Introduction

The heart is richly innervated by the sympathetic and parasympathetic (vagal) nerves1,2,3. Normal autonomic nervous system (ANS) coordination is critical for cardiac function maintaining1,2,3. However, disorganized ANS activation can induce atrial fibrillation (AF)1,2,3,4. The overall role of sympathetic nerves in the heart is to modulate the automaticity of sinoatrial node and the conductivity of atrioventricular node1,2,3. Abnormal adrenergic activation may promote focal activity of atrial myocytes via enhanced automaticity, early afterdepolarization or delayed afterdepolarization; which may act as a trigger of AF or an AF-maintaining driver1,2,3. The general effect of vagal nerve on the heart is to reduce the automaticity of sinoatrial node and the conductivity of atrioventricular node1,2,3. Abnormal parasympathetic activation produces spatially heterogeneous action potential and refractory period abbreviation, which may promote the occurrence and maintenance of re-entrant activity3.

Recent basic science concepts and clinical catheter-based AF-ablation techniques showed that most paroxysmal AF is triggered by ectopic firing originating from the cardiomyocyte sleeves extending to the pulmonary vein (PV) and the junction between PV and left atrium5,6,7,8,9. Thus, PV isolation (PVI) rapidly replaced focal ablation and emerged as a main therapeutic strategy to treat AF9,10,11,12. The efficacy and major complications of catheter ablations are now approximately 70–80% and 5% respectively5,8,9.

Despite encouraging achievements of catheter ablation in AF therapy, how to improve the long-term efficacy requires further knowledge on how ablation rectified the ANS modulation in cardiac electrophysiology5,9. Current measures of cardiac electrophysiology, including conventional heart rate variability (HRV), are hard to distinguish the respective role of sympathetic and parasympathetic nerves in the cardiac pathophysiology of AF13,14. In 2006, Baver et al. created a series of signal processing technology and algorithm to separately characterize the deceleration and acceleration capacities of the heart rate using 24-h ambulatory electrocardiogram15. Their calculation principle and later reports suggested that AC and DC might distinguish and quantify between vagal and sympathetic nervous system roles that affect cardiac electrophysiology15,16,17.

In this report, to evaluate whether AC/DC have the potential to discriminate AF recurrence, 154 qualified paroxysmal AF subjects were assembled. The dynamic change trend of AC and DC before and after CPVI was compared between recurrent and recurrence-free subjects.

Patients and Methods

Ethical issues

The review board of the First Affiliated Hospital of Soochow University approved this protocol in accordance with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. Written informed consent regarding procedures and medical data were obtained from all of the patients according to the guidelines of the Chinese National Ethics Regulation Committee. All patients were informed of their rights to withdraw consent personally or via kin, caretakers, or guardians.

Subjects and preoperative preparation

All subjects involved in this study were drug-resistant paroxysmal AF patients. Paroxysmal AF is defined as an episode of AF that terminates spontaneously or with intervention within 7 days of onset or episodes may recur with variable frequency3. Drug-resistant paroxysmal AF was defined at least 2 ineffective antiarrhythmic drugs in suppressing AF recurrence3.

Initially, 219 patients with paroxysmal AF were treated with radiofrequency ablation in our Department of Cardiology from January 2013 to November 2015 (For details, see results). 24 hours before CPVI, transesophageal echocardiography was performed to exclude the possibility of thrombus in left atrial appendage and left atrium, the anatomical structures of the left atrium and pulmonary vein were determined by CT scan. Patients with a previous catheter ablation; coronary insufficiency; recent myocardial infarction; clinical symptoms of heart failure; relevant valvular dysfunction; terminal renal insufficiency, obstructive sleep apnea, chronic obstructive pulmonary disease or any other relevant pulmonary disease; hyperthyroidism/hypothyroidism; inflammatory diseases; diabetes and sick sinus syndrome were excluded from this study.

Circumferential pulmonary vein isolation (CPVI)

Under the guidance of three-dimensional mapping system (EnSite 3000, St. Jude Medical Inc., Little Canada, MN), CPVI was performed using cold saline perfusion (flow rate 17 mL/min, 30–35 Watts). Under local anesthesia; puncture the left femoral vein; insert 10 polar sinus electrodes to the distal coronary vein and fix with skin paste; puncture the right femoral vein twice; insert the Agilis and SL1 long sheaths (St. Jude Medical Inc., Little Canada, MN) respectively. The puncture needle (St. Jude Medical Inc., Little Canada, MN) is sent to the atrial septum via the SL1 long sheath; SL1 long sheath are sent into the left atrium after piercing atrial septum. The electrodes of saline infusion radiofrequency catheter ablation are sent to left atrium via the Agilis long sheath. Under the positioning of the EnsiteNavx (St. Jude Medical Inc., Little Canada, MN), complete the following procedures: construction of the three-dimensional model of left atrium; mapping the potentials of the left superior, left inferior, right superior and right inferior pulmonary veins; determination of pulmonary vein orifice and left atrial appendage position; setting the circumferential pulmonary vein ablation line; loading the electrodes of saline infusion radiofrequency catheter ablation; the left and right pulmonary vein macrocyclic isolation along the ablation line. Dormant conduction was assessed by isoproterenol infusion, pulmonary vein pacing, coronary vein and intra atrial pacing.

Postoperative treatment and follow-up

All patients received postoperative dabigatran (110 mg, bid) for anticoagulation; amiodarone (0.2 g, qd) to maintain sinus rhythm; and omeprazole (20 mg, qd) to suppress gastric acid for 3 months. Patients with discomfort will be submitted for the ECG monitoring. The cardiac electrophysiology and echocardiography of all patients will be evaluated using 24-h ambulatory electrocardiograms and a Sonos 5500 type ultrasound machine (Philips, Best, Netherlands) three days, 3 months and 6 months after operation to learn the efficacy and complications of CPVI.

Definition of postoperative recurrence of AF

The time point to define postoperative recurrence of AF has not yet unified. Three months after radiofrequency ablation is a blank period3. According to the “2017 HRS/EHRA/ECAS/APHRS/SOLAECE expert consensus statement on catheter and surgical ablation of atrial fibrillation”, the recurrence of AF within 3 months was defined as early recurrence3. In this study, the average time of the third time point (3 months after CPVI) was 90 ± 5 days, so, paroxysmal atrial arrhythmias; such as atrial tachycardia, atrial flutter and AF lasting more than 30 seconds detected at this time point was defined as early recurrence. AF recurrence > 30 s recorded at the fourth time point (6 months after CPVI) was defined as late recurrence.

Calculation of DC and AC using 24 h ambulatory electrocardiograms

To monitor the cardiac electrophysiology of all participates before and after ablation, a Holter monitor test was performed using a Holter Monitoring DigiTrak XT Holter System (Philips, Best, Netherlands).

A phase-rectified signal averaging (PRSA) algorithm17 to the RR interval series, which is capable of detecting and quantifying quasi-periodic oscillations masked by nonperiodic components, artifacts and ectopic beats, was adopted in analysis of 24-h ambulatory electrocardiograms. The DC and AC were computed with timescales (T) = 1 and wavelet scales (s) = 2. To avoid errors caused by artifacts and ectopic rhythm; abnormal RR intervals, defined as RR intervals that change by more than 20% from the previous RR interval, were removed automatically; sinus arrhythmia and ventricular rhythm were further excluded manually according to the P waves morphology and QRS waves shape respectively.

Briefly, step 1, Definition of decelerating and accelerating anchors: heartbeat intervals longer than the preceding interval were identified as decelerating anchors; heartbeat intervals shorter than the preceding interval were identified as accelerating anchors. Step 2, definition of cardiac electrical segments: segments of interval data around the decelerating and accelerating anchors are selected. Step 3, phase rectification: the cardiac electrical segments described above were aligned at the decelerating and accelerating anchors. Step 4, signal averaging: the phase-rectified signal averaging signal X(i) was obtained by averaging the signals within the aligned cardiac electrical segments. Step 5, quantification of deceleration and acceleration capacities using the following formula: DC (AC) = [X(0) + X(1) − X(−1) − X(−2)]/4.

HRV measure and analysis

According to the guidelines: Heart rate variability, Standards of measurement, physiological interpretation, and clinical use14, the HRV parameters were computed using same set of 24 h ambulatory electrocardiograms while calculating AC and DC. The quality control of the RR interval series was performed as AC/DC computation, that is, abnormal RR intervals, defined as RR intervals that change by more than 20% from the previous RR interval, were removed automatically; sinus arrhythmia and ventricular rhythm were further excluded manually according to the P waves morphology and QRS waves shape respectively. Indices of time-domain methods include the full-course standard deviation of NN (NN is used in place of RR to emphasize that the processed beats are normal beats) intervals (SDNN); root mean square of successive differences (RMSSD, which refers to the square root of the mean of the squares of the successive differences between adjacent NNs); indices of frequency-domain methods, which assign bands of frequency and then count the number of NN intervals that match each band, these include high frequency (HF), low frequency (LF) and the LF/HF; and average heart rate; were computed using Kubios HRV analysis software (http://kubios.uef.fi).

Echocardiography

As previously described16, a transthoracic echocardiographic examination was performed using a Sonos 5500 type Ultrasound machine (Philips, Best, Netherlands) with a 2.5 Hz transducer. The parameters measured with the M-mode technique included left ventricular end-diastolic, end-systolic and left atrial diameters. The measurement of the left ventricular ejection fraction was performed using the Simpson’s biplane method.

All of the echocardiographic and above cardiac electrophysiological examinations were performed and analyzed licensed technicians who were blinded to clinical data and group division.

Statistical analysis

Continuous variables are presented as means ± standard deviation (SD); a non-paired Student’s t-test were used for comparison between groups. Categorical data are summarized as frequencies and percentages and compared using chi-square test and/or Fisher’s exact test. Kaplan–Meier analyses with log-rank tests were used to calculate AF recurrence-free survival over time and compare recurrence rates between groups. The performance of discrimination was evaluated by an area under receiver operating characteristic (ROC) curve (AUC). P values less than 0.05 (two-sided) were deemed statistically significant. Statistical analyses were performed using SPSS version 17.0 (SPSS Inc., Chicago, IL, USA).

Results

Participants

219 patients with drug refractory paroxysmal AF were collected initially from January 2013 to November 2015. Of whom, 16, 5, 2, 5, 12, 5, 10, 4 and 6 individuals were excluded from this reports due to typical atrial flutter, refusing ablation, incomplete CPVI due to pericardial tamponade, tricuspid isthmus ablation, left atrial linear ablation, complication of atrial tachycardia, superior vena cava isolation, ablation of left-sided atrioventricular accessory pathways and lost to follow-up, respectively. Finally, 154 qualified patients with paroxysmal AF were involved in this study (Fig. 1). Of the 154 participants; their average age was 60.0 ± 10.1 years old; 94 (61.0%) and 60 (39.0%) were males and females; the mean PAF duration was 3.5 ± 6.0 years; 86 (55.8%) and 7 (4.6%), complicated with hyptertension and CHD respectively; 85 (55.2%), 26 (16.9%) and 74 (48.1%) were treated with aminodarone, propafenone and metoprolol respectively; the DC, AC, AHR, SDNN, RMSSD and LF/HF were 7.15 ± 2.15 ms, −7.46 ± 2.21 ms, 66.6 ± 13.0 bpm, 144.1 ± 55.5 ms, 38.2 ± 23.9 ms and 1.76 ± 1.33 respectively; the LAd and LVEF were 40.3 ± 5.6 mm and 0.68 ± 0.06 respectively (Table 1).

Figure 1
figure 1

Study flowchart of patient selection. AF, atrial fibrillation; CPVI, circumferential pulmonary vein isolation.

Full size image
Table 1 Characteristics of the subjects (n = 154).
Full size table

AC, DC and HRV changes in AF patients before and after CPVI

First, let’s look at DC: the DCs of the recurrent group before CPVI and three days, 3 months and 6 months after CPVI were 7.39 ± 2.90, 4.36 ± 1.46, 6.08 ± 2.09 and 5.89 ± 1.65 ms respectively; the DC decreased significantly at three days after CPVI; although DCs elevated 3 and 6 months after CPVI, it kept still significantly lower than that before CPVI (Fig. 2A). The DCs of the recurrence-free group before CPVI and three days, 3 months and 6 months after CPVI were 7.06 ± 1.77, 3.79 ± 1.18, 4.22 ± 1.96 and 3.97 ± 0.98 ms respectively; the DC also decreased significantly three days after CPVI and kept at relatively low level through out 6 months; impressively, the DCs were significantly lower in recurrence-free group than these in recurrent group at each time point (Fig. 2A).

Figure 2
figure 2

The change tendency of each index before and after CPVI of the recurrence-free and recurrence groups. *indicates the significance at each time point after CPVI when compared with that before CPVI. #indicates the significance at a time point between recurrence-free group and recurrence group. For abbreviation, please see Table 1 or Table 2.

Full size image

Next, let’s look at AC: the ACs of the recurrent group before CPVI and three days, 3 months and 6 months after CPVI were −7.78 ± 2.76, −4.69 ± 1.62, −6.36 ± 2.34 and −6.16 ± 1.78 ms respectively; the AC increased significantly three days after CPVI; although ACs declined 3 and 6 months after CPVI, it kept still significantly higher than that before CPVI (Fig. 2B). The ACs of the recurrence-free group before CPVI and three days, 3 months and 6 months after CPVI were −7.33 ± 1.95, −4.09 ± 1.49, −4.38 ± 2.10 and −3.92 ± 1.01 ms respectively; the AC also increased significantly three days after CPVI and kept at relatively high level through out 6 months; impressively again, the DCs were significantly higher in recurrence-free group than these in recurrent group at each time point (Fig. 2B).

Then, let’s look at other indices: of the recurrent group, three days after the CPVI, the AHR was significantly higher and all others except RMSSD were significantly lower than those before CPVI; of the recurrence-free group, three days after the CPVI, the AHR was significantly higher and all others indies were significantly lower than those before CPVI (Fig. 2C–F); different from AC and DC, no stable difference trend was found between two groups, although “significant differences” emerged frequently (Fig. 2C–F).

In conclusion, AC and DC displayed a relative superiority to discriminate the recurrence-free and recurrence subjects.

CPVI induces the AC and DC changes

To know further the changes of AC, DC and the differences of these indices between recurrence-free and recurrent groups were caused by ablation, the baseline data of recurrence-free and recurrence groups were compared. As showed in Table 2, before the ablation, no significant difference between two groups, including disease history, complications, treatment, electrophysiology and echocardiography was found. As Amiodarone, propafenone and metoprolol administration might alter the electrocardiographic activity. The effect of these medicines on DC and AC was further assessed by comparing the mean AC and DC of treated and untreated subjects. As showed in Table 3, none of the three medicines changed the HRV indexes or AC and DC significantly. The above analyses suggest that CPVI induces the AC and DC changes.

Table 2 Baseline data of recurrence-free and early recurrent groups before CPVI.
Full size table
Table 3 Medication had not changed the HRV, AC and DC significantly.
Full size table

Recurrence-free rates delimited by the cut off values of AC and DC

The above data showed that AC and DC displayed certain potentials to discriminate recurrence and recurrence-free groups after ablation, we then tried to perform Kaplan–Meier analysis using cut off values of AC and DC.

Firstly, the receiver operating characteristic (ROC) curve was plotted using AC and DC values of each patient. The area under ROC curve (AUC) of AC and DC were 0.747 (95% CI 0.618 to 0.875) and 0.758 (95% CI 0.630to 0.886) respectively. As expected, AC and DC displayed good performance to discriminate recurrence-free subjects. The pointcuts of AC and DC were −5.1 and 4.8 ms respectively.

Next, the recurrence-free rates were analyzed using above cut off values. During 6 months of follow-up, 34 (22.1%) and 54 (35.1%) of the 154 patients showed AF recurrence within 3 and 6 months following up, respectively. Kaplan–Meier analysis according to DC (DC < 4.8 ms, log-rank, Chi-square = 159.149, P < 0.001; Fig. 3) showed significant benefit for clinical outcome after catheter ablation. Also, Kaplan–Meier analysis according to AC (AC ≥ -5.1 ms,log-rank, Chi-square = 163.907, P < 0.001; Fig. 3) showed significant high AF-free survival rate after catheter ablation.

Figure 3
figure 3

Kaplan–Meier analysis of AF recurrence-free rate according to AC and DC. A. Recurrence-free rates in 6 months follow-up of DC < 4.8 ms vs. DC ≥ 4.8 ms. B. Recurrence-free rate in 6 months follow-up of AC ≥ −5.1 ms vs. AC < −5.1 ms.

Full size image

Discussion

In this report, a novel measure was adopted to explore the eletroneurophysiological properties in AF patients before and after ablation. Firstly, ablation significantly reduced the DCs both in recurrent and recurrence-free groups; the mean DC of recurrent group had recovered significantly 3 and 6 months after ablation; however, the mean DC of recurrence-free group kept at a relative low level through out the 6 months; the mean DCs were significantly lower in recurrence-free group than these in recurrent group at each time point. Secondly, ablation significantly increased the ACs both in recurrent and recurrence-free groups; the mean AC of recurrent group had dropped down significantly 3 and 6 months after ablation; impressively, the mean AC of recurrence-free group kept at a relative high level; the ACs were significantly higher in recurrence-free group than these in recurrent group at each time point. Thirdly, further Kaplan-Meier analysis showed that subjects with DC < 4.8 ms or AC ≥ −5.1 ms displayed significantly higher recurrence-free rates compared to subjects with DC ≥ 4.8 ms or AC < −5.1 ms. Finally, other HRV indices, such as AHR, SDNN, RMSSD and LF/HF were erratic in two groups before and after ablation. These results suggested that AC and DC displayed a relative superiority to early discriminate the recurrence-free and recurrent subjects.

As we known, to quantify and/or to distinguish between the cardiac sympathetic and parasympathetic neural modulations are hard problems13,14. Theoretically, the AC and DC could reflect the cardiac sympathetic and parasympathetic modulations on heart respectively15,16,17. Suppose above theory is correct; in our report; the significant decline of the absolute values of AC and DC three days after ablation suggests patients with AF were in a higher DC and lower AC state before ablation; alternatively; patients with AF were in a lower sympathetic and higher parasympathetic modulating state before ablation; this deduction is in line with some reports18,19. In addition, lower DC associated with higher AF-free rate after ablation suggest that vagal denervation and destroy of ectopic firing originating from the cardiomyocyte sleeves are critical for AF therapy19. While, higher AC should be a secondary phenomenon, that is, high sympathetic activity is most likely to be caused by vagal denervation, which weakened antagonism to sympathetic activity.

A recent study using a cardiovascular system model showed that the eletroneurophysiological properties of DC and AC were highly influenced by the timescales (T) and wavelet scales (s) used in the computation20. Either DC or AC was solely dependent on vagal activity under the scales of T = 1 and s = 2; however, with the scales of T = 3 and s = 5, both DC and AC were correlated positively to sympathetic activity and negatively to vagal activity. In our study, the DC and AC were computed with T = 1 and s = 2. If the above theory is adopted to interpret our results, high DC and low AC might predict more severe disorders of vagal modulation in heart eletroneurophysiology and then easily cause early AF recurrence.

The recurrence of AF might be caused by inadequate autonomic denervation or neuroplasticity2,3,21. In our study, 35.1% subjects displayed clear evidence of AF recurrence 3 months after CPVI. Since dormant conduction was assessed by isoproterenol infusion, pulmonary vein pacing, coronary vein and intra atrial pacing during CPVI, the early recurrence of AF is more likely to be caused by the ganglionic plexi renovation and/or PV-left atrial conductions recovery. Interestingly, AC and DC displayed a potential performance to discriminate the recurrence.

Our data showed that all the SDNN, RMSSD and LF/HF of recurrence-free and recurrence groups displayed a separation and coincidence tendency after the ablation, which might reflect the injury and repair of PV after ablation. Nevertheless, no stable tendency of the HRV parameters was found. A study examining 636 Chinese subjects without heart disease showed that DC, SDNN, RMSSD varied by age, gender and circadian rhythm22. DC decreased gradually with age increase in normal population, the average DCs of subjects aged 50–59 and 60–70 years old were 6.61 ± 1.38 and 6.18 ± 1.93 ms respectively. In our study, the average age of the subjects was 60.0 ± 10.1 years old, their average DC before ablation (7.15 ± 2.15 ms) was higher than above normal Chinese with similar age. This difference may also be explained by that AF associated with higher vagal tension18,19.

In our report, 85 (55.2%), 26 (16.9%) and 74 (48.1%) subjects were treated with amiodarone,propafenone and metoprolol, respectively. As we know, amiodarone slows conduction rate and prolongs the refractory period of the SA and AV nodes;23 propafenone slows the influx of sodium ions into the cardiac muscle cells, causing a decrease in excitability of the cells;24 and metoprolol blocks β1 adrenergic receptors in heart muscle cells, thereby decreasing the slope of phase 4 in the nodal action potential and prolonging repolarization of phase 325. In general, these medicines are mainly to control arrhythmia by affecting the action potential of cardiac myocytes. DC and AC are derived from the heartbeat intervals. Thus, amiodarone, propafenone and metoprolol administration should have limited effect on AC and DC. Be the evidence, our data showed that none of the three medicines changed the HRV indexes or AC and DC significantly, which is also consistent with other reports partially26,27.

Since Kantelhardt JW et al. applied the PRSA algorithm in AC and DC of the heart rate17, more and more studies have found that AC and DC are related not only to physiological activity but also to a number of pathophysiological states. DC decreased gradually with age increase in normal population;22 the magnitudes of DC and AC increased progressively as breathing frequency decreased;28 lower DC of heart rate is a strong and independent predictor of 1-year mortality in patients with severe aortic stenosis undergoing transcatheter aortic valve implantation;29 AC is positively associated with heart failure grade in patients with dilated cardiomyopathy;16 lower DC assessed from short-term recordings is a strong and independent predictor of mortality and cardiovascular mortality after myocardial infarction30. These reports showed the complex role of ANS in the physiological and pathological process. More extensive and deeper researches are needed to figure out the overview of ANS.

There are several limitations in our study, firstly, we had only followed up the patients for only 6 months, we will continue to observe these patients and will provide long-term data in our next study; secondly; we computed the AC and DC under scales T = 1 and s = 2 only, we had not computed the AC/DC using other scales.