The Impact of Burst Motor Cortex Stimulation on Cardiovascular Autonomic Modulation in Chronic Pain: A Feasibility Study for a New Approach to Objectively Monitor Therapeutic Effects

Abstract

Introduction

Chronic refractory pain of various origin occurs in 30–45% of pain patients, and a considerable proportion remains resistant to pharmacological and behavioral therapies, requiring adjunctive neurostimulation therapies. Chronic pain is known to stimulate sympathetic outflow, yet the impact of burst motor cortex stimulation (burstMCS) on objectifiable autonomic cardiovascular parameters in chronic pain remains largely unknown.

Methods

In three patients with chronic pain (2 facial pain/1 post-stroke pain), we compared pain intensity using a visual analog scale (VAS 1–10) and parameters of autonomic cardiovascular modulation at supine rest, during parasympathetic challenge with six cycles per minute of metronomic deep breathing, and during sympathetic challenge (active standing) at baseline and after 4 months of burstMCS compared to age-/gender-matched healthy controls.

Results

While two out of three patients were responsive after 4 months of adjunctive burstMCS (defined as pain reduction of > 30%), no differences were found in any of the three patients regarding the R-R intervals of adjacent QRS complexes (RRI, 642 vs. 676 ms) and blood pressure (BP, 139/88 vs. 141/90 mmHg). Under resting conditions, parameters of parasympathetic tone [normalized units of high-frequency oscillations of RRI (RRI-HFnu power) 0.24 vs. 0.38, root-mean-square differences of successive RRI (RRI-RMSSD) 7.7 vs. 14.7 ms], total autonomic cardiac modulation [RRI total power 129.3 vs. 406.2 ms², standard deviation of RRI (RRI-SD) 11.6 vs. 18.5 ms, coefficient of variation of RRI (RRI-CV) 1.9 vs. 3.7%], and baroreceptor reflex sensitivity (BRS, 1.9 vs. 2.3 ms/mmHg) increased, and parameters of sympathetic tone [normalized units of low-frequency oscillations of RRI (RRI-LFnu power) 0.76 vs. 0.62] and sympatho-vagal balance [ratio of RR-LF to RRI-HF power (RRI-LF/HF ratio) 3.4 vs. 1.9] decreased after 4 months of burstMCS. Low-frequency oscillations of systolic blood pressure (SBP-LF power), a parameter of sympathetic cardiovascular modulation, increased slightly (17.6 vs. 20.4 mmHg²). During parasympathetic stimulation, the expiratory–inspiratory ratio (E/I ratio) increased slightly, while upon sympathetic stimulation, the ratio between the shortest RRI around the 15th heartbeat and the longest RRI around the 30th heartbeat after standing up (RRI 30/15 ratio) remained unchanged.

Conclusion

Four months of adjunctive burstMCS was associated with an increase in parameters reflecting both total and parasympathetic autonomic modulation and baroreceptor reflex sensitivity. In contrast, sympathetic tone declined in our three patients, suggesting stimulation-associated improvement not only in subjectively perceived VAS pain scores, but also in objectifiable parameters of autonomic cardiovascular modulation.

Key Summary Points

Neurostimulation therapies—in particular motor cortex stimulation (MCS)—for chronic pain lack high-quality evidence.

Commonly applied subjective score-based outcome measures are biased by intra- and inter-individual variability.

The effects of burstMCS on pain levels and parameters of cardiovascular modulation remain largely unknown.

Our preliminary findings indicate that burstMCS evokes pain relief and objectifiable changes in sympathetic and parasympathetic tone.

Large-scale, sham-controlled trials are needed to clearly characterize these clinical effects on both pain and autonomic cardiovascular modulation.

Introduction

Chronic pain of various origin (post-stroke, post-surgical, post-infection) affects approximately 30–45% of pain patients and is characterized by a broad clinical spectrum of pain phenotypes including, but not limited to, central post-stroke pain, complex regional pain syndrome, facial pain, extremity pain, back pain, and other pain disorders. Given the fact that these pain phenotypes derived from distinct and not well-understood pathophysiological pathways, clinical diagnosis and management of chronic pain appear to be challenging, thus slowing recovery in a considerable proportion of affected individuals [1–11]. In the case of failure and/or limited responsiveness to first-line pharmacological and behavioral therapies, central (transcranial magnetic stimulation (TMS), transcranial direct current stimulation, transcranial alternating stimulation, motor cortex stimulation (MCS), deep brain stimulation (DBS), and peripheral neurostimulation therapies [vagal nerve stimulation, peripheral nerve stimulation (PFS), spinal cord stimulation (SCS)] have been applied adjunctively to promote pain relief under these challenging circumstances, yielding response rates ranging from 25 to 45% [12–22]. However, noninvasive techniques such as repetitive transcranial magnetic stimulation (rTMS), transcranial direct current stimulation (tDCS), vagal nerve stimulation (VNS), and nerve field stimulation (NFS) have provided heterogeneous results of moderate quality. The level of evidence for minimally invasive spinal cord stimulation (SCS) and peripheral nerve field stimulation (PNFS) has been classified as low to moderate depending on the pain condition treated (e.g., neuropathic extremity pain, headache). In summary, interdisciplinary guidelines confirmed the heterogeneous and, in part, inconclusive state of neurostimulation therapies for chronic pain disorders [23]. Uncontrolled study design, short-term follow-up, heterogeneity of implantation procedure, applied stimulation protocols, different pain disorders/diagnosis, and participant selection represent the main barriers for high-class evidence [23]. Invasive brain stimulation techniques such as epidural MCS or DBS have been reported to promote pain relief in uncontrolled clinical observational studies. Notably, neither cranial neurostimulation therapy is approved (FDA/CE), and thus both are used off-label. However, conventional MCS waveforms have generally been used in recent decades to treat different chronic pain disorders [17, 19–22, 26]. MCS using the burst mode has been explored in a few patients and was initially developed as an SCS waveform [13, 18, 24–26].

Chronic pain evokes an increase in sympathetic outflow pathways, which in turn may lead to a deterioration in cardiovascular-autonomic modulation, causing autonomic dysfunction. This characteristic decline in parasympathetic tone and an increase in sympathetic activity co-occurs frequently along with sensorimotor, cognitive, metabolic, and sleep impairment, and contributes essentially to a deteriorated functional state and a poor long-term outcome of pain management [27–31]. However, the impact of MCS using burst stimulation waveforms (burstMCS) on a potentially altered cardiovascular imbalance in chronic pain remains largely unknown. This is the first report assessing the mechanistic effects of MCS using burst stimulation waveforms on parameters of the cardiovascular-autonomic system in chronic pain.

Methods

This single-center, prospective observational study was performed to assess the feasibility of autonomic modulation characteristics as potential outcome measures for MCS therapy. The study included three patients with chronic pain (two facial pain and one post-stroke pain), where we evaluated pain-perception (VAS) and parameters of autonomic cardiovascular modulation at supine rest, during parasympathetic challenge with six cycles per minute of metronomic deep breathing, and during sympathetic challenge, i.e., active standing at baseline (burstMCS Off) and after 4 months of burstMCS. Autonomic parameters were compared to age-/gender-matched healthy controls. The goal was to obtain preliminary clinical data and results as a proof of concept, which may later guide the conceptualization of a randomized controlled blinded study protocol. According to federal regulation and FDA guidance (45 CFR 46.116(f), Common Rule), German Federal Law for data protection (BDSG §4), and the Bavarian Hospital Law for data protection (§27, section 4), institutional review board (IRB) approval was waived, as the autonomic testing procedure represents a noninvasive standard examination in the clinical routine of different neurological diseases and is associated with “none-to-minimal risks to subjects” assessed in our study [irb.ucsf.edu/waiving-informed-consent]. All study participants provided written informed consent prior to participating in the study and for data analysis. The study was performed in accordance with the Helsinki Declaration of 1964 and its later amendments.

fMRI-Guided BurstMCS Lead Implantation and Stimulation Protocol

Two four-contact paddle leads (Abbott Inc. Plano, TX USA) were inserted under general sedation, targeting the motor strip of the cortex. The application of burst patterns for MCS was off-label, as burst stimulation waveforms have been approved (FDA/CE) only for use in spinal cord stimulation. Notably, MCS in general is still off-label, although many practicing neurosurgeons consider MCS for interventional pain treatment. MCS leads were accurately implanted using image guidance with fused preoperatively acquired CT and functional MRI (fMRI) datasets. For fMRI, we used a 1.5 T MR scanner with echo planar imaging (Magnetom Sonata, Siemens Medical Solutions, Erlangen, Germany). Measurements were performed with 25 slices of 3 mm thickness and resolution. We used TR = 2470, TE = 60. Stimulation was done in a block paradigm (“boxcar”) with 180 measurements in six blocks. The procedure consisted of 30 measurements during a resting condition alternating with 30 measurements during movement of the respective body parts. Patients were informed via headphones when to start or stop movements. For motion correction, we applied an image-based prospective acquisition correction applying interpolation in the k-space [32]. We constructed activation maps by analyzing the correlation between signal intensity and a square wave reference function for each pixel according to the paradigm. Pixels exceeding a significance threshold (typical correlations above a threshold of 0.3 with p < 0.000045) were displayed if at least six contiguous voxels built a cluster to eliminate isolated voxels. We aligned the functional slices to MPRAGE (magnetization-prepared rapid acquisition gradient echo) images using 160 slices of 1 mm slice thickness and resolution obtained from the same patient position (Fig. 1).

Fig. 1.

fMRI (1.5 T MR scanner; Magnetom Sonata, Siemens Medical Solutions, Erlangen, Germany). Derived from measurements consisting of 25 slices of 3 mm thickness and resolution. Stimulation was applied with 180 measurements in six blocks including 30 measurements in resting condition followed by 30 measurements during movement of the target body part (facial area)

In a second operative step, the MCS leads were connected to a subcutaneous, sub-clavicular rechargeable IPG (Prodigy IPG; Abbott Inc. Plano, TX, USA). Postoperative sequential CT (3D reconstructions) scans demonstrated appropriately and perpendicularly placed electrodes relative to the central strip (Fig. 2). In all three patients, stimulation intensity started at 20% of the motor threshold and was subsequently increased to 80% of the motor threshold, confirmed clinically by motor response of the arm or facial muscles. The following stimulation parameters were applied: 1.4–3.2 mA intensity (stimulation parameters were adjusted within 14 days after implantation), bipolar configuration, burst rate 40 Hz, intra-burst 500 Hz, pulse width 1 ms. The activated contacts were chosen accordingly to ensure sufficient stimulation of the arm/leg in one patient and facial area in two patients of the motor strip (Table 1).

Fig. 2.

CT scans using 3D surface rendering demonstrating the location of the two four-contact paddle leads implanted via a small craniotomy targeting the central strip

Table 1.

Pre-implantation clinical characteristics of the study participants (age, gender, pain disorder, medication, previous treatments, comorbidities)

Patient	Age (years)	Gender	Pain disorder	Medication	Previous treatment	Pre/post-burstMCS	Stimulation parameters	SAE
1	42	F	TN V2/V3 right 5 years	Pregabalin NSAIR Opioids Dronabinol	MVD SRS TC Botox	VAS 9/10 vs. VAS 8/10	1.4–2.7 mA Bipolar Burst rate 40 Hz, intra-burst 500 Hz, pulse width 1 ms	–
2	62	F	TN V2/V3 right 10 years	Carbamazepine Mirtazapine	SRS TC Botox	VAS 10/10 vs. VAS 6/10	1.4–3.0 mA Bipolar Burst rate 40 Hz, intra-burst 500 Hz, pulse width 1 ms	–
3	74	M	PSP Hemibody right 10 years	Opioids Amitriptyline Antidepressants	–	VAS 10/10 vs. VAS 3/10	2.0–3.2 mA Bipolar Burst rate 40 Hz, intra-burst 500 Hz, pulse width 1 ms	–

MCS motor cortex stimulation, VAS visual analog scale, SAE serious adverse events, F female, M male, TN trigeminal neuralgia pain, PSP post-stroke pain, MVD microvascular decompression, SRS stereotactic radiosurgery, TC thermocoagulation of ganglion gasseri , mA milliamperes, Hz hertz, ms milliseconds

Protocol of Autonomic Testing and Data Analyses

Measurement of RR Intervals, Blood Pressure and Respiration at Rest and During Autonomic Challenge Maneuvers

We studied autonomic cardiovascular modulation at resting conditions and autonomic responses to parasympathetic challenge with six cycles per minute (cpm) of metronomic deep breathing, and sympathetic challenge, i.e., active standing (Ewing test). After resting for 40 min to ensure a stable cardiovascular situation, autonomic testing was performed between 9 am and 2 pm in a quiet room with ambient temperature of approximately 24 °C and stable humidity [33].

Using a three-lead electrocardiogram (ECG), we monitored electrocardiographic RR intervals (RRI). Beat-to-beat systolic and diastolic blood pressure (SBP, DBP) were measured using noninvasive finger pulse photoplethysmography at the index or middle finger. Respiratory frequency was monitored using a piezoelectric respiratory belt attached to the lower thorax. Bio-signals were recorded for 5 min in supine position, for 2 min when breathing was paced at six cpm to induce parasympathetic activation, and during 10 min of sympathetic stimulation, i.e., active standing.

Data Storage, Analyses of Time-Domain Parameters and Spectral Power of Autonomic Modulation

Data were sampled on a custom-designed data acquisition and analysis system (SUEmpathy™, SUESS GmbH, Germany) for offline analysis.

From 2-min segments under resting conditions with clear signals and without arrhythmias, mean values and standard deviation (SD), along with time-domain parameters of cardiovascular autonomic modulation, were calculated in a linear analysis. Parameters thus obtained were the RRI-SD and coefficient of variation of RRIs (RRI-CV), both reflecting total autonomic cardiac modulation, and the root-mean-square of successive differences in RRIs (RRI-RMSSD), which in turn represents vagal cardiac modulation. RRI and BP values show underlying fluctuations that are largely mediated by the undulating activity of the autonomic nervous system. We applied trigonometric regressive spectral analysis (TRS) to sympathetic and parasympathetic oscillations. In short-term recordings, three main peaks of oscillation can be identified using TRS algorithms: at very low frequency (VLF; < 0.04 Hz), at low frequency (LF; 0.04–0.14 Hz), and at high frequency (HF; 0.15–0.40 Hz). The magnitude of spectral power can be compared by determining the area under the power spectral density curves within the respective frequency range, and is then defined as VLF , LF and HF power of RRI or SBP. VLF oscillations of RRI are influenced by multiple factors, such as the renin–angiotensin system, some endothelial factors, and thermoregulation, and are therefore not used for analyses in the following. LF oscillations of RRI at rest (RRI-LF power) represent mainly sympathetic and to an unclear extent also vagal outflow, while HF oscillations of RRI at rest (RRI-HF power) are assumed to be influenced by vagal outflow only. BP fluctuations in the HF range are primarily a mechanical consequence of respiration-induced fluctuations in venous return and in cardiac output, whereas BP LF fluctuations are mainly related to oscillations in sympathetic outflow.

Besides describing the LF and HF power components in absolute values (ms²/Hz), we also calculated normalized units (RRI-LFnu and RRI-HFnu power) from the raw values of RRI-LF or RRI-HF power divided by the spectral power, i.e., the sum of RRI-LF and RRI-HF power. Overall, normalization results in minimizing the effects of the changes in the total power on the values of the LF and HF components. Furthermore, RRI-LF and RRI-HF power indicate another parameter of total autonomic cardiac modulation (RRI total power). Finally, RRI-LF/HF ratios were determined as an index of sympatho-vagal balance.

Furthermore, baroreceptor reflex sensitivity (BRS) was determined using the alpha index, which derives BRS in ms/mmHg from the square root of the ratio between the power of simultaneous spectral analyses of spontaneous variability in RRIs and SBP for coherence above 0.5, i.e., a significantly stable phase relation between the two oscillations.

Metronomic deep breathing induces sinus arrhythmia mostly influenced by vagal outflow. From 2-min segments when breathing was paced at six cpm, we again calculated the RRI-SD, RRI-CV, and RRI-RMSSD. Furthermore, the expiratory–inspiratory ratio (E/I ratio) was determined from the maximum and minimum HR during this maneuver.

Finally, active standing was used to assess HR responses during the initial phase of orthostatic challenge. We calculated the RRI-30/15 ratio—i.e., the ratio between the shortest RRI around the 15th heartbeat and the longest RRI around the 30th heartbeat after standing up—as a measure of baroreceptor reflex-induced HR responses to sympathetic activation.

Autonomic Testing Protocol

Autonomic testing was performed at two distinct time points: (1) within 5 days after implantation (burstMCS Off) and (2) after 4 months of burstMCS. Due to the exploratory nature of this study and the rather small sample size of three patients, data analyses were performed using intra-individual comparison of mean values of the time- and frequency-domain parameters under resting conditions as well as the absolute values of the E/I and RRI-30/15 ratios between the two time points of measurement. Data were compared with healthy age- and sex-matched controls. Furthermore, age-normative values of time-domain parameters of RRI-CV and RRI-RMSSD under resting condition and during deep metronomic breathing, as well as the E/I and RRI-30/15 ratios for autonomic challenge maneuvers, were used to further distinguish between pathological and non-pathological autonomic cardiovascular modulation.

Results

Patient Characteristics and Subjective Pain Scores (VAS) at Baseline and After 4 Months of BurstMCS

Patient 1 (42 years old, female) received neurosurgical intervention after 5 years of chronic pain due to right-sided trigeminal neuralgia. She was co-diagnosed with obesity, depression, fibromyalgia and was regularly co-medicated with pregabalin (450 mg/day), tramadol (37.5 mg/day), paracetamol (332.5 mg/day) and dronabinol (5 mg/day). Prior to MCS implantation, microvascular decompression, stereotactic radiosurgery (70 Gy), thermocoagulation of the trigeminal ganglion and repetitive botulinum toxin injections failed to achieve a sustained pain suppression and was described as a combination of burning sensations combined with sharp attacks in the second and third branch of the trigeminal nerve. According to the VAS, the intensity of pain was classified as 9/10 at baseline (5 days after burstMCS implantation/Off) and at time point 2 of measurement, i.e., after 4 months of adjunctive burstMCS declined to 8/10 (10% reduction).

Patient 2 (64 years old, female) suffered from chronic facial pain for 10 years due to trigeminal neuralgia characterized by a constant sharp burning pain along with attacks occurring 25 days per month in the distribution area of the second and third branch of the trigeminal nerve. Other prior comorbidities were depression, polyneuropathy, and chronic vertigo. The pre-MCS pharmacotherapies consisted of carbamazepine (1100 mg/day) and mirtazapine (15 mg/day). Adjunctive burstMCS improved the patient’s pain levels after 4 months, as quantified by the VAS compared to baseline (VAS 10/10 versus 6/10; 40% reduction).

Patient 3 (73 years old, male) received burstMCS after a 10-year history of chronic central pain due to posterior cerebral artery stroke described as constant burning pain affecting the right-sided upper/lower extremities. Other pre-existing conditions were myocardial infarction, arterial hypertension, hypercholesterolemia, and obesity. Long-term medication consisted of torasemide (10 mg/day), β-blocker (95 mg/day), ACE inhibitor (10 mg/day), statins (30 mg/day), allopurinol (300 mg/day), tapentadol (100 mg/day), amitriptyline (50 mg/day), and antidepressants (citalopram; 20 mg/day). The most pronounced decline of pain intensity was observed in patient 3 and decreased markedly with adjunctive burstMCS compared to baseline (arm pain: baseline VAS 10/10 versus 3/10; leg pain: VAS 9/10 to VAS 3/10).

In all three patients, unchanged dosage of the pre-MCS pain medication were recorded within 4 months of burstMCS. No implantation and/or stimulation associated adverse events occurred nor did we observe burstMCS-evoked seizures (Table 1).

Autonomic Cardiovascular Modulation at Resting Conditions

RRI and BP values did not differ significantly between baseline and after 4 months burstMCS [patient 1: RRI 678.5 vs. 742.1 ms, SBP 147.8 vs. 155.7 mmHg, DBP 96.8 vs. 99.7 mmHg; patient 2: RRI 569.4 vs. 649.8 ms, SBP 158.4 vs. 144.2 mmHg, DBP 96.4 vs. 91.1 mmHg and patient 3: RRI 677.6 vs. 638.2 ms, SBP 110.1 vs. 122.5 mmHg, DBP 71.7 vs. 79.9 mmHg]. The RRI-CV and RRI-RMSSD were within the normative range at both time points in patients 1 and 2, while they were pathologically low in patient 3 at both time points of measurement [patient 1: RRI-CV 2.0 vs. 2.6/RRI-RMSSD 8.8 vs. 15.6 ms; patient 2: RRI-CV 2.7 vs. 4.9/RRI-RMSSD 9.2 vs. 25.2 ms; patient 3: RRI-CV 2.7 vs. 4.9%/RRI-RMSSD 5.1 vs. 3.2 ms]. BRS at baseline vs. 4 months burstMCS for patient 1 was 1.8 vs. 2.9 ms/mmHg, for patient 2 was 1.0 vs. 2.4 ms/mmHg, and for patient 3 was 2.9 vs. 1.5 ms/mmHg.

Values of frequency-domain parameters at baseline vs. burstMCS were as follows: for patient 1, RRI-total power 167.4 vs. 346.5 ms², RRI-LFnu power 0.67 vs. 0.50, RRI-HFnu power 0.33 vs. 0.50, and RRI-LF/HF ratio 2.0 vs. 1.0; for patient 2, RRI-total power 190.7 vs. 861.8 ms², RRI-LFnu power 0.73 vs. 0.66, RRI-HFnu power 0.27 vs. 0.34, and RRI-LF/HF ratio 2.4 vs. 2.3; and for patient 3, RRI-total power 29.8 vs. 10.1 ms², RRI-LFnu power 0.87 vs. 0.69, RRI-HFnu power 0.12 vs. 0.31, and RRI-LF/HF ratio 5.9 vs. 2.4. The mean values of autonomic cardiovascular parameters between time points increased in all parameters of total autonomic modulation (RRI-total power 129.3 vs. 406.2 ms²/Hz, RRI-SD 11.6 vs. 18.5 ms and RRI-CV 1.9 vs. 3.7) and in parameters of vagal cardiac modulation (RRI-HFnu power 0.24 vs. 0.38 and RRI-RMSSD 7.7 vs. 14.7 ms). In contrast, the parameters of sympathetic cardiac modulation (RRI-LFnu power 0.76 vs. 0.62) and sympathovagal balance (RRI-LF/HF ratio 3.4 vs. 1.9) decreased. Furthermore, BRS also increased (BRS 1.9 vs. 2.3 ms/mmHg) from baseline to 4 months after adjunctive burstMCS.

Nevertheless, compared to three healthy age- and sex-matched controls, at the second time point of measurement parameters of total autonomic modulation (e.g. RRI-total power 406.2 vs. 877.2 ms²), vagal cardiovascular modulation (e.g. RRI-RMSSD 14.7 vs. 35.4 ms), and baroreceptor reflex sensitivity (BRS 2.3 vs. 7.5 ms/mmHg) revealed lower values while parameters of sympathetic cardiovascular modulation (e.g. RRI-LFnu power 0.62 vs. 0.43) and sympathovagal balance (RRI-LF/HF ratio 1.9. vs. 1.2) demonstrated higher values in our patient cohort (Table 2).

Table 2.

Parameters of autonomic cardiovascular modulation in controls and patients at both measurement time points before and after 4 months adjunctive burstMCS

	First measurement	Second measurement	Controls
Sympathetic parameter
RRI-LFnu power	0.76	0.62	0.43
SBP-LF power (mmHg2)	17.6	20.4	12.4
Parasympathetic parameter
RRI-HFnu power	0.24	0.38	0.57
RRI-RMSSD (ms)	7.7	14.7	35.4
Total autonomic parameter
RRI-total power (ms2)	129.3	406.2	877.2
RRI-SD (ms)	11.6	18.5	32.3
RRI-CV	1.9	3.7	3.9
Sympathovagal balance
RRI-LF/HF ratio	3.4	1.9	1.2
Baroreceptor reflex sensitivity (ms/mmHg)	1.9	2.3	7.5

Autonomic Cardiovascular Modulation During Metronomic Deep Breathing

Baseline and follow-up E/I ratios for patient 1 were 1.10 vs. 1.17, for patient 2 were 1.10 vs. 1.27, and for patient 3, 1.02 vs. 1.03. The mean value of the E/I ratio increased between time points from 1.01 to 1.16. According to age-normative values, two patients had an E/I ratio below the normative range at measurement 1, while the E/I ratio at measurement 2 was still pathological only in one patient. Additionally, RRI-RMSSD increased in two patients from pathological to non-pathological values (patient 1 8.8 vs. 15.6 ms and patient 2 9.2 vs. 25.2 ms) while it decreased in patient 3 and remained in a pathological range. Finally, RRI-CV was non-pathological at both time points in patient 1 (2.0 vs. 2.6%), changed from pathological to non-pathological in patient 2 (2.7 vs. 4.9%) and stayed in the pathological range at both time points in patient 3 (0.9 vs. 0.6%), as summarized in Table 3.

Table 3.

Parameters with comparable age-normative values during deep metronomic breathing at both measurement time points: E/I ratio (A) and RRI-RMSSD and RRI-CV (B) before and after 4 months adjunctive burstMCS

(A)	First measurement	Second measurement
E/I-ratio
Patient 1	1.10*	1.17
Patient 2	1.10	1.27
Patient 3	1.02*	1.03*

(B)	First measurement		Second measurement
	RRI-RMSSD	RRI-CV	RRI-RMSSD	RRI-CV
Patient 1	8.8*	2.0	15.6	2.6
Patient 2	9.2*	2.7*	25.2	4.9
Patient 3	5.1*	0.9*	3.2*	0.6*

*indicate pathological values

Autonomic Cardiovascular Modulation During Active Standing

Between the two measurements, the range from RRI minimum to maximum in patient 1 was 81.8 vs. 85.2 ms, in patient 2 was 59.2 vs. 48.6 ms, and in patient 3, 22.2 vs. 25.2 ms, while the maximum decrease in SBP/DBP demonstrated the following pattern: patient 1: 47.5/28.1 mmHg vs. 36.7/19.7 mmHg; patient 2: 49.9/23.2 vs. 53.3/23.6 mmHg; patient 3: 32.7/24.8 vs. 30.4/18.9 mmHg. At measurement 1 vs. 2, the 30/15 ratios for patient 1 displayed 1.15 vs. 1.16, for patient 2 0.98 vs. 0.98, and for patient 3 1.01. vs. 1.02, while the mean values of the 30/15 ratio were 1.05 and did not differ between baseline and 4 months adjunctive burstMCS (Table 4).

Table 4.

Parameters with comparable age-normative values during active standing at both measurement time points: RRI-30/15 ratio before and after 4 months adjunctive burstMCS

	First Measurement	Second Measurement
RRI-30/15 ratio
Patient 1	1.15	1.16
Patient 2	0.98*	0.98*
Patient 3	1.01	1.02

*indicate pathological values

Discussion

Summary of Our Findings and Comparison with BurstMCS in Human Studies for Chronic Pain

We herein report preliminary data assessing the influence of burstMCS waveforms on sympathetic and parasympathetic tone in chronic pain patients. Our findings indicate that burstMCS not only improved pain levels but also evoked changes in the autonomous nervous system relevant for cardiovascular modulation which is well known to be dysfunctional towards an increased sympathetic and decreased parasympathetic state in chronic pain disorders. Although we report results stemming from short-term observation period, no implantation and stimulation related adverse events occurred. Notably, medication relevant for pain and hemodynamics remained unchanged throughout 4 months of adjunctive burstMCS. Contrary to most MCS studies, we entirely applied burst mode driven MCS patterns, which have been under investigation only in a few pain patients so far [12, 26]. Despite transcranial magnetic stimulation (TMS) and pharmacological agents (opioids, ketamine), no data exist evidently providing pre-implantation factors to predict MCS responsiveness. In our small scale cohort study noninvasive TMS was not performed prior to burstMCS, nor did we include a MCS test stimulation with externalized leads, as MCS trial has been a subject of controversy [12].

Certainly, our findings need to be re-examined in a larger sample under a randomized controlled study protocol which might help to better understand the effects reported in our feasibility trial. Along these lines, sham stimulation should be integrated to account for confounding variables by means of the placebo effect [34, 35].

Dysfunctional Autonomic Cardiovascular Modulation in Chronic Pain

Pain perception and processing within the peripheral and central nervous systems is closely linked to the modulation of the autonomic nervous system. From a neuroanatomical perspective, many signaling pathways are similar for the processing of nociceptive and viscerosensory information, and the cortical integration of both systems often takes place in the same brain regions (Fig. 3, left side). In the clinical context, there is also a close link between pain perception and autonomic outflow, as nociceptive stimulation leads to sympathetic activation, such as with increases in blood pressure and heart rate [36, 37]. The nociceptive Aδ- and C-fibers first transmit their information to the spinal dorsal horn and from there via the tractus spinothalamicus to the contralateral thalamus. This tractus strongly projects to the sympathetic thoracolumbar system of the spinal cord and thus forms spinospinal loops as the basis of somato- and viscero-autonomous reflexes. Via spinoreticular and spinomesencephalic pathways, connections are formed with two autonomic brainstem regions, namely the bulbopontine and the pontomedullary parts, for modulation of pain perception and behavioral control. Furthermore, a connection for regulating homeostasis is also established via projections into and from the hypothalamus. Finally, different thalamic nuclei project integrative control areas of the forebrain (i) via the nucleus ventrolateralis into the primary (S1) and secondary (S2) somatosensory cortex for precise pain localization, and (ii) via the nucleus ventromedialis posterior to the insular cortex for body sensation awareness. A third pathway projects through the dorsomedial and intralaminar nuclei of the thalamus to the anterior cingulate cortex (ACC), modulating emotional components of pain perception and behavioral change. In conclusion, these areas as well as parts of the amygdala and prefrontal cortex represent the so-called pain matrix, i.e. areas closely related for processing nociceptive and autonomic information [36].

Fig. 3.

Afferent autonomic and nociceptive pathways and alterations in chronic pain. Left side: nociceptive information enters the thalamus via peripheral Aδ-/C-fibers and the spinothalamic tract. From there, pathways travel to the primary (S1) and secondary (S2) somatosensory cortex, the formatio reticularis of the brainstem, the hypothalamus (HT) and the limbic system. Along the way from stimulus reception to cortical areas, several connections are made with the autonomic nervous system on different levels: connections with the sympathetic thoracolumbar system form the basis of spinospinal loops and a strong connectivity with the autonomic centers of the bulbopontine (BP) and pontomedullary (PM) levels of the brainstem are established through the spinoreticular and spinomesencephalic tracts. Finally, processing of nociceptive information and integration in areas for autonomic control takes place in the forebrain. This includes the insular cortex (I), amygdala (A), the anterior cingulate cortex (ACC) and the prefrontal cortex (PFC) to ensure cognitive and emotional interpretation for regulation of homeostasis and behavioral modulation. Right side: as a result of chronic pain, structural and functional changes are evoked at all peripheral and central levels including alterations in corticolimbic structures, changes in pain modulation of the brainstem and central and peripheral sensitization

Pain signaling can be regulated by a descending pain modulating system: the amygdala, parts of the stria terminalis, the hypothalamus and especially the PAG control autonomic, endocrine, and motor outputs. The PAG itself represents a critical interface within this system; it exerts dual control, with both inhibitory and excitatory effects on pain perception via tonic control of noradrenergic and serotonergic pathways and the release of endogenous opioids. The processing and conscious perception of nociceptive information is significantly influenced by various parameters such as attention, emotion, mood or expectation of pain [37, 38]. However, in addition to the context, the duration of the stimulation and the associated sympathetic modulation also seem relevant in this context: short-term activation of the SNS physiologically suppresses pain in healthy individuals, while long-term activation amplifies pain. Functional MRI studies have shown that acute nociceptive stimulation, i.e. normal pain stimulation, induces activation of the contralateral thalamus, S1, S2, insula, dorsal ACC, and dorsolateral prefrontal cortex. In patients with chronic somatic pain, activation of the contralateral thalamus is reduced. Instead, there is bilateral activation of the insula (involved in awareness of body sensation), the dorsal ACC (involved in behavioral arousal) and the prefrontal cortex (involved in attention mechanisms), suggesting that cognitive and emotional factors strongly influence chronic pain [37–39].

In addition to a central reorganization of the pain matrix, chronic pain induces dysfunctional pain modulation via sensitization of peripheral nociceptors (including increased expression of α-1 adrenoreceptors), and enhanced sympathetic sprouting in spinal ganglia of the dorsal root as well as conditioning (Fig. 3, right side). According to these close anatomical links between nociceptive and autonomic processing, measurement of HR- and BP-variability seems intuitive to assess autonomic networks in chronic pain. Because heart rate variability (HRV) correlates with the resting state connectivity of the medial prefrontal cortex and the functional connectivity of the dorsal ACC to the thalamus and brainstem, enhanced spectral power of the HF component has been associated with pain-suppressive effects [39]. In contrast, reduced RRI-HF along with increased RRI-LF and RRI-LF/HF ratios and reduced baroreceptor reflex sensitivity have been suggested in chronic pain syndromes, such as in fibromyalgia [37]. In line with these findings, our patients had higher spectral power in the LF-range, indicating enhanced sympathetic outflow, and deteriorated vagal modulation, i.e. lower HF power compared to healthy controls.

Although bio-signals (HR, BP) did not differ meaningfully between groups, subtle changes in autonomic cardiovascular modulation as demonstrated by altered HR and BP variability testing has been suggested to negatively impact outcome, and correlate with higher mortality rates in various diseases, such as ischemic stroke [27]. In line with altered spectral power of autonomic modulation, we further found impaired HRV values in the time-domain analysis—i.e. a test method with existing age-normative values. Therefore, our results suggest impaired autonomic modulation etiologically linked to an altered pain matrix in our chronic pain patients. On the other hand, changes in autonomic cardiovascular modulation may be associated with therapeutic effects in pain patients. In patients with trigeminal neuralgia, increased RRI-LF and reduced RRI-HF have been found during pain attacks, and the presence of autonomic symptoms has been suggested to correlate with lower success rates of pain reduction after microvascular decompression [37]. In contrast to these results, we found improvement of cardiovascular autonomic modulation, with a decrease in sympathetic tone and an increase in vagal outflow after only 4 months of burstMCS. Along with improvement in subjective pain perception in our three patients, we therefore suggest that HR and BP variability testing may offer a new approach to objectively monitor therapeutic effects after neurosurgical treatment in patients with chronic pain.

Limitations and Future Directions

Because of the small sample size, we cannot derive a clear pattern of correlation between declining VAS and sympathetic tone. While patients 1 and 2 showed moderate changes in both pain perception and autonomic modulation after 4 months of stimulation, patient 3 reported a massive reduction in VAS despite minimal changes in autonomic control. This may be explained by the various factors influencing autonomic modulation. These relevant influencing factors include sex and age as well as comorbidities (especially with involvement of the cardiovascular system) and previous medication (mainly antihypertensives, antiarrhythmics and various analgesics, centrally acting opioids) [40, 41]. In addition to chronic pain, patient 3 for example suffers from severe cardiovascular disease, which has previously been associated with autonomic dysfunction [27]. In order to be able to assess confounding variables more precisely, a larger cohort of patients is needed.

Since the influence of the autonomic nervous system on immune-inflammatory processes is also a relevant factor for the increased morbidity and mortality in chronic pain, a closer look at the pathologically increased sympathetic tone and reduced vagal tone—including the cholinergic anti-inflammatory pathway—would be of great importance for a better understanding of disease development and maintenance in chronic pain [42, 43].

In order to quantify the different dimensions of pain as a complex multisystem disorder, other subjective surveys such as assessments of mood and sleep might be useful in addition to the subjective pain perception determined by VAS.

Furthermore, evaluation of the adjunctive effect of MCS on different classes of analgesic drugs should be addressed in larger validation studies, since different classes of analgesics may be more suitable for combination with stimulation procedures for pain therapy such as MCS.

Finally, the impact of different stimulation paradigms, for instance, DBS, SCS, TMS, or VNS. on autonomic nervous system modulation may be part of future research, as so far there are no studies investigating autonomic cardiovascular modulation under different stimulation procedures particularly in chronic pain patients [44, 45].

Although correlations may therefore not be derived so far, our results suggest promising hints that may guide clinical practice and surveillance after validation in larger cohorts.

Conclusions

In summary, we found a meaningful clinical response related to the chronic pain disorders treated with MCS quantified by subjective score-based assessment along with objective autonomic outcome measures after 4 months of adjunctive burstMCS.

This is the first study investigating the relationship between burstMCS and changes in cardiovascular modulation in patients with chronic pain. Obviously, several biases limit the generalization of our findings; however, this does not limit the worthiness of our approach to explore objective outcome parameters relevant for interventional pain therapy. Further investigations are needed to determine the usefulness of HR and BP variability in patients with chronic pain as a simple, noninvasive, and objectifiable tool to assess treatment effects.