Atrial fibrillation (AF) is the most common permanent cardiac arrhythmia and a further rise in prevalence is still expected for Western nations [
1]. Since the disease is associated with a significant burden of morbidity and mortality as well as financial strain, early detection and effective treatment of the arrhythmia remains an important goal. Regarding treatment, progress has been made using either cryo- or radiofrequency ablation for catheter-guided pulmonal vein isolation (PVI), which represents an effective tool to achieve rhythm control and, according to recent guidelines, should be considered as first-line therapy for a specific group of patients [
1]. However, even for this patient group, a significant proportion of patients suffer from recurrence, with a single-procedure success rate between 66.6% for paroxysmal and 51.9% for non-paroxysmal AF 1 year after ablation [
2], showing the need for additional therapeutic approaches. Concerning ablation success, a variety of predictors have been discussed, ranging from certain comorbidities to general patient characteristics such as age or sex. A decreased heart rate variability (HRV) is among these predictors; it can be obtained relatively easily through 24‑h Holter or (ultra-)short-term recordings and represents autonomous nervous system (ANS) activity. Since the origin and maintenance of AF are partly attributed to a disbalance between the sympathetic and parasympathetic nervous system [
3] and modulation of the intrinsic ganglionated plexi as part of the ANS in addition to PVI has been shown to increase ablation success rate [
4], HRV could conceivably be a useful proxy for predicting ablation success.
We therefore evaluated the relation between HRV and AF recurrence in patients after PVI using first-generation cryoballoon (CB1), second-generation cryoballoon (CB2), or radiofrequency ablation (RF).
Methods
Study population
In total, 497 patients were included in this retrospective observational study (for patient characteristics, see Table
1). Initially, we gathered data from 1522 patients who underwent either CB1, CB2, or RF from January 2011 to August 2017 in our center. CB2 was introduced in late April 2014 and completely superseded CB1 from that point on. We excluded patients with repeated ablation (
n = 378), who had no recordings or unsuitable Holter recordings (
n = 588), who had less than 15 h recording time (
n = 5), and those who had missing follow-up data (
n = 54). The follow-up period was limited to 24 months.
Table 1
Study population characteristics
Number | 129 | 172 | 196 | – |
Age | 58.4 ± 12.5 | 64.8 ± 10.0 | 60.7 ± 12.8 | < 0.001 |
Gender | Male (%) | 59 (46) | 99 (58) | 92 (47) | 0.061 |
Female (%) | 70 (54) | 73 (42) | 104 (53) | – |
BMI | 28.7 ± 4.3 | 29.4 ± 4.9 | 28.0 ± 4.4 | 0.043 |
AF type | Paroxysmal (%) | 118 (92) | 135 (79) | 127 (65) | < 0.001 |
Persisting (%) | 11 (8) | 37 (21) | 69 (35) | – |
Smoking state | Non-smoker (%) | 118 (91) | 157 (91) | 189 (96) | 0.008 |
Smoker (%) | 10 (8) | 6 (4) | 4 (2) | – |
Ex-smoker (%) | 1 (1) | 9 (5) | 3 (2) | – |
CHF NYHA | No CHF (%) | 118 (91) | 147 (85) | 177 (90) | 0.536 |
I (%) | 4 (3) | 5 (3) | 3 (2) | – |
II (%) | 6 (5) | 15 (9) | 12 (6) | – |
III (%) | 1 (1) | 5 (3) | 4 (2) | – |
IV (%) | 0 (0) | 0 (0) | 0 (0) | – |
CAD (%) | 13 (10) | 45 (26) | 22 (11) | < 0.001 |
Status post AMI (%) | 2 (2) | 7 (4) | 3 (2) | 0.217 |
Diabetes (%) | 14 (11) | 29 (17) | 19 (10) | 0.094 |
Hypertension (%) | 84 (65) | 136 (79) | 121 (62) | 0.001 |
Status post stroke (%) | 3 (2) | 8 (5) | 7 (4) | 0.564 |
Amiodaron (%) | 38 (30) | 42 (24) | 55 (28) | 0.583 |
Flecainide (%) | 47 (36) | 30 (17) | 40 (20) | < 0.001 |
Flecainide PIP (%) | 0 (0) | 2 (1) | 7 (4) | 0.045 |
β‑Blocker (%) | 112 (87) | 144 (84) | 177 (90) | 0.169 |
CHA2DS2-VASc Score | 1.8 ± 1.4 | 2.6 ± 1.5 | 2.0 ± 1.5 | 0.007 |
Mean timepoint of latest follow-up | 31.4 ± 26 | 22.8 ± 13.7 | 31.7 ± 22 | < 0.001 |
Recurrence-free rate after 2 years | 57.3% | 69.2% | 72.2% | 0.032 |
Ablation procedure
The decision of whether to schedule patients for cryoballoon ablation or RF was made according to clinical judgment with regard to patient characteristics and the preference of the examiner.
The Cryoballoon (CB) and RF procedure and their preparation concerning imaging and anticoagulation have been previously described in detail [
5].
In brief, the CB or RF catheter were delivered via transseptal puncture.
For CB, using the medtronic arctic front (CB1) or medtronic arctic front advance (CB2), each pulmonary vein was isolated (confirmed via entrance or exit block using Achieve, Medtronic spiral mapping catheter) after verifying occlusion.
For RF, we performed encircling PVI using the Thermocool (Biosense Webster) irrigated tip catheter and utilized a Lasso-catheter (Biosense Webster) to confirm isolation of the pulmonary veins.
24-h Holter recordings and HRV
All patients underwent at least 15-h-long Holter recordings (AMEDTEC EP8-S) 1–2 days after the procedure. The HRV parameters were analyzed using AMEDTEC ECGpro. The five following time-domain HRV parameters were analyzed: standard deviation of normal-to-normal (NN) intervals (SDNN; ms), standard deviation of the average NN intervals for each 5‑min segment (SDANN; ms), mean of the standard deviations of all the NN intervals for each 5‑min segment (SDNN-Index; ms), root mean square of the differences between adjacent NN intervals (rMSSD; ms); and the percentage of adjacent NN interval differences greater than 50 ms (pNN50; %).
Patient follow-up
After ablation and discharge from our hospital, medical attendance of our patients was taken over by outpatient care, which scheduled follow-up visits 3–6 months after the ablation and then yearly after the first follow-up visit (thus 15–18 months after the ablation in the observation period for our study). Additionally, patients were encouraged to call in for an unscheduled visit if they were to experience recurrence-related symptoms. Recurrence was defined as ECG-recorded patterns typical for AF lasting at least 30 s and occurring at least 3 months after the ablation.
Statistical analysis
All continuous variables are expressed as mean values (± standard deviation). Differences in baseline characteristics between the different ablation methods were compared using the Kruskal–Wallis test due to missing normal distribution, whereas differences in results were compared using the t test or Welch test if there was a difference in variances. Categorical variables are expressed as numbers and percentages and were compared using the chi-squared test. To test for homogeneity of variance, we used Levene’s test. Recurrence-free survival was analyzed using the Kaplan–Meier estimator with a log-rank test and adjustment for other possible risk factors was performed using Cox regression analysis. Furthermore, we used receiver operating characteristics (ROC) to determine cut-off values for our HRV parameters to conduct survival analysis. Statistical significance was defined as a two-sided p value smaller than 0.05. The software used to manage our database and to conduct statistical analysis was SPSS 26.0.0.0.
Discussion
Catheter-based ablation methods are a powerful tool to achieve long-term rhythm control when treating AF. However, not all patients seem to benefit equally from the procedure, as approximately one third of them suffer from recurrences within 1 year. To date, there is still a lack of viable characteristics to identify this group, which could benefit from a different follow-up-treatment. With our study, we demonstrate that low rMSSD values a few days after the ablation can independently predict AF recurrence. This relationship is strongest for CB2 and therefore better suited to predict recurrence after CB2 than after RF. When regarding only the mostly superseded CB1, this association is the weakest and insignificant.
As reported by Pandya et al. [
7], in our data too recurrence-free survival was improved by CB2 (
p = 0.057) compared to CB1, falling just short of significance. Likewise, as stated by the most recent guidelines [
1], we also found CB2 to be noninferior to RF regarding efficiency, thus confirming the validity of our dataset.
Before we identified rMSSD as a predictor for recurrence, we first examined the differences in mean HRV between the recurrence and non-recurrence group, finding that suffering from AF recurrence is associated with significantly higher HRV after CB2 (except for SDANN). This led us to attempt to identify specific cut-offs for differentiating between groups with either high or low risk for recurrence, which we found for all HRV parameters. Thus, our next step was to compare them and find the most applicable: rMSSD, as confirmed by the first Cox regression and validated as an independent predictor by the second analysis. Our earlier results also suggested that the calculated cut-off for rMSSD might show different predictive capabilities depending on the ablation method. Therefore, instead of only using the whole population as basis for our cut-off calculation, we recalculated the best cut-off for every ablation method, confirming our hypothesis that rMSSD is best suited to evaluate recurrence risk after CB2.
Several studies in the past have demonstrated a relation between low HRV after ablation and lower recurrence risk. However, the HRV parameters that have been found to be predictive for AF recurrence differ to some extent between those studies. Wang et al. reported significantly reduced SDNN and rMSSD for the non-recurrence group, but unlike us determined SDNN as the sole independent predictor [
8]. Two other studies identified only frequency-domain HRV parameters as independent predictors for recurrence [
9,
10] and Zhu et al. found rMSSD, pNN50, and high-frequency components (HF) as independent predictors for AF recurrence [
11]. These three studies, however, only included patients who received RF, which could explain some of the differences compared to our study given we included also the first two generations of cryoballoon ablation. We found only one study including patients treated with CB2, which observed significantly reduced rMSSD, HF, and low-frequency components after ablation only in the non-recurrence group but measured not until 3 months after the procedure [
12]. Oswald et al. did not find a relation between AF recurrence and attenuated HRV but they used only CB1, and agreed with our estimation that HRV is not suited to predict recurrence for this superseded method [
13]. To the best of our knowledge, this is the first study to examine the relationship between HRV and AF recurrence for more than 300 patients while at the same time including patients after CB1, CB2, and RF. Thus, while comparing well with the aforementioned studies in confirming the existence of said relationship with a larger sample, we also add novelty regarding the notion that the strength of this relation might depend on the ablation method.
Taking into consideration that rMSSD “is the primary time-domain measure used to estimate vagally-mediated changes reflected in HRV” [
14], our finding of rMSSD as the sole independent predictor for recurrence is consistent with the hypothesized vagal withdrawal happening during ablation. Pappone et al. were among the first to report a link between HRV attenuation and AF recurrence. They mapped parasympathetic innervation around the pulmonary veins and proposed the addition of complete vagal denervation to PVI after they found it to significantly reduce AF recurrence [
15], thus identifying a new potential target site for ablation. The intrinsic part of the extensive autonomous innervation of the heart is made of ganglionated plexi (GP), which are located mainly in fat pads around the pulmonary vein ostia [
4]. While it has been shown that adjunctive ablation of these GP can increase the success rate of pulmonary vein ablation [
4], this technique is not well established. It has been demonstrated that PVI alone attenuates HRV [
12,
13] and a more recent study observed vagal responses and HRV attenuation in a similar fashion for both PVI and PVI with GP ablation [
16], suggesting that PVI alone might produce GP damage to at least some degree. With the knowledge that the origin and maintenance of AF are partly attributed to a disbalance in the ANS [
3], this could explain why HRV could be a useful predictor for AF recurrence.
Concerning our conclusion that HRV is better suited to predict recurrence after CB2 than after RF, we hypothesized that this could be due to the circumstance that ablation via cryoballoon increases myocardium-specific enzymes more strongly after the ablation and causes wider lesions than RF [
17], suggesting that CB2 creates more extensive lesions and thus modulates/destroys more GP. The same mechanism could explain why for CB1, HRV is not as applicable for predicting AF recurrence: only four vs. six cryosource emission holes as used with CB2 result in a smaller cooling surface area [
18] and potentially in fewer destroyed GP.
Apart from confirming rMSSD as an independent predictor for AF recurrence, our Cox regression also suggests that suffering from heart failure increases the risk for recurrence, which is mostly consistent with previous findings [
19]. However, suffering from coronary disease or hypertension seems to be a protective factor in our study, which stands in contrast to previous findings [
20]. There are multiple possible explanations for this discrepancy: Firstly, Hiraya et al. reported a lower risk of recurrence for patients with coronary disease who underwent percutaneous coronary intervention [
20] and we did not document how many of our patients with coronary disease might have profited from that measure. Secondly, since our co-occurrence of coronary disease and heart failure (see Tables 4 and 5 in supplementary data) is far lower than described in the literature [
21] and our study was conducted retrospectively, it may be that one or both comorbidities were poorly documented. The third approach to an explanation also applies to hypertension being a protective factor, because it suggests that patients suffering from one of both conditions might profit from a more extensive antiarrhythmic drug treatment. The percentage of patients with coronary disease/hypertension receiving beta-blockers is significantly higher than the percentage of patients with beta-blockers without those conditions (see Tables 6 and 7 in supplementary data). However, our multivariate Cox-regression analysis could not detect a significantly decreased risk for recurrence for patients taking beta-blockers and on the contrary the trend pointed toward an increased risk for this group. A future prospective study would not have to rely on proper documentation in the past and could thus circumvent part of these issues.
Limitations
Our study was conducted retrospectively in a single center and the baseline characteristics were not all similar among the three different ablation methods. Since we only obtained HRV data shortly after the ablation, it is not safe to say that the difference between the recurrence and non-recurrence group appeared only after the procedure and did not already exist before it. To prevent patients from being wrongly categorized into a high-/low-risk group, it would have been more accurate to measure the difference in HRV before and after the ablation. Furthermore, our study might have experienced selection bias since we excluded patients whose Holter recordings were not suitable for HRV analysis, thus possibly ignoring a large fraction of patients with early recurrence.