Spinal Anesthesia Reduces Myocardial Ischemia–triggered Ventricular Arrhythmias by Suppressing Spinal Cord Neuronal Network Interactions in Pigs

Nine Yorkshire pigs (38.1 ± 2.5 kg, seven males, two females) were used in this study. All animal studies followed the recommendations of the Institutional Animal Care and Use Committee (University of California, Los Angeles, California) and the National Institutes of Health (Bethesda, Maryland) Guide for the Care and Use of Laboratory Animals (NIH Pub. No. 85-23, Revised 1996). All animal experiments were approved by the University of California, Los Angeles, Chancellor’s Animal Research Committee and performed in our laboratories at the University of California, Los Angeles. Animals were euthanized by inducing terminal ventricular arrhythmia under general anesthesia.

Animals were sedated with intramuscular telazol (4 to 6 mg/kg), intubated, and mechanically ventilated. General anesthesia was maintained with inhaled isoflurane (1.5 to 2.5%) and intravenous boluses of fentanyl (total 20 μg/kg) for analgesia during surgical preparation.^7,11,29  Continuous intravenous saline was infused throughout the procedure. During the surgical preparation, a partial T1-T5 laminectomy was performed, followed by the sternotomy. T2 paravertebral ganglia and the heart were exposed after the sternotomy. T2 electrodes were impaled into the ganglia, and the epicardial electrodes were placed around the heart. Laminectomy was completed in the lateral position, and a small part of the dura was cut to advance the intrathecal and place the neural recording probe. After surgical preparation was completed, inhaled isoflurane was changed to intravenous α-chloralose (50 mg/kg initial bolus followed by a 20 mg/kg per hour continuous infusion). Myocardial ischemia and cardiac load–dependent stressors were performed before and after bupivacaine injection. Electrocardiogram (ECG) data (12-lead) were continuously recorded via a Prucka CardioLab system (GE Healthcare, USA). Left ventricular systolic pressure and maximum and minimum rate change of left ventricular pressure were also measured using a 12-pole conductance, high-fidelity pressure monitoring pigtail catheter (5-French) inserted into the left ventricle via the left carotid artery and connected to an MPVS Ultra Pressure-Volume Loop System (Millar Instruments, USA). Arterial blood gas was tested every hour and corrected as necessary by adjusting the ventilation.

The activity generated by neural soma in the T2 level of the spinal cord was directly recorded in situ. The T2 level of the spinal cord was accessed after laminectomy and the opening of a small window in the spinal cord dura. A high-density two-dimensional penetrating microarray (64 electrode recording sites; NeuroNexus, USA), mounted on a micromanipulator, was inserted 2 mm into the spinal cord at the T2 level, lateral to the midline (fig. 1A). This position allowed the recording sites to capture both dorsal horn and intermediolateral neurons. Neural signals were recorded directly from the thoracic spinal cord, amplified, and digitized (SmartBox acquisition system, NeuroNexus, USA). Using the high impedance headstage and the filter setting, the two-dimensional microarrays were only able to record the extracellular action potential from the adjacent soma and not axons of passage.^30,31  ECG and left ventricular pressure were recorded in the same file as the neural data.

The cardiac electrophysiologic parameters in this porcine model have been extensively studied and well-characterized by our laboratory, thus providing us with a promising model for studying myocardial sympathoexcitation and arrhythmogenesis.^7,11,29  In order to acquire unipolar ventricular epicardial electrograms, a high-fidelity 56-electrode sock array was placed around the heart (fig. 1B), and electrograms were recorded by the Prucka CardioLab system (GE Healthcare). Whole heart activation recovery interval, a well-validated measure for action potential duration, was analyzed as previously described.^29,32  A few supraventricular tachycardia episodes occurred during the experiments and were removed from the analysis. Activation recovery interval measurement was in accordance with guidelines described by Haws and Lux.³³  Shortening of activation recovery interval is seen with an increase in myocardial sympathetic activity. To ensure accuracy, each electrogram with ST-segment changes was measured by software and then checked manually across five beats. The variance of all activation recovery intervals in the whole heart was used to calculate the epicardial dispersion of repolarization. Epicardial dispersion is correlated to the heterogeneity of repolarization time and is a measure of the increased risk for ventricular arrhythmias.²⁹  Acute ischemia is associated with shortened activation recovery interval duration and increased dispersion of repolarization. We recorded whole heart activation recovery interval and epicardial dispersion of repolarization to measure the electrophysiologic effects elicited by cardiac stressors and the effect of intrathecal bupivacaine on these electrophysiologic parameters. The customized software iScalDyn (University of Utah, USA) was used to analyze activation recovery interval and epicardial dispersion data.^11,29,31,34  Manual classification of premature ventricular contractions during acute ischemia was performed using a continuous 12-lead ECG. Abnormal heartbeats containing an irregularity in QRS complex morphology, a decrease in R wave–to–R wave time interval, and an absence of a P wave were classified as premature ventricular contractions. There were no episodes of sustained ventricular tachycardia or fibrillation.

Animals underwent median sternotomy to expose the anterior surface of the heart. Umbilical tapes were placed around the base of the inferior vena cava and descending thoracic aorta to occlude the inferior vena cava and the descending aorta, respectively. To create acute myocardial ischemia, a 4-0 Prolene suture (Ethicon Inc., USA) was placed around the left anterior descending (LAD) coronary artery near the third diagonal branch. The suture was led through a short polyethylene tubing segment for occlusion of the LAD. Inferior vena cava (1 min), descending aorta (1 min), and the LAD coronary artery (5 min) were occluded before (control) and after thoracic intrathecal bupivacaine (fig. 2A). These cardiac stressors were applied to functionally delineate spinal cord neuronal subpopulations responsive to acute myocardial ischemia and/or cardiac load–dependent perturbations.¹⁰  There was a waiting period of 15 min between cardiac load–dependent stressors and 30 min after acute ischemia to allow hemodynamic and electrophysiologic indices to return to baseline.

Local anesthesia with 0.5% isobaric bupivacaine bolus of 1 ml and continuous infusion of 0.1% bupivacaine 1 ml/h was administered via an intrathecal catheter at the T2 level. The catheter position was confirmed during and at the end of the protocol. Previous studies have shown that the onset, spread, and duration of single-injection isobaric bupivacaine spinal anesthesia, used at various concentrations and volumes, is primarily related to dosage.^35,36  Based on these reports, we used 5 mg of intrathecal bupivacaine, waited 10 min to begin our protocol to allow for the onset of bupivacaine, and finished all cardiac stressor procedures within 60 min of bolus thoracic intrathecal bupivacaine injection.

Neuronal activity was identified as action potentials with a signal to noise ratio greater than 2:1.³¹  The activity generated by individual neurons was identified using available spike sorting tools (principal component analysis and cluster on measurements techniques) in the Spike2 software program (Cambridge Electronics Design, United Kingdom; fig. 2B). Simultaneously occurring activities displaying a similar appearance across all electrode channels were interpreted as being artifactual. Artifacts arising from exogenous stimuli and electrical stimuli were removed from the analysis. After artifact removal, the activity created by individual neuron somata from each of the 64 electrode channels was characterized by their specific amplitudes and waveforms.³⁰  Using these methods, action potentials produced by individual soma, rather than axons of passage, can be recorded for prolonged periods of time.³¹ 

To investigate the intraspinal neuronal mechanisms contributing to the cardioprotective effects of bupivacaine, we first defined the response characteristics of T2 spinal dorsal horn and intermediolateral neurons, which were functionally delineated into subpopulations responsive to acute ischemia and cardiac load–dependent perturbations. Statistically significant increases or decreases in activity for each neuron between a 1-min baseline period before each intervention and the neural activity during the intervention were analyzed.^25,26,30,31  Neurons with significant changes in activity were characterized as cardiac-related neurons.¹⁰  We then evaluated the effect of bupivacaine on the modulation of neural activity and interactions in the spinal cord to such destabilizing inputs. Cardiac electrical responses were also evaluated.

Mathematical methods characterizing single neuron activities are typically based on the measurement of spike times or interspike intervals in spike trains. However, spike trains are discrete rather than continuous functions. Therefore, kernel-estimation techniques are often used. A well-validated method is convolving the spike train with a continuous function as the kernel to achieve a smooth and continuous signal, termed spike-density function.^37,38  In this study, we utilized the Gaussian kernel to calculate the spike density function. Correlation analysis was used to evaluate the potential correlation of neural firing between dorsal horn and intermediolateral neurons. The occurrence of a spike at one timepoint is not independent of the occurrence of spikes at other times, both within spike trains from a single neuron and across spike trains from multiple neurons. The presence of these correlations has led to the proposal that such correlations might form a key element of the neural code. In this study, Pearson’s linear correlation coefficient was used to measure the similarity between the spike density functions of a pair of neurons. Correlated neurons were defined as a pair of neurons with a correlation coefficient of higher than 0.7.³⁹ 

Sympathetic efferent preganglionic neurons related to cardiac control originate in the intermediolateral cell column of the spinal cord and project their axons into the paravertebral chain.^21,22  Antidromic stimulation of the T2 paravertebral chain was used to identify the sympathetic preganglionic neurons located in the intermediolateral nucleus of the thoracic spinal cord.¹⁸  Left T2 paravertebral chain stimulation was performed using bipolar needle electrodes. Stimulation current threshold was defined as the stimulation current amplitude, which elicited a 10% increase in systolic blood pressure at 4 Hz and 4-ms pulse durations. Stimulation pulses were delivered at 1-ms durations and 1 Hz at stimulation current threshold (mean ± SD, 7.0 ± 1.0 mA) using a Grass S88 Stimulator (Grass Co., USA). Intermediolateral neurons were identified as those that had more than 60% one-to-one firing within 50 ms after T2 paravertebral chain stimulation pulses. Intermediolateral neural activity was recorded on electrode sites that were clustered together in one region on the electrode that entered the intermediolateral region. Dorsal horn neurons were classified, putatively, as all neurons outside the identified intermediolateral region.

Statistical power calculation was not conducted before the study. The sample size was selected based on our previous published work in which similar extracellular neural recordings were performed and neural network interactions in response to different cardiac stressors were compared.^7,30,40  To determine statistically significant neural activity changes and, thus, functional identity for each recorded neuron during bupivacaine application and cardiac perturbations, a statistical test derived from the Skellam distribution⁴¹  was utilized as used previously in other peripheral ganglia.^30,31  The null hypothesis of this analysis is that the two firing rates are equal, and the assumption is that the number of action potentials identified follows a Poisson distribution. Based on this assumption, the difference in the activities follows a Skellam distribution,⁴²  and the probability that the difference in the number of firings is larger than the observed value provides the desired P value (unilateral test) using the Skellam cumulative distribution function. This analysis compares 1 min of neural activity before the cardiac perturbation (baseline) to neural activity during the perturbation. A two-way chi-square test was used to compare the percentage of neurons that responded to each intervention and the number of correlated dorsal horn–intermediolateral nucleus neurons before and after bupivacaine administration. The Shapiro–Wilk normality test was used to test if the data were normally distributed. Activation recovery intervals, dispersion in activation recovery intervals, and hemodynamics measures passed the Shapiro–Wilk normality test, and therefore, a paired t test was used to compare these measures. A Wilcoxon matched-pairs signed-rank test was used to compare the firing responses, correlation coefficients, and number of premature ventricular contractions that did not pass the Shapiro–Wilk normality test. Two-tailed testing was used for all statistical tests, and a P value of 0.05 was considered statistically significant. P values were corrected using the Bonferroni correction method due to multiple comparisons for correlation data analysis. Statistical analyses were performed using Prism (GraphPad Software, USA). Data are reported as median (25th, 75th percentiles) or mean ± SD.

To evaluate the change in hemodynamics associated with intrathecal administration of bupivacaine, hemodynamic parameters were recorded at baseline and 10 min after bupivacaine injection into the intrathecal space. Bupivacaine injection led to a decrease in left ventricular systolic pressure (87 ± 19 to 75 ± 18 mmHg; P = 0.015; table 1). Heart rate (79 ± 16 to 80 ± 12 bpm), maximum rate change of left ventricular pressure (1,826 ± 500 to 1,669 ± 544 mmHg/s), and minimum rate change of left ventricular pressure (–1,900 ± 667 to –1,543 ± 473 mmHg/s) were not significantly changed from baseline to 10 min after thoracic intrathecal bupivacaine placement (table 1). Hemodynamic responses to the cardiac stimuli were measured by calculating the percentage difference in the hemodynamic parameters before the onset of a cardiac stimulation to during the cardiac intervention. Heart rate, left ventricular systolic pressure, maximum rate change of left ventricular pressure, and minimum rate change of left ventricular pressure responses to LAD coronary artery, aortic, and inferior vena cava occlusions did not change before and after bupivacaine administration (table 2).

Whole heart electrophysiologic measures were recorded throughout the entire protocol. Seven of nine animals had full data sets with good signal quality, and two animals were removed due to artifactual/noisy electrophysiologic signal. The global ventricular mean activation recovery interval shortened during acute ischemia at control and bupivacaine application (control: 448 ± 113 ms to 396 ± 93 ms, P = 0.012; bupivacaine: 426 ± 86 ms to 393 ± 85 ms, P = 0.028; fig. 3B). However, after thoracic spinal anesthesia, the magnitude of global ventricular activation recovery interval shortening during acute ischemia was attenuated (–11 ± 7 vs. −8 ± 6%; P = 0.032; fig. 3C). Dispersion of repolarization also increased during acute ischemia both at control and bupivacaine application (control: 582 ± 229 to 3,743 ± 1945 ms², P = 0.015; bupivacaine: 540 ± 169 to 2,582 ± 1,008 ms², P = 0.008; fig. 3D). However, the magnitude of increase in dispersion of repolarization during acute ischemia was suppressed after thoracic intrathecal anesthesia (677 ± 560% at control vs. 468 ± 391% at bupivacaine; P = 0.030; fig. 3E). Ventricular arrhythmias were measured in episodes of premature ventricular contractions. Acute ischemia was associated with fewer premature ventricular contractions in the bupivacaine state (2.5 [1.75–5.25] premature ventricular contractions during ischemia before bupivacaine vs. 1 [0–1] premature ventricular contractions during ischemia after bupivacaine, P = 0.031, fig. 3F), indicating a functional, antiarrhythmogenic benefit of thoracic spinal anesthesia.

Neural recordings were obtained in all nine animals, and all animals were used for neural analysis. Out of 1,184 spinal cord neurons identified, 824 cardiac-related neurons (dorsal horn: 791, intermediolateral: 33) were classified based on their statistically significant response to cardiac stimuli. Twenty-nine (3.5%) cardiac-related neurons responded to ischemia only, 57.5% responded to cardiac load–dependent stimuli only, and 39.0% responded to both ischemia and cardiac load–dependent stimuli. All further analysis has been performed on the 824 cardiac-related neurons.

Bupivacaine administration alone did not change the activity of dorsal horn and intermediolateral neurons before the cardiac interventions (dorsal horn: from 0.00 [0.00–0.02] Hz at baseline to 0.00 [0.00–0.02] Hz during bupivacaine; intermediolateral: from 0.00 [0.00–0.02] Hz at baseline to 0.00 [0.00–0.02] Hz during bupivacaine).

The number of spinal cord neurons that responded to the ischemia was decreased by bupivacaine application (dorsal horn: from 43% to 18%, P < 0.0001; intermediolateral: from 39% to 9%, P = 0.004; fig. 4, A and B).

The absolute value of the neuronal response to the ischemia was changed by bupivacaine. For dorsal horn neurons, bupivacaine suppressed the neuronal response to ischemia from 0.03 (0.00–0.13) Hz (95% CI, 0.02–0.04 Hz) before bupivacaine to 0.01 (0.00–0.02) Hz (95% CI, 0.00–0.01 Hz) after bupivacaine (P < 0.0001; fig. 4C). Spinal bupivacaine application also decreased the intermediolateral neurons’ response to ischemia from 0.03 (0.01–0.14) Hz before bupivacaine to 0.01 (0.00–0.02) Hz after bupivacaine (P = 0.001; fig. 4D). Intermediolateral neurons’ response to reperfusion did not change after bupivacaine application (from 0.02 [0.00–0.14] Hz [95% CI, 0.01–0.14] Hz before bupivacaine to 0.02 [0.00–0.07] Hz [95% CI, 0.00–0.02] after bupivacaine; P = 0.357; fig. 4D).

The temporal response of dorsal horn and intermediolateral neurons to myocardial ischemia was also evaluated in the absence of bupivacaine and after bupivacaine administration (fig. 5). Dorsal horn and intermediolateral neurons show periodic spikes in activity with a correlated temporal response between dorsal horn and intermediolateral neurons during myocardial ischemia (fig. 5, A and B). This correlation in the temporal response of dorsal horn and intermediolateral neurons to myocardial ischemia was abolished after bupivacaine administration (fig. 5, C and D). This observation led us to investigate the neuronal activity correlation between the dorsal horn and intermediolateral neurons and within the dorsal horn neurons during the ischemia before and after spinal anesthesia.

The number of correlated pairs of dorsal horn–dorsal horn and dorsal horn–intermediolateral nucleus neurons increased from baseline to during ischemia (dorsal horn–dorsal horn: 324 correlated pairs of neurons to 931 correlated pairs of neurons out of 1,189 pairs, P < 0.0001; dorsal horn–intermediolateral nucleus: 11 correlated pairs of neurons to 69 correlated pairs of neurons out of 1,135 pairs, P < 0.0001). This network response to ischemia was blunted after bupivacaine administration (dorsal horn–dorsal horn: 931 correlated pairs of neurons vs. 38 correlated pairs of neurons out of 1,189 pairs, P < 0.0001; dorsal horn–intermediolateral nucleus: 69 correlated pairs of neurons vs. 1 correlated pair of neurons out of 1,135 pairs, P < 0.0001; fig. 6, A and C). The magnitude of correlation of the correlated neurons in the control state also increased (dorsal horn–dorsal horn: from 0.19 [0.07–0.73] [95% CI, 0.16–0.20] before ischemia to 0.81 [0.71–0.92] [95% CI, 0.79–0.82] during ischemia, P < 0.0001; dorsal horn–intermediolateral nucleus: from 0.14 [0.06–0.41] [95% CI, 0.09–0.25] before ischemia to 0.80 [0.75–0.88] [95% CI, 0.77–0.83] during ischemia, P < 0.0001). This myocardial ischemia–induced augmentation in dorsal horn–dorsal horn and dorsal horn–intermediolateral nucleus correlation was suppressed after bupivacaine (dorsal horn–dorsal horn: 0.81 [0.71–0.92] [95% CI, 0.79–0.82] before bupivacaine to 0.08 [0.03–0.18] [95% CI, 0.07–0.08] after bupivacaine, P < 0.0001; dorsal horn–intermediolateral nucleus: 0.80 [0.75–0.88] [95% CI, 0.77–0.83] before bupivacaine to 0.06 [0.03–0.13] [95% CI, 0.03–0.08] after bupivacaine, P < 0.0001; fig. 6, B and D).

Similar to the responses seen during acute myocardial ischemia, spinal bupivacaine reduced the activity of cardio-spinal neurons in the thoracic spinal cord.

Bupivacaine administration reduced the number of spinal cord neurons that responded to the aortic occlusion (dorsal horn: from 70% to 34%, P < 0.0001; intermediolateral: from 70% to 24%, P < 0.001) and inferior vena cava occlusion (dorsal horn: from 63% to 53%, P < 0.0001; intermediolateral: from 39% to 9%, P = 0.014).

Spinal bupivacaine suppressed the dorsal horn and intermediolateral neuronal response to aortic occlusion (dorsal horn: from 0.15 [0.02–0.42] Hz before bupivacaine to 0.02 [0.00–0.05] after bupivacaine, P < 0.0001; intermediolateral: from 0.23 [0.02–0.97] Hz before bupivacaine to 0.02 [0.00–0.07] after bupivacaine, P = 0.003) and release of the aortic occlusions (dorsal horn: from 0.03 [0.00–0.15] Hz before bupivacaine to 0.00 [0.00–0.02] after bupivacaine, P < 0.0001; intermediolateral: from 0.02 [0.00–0.38] Hz before bupivacaine to 0.00 [0.00–0.02] after bupivacaine, P = 0.060). Spinal cord neuronal response to inferior vena cava occlusion was also mitigated by thoracic spinal bupivacaine application (dorsal horn: from 0.13 [0.03–0.42] Hz before bupivacaine to 0.05 [0.00–0.17] after bupivacaine, P < 0.0001; intermediolateral: from 0.44 [0.13–1.40] Hz before bupivacaine to 0.04 [0.01–0.23] after bupivacaine, P < 0.0001). The evoked spinal cord neuronal response to release of inferior vena cava occlusion was diminished after application of thoracic intrathecal anesthesia (dorsal horn: from 0.06 [0.02–0.22] Hz before bupivacaine to 0.03 [0.00–0.17] after bupivacaine, P = 0.016; intermediolateral: from 0.32 [0.10–2.40] Hz before bupivacaine to 0.06 [0.03–0.36] after bupivacaine, P = 0.017).

Dorsal horn–dorsal horn and dorsal horn–intermediolateral nucleus interactions increased during aortic and inferior vena cava occlusions. The number of correlated dorsal horn–dorsal horn neurons increased from baseline to during aortic and inferior vena cava occlusions (aortic: 192 correlated pairs of neurons to 957 correlated pairs of neurons out of 1,111 pairs, P < 0.0001; inferior vena cava: 69 correlated pairs of neurons to 304 correlated pairs of neurons out of 368 pairs, P < 0.0001). This increase in the number of correlated afferent neurons during great vessels occlusion was suppressed by bupivacaine administration (aortic: 957 correlated pairs of neurons vs. 81 correlated pair of neurons out of 1,111 pairs, P < 0.0001; inferior vena cava: 304 correlated pairs of neurons vs. 39 correlated pair of neurons out of 368 pairs, P < 0.0001). The number of correlated dorsal horn–intermediolateral nucleus neurons also increased from baseline to during great vessels’ occlusions (aortic: 0 correlated pairs of neurons to 99 correlated pairs of neurons out of 844 pairs, P < 0.0001; inferior vena cava: 9 correlated pairs of neurons to 28 correlated pairs of neurons out of 1041 pairs, P = 0.000). The number of dorsal horn–intermediolateral nucleus interactions during the cardiac load dependent stressor was suppressed after bupivacaine administration (aortic: 99 correlated pairs of neurons vs. 12 correlated pairs of neurons out of 844 pairs, P < 0.0001; inferior vena cava: 28 correlated pairs of neurons vs. 5 correlated pairs of neurons out of 1,041 pairs, P < 0.0001).

In the control state, aortic and inferior vena cava occlusions augmented the interactions between dorsal horn neurons and the magnitude of the correlation of neural firing between dorsal horn–dorsal horn pairs of neurons was increased during great vessels’ occlusions (aortic: from 0.13 [0.06–0.48] at baseline to 0.77 [0.72–0.84] during aortic occlusion, P < 0.0001; inferior vena cava: from 0.15 [0.07–0.31] at baseline to 0.77 [0.72–0.84] during inferior vena cava occlusion, P < 0.0001). This increase in the correlation magnitude of neural firing between dorsal horndorsal horn pairs neurons was suppressed by bupivacaine (during aortic: from 0.77 [0.72–0.84] before bupivacaine to 0.11 [0.06–0.27] after bupivacaine, P < 0.0001; during inferior vena cava: from 0.77 [0.72–0.84] before bupivacaine to 0.19 [0.09–0.39] after bupivacaine, P < 0.0001). The magnitude of correlation between dorsal horn and intermediolateral neuronal activity also increased from baseline to during aortic and inferior vena cava occlusions (aortic: from 0.13 [0.08–0.33] at baseline to 0.82 [0.76–0.86] during aortic occlusion, P < 0.0001; inferior vena cava: from 0.14 [0.08–0.58] at baseline to 0.75 [0.57–0.81] during inferior vena cava occlusion, P = 0.002). Thoracic spinal anesthesia reduced the magnitude of correlation of the dorsal hornintermediolateral nucleus correlated neurons in response to aortic and inferior vena cava occlusions (during aortic: from 0.82 [0.76–0.86] before bupivacaine to 0.08 [0.05–0.14] after bupivacaine, P < 0.0001; during inferior vena cava: from 0.75 [0.57–0.81] before bupivacaine to 0.16 [0.07–0.33] after bupivacaine, P < 0.0001).

Neural networks involved in cardiac control can generate exaggerated sympathetic reflex responses in the presence of cardiovascular dysfunction and lead to lethal arrhythmias.^4,9  The primary aim of this study was to investigate spinal bupivacaine suppression of the cardio-spinal neuronal network responses to myocardial ischemia. To characterize the neural interactions at the level of spinal cord, we simultaneously recorded the activity of the network of intermediolateral and dorsal horn neurons with high-density microelectrode silicon arrays during myocardial ischemia and spinal anesthesia, while assessing sympathoexcitation induced the cardiac arrhythmogenesis. Our major findings are as follows: (1) at baseline, spinal anesthesia had no significant effect on the dorsal horn and intermediolateral neuronal activity; (2) during myocardial ischemia and cardiac load–dependent stressors, there was a significant increase in dorsal horn and intermediolateral cardiac neuron firing rate; (3) during myocardial ischemia, dorsal horn–dorsal horn cardiac neuron network interactions increased more than threefold, and this afferent activity synchronization during ischemia was diminished by bupivacaine; (4) dorsal horn–intermediolateral nucleus cardiac neuron network coordination was increased during ischemia, and this short-term coordination between afferent and efferent network was suppressed by bupivacaine; (5) spinal anesthesia prevented the sympathetically mediated shortening of ventricular activation recovery interval during myocardial ischemia and suppressed the dispersion of repolarization (arrhythmogenic substrate); and (6) spinal anesthesia reduced the episodes of premature ventricular contractions during myocardial ischemia.

Cardiac sympathoexcitation is one of the leading causes of cardiac arrhythmias and sudden cardiac death. Spinal or stellate ganglion blocks are being used to treat sympathetically driven ventricular arrhythmias that are resistant to conventional antiarrhythmic drug therapy, and novel spinal neuromodulation therapies are being developed to reduce cardiac arrhythmia and heart failure burden.^{11,27,43–45}  There exists a need to understand how these sympathetically driven arrhythmias develop and how best to optimize therapies for the patients. Providing insight into the coordination and integration of neural signals in the dorsal horn and intermediolateral neurons during myocardial ischemia is critical to the understanding of the spinal cord control of cardiac pathophysiology and optimizing new pharmacologic and bioelectronic neuromodulation therapies that can be tailored to the regulation of specific network activity.^4,46  While the myocardial substrate electrophysiologic changes underlying the antiarrhythmic effects of spinal anesthesia were recently described,¹¹  the neural network interactions in the spinal cord, through which the reduction in sympathetic output from the preganglionic neurons in the intermediolateral occurs, has not been characterized. In this study, we have investigated a possible mechanism through which spinal anesthesia reduces cardiac sympathoexcitation and thus provide important new insight into cardiac autonomic modulation at the level of the spinal cord.

Our results provide a foundation for future studies that may delineate complex neuronal interactions and network coordination in the spinal cord with other pharmacologic therapeutic agents. Administration of specific antagonist pharmacologic agents while recording neuronal activity can help identify the neurochemical signaling involved in the excitatory or inhibitory neural circuits in different spinal cord laminae. Further, alterations in spinal cord gene expression, including those involved in stress response and inflammation, are observed in chronic myocardial infarction and heart failure.⁴⁷  The impact of these changes on neural network interactions and their contribution to exaggerated sympathetic response in chronic heart diseases are areas of future investigation.

We show that a critical aspect of the cardio-spinal reflex response to cardiac stress is the significant increase in short-term coordination between dorsal horn to dorsal horn and dorsal horn to intermediolateral sites, indicating that local afferent–afferent and afferent–efferent neuronal network interaction at the level of the spinal cord contributes significantly to amplification of cardiac sympathetic activity. This network interaction is blocked by bupivacaine, reducing sympathetically mediated cardiac arrhythmogenesis. Our results showing bupivacaine induced reduction in dorsal horn–dorsal horn and dorsal horn–intermediolateral nucleus correlations during myocardial ischemia suggest that the antiarrhythmic effect of spinal bupivacaine is linked to its ability to disrupt the heightened spinal cord neural interactions and synchronization during sympathoexcitation. Our observations are also supported by a recent investigation by Ardell et al.⁴⁸  demonstrating that ischemia-induced neuronal transmission in the spinal cord amplified sympathoexcitation with an increase in cardiac norepinephrine release that triggered ventricular arrhythmias.

All applied cardiac stimuli in our study increased the number of correlated dorsal horn–dorsal horn and dorsal horn–intermediolateral nucleus pairs of neurons and the correlations of these neurons in the control state, suggesting that cardiac stimuli induce a hyperexcitatory state that increases neural interactions within the afferent and between the afferent and the efferent sites of the spinal cord. In this study, we have not tested the response to ischemia after the anesthesia wore off; however, this could be an area of investigation in the future to assess the memory effect of the bupivacaine.

Our results show that bupivacaine did not alter basal spinal neuronal activity (unchanged firing rate), but notably, it suppressed the cardiospinal neuronal response to myocardial ischemia and cardiac load–dependent stressors. This result sheds light on the underlying mechanisms in our previous study that showed thoracic epidural anesthesia did not impact cardiac electrophysiology at baseline, but rather reduced arrhythmogenicity during myocardial ischemia induced sympathetic hyperactivity.¹¹  Dorsal horn and intermediolateral neurons show periodic spikes in activity in response to transduction of sensory signals from myocardial ischemia, with a correlated temporal response between dorsal horn and intermediolateral neurons. This was abolished after bupivacaine administration. Our study specifically focused on the effect of bupivacaine on cardiac spinal neuronal activity. Previous single-unit studies show that noxious stimuli, such as myocardial ischemia, leads to the release of neuropeptides, including excitatory neurotransmitter glutamate and substance P, in the dorsal laminae of the thoracic spinal cord.^49–51  In response to cardiac ischemia, transient receptor potential vanilloid 1 receptors in the thoracic spinal dorsal horn mediate neuronal activation and release of substance P from the central terminal endings of cardiac ischemic–sensitive afferent neurons.⁵²  Bupivacaine has been shown to act on a multitude of channels and receptors to reduce neurotransmission.^53–55  Yanagidate and Strichartz⁵³  demonstrated that bupivacaine inhibits transient receptor potential vanilloid 1-induced excitatory transmission in the spinal dorsal horn during myocardial ischemia. Our result that bupivacaine attenuates ischemia-induced neuronal firing in the dorsal horn and intermediolateral may indicate that, in part, the attenuated neurons in the dorsal horn could be related to an ischemia-induced excitatory pathway such as the transient receptor potential vanilloid 1-mediated sympathetic preganglionic pathway.

Compared to the action of bupivacaine on dorsal horn neurons, its direct action on the neurons in the intermediolateral is less clear. We identified 33 efferent sympathetic preganglionic neurons through T2 paravertebral stimulation and observed that the firing pattern of these neurons in response to myocardial ischemia changed after bupivacaine application. Decreased cross-correlation between dorsal horn and intermediolateral neurons indicates a reduction in network interactions and suppression of afferent–efferent “cross-talk” that is responsible for sympathetic hyperactivity in the preganglionic neurons. Although further studies are necessary to identify whether bupivacaine has a direct effect on neurons in the intermediolateral nucleus, our findings suggest that bupivacaine acts on both afferent signaling and efferent sympathetic outflow, likely changing the interactions between the dorsal horn and intermediolateral during excess sympathoexcitation.

Sympathoexcitation leads to increased myocardial excitability and creates an arrhythmogenic substrate characterized by shortening of activation recovery intervals and dispersion of repolarization.^11,27  From an arrhythmogenicity standpoint, the difference of 15 to 20 ms in activation recovery interval and greater than 500 ms change in dispersion of repolarization is electrophysiologically very significant to create clinical arrhythmias.¹  Studies evaluating spinal neural activity modulation have also demonstrated equivalent changes in action potential duration-90 and effective refractory period in conjunction with the prevention of arrhythmias.^56,57  Increased dispersion of repolarization creates electrical heterogeneity in the myocardium and leads to cardiac reentry mechanism for ventricular arrhythmias.^10,11,29 

We have previously shown that thoracic epidural anesthesia selectively modulates the sympathetic driven cardiac excitability in the ischemic myocardium as compared to the remote myocardium by preventing reduction of activation recovery intervals and decreasing dispersion of repolarization.¹¹  Similar results were seen in the current study with spinal bupivacaine. Bupivacaine suppressed the magnitude of increase in dispersion of repolarization during acute ischemia from a mean of 677% in the controls to 468% with bupivacaine (P = 0.030) and reduced premature ventricular contraction burden during the brief ischemia period from a mean of 2.5 to 1. The short duration of ischemia triggered premature ventricular contractions, which, if the ischemia was to be further prolonged, would lead to lethal VTs due to reentry.

The attenuation of activation recovery interval shortening, reduction in dispersion of repolarization, and reduction in premature ventricular contractions were included to validate the model and demonstrate the effect of bupivacaine on cardiac sympathoexcitation, as previously described,^11,14,58  and confirm that the recorded neural responses were due to the effects of bupivacaine. The primary focus of this study was to show that one of the mechanisms behind the suppression of the arrhythmogenicity by bupivacaine could be altering the spinal neural network and interactions between spinal afferent and efferent sites. We showed that spinal bupivacaine blocked the integration and processing of afferent neural signals in the dorsal horn during myocardial ischemia and prevented coordinated activity of dorsal horn–intermediolateral nucleus to reduce the cardiac sympathoexcitation.

Both sympathetic and vagal influences work together in the autonomic nervous system toward the reflex control of the heart. Our studies were designed to study the impact of alternations in afferent signaling on the spinal sympathetic output in an in vivo model of intact vagal control. In our model, we investigated neural interactions solely within the T2 level of the spinal cord. Multilevel functional assessment will entail simultaneous insertion at two or more levels. It is known that bupivacaine blocks sodium channels on afferent root fibers as well as ventral root preganglionic sympathetic fibers.⁵⁹  Since this study did not focus on recording from the dorsal root ganglion and sympathetic trunk ganglion, it is not apparent whether root fibers were affected by bupivacaine administration. The extent to which sodium channel blockage in these fibers may have affected the attenuation of signals observed in the dorsal horn and intermediolateral is also unknown. Additionally, there are supraspinal inputs from higher centers that synapse onto the dorsal horn and intermediolateral regions.^23,60  Our study was not designed to delineate the influence of the higher centers on spinal cord processing and integration of cardiac signals. Also, the suppression in the neural interaction to the ischemia by bupivacaine that is reported in this study might be partially due to the ischemia preconditioning, although previous studies in this model suggest a limited effect with the brief periods of myocardial ischemia.³¹  In this study we did not test the control ischemia after the bupivacaine wears off; however, based on our other studies with a similar experimental model,^26,30,31,40  the time controls during model validation did not show a significant difference in the neural response to multiple stressors. We acknowledge that there are other novel approaches for cardioprotection from myocardial ischemia, such as hypoxia signaling, circadian rhythm, or adenosine signaling. We did not evaluate them in the current study.

Thoracic spinal anesthesia with bupivacaine suppresses the cardiospinal neuronal network responses to myocardial ischemia and cardiac stressors at the level of the spinal cord. Increased short-term coordination between dorsal horn to dorsal horn and dorsal horn to sympathetic preganglionic neurons at the level of the spinal cord contributes significantly to the increase in cardiac sympathetic activity. This increased cardiospinal network interaction during myocardial ischemia is blocked by bupivacaine, reducing sympathetically mediated cardiac arrhythmogenesis. Future characterization of types of responsive neurons, neurochemicals involved in network signaling and synchronization, and their responses to various modulatory approaches can allow more specific drugs and bioelectric therapies to be developed.

Supported by grants from the National Institutes of Health (Bethesda, Maryland). The study is funded by NIH RO1 HL136836. Dr. Mahajan is supported by NIH RO1 HL136836 and NIH R44 DA049630. Dr. Howard-Quijano is supported by NIH K08 HL135418.

The authors declare no competing interests.