J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y VO L . 4 , N O . 5 , 2 0 1 8

ª 2 0 1 8 B Y T H E A M E R I C A N C O L L E G E O F C A R D I O L O G Y F O U N D A T I O N

P U B L I S H E D B Y E L S E V I E R

Effects of Stellate Ganglion Cryoablationon Subcutaneous Nerve Activity andAtrial Tachyarrhythmias in a CanineModel of Pacing-Induced Heart Failure

Richard S. Shelton, MD,a Masahiro Ogawa, MD, PHD,a,b Hongbo Lin, MS,c Changyu Shen, PHD,c JohnsonWong, BS,a

Shien-Fong Lin, PHD,d Peng-Sheng Chen, MD,a Thomas H. Everett, IV, PHDa

ABSTRACT

ISS

Fro

Me

Ca

En

Gr

me

an

ha

Ch

All

an

Cli

Ma

OBJECTIVES This study aimed to test the hypothesis that subcutaneous nerve activity (SCNA) can adequately estimate

the cardiac sympathetic tone and the effects of cryoablation of the stellate ganglion in dogs with pacing-induced

heart failure (HF).

BACKGROUND Recording of SCNA is a new method to estimate sympathetic tone in dogs. HF is known to increase

sympathetic tone and atrial arrhythmias.

METHODS Twelve dogs with pacing-induced HF were studied using implanted radiotransmitters to record the stellate

ganglia nerve activity (SGNA), vagal nerve activity, and SCNA. Of these, 6 dogs (ablation group) underwent bilateral

stellate ganglia cryoablation before the rapid ventricular pacing; the remaining 6 dogs (control group) had rapid ven-

tricular pacing only. In both groups, SCNA was compared with SGNA and the occurrence of arrhythmias.

RESULTS SCNA invariably increased before the 360 identified atrial tachyarrhythmia episodes in the 6 control dogs

before and after HF induction. SCNA and SGNA correlated in all dogs with an average correlation coefficient of 0.64 (95%

confidence interval: 0.58 to 0.70). Cryoablation of bilateral stellate ganglia significantly reduced SCNA from 0.34 �0.033 mV to 0.25 � 0.028 mV (p ¼ 0.03) and eliminated all atrial tachyarrhythmias.

CONCLUSIONS SCNA can be used to estimate cardiac sympathetic tone in dogs with pacing-induced HF.

Cryoablation of the stellate ganglia reduced SCNA and arrhythmia vulnerability. (J Am Coll Cardiol EP 2018;4:686–95)

© 2018 by the American College of Cardiology Foundation.

S ympathetic tone measured by stellate ganglionnerve activity (SGNA) has been shown to influ-ence atrial electrophysiology and the onset of

atrial arrhythmias. We (1–3) recently proposed a newmethod to simultaneously record electrocardiogram(ECG) and subcutaneous nerve activity (SCNA) using

N 2405-500X/$36.00

m the aKrannert Institute of Cardiology and the Division of Cardiology, D

dicine, Indianapolis, Indiana; bDepartment of Cardiology, Fukuoka Univer

rdiology, Department ofMedicine, Indiana University School ofMedicine, I

gineering, National Chiao Tung University, Hsin-Chu, Taiwan. This study w

ants P01HL78931, R01HL71140, R41HL124741, and R42DA043391, U18TR00

nt; the Charles Fisch Cardiovascular ResearchAward endowed byDr. Suzan

d the Indiana University Health-Indiana University School of Medicine Stra

veequity interest inArrhythmotechLLC.,Cyberonics, andMedtronic; andSt

en’s research laboratory.

authors attest they are in compliancewith human studies committees and a

d Food and Drug Administration guidelines, including patient consent whe

nical Electrophysiology author instructions page.

nuscript received July 27, 2016; revised manuscript received January 23,

bipolar electrodes placed under the skin. The electricalsignals were low-pass filtered to optimize ECG andhigh-pass filtered to reveal SCNA. Because a good cor-relation has been shown between SCNA and the simul-taneously recorded SGNA (1), we proposed that SCNAmay be used to estimate cardiac sympathetic tone.

https://doi.org/10.1016/j.jacep.2018.02.003

epartment of Medicine, Indiana University School of

sity School of Medicine, Fukuoka, Japan; cDivision of

ndianapolis, Indiana; and the dInstitute of Biomedical

as supported in part by National Institutes of Health

2208, and R01HL139829; a Medtronic-Zipes Endow-

ne B. Knoebel of theKrannert Institute of Cardiology;

tegic Research Initiative. Drs. Lin, Wong, and Everett

. JudeMedical Inc. donated researchequipment toDr.

nimalwelfare regulations of the authors’ institutions

re appropriate. For more information, visit the JACC:

2018, accepted February 8, 2018.

AB BR E V I A T I O N S

AND ACRONYM S

aSGNA = averaged stellate

ganglia nerve activity

aSCNA = averaged

subcutaneous nerve activity

aVNA = averaged vagus nerve

activity

ECG = electrocardiogram

HF = heart failure

LCSD = left cardiac

sympathetic denervation

LSG = left stellate ganglion

PAT = paroxysmal atrial

tachycardia

RSG = right stellate ganglion

SCNA = subcutaneous nerve

activity

SGNA = stellate ganglia nerve

activity

VNA = vagus nerve activity

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8 Shelton et al.M A Y 2 0 1 8 : 6 8 6 – 9 5 SCNA in Heart Failure

687

The noninvasive estimation of sympathetic tone mayprovide new insights into the mechanisms of cardiacarrhythmias and may also provide a direct method tomeasure the success of neuromodulation procedures.Ogawa et al. (4,5) previously performed 2 studies indogs with pacing-induced heart failure (HF). In thefirst study, it was shown that HF increases SGNA andoccurrence of cardiac arrhythmias (4). In a secondstudy using the same HF model, cryoablation of thecaudal half of the left and right stellate ganglia andT2 to T4 thoracic sympathetic ganglia was performed.It was shown that cryoablation significantly reducedboth SGNA and paroxysmal atrial tachycardia (PAT)episodes (5). The data from these studies have notbeen analyzed to determine if SCNA increased after in-duction of HF and decreased after cryoablation of thestellate ganglion. We retrieved the data from bothstudies and manually analyzed de novo the SGNAand atrial tachyarrhythmia episodes. We also per-formed analyses of SCNA usingmethods previously re-ported (1). The purpose of the present study was to testthe hypothesis that: 1) HF increases both SCNA andSGNA; 2) both SCNA and SGNA preceded the onset ofPAT at baseline and after induction of HF; and 3) cryoa-blation of the stellate ganglion can reduce both SCNAand SGNA in ambulatory dogs.

SEE PAGE 696

METHODS

Original data from 2 previously published reports(4,5) were retrieved and manually analyzed. Theresearch protocols were approved by the InstitutionalAnimal Use and Care Committees of the Cedars-SinaiMedical Center, Los Angeles, California, and theIndiana University School of Medicine at the Meth-odist Research Institute, Indianapolis, Indiana.SURGICAL PREPARATIONS. The surgical prepara-tions have been described previously. Briefly, allsurgeries were performed under isoflurane anes-thesia. Subcutaneous bipolar electrodes were placedin the left thorax of dogs, 6 to 10 cm apart, to recordSCNA. The electrodes were stainless steel wires withthe terminal 5 mm of the wires stripped of theirinsulation and used for electrical recording. The leadswere attached to a Data Science International D70-EEE radiotransmitter (DSI, St. Paul, Minnesota),which was also implanted in the subcutaneous space.In the same procedure, through a left thoracotomy, apair of bipolar electrodes from the DSI radio-transmitter was sutured onto the left stellate ganglion(LSG) to record SGNA. Another set of bipolar elec-trodes was placed on the left thoracic vagal nerve torecord vagal nerve activity (VNA).

EXPERIMENTAL PROTOCOLS. In all dogs, apacing lead was implanted in the right ven-tricular apex and connected to a MedtronicItrel neurostimulator for high-rate ventricu-lar pacing. After recovery from the initialsurgery, the dogs had 2 weeks of continuousbaseline ambulatory monitoring. After thebaseline recording, the first group of 6 dogs(group 1) underwent right ventricular pacingat 150 beats/min for 3 days, at 200 beats/minfor 3 days, and then at 250 beats/min for 3weeks to induce HF, which was confirmed byECG (6). The pacemakers were then turnedoff to allow ambulatory monitoring for anadditional 2 weeks. In the second group of 6dogs (group 2), the caudal half of the leftstellate ganglion and T2 to T4 thoracic sym-pathetic ganglia were cryoablated throughthe left thoracotomy using a 7-cm SurgiFrostprobe (CryoCath Technologies, Inc., Mon-treal, Canada) during the same surgery in

which the DSI radiotransmitter was implanted formonitoring of SCNA, SGNA, and VNA. The dogs thenunderwent the same surgery and protocol for im-plantation of the Medtronic Itrel neurostimulator forsubsequent high-rate ventricular pacing to induceHF. These dogs were also monitored for 2 weeksbefore and after HF induction.

DATA ANALYSES. Recordings from the bipolar elec-trodes connected to the radiotransmitter wereanalyzed to obtain SCNA, SGNA, VNA, and ECG usingcustom-written software. The impedance of theelectrodes was 0.7 to 0.8 U and the recordings wereamplified 10,000 times and sampled at 1,000 Hz. Weobtained an ECG for analyses by applying a bandpassfilter (5 to 100 Hz) on the vagus nerve channel orsubcutaneous nerve channel of the radiotransmitter.SCNA, SGNA, and VNA were obtained by high-passfiltering at 150 Hz. The SGNA and VNA were recti-fied and integrated over 1 min. The SCNA was alsorectified and integrated over the same minute afteradjusting the SCNA with wavelet analysis asdescribed previously (7). In particular, spike-triggered averaging was performed to allow removalof the ventricular electrogram from the SCNA bysubtracting a ventricular electrogram template ob-tained from averaged ventricular electrograms in theobservational window. Finally, the ECG was used tocalculate the heart rate over the same minute. Thesum of all digitized nerve activity was then divided bythe total number of samples during the same periodto obtain averaged SGNA (aSGNA), averaged SCNA(aSCNA), and averaged VNA (aVNA) per sample.

FIGURE 1 Effect of Cryoablation on Nerve Activity Before Ventricular Pacing

Nerve activity was quantified by averaging the rectified, high-pass signal every minute for 10 min every hour for 24 h. Data before pacing are

shown as a box-and-whisker plot comparing the control group (group 1) versus dogs that had stellate ganglion cryoablation (group 2). A

significant decrease in nerve activity is seen for the SCNA, SGNA, and VNA in the ablation group. aSCNA ¼ averaged SCNA; aSGNA ¼ averaged

SGNA; aVNA ¼ averaged VNA; SCNA ¼ subcutaneous nerve activity; SGNA ¼ stellate ganglia nerve activity; VNA ¼ vagal nerve activity.

Shelton et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8

SCNA in Heart Failure M A Y 2 0 1 8 : 6 8 6 – 9 5

688

Nerve activity was also quantified by determining thenumber of spikes and bursts within the nerve activity(8). The filtered and rectified signals from a 24-hrecording were integrated over a 100-ms slidingwindow. Within this integrated signal, nerve spikesand bursts of nerve activity were detected. A spikewas defined as the continuous period in which theamplitude is higher than the threshold value. Eachspike’s duration and average amplitude weremeasured within that period. The threshold valuewas calculated as 3 times the baseline noise level,which was considered to be the lowest 10th percentileof all values during the 24-h recording. Spikes with aninter-spike interval of <5 s were grouped into bursts.

The aSGNA, aSCNA, aVNA, and heart ratewere calculated every hour of a 24-h period (2:00 PM

to 2:00 PM the next day) before pacing at baseline andof every hour of the 24-h period (2:00 PM to 2:00 PM

the next day) immediately after pacing was stopped.To avoid periods of dropped radiotransmission orartifactual noise caused by movement of the dog,which was greater than nerve activity, the first10 analyzable minutes of each hour were used forthe calculation of the averaged nerve activity andheart rate.

Paroxysmal atrial tachyarrhythmias were definedas episodes of tachycardia ($175 beats/min) withabrupt onset (>50 beats/min/s) and a duration of >5 s.

Of these episodes, one-half were randomly selectedfor P-wave analysis. This analysis involvedcomparing the P waves before the initiation of thetachycardia to the P waves during the tachycardiaepisode and analyzing the changes in P-wavemorphology and PR interval.

STATISTICAL ANALYSIS. The averaged nerve activ-ities were compared between the control dogs andcryoablated dogs using 1-sided Student’s t-tests bothbefore and after pacing. In addition, the aSCNA foreach dog was used to look at the effect of ventricularpacing on both control and cryoablated dogs by 1-sidedpaired t tests. The data were presented as mean � SD.To test for circadian variation, generalized additivemixed-effects models were fitted to the longitudinaldata. A Pearson correlation coefficient was used tocompare the SCNA and SGNA between the control andthe ablation groups for 24 h before and after pacing. A pvalue #0.05 was considered significant.

RESULTS

EFFECT OF STELLATE GANGLION CRYOABLATION

ON NERVE ACTIVITY. The group 1 dogs did not haveablation, whereas the group 2 dogs had stellate gan-glion ablation during the first surgery. Figure 1 showsthat, at baseline (before rapid ventricular pacing),aSGNA was significantly lower in the group 2 dogs

FIGURE 2 Effect of Cryoablation on Nerve Activity After Ventricular Pacing

Nerve activity was quantified by integrating the rectified, high-pass signal every minute for 10 min every hour for 24 h. Data after pacing are

shown as a box-and-whisker plot comparing the control group (group 1) versus dogs that had stellate ganglion cryoablation (group 2).

A significant decrease in nerve activity is seen for the VNA in the ablation group. Abbreviations as in Figure 1.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8 Shelton et al.M A Y 2 0 1 8 : 6 8 6 – 9 5 SCNA in Heart Failure

689

(0.56 � 0.13 mV) than group 1 dogs (1.14 � 0.11 mV)(p ¼ 0.003), indicating ablation significantly reducedaSGNA. aSCNA was also significantly lower in group2 versus group 1 dogs (0.25 � 0.028 mV vs. 0.34 �0.033 mV, p ¼ 0.03). The aVNA was lower in group 2(0.21�0.0068 mV) than in group 1 (0.48�0.066 mV, p¼0.001) dogs. The heart rate was also lower in group 2(88� 7.9 beats/min) than group 1 (109 � 4.0 beats/min,p ¼ 0.02). Consistent with a previous report (4), rapidventricular pacing induced HF increased the nerveactivities in both groups of dogs. Figure 2 shows thedifferences between group 1 and group 2 dogs after HFhad been induced. As shown in that figure, the group 2dogs had lower nerve activities than group 1 dogs, butthe differences did not reach statistical significance foraSCNA and aSGNA. The aSCNA was 0.28 � 0.024 mV ingroup 2 and 0.43 � 0.079 mV in group 1 (p ¼ 0.06). TheaSGNAwas 0.91�0.27 mV for group 2 and 1.23�0.27 mVfor group 1 (p ¼ 0.21). The VNA was 0.21 � 0.011 mV forgroup 2 and 0.55�0.082 mV for group 1 (p¼0.001). Theheart rate was 87 � 4.9 beats/min for group 2 and 96 �3.9 beats/min for group 1 (p ¼ 0.11). The nerve activitywas also quantified by detecting the spikes and burstsof activity within the nerve recording. Figure 3 showsthat the amplitude of the bursts was significantlydecreased after cryoablation for each type of nerverecording (SCNA, SGNA, and VNA). Although there wasno significant change in the burst frequency, becauseof the significant reduction in burst amplitude, totalnerve activity is reduced.

SCNA AND ATRIAL TACHYARRHYTHMIAS. Dogsnormally have PAT at baseline (9). Consistent withthat report, we were able to identify a total of 243 PATepisodes in 6 group 1 dogs during a 24-hour period ofmonitoring at baseline and 117 PAT episodes afterinduction of HF by rapid pacing. Our new analysesconfirmed the previous report that no PAT episodeswere present in group 2 dogs either at baseline orafter rapid ventricular pacing (6).

In all PAT episodes, an increase in the SCNA andSGNA correlated with the onset of the tachycardia(Figure 4). At the arrows labeled A and B, the SGNAand SCNA respectively begin to increase dramatically.The heart rate also increased suddenly, increasing ata rate of 87 beats/min/s (abrupt onset). The heart rateincreased to >175 beats/min for over 7 s, fulfilling thediagnostic criteria for PAT. In all episodes, as in thisexample, increased SCNA was closely related to in-creases in SGNA during PAT episodes.

P-wave changes were found in most randomlyselected episodes of arrhythmias, including P-wavemorphology changes and PR interval shortening, asshown in Figure 5. In the atrial tachyarrhythmia inFigure 4, the P-wave before the tachycardia startedwith a biphasic P-wave with an initial upward phasefollowed by a negative phase (Figure 5A), but after thetachycardia, the P-wave was initially negative with asmall positive phase, and thenwas evenmore negative(Figure 5B). Furthermore, the PR interval before thetachycardia was 125 ms and was shortened to 105 ms

FIGURE 3 Effect of Cryoablation on Nerve Activity After Ventricular Pacing

Nerve activity was quantified by detecting spikes within the integrated signal. Spikes with an inter-spike interval <5 s were grouped into bursts.

Data after pacing are shown as a box-and-whisker plot comparing the control group (group 1) versus dogs that had stellate ganglion cryoa-

blation (group 2). (A) A significant decrease in the amplitude of the burst activity is seen in the ablation group for each type of nerve recording.

(B) Results of calculating the frequency of the bursts. Cryoablation of the stellate ganglion did not have a significant effect on the frequency of

the bursts of nerve activity. Abbreviations as in Figure 1.

Shelton et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8

SCNA in Heart Failure M A Y 2 0 1 8 : 6 8 6 – 9 5

690

during the tachycardia. These findings indicate thatthe PAT episodes are unlikely to be sinus tachycardia.

CORRELATION BETWEEN SGNA AND SCNA. Thenerve activity of stellate ganglia and subcutaneousnerves for both groups 1 and 2 at baseline and afterpacing were compared by plotting SGNA by SCNA.Linear regression was then used to fit the data andcalculate the correlation coefficient. Figure 6 showsthe results for dog A, including the correlation coeffi-cient of r ¼ 0.61 and p < 0.0005. The correlation co-efficient and p value for each dog both before and afterpacing is shown in Table 1. In general, SCNA correlatedmoderately well with SGNA, with an average of allcorrelation coefficients of all dogs at 0.64 (95% confi-dence interval: 0.58 to 0.70). To compare the 2 groups,before pacing, the mean difference between the abla-tion group and the control group (ablation – control) is0.2567 (p ¼ 0.01). After pacing, the mean differencebetween the ablation group and the control group(ablation – control) is �0.035 (p ¼ NS).

CIRCADIAN VARIATION. Figure 7 summarizes theSCNA data for all dogs. The SCNA had a significantcircadian variation (p < 0.001) when fitted to ageneralized additive mixed-effects model. Figure 8shows the generalized additive mixed-effect modelfitted to the SCNA data. Previous studies have showna similar significant circadian variation for the SGNA(5). Compared with the SCNA in Figure 7A, they bothshow a progressive increase in the morning hours,with a secondary increase in the late afternoon.

DISCUSSION

USING SCNA TO MEASURE SYMPATHETIC TONE. HFis characterized by increased sympathetic tone andcardiac arrhythmias (4). The sympathetic nerve ac-tivity as measured by microneurography is reduced bycardiac resynchronization therapy in the responders,but not in nonresponders (10). Because micro-neurography techniques are difficult to perform and

FIGURE 4 Atrial Tachycardia Correlated With an Increase in SCNA and SGNA

The arrows indicate the increase in nerve activity for the SGNA (A) and the SCNA (B). ECG¼ electrocardiogram; other abbreviations as in Figure 1.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8 Shelton et al.M A Y 2 0 1 8 : 6 8 6 – 9 5 SCNA in Heart Failure

691

cannot be used in ambulatory patients however,sympathetic nerve activity measurements have notbeen clinically useful in assessing the effects of cardiacresynchronization therapy or other neuromodulation

FIGURE 5 Changes in P Waves During Atrial Tachycardia

The onset of tachycardia was preceded by the onset of SCNA (red arrow

atrial tachycardia. (a) P-wave before arrhythmia onset; (b) the P-wave a

activity correlating to an increase in heart rate. Abbreviations as in Figu

methods. The results of this study suggest that theSCNA might be a useful method to directly measuresympathetic tone in HF and determine the effects ofHF therapy. A second possible use of SCNA is to

). A morphology change in the P-wave is shown with the onset of an

fter arrhythmia onset. Also shown is an increase in SGNA and SCNA

res 1 and 4.

FIGURE 6 Correlation of SGNA and SCNA Over 24 h

Relationship between SGNA and the SCNA integrated nerve activity (mV) over a 24-h period

for dog A. Abbreviations as in Figure 1.

TABLE 1 Pearson Co

Ablation Groups for 2

Con

Dog Befor

A r 0.

p value <0.

B r 0.

p value <0.

C r 0.

p value <0.

D r 0.

p value <0.

E r 0.

p value <0.

F r 0.

p value <0.

SCNA ¼ subcutaneous nerv

Shelton et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8

SCNA in Heart Failure M A Y 2 0 1 8 : 6 8 6 – 9 5

692

determine if cardiac arrhythmias are triggered bysympathetic activation. A previous study showed thatSCNA was observed before a majority of episodes ofventricular tachycardia and ventricular fibrillation inan ambulatory canine model of sudden cardiac death(2). In the present study, we have shown that SCNApreceded the onset of all PAT episodes both at baselineand after the induction of HF. These findings suggestthat SCNA can be useful in determining whether PATwas caused by sympathetic activation.

rrelation Coefficient Between SCNA and SGNA for the Control and

4 h Before and After Pacing

trol Ablation

e Pacing After Pacing Dog Before Pacing After Pacing

61 0.74 G r 0.80 0.65

0005 <0.0005 p value <0.0005 <0.0005

57 0.62 H r 0.76 0.81

0005 <0.0005 p value <0.0005 <0.0005

35 0.83 I r 0.56 0.72

0005 <0.0005 p value <0.0005 <0.0005

21 0.53 J r 0.70 0.61

0005 <0.0005 p value <0.0005 <0.0005

56 0.72 K r 0.84 0.59

0005 <0.0005 p value <0.0005 <0.0005

58 0.74 L r 0.76 0.59

0005 <0.0005 p value <0.0005 <0.0005

e activity; SGNA ¼ stellate ganglia nerve activity.

SOURCES OF SCNA. Axonal tracer studies haveshown that a significant portion of the skin sympa-thetic nerves of the neck and upper thorax originatefrom the ipsilateral stellate ganglion (11,12). Severalother studies have demonstrated a strong relation-ship between sympathetic nerve structures in thechest and skin nerves. Donadio et al. have shown inskin biopsies an abundance of sympathetic nerves inarteriovenous anastomosis, arrector pilorum muscles,and arterioles (13). Baron et al. showed in axonaltracer studies that nearly all skin sympathetic nervesomata are located in the middle cervical and stellateganglia (11). The middle cervical ganglion and thestellate ganglion are highly interconnected tointrinsic cardiac neurons. In addition, it has beenshown that SCNA can be used as an estimate ofsympathetic tone (1) and is more accurate than heartrate variability (14). The current study confirms thatSGNA and SCNA are closely related because they bothincrease before the onset of paroxysmal atrial tachy-arrhythmias and both decrease after cryoablation.

EFFECTS OF LSG ABLATION ON SCNA. Previousstudies have shown that the right stellate ganglion(RSG) controlled heart rate and only had an effect onblood pressure after the left stellate ganglion wascompletely resected. This suggests that the RSG in-nervates the sinus node and plays a role in regulatingthe heart rate. The LSG controls blood pressure and isthought to be the main influence in cardiac sympa-thetic control (15). In a separate canine study, it wasdemonstrated that LSG stimulation significantlyincreased atrial fibrillation inducibility over RSGstimulation. Removing the LSG significantly reducedthe incidence of atrial fibrillation (16); however, in aclinical study, both RSG and LSG block showed similarresults in reduced atrial fibrillation inducibility andduration (17). In addition to controlling atrial arrhyth-mias in caninemodels (18), LSG ablation is known to beeffective in decreasing sympathetic outflow and con-trolling ventricular arrhythmias (19–22). In animal ex-periments, it has also been shown that partial LSGablation can reduce or eliminate PAT in ambulatorycanine models (5,18). Left cardiac sympathetic dener-vation (LCSD) has been used as a therapeutic option forpatients with catecholaminergic polymorphic ven-tricular tachycardia that have recurrent ventricularfibrillation/ventricular tachycardia episodes evenwithb-blocker therapy. Surgical approaches include thor-acoscopic, transaxillary, and supraclavicular to exposethe left-sided sympathetic chain from T4 to T1.Resection or ablation of the sympathetic ganglia fromT1 to T2 then has shown to have a positive effect inthese patients. A more minimally invasive procedure

FIGURE 7 Circadian Variation in SCNA

Summary data of the averaged integrated nerve activity (mV) for every hour over a 24-h period for the (A) control dogs and (B) dogs with

cryoablation. Abbreviation as in Figure 1.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8 Shelton et al.M A Y 2 0 1 8 : 6 8 6 – 9 5 SCNA in Heart Failure

693

using video-assisted thoracic surgery to perform LCSDhas also proven to be effective in reducing cardiacevents. Even though we have previously shown thatLCSD therapy with cryoablation can be used as a ther-apy for atrial arrhythmias, it is still not a commonlyused approach as a clinical therapeutic option for atrialarrhythmias.

If SGNA is a source of SCNA, then partial LSGablation should reduce SCNA. The latter findingswould further support the validity of SCNA in esti-mating SGNA. As shown in Figure 1, however, the LSGablation seems to have more effects on aSGNA thanaSCNA. One reason is that the subcutaneous nerves inthe thorax may not all come from the LSG; therefore,the baseline correlation between aSGNA and aSCNAwas only moderate. A second reason is that LSG is alarge nerve structure that generates strong electricalsignals. Because of the high signal-to-noise ratio, theamplitude of reduction of SGNA after cryoablation iseasier to visualize. In contrast, the subcutaneousnerves are small (average <1 mV at baseline) and havea smaller signal-to-noise ratio. Given that inherentnoise of the recording system is a significant portionof the signal, a reduction of SCNA may not be as

apparent after LSG ablation. Monitoring SCNA hasmany advantages over monitoring SGNA as an indi-cator of sympathetic nerve activity. Monitoring SGNAor other indicators of sympathetic nerve activity haveinvolved an open chest procedure to implant elec-trodes on the stellate ganglion or other nerve struc-tures. On the other hand, implantation ofsubcutaneous electrodes in the chest could beaccomplished much less invasively. In addition, thestudy of subcutaneous nerve activity may lead to theuse of surface electrodes for the monitoring of sym-pathetic nerve activity. Preliminary results haveshown that it is possible to record the skin sympa-thetic nerve activity using ordinary ECG patch elec-trodes (3,23). For these reasons, SCNA may be a usefultool in minimally invasively monitoring sympatheticnerve activity involved in the arrhythmogenesis ofatrial tachyarrhythmias and in estimating the efficacyof neuromodulation procedures.

STUDY LIMITATIONS. The electrical signals withinthe recordings from subcutaneous electrodes maycome from multiple sources, including low-frequencymotion artifacts, ECG, respiratory muscle activity,

FIGURE 8 Generalized additive Mixed-Effect Model Fitted to the SCNA Data

Circadian variation of the SCNA inmV for the (A) control (group 1) dogs and (B) the dogswith cryoablation (group2).Solid lines representmean integrated

SCNA calculated over 1 min for 10 min every hour. Dotted lines represent 95% upper and lower confidence intervals. Abbreviation as in Figure 1.

Shelton et al. J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8

SCNA in Heart Failure M A Y 2 0 1 8 : 6 8 6 – 9 5

694

and nerve activities. Although some noise can beeffectively filtered with conventional filters, the ECGcan contain high-frequency content that is withinthe range of the frequency bandwidth of the nerveactivity; therefore, we decided to use a wavelet filterto filter out much of the ECG as we have done inprevious studies with large ECG signals accompa-nying the nerve activity of interest (7). Although in-tegrated SCNA showed a significant correlation withintegrated SGNA, integrated wavelet SCNA appearedto have a greater correlation with integrated SGNA.The dogs do not have sweat glands in the thoraxwhere the recordings were made. Whether sweatingaffects the SCNA recording cannot be determined inthis study.

In addition, our study used widely spaced elec-trodes to monitor subcutaneous nerve activity. Otherstudies in our laboratory have used more narrowlyspaced electrodes with varying success (1,3). Whenusing the DSI D70-EEE radiotransmitter withnarrowly spaced electrodes, the bandwidth andsampling rate of the device makes the signal-to-noiseratio insufficient to accurately monitor subcutaneousnerve activity; however, when using narrowly spaced

electrodes with recording equipment capable ofrecording with higher bandwidth and higher sam-pling rate of 10,000 Hz, we were able to high-passfilter at 500 Hz, improving the signal-to-noise ratioand increasing the accuracy of monitoring the sub-cutaneous nerve activity (1).

CONCLUSIONS

Subcutaneous nerve activity correlated with SGNAand the onset of arrhythmias. Cryoablation of thestellate ganglia reduced the SCNA, further validatingthe use of SCNA to estimate cardiac sympathetic tone.This study suggests that using equipment with anappropriate bandwidth and sampling rate, recordingsubcutaneous nerve activity with an implantable de-vice may be possible and the SCNA can be used as asurrogate for SGNA.

ADDRESS FOR CORRESPONDENCE: Dr. Thomas H.Everett IV, Krannert Institute of Cardiology, IndianaUniversity School of Medicine, 1800 North CapitolAvenue, Indianapolis, Indiana 46202. E-mail:theveret@iu.edu.

PERSPECTIVES

COMPETENCY IN MEDICAL KNOWLEDGE: Micro-

neurography is used to directly measure sympathetic

nerve activity; however, microneurography techniques

are difficult to perform and cannot be used in ambulatory

patients. The results of this study suggest that the SCNA

might be a useful method to directly measure sympa-

thetic tone in HF, to determine the effects of HF therapy,

and to determine if cardiac arrhythmias are triggered by

sympathetic activation.

TRANSLATIONAL OUTLOOK: Neuromodulation ther-

apy including the ablation of the ganglionated plexi has

been used for controlling atrial fibrillation. Techniques

that can be used as a surrogate for microneurography can

increase the clinical usefulness of sympathetic nerve ac-

tivity measurements to help in assessing the efficacy of

neuromodulation.

J A C C : C L I N I C A L E L E C T R O P H Y S I O L O G Y V O L . 4 , N O . 5 , 2 0 1 8 Shelton et al.M A Y 2 0 1 8 : 6 8 6 – 9 5 SCNA in Heart Failure

695

RE F E RENCE S

1. Robinson EA, Rhee KS, Doytchinova A, et al.Estimating sympathetic tone by recording subcu-taneous nerve activity in ambulatory dogs.J Cardiovasc Electrophysiol 2015;26:70–8.

2. Doytchinova A, Patel J, Zhou S, et al. Subcu-taneous nerve activity and spontaneous ventricu-lar arrhythmias in ambulatory dogs. Heart Rhythm2015;122:612–20.

3. Jiang Z, Zhao Y, Doytchinova A, et al. Using skinsympathetic nerve activity to estimate stellateganglion nerve activity in dogs. Heart Rhythm2015;12:1324–32.

4. Ogawa M, Zhou S, Tan AY, et al. Left stellateganglion and vagal nerve activity and cardiac ar-rhythmias in ambulatory dogs with pacing-inducedcongestive heart failure. J Am Coll Cardiol 2007;50:335–43.

5. Ogawa M, Tan AY, Song J, et al. Cryoablation ofstellate ganglia and atrial arrhythmia in ambula-tory dogs with pacing-induced heart failure. HeartRhythm 2009;6:1772–9.

6. Ogawa M, Tan AY, Song J, et al. Cryoablationof extrinsic cardiac sympathetic nerves mark-edly reduces atrial arrhythmias in ambulatorydogs with pacing-induced heart failure. HeartRhythm 2008;5:S54.

7. Park HW, Shen MJ, Han S, et al. Neural controlof ventricular rate in ambulatory dogs withpacing-induced sustained atrial fibrillation. CircArrhythm Electrophysiol 2012;5:571–80.

8. Hart EC, Head GA, Carter JR, et al. Recordingsympathetic nerve activity in conscious humansand other mammals: guidelines and the road tostandardization. Am J Physiol Heart Circ Physiol2017;312:H1031–51.

9. Choi EK, Shen MJ, Han S, et al. Intrinsic cardiacnerve activity and paroxysmal atrial

tachyarrhythmia in ambulatory dogs. Circulation2010;121:2615–23.

10. Najem B, Unger P, Preumont N, et al.Sympathetic control after cardiac resynchroni-zation therapy: responders versus non-responders. Am J Physiol Heart Circ Physiol2006;291:H2647–52.

11. Baron R, Janig W, With H. Sympathetic andafferent neurones projecting into forelimb andtrunk nerves and the anatomical organization ofthe thoracic sympathetic outflow of the rat.J Auton Nerv Syst 1995;53:205–14.

12. Taniguchi T, Morimoto M, Taniguchi Y,Takasaki M, Totoki T. Cutaneous distributionof sympathetic postganglionic fibers fromstellate ganglion: a retrograde axonal tracingstudy using wheat germ agglutinin conjugatedwith horseradish peroxidase. J Anesth 1994;8:441–9.

13. Donadio V, Nolano M, Provitera V, et al. Skinsympathetic adrenergic innervation: an immuno-fluorescence confocal study. Ann Neurol 2006;59:376–81.

14. Chan YH, Tsai WC, Shen C, et al. Subcutaneousnerve activity is more accurate than heart ratevariability in estimating cardiac sympathetic tonein ambulatory dogs with myocardial infarction.Heart Rhythm 2015;12:1619–27.

15. Wu G, DeSimone CV, Suddendorf SH, et al.Effects of stepwise denervation of the stellateganglion: Novel insights from an acute caninestudy. Heart Rhythm 2016;13:1395–401.

16. Zhou Q, Hu J, Guo Y, et al. Effect of the stel-late ganglion on atrial fibrillation and atrial elec-trophysiological properties and its left-rightasymmetry in a canine model. Exp and Clin Cardiol2013;18:38–42.

17. Leftheriotis D, Flevari P, Kossyvakis C, et al.Acute effects of unilateral, temporary stellateganglion block on human atrial electrophysiologicproperties and atrial fibrillation inducibility. HeartRhythm 2016;13:2111–7.

18. Tan AY, Zhou S, Ogawa M, et al. Neuralmechanisms of paroxysmal atrial fibrillation andparoxysmal atrial tachycardia in ambulatory ca-nines. Circulation 2008;118:916–25.

19. Moss AJ, McDonald J. Unilateral cervico-thoracic sympathetic ganglionectomy for thetreatment of long QT interval syndrome. N Engl JMed 1971;285:903–4.

20. Schwartz PJ, Priori SG, Cerrone M, et al.Left cardiac sympathetic denervation in themanagement of high-risk patients affected bythe long-QT syndrome. Circulation 2004;109:1826–33.

21. Vaseghi M, Rn Msn Np JG, Kanaan C, et al.Cardiac sympathetic denervation in patients withrefractory ventricular arrhythmias or electricalstorm: intermediate and long-term follow-up.Heart Rhythm 2014;11:360–6.

22. Buckley U, Yamakawa K, Takamiya T, AndrewArmour J, Shivkumar K, Ardell JL. Targeted stel-late decentralization: implications for sympatheticcontrol of ventricular electrophysiology. HeartRhythm 2015;13:282–8.

23. Doytchinova A, Hassel J, Yuan Y, et al. Usingskin sympathetic nerve activity to estimate sym-pathetic tone in humans (abstr). Heart Rhythm2015;12:S297.

KEY WORDS arrhythmias, autonomicnervous system, heart failure, stellateganglion, vagus nerve