A novel and simple scoring system for assessing the indication for catheter ablation in patients with atrial fibrillation: The HEAL‐AF Score

Abstract Introduction A scoring system to determine indications for catheter ablation (CA) in atrial fibrillation (AF) is desired. Methods and Results Among 2898 consecutive patients with AF, CA was performed in 938 (32.4%). A new HEAL‐AF score has been developed by six variables, all of which were independently associated with CA by multivariate analysis and for each 1 point was assigned: heart failure ≥ NYHA II, elderly patients (age ≥75 years), asymptomatic AF, long‐standing persistent AF, atrial dilation (left atrial diameter ≥ 50 mm), and female sex. Low HEAL‐AF score was associated with high incidence of CA performance (52.0% for 0, 36.5% for 1, 15.1% for 2, and 5.6% for ≥ 3) and the predictive capability of this score by AUC of ROC curve was 0.720 (95% CI 0.701‐0.739, P < .001). The rates of freedom from AF/AT recurrence were 73.2% in HEAL‐AF score 0, 71.0% in 1, 60.0% in 2, and 50.0% in ≥ 3 (log‐rank test, P = .004). HEAL‐AF score 2 and ≥ 3 were significantly associated with recurrence of atrial tachyarrhythmia as compared with HEAL‐AF 0 (HR 1.755, P = .002, and HR 2.211, P = .007, respectively). Conclusions A new HEAL‐AF score was associated with patient indication for and the recurrence of atrial tachyarrhythmia after CA in patients with AF. Prescription of CA should be considered carefully in AF patients with HEAL‐AF score of 2 and ≥ 3.

degree of symptoms, presence or absence of significant heart failure (HF), age, left atrial (LA) size, number of failed AADs, likelihood of complications, and patient preference. Despite recent revisions of the guidelines, [4][5][6][7] there remains a debate how to select appropriate patients for CA, thereby confusing general cardiologists and primary care physicians.
Several scoring systems, such as CHADS2 score 8 and CHA2DS2-VASc score, 9 have played a significant role for selecting the AF patients who needs anticoagulation therapy for prevention of thromboembolic events. Similarly, a scoring system to determine the indication for CA in AF, if any, would be useful, but currently has not been well established. The aim of the present study was to develop a scoring system for the patient selection for CA in patients with AF using data from a single hospital-based cohort, and to evaluate whether the score is associated with the rhythm outcome after CA.

| Subjects
The Shinken Database was established for all new patients who visited The Cardiovascular Institute, Tokyo, Japan (Shinken is an abbreviation of the name of the hospital in Japanese). Patients with active cancer and foreign travelers were excluded. The principal aim of this hospital-based database is to survey the prevalence and prognosis of cardiovascular diseases in a cardiovascular-specialized hospital, including arrhythmic specialists, in an urban area of Japan. 10,11 The registry was started in June 2004, and thereafter, patients have been continually registered in the database. The data in the present study were derived from this database between April 2009 and March 2017, when the annual number of AF patients who underwent CA exceeded 50 cases per year. During this period, 12 760 patients were newly registered in the total database, among whom AF was diagnosed in 2898.
In the present study, AF was diagnosed by electrocardiography, including 12-lead surface electrocardiograms (ECG) and 24-hour Holter ECG performed within 3 months after the initial visit, and by a medical history of AF from the referring physician, and new-onset AF that occurred> 3 months after the initial visit was not included in the diagnosis of AF. Any AF episodes within 3 months after the initial visit, including first detected AF and post-operative AF, were diagnosed as AF. Definitions of paroxysmal AF, persistent AF, and long-standing persistent AF were according to the ACC/AHA/ESC guidelines 4 : paroxysmal AF was defined as AF that terminates spontaneously or with intervention within 7 days of onset; persistent AF was defined as continuous AF that lasted more than 7 days; and long-standing persistent AF was defined as continuous AF with a duration > 12 months. The definition of asymptomatic AF was the lack of the usual AF-related symptoms, including palpitations, shortness of breath, chest oppression, dizziness, etc The concomitance of acute or chronic HF in patients with asymptomatic long-standing persistent AF was also categorized as asymptomatic AF.

| Catheter ablation
Our ablation protocol was described previously. 12 All patients had anticoagulation therapy for more than 3 weeks before ablation and underwent transesophageal echocardiography to exclude atrial thrombus within 3 days before ablation. All AADs were discontinued for at least 5 half-lives before the procedure, except amiodar- with electrophysiological and pharmacological methods as described previously. 12 If AF or atrial tachycardia (AT) lasting more than 5 minutes was induced by atrial pacing, additional ablation including targeting complex fractionated atrial electrograms (CFAE) and LA linear ablation were performed. Repetitive non-PV ectopic beats were also eliminated if these were induced with isoproterenol infusion. Finally, the dormant conduction provoked by administration of adenosine triphosphate (ATP, 20 mg) was ablated at the end of the procedure.
All patients were followed up at our outpatient clinic every month for 3 months after the procedure, and thereafter every 2-3 months for 9 months after the procedure. Oral anticoagulants were maintained for at least 3 months after the procedure. AADs except beta-blockers were continued for 1-2 months and then discontinued if the patient had no recurrence of AF/AT. The recurrence of AF/AT was evaluated based on clinical symptoms and ECG, including 12lead ECG at every visit, 24-hour Holter ECG at 3 and 12 months, and 30-s ECG recorded with a mobile event recorder at a minimum of 1-2 times a day for 3-6 months after the procedure. Recurrence of AF/ AT defined as any episode of atrial tachyarrhythmia lasting > 30 s after 3 months of the blanking period without AAD.

| Statistical analysis
Continuous data are given as the mean ± SD. All reported P-values are two-sided, and P < .05 was taken to indicate statistical significance. Statistical analyses were performed using spss 19.0 (SPSS Inc).
First, total patients were divided into those who underwent CA (Ablation group) and who did not (No-ablation group). The differences in continuous and categorical variables between the two groups were tested by unpaired Student's t test and χ 2 test, respectively. To identify the clinical variables that affected CA performance, a multivariate model of the logistic regression analysis was developed. In this model, clinical factors that are listed as the major issues to affect the indication of CA in recent guidelines 4-7 (age ≥ 75 years, NYHA class ≥ II, long-standing persistent AF, asymptomatic AF, female sex, left atrial diameter ≥ 50 mm, left ventricular ejection fraction < 40%, and cardiomyopathy) were forcedly introduced. Assigning 1 point for each independent variable, we developed a new scoring system. The predictive capability of the new score for the incidence of CA performance was evaluated by the area under the receiver operating characteristic (ROC) curve.
Second, the effects of the new score on the procedural results of CA and the rhythm outcome after initial CA were evaluated. The cumulative event-free rate of recurrent AF/AT was estimated by the Kaplan-Meier method, and the difference by the score was tested by the log-rank test.
Third, to understand the precise mechanisms in the relationship between the new score and the procedural results or the rhythm outcome, further analyses were performed using the components of the scoring system. The effect of the components on non-PV triggers and substrate modification during CA were evaluated by the Values are shown as mean ± SD or n (%).
TA B L E 1 Patient characteristics multivariable models with logistic regression analysis. The effect of the components on AF/AT recurrence was evaluated by the univariate and multivariate models with Cox regression analysis.

| Development of HEAL-AF score
The predictive capabilities of the HEAL-AF score for CA as determined by the area under the curve (AUC) was 0.720 (95% CI 0.701-0.739, P < .001). F I G U R E 1 Distribution of patients according to the number of HEAL-AF score and each components of the HEAL-AF score. HEAL-AF score, NYHA class ≥ II = 1, age ≥ 75 years = 1, asymptomatic AF = 1, long-standing persistent AF = 1, left atrial diameter ≥ 50 mm = 1, female sex = 1 and linear ablation in LA (P < .001), significantly increased. Figure 3 shows the Kaplan-Meier curves of the event-free rate of AF/AT re-

| Adverse outcomes after CA
Pericardial effusion requiring pericardiocentesis after the ablation procedure occurred in two patients with HEAL-AF 0 and 1 patient with HEAL-AF 1 (asymptomatic AF). Symptomatic stroke after the ablation procedure occurred in 1 patient with HEAL-AF 1 (female).
There were no patients complicated by symptomatic PV stenosis or atrioesophageal fistula.

| Main findings
In this single hospital-based cohort study, we identified the independent predictors for the actual CA performance in patients with AF, consequently developing a new scoring system. The HEAL-AF score, which are based on simple and easily obtained clinical variables, could predict the incidence of CA with a moderate predictive value. In addition, the HEAL-AF score was associated with complex ablation strategy, and AF/AT recurrence after CA.

| Clinical implications of HEAL-AF score
First, the HEAL-AF score was significantly associated with the CA incidence of AF in the present study. Although recent guidelines 4-7 have demonstrated that types and duration of AF, degree of symptoms, concomitance of significant HF, and older age are major issues for CA indication, it has not been well established how to select appropriate patients for CA in AF population. For the HEAL-AF scoring system, we adopted 6 variables, including these 4 issues, which independently associated with the incidence of CA performance, thereby stratifying the physicians' selection tendency with CA in AF patients. In addition, patient preference for CA is one of the most important factors in the decision regarding CA, especially in patients with asymptomatic AF and elderly patients, which was incorporated into the new score. Therefore, the HEAL-AF score could reflect both medical judgment and possibly also patient preference in CA.
Second, the present study demonstrated a significant association between the HEAL-AF score and the outcome of CA. As HEAL-AF score increased, the frequency of substrate modification was significantly increased in the present study. Moreover, the HEAL-AF score was significantly associated with AT/AT recurrence after the CA procedure: compared with HEAL-AF-0, HEAL-AF 1 had a similar incidence, while HEAL-AF 2 and ≥ 3 had 1.8 and 2.2 fold incidence of AF/AT recurrence, respectively. Interestingly, some components of the HEAL-AF score overlapped with other scoring systems predicting low voltage zone in LA, 13 and AF recurrence after CA. 14,15 Therefore, the HEAL-AF score classification could provide a measure to select not only suitable patients, but also favorable outcomes with suitable strategies for CA of AF. In addition, due to the unique features of simple and easily obtained clinical variables, without multinominal categories of age and left atrial diameter, and without assessing renal function, the HEAL-AF score may gain an advantage as a more convenient tool in the clinical practice over other scoring systems. [13][14][15] In patients with HEAL-AF 0, CA, especially the PVI only strategy, will be adequate in most patients. In patients with HEAL-AF 1, CA may be reasonable with a sufficient procedural success rate, but additional ablation strategy beyond PVI may considered in some patients. In patients with HEAL-AF 2, careful patient judgment and specific ablation strategies will be required. In patients with HEAL-AF ≥ 3, CA for AF was less experienced in the present study. Indication of CA in this population should be more carefully considered because of the possibly limited efficacy.

| HEAL-AF score and performance of CA in HF patients
Several randomized trials in patients with AF and HF have reported improvement in both soft endpoints (improving LVEF and maintaining sinus rhythm) and hard endpoints (death and hospitalization for heart failure) by CA. [16][17][18] Most patients enrolled in these trials have generally been younger men with paroxysmal or persistent AF, who would be classified as HEAL-AF 1 in the present study. These patients would belong to HEAL-AF 1 who have HF without other components of the HEAL-AF score, which suggest favorable outcome. There are controversies regarding the procedural strategy of CA in patients with HF.
The AATAC trial, 17 mostly including patients with possible HEAL-AF 1, demonstrated that patients undergoing PVI and posterior wall isolation had significantly better outcome than patients with PVI alone.
In contrast, the CAMTAF trial, 16

| HEAL-AF score and performing CA in elderly and female patients
Although there was no significant association between AF/AT recurrence and elderly patients in the present study, the efficacy of CA for elderly patients has remained controversial. A pooled analysis reported by Kautzner et al 19 showed that the success rate of a single CA procedure was not different between older and younger groups.
On the other hand, in several scoring system to predict AF recurrence after CA, older age has been introduced as one of the most common predictors. 13 21 ), may also affect the decision making for CA in elderly patients. To determine the optimal indication of CA for elderly patients, more specific approach will be needed.
The present study also showed that female sex was an independent predictor of AF/AT recurrence. Interestingly, several studies showed an increased rate of AF recurrence in women after CA. [22][23][24] Reasons for this gender difference may be explained which are categorized into patient selection bias and AF mechanisms. Several studies showed that women who underwent CA are older, have high comorbidities, have higher rate of long-standing persistent AF, and referred for CA later in their clinical course. 25,26 In addition, similar to the present study, Patel et al 26 reported that women had more non-PV triggers than men.

| HEAL-AF score and performing CA in asymptomatic AF or long-standing persistent AF patients
The prevalence of asymptomatic AF has been reported to be 10%-40% in a variety of cohorts. 27 For this large population of asymptomatic AF, the indication of CA is still controversial. Forleo et al 28 found similar success rates of ablation between symptomatic (mostly classified as HEAL-AF 0) and asymptomatic patients (mostly classified as HEAL-AF 1) with paroxysmal and persistent AF. On the other hand, in a study by Wu et al 29 the success rates of ablation were much lower in asymptomatic patients with persistent or long-standing persistent AF (mostly classified as HEAL-AF 2) than in a matched group of symptomatic patients (mostly classified as HEAL-AF 1). In the present study, asymptomatic AF was not significantly associated with AT/AT recurrence, while long-standing persistent AF was an independent predictor of AF/AT recurrence. Therefore, for younger asymptomatic AF patients, classified as HEAL-AF 1 in the present study, CA could be considered appropriate before or soon after progression to long-standing persistent AF. Asymptomatic AF, as well as long-standing persistent AF and LAD ≥ 50 mm were independent predictors of substrate modification in the present study. Further studies are necessary to determine the optimal CA strategy for patients who have these HEAL-AF components.

| Study Limitations
The present study had several limitations. First, our database consisted of patients from a single cardiovascular hospital which included only Japanese patients. In addition, patients in our cohort were younger than those in other areas or population-based registries. 30

| CON CLUS IONS
The HEAL-AF score, which is based on simple and easily obtained clinical variables, was significantly associated with CA for and AF/AT recurrence after CA in patients with AF. In patients with HEAL-AF 0, CA, with the PVI only strategy, will be useful for most patients; In HEAL-AF 1, CA may be reasonable with sufficient procedural success rate; In HEAL-AF 2, careful patient judgment and specific ablation strategy should be considered; In HEAL-AF ≥ 3, CA for AF was less experienced.

ACK N OWLED G EM ENT
This research was supported by a research grant from Nippon Boehringer Ingelheim.

D I SCLOS U R E S
The Institutional Review Board of the Cardiovascular Institute approved the study (Date of IRB approval; May, 2019; Approval number, 169).

CO N FLI C T O F I NTE R E S T
The following authors have potential conflicts of interest: SS received research funding from Daiichi-Sankyo and Mitsubishi-Tanabe.
TI received research funding from Medtronic Japan, Japan Lifeline,