Pulmonary vein isolation in patients with paroxysmal atrial fibrillation is associated with regional cardiac sympathetic denervation

Patients

Radionuclide (¹²³I-mIBG) imaging

According to a proposal for standardization of ¹²³I-mIBG cardiac sympathetic imaging by the EANM Cardiovascular Committee and the European Council of Nuclear Cardiology [13], all patients received 370 MBq of ¹²³I-mIBG (AdreView, GE Healthcare, Little Chalfont, Buckinghamshire, UK) and underwent anterior planar imaging of the thorax SPECT/CT imaging of the heart at 15 min and 4 h post injection. Data acquisition was performed on a hybrid two-slice SPECT/CT device (Symbia T2, Siemens Medical Solutions, Malvern, PA, USA). Emission data were acquired with parallel-hole, medium-energy, high-resolution collimators, the two detector heads positioned in a 90° angle, with a 20% symmetric energy window centered at 159 keV. Further acquisition parameters were 32 rotation steps, with 2.8° rotation per stop, 90° each head, and 25 s per rotational projection. All SPECT data were acquired electrocardiogram (ECG)-gated (eight ECG gates). The left ventricular end-systolic and end-diastolic volumes (ESV, EDV) and left ventricular ejection fraction (LVEF) were automatically calculated from the ECG-gated SPECT data of the late (4 h p.i.) study by means of the Corridor 4D-MSPECT software package (version 5.1, INVIA Medical Imaging Solutions, Ann Arbor, MI, USA). All SPECT data were corrected for attenuation by CT-attenuation correction through a low-dose spiral CT scan performed directly after acquisition of the emission data. CT scans were acquired during free breathing without ECG-gating, at spiral mode with pitch 1.4, tube voltage of 130 kV, 30 mAs, scan time 14.67 s, CTDI (computed tomography dose index) of approximately 3.2, and DLP (dose-length product) of 55. The CT images were reconstructed at a 5.0-mm thickness by using a reconstruction algorithm with a 512 × 512 matrix and a full-chest-size-adapted FOV of 50 × 50 cm².

Analysis

Computed tomography

CT angiography of the atria and pulmonary veins was performed in 15 out of 16 patients for preinterventional of PVI. All CT examinations were acquired using dual-source CT systems (either Somatom Definiton or Somatom Definiton Flash, Siemens Medical Solutions, Malvern, PA, USA). Examinations were performed at 100 kV according to a non-ECG-triggered high-pitch (3,2) protocol. Collimation was 64 × 0.6 mm² for the Somatom Definition and 128 × 0.6 mm² for the Somatom Definition Flash. A total of 60 ml of non-ionic, iodinated contrast agent (Ultravist 370, Bayer-Schering-Pharma, Berlin, Germany) was injected through a right antecubital vein at a flow rate of 5 ml/s, followed by a 50-ml saline chaser bolus at the same flow rate. Data acquisition was initiated by a bolus-triggering technique (CARE Bolus, Siemens Medical Solutions) with the region of interest placed within the left atrium. The threshold above the native reference scan and the delay before scanning started were 100 HU respectively 6 s for the Somatom Definition and 150 HU respective 10 s for the Somatom Definition Flash. Measurements of the left atrial size were performed on transverse 1-mm slices with a reconstruction increment of 1 mm using commercially available software (SyngoVolumetry, Siemens Medical Systems, Malvern, PA, USA). If ECG-trigger was available (1 case), a diastolic reconstruction interval was chosen by ECG; otherwise, the diastole (11 cases) was identified by evaluation of the valve's status. The pulmonary veins were excluded from the atrial volume.

Pulmonary vein isolation

PVI was carried out as described previously [3, 18]. All patients signed written informed consent and transesophageal echocardiography was performed in all patients prior to the procedure. The ablation was performed in the fasting state under sedation with midazolam and pain medication with piritramide as needed. Surface ECG and endocardial electrograms were continuously monitored on a computer-based digital amplifier/recorder system (Siemens Axiom Sensis XP or Prucka, GE Medical Systems Inc., Milwaukee, WI, USA). Two 6-F sheaths were placed in the left groin. A decapolar steerable catheter (St. Jude Medical, Inc., St. Paul, MN, or Inquiry, IBI, Irvine Biomedical, Inc., Irvine, CA, USA) was placed in the coronary sinus via one of the sheaths. Two 8-F sheaths were placed in the right groin, one being an SL1-sheath (Daig SL1, St. Jude Medical, Inc., Little Canada, MN, USA) for transseptal puncture. Selective angiography of the PV was performed with about 40 ml of non-ionic contrast (Ultravist 370, Bayer, Berlin, Germany), in posterior-anterior and left anterior oblique 60° projections. The 3D geometry of the left atrium was acquired using the Ensite NavX Velocity (St. Jude Medical) or the CARTO System (Biosense Webster, Diamond Bar, CA, USA) or the NaVX-Fusion System (St. Jude Medical, Inc.). The acquired geometry was fused with the left atrial 3D reconstruction of the computed tomographic (CT) data, previously performed with the Ensite Verismo software. Thereafter, an antral point-by-point circumferential ablation around the ipsilateral PV, with a distance of 0.5 to 1.0 cm from the ostia, using a 4-mm open-tip irrigated catheter (IBI Therapy Coolpath Duo 7-Fr, St. Jude Medical), was performed. The maximum power was set to 30 W, going selectively up to 40 W if the PV isolation could not be achieved, especially at the anterior ridge border of the lateral PVs. Temperature was limited to 43°C. Irrigation was adjusted manually between 17 and 30 ml/min. The complete electrical isolation was monitored and confirmed by the decapolar catheter during sinus rhythm and differential pacing maneuvers.

Statistics

Values are expressed as mean ± standard deviation. Differences in mean values were compared by either a Wilcoxon test or a Mann–Whitney U test for unpaired samples. A p < 0.05 indicated a statistically significant difference. All statistical analyses were performed using a commercially available software package (SPSS, version 17.0, SPSS Inc., Chicago IL, USA).

Cardiac sympathetic activity before and after PVI

Analysis of global sympathetic cardiac innervation

Mean late H/M ratios and global WRs did not change from baseline to the follow-up scan after PVI (pre, 2.9 ± 0.5 vs. post, 2.7 ± 0.6, p = 0.53 and pre, 26.8% ± 6.0% vs. post, 26.9% ± 3.6%, p = 0.67, respectively; Figure 1A,B).

Analysis of regional sympathetic cardiac innervation

At baseline, no regional defects were observed in the early (15 min p.i.), except in four patients (25%) in the late (4 h p.i.) SPECT images. These defects were located in the inferolateral (three patients) or the lateral wall (one patient). After PVI, regional deficits were present in two patients in the early but in the ten patients (62.5%) in the late images including the four patients with preexisting deficits which remained unchanged. Newly emerging defects post PVI were localized in the inferolateral (four patients) and the apical or basal posterior wall (one patient each), respectively (patient example, Figure 2). Semiquantitative analysis of all subjects showed a significant increase in the mean summed defect score (4 h p.i.) from baseline to post PVI (pre, 0.8 ± 1.3 vs. post, 2.6 ± 2.7, p = 0.011; Figure 3). The quantitative SPECT analysis comparing the late (4 h p.i.) ¹²³I-mIBG uptake before and after PVI showed a significant reduction of uptake in the left ventricular segments 4, 5, and 10 and a significantly increased regional washout in the segments 3, 4, 5, and 11 (Figure 4). In contrast, H/M ratios did not differ significantly between patients with (n = 10) and patients without regional defects (n = 6) after PVI (2.8 ± 0.6 vs. 2.7 ± 0.7, p = not significant (ns)). Accordingly, global WRs were similar between these both groups (26.8% ± 5.0% vs. 27.8% ± 7.2%, p = ns).

Innervation defects and AF relapses

In the short-term follow-up period of 6 months, four out of ten patients (40%) with regional innervation defects after PVI suffered from an early recurrence of AF. In contrast, only one of the six patients (17%) without detectable regional defects of myocardial ¹²³I-mIBG uptake after PVI had a clinically apparent early relapse of AF (Figure 5). In case of the four patients with preexisting and persisting defects, one patient (25%) suffered from a clinical relapse. However, the mean SDS after PVI were slightly but not significantly higher in patients with and AF relapses compared with patients without relapses (no relapse n = 11:2.4 ± 1.5 vs. relapse n = 4:2.7 ± 3.1; p = ns). The formally calculated sensitivity, specificity, positive predictive value and negative predictive value were 80%, 45%, 40%, and 83%.

Left atrial volumes and AF relapse

Except one patient, left atrial volumes were calculated on the basis of contrast enhanced CT data. Patients suffering from a relapse showed a tendency towards larger left atrial volumes [95 ± 23 ml vs. 132 ± 36 ml (no relapse vs. relapse; n = 11 vs. 4)]. However, statistical analysis revealed no significant differences (p = ns).

Left ventricular function and AF relapse

Left ventricular volumes (ESV, EDV) and LVEF were preserved in every patient and did not change significantly after PVI (ESV, pre 36 ± 14 ml vs. post 35 ± 15 ml; EDV, pre 107 ± 24 ml vs. post 102 ± 28 ml; LVEF, pre 66% ± 7% vs. post 67% ± 9%; all p = ns). These parameters did not differ between patients without or with AF relapse either [ESV, 37 ± 10 ml vs. 33 ± 20 ml; EDV, 106 ± 19 ml vs. 108 ± 44 ml; LVEF, 65% ± 6% vs. 70% ± 7%; all p = ns (no relapse vs. relapse; n = 11 vs. 4)].

PVI causes changes in cardiac sympathetic innervation

Solid evidence supports the hypothesis that the sympathetic cardiac nervous system plays a key role in the initiation and propagation of cardiac arrhythmias including paroxysmal AF [19–22]. Data from animal studies suggest that sympathetic innervation changes lead to alterations rendering the atrium more vulnerable to the arrhythmogenic triggers originating from the pulmonary veins [23–26]. Recently, Akutsu et al. performed a cardiac ¹²³I-mIBG study in patients with paroxysmal AF [8]. They found that a reduced H/M ratio was a powerful independent predictor of the development of permanent AF and heart failure plus permanent AF. Arimoto et al. were the first to perform cardiac ¹²³I-mIBG imaging in patients with AF after treatment with PVI [9]. In contrast to our data, they reported a high WR as an independent predictor of AF recurrence in patients with paroxysmal or permanent AF, a finding which is possibly missed by our study due to the size of the patient cohort. On the contrary, a subgroup of 20 patients underwent ¹²³I-mIBG imaging before and after PVI. In this group neither the H/M ratio nor the global WR changed significantly after treatment. This finding is in line with our results, but in contrast to the study by Arimoto et al. [9], we additionally assessed segmental tracer uptake known to improve the sensitivity and specificity of the imaging technique [27]. Hence, we observed regionally increased washout and reduced late segmental ¹²³I-mIBG uptake in 25% of the patients already before PVI and newly emerging innervation defects in 37.5% of patients after PVI.

Changes in sympathetic innervation are located in the posterolateral wall

¹²³I-mIBG SPECT/CT analysis shows that the regional defects are located in the inferior lateral wall of the left ventricle, both before and after PVI, but more frequently after PVI. Interestingly, while atrial ganglionated plexuses seem to be located in multiple locations on atrial chamber walls [21], different studies suggest that ventricular ganglionated plexuses appear to be primarily associated with the origins of major cardiac blood vessels. In addition, there is a high density of nerves near the venous ostia and along the pulmonary veins. Various anatomical studies (reviewed in [21]) implicate that these nerves appear to originate in the ganglionated plexuses along the middle and dorsum of the right atrium, while the corresponding nerve endings seem to penetrate mostly at the root of the veins.

As a firstly intriguing explanation, the observed innervation defects could be interpreted as a consequence of direct denervation due to ablation of sympathetic nerve fibers around the pulmonary veins similar to the results of a study performed by Mabucchi et al. [28]. These authors described a partial left ventricular sympathetic denervation assessed by cardiac ¹²³I-mIBG imaging after mitral valve surgery. Alternatively, the degree of rhythm control following PVI might offer another interpretation of our results. In this model, the interplay between sinus rhythm and atrial fibrillation episodes alters sympathetic innervation and therefore leads to regional innervation imbalances. More specifically, the occurrence of paroxysmal AF with short duration may cause the appearance of regional denervation. Hence, ventricular innervation imbalances may reflect primarily atrial innervation imbalances.

Regional changes in sympathetic innervation affect both uptake and washout

The results from our study show convincingly that the changes in sympathetic innervation affect both uptake and washout. A number of studies dealing with cardiac ¹²³I-mIBG imaging in several cardiac diseases have noted decreased global cardiac ¹²³I-mIBG uptake (H/M ratio) in combination with increased global washout. In a variety of different conditions such as congestive heart failure, coronary artery disease, and inherited arrhythmias, such changes have been demonstrated to indicate disease progression or an adverse prognosis [8, 29–35]. Additionally, pharmaceutical and device therapy in some of these conditions has resulted in recession of these findings [36–38]. The fact that different cardiac diseases are associated with the changes in cardiac uptake or washout of ¹²³I-mIBG suggests the presence of a common reaction of the cardiac sympathetic nervous system [39]. It has been suggested that one such mechanism could be a high rate of post-ganglionic cardiac sympathetic traffic whereby the locally increased rate of translocation of neurotransmitter vesicles to the membrane surface and an increased rate of exocytosis result in decreased regional ¹²³I-mIBG uptake and higher washout. The results of our study point to the development of regional imbalances of sympathetic innervation with either increased regional sympathetic tone (reflected by the increased regional washout) or innervation defects in patients with PAF (paroxysmal atrial fibrillation) both before and after PVI.

Since sympathetic innervation is considered to be a highly dynamic process with an ability to self-regenerate [28], the findings suggest that the cardiac sympathetic nervous system might play a role both in the remodeling process that is associated with paroxysmal AF and the reverse remodeling following PVI. Thus, the remodeling process of the sympathetic nervous system might have an impact on the likelihood to develop an AF relapse in the short-term course following PVI. Clinical studies indicate that this period of time is crucial for the atrial reverse remodeling process that determines the probability of sinus rhythm maintenance [2, 40–42]. In patients with more intense atrial remodeling the remodeling process may result in a more extensive sympathetic damage and subsequently to more severe left ventricular denervation.

Regional sympathetic innervation deficits might help identify patients at risk for AF relapse

Hence, we hypothesize that regional defects of cardiac ¹²³I-mIBG uptake and increased regional washout are hallmarks of sympathetic innervation imbalances, which may be useful parameters for the non-invasive risk stratification of AF occurrences. Although based on findings in a relatively small patient cohort the results of our study suggest a certain prognostic value of ¹²³I-mIBG SPECT/CT. As the absence of innervation deficits was indicative for a better outcome with respect to the occurrence of AF relapses, the negative predictive value of ¹²³I-mIBG SPECT/CT may be high, a preliminary finding which has to be interpreted with caution and has to be proved in a larger patient cohort. Apart from that, patients with larger left atrial volumes tend to have more AF relapses following PVI [40, 41]. Accordingly, patients with AF relapses in our study tended to have larger left atrial volumes. However, the differences in atrial size in comparison to patients without AF relapse did not reach the level of statistical significance.

Left ventricular function was preserved in our patient cohort, did not change after PVI, and did not correlate with AF relapses. Therefore, our imaging results should not be confounded by the presence of heart failure. Moreover, relevant coronary artery disease was excluded in our patient cohort.

Further studies with larger patient cohorts are needed to clarify the value of ¹²³I-mIBG SPECT/CT for risk stratification and adequate patient selection of AF patients for PVI. It remains to be determined whether regional sympathetic innervation deficits are a surrogate parameter for incomplete PVI with respect to the middle and/or long-term outcome. In addition, further research is necessary to decipher the effect of cardiac sympathetic innervation inhomogeneity on susceptibility to AF in patients suffering from AF paroxysms and on AF relapses in patients following PVI.

Limitations

This observational study does not contain data from a control group treated without PVI, since we compare cardiac innervation status before PVI with that after PVI. Beyond patient-related issues, the imaging method chosen in this clinical study allows for ventricular imaging only, since due to the low mass of the atria only left ventricular innervation can be assessed and local sympathetic innervation changes around the pulmonary veins could not be adequately measured.

Measurements of left atrial volumes are not exact since the majority of CT scans were performed non-ECG-triggered.