Novel Predictors For New Onset Atrial Fibrillation After Typical Atrial Flutter Ablation: An Invasive Prospective Study

Purpose: Patients undergoing cavotricuspid isthmus (CTI) ablation for typical utter (AFL) have a high incidence of new onset atrial brillation (AF). We aimed to analyze the incidence and predictors for new onset AF in this subset of patients to stratify thromboembolic risk. Methods: Between 2016 and 2019, 70 patients without history of AF but with high-risk for developing AF, based on a recent AFL ablation or a high PACE score for AF risk, were prospectively included. All patients were monitored continuously by implantable loop recorder and followed by remote monitoring. Results: Overall 48 patients were included after CTI ablation and 22 patients were included based on a high PACE score. New onset AF rate at 12 months was signicantly higher in the AFL group compared to PACE group (56.3% vs 22.7%, p=0.011). History of AFL was the only independent predictor for new onset AF (HR:3.82; 95% CI:1.46-10.03; p=0.006) at a median follow-up of 12 months (Q1-Q3:4-19 months). In the AFL group, two very strong independent predictors for new onset AF were a PACE score ³30 (HR:6.9; 95% CI:1.71-27.91; p=0.007) and HV interval ³55 (HR:11.86; 95% CI:2.57-54.8; p=0.002). Conclusions: AFL is the most important predictor for new onset AF. In patients undergoing AFL ablation, a high PACE score and/or long HV interval predict even higher risk, and may be useful in decision for empiric long-term anticoagulation.


Introduction
Typical atrial utter (AFL) is caused by a macroreentry in the right atrium and can be easily eliminated by cavotricuspid isthmus (CTI) catheter ablation, with a recurrence-rate of 6%, decreasing to 1.5% if very speci c techniques are used to assess complete CTI block [1][2][3]. However, after AFL ablation, new onset atrial brillation (AF) is diagnosed in 20 to 50% of cases at 2-3 years [4][5][6][7][8][9][10][11][12][13]. AF detection increases with more frequent monitoring and/or longer follow-up duration [8,9]. Several studies have identi ed predictors for AF after AFL ablation, but all are retrospective and non-invasive. On note guidelines aim for documentation of AF in the decision making process of continuation/discontinuation of long-term anticoagulation.
To date, the strongest predictors for AF after AFL ablation are left atrium dilatation [6,7], its surface ECG surrogate, the presence of advanced interatrial block [10], and its electrophysiological surrogate, the presence of a long interatrial conduction time [11]. Other variables associated with new onset AF are a high HATCH score [12], AF inducibility during the CTI ablation procedure [13], and the presence of obstructive sleep apnea (OSA) [9,14].
In this study we sought to analyze the incidence and predictors for new onset AF after CTI ablation for typical AFL in a prospective manner, and compare with patients without AFL, but with high risk for AF based on the PACE calculator score. All patients underwent internal loop recorder (ILR) implantation or had preexisting implantable cardiac device with an atrial lead (CDAL) for continuous long-term monitoring.

Study setting
This study was supported by the 2014 La Marató de TV3 Foundation Program, which was dedicated to heart diseases and aims to promote biomedical research through the funds raised by Catalonian donations. The study was reviewed and approved by the Hospital del Mar ethics committee and all ablation procedures and ILR implantations were performed after written informed consent was obtained.
Patient population Between 2016 and 2019, patients with high risk to develop AF and no history of AF were considered to be included in this prospective single-centre study. Patients were considered high risk if they had at least one of the two following conditions: 1. Patients presenting with typical AFL, undergoing CTI ablation with resultant complete bidirectional block.
2. Patients with a PACE calculator risk score ≥ 30% for new onset AF at 3 years. The PACE calculator score uses a composite of 4 independent predictors for new onset AF (age, heart failure/cardiomyopathy, supraventricular ectopy > 0.2% or atrial tachycardia on 24h-holter monitoring (SVE), and PR interval) to predict new onset AF at 3 years [15]. This calculator is available at https://a bcalculator.com. An example can be seen in Fig. 1.
All patients underwent a 24h-holter monitoring to evaluate atrial ectopy, and an ILR implantation except for those already wearing a CDAL, and were followed by remote monitoring. The study protocol conforms to the ethical guidelines of the 1975 Declaration of Helsinki as re ected in a priori approval by our institution's human research committee.

Study endpoints
The introductory endpoint was the incidence of new onset AF after CTI ablation compared to other high risk patients with no history of AFL. Any AF episode of at least 2 minutes was de ned as positive. The precise primary endpoint was the study of multiple clinical, echo, ECG, and electrophysiological variables to predict new onset AF after CTI ablation.
Electrophysiological study, CTI ablation and ILR implantation.
The description of the procedure has been described previously [16]. Brie y cavotricuspid isthmus ablation was performed under local anesthesia and conscious sedation. Anticoagulation therapy was not discontinued. Two diagnostic multipolar steerable 6F catheters were placed through the right femoral vein into the low lateral right atrium and the coronary sinus ostium (CSO). A 7F irrigated-tip catheter was used for CTI mapping and ablation. Radiofrequency ablation was performed during atrial pacing from the CSO or during AFL. Established criteria for the diagnosis of bidirectional CTI complete block were assessed, including incremental pacing maneuvers [16][17][18]. Conduction parameters, including AH and HV interval were measured. During the same procedure, an ILR was placed under local anesthesia and antibiotic prophylaxis with Cefazolin.
Long-term follow-up All antiarrhythmic drugs were discontinued after ILR implantation. Patients had follow-up in the outpatient clinics 2 months after the procedure and every 6 months thereafter for a minimum period of time of 1 year. Patients were monitored with monthly remote transmissions by means of Home Monitoring Service Center (Biotronik ILR) or Merlin.net (Abbott ILR).

Statistical analysis
Continuous variables are expressed as mean ± SD. Comparisons between groups were performed using an unpaired Student t test, a Chi-square test or a Fisher exact test analysis as appropriate. The primary end point was AF documentation, rst of all taking into account the whole population and afterwards, restricted to the population having with previous AFL. Kaplan-Meier survival curves were performed to compare patients with and without previous AFL, using log-rank test to check differences between survival curves. Cox regression models were used to estimate hazard ratios associated to AF. Univariate and multivariate analysis were performed. Variable selection on multivariate analysis was based on previous knowledge (clinical criteria) and variables with p-values < 0.1 on univariate analysis (both criteria combined). Finally, and restricted to the analysis of patients with previous AFL, optimal cutoffs points were calculated for the variables PACE calculator score and HV, using the Youden criteria: sample point that maximizes the Youden index (sensitivity + speci city − 1). These dichotomized variables were also analyzed through Kaplan-Meier survival curves and included in the multivariate analysis. A p value of ≤ 0.05 was considered statistically signi cant. All statistical analyses were performed using STATA version 15.1 (StataCorp, College Station, TX, USA).

Population characteristics
Between 2016 and 2019, 70 patients (79% male, 78.54 ± 10.87 years) with high risk to develop AF and no history of AF were included: 48 patients (85% male, 76.11 ± 10.68 years) were consecutive non-selected patients included after CTI ablation with complete bidirectional block (AFL group) independent of the PACE calculator risk score, and 22 patients (64% male, 80.59 ± 10.74 years) were included based on a high PACE calculator score (PACE group). Overall 60 patients (86%) underwent an ILR implantation during admission, while the remaining 10 patients (14%) had a previously implanted CDAL. There were either no complications related to the ablation procedure or to the ILR implantation. Patients with previous AFL had less comorbidities, including renal failure, hypertension, dyslipidemia, left ventricle hypertrophy, previous stroke, ischemic coronary disease and valvular disease ( Table 1). As expected, the PACE calculator score was lower among patients with previous AFL (20.09 ± 19.3 with AFL vs. 47.05 ± 20.77 without AFL; p < 0.001), as the history of AFL was the only reason for them to be enrolled, whereas patients without AFL were selected for their higher PACE calculator score. For the same reason CHA2DS2-VASC score was higher in patients without AFL compared to those with AFL (4.5 ± 1.4 vs 2.7 ± 1.6, respectively; p < 0.001). No differences were observed in terms of ECG, echocardiography, Holter or polysomnography parameters between both groups. Table 1 Comparison of characteristics in patients who had prior AFL ablation and patients who did not, but had high-risk score for AF  Table 2).  Given the fact that SVE is included in the PACE calculator and that SVE and HV interval showed a limited discrimination potential as continuous variables, we carried out a new analysis using the PACE calculator and HV interval as categorical variables (Table 4). Multivariate analyses showed that a PACE calculator score of ≥30% predicted the occurrence of new onset AF at a median follow-up of 12 months (Q1-Q3: 4-19 months), with a HR of 6.9 (95% CI: 1.71-27.91; p = 0.007) and HV interval of ≥55msec predicted the occurrence of new onset AF with a HR of 11.86 (95% CI: 2.57-54.8; p = 0.002). Interestingly LA dilatation did not predict AF incidence in our population. Figure 3shows survival curves for freedom from AF incidence related to the PACE calculator score and HV interval, respectively, in patients with previously ablated AFL. Predicting high-risk patients We analyzed the added value of considering the PACE calculator score and/or HV interval in order to better stratify the risk for new onset AF in patients undergoing an AFL ablation procedure. Concerning the PACE calculator, a score of ≥30% predicted the incidence of new onset AF during follow-up with a speci city of 90.5%, a sensitivity of 38.5%, a positive predictive value (PPV) of 83.3% and a negative predictive value (NPV) of 54.3%. Regarding the HV interval, a value of ≥55 msec predicted the incidence of new onset AF during the follow-up with a speci city of 88.9%, a sensitivity of 62.5%, a positive predictive value (PPV) of 88.2% and a negative predictive value (NPV) of 64%. Therefore if the PACE score was ≥ 30 or the HV was ≥ 55msec, the probability of developing AF at a median of 12 months follow-up was greater than 83% and 88%, respectively. Figure 4 shows area under ROC curve to predict new onset AF in patients with AF with PACE calculator score ≥ 30 or with HV interval ≥ 55msec, respectively. Finally when combining both variables, the NPV of having a PACE calculator score of < 30 and an HV interval of < 55msec was 75% (95% CI: 50.9%-91.3%).

Discussion
Typical utter as a risk factor for AF New onset AF can be observed in a high proportion of patients after an AFL ablation [4][5][6][7][8][9][10][11][12][13] but studies addressing this issue are retrospective and most of them are based on 24-h holter monitoring, and it is well known that AF detection increases with more frequent and continuous monitoring [8,9]. We prospectively included seventy patients and monitored them continuously with ILR or CDAL. It was expected that the patients with AFL had less comorbidities and lower PACE scores than the patients without AFL, as they were enrolled solely for a history of AFL undergoing ablation. We found that the presence of AFL was associated with a much more increased risk that we expected for developing new  [7]. We did not nd any association between LA enlargement and new onset AF but the limited size of our population could have in uenced the results. The presence of advanced interatrial block, which is linked to LA enlargement [10], has been also described as a predictor for new onset AF in patients with AFL. This parameter is useful since it can be observed in a 12-lead ECG. Unfortunately we did not collect this information. Instead we collected the PR interval, which takes part in the PACE calculator, and other surface and intracardiac electrophysiological parameters such as QRS duration, AH interval and HV interval. We found a PACE calculator score of ≥30% and/or an HV interval of ≥55 msec to be the only, but strong predictors for new onset AF at follow-up. This nding reinforces the PACE calculator as a useful tool to predict AF in the general population as well as to further risk stratify patients after AFL ablation, since it behaved as independent predictor, providing additional and complimentary information. Of importance the PACE calculator includes signi cant atrial ectopy as a predictor, which was associated to new onset AF at follow-up in the general population. We could not nd any other predictor, including left atrium dilatation or OSAS. Finally we found that both a high PACE calculator score and/or a long HV interval are very speci c predictors for new onset AF in this subset of patients, with a very good PPV. Concerning our ndings with the HV interval, this parameter has been previously correlated inversely with the left ventricular ejection fraction in patients with structural heart disease [19], and an HV interval of ≥ 55 msec has also been associated to atrial arrhythmias [20]. Given that patients with isolated AFL develop not only new onset AF but also stroke [21] at a higher rate than general population we believe that both variables could be of help in the decision for rigorous monitoring for AF and even in the decision for longterm anticoagulation in determinate cases: if either the PACE score is ≥ 30 or the HV interval is ≥ 55msec, the probability of developing AF at a median of 12 month follow-up is greater than 83% and this would favor the decision of not to stop anticoagulation after AFL ablation; on the other hand if both, the PACE calculator score and the HV interval, are below those limits, the probability maintaining sinus rhythm is 75%, indicating that discontinuation of anticoagulation would be probably safe.
Interestingly, some centers have performed prophylactic pulmonary vein isolation in patients with AFL but without documented AF [22,23]. Results have revealed a signi cant reduction in recurrent atrial arrhythmias. From our point of view such a strategy could be of use in individualized high-risk cases, namely young patients with AFL with a high PACE calculator score or long HV interval, but this should be assessed in a prospective randomized study.

Limitations
It is likely that the high incidence of new onset AF in this study is due to both the high risk pro le of patients included (average CHA2DS2-VASC score of 3.2) and the use of continuous long-term monitoring with ILR or CDAL in all patients. This study may have been limited by the small sample size, which may have accounted for the lack of predictive value of left atrial dilation or sleep apnea. Importantly despite this inconvenience we have been able to reach the two most relevant independent predictors for new onset AF in this setting of previous AFL ablation. This should be complemented with further research of predictors in a more extensive sample size.

Conclusions
In summary, we nd that AFL is the most important predictor for new onset AF in a prospective study evaluating comprehensive clinical, noninvasive and invasive electrophysiological parameters, with longterm continuous cardiac monitoring. Patients with AFL have even greater risk for AF than patients with signi cantly increased comorbidities, traditional risk factors for AF, and higher PACE calculator score. Additionally, in patients who undergo AFL ablation, a high PACE score or long HV interval predicts very high risk for developing AF. These patients should be considered for rigorous monitoring for AF or even for empiric long-term anticoagulation in determinate cases. A large prospective randomized trial should be performed in order to con rm these ndings.