Effects on Heart Rate from Direct Current Block of the Stimulated Rat Vagus Nerve

Abstract

Objective:

Although electrical vagus nerve stimulation has been shown to augment parasympathetic control of the heart, the effects of electrical conduction block have been less rigorously characterized. Previous experiments have demonstrated that direct current (DC) nerve block can be applied safely and effectively in the autonomic system, but additional information about the system dynamics need to be characterized to successfully deploy DC nerve block to clinical practice.

Approach:

The dynamics of the heart rate from DC nerve block of the vagus nerve were measured by stimulating the vagus nerve to lower the heart rate, and then applying DC block to restore normal rate. DC block achieved rapid, complete block, as well as partial block at lower amplitudes.

Main Results:

Complete block was also achieved using lower amplitudes, but with a slower induction time. The time for DC to induce complete block was significantly predicted by the amplitude; specifically, the amplitude expressed as a percentage of the current required for a rapid, 60 second induction time. Recovery times after the cessation of DC block could occur both instantly, and after a significant delay. Both blocking duration and injected charge were significant in predicting the delay in recovery to normal conduction.

Significance:

While these data show that broad features such as induction and recovery can be described well by the DC parameters, more precise features of the heart rate, such as the exact path of the induction and recoveries, are still undefined. These findings show promise for control of the cardiac autonomic nervous system, with potential to expand to the sympathetic inputs as well.

Introduction

Clinical Needs

Long-term changes in heart rate, specifically tachycardia, have been shown to significantly increase mortality rates [1]. An increase in heart rate is indicative of an increased sympathetic input to the heart, which has also been associated with the precipitation of cardiac arrythmias and sudden cardiac death [2-4]. Traditional treatments for these afflictions come in the form of pharmacological solutions (Beta blockers) or direct electrical stimulation of the heart muscle (implantable cardiac defibrillators, pacemakers) [5]. Electrical stimulation of the vagus nerve (VNS) has shown to be a useful neuromodulation technique to increase parasympathetic input to the heart; this has been seen to be an effective treatment for arrythmias such as ventricular tachycardia and fibrillation [6-9]. Some arrythmias, such as atrial fibrillation, occur when both the parasympathetic and sympathetic inputs are high [10]. Treatments such as sympathetic chain ablation can solve these conditions but are irreversible and can have significant side-effects [10]. Electrical block of the autonomic inputs to the heart may be a better alternative to ablation because it can offer similar reductions in the neural conduction but has the advantages of gradeability and reversibility [11, 12]. While these experiments focus on vagus stimulation and block, these principles apply to the sympathetic inputs to the heart as well, allowing for a much wider range of treatment options. Together, electrical stimulation and block supply complimentary approaches for augmenting the heart’s autonomic inputs that have the potential to create a robust treatment method for arrhythmic patients.

Vagus Background

The vagus nerve (cranial nerve X) supplies parasympathetic control to organ systems throughout the body, among them the cardiac, pulmonary, and digestive systems [13-15]. Recently, researchers have been investigating effects of electrical stimulation of the vagus as a treatment for obesity, depression, and cardiac arrythmias [16-19]. In the cardiac system, VNS produces a negative chronotropic effect by eliciting an increased release of acetylcholine at the neural endpoints on the heart; acetylcholine then activates the M₂ receptors on the cardiac myocytes, leading to a decrease in heart rate (HR) [2]. Because the vagus also contains afferent cardiac nervous pathways, and both afferent and efferent pathways from other organ systems, the effects from VNS on the heart and body are not precise, and the exact cumulative effects are not well defined[20].

Vagal Modulation as a Treatment

VNS has been shown to be an effective treatment method for several types of arrythmias, having been used as a method of counteracting sympathetic activity as well as inducing remodeling of cardiac innervation [21-23]. Conversely, ablation of the vagus can be beneficial when treating other forms of arrythmias such as atrial fibrillation [24]. With increases and decreases in vagal input to the heart offering treatments for different symptoms, it would be clinically advantageous to develop a method for reversibly decreasing vagal input. Electrical nerve block has been shown to be an effective method for fast, reversible disruption of action potentials in the motor system [25-34]. Translating these techniques to the autonomic nervous system could be a powerful tool for treatment.

Aside from arrhythmias, patients suffering from vasovagal syncope could likewise benefit from a reversible block of vagus conduction. Ablation of the vagal inputs to the heart can be very effective at treating vasovagal syncope [35, 36], but this is an irreversible solution; the effects from the ablation are always present, not just during a syncope episode. Applying electrical vagal block during the onset of syncope may reduce the severity of the event, while allowing normal vagal function when the patient is not experiencing symptoms.

Experimental Aims

These experiments sought to characterize the response of heart rate to electrical vagus nerve block used in the presence of electrically elevated vagal tone. VNS causes bradycardia and was used in this study to simulate an over-active vagal tone. The electrical block was implemented to restore the heart rate to the normal level, thus simulating the reversal of over-active vagal tone.

A primary goal of these experiments was to characterize the response of the heart rate to the application of direct current (DC) block, which is one approach to producing an electrical block. Specifically, they were designed to characterize the dynamics of the DC-blocked vagal system, with the eventual goal of developing a closed-loop controller for vagus nerve block. Studies using closed-loop control of vagal stimulation have found that a conventional proportional-integral-derivative (PID) controller may not be the most efficient control scheme [37]. Measuring the response of the heart rate from a step change in block amplitude will give us information on the processes involved in vagus nerve block. Defining functions to describe these processes, including their time constants, will allow us to design more effective control schemes. A well-functioning control system will allow further development of devices utilizing electrical nerve block as a treatment method.

In these experiments, the relationships between the applied block and its effect on the heart rate, the block “induction duration,” and the “recovery duration” after the block was turned off were found. One can use these system parameters to determine the responses and delays that must be incorporated into a controller design. “Induction duration” was defined to be the time between initiation of the DC block during simulated bradycardia and the return of the heart rate to pre-trial, baseline levels. This metric quantifies the transient response of a positive change in the block application. Similarly, “recovery duration” was defined as the time it takes the heart rate to return to pre-block bradycardia after the block was turned off. Our analysis used recovery metrics to measure the response of a negative step change in the system, but the system responses were found to be highly dependent on the prior block parameters.

Each experiment performed was one of 3 separate protocols, each designed to investigate a different aspect of the block phenomenon. The first protocol (“Sub- vs Supra-Block Threshold Effect on Block Percentage”) was designed to investigate the relationship between the DC block amplitude and the amount of change in the heart rate. Based on previous DC somatic nerve block studies [25, 26, 29, 38, 39], one would expect applying a higher amplitude blocking current to increasingly reduce vagus nerve conduction, resulting in a greater change in heart rate. The second protocol (“Sub-Block Threshold Induction Characterization”) focused on the effect of block amplitude on the induction duration of the block. A property of DC nerve block is that it is duration-dependent and will become more effective over time [25, 26, 38, 39], which is referred to as the induction effect. The last protocol (“Effect of Blocking Duration on Recovery Duration”) inspected the effect of the duration of the block (and additionally, net charge) on the duration it took to recover to the VNS baseline after the cessation of block. This delayed recovery of DC block has previously been seen in the peripheral nervous system [40], where an increase in total delivered charge led to longer recovery durations.

Methods

Electrodes

The nerve block implemented in these experiments uses a cathodic (depolarizing) DC waveform delivered via a Carbon Separated-Interface Nerve Electrode (CSINE). The CSINE has been shown to mitigate the effects of harmful chemical species created at the electrode during the application of monophasic current [40]. The CSINE uses a high-capacitance carbon slurry that was physically electrically connected to the current source, and ionically connected to the nerve through an electrolytic solution. The slurry was composed of two carbon powders: 22 wt% YP-50 (Kuraray, Canoga Park, CA, USA) and 2 wt% XC-72R (Cabot Corporation, Boston, MA, USA) suspended in a glycerol/NaCl “gel.” Approximately 1 mL of this slurry was placed into a capsule with a woven carbon fiber wire electrically connecting the slurry to the power supply. An ionically conducting membrane (Daramic 175, Charlotte, NC, USA) coated with non-porous polyvinyl alcohol lined one end of the capsule and separated the slurry from the isotonic saline used as an ionic conductor. Silicone and polyurethane tubes filled with the isotonic saline connected the carbon slurry capsule to the target nerve. The target nerve was secured in a small notch in the side of the tube (Figure 1c) for consistent, repeatable delivery of DC block. The high surface area of the carbon slurry reduces the amount of harmful chemical species that are released into the solution, and the ionic connection keeps any reactant formation from reaching the nerve. The carbon slurry creates a large surface for capacitive (double layer) charging, which increases the amount of time before Faradaic reactions occur. The slurry itself is contained in the capsule by an ion selective membrane.

Figure 1. Experimental Setup and CSINE:

a) Stimulating bipolar cuff electrodes, b) Vagus nerve, c) Saline-filled tube connecting the CSINE and the nerve, d) Needle electrode for return current, e) Carbon fiber wire electrode inside the capsule, f) Carbon slurry inside the capsule, g) Ionically conducting membrane to separate the slurry and the saline

While some heating of the fluid is likely due to the blocking current being pushed through the saline-filled tube, the power lost due to heat can be easily exchanged over the length of the tube at a difference of only a few degrees, as wall thicknesses were kept under 1 mm. As seen in the results herein, we are able to achieve at very low power, and over quicker time courses than if nerve heating were the mechanism of action. Over long blocking durations at high power, we would still expect the milliwatts of heat to be easily sunk into tubing and radiated into the air.

The stimulating electrodes were custom-made bipolar cuff electrodes comprised of two platinum foil contacts secured between pieces of silicone sheeting. An opening 1 mm in diameter was cut through the silicone over each platinum sheet to provide a small contact area; these openings were 2 mm apart on-center. The cuff was bent to form a ‘J” shape, and the nerve was placed inside. The relatively small openings increase the charge density emitted by electrodes, which lowered the current required to elicit action potentials [41].

Hardware

The stimulating electrodes were driven by a current-controlled pulse stimulator (Grass Instruments S88 with PSIU88, West Warwick, RI, USA) delivering monophasic VNS at 20 Hz with 50 μs pulses. A variable DC power supply (Keithley 2450, Solon, OH, USA) was used to provide a current-controlled blocking waveform to the nerve through the CSINE. The greatest blocking current used in these experiments was −4 mA, and the smallest was −0.125 mA. The heart rate was measured in one of two ways: 3-lead ECG or arterial blood pressure. The 3-lead ECG was measured by placing leads on the left front paw, right front paw, and a reference lead on the lower abdomen. The blood pressure was measured using a Transonic solid-state catheter (Transonic SP200, Ithaca, NY, USA) that was inserted into the rat’s left femoral artery. Signals from both techniques were collected using an amplifier and DAQ combination (CED 1902 and 1401-3, respectively Cambridge, EN, UK) and processed to calculate the beat-to-beat frequency of the heart rate. Data collection, post-processing, and feature extraction were done using CED’s Spike2 v8 software.

Surgery

A total of 25 rats were used in these experiments under protocols approved by the Case Western Reserve University IACUC committee. Each rat was anesthetized using inhaled isoflurane (induction 3-5%, maintenance 1.5-2.5%). The rat was laid supine, and the neck was dissected to expose the cervical vagus nerve on both sides. A bipolar platinum cuff stimulation electrode was placed on each side superiorly. The vagus nerve was then crushed on both sides using fine forceps, proximal to the stimulation electrodes to eliminate any afferent activity produced by the VNS. This afferent activity reduced the breathing rate in preliminary experiments, and crushing was used to maintain breathing rate during the experiments. After crushing, the effects of the vagus nerve stimulation are greatly isolated and allow us to specifically test its effect on the heart rate without off-target effects interfering with our results; while this does not remove every off-target effect and compensatory input to the heart, this preparation was able to create a setup that was stable enough to perform the experiments while minimizing invasiveness to the system. After crushing, the responses of unilateral VNS for both sides were analyzed and placed the CSINE blocking electrode on the side with the larger stable response. The complete surgical setup can be seen in Figure 1. Although in large animal models and human testing show that there are distinct differences in left- and right-side VNS, the literature indicates that these differences are not as significant in rats [42, 43]. Therefore, in these experiments, the side with the greater effect on heart rate was used. The exposure was left open to air, and excess fluid buildup was continually removed via a cotton swab to limit the potential for current to leak out of the electrode. To further reduce the chance of leakage currents, sheets of nitrile were placed under the vagus nerve to insulate the surrounding tissue from the electrical stimulation. Trials performed without this insulative layer were seen to cause local activation of the neck muscles, and the addition of the sheet successfully prevented this.

Experimental Procedure

To start an experiment, an appropriate VNS amplitude to simulate bradycardia determined by slowly increasing the stimulation current until a sharp decrease in heart rate (>25 bpm) occurred, generally requiring current in the range of 0.5 to 5 mA. Once the initial drop occurred, increasing the amplitude of the stimulation did not further decrease the steady-state heart rate, but only added overshoot that would decay back to the same steady-state level over short time periods. The minimum amplitude at which this low steady-state heart rate was elicited was defined to be the saturation amplitude for vagal stimulation and was used as the stimulation amplitude for the subsequent trials.

The DC blocking trials followed the following format: at least 30 s of baseline heart rate, 2 min of vagal stimulation to lower the heart rate and let it stabilize, selected amplitudes and varying durations of DC blocking current with continuing vagal stimulation, at least 2 minutes of vagal stimulation after block was turned off, and finally at least 2 minutes of recovery to baseline (Figure 2). If the return to VNS heart rate post-block, or the return to normal heart rates post-VNS took longer than 2 minutes, more time was added to these sections by waiting for the heart rate to reach the expected value, and then adding 2 additional minutes before moving to the next segment or trial. Trials were organized into sets of three; each set contained each of the three input values for either amplitude or duration, as shown in Table 1.

Figure 2. Example 360 s Trial:

a.) ΔHR, where a heart rate equal to the baseline is 100%, and one equal to the VNS level is 0%. b.) Induction delay, where there is no change in heart rate after the block is turned on. c.) Induction Ramp, where the heart rate increases as a result of the applied block, d.) Recovery Plateau, where the block heart rate remains high after block has been turned off. e.) Recover Ramp, where the heart rate trends down toward the VNS level after block has been turned off. f.) Block-Duration Integral, which is the area under the heart rate trace as a function of duration.

Table 1.

List of input and output parameters for each type of experiment.

Experiment Name	Block Length(s)	Block Amplitude(s)	Input Parameters of Interest	Output Parameters of Interest
Sub- vs Supra-Block Threshold Effect on Percent Change in Heart Rate	60s	50%-75% BT60 100% BT60 125%-200% BT60	Block Amplitude	% ΔHR at 60 s
Sub-Block Threshold Induction Characterization	360 s	50% BT60 75% BT60 100% BT60	Block Amplitude	% ΔHR at 60 and 360 s Block-duration integral Induction metrics
Effect of Blocking Duration on Recovery Duration	120 s 240 s 360 s	100% BT60	Block Duration Total Injected Charge	Recovery metrics

In these experiments, ‘block threshold’ was defined as the lowest block amplitude that was able to raise the heart rate to the pre-VNS level within 60 s (BT₆₀). BT₆₀ was found using a binary search with a typical resolution of 0.25 mA; if the block threshold was less than 1.0 mA, a resolution of 0.125 mA was used. The binary search method limited the number of required test pulses to approximately 5 different amplitudes. BT₆₀ was seen to trend both up and down by as much as 50% throughout an experiment; additionally, BT₆₀ would also change significantly when the blocking electrode was moved, even when re-placing it in the same spot. If BT₆₀ appeared to have changed, BT₆₀ would be remeasured between sets, but not between individual trials. Muscle contraction at the onset of DC was seen in some animals, but it has been shown that ramping to the therapeutic DC level can fully remove this onset response [44-46]. While this onset effect is clinically undesirable, and a hypothetical device would use the ramps to mitigate this issue, a square waveform was still used in these experiments to limit the number of experimental variables.

VNS saturation levels also increased throughout the experiment, although removing excess fluid from the electrode site helped solve this issue. VNS amplitude was adjusted as needed so that the same change in heart rate was achieved for every trial; the adjustments needed for the stimulation amplitude were typically much smaller (~10%) and changes occurred more slowly than for BT₆₀. Changing the amplitude for each trial was deemed appropriate, to ensure stimulation effects remained constant throughout all trials. If a change in the VNS saturation level did occur quickly (within a single trial), stimulation would be increased to an amplitude slightly higher than the saturation level (~10% increase). This increase in stimulation amplitude, along with reevaluating VNS efficacy more often, was able to prevent a diminished response from VNS.

Sub- vs Supra-Block Threshold Effect on Percent Change in Heart Rate

This experiment type compared sets of 3 different blocking amplitudes applied for 60 s and their effects on the relative change in heart rate (ΔHR). The amplitudes consisted of the block threshold (BT₆₀). one subthreshold level (50%-75% BT₆₀), and one suprathreshold level (125%-200% BT₆₀). These trials defined the relationship between blocking current amplitude and the resulting change in the heart rate. Specific sub- and supra-BT₆₀ percentages in each animal were chosen based on being able to measure a blocking effect at the sub-BT₆₀ level (i.e., the DC was high enough to affect heart rate) and by avoiding excessive recovery durations at supra-BT₆₀ levels. For a particular animal, the same sub- and supra-BT₆₀ percentages were used for all trials, even if BT₆₀ changed.

Sub-Block Threshold Induction Characterization

This type of experiment investigated how sub-BT₆₀ amplitudes related to the ‘induction’ effect seen in DC block (Figure 2b-c). Block was applied at amplitudes of 50%, 75%, and 100% BT₆₀ for 360 s, and the corresponding changes in heart rate were measured to determine induction durations for conduction block. While the 100% BT₆₀ value could be expected to block in 60 s, it was still measured to account for changes in the experiment preparation over time, such when BT₆₀ would change within a set.

Effect of Blocking Duration on Recovery Duration

This experiment characterized the relationship between blocking duration and the ‘recovery’ duration seen after block was turned off (Figure 2d-e). A current at 100% BT₆₀ was applied for sets of 2, 6, and 10 minutes, and the resulting recovery durations were recorded. The overall injected charge was also calculated as a potential predictor of delayed recovery. A list of all the different experiment types can be found in Table 1.

Data Analysis

After performing the trials described above, the data were analyzed to extract features of interest, some of which are defined below and illustrated in Figure 2:

Relative Blocking Amplitude, BT₆₀ (%) – the ratio of the blocking amplitude to the block threshold e.g., a blocking current of −1 mA in an experiment with a BT₆₀ of −2 mA is at 50% BT₆₀

Relative Change in Heart Rate, ΔHR (%) – the ratio of the change in heart rate caused by the blocking current to the change in heart rate caused by the vagal stimulation e.g. a trial with a baseline HR of 320, a HR after vagal stim of 280, and a measured heart rate of 310 during block would be a 75% ΔHR; a trial with the same baseline and vagal stimulation HRs where the HR drops further during block (e.g. no conduction block with baseline drift) to 270 would have a −25% ΔHR (Figure 2a)

Induction Delay Duration (s) – the duration after the block was turned on that the heart rate stays flat at the vagal stim level (Figure 2b.)

Induction Ramp Duration (s) – the duration between when the heart rate starts changing (after the induction delay) and before the heart rate levels off during application of a blocking current (Figure 2c.)

Recovery Plateau Duration (s) – the duration after block was turned off where the heart rate stays plateaued at the blocked level before it trends back down to vagal stim levels (Figure 2d.)

Recovery Ramp Duration (s) – the duration between the end of the recovery plateau and the heart rate setting at the vagal stim level after block. This measures the duration that the heart rate was trending down (Figure 2e.)

Block-Duration Integral (%×s) – the area under curve of the heart rate while the blocking current was applied, expressed as ΔHR × duration (Figure 2f.)

Statistical Analysis

Our analysis of the data collected first labelled each experiment as one of the three types of experiments. For each experiment type, the predictor variables were 3 categorical variables (3 amplitudes or 3 durations). The first type included 11 dependent variables, the second had 12, and the third had 9. Predictor variables included block level as a fixed effect and the experimental unit rat as a random effect. Since the fixed effect predictor variable level was categorical, the outcome variables were modeled using one-way analysis of variance mixed models. A variance component type was used to describe the covariance structure of the model. Residual plots were employed compliant with a normally distributed error structure of the outcome variables. Additionally, trials with heart rate drops (difference between baseline and the VNS level) that were less than 15 bpm were excluded (12% of trials), as change in heart rate could not be accurately evaluated due to normal heart rate variability. Pairwise comparisons between level means were assessed by pairwise contrasts among the 3 levels with no adjustment for multiple comparisons. PROC MIXED in SAS v9.4 was used to do the analysis. Regression analyses were done using Stata v17 and JMP v16. Regression models included linear, quadratic, exponential, and sigmoidal equations. Statistical significance throughout the experiments was α 0.005.

Results

Baseline and VNS Heart Rates

To ensure that our preparation did not alter the typical heart rate dynamics of vagal stimulation excessively, we compared our observed heart rates to those reported elsewhere. Our measured average baseline and VNS heart rates from before the block can be seen in Table 2, and these data are consistent with other researchers’ findings [47, 48]. The differences in ΔHR between each experiment type are potentially due to better surgical preparations and/or due to variations in the rat anatomy and physiology.

Table 2.

Average baseline and VNS heart rates for each experiment type

	Baseline HR	VNS HR	ΔHR bpm (%)
Sub- vs Supra-Block Threshold Effect on Percent Change in Heart Rate	311.7	279.2	32.5 (10.4%)
Sub-Block Threshold Induction Characterization	292.1	245.0	47.1 (16.1%)
Effect of Blocking Duration on Recovery Duration	275.2	222.2	53.0(19.3%)

Sub- vs Supra-Block Threshold Effect on Percent Change in Heart Rate

When designing a therapy using nerve block in this application, it would be useful to define the relationship between the amount of current applies and the change in heart rate seen. In these experiments, the overall trend was that an increase in DC amplitude corresponds to an increase in the amount that the HR was raised within 60 s. This was an expected result, but the relationship between amplitude and the heart rate does not appear to be linear, as the data show an asymptote at a maximum change in heart rate. While this is a useful result, it indicates that any automated implanted device will need more complex models in order to apply an appropriate amount of current.

Demonstrating this, a categorical analysis of variance (ANOVA) compared the three amplitude levels, two at a time. This analysis indicated that the differences between sub-BT₆₀ vs. BT₆₀ levels and the sub- vs. supra-BT₆₀ levels were significant (p<0.0001 for both), while the difference between the BT₆₀ and supra-BT₆₀ levels was not (p=0.1086). From these findings, it is believed that at supra-BT₆₀ levels, ΔHR could not be increased, i.e., the blocking effect has an upper limit as shown in the box plot in figure 3. However, in this experiment, the linear regression model (illustrated in Figure 4 as the blue line) indicated that ΔHR after 60 s was positively proportional to %BT₆₀ (equation: p<0.0001). Specifically, the linear term (α₁, figure 4) returned as significant.

Figure 3. Percent Change in HR After 60 s vs Relative Blocking Amplitude:

Categorical plot comparing changes in heart rate at block threshold, sub-threshold, and supra-threshold blocking amplitudes.

Figure 4. Percent Change in HR after 60 s vs Percentage of BT₆₀:

Plots, equations, and variables for linear and sigmoidal fits of percent of BT₆₀ and its corresponding change in heart rate.

To find a more accurate relationship function for the data, quadratic, exponential, and sigmoidal models were fitted to the same data. In the quadratic model, the linear term was significant, with the exponential term being strong but just outside the threshold. The analysis did not find any of the terms in the exponential model to be significant, and the fit from this model was approximately linear. In the sigmoidal model (Figure 4, green, the asymptotic (δ₀) and inflection point (δ₂) terms were significant, while the growth rate (δ₁) was not. The plot of this fit shows an asymptote at 84% ΔHR, reaching that level between 100% and 150% BT₆₀. The results from these models support the hypothesis that increases in blocking currents greater than BT₆₀ cannot raise the heart rate significantly higher. This was especially clear in the asymptotic term of the sigmoidal. Looking at the ΔHR asymptote position, the pre-trial baseline can be seen to be the rough upper limit of the heart rate. This may seem to be an expected result, but unlike the experiments in somatic nerves, there was a significant amount of background baseline activity in the vagus nerves. The heart rate is also not modulated directly, as it is dictated by many other factors; this is not the case with nerves that innervate skeletal muscle.

Sub-Block Threshold Induction Characterization

With induction times taking up to several minutes to reach complete block, an automated device will need to be able to predict both the change in heart rate and the speed of the effect. Since this is far from an instant change, properly defining the induction time courses will be useful to ensure that the device will not overreact to the slowly changing system. In this set of experiments, the data corroborated the %BT60 vs ΔHR relationship from the previous experiment type, as well as defined a relationship between DC amplitude and induction duration. Significantly quicker induction durations were achieved by raising the DC amplitude.

The analysis of the data collected in these experiments returned a linear regression that showed a significant positive relationship between %BT₆₀ and ΔHR. This was the case for both measurements 60 s into the block as well as at after the full 360 s (p<0.0001, both). A categorical ANOVA model to comparing these same variables at 60 s additionally found all levels to be significantly different (50%-75% p=0.0006; 50%-100% p<0.0001; 75%-100% p<0.0001). This aligned with the regression analysis and corroborated the results of the 60 s trials in the previous experiment type. After 360 s, the ANOVA showed the differences of ΔHR between the 50%-100% BT₆₀ levels were significant (p<0.0001), but that the differences between the 50%-75% BT₆₀ and 75%-100% BT₆₀ levels were not (p=0.0057, p=0.0118). This indicates that the block effectiveness at these different levels converge in this longer time frame. The use of the ANOVA is simply to confirm our linear results, as the distribution of %BT₆₀ is clustered at the three tested values.

To test this assumption of convergence, the 60 s and 360 s trials in were analyzed with a paired t-test. The mean difference of 360 s – 60 s ΔHRs was 29.86% (p<0.0001, 95% CI [23.21-36.52]), indicating that the trials overall showed a strong increase in block effectiveness as the trial continued past 60 s. Separating by %BT₆₀, the 50% BT₆₀ group saw a mean difference of 37.57% (p<0.0001, 95% CI [24.35, 50.80]), the 75% BT₆₀ group saw 33.77% (p<0.0001, 95% CI [22.10, 45.44]), and the 100% BT₆₀ group saw 17.10% (p = 0.0002, 95% CI [8.42, 25.78]). While the 100% BT₆₀ trials did see a significant change in ΔHR between 60 s and 360 s, the difference between them was smaller compared to the other two groups. The difference between the two time periods can be explained by variations of BT₆₀ within a single set, such that the new BT₆₀ was larger, as well as natural variations in heart rate over time.

Many of the trials in of this experiment type did not achieve complete block and restore the heart rate to the pre-trial baseline within the 6 minutes of block. Most of these trials were at the 50% BT₆₀ level, but some were at higher amplitudes. The distribution of induction durations and non-blocking trials for each block level is seen in Figure 6. The trials that did not block (DNB) are shown as striped bars. As seen in the figure, the 50% BT₆₀ trials were less likely to block (58% DNB) than the 75% and 100% BT trials (10% and 4% DNB, respectively). Comparing the proportion of DNB trials at each level using Fisher’s exact test, each level was found to be significantly different from each of the others.

Figure 6. Distribution of Total Induction Durations:

Histogram of induction durations binned over 1 minute. The striped bars on the right side of each section indicate the number of trials that did not block within 6 minutes (DNB).

To account for these trials that did not reach the pre-trial baseline, the analysis compared %BT₆₀ to the cumulative block effect (as measured by the integral of ΔHR as a function of duration (block-duration integral, Figure 2f.)). A linear regression analysis showed a positive correlation between these two variables (p<0.0001). A categorical ANOVA additionally showed significance between all levels (50%-75% BT₆₀, p<0.0001; 50%-100% BT₆₀, p<0.0001; 75%-100% BT₆₀, p=0.0001) (Figure 7.). This was consistent with previous findings that a lower amplitude waveform leads to smaller change in heart rate.

Figure 7. Block-Duration Integral vs Percentage of BT₆₀:

Comparison of the block-duration integral, which is the area under the normalized heart rate trace, at the 3 block amplitude levels.

Looking at the data for only the trials that did achieve full vagus conduction block and returned to the pre-trial HR level, a linear regression analysis showed a significant positive relationship between %BT₆₀ and induction ramp duration (p<0.0001). However, the induction delay and the total induction durations were not seen to be linearly correlated (p=0.0255, p=0.0467, respectively). After performing an ANOVA, the results (Figure 8.) showed similar results: the induction ramp showed significant differences comparing the 50%-100% BT₆₀ and 50%-75% BT₆₀ levels (p<0.0001, p=0.0003, respectively), but no significance between the 75%-100% (p=0.3707). Similarly, to the linear model, neither the induction delay nor total induction duration show significance differences between any levels. This suggests that while the slope of the block effect is affected directly by amplitude, there may be other factors at play affecting the total induction duration at sub-BT₆₀ levels. Figure 8 also illustrates that the induction ramp was a much larger portion of the total induction duration than the flat period.

Figure 8. Induction Durations vs Percentage of BT₆₀.

Induction metrics for completely blocking 360 s trials for each of the 3 block amplitudes.

Effect of Blocking Duration on Recovery Duration

The next set of experiments focused on characterizing the relationship between block duration and recovery durations. Defining this relationship has two potential applications for a future device: the device can use this information to keep the current low so that recovery can be as short as possible without sacrificing block efficacy, or it can use higher amplitudes to extend the recovery so that the effect lasts longer than the applied therapy, which may have user-comfort or power-saving benefits.

The linear regression model of these data showed a significant positive correlation between the block duration and the three different recoveries (plateau duration: p<0.0001, ramp duration: p=0.0027, and total duration: p<0.0001). The categorical ANOVA (figure 9 middle row) showed statistically significant differences between 2 min-10 min and 2 min-6 min levels in both the recovery plateau (p<0.0001, p=0.0014, respectively) and the total recovery durations (p<0.0001, p = 0.0050, respectively). The 6 min-10 min levels were not different for any recovery metric, and the recovery ramp showed no significance between any of the levels. Together, these indicate that an increase in the duration of applied DC block increases the duration until full recovery, but that the relationship gets weaker as durations increase.

Figure 9. Analysis of Recovery Metrics:

The leftmost column (blue) is the recovery plateu, the middle column (green) is the recovery ramp, and the rightmost column (red) is the total recovery.

Top Row: The region of the heart rate responses that were measured. These are taken from the example trial in Figure 2.

Middle Row: Plots of the relationship between the duration of the cathodic DC block and the duration of each recovery metric.

Bottom Row: Plots of the relationship between the total charge of the DC block (measured in negative milliCoulombs) and the recovery metrics.

In addition to block duration, the total injected charge of the block (duration × amplitude, expressed as negative mC) was compared to recovery durations. Using a linear regression, the injected charge was positively correlated to all three recovery metrics (all p<0.0001) (figure 9 bottom row).

Discussion

The results gathered from these experiments show that DC block of the rat vagus nerve share several characteristics to DC block of motor nerves [25, 26, 38-40] but the temporal dynamics of these two systems differ significantly. In the motor system, the rate-limiting process occurs in the block of the nerve, as the muscle responds very quickly to changes in stimuli. In the autonomic system, stimulation does not directly innervate the heart muscle, but instead modulates heart rate with the release and diffusion of neurotransmitters. These chemical processes have their own rate constants, and it is difficult to differentiate which delay is predominant at any point in time. Additional autonomic systems (such as the sympathetic response and baroreceptor reflex) are also influencing the heart rate, and these effects will contribute to the differing dynamics of this system from the musculoskeletal system. More work is needed to precisely define the relationship between the neural and cardiac delays; a fully developed model of this system would be invaluable to anyone implementing these techniques in a clinical setting.

Experimental Setup

All the experiments described above were performed under general anesthesia via inhaled isoflurane. Isoflurane can affect the sympathetic and parasympathetic inputs to the heart, with varying mechanisms proposed for these changes [47]. Our baseline heart rates were within the observed ranges seen for rats under isoflurane concentrations between the 1.5% and 2.5% used in these experiments [47, 48].The exact concentration used varied both between and within experiments, and the concentration was adjusted with priority given to the anesthetic depth of the animal.

The effects of crushing the proximal vagus on both sides did not make a significant long-term change in resting heart rate, likely due to the compensation of the intact sympathetic inputs. While stimulus parameters were adjusted to maximize the effects from VNS, the changes in heart rate (average 10-20%) were in line with previously observed changes [48]. Variance in the baseline and VNS heart rates were seen throughout the experiments, but these were accepted to be a natural variance in heart rate. It is also known that Isoflurane can affect the cardiac system dynamics, which may have an effect of some of the effects observed here.

As stated above, BT₆₀ was seen to vary significantly within and between experiments. The causes of this are unknown, but care was taken to ensure that the block threshold was measured and maintained consistently throughout the experiments. Chemical damage was mitigated by using the CSINE, and a sufficient diameter of tubing was used to avoid excess heating of the saline and nerve [40]. The primary hypothesis for these differences in block thresholds is simply the differences in preparation. Each rat and surgery come with their own idiosyncrasies, and small changes in the position and rotation of the nerve and changes in the exposed preparation site could potentially contribute significantly to the block threshold.

Sub- vs Supra-Block Threshold Effect on Percent Change in Heart Rate

The results from these experiments show a clear increase in the change in the heart rate as the blocking current increases. This illustrates the gradeability of this cardiac modulation technique, which could allow for clinical applications where heart rate is actively adjusted to suit the transitory needs of the patient. An upper limit to the change in heart rate was seen, which can be easily explained physiologically by the fact that once the proximal vagus is crushed, little-to-no background vagus tone is present. Without this tonic activity, the only activity on the nerve to block is what is produced by stimulation; therefore, if all VNS is blocked, this will look the same as if the VNS were not on at all (i.e., the similar to our baseline measurement). Had the vagus nerve been left uncrushed, blocking may have allowed an increase the heart rate to above the resting baseline HR by downregulation of the tonic vagus activity. Blocking only the vagal tone could potentially allow us to modulate the heart rate without the use of VNS. This effect would explain some of the >100% blocking percentages seen during these experiments, as a portion of the vagal tone would then remain to be blocked. Another explanation of these observed results is the natural drift in the baseline heart rate throughout the duration of the trial.

Sub-Threshold Induction Characterization

These longer trials expanded upon the effects seen in the first experiment type. Comparing ΔHR to %BT₆₀ at 60 s into the block application, the conclusions from the first experiments were confirmed: increasing the current leads to an increase in change in heart rate. However, measured after 360 s, the differences between these levels decrease, and they all approach the maximum possible conduction block. This was illustrated by the paired 60 s vs 360 s analysis, which showed that the differences in ΔHR between the two time points decreased as the DC amplitude was increased, showing a plateauing of the effect. This data also serves as an illustration of the DC block induction effect, where complete block can still be achieved using lower amplitudes for longer periods of time, which could greatly improve power requirements and device safety. The presence itself of the delayed induction indicates that there are some time- or charge-dependent features of block that are not strictly related to the instantaneous current at the nerve. These features are seen in different DC block applications as well [25, 39], so it is unlikely that it is due to the CSINE, and more likely a physiological process.

A practical limit to this induction effect was seen when looking at the number of trials that did not block. As seen in figure 6, a significantly higher proportion of trials that were run at 50% BT₆₀ did not block compared to 75% and 100% BT₆₀. Some of the DNB trials achieved a partial ΔHR within the 6 minutes of block, but others showed very little change in the heart rate. This suggests that there is a minimum amplitude needed to elicit the induction effect; based on the large increase in DNB trials seen in Figure 6, this minimum threshold likely occurs around 50% BT₆₀.

The ΔHR-duration integral was used to normalize the induction effects between trials that both did and did not fully block. This metric was used to characterize the partial blocks that are not captured by measuring the duration it takes to reach complete conduction block. A fast, complete induction will have a large ΔHR-duration integral, with slower and incomplete inductions decreasing the value. This allows us to analyze more of the data for induction effects, as a trial that reached only a 50% change in heart rate was marked as DNF, but the ΔHR-duration integral includes this effect. The ΔHR-duration metric showed a strong positive correlation in response to the relative block amplitude, which further confirms the blocking amplitude to induction effect relationship. Many of the sub-BT trials appeared to level off at a partial block effectiveness, which could be a beneficial clinical treatment. A well-defined relationship between the induction effect could potentially indicate whether an upward-trending heart rate will reach full block, or level out.

To explore the specific features of the induction effect, the analysis looked at results from only completely blocking trials. This indicated that the ramp component of induction appears to be the most correlated to the amplitude. The physiological reason for the initial flat portion of the induction is not known, nor why it appears to be poorly related to amplitude. It is possible that this may be influenced by the anatomy of a specific animal, the surgical preparation, or some of many other environmental variables. The reduced number of completely-blocking trials may have also reduced our ability to identify trends in this metric. Since only part of the induction effect can be characterized from these experiments, more work needs to be done to understand its relationship to an applied block.

Effects of Blocking Duration on Recovery

These experiments show a strong relationship between the duration of block and delay to recover back to normal conduction. Specifically, the duration of the recovery plateau seems to vary with block duration more significantly than the recovery ramp back to full vagus conduction. From these data, a potential mechanism that describes this response is proposed: by modeling the vagus as a bundle of individual nerves, the elongated plateau response can be described as the period where no fibers are recovered. The ramp can be an effect caused by individual axons returning to normal conduction after varying delays. Each fiber thus experiences a unique delay before returning to normal conduction. While a large portion of these delays can be attributed to the amount of DC block, the variance in these responses may be caused by the characteristics of individual axons (myelination, diameter, and geometric distance from the electrode). Further investigation should be done to determine what specific characteristics affect the recovery duration. A well-defined model of the recovery effects would be clinically significant in predicting the response to DC nerve block. Both the limiting and extension of the recovery duration could be useful in different treatment therapies; for example, a long recovery duration may be useful in a chronic application (ex. chronic arrythmias), but instant recovery may be desired for acute scenarios (ex. vasovagal syncope episodes).

The relationship between injected charge and the recovery duration provides more insights for characterizing the recovery effect. Since the relationships between charge and recovery durations are so strong, it is possible that absolute amplitude (not relative to BT), in addition to duration, plays a factor in the delayed recovery. This was particularly interesting, as induction seems to be more related to amplitude relative to block threshold, as opposed to absolute amplitude; this fact suggests that the induction and recovery effects may be caused by distinctly different physiological mechanisms. Recent research shows that the delay in recovery may be related to K⁺ diffusion and accumulation in the extracellular space [49-51], but it is unclear what role this plays in the induction effect, as complete conduction block has been achieved instantly without the delay needed for significant ion movement. If the diffusion of K+ ions is the cause of these features, the net injected charge would be a very useful tool, as the ions are the carriers of this charge. However, this is not confirmed, and there may be other mechanisms at play that may make charge an inferior predictor compared to a metric like block duration.

Application Limitations

Several issues become clear when trying to extend the results of this study to a broader application. The largest by far being the crushing of the proximal vagus nerve to limit afferent activity. While this was useful to limit the number of variables in this study, it is not a proper model for clinical behavior; however, it was an effective tool to narrow down the effects seen from VNS to just those on heart rate. The issues seen that necessitated the crushing of the vagus were mostly due to off-target effects of stimulation and by autonomic pathways trying to negate the effects of the stimulation. These issues will not be relevant in a clinical application of DC nerve block, as the stimulation is no longer needed. The effects in the study may have been affected by the intrinsic cardiac nervous system’s responses to VNS (e.g., rebound tachycardia after VNS) which has the potential to change the responses of the system when VNS is not included. However, our previous work in the motor system shows a similar effect which is not influenced by any intrinsic response.

The general surgical preparation is also not a perfect representation of an implanted device. Primarily, it used an open dissection, whereas a clinical device would be completely closed and sealed. Repair responses such as healing and scarring could also change the local anatomy, and potentially create insulating and conducting paths that will direct current. These could potentially change the current it takes to reach block threshold, but the normalization of the data to the block threshold has the potential to compensate for these effects. Chronic studies will need to be performed to assess the changes in blocking characteristics over time in an implanted device, as well as assess any long-term effects of repeated DC application and recharge

The off-target effects of blocking the vagus were also not comprehensively observed. The vagus nerve contains fibers of varying anatomy innervating a wide assortment of organs. Indiscriminate blocking of vagus conduction may then cause issues for other bodily functions. A method for selective block of cardiac fibers may then be required; this could be achieved using more complex electrical field manipulation strategies, or by simply putting the electrode closer to the neural endpoints on the heart, after most off-target fibers have branched off. This would be more easily accomplished in a larger animal model, however that will come with a different set of issues due to different fascicular structure, as well as lower heart rates in general.

Conclusions

From these experiments, it was seen that the responses in the system are dependent on more than just the applied amplitude and duration. Changes in heart rate are not induced immediately, nor do they revert immediately after the block has commenced. These delays have been shown to be, in part, related to the parameters of the applied DC current, but not all the measured findings can be fully explained. Specifically, the DC amplitude was a significant predictor of the amount of block after 60 s, the block-duration integral, and the induction ramp; DC duration and charge were significant predictors of the recovery plateau duration and total recovery duration.

Further experiments will be needed to better characterize these induction and recovery effects, but the data shown provides evidence about which factors may be significant predictors. Additionally, to represent a more real-world clinical setting, future experiments may exclude the crushing of the proximal vagus. With the vagus still functioning otherwise normally, effects of stimulation and blocking can be explored, as well as blocking without the use of stimulation.

For a clinical application, a smaller electrode system would need to be developed. This study used quite large volumes of carbon slurry in the CSINEs in order to be sure that they would remain effective for the full duration of the experiment. This size would not be conducive to implantation. Reducing the volume of the CSINE could bring its operating times down from several hours to a few minutes before a repolarizing recharge cycle is required, depending on the volume of slurry used. A device would need to find a balance between electrode size and capacity. Carbon slurry in this form is also not currently used in medical devices; however, the slurry does not come in direct contact with the tissue, and the ion-permeable membrane should be considered as the “electrode surface.” The membrane is based on polyethylene, which is biocompatible, but more studies need to be done to gauge membrane efficacy over time as immune responses effect the surface.

Overall, these data show promise that there is a path forward closed-loop control of the HR via vagus nerve block. Preliminary attempts at closed-loop control have shown promise [52], but more work is required to fully model the system and account for every possible off-target effect. DC block has advantages over other block schemes in that it can apply therapy for long periods of time and can be modulated with a single parameter. AC blocking techniques have some limitations that DC does not, such as a lack of block during transition periods in low-frequency block [53], and onset activity on high-frequency block. Looking forward, it is also possible to apply these blocking techniques to the sympathetic inputs to the heart as well. This would expand electrical nerve block treatments to both bradycardia and tachycardia, allowing these techniques to benefit from a larger pool of patients. Adapting these results to the sympathetic chain will also require additional effort to analyze the different possible electrode locations (preganglionic vs postganglionic, etc.). Measuring cardiac outputs other than the heart rate may also reveal other effects that can be modulated. Combining all these effects into an automated device utilizing electrical nerve block would be a powerful tool for to provide personalized and actively adapting treatments to patients without existing options.

Figure 5. Percent Change in HR after 60s and 360s vs Percentage of BT₆₀:

Plot comparing the percent change in heart rate at 60 s and 360 s into the block against the proportion of the block threshold.

Definitions:

VNS

Vagus nerve stimulation

HR

Heart Rate

ΔHR

Change in heart rate, expressed as a percent normalized to baseline and VNS levels

DC

Direct current

BT₆₀

60 second block threshold DC amplitude required to fully block in 60 seconds

CSINE

Carbon separated interface nerve electrode, the electrode used to deliver DC block