10; p ¼ 0.015); PVFast/LAA: 0.71 � 0.11 vs. 0.57� 0.10; p < 0.001). Significantly more patients in theHF group had a ratio of PV4PVAverage/LAA or PVFPVA-

verage/LAA >1 (PV4PVAverage/LAA: 95% vs. 45%; p < 0.001;PVFPVAverage/LAA: 60% vs. 25%; p ¼ 0.025). There wasa significant correlation between PV4PVAverage andbipolar voltage (R ¼ �0.62; p < 0.001) and complexfractionated activity (R ¼ 0.46; p ¼ 0.001) in theantrum (Figure 3).

FIGURE 2 Biatrial Bipolar Voltage Maps in HF and NLV Function

Bipolar voltage in the LA and RA in a patient with persistent AF and dilated cardiomyopathy (LVEF 31%) and a patient with persistent AF and

normal LV function (LVEF 65%). (A) Areas of low voltage in heart failure congregating around the posterior and antral regions. (B) Mean

voltage across the RA and LA in HF and NLV. Sample of pulmonary vein activity with the average PVCL is also indicated for HF and NLV

showing prolongation of PVCL in HF relative to NLV. AF ¼ atrial fibrillation; HF ¼ heart failure; IVC ¼ inferior vena cava; LA ¼ left atrium;

LVEF ¼ left ventricular ejection fraction; NLV ¼ normal left ventricle; PA ¼ posteroanterior; SVC ¼ superior vena cava; other abbreviations as in

Figures 1 and 3.

Prabhu et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8

Atrial Remodeling in Heart Failure and AF J A N U A R Y 2 0 1 8 : 8 7 – 9 6

92

DISCUSSION

AF and HF frequently coexist; however, whether HFconfers a cumulative impact on atrial structuralremodeling over and above the impact of AF itself hasnot been previously explored. In the present study,

we undertook detailed biatrial substrate analysisin patients with persistent AF both with and withoutLV dysfunction. In persistent AF, patients with HFdemonstrated a significant reduction in biatrialunipolar and bipolar tissue voltages; increase inbiatrial voltage heterogeneity, low voltage, and scar;

TABLE 3 Regional Tissue Voltage

HF Group NLV Group p Value

LA septum

Bipolar voltage (mV) 1.13 � 0.55 2.16 � 0.83 <0.001

Unipolar voltage (mV) 1.90 � 0.91 3.22 � 1.37 <0.001

LA anterior

Bipolar voltage (mV) 1.71 � 0.9 2.33 � 0.95 0.051

Unipolar voltage (mV) 2.36 � 1.26 3.45 � 1.36 0.012

LA lateral

Bipolar voltage (mV) 1.84 � 0.90 2.54 � 0.85 0.016

Unipolar voltage (mV) 2.70 � 1.13 3.85 � 1.43 0.007

LA inferior

Bipolar voltage (mV) 1.79 � 1.07 2.48 � 1.19 0.060

Unipolar voltage (mV) 2.96 � 1.35 4.20 � 1.63 0.013

LA posterior

Bipolar voltage (mV) 1.34 � 0.57 2.18 � 0.77 <0.001

Unipolar voltage (mV) 2.23 � 0.89 3.59 � 1.19 <0.001

Values are mean � SD.

Abbreviations as in Tables 1 and 2.

TABLE 4 PV Antrum and Posterior Left Atrium: Electrophysiological and

Electroanatomic Parameters

HF Group(n ¼ 20)

NLV Group(n ¼ 20)

pValue

PVCL

PV4PVAverage (4-vein average over 100 cycles) (ms) 185 � 27 165 � 19 0.016

PVFVAverage (fastest vein over 100 cycles) (ms) 172 � 24 155 � 17 0.013

PVFast (shortest CL of any vein) (ms) 126 � 23 94 � 17 <0.001

LAA average (ms) 170 � 18 166 � 15 0.45

PV4PVAverage (4-vein average)/LAA average 1.07 � 0.09 0.99 � 0.10 0.030

PVFPVAverage (fastest vein average)/LAA average 1.01 � 0.10 0.93 � 0.10 <0.001

PV4PVAverage/LAA average >1 (% of patients) 95 45 <0.001

PVFPVAverage/LAA average >1 (% of patients) 60 25 0.025

Electroanatomic mapping of PV antrum

Bipolar voltage (mV) 1.18 � 0.51 2.00 � 0.68 <0.001

Unipolar voltage (mV) 1.95 � 0.80 2.98 � 0.89 <0.001

Complex electrograms 40.0 � 15.4 14.5 � 14.0 <0.001

Low voltage (<0.5 mV bipolar) 25.4 � 24.5 9.5 � 11.0 0.012

Presence of any scar (<0.05 mV bipolar) (% of patients) 50 10 0.006

Conduction velocity (m/s) 0.93 � 0.20 1.05 � 0.19 0.07

Electroanatomic mapping of posterior left atrium

Bipolar voltage (mV) 1.34 � 0.57 2.18 � 0.77 <0.001

Unipolar voltage (mV) 2.23 � 0.89 3.59 � 1.19 <0.001

Low voltage 26.0 � 18.0 5.7 � 7.0 <0.001

Scarring 2.4 � 3.3 0.1 � 0.4 0.005

Presence of any scar (% of patients) 50 10 0.006

Complex electrograms 31.5 � 18.7 8.8 � 8.3 <0.001

Global conduction velocity (m/s) 0.89 � 0.23 0.98 � 0.19 0.17

Values are mean � SD or %.

CL ¼ cycle length; LAA ¼ left atrial appendage; PV ¼ pulmonary vein; PVCL ¼ pulmonary vein cycle length;other abbreviations as in Table 1.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8 Prabhu et al.J A N U A R Y 2 0 1 8 : 8 7 – 9 6 Atrial Remodeling in Heart Failure and AF

93

increase in biatrial complex fractionated activity;and slowing in PV electrical activity in AF thatcorrelated with voltage and complex atrial activity atthe PV antrum.

ATRIAL SUBSTRATE IN HF. Pathophysiologicalmechanisms responsible for AF and HF create acomplex interplay and “chicken-and-egg” relation-ship between 2 common cardiac conditions. Sanderset al. (13) elegantly demonstrated structural remod-eling in the RA in patients with HF in the absence ofAF. HF was associated with reduced bipolar voltage,increased low voltage and scarring, and conductionslowing (13). The present study supports these earlierfindings but extends mapping to the LA and PVs inthe persistent AF population.

Previous studies looking at atrial structuralremodeling as detected by atrial delayed gadoliniumenhancement have shown mixed results. Oakes et al.(14) and McGann et al. (15) demonstrated no rela-tionship between the extent of atrial delayedenhancement and systolic dysfunction. In contrast,Akkaya et al. (16) specifically compared the extent ofLA delayed enhancement between patients withLVEF >50% and those with LVEF #50% and foundsignificantly higher percentage of delayed enhance-ment in the reduced LVEF group. Furthermore, theextent of enhancement predicted the extent of LVrecovery. However, no studies performed invasiveassessment, and all included mixed etiologies of HF.Haldar et al. (17) demonstrated a significant increasein atrial fractionation in patients with persistent AFand HF of variable etiology. Although consistent withthe findings in the present study, their retrospective

analysis was performed during AF rather than in si-nus rhythm; HF etiology included ischemic andvalvular heart disease; and substrate analysis wasconfined to fractionation alone. In the present studyin patients with equivalent durations of persistentAF, HF was associated with biatrial electrical andstructural remodeling with reductions in tissuevoltage, low voltage, and complex atrial activity. Theadverse atrial remodeling demonstrated in the HFpopulation in the present study may be representa-tive of a primary global cardiomyopathy or may bedriven by AF in patients vulnerable to systolicdysfunction.

CV was preserved in the LA despite being signifi-cantly reduced in the RA in the HF population. Dif-ferences in atrial architecture, such as the cristaterminalis, may have enhanced conduction slowingin the RA and may explain the observed differencesbetween atrial chambers.

PV ANTRUM AND REGIONAL ANALYSIS. In the pre-sent study, we demonstrated a significant reductionin voltage and increased fractionation in all regions of

FIGURE 3 Relationship Between Antral Voltage and Complex Fractionated Activity With Pulmonary Vein Cycle Length

Correlation of antral bipolar tissue voltage (A) and antral complex electrograms (B) with average PVCL (average of all measured pulmonary

veins). Antral region is the area defined in Figure 1. PVCL ¼ pulmonary vein cycle length.

Prabhu et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8

Atrial Remodeling in Heart Failure and AF J A N U A R Y 2 0 1 8 : 8 7 – 9 6

94

the LA in HF, most notably in the posterior and septalLA. In addition, tissue voltage was significantly lowerin the PV antrum compared with surrounding non-antral sites with HF, a difference not observed in theNLV group. This study is the first to demonstrate thatthis remodeling in the PV antrum is associated with aslowing of PV electrical activity. This more extensiveremodeling of the PV antrum reinforces the role ofwide encirclement rather than ostial ablative strategyin this patient population. Cabrera et al. (18,19)demonstrated histological changes in the PV antralregion, particularly the interpulmonary and PV/LAAridge, predisposing to the spread of AF activity, and arecent clinical trial highlighted the clinical implica-tions of remodeling this region (20). Recent meta-analyses have demonstrated that a wide antralapproach to pulmonary vein isolation (PVI) may bemore effective in achieving long-term freedom fromAF in patients with persistent AF (21,22).

Roberts-Thomson et al. (23) demonstrated theimportance of the posterior LA in AF in an open chestepicardial human study in patients with structuralheart disease and LV dysfunction. CMR-detectedatrial late gadolinium enhancement has similarlybeen shown to predominate in the posterior LA(14,15). These findings have important clinical impli-cations and may in part explain a recent observationfrom a randomized study of catheter ablation versusamiodarone in AF and HF. A retrospective analysis

demonstrated that catheter ablation was more suc-cessful if PVI included posterior LA isolationcompared with PVI alone (79% vs. 26%; p < 0.001) inpatients with HF (24). The presence of low-voltageregions within the PV antrum and posterior LA pro-vides mechanistic support to the observation ofimproved outcomes when isolation of the posteriorLA is included with PVI during catheter ablation forAF in HF.

PV AF cycle length was significantly slower in HFpatients with persistent AF, a finding that to ourknowledge has not been previously demonstrated.Pascale et al. (25) demonstrated that the shortestmeasured PVCL relative to LAA AF cycle length waspredictive of recurrence after AF ablation. Walterset al. (26) demonstrated that the AF cycle lengthwithin the coronary sinus correlated with LA struc-tural remodeling (bipolar voltage, CV, and complexelectrograms) in patients with long-standing persis-tent AF. However, their study did not specificallycorrelate structural remodeling with PVCL and didnot include patients with HF. Intriguingly, in thepresent study there was a significant relationshipbetween PVCL and PV antral voltage. Lower tissuevoltage was associated with slowing of PVCL in AF inHF, and one may speculate that reduced PV firing andmore extensive atrial substrate may provide someinsight into the reported lower success of PVI inpatients with HF (8,27).

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: Both AF and HF

exert adverse remodeling influences on the atria; however, the

contribution of systolic dysfunction, over and above that of AF

itself, is unknown. Concurrent systolic impairment and

persistent AF is associated with a significantly more advanced

atrial substrate than in persistent AF without systolic

impairment. This may explain the poorer outcomes seen

with rhythm control strategies in patients with systolic HF

and AF.

TRANSLATIONAL OUTLOOK: Patients with concurrent AF

and systolic dysfunction may require more extensive treatment

approaches, including aggressive risk-factor modification,

additional substrate modification, and other anti-HF treatments

to optimize rhythm control success. The reversibility of this

additional remodeling after improvement in systolic function

remains to be determined.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8 Prabhu et al.J A N U A R Y 2 0 1 8 : 8 7 – 9 6 Atrial Remodeling in Heart Failure and AF

95

CLINICAL IMPLICATIONS. In the present study, pa-tients with persistent AF and HF demonstrated bia-trial electrical and structural changes that may in partexplain the reduced success of pharmacotherapy andcatheter ablation compared to patients without sys-tolic impairment (27,28). This has important impli-cations, as recent studies have suggested thatcatheter ablation in HF may have additional benefitsof recovery of systolic dysfunction after restoration ofsinus rhythm (4,6,29–31). Ling et al. (29) demon-strated that in patients with AF and systolicdysfunction without delayed ventricular enhance-ment on CMR, LV function normalized after success-ful catheter ablation. In the STAR AF II (Substrate andTrigger Ablation for Reduction of Atrial Fibrillation II)trial, there was no significant difference in ablationoutcomes for persistent AF with PVI versus PVI plussubstrate modification; however, the proportion ofpatients with HF was just 4% (32). The presence of aslower PVCL and more extensive atrial substrate mayrequire an alternate ablation strategy to improveoutcomes in this population.

To date, no randomized trials have comparedablation strategies in patients with normal andreduced ejection fraction, although importantly,substrate modification strategies such as posteriorwall isolation have yet to be prospectively evaluatedin a randomized fashion. The AATAC-AF (AblationVersus Amiodarone for Treatment of Persistent AtrialFibrillation in Patients With Congestive Heart Failureand an Implanted Device) trial demonstrated thesuperiority of catheter ablation over amiodarone inpatients with AF and HF (24). Retrospective analysisidentified improved outcomes when posterior wallisolation was included beyond PVI alone (79% vs.36%; p < 0.001) (24). This present study demon-strating extensive biatrial substrate in the AF/HFpopulation offers a pathophysiological explanation tosupport ongoing research into substrate modificationstrategies in this specific patient population.

The aim of the present study was to determine thebiatrial substrate in patients with persistent AF andHF not explained by ischemia or valvular dysfunction.A significant proportion of HF patients in this studyhad an underlying arrhythmia-mediated cardiomy-opathy with normalization of LVEF after restoration ofsinus rhythm with catheter ablation. These patientsdisplayed less advanced atrial substrate compared tothose in whom LV function failed to normalize afterablation. This may explain the improved freedomfrom AF with catheter ablation in patients witharrhythmia-mediated cardiomyopathy compared withknown structural heart disease (31,33).

STUDY LIMITATIONS. The exclusion of patients withstructural heart disease such as ischemic or valvularheart disease may limit the extrapolation of the studyfindings to these cardiac conditions. In addition, pa-tients with paroxysmal AF were not included. Thismechanistic study was designed to characterize thebiatrial substrate in patients with persistent AF andHF, rather than outcome with catheter ablation. Useof a 3.5-mm bipolar ablation catheter with a contactforce >10 g may have resulted in both undersensing oflocal electrograms and oversensing of far-field elec-trograms at the LA septum. The cutoff value of <0.5mV for low voltage, although consistent with othersubstrate analysis studies, has not yet been histo-logically validated as a marker of structural remod-eling in humans.

CONCLUSIONS

HF is associated with significant reductions in biatrialtissue voltage, fractionation, and prolongation of AFcycle length within the PVs. More advanced biatrialremodeling may have implications for ablation stra-tegies and explain the relative ineffectiveness ofantiarrhythmic therapy in patients with AF and HF.

ADDRESS FOR CORRESPONDENCE: Prof. Peter M.Kistler, Baker Heart and Diabetes Institute, 75Commercial Road, Melbourne, Victoria 3004,Australia. E-mail: peter.kistler@bakeridi.edu.au.

Prabhu et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 1 , 2 0 1 8

Atrial Remodeling in Heart Failure and AF J A N U A R Y 2 0 1 8 : 8 7 – 9 6

96

RE F E RENCE S

1. Chugh SS, Havmoeller R, Narayanan K, et al.Worldwide epidemiology of atrial fibrillation: aGlobal Burden of Disease 2010 Study. Circulation2014;129:837–47.

2. Wolowacz SE, Samuel M, Brennan VK, Jasso-Mosqueda JG, Van Gelder IC. The cost of illness ofatrial fibrillation: a systematic review of the recentliterature. Europace 2011;13:1375–85.

3. Anter E, Jessup M, Callans DJ. Atrial fibrillationand heart failure: treatment considerations for adual epidemic. Circulation 2009;119:2516–25.

4. Hunter RJ, Berriman TJ, Diab I, et al.A randomized controlled trial of catheter ablationversus medical treatment of atrial fibrillation inheart failure (the CAMTAF trial). Circ ArrhythmElectrophysiol 2014;7:31–8.

5. Gentlesk PJ, Sauer WH, Gerstenfeld EP, et al.Reversal of left ventricular dysfunction followingablation of atrial fibrillation. J Cardiovasc Elec-trophysiol 2007;18:9–14.

6. Jones DG, Haldar SK, Hussain W, et al.A randomized trial to assess catheter ablationversus rate control in the management of persis-tent atrial fibrillation in heart failure. J Am CollCardiol 2013;61:1894–903.

7. Nedios S, Sommer P, Dagres N, et al. Long-termfollow-up after atrial fibrillation ablation in pa-tients with impaired left ventricular systolic func-tion: the importance of rhythm and rate control.Heart Rhythm 2014;11:344–51.

8. Cha YM, Wokhlu A, Asirvatham SJ, et al. Successof ablation for atrial fibrillation in isolated leftventricular diastolic dysfunction: a comparison tosystolic dysfunction and normal ventricular func-tion. Circ Arrhythm Electrophysiol 2011;4:724–32.

9. McLellan AJ, Schlaich MP, Taylor AJ, et al.Reverse cardiac remodeling after renal denerva-tion: atrial electrophysiologic and structuralchanges associated with blood pressure lowering.Heart Rhythm 2015;12:982–90.

10. Pascale P, Shah AJ, Roten L, et al. Pulmonaryveins to left atrium cycle length gradient predictsprocedural and clinical outcomes of persistentatrial fibrillation ablation. Circ Arrhythm Electro-physiology 2014;7:473–82.

11. Li C, Zhang J, Zhou C, Huang L, Tang H, Rao L.Will simultaneous measurement of E/e’ indexfacilitate the non-invasive assessment of leftventricular filling pressure in patients with non-valvular atrial fibrillation? Eur J Echocardiogr2010;11:296–301.

12. Kusunose K, Yamada H, Nishio S, et al. Clinicalutility of single-beat E/e’ obtained by simulta-neous recording of flow and tissue Doppler ve-locities in atrial fibrillation with preserved systolicfunction. J Am Coll Cardiol Img 2009;2:1147–56.

13. Sanders P, Morton JB, Davidson NC, et al.Electrical remodeling of the atria in congestiveheart failure: electrophysiological and electro-anatomic mapping in humans. Circulation 2003;108:1461–8.

14. Oakes RS, Badger TJ, Kholmovski EG, et al.Detection and quantification of left atrial struc-tural remodeling with delayed-enhancementmagnetic resonance imaging in patients withatrial fibrillation. Circulation 2009;119:1758–67.

15. McGann C, Akoum N, Patel A, et al. Atrialfibrillation ablation outcome is predicted by leftatrial remodeling on MRI. Circ Arrhythm ia Elec-trophysiol 2014;7:23–30.

16. Akkaya M, Higuchi K, Koopmann M, et al.Higher degree of left atrial structural remodelingin patients with atrial fibrillation and left ventric-ular systolic dysfunction. J Cardiovasc Electro-physiol 2013;24:485–91.

17. Haldar SK, Jones DG, Khan H, et al. Charac-terising the difference in electrophysiologicalsubstrate and outcomes between heart failure andnon-heart failure patients with persistent atrialfibrillation. Europace 2017;Jan 20:pii euw380.

18. Cabrera JA, Ho SY, Climent V, Sanchez-Quintana D. The architecture of the left lateralatrial wall: a particular anatomic region with im-plications for ablation of atrial fibrillation. EurHeart J 2008;29:356–62.

19. Cabrera JA, Ho SY, Climent V, Fuertes B,Murillo M, Sánchez-Quintana D. Morphological ev-idence of muscular connections between contig-uous pulmonary venous orifices: relevance of theinterpulmonary isthmus for catheter ablation inatrial fibrillation. Heart Rhythm 2009;6:1192–8.

20. McLellan AJA, Ling L-H, Azzopardi S, et al.A minimal or maximal ablation strategy to achievepulmonary vein isolation for paroxysmal atrialfibrillation: a prospective multi-centre randomizedcontrolled trial (the Minimax study). Eur Heart J2015;36:1812–21.

21. Parkash R, Tang AS, Sapp JL, Wells G.Approach to the catheter ablation technique ofparoxysmal and persistent atrial fibrillation: ameta-analysis of the randomized controlled trials.J Cardiovasc Electrophysiol 2011;22:729–38.

22. Proietti R, Santangeli P, Di Biase L, et al.Comparative effectiveness of wide antral versusostial pulmonary vein isolation: a systematic re-view and meta-analysis. Circ Arrhythm Electro-physiol 2014;7:39–45.

23. Roberts-Thomson KC, Stevenson IH,Kistler PM, et al. Anatomically determined func-tional conduction delay in the posterior left atriumrelationship to structural heart disease. J Am CollCardiol 2008;51:856–62.

24. Di Biase L, Mohanty P, Mohanty S, et al.Ablation Versus Amiodarone for Treatment ofPersistent Atrial Fibrillation in Patients WithCongestive Heart Failure and an Implanted Device:results from the AATAC multicenter randomizedtrial. Circulation 2016;133:1637–44.

25. Pascale P, Shah AJ, Roten L, et al. Pulmonaryveins to left atrium cycle length gradient predictsprocedural and clinical outcomes of persistentatrial fibrillation ablation. Circ Arrhythm Electro-physiol 2014;7:473–82.

26. Walters TE, Teh AW, Spence S, Kistler PM,Morton JB, Kalman JM. Relationship between theelectrocardiographic atrial fibrillation cycle lengthand left atrial remodeling: a detailed electro-anatomic mapping study. Heart Rhythm 2014;11:670–6.

27. Ullah W, Ling LH, Prabhu S, et al. Catheterablation of atrial fibrillation in patients with heartfailure: impact of maintaining sinus rhythm onheart failure status and long-term rates of strokeand death. Europace 2016;18:679–86.

28. Roy D, Talajic M, Nattel S, et al. Rhythmcontrol versus rate control for atrial fibrillationand heart failure. N Engl J Med 2008;358:2667–77.

29. Ling LH, Taylor AJ, Ellims AH, et al. Sinusrhythm restores ventricular function in patientswith cardiomyopathy and no late gadoliniumenhancement on cardiac magnetic resonanceimaging who undergo catheter ablation foratrial fibrillation. Heart Rhythm 2013;10:1334–9.

30. Anselmino M, Matta M, D’Ascenzo F, et al.Catheter ablation of atrial fibrillation in patientswith left ventricular systolic dysfunction: a sys-tematic review and meta-analysis. Circ ArrhythmElectrophysiol 2014;7:1011–8.

31. Prabhu S, Ling LH, Ullah W, et al. The impact ofknown heart disease on long-term outcomes ofcatheter ablation in patients with atrial fibrillationand left ventricular systolic dysfunction: a multi-center international study. J Cardiovasc Electro-physiol 2016;27:281–9.

32. Verma A, Jiang CY, Betts TR, et al. Approachesto catheter ablation for persistent atrial fibrilla-tion. N Engl J Med 2015;372:1812–22.

33. Addison D, Farhad H, Shah RV, et al. Effect oflate gadolinium enhancement on the recovery ofleft ventricular systolic function after pulmonaryvein isolation. J Am Heart Assoc 2016;5:e003570.

KEY WORDS atrial fibrillation, atrialremodeling, catheter ablation,electroanatomic mapping, heart failure