The usefulness of catheter ablation even in patients with myocardial impairment: Myocardial extracellular volume assessed by preablation planning computed tomography cannot predict atrial fibrillation recurrence

Abstract

Background

Increased left ventricular extracellular volume (ECV) measured by cardiac magnetic resonance imaging is associated with myocardial damage and has been considered to predict atrial fibrillation (AF) recurrence after catheter ablation (CA). However, recent reports suggest that AF recurrence is infrequent even with high ECV owing to advancing ablation technology.

Objective

This study aimed to evaluate the relationship between ECV quantified by computed tomography (CT-ECV), commonly performed before CA, and AF recurrence.

Methods

Consecutive 467 patients undergoing their first CA for AF at our hospital between January 2021 and June 2023 received pre-CA contrast-enhanced CT. The relationship between CT-ECV and AF recurrence within 400 days after CA was examined.

Results

AF recurrence occurred in 77 patients (16.5%), and CT-ECV in the nonrecurrence group was not significantly different from that in the recurrence group (28.0% vs 28.1%; P = .502). In the Cox proportional hazards model, CT-ECV did not predict AF recurrence (adjusted hazard ratio 1.018; P = .376). After dividing patients into low and high CT-ECV groups based on median CT-ECV (28.0%), Kaplan–Meier analysis showed no significant difference in AF recurrence between the groups (log-rank P = .727). Brain natriuretic peptide level, left ventricular ejection fraction, and left atrial volume index significantly improved even in the high CT-ECV group without AF recurrence.

Conclusion

CT-ECV was not significantly associated with AF recurrence within 400 days after the first CA. Regardless of the severity of ventricular myocardial damage, CA for AF is beneficial for preventing AF and improving cardiac function.

Graphical abstract

Key Findings.

• ▪Left ventricular extracellular volume (ECV) quantified by preablation planning computed tomography (CT-ECV) could not predict atrial fibrillation (AF) recurrence after catheter ablation (CA), with emphasis on pulmonary vein isolation.
• ▪When patients with AF were divided into 2 groups by CT-ECV median values (28%), 35%, 40%, 50%, and 60%, no significant difference in AF recurrence was observed after CA between high and low CT-ECV groups.
• ▪These findings suggest that, regardless of the extent of the left ventricular damage, the elimination of AF triggers is critical for maintaining sinus rhythm.

Introduction

Catheter ablation (CA) for atrial fibrillation (AF) is a common procedure in daily practice owing to its improved outcomes and reduced complications. However, thorough preoperative patient evaluation and assessment are crucial for safe and reliable, recurrence-free CA. Recently, preoperative evaluation using computed tomography (CT) has become common. CT not only provides a 3-dimensional (3D) model of the pulmonary veins (PVs) and left atrium but also enables the evaluation of the esophagus location, venous access routes, coronary arteries, and thrombus presence when contrast is applied. Guidelines recommend transesophageal echocardiography to exclude the left atrial thrombus before CA.¹ However, delayed-phase contrast-enhanced CT has demonstrated a 100% negative predictive value for ruling out left atrial thrombus.² Therefore, instead of transesophageal echocardiography, contrast-enhanced CT is increasingly used to evaluate intracardiac thrombus. Moreover, contrast-enhanced CT can assess cardiac damage, such as increased interstitial tissue including fibrosis and decreased myocardial cell volume by quantifying the extracellular volume (ECV) of the myocardium. Numerous studies have reported the utility of left ventricular CT-quantified ECV (CT-ECV) in evaluating and screening for cardiomyopathy, including cardiac amyloidosis.3, 4, 5, 6 In our institution, we have also measured ECV in all necessary and possible patients with arrhythmia using preablation planning CT to evaluate and identify the patient’s background and myocardial damage.³

In contrast, left ventricular fibrosis has been observed in patients with AF,⁷ with evidence suggesting that longer AF duration are associated with more extensive fibrosis.^8,9 Furthermore, cardiac magnetic resonance imaging (MRI)–quantified ECV (MRI-ECV) elevation, as represented by fibrosis, has been linked to AF recurrence after CA.^10,11 However, conflicting reports indicate that MRI-ECV may not reliably predict AF recurrence after CA.¹² In addition, it is challenging to measure ECV by MRI during irregular heart rate in patients with AF. Previous reports suggested that CT-ECV is reliable even in patients with AF owing to its rapid image acquisition.¹³ Given the logistical challenge of performing MRI in all patients before CA, there is a need to evaluate ECV using preablation planning CT before CA in more consecutive patients to assess its association with AF recurrence. To the best of our knowledge, this relationship has not been previously reported.

Methods

Study design and study population

This retrospective cohort study included 581 consecutive patients who underwent their first sessions of CA for AF at Kumamoto University Hospital using the EnSite 3D mapping system (Abbott, Chicago, IL) or the CARTO 3D mapping system (Biosense Webster, Irvine, CA) between January 2021 and June 2023. Patients with a follow-up period of less than 90 days, missing CT-ECV data, or intraoperative complications preventing PV isolation were excluded. To evaluate the pure effects of CA on AF, patients who had a possibility of having masked AF recurrence owing to the use of antiarrhythmic drugs were also excluded. Ultimately, 467 patients with AF (269 paroxysmal AF [PAF] and 198 persistent AF [PeAF]) were enrolled in this study (Supplemental Figure 1). The types of AF, PAF, and PeAF were determined according to the 2017 expert consensus.¹ The study was conducted according to the principles outlined in the Declaration of Helsinki and was approved by the institutional review board and ethics committee of Kumamoto University (approval no. 1393). The requirement for informed consent was waived owing to the low-risk nature of this retrospective study, and direct consent could not be obtained from all subjects. Therefore, we widely advertised the study protocol at Kumamoto University Hospital and on our website (http://www.kumadai-junnai.com), providing patients with the opportunity to opt out.

Study protocol

Before CA for AF, all patients without severe renal dysfunction underwent contrast-enhanced CT for constructing the 3D model of the PVs and left atrium, ruling out intracardiac thrombus, assessing the access route, locating the esophagus, and detecting coronary artery stenosis. Furthermore, the CT-ECV was quantified in these patients. The primary endpoint was AF recurrence within 400 days after CA. The relationship between the proportion of patients with AF recurrence and CT-ECV was also evaluated.

CT imaging protocol

The preablation CT imaging method used at our institution has previously been reported⁴ and is shown in Figure 1 and Supplemental Appendix. A 320 × 0.5 mm detector-row CT unit (Aquilion One Genesis edition; Canon Medical Systems, Tochigi, Japan) was used for CT examinations. Firstly, precontrast CT scans were performed. After the precontrast cardiac CT scan, the first contrast injection (contrast material dose of 450 mg iodine/kg, Iopamiron 370 [370 mg iodine/mL]; Bayer Healthcare) was performed, and an arterial phase CT scan was conducted. Additional contrast was injected immediately afterward (a total dose of approximately 550 mg iodine/kg), and delayed-phase CT scans were conducted 7 minutes after the first contrast injection.

Figure 1.

Cardiac CT angiography protocol and extracellular volume analysis. This figure illustrates the timeline of the cardiac CT protocol used in this study. The left-side CT image shows a precontrast short-axis multiplanar reconstruction image. The right-side CT image shows a postcontrast delayed-phase short-axis multiplanar reconstruction image. The regions of interest (ROIs) are placed on the septal midventricular wall (red dotted circle) and left ventricular blood pool (yellow dotted circle) in the pre- and postcontrast delayed phases, respectively. CT-ECV was calculated as follows: CT-ECV (%) = (1 − hematocrit) × (Δ myocardial density in the myocardium) / (Δ blood pool density in the left ventricle) × 100. Δ myocardial density in the myocardium = myocardial Hounsfield units (delayed-phase cardiac CT − precontrast cardiac CT) and Δ blood pool density in the left ventricle = blood pool Hounsfield units (delayed-phase cardiac CT − precontrast cardiac CT). CT = computed tomography; CT-ECV = computed tomography–quantified extracellular volume.

CT-ECV measurement

Myocardial CT-ECV was quantified on pre- and postcontrast delayed-phase CT images using the subtraction method. Regions of interest (ROIs) were manually drawn on the septal midventricular wall and left ventricular blood pool in the left ventricle short-axis multiplanar reformation images (Figure 1). ROIs were initially drawn on the delayed-phase images and then copied to the corresponding precontrast short-axis multiplanar reconstruction images. The CT-ECV was calculated as follows: CT-ECV (%) = (1 − hematocrit) × (Δ myocardial density in the myocardium) / (Δ blood pool density in the left ventricle) × 100, where Δ myocardial density in the myocardium = myocardial Hounsfield units (delayed-phase cardiac CT − precontrast cardiac CT), and Δ blood pool density in the left ventricle = blood pool Hounsfield units (delayed-phase cardiac CT − precontrast cardiac CT). For the CT-ECV measurements, the septal segment was chosen because it had demonstrated improved accuracy in previous studies using a septal ROI.¹⁴

Regarding atrial CT-ECV, the atrial wall is extremely thin, making accurate visualization challenging with current CT spatial and contrast resolution. Furthermore, creating CT-ECV requires subtracting precontrast CT and delayed-phase contrast CT images. The thin atrial wall is prone to misregistration during subtraction and also makes ROI measurement difficult. Furthermore, thin structures within the plane are susceptible to the partial volume effect, where surrounding structures (in the case of the left atrium, the left atrial lumen) influence CT number (Hounsfield units), making accurate CT measurements challenging. For these reasons, measuring atrial CT-ECV is difficult, and the CT-ECV measured in the ventricular septum was used in this study.

CA

Extensive encircling of the ipsilateral PV isolation (EEPVI) was initially performed using contact force–sensing, open-irrigation radiofrequency catheter (TactiCath sensor-enabled catheter, Abbott, or ThermoCool Smart Touch SF catheter, Biosense Webster) during radiofrequency ablation or a cryoballoon catheter (Arctic Front Advance Pro; Medtronic, Minneapolis, MN) during cryoablation, with the support of a 3D mapping system. During radiofrequency ablation, the target lesion size index was set to 5.2 using the EnSite system,¹⁵ and the target ablation index was set to 350–400 for the posterior wall and 400–450 for the nonposterior wall using the CARTO system.¹⁶ The radiofrequency tags were set at 4 mm and overlapped with reference to the CLOSE protocol.¹⁷ In the cryoballoon ablation, cryoablation was applied to all PVs for 180–240 seconds while monitoring the esophageal temperature and phrenic nerve pacing capture. After the EEPVI was performed, 10–20 μg isoproterenol (Proternol; Kowa, Nagoya, Japan) and 20 mg adenosine triphosphate (Adephos; Kowa) were intravenously administered to identify non-PV AF triggers.¹ When non-PV AF triggers were detected, additional ablation, such as superior vena cava (SVC) isolation (including the persistent left SVC isolation), posterior wall box isolation, or focal ablation of the non-PV AF trigger site, was performed. Cavotricuspid isthmus (CTI) ablation was also performed if CTI-dependent atrial flutter was documented, whereas linear ablation was performed for non-CTI-dependent atrial flutter. Prophylactic SVC isolation or CTI ablation was performed at the operator’s discretion. In the cryoablation strategy, ablations other than EEPVI were performed using a cryoablation catheter (Freezor MAX; Medtronic) or an open-irrigation radiofrequency catheter (FlexAbility, Abbott). If AF persisted after EEPVI, cardioversion was performed. After EEPVI, a left atrial voltage map was generated under sinus rhythm or atrial pacing using a multielectrode mapping catheter (Advisor HD Grid mapping catheter, Abbott; Pentaray mapping catheter, Biosense Webster; or Octaray mapping catheter, Biosense Webster), and EEPVI completion was confirmed. During the second session of CA for AF recurrence, reconnection of the PVs was confirmed using a multielectrode mapping catheter, and PV reisolation was conducted.

Clinical variables

The definition of cardiomyopathy, CHA₂DS₂-VASc score, heart failure, hypertension, diabetes mellitus, and significant valvular disease are presented in the Supplemental Appendix.

Follow-up

The 12-lead electrocardiogram (ECG) was performed at 1, 3, 6, and 12 months after CA and at unscheduled visits with signs or symptoms of arrhythmia recurrence. One-week Holter ECG monitor was recommended for all patients 6 and 12 months after the CA. Furthermore, AF recurrence was evaluated using device interrogations in patients with cardiac implantable electronic devices (CIEDs). As a result, rhythm checks using a 1-week Holter ECG monitor or CIED interrogation were performed in 397 patients (85.0%). Given that the outpatient clinic of 12 months after CA was conducted within the 400 days after CA for almost all of the study participants, follow-up duration was set at 400 days in this study. AF recurrence was defined as atrial arrhythmic events including AF, atrial flutter, and atrial tachycardia lasting more than 30 seconds.¹ Regarding the antiarrhythmic drugs, class I, III, and IV (bepridil), according to the Vaughan Williams classification, were defined as antiarrhythmic drugs in this study; however, class II antiarrhythmic drugs (β-blockers) were excluded owing to their common prescription for conditions such as heart failure, cardiomyopathy, and hypertension. Antiarrhythmic drugs were discontinued after a 3-month blanking period after CA, if possible.

Statistical analysis

The Shapiro–Wilk test was used to assess the normality of all continuous variables. Continuous variables are expressed as mean ± standard deviation or median (interquartile range), whereas categorical data are expressed as numbers with percentages. Normally distributed continuous variables were analyzed using the Student t test for independent samples and the paired t test for paired samples. Non-normally distributed continuous variables were analyzed using the Mann–Whitney U test for independent samples and the Wilcoxon signed-rank test for paired samples. Categorical variables were analyzed using the chi-square test; however, if 1 or more expected values were less than 5, Fisher’s exact test was conducted. Univariate and multivariate analyses were performed using the Cox proportional hazards model, with hazard ratios (HRs) expressed as 95% confidence intervals (CIs). Variables with P < .1 in the univariate analysis were included in the multivariate analysis. Variables associated with CT-ECV expansion were also included in multivariate analysis for baseline adjustment. Covariates with significant multicollinearity were excluded from the multivariate analysis. Kaplan–Meier analysis was performed using the log-rank test. Statistical significance was set at P < .05 for all analyses. All analyses were performed using IBM SPSS Statistics for Windows, version 25 (IBM Corp).

Results

Baseline clinical characteristics

The baseline characteristics of the study participants are presented in Table 1. Compared with patients with PAF, brain natriuretic peptide (BNP) and left atrial volume index (LAVI) were significantly higher in patients with PeAF. In addition, CT-ECV in patients with PeAF was significantly higher than in patients with PAF (27.6% [25.2%–30.0%] [PAF] vs 28.9% [26.0%–31.1%] [PeAF]; P = .004). The prevalence of male sex, heart failure, β-blocker use, and significant valvular disease was higher in patients with PeAF than in those with PAF. The left ventricular ejection fraction (LVEF) was significantly lower in patients with PeAF than in those with PAF. Cryoablation was more likely to be performed in patients with PAF than in those with PeAF. Among the 467 eligible patients, 77 (16.5%) experienced AF recurrence during a median follow-up period of 380 days (370–400). However, AF recurrence rate in patients with PeAF was higher than in patients with PAF, but the difference was not significant (13.8% [PAF] vs 20.2% [PeAF]; P = .064).

Table 1.

Baseline characteristics of the study participants (N = 467)

Variable	All (N = 467)	PAF (n = 269, 57.6%)	PeAF (n = 198, 42.4%)	P value
Baseline clinical variables
Age, y	70.0 (60.0–75.0)	70.0 (60.5–75.0)	69.0 (60.0–75.0)	.445
Men, n (%)	303 (64.9)	160 (59.5)	143 (72.2)	.004
BMI, kg/m2	24.0 (21.8–26.7)	23.8 (21.5–26.2)	24.2 (22.5–27.0)	.063
Cardiomyopathy,∗ n (%)	47 (10.1)	23 (8.6)	24 (12.1)	.205
Cardiac amyloidosis, n (%)	15 (3.2)	6 (2.2)	9 (4.5)	.161
CHA2DS2-VASc score†	2.0 (1.0–4.0)	2.0 (1.0–3.5)	3.0 (2.0–4.0)	.127
Heart failure, n (%)	106 (22.7)	19 (7.1)	87 (43.9)	<.001
Hypertension, n (%)	291 (62.3)	161 (59.9)	130 (65.7)	.201
Diabetes mellitus, n (%)	79 (16.9)	43 (16.0)	36 (18.2)	.531
Previous stroke, n (%)	40 (8.6)	24 (8.9)	16 (8.1)	.748
Creatinine clearance, mL/min	71.5 (56.1–94.1)	71.9 (57.0–95.0)	70.1 (55.0–89.0)	.278
BNP, pg/mL	58.1 (19.5–116.1)	27.3 (11.6–65.5)	105.7 (65.6–156.7)	<.001
β-blocker use, n (%)	171 (37.3)	70 (26.7)	101 (51.3)	<.001
Echocardiogram
LAVI, mL/m2	41.1 (34.1–52.6)	36.6 (30.2–45.3)	48.5 (39.7–59.4)	<.001
LVEF, %	60.3 (54.6–64.4)	62.4 (58.5–66.0)	56.0 (47.0–60.6)	<.001
Significant valvular disease,‡ n (%)	33 (7.1)	12 (4.5)	21 (10.6)	.010
Computed tomography
CT-ECV, %	28.0 (25.6–30.2)	27.6 (25.2–30.0)	28.9 (26.0–31.1)	.004
Catheter ablation
EEPVI, n (%)	467 (100.0)	269 (100.0)	198 (100.0)	-
CTI ablation, n (%)	127 (27.2)	75 (27.9)	52 (26.3)	.698
SVC isolation, n (%)	80 (17.1)	45 (16.7)	35 (17.7)	.788
Linear or box ablation, n (%)	11 (2.4)	5 (1.9)	6 (3.0)	.300
Cryoablation, n (%)	63 (13.5)	56 (20.8)	7 (3.5)	<.001
Non-PV triggers, n (%)	19 (4.1)	12 (4.5)	7 (3.5)	.617
Postprocedure variables
AF recurrence, n (%)	77 (16.5)	37 (13.8)	40 (20.2)	.064

Values are presented as median value with interquartile range or frequencies and percentages (%).

AF = atrial fibrillation; BMI = body mass index; BNP = brain natriuretic peptide; CT = computed tomography; CTI = cavo-tricuspid isthmus; ECV = extracellular volume; EEPVI = extensive encircling of the ipsilateral pulmonary vein isolation; LAVI = left atrial volume index; LVEF = left ventricular ejection fraction; PAF = paroxysmal atrial fibrillation; PeAF = persistent atrial fibrillation; PV = pulmonary vein; SVC = superior vena cava.

^∗Cardiomyopathy includes dilated cardiomyopathy, hypertrophic cardiomyopathy, old myocardial infarction, and cardiac amyloidosis.

^†CHA₂DS₂-VASc score was calculated by assigning 1 point each for ages 65–74 years, hypertension, diabetes mellitus, congestive heart failure, vascular disease, and female sex and 2 points for a previous stroke or transient ischemic attack and age ≥75 years.

^‡Significant valvular disease denotes moderate or severe mitral regurgitation and/or tricuspid regurgitation.

Clinical variables associated with AF recurrence

The clinical variables of patients with and without AF recurrence after CA are presented in Table 2. LAVI and BNP levels were significantly higher in the AF recurrence group than in the nonrecurrence group. In addition, β-blocker use was more frequent in the AF recurrence group. However, no significant differences were observed between the AF nonrecurrence and recurrence groups for CT-ECV (28.0% [25.6%–30.2%] [nonrecurrence group] vs 28.1% [25.7%–30.6%] [nonrecurrence group]; P = .502) or other variables including the prevalence of cardiomyopathy, heart failure, and CA procedures. The univariate and multivariate Cox proportional hazards models used to predict AF recurrence after CA are presented in Table 3. In univariate analysis, LAVI (HR 1.021; 95% CI 1.009–1.032; P < .001) and β-blocker use (HR 1.851; 95% CI 1.184–2.894; P = .007) were significant predictors of AF recurrence. However, in the multivariate analysis, only LAVI remained an independent predictor of AF recurrence (adjusted HR 1.021; 95% CI 1.007–1.035; P = .004). The CT-ECV was not an independent predictor of AF recurrence (adjusted HR 1.018; 95% CI 0.978–1.059; P = .376) (Table 3).

Table 2.

Clinical variables of patients with and without AF recurrence after catheter ablation (N = 467)

Variable	Recurrence (−) (n = 390, 83.5%)	Recurrence (+) (n = 77, 16.5%)	P value
Baseline clinical variables
Age, y	70.0 (60.0–75.0)	70.0 (60.0–75.0)	.930
Men, n (%)	253 (64.9)	50 (64.9)	.992
BMI, kg/m2	24.0 (21.9–26.7)	23.9 (21.4–26.5)	.551
Cardiomyopathy, n (%)	39 (10.0)	8 (10.4)	.917
Cardiac amyloidosis, n (%)	10 (2.6)	5 (6.5)	.083
CHA2DS2-VASc score	2.0 (1.0–4.0)	3.0 (2.0–4.0)	.370
Heart failure, n (%)	86 (22.1)	20 (26.0)	.453
Hypertension, n (%)	245 (62.8)	46 (59.7)	.610
Diabetes mellitus, n (%)	66 (16.9)	13 (16.9)	.993
Previous stroke, n (%)	30 (7.7)	10 (13.0)	.129
Creatinine clearance, mL/min	72.3 (56.9–94.2)	70.0 (51.0–92.6)	.203
AF type, PeAF, n (%)	158 (40.5)	40 (51.9)	.064
BNP, pg/mL	51.9 (16.8–116.2)	69.8 (29.4–117.8)	.034
β-blocker use, n (%)	132 (34.6)	39 (50.6)	.008
Echocardiogram
LAVI, mL/m2	39.9 (33.7–51.2)	46.4 (37.4–62.0)	<.001
LVEF, %	60.5 (54.7–64.5)	58.4 (52.4–64.2)	.171
Significant valvular disease, n (%)	26 (6.7)	7 (9.1)	.448
Computed tomography
CT-ECV, %	28.0 (25.6–30.2)	28.1 (25.7–30.6)	.502
Catheter ablation
EEPVI, n (%)	390 (100.0)	77 (100.0)	-
CTI ablation, n (%)	105 (26.9)	22 (28.6)	.766
SVC isolation, n (%)	70 (17.9)	10 (13.0)	.291
Linear or box ablation, n (%)	9 (2.3)	2 (2.6)	.565
Cryoablation, n (%)	54 (13.8)	9 (11.7)	.612
Non-PV triggers, n (%)	14 (3.6)	5 (6.5)	.189

Abbreviations as in Table 1.

Table 3.

Univariate and multivariate Cox proportional hazards models for predicting the AF recurrence after catheter ablation (N = 467)

Variable	Univariate analysis		Multivariate analysis (model 1)		Multivariate analysis (model 2)
	HR (95% CI)	P value	HR (95% CI)	P value	HR (95% CI)	P value
Age (per y)	0.999 (0.978–1.021)	.942	0.983 (0.961–1.006)	.155	0.985 (0.963–1.008)	.193
Men (yes)	1.037 (0.649–1.656)	.880
BMI (per kg/m2)	0.974 (0.917–1.036)	.405	0.969 (0.909–1.033)	.332	0.967 (0.907–1.031)	.304
Cardiomyopathy (yes)	1.045 (0.502–2.173)	.907	0.744 (0.309–1.794)	.510	0.876 (0.404–1.899)	.738
Cardiac amyloidosis∗ (yes)	2.299 (0.928–5.695)	.072	Not selected		Not selected
CHA2DS2-VASc score† (pre 1 point)	1.043 (0.904–1.202)	.565	Not selected		Not selected
Heart failure (yes)	1.216 (0.730–2.024)	.452	0.846 (0.480–1.492)	.564	0.882 (0.505–1.540)	.660
Hypertension (yes)	0.848 (0.538–1.338)	.480
Diabetes mellitus (yes)	0.985 (0.543–1.788)	.961	0.880 (0.477–1.626)	.684	0.909 (0.492–1.681)	.762
Previous stroke (yes)	1.610 (0.828–3.130)	.160
Creatinine clearance† (per mL/min)	0.995 (0.987–1.003)	.193	Not selected		Not selected
BNP‡ (per pg/mL)	1.001 (1.000–1.003)	.125	Not selected		Not selected
β-blocker use (yes)	1.851 (1.184–2.894)	.007	1.635 (0.990–2.700)	.055	1.572 (0.959–2.577)	.073
AF type‡ (PeAF, yes)	1.476 (0.944–2.308)	.088	Not selected		Not selected
LAVI (per mL/m2)	1.021 (1.009–1.032)	<.001	1.021 (1.007–1.035)	.004	1.023 (1.009–1.037)	.001
LVEF (per 1 %)	0.987 (0.967–1.008)	.231
Significant valvular disease (yes)	1.330 (0.611–2.892)	.472	0.715 (0.303–1.686)	.443	0.727 (0.309–1.712)	.466
CTI ablation (yes)	1.044 (0.637–1.712)	.864
SVC isolation (yes)	0.706 (0.363–1.371)	.304
Linear or box ablation (yes)	1.151 (0.283–4.687)	.845
Cryoablation (yes)	0.871 (0.435–1.746)	.697
High CT-ECV§ (yes)	1.083 (0.691–1.696)	.728			0.935 (0.581–1.503)	.780
CT-ECV (per 1 %)	1.017 (0.988–1.046)	.250	1.018 (0.978–1.059)	.376

CI = confidence interval; HR = hazard ratio. Other abbreviations as in Table 1.

High CT-ECV was defined as CT-ECV of ≥28.0% (median value).

^∗Cardiac amyloidosis was excluded from the multivariate analysis owing to the significant correlation with cardiomyopathy (correlation coefficient = 0.545).

^†CHA₂DS₂-VASc score and creatinine clearance were excluded from the multivariate analysis owing to the significant correlation with age (correlation coefficient = 0.703 and −0.755, respectively).

^‡BNP and AF type were excluded from the multivariate analysis owing to the significant correlation with LAVI (correlation coefficient = 0.559 and 0.423, respectively).

^§High CT-ECV was defined as CT-ECV of ≥28.0% (median value).

Analysis between the low and high CT-ECV groups

Table 4 compares the baseline characteristics of patients in low (CT-ECV <28.0% [median value]) and high CT-ECV groups (CT-ECV ≥28.0% [median value]). The patients in the high CT-ECV group were older; had lower body mass index and creatinine clearance; had higher frequencies of cardiomyopathy, cardiac amyloidosis, heart failure, diabetes mellitus, significant valvular disease, and PeAF; and had a higher CHA₂DS₂-VASc score, BNP levels, and LAVI. However, other clinical variables, including LVEF, CA strategy, and AF recurrence rate, were not significantly different between the low and high CT-ECV groups. The Kaplan–Meier analysis of freedom from AF recurrence after CA for AF is presented in Figure 2A. No significant difference was observed between the low and high CT-ECV groups (log-rank P = .727). Patients were also divided into low and high CT-ECV groups using cutoff values of 35%, 40%, 50%, and 60%, and the Kaplan–Meier analysis for each cutoff value is presented in Figure 3A, 3B, 3C, and 3D, respectively. Although the AF recurrence rates after CA in the high CT-ECV group were higher than in the low CT-ECV group at cutoff values of 40%, 50%, and 60%, the differences were not statistically significant (log-rank P = .801 for 35%, .179 for 40%, .461 for 50%, and .200 for 60%). Furthermore, Kaplan–Meier analysis was extended to 2 years (censored at 720 days) for patients with follow-up beyond 1 year. When the cutoff value for the CT-ECV value was not only 28% (Supplemental Figure 2A) but also 35%, 40%, 50%, and 60% (Supplemental Figure 3A–3D), no significant difference was observed in the recurrence rate of AF between the high and low CT-ECV groups (log-rank P = .628, .879, .155, .333, and .444, respectively).

Table 4.

Baseline characteristics of patients in low and high CT-ECV groups (N = 467)

Variable	Low CT-ECV group (CT-ECV of <28.0%, n = 224)	High CT-ECV group (CT-ECV of ≥28.0%, n = 243)	P value
Baseline clinical variables
Age, y	68.0 (58.3–74.0)	71.0 (62.0–76.0)	.002
Men, n (%)	153 (68.3)	150 (61.7)	.137
BMI, kg/m2	24.4 (22.4–26.7)	23.7 (21.2–26.7)	.043
Cardiomyopathy	13 (5.8)	34 (14.0)	.003
Cardiac amyloidosis, n (%)	0 (0)	15 (6.2)	<.001
CHA2DS2-VASc score	2.0 (1.0–3.0)	3.0 (2.0–4.0)	.002
Heart failure, n (%)	37 (16.5)	69 (28.4)	.002
Hypertension, n (%)	137 (61.2)	154 (63.4)	.622
Diabetes mellitus, n (%)	29 (12.9)	50 (20.6)	.028
Previous stroke, n (%)	17 (7.6)	23 (9.5)	.469
Creatinine clearance, mL/min	77.2 (60.2–102.3)	66.3 (52.3–88.1)	<.001
BNP, pg/mL	42.6 (12.3–82.7)	76.8 (24.8–137.5)	<.001
β-blocker use, n (%)	73 (33.0)	98 (41.2)	.071
AF type, n (PeAF, %)	80 (35.7)	118 (48.6)	.005
Echocardiogram
LAVI, mL/m2	38.8 (32.3–47.8)	43.8 (35.0–57.0)	<.001
LVEF, %	60.1 (55.7–64.5)	60.3 (53.7–64.4)	.647
Significant valvular disease, n (%)	9 (4.0)	24 (9.9)	.014
Computed tomography
CT-ECV, %	25.4 (24.0–26.8)	30.0 (29.0–32.8)	<.001
Catheter ablation
EEPVI, n (%)	224 (100.0)	243 (100.0)	-
CTI ablation, n (%)	58 (25.9)	69 (28.4)	.544
SVC isolation, n (%)	39 (17.4)	41 (16.9)	.877
Linear or box ablation, n (%)	5 (2.2)	6 (2.5)	.866
Cryoablation, n (%)	36 (16.1)	27 (11.1)	.117
Non-PV triggers, n (%)	12 (5.4)	7 (2.9)	.176
Postprocedure variables
AF recurrence, n (%)	35 (15.6)	42 (17.3)	.629

Abbreviations as in Table 1.

Figure 2.

Kaplan–Meier analysis of freedom from AF recurrence 400 days after the catheter ablation. Kaplan–Meier analysis illustrates the freedom from AF recurrence within 400 days after catheter ablation for AF in the low and high CT-ECV groups, using the cutoff value of the median (28%). Panels A, B, and C depict analyses for all patients, patients with paroxysmal AF, and patients with persistent AF, respectively. The blue line shows the low CT-ECV group, and the red line shows the high CT-ECV group. Adjusted HRs with CIs for the high CT-ECV group are compared with the low CT-ECV group as a reference. AF = atrial fibrillation; CI = confidence interval; CT-ECV = computed tomography–quantified extracellular volume; HR = hazard ratio.

Figure 3.

Kaplan–Meier analysis of freedom from AF recurrence 400 days after the catheter ablation with different cutoff values. Kaplan–Meier analysis demonstrates the freedom from AF recurrence within 400 days after catheter ablation for AF in the low and high CT-ECV groups, using cutoff values of 35% (A), 40% (B), 50% (C), and 60% (D). The blue line shows the low CT-ECV group, and the red line shows the high CT-ECV group. AF = atrial fibrillation; CT-ECV = computed tomography–quantified extracellular volume.

Subgroup analysis in patients with PAF and PeAF

The clinical variables of the patients with and without AF recurrence after CA in patients with PAF are presented in Supplemental Table 1. CT-ECV was not significantly different between the nonrecurrence and recurrence groups in patients with PAF (27.6% [25.0–29.8%] [nonrecurrence group] vs 27.6% [25.5–30.0%] [recurrence group]; P = .626). In both the univariate and multivariate Cox proportional hazards analyses, CT-ECV did not predict AF recurrence after CA for PAF (adjusted HR 1.001; 95% CI 0.925–1.084; P = .974) (Supplemental Table 2). After stratifying patients with PAF into low (CT-ECV <28.0% [median]) and high ECV groups (CT-ECV ≥28.0% [median]), AF recurrence was not significantly different between the groups (13.2% [low CT-ECV] vs 14.4% [high CT-ECV]; P = .775) (Supplemental Table 3). Kaplan–Meier analysis demonstrated no significant difference in freedom from AF recurrence after CA for PAF between the low and high CT-ECV groups (log-rank P = .879) (Figure 2B).

The clinical variables of patients with PeAF, with and without AF recurrence after CA, are presented in Supplemental Table 4. CT-ECV levels were not significantly different between the nonrecurrence and recurrence groups in patients with PeAF (28.9% [26.0%–31.1%] [nonrecurrence group] vs 28.7% [25.9%–31.6%] [recurrence group]; P = .867). In both the univariate and multivariate Cox proportional hazards analyses, CT-ECV did not predict the AF recurrence after CA in patients with PeAF (adjusted HR 0.967; 95% CI 0.913–1.024; P = .246) (Supplemental Table 5). Stratifying patients with PeAF into low (CT-ECV <28.0% [median]) and high ECV groups (CT-ECV ≥28.0% [median]) similarly revealed no significant difference in AF recurrence between the groups (20.0% [low CT-ECV] vs 20.3% [high CT-ECV]; P = .954) (Supplemental Table 6). Kaplan–Meier analysis confirmed no significant difference in freedom from AF recurrence rate after CA for PeAF between the low and high CT-ECV groups (log-rank P = .981) (Figure 2C).

Subsequently, Kaplan–Meier analysis for the extended follow-up was conducted in the PAF (Supplemental Figure 2B) and PeAF groups (Supplemental Figure 2C). No significant difference was observed in AF recurrence rate between the PAF (log-rank P = .856) and PeAF groups (log-rank P = .854).

Subgroup analysis in patients with cardiac amyloidosis, heart failure, and low LVEF

There were 12 patients with definite and 3 patients with probable diagnoses of transthyretin cardiac amyloidosis in this study. Among them, 14 patients (93.3%) were positive for technetium 99m pyrophosphate scintigraphy, 12 patients (80.0%) were positive for amyloid deposition in the biopsy sample, and none of the patients had a mutation in the transthyretin gene. In this amyloidosis subgroup, CT-ECV was not a predictor for AF recurrence within 400 days after their first session of CA (HR 1.009; 95% CI 0.940–1.084; P = .802) (Supplemental Table 7).

In addition, we conducted a primary outcome assessment in the heart failure subgroup (n = 106) and the low LVEF subgroup (preablation LVEF <50%) (n = 74). In both subgroups, there was no significant association between CT-ECV and AF recurrence (heart failure subgroup, HR 1.008, 95% CI 0.970–1.047, P = .687; low LVEF subgroup, HR 1.018, 95% CI 0.983–1.053, P = .318) (Supplemental Tables 8 and 9).

PV reconnection rate in the second session of CA

Among the 77 patients with AF recurrence, 35 underwent a second session of CA. Eight patients (22.9%) had a reconnection of PV. In the per-PV analysis, 10 of 140 PVs (7.1%) were reconnected. Dividing those 35 patients into low (CT-ECV <28.0%) and high CT-ECV groups (CT-ECV ≥28.0%), the PV reconnection rate was not significantly different between the groups (7 of 84 PVs [8.3%] [low CT-ECV] vs 3 of 56 PVs [5.4%] [high CT-ECV]; P = .377). All PVs were reisolated during the second session.

Changes in biochemical markers and echocardiographic parameters after CA

Blood biochemistry tests and echocardiography were conducted at 6 or 12 months post-CA. In the absence of recurrence, BNP level, LVEF, and LAVI significantly improved in the low CT-ECV group (Table 5 and Supplemental Figure 4). Furthermore, in the high CT-ECV group, BNP level, LVEF, and LAVI also significantly improved in the absence of recurrence. We also evaluated the relationship between CT-ECV and postablation LVEF improvement in patients with PeAF with low LVEF (preablation LVEF <50%) but without AF recurrence (n = 45). LVEF improvement was defined as LVEF of ≥50% at 6 or 12 months after CA and was observed, regardless of whether CT-ECV was high or low (70.6% [low CT-ECV] vs 78.6% [high CT-ECV]; P = .398).

Table 5.

Postoperative biochemical marker and echocardiographic parameter analysis

Low CT-ECV group (n = 224)
Variable	Recurrence (−) (n = 189)			Recurrence (+) (n = 35)
	Preablation	Postablation	P value	Preablation	Postablation	P value
BNP, pg/mL	39.8 (11.1–80.8)	16.3 (6.6–30.2)	<.001	57.0 (22.2–90.3)	35.1 (18.2–72.9)	.068
LVEF, %	60.5 (56.1–64.8)	62.1 (59.3–65.4)	<.001	58.4 (53.5–63.4)	59.5 (56.7–65.5)	.035
LAVI, mL/m2	37.9 (30.6–46.7)	32.5 (26.8–40.8)	<.001	46.1 (36.0–56.0)	38.1 (32.3–49.2)	.002
High CT-ECV group (n = 243)
	Recurrence (−) (n = 201)			Recurrence (+) (n = 42)
Variable	Preablation	Postablation	P value	Preablation	Postablation	P value
BNP, pg/mL	75.2 (23.2–137.4)	32.1 (12.3–55.9)	<.001	87.5 (41.8–158.2)	55.2 (34.1–105.7)	.011
LVEF, %	60.6 (54.1–64.4)	62.8 (58.5–65.6)	<.001	59.0 (47.3–64.5)	61.5 (53.4–65.4)	.210
LAVI, mL/m2	42.3 (34.9–55.6)	36.3 (29.4–44.2)	<.001	48.2 (39.7–66.4)	44.0 (30.9–59.3)	.010

Abbreviations as in Table 1.

Discussion

This study demonstrated that CT-ECV could not predict AF recurrence in the absence of antiarrhythmic drugs within a 400-day follow-up after the first session of CA, either in the overall cohort or in subgroups of patients with PAF and PeAF. Furthermore, Kaplan–Meier analysis revealed no significant difference in the AF recurrence-free rate between the high and low CT-ECV groups, with the cutoff value of CT-ECV set not only at the median (28%) but also at 35%, 40%, 50%, and 60%.

ECV refers to the space or volume that is not occupied by cells. Regarding the cardiac tissue, ECV expansion is associated with inflammation with edema, replacement fibrosis, interstitial reactive fibrosis, and amyloidosis.¹⁸ Because the excessive collagen deposition is the main cause of ECV elevation, ECV mainly reflects myocardial fibrosis.¹⁹ ECV can be evaluated with contrast-enhanced CT or MRI. Similarly to previous reports,²⁰ our facility has also demonstrated a strong correlation between CT-ECV and MRI-ECV.²¹ In this study, 11 patients underwent cardiac MRI within 4 weeks before or after cardiac CT acquisition. Similarly to previous reports,^20,21 a significant correlation was also observed between CT-ECV and MRI-ECV (correlation coefficient 0.891; P < .001) (Supplemental Figure 5), suggesting that CT can evaluate ECV with the same accuracy as MRI. The reference values for CT-ECV at our institution and scanner were 27.5% ± 2.9%, calculated from 86 cases (excluding cardiomyopathy, coronary artery disease, severe valvular disease, arrhythmia, and reduced LVEF) in a previous report.⁵ In the pathophysiology of AF, ventricular ECV elevation can lead to ventricular diastolic dysfunction²² and atrial fibrosis, resulting in structural remodeling and, ultimately, the onset of atrial fibrillation.²³

In contrast, AF can also lead to ventricular remodeling owing to cellular, extracellular, and neurohormonal mechanisms, referred to as AF-mediated cardiomyopathy,²⁴ and previous studies have highlighted the influence of AF on left ventricular structural remodeling. Shantsila et al⁸ reported that prolonged AF duration correlates with increased ventricular myocardium fibrosis. They suggested that the inflammation or neurohormonal mechanism such as the activation of the renin-angiotensin-aldosterone system or transforming growth factor-β pathway may contribute to the ventricular fibrosis in the presence of AF. Furthermore, a previous report that examined autopsy cases found that AF is related to histopathologic atrial and ventricular fibrosis, along with chronic inflammation.²⁵ Another explanation for the relationship between AF and ventricular fibrosis is tachycardia-mediated cardiomyopathy. In animal experiments, left ventricular fibrosis is considered to occur because of a rapid ventricular response.²⁶ These mechanisms lead to the genesis of myocardial fibrosis, which is not limited to the atrium but expands to the whole heart, increasing with prolonged AF duration.⁸ In this study, the CT-ECV was significantly higher in the PeAF group than in the PAF group, consistent with previous reports from our institution.²⁷ This result may reflect the relationship between longer AF duration and the progression of left ventricular fibrosis.

Regarding ventricular ECV and AF recurrence after CA, CT-ECV did not predict AF recurrence. However, our results contradict those of 2 previous studies that used MRI for ECV quantification. First, Neilan et al¹⁰ reported that, in patients with AF and hypertension, expansion of MRI-ECV could predict AF recurrence after CA. However, the follow-up duration of the study was 2 years, and their Kaplan–Meier analysis between the low and high ECV groups did not seem to be different at the time of 1 year after CA and diverged after 1 year. Thus, if their follow-up period had been limited to 1 year, MRI-ECV might not have predicted AF recurrence after CA; conversely, extending the follow-up period of our study beyond 2 years, CT-ECV might have predicted AF recurrence after CA. This temporal aspect may explain the discrepancies between studies. Second, Li et al¹¹ also reported a positive correlation between MRI-ECV and AF recurrence after CA. However, in their study, nearly half of the patients with PeAF experienced AF recurrence, whereas more than 70% of patients with PAF maintained sinus rhythm. They routinely performed roof line or mitral isthmus line ablation only in the PeAF group, in addition to PV isolation. The reconnection rate of the linear lines is approximately 30%–50% according to previous reports,28, 29, 30 and the incomplete block lines are likely to be a source of atrial tachycardia or atrial flutter.³¹ These additional ablations in the PeAF group might have affected atrial arrhythmia recurrence rates in their study. Although their study did not explicitly compare ECV between PAF and PeAF groups, the ECV of patients with PeAF was probably higher than of patients with PAF. Hence, the difference in the AF-free survival rate caused by the difference in strategy between the PeAF and PAF groups might have been positively correlated with ECV in the study by Li et al.¹¹

Conversely, Gunasekaran et al¹² reported that MRI-ECV is not associated with AF recurrence after CA. In contrast to the 2 previously mentioned studies, they used cryoballoons for CA. Cryoballoon ablation has been demonstrated to achieve more durable PV isolation compared with radiofrequency CA,^32,33 with some reports showing that electrical PV isolation is maintained in 91% of PVs (68 of 75 PVs) when evaluated on a per-PV basis.³⁴ In fact, although the recurrence-free rate is 69% over a median follow-up period of 18 months in Neilan et al’s¹⁰ study (n = 145, excluding PeAF) and 62% over a median follow-up period of 13 months in Li et al’s¹¹ study (n = 130, which included substrate modification for PeAF), Gunasekaran et al’s¹² study (n = 100, cryoablation only) has demonstrated the highest recurrence-free rate of 72% over a median follow-up period of 457 days. Moreover, regarding AF duration, their study found no significant difference in AF recurrence between PAF and PeAF, which is consistent with our findings. These results suggest that durable PV isolation may be important, regardless of substrate modification, during CA for AF. The present study is also superior to previous reports in terms of the larger sample size (n = 467), no exclusion criteria for study participants (including PAF and PeAF), and a consistent ablation strategy (trigger elimination-based, substrate modification was performed only in 2.4%).

The importance of AF trigger elimination

The rate of maintaining the sinus rhythm during the 400-day follow-up period after the CA for AF was 83.5% in this study. Furthermore, although not all patients were examined, 92.9% of PVs (130 of 140 PVs) were isolated in 35 patients who underwent a second CA session owing to the arrhythmia recurrence. The low PV reconduction and AF recurrence rates after CA might explain why CT-ECV could not predict AF recurrence after CA. Regarding the ablation strategies, only 2.4% patients received substrate modifications, such as left atrial linear or box ablation. Because fibrosis occurs in both the atrium and ventricle in patients with AF,^7,25,35 left atrial fibrosis might have progressed with high CT-ECV. However, many patients with high CT-ECV can maintain sinus rhythm without substrate modification during CA. Notably, in our report on CA in patients with wild-type transthyretin amyloid cardiomyopathy, which is associated with an extremely high ECV, recurrence of AF itself was rare in patients with PeAF, and ECV was not significantly or independently associated with recurrence of AF, atrial flutter, or atrial tachycardia.³⁶ Furthermore, in the present study, patients without AF recurrence exhibited significant improvements in BNP levels, LVEF, and LAVI after CA compared with those before CA, even in the high CT-ECV group. The usefulness of CA for patients with AF with reduced cardiac function has also been shown in many reports.37, 38, 39 These findings suggest that eliminating PV triggers may be important for maintaining sinus rhythm and improving cardiac function after CA, regardless of substrate modification or CT-ECV value.

However, durable PV isolation alone does not guarantee freedom from AF recurrence after CA. In terms of very late AF recurrence (more than 1 year after CA), 1 contributing factor is non-PV triggers. The non-PV triggers are considered a negative factor for maintaining the sinus rhythm after CA.^40,41 Early AF recurrence is often associated with PV triggers (ie, PV reconnection), whereas very late AF recurrence is more likely to involve non-PV triggers.⁴² In addition, atrial myocardial fibrosis is said to be the source of non-PV triggers.⁴³ Given that AF is related to both the ventricular and atrial fibrosis,^7,25,35 longer follow-up periods may reveal late-onset AF recurrence owing to the non-PV triggers in the high CT-ECV group. Further studies are needed to investigate whether high CT-ECV is related to late AF recurrence owing to non-PV triggers.

Study limitations

First, the follow-up period was approximately 1 year after CA. Longer follow-ups may reveal different outcomes regarding the association between CT-ECV and AF recurrence rates after CA. Second, a 1-week Holter ECG monitor or CIED interrogation was not conducted for 15% of the eligible patients. Consequently, some instances of AF recurrence may have been missed, potentially affecting the results. Third, the left atrial substrate was not analyzed in this study. Therefore, we could not assess the relationship between atrial substrate and AF recurrence. Fourth, we could not eliminate the antiarrhythmic effect of β-blockers because many patients were unable to discontinue β-blockers owing to heart failure or cardiomyopathy, rather than arrhythmia alone. This introduced a selection bias regarding the β-blocker use after CA. Finally, the present study has some statistical limitations. As a single-center, retrospective cohort study with a small number of patients, it lacked broader applicability. Regarding the Cox proportional hazards model (Table 3, Supplemental Tables 2 and 5), CT-ECV seemed to have a minor effect on AF recurrence after CA, even if there was a significant difference, because HR was close to 1 with narrow 95% CIs when CT-ECV was treated as a continuous variable. However, when the CT-ECV was categorized based on the median value of 28.0% and used in the Cox proportional hazards model as a categorical variable, the 95% CI widened and uncertainty increased. If CT-ECV had a small effect on AF recurrence after CA, this effect may not have been detected because of the small sample size. Although CT-ECV is not a crucial risk factor for predicting AF recurrence after CA, this study could not determine whether it is associated with AF recurrence after CA. A multicenter study with a larger sample size is required to confirm these findings.

Conclusion

CT-ECV was not associated with AF recurrence within 400 days after the first CA session for AF and could not predict AF recurrence during this period. Even in patients with high ECV, simple and high-quality elimination of AF triggers, namely PV isolation, could sufficiently suppress AF recurrence without substrate modification. These findings suggest that regardless of the severity of left ventricular myocardium damage, PV isolation is useful for preventing AF and improving cardiac function.

Disclosures

Dr Kanazawa and Dr Kaneko received grants from Medtronic Japan, Nihon Kohden, Abbott Medical Japan, Fukuda Denshi, Boston Scientific Japan, Japan Lifeline, Nipro, and Biotronik Japan. Dr Tsujita has received honoraria from Bayer Yakuhin, Daiichi Sankyo, Kowa, MSD, Sanofi, and Takeda Pharmaceuticals and grants from Astellas Pharma, Abbott Vascular Japan, Bayer Yakuhin, Boehringer Ingelheim Japan, Boston Scientific Japan, Bristol Myers Squibb, Chugai Pharmaceutical, Daiichi Sankyo, Goodman, Japan Lifeline, Medtronic Japan, Mitsubishi Tanabe Pharma, MSD, Novartis Pharma, Otsuka Pharmaceutical, Sanofi, Takeda Pharmaceutical, and Terumo. All remaining authors have declared no conflicts of interest.