Innovations in spinal cord stimulation for pain

Abstract

Spinal cord stimulation (SCS) is a neuromodulation therapy used to treat medically refractory chronic pain. In SCS, an implanted pulse generator produces electrical signals that are conveyed through electrode arrays located in the region of the spinal cord. The goal of SCS is to modulate neural signaling through spinal and supraspinal mechanisms to reduce pain. Although available for decades, SCS still enjoys only limited clinical success, limited quality-of-life improvement, and limited long-term efficacy. To improve SCS outcomes, advances in lead design, stimulator features, and waveform paradigms have been recently introduced. While it is an exciting time for the neuromodulation field, empirical SCS advances have surpassed scientific understanding of SCS mechanisms of action. We still do not know why SCS works in some patients but not in others. We also lack information-rich biomarkers of pain and pain relief through which to optimize SCS programming. To optimize both system designs and clinical implementations of SCS, it is critical that we address these scientific and mechanistic knowledge gaps.

Introduction

Spinal cord stimulation (SCS) is a pain management therapy approved by the United States Food and Drug Administration (FDA) for a primary indication of neuropathic limb pain that is refractory to conventional medical management (e.g. failed back surgery syndrome or complex regional pain syndrome) [1,2]. SCS has also proven effective for visceral pain, including refractory angina, and peripheral vascular disease [3]. It is estimated that as many as 50,000 spinal cord stimulators are implanted annually and the clinical demand for SCS continues to increase [4].

SCS implantation typically involves a two-stage procedure. First, patients undergo a short-duration (e.g. 3–10 days) trial phase in which electrode arrays are implanted in the epidural space dorsal to the spinal cord a few levels above the affected spinal segments (e.g. lower thoracic levels for lower limb pain, cervical levels for upper limb pain) and connected to an external trial stimulator. This procedure can be performed on an outpatient basis under local anesthesia with cylindrical electrode arrays that are implanted percutaneously through a Touhy needle. The goal of the trial phase is to determine the likelihood that a patient will receive sufficient pain relief from SCS. During this trial phase, several stimulation parameters (e.g. amplitude, pulse width, frequency, stimulation configuration) are tested in an attempt to maximize pain-paresthesia overlap. If a patient achieves sufficient pain relief (e.g. ≥ 50%), they can proceed to the second stage in which electrode array(s) are permanently implanted and connected to an implanted pulse generator (IPG) (most commonly placed in the posterior hip area) (Figure 1). SCS can also utilize paddle- or plate-style electrode arrays in which a surgeon performs a laminotomy for implantation. Paddle leads have potential advantages over percutaneous leads (e.g. decreased lead migration, stable configuration of the stimulation array, unidirectional stimulation) at the cost of being more invasive [5].

Figure 1.

Spinal cord stimulation (SCS) for pain. In conventional SCS, an electrode array(s) implanted in the epidural space applies electrical pulses at a moderate frequency (e.g. 40–60 Hz) to excite large-diameter Aβ somatosensory fibers in the dorsal columns. Antidromic action potential propagation is believed to inhibit output of projection neurons in the dorsal horn via activation of inhibitory interneurons that close the “gate” and prevent the transmission of pain signals to the brain.

Conventional SCS utilizes tonic stimulation at amplitudes that produce a perceptible paresthesia in the region of pain and has been a widely-available therapy for decades. Despite improvements in lead design and advances in IPG technology, the outcomes of conventional SCS have been characterized by limited success rates, limited improvements in patient quality of life, and a decline in efficacy over the long term [6,7]. However, over the past 5 years, there have been dramatic advancements in SCS technology that include novel stimulation waveforms, further innovations in lead design, and substantial advances in IPG capabilities. These advancements have heralded an exciting time in the field for clinicians. At the same time, these new technologies have advanced empirically, and clinical results have not been accompanied by new scientific understanding of the mechanisms of action of SCS. Yet, improved scientific understanding of conventional and novel forms of SCS is necessary to drive patient selection, to optimally select stimulation patterns, and to individualize the therapy. The purpose of this review is to summarize recent advancements in SCS and to describe the knowledge gaps that merit further careful study.

Innovations in stimulation waveform

Conventional SCS

SCS was first tested in patients in 1967, two years after Melzack and Wall published the gate control theory of pain [8,9]. Conventional SCS involves tonic electrical stimulation at a moderate frequency (e.g. 40–60 Hz) with the goal of exciting large-diameter (Aβ) sensory afferents in the dorsal columns to create paresthesias (e.g. tingling, buzzing, pin and needles, pressure) over the painful areas (Figure 2A) [10]. According to the gate-control theory, some of the Aβ fibers have collaterals that project to the affected spinal levels. SCS results in antidromic activation of the Aβ fibers that in turn (via their collaterals) lead to inhibition of nociceptive-specific or wide dynamic range projection neurons in the dorsal horn (Figure 1). In neuropathic pain conditions, excessive nociceptive inputs are believed to produce hyperexcitability in these projection neurons (i.e. central sensitization). Antidromic activation of Aβ fibers from SCS is hypothesized to inhibit output from these projection neurons largely through the activation of inhibitory interneurons that help close the “gate” and prevent the transmission of pain signals to the brain [11]. There is additional experimental evidence that orthodromic Aβ axonal activation may also help reduce pain via supraspinal mechanisms, such as descending inhibition [12].

Figure 2.

Spinal cord stimulation (SCS) waveform paradigms. A) Conventional SCS applies short-duration pulses in a frequency range of 40–60 Hz. B) Burst SCS is a new type of stimulation that applies bursts of stimulation at a burst-rate of 40 Hz, intra-burst frequency of 500 Hz, and five pulses per burst. Each pulse has a pulse width of 1 ms and an inter-pulse interval of 1 ms. C) Another novel form of SCS applies 30-μs pulses at a rate of 10 kHz to provide pain relief without concomitant paresthesias.

Stimulator programming involves assessing a variety of parameter combinations (e.g. amplitude, pulse width, frequency, stimulation configuration) with the goal of maximizing the overlap of SCS-induced paresthesias with a patient’s painful areas. Clinical experience suggests that sufficient pain-paresthesia overlap increases the likelihood for pain relief [13]. It is important to note, that SCS-induced pain relief occurs over a time course of seconds to several hours. However, SCS-induced paresthesias occur rapidly (~1 s) and serve as a surrogate for pain relief that make it practical to test several sets of stimulation parameters within a standard clinical visit. Although widely used for decades, the success rate of SCS, defined as the proportion of patients receiving ≥ 50% pain relief, is approximately 58% [6].

Burst-SCS

In an attempt to improve the clinical efficacy of SCS, several novel waveform paradigms have emerged on the clinical market within the last several years. Burst-SCS is one of these new approaches that delivers intermittent bursts of electrical pulses (five pulses at 500 Hz, delivered 40 times per second) to mimic thalamic bursting within the nervous system (Figure 2B) [14]. Burst-SCS demonstrates two potential advantages over conventional low-frequency SCS: 1) improved pain relief, and 2) pain relief without concomitant paresthesias. Paresthesia-free pain relief can be extremely beneficial because SCS-induced paresthesias may disturb sleep, be experienced as excessive or uncomfortable, and vary with body position [15].

Burst-SCS was approved by the FDA in 2016 based on the results of a large multi-center clinical trial that reported Burst-SCS to be significantly more effective than conventional SCS (60% v. 51% success rate) [16]. Preliminary evidence suggests that Burst-SCS modulates both ascending pain-evoking and descending pain-inhibitory pathways involved in pain processing and may function through different mechanisms than conventional SCS [12,17]. However, these potential mechanisms of action require validation with additional studies.

Kilohertz-frequency SCS (KHFSCS)

KHFSCS is another novel form of SCS that applies tonic stimulation pulses at a rate > 1 kHz. In 2015, the FDA approved a KHFSCS system that applies stimulation at a rate of 10 kHz and provided dramatic pain relief (~80%) without generating paresthesias (Figure 2C) [18]. The pain-relief mechanisms of KHFSCS are currently unknown. Kilohertz stimulation frequencies have shown the ability to generate rapid and reversible conduction block in peripheral nerve models [19]. This idea of conduction block was a driving force behind the development of KHFSCS with the idea that it could block the propagation of painful signals to the brain. However, theoretical and experimental data suggest that conduction block is unlikely at the low stimulation amplitudes used clinically (i.e. 0.5–5.0 mA) [20–23]. These studies also suggest that the direct activation of dorsal column axons is also unlikely with clinical KHFSCS. A recent study examined the paresthesias generated by low-frequency stimulation in a group of patients that had been receiving KHFSCS as part of their standard clinical care [24]. In these KHFSCS patients, the paresthesias generated by low-frequency stimulation (utilizing the same electrodes as with KHFSCS) did not overlap with the patients’ painful areas. This data suggested that pain-paresthesia overlap is not necessary and that KHFSCS may be functioning through different mechanisms of action relative to low-frequency SCS. Several additional mechanisms of action of KHFSCS have also been presented, such as asynchronous activation, desynchronization of clusters of neurons firing in synchrony, reduced “wind-up” and suppression of spontaneous activity in dorsal horn cells, and excitability changes from possible tissue heating due to the increased power deposited in tissue during KHFSCS [12,25,26].

While KHFSCS is an exciting new approach, there are several unanswered questions and limitations with current clinical implementation of this technology. For example, it is not clear what stimulation rate in the kilohertz frequency range provides the optimal pain relief. A recent clinical study demonstrated equal pain relief at several frequencies in the kilohertz range (i.e. 1, 4, 7, and 10 kHz) and stimulation at a lower frequency (e.g. 1 kHz) may provide equivalent pain relief at lower energy demands [27]. Additionally, although it can be advantageous to provide pain relief without concomitant paresthesias, these rapid-onset paresthesias serve as a surrogate to select clinically-effective stimulation parameters with paresthesia-based low-frequency SCS. With paresthesia-free KHFSCS, parameter efficacy can only be determined by patient-reported pain relief after several hours of stimulation.

Innovations in lead design

High-density electrode arrays

The efficacy of SCS is dependent on the ability to stimulate the desired neural targets within the spinal cord. Therefore, several technical innovations have focused on improving the lead design, such as increasing the number of electrodes. A higher electrode count theoretically allows users to shape the electric field to optimize the neural selectivity by strategic selection of anodes and cathodes. For conventional SCS, increased electrode number improves the ability to select stimulation parameters that produce sufficient pain-paresthesia overlap and decreases the likelihood of revision surgery necessary to maintain concordant paresthesia [13,28,29]. For several years clinical arrays included four electrodes (Figure 3A) [13,30,31]. However, cylindrical percutaneous arrays now include 8–16 electrodes and paddle electrodes now include 16–32 electrodes and up to five columns of electrodes (Figure 3B) [3,10]. A higher number of electrodes along the length of a lead increases the number of spinal levels that can be targeted with stimulation and allows for stimulation configurations that improve targeting of the dorsal columns over the dorsal roots (e.g. longitudinal guarded cathode). Paddle electrode arrays also include 2–5 columns of electrodes that theoretically improve the mediolateral resolution of the stimulation to improve the ability to target the desired dermatomes within the spinal cord.

Figure 3.

Spinal cord stimulation (SCS) lead designs. A) Older generations of percutaneous and paddle arrays typically included four electrodes. B) To help improve stimulation selectivity and flexibility, newer generations of percutaneous arrays include 8–16 electrodes and paddle arrays include 16–32 electrodes.

Flexible lead arrays

Clinical implementation of SCS originally utilized intrathecal or subdural stimulation, but epidural stimulation eventually became the preferred method due to potential complications of subdural SCS (e.g. cerebrospinal fluid leakage, acute neurological deficit) [10]. Recently, there has been renewed interest in subdural SCS and several groups are also attempting to develop flexible SCS leads to help prevent deformation and damage to the spinal cord [32–34]. These flexible arrays could be implanted directly on or near the spinal cord and would allow for more direct targeting of the spinal cord at lower stimulation amplitudes.

Dorsal root stimulation

SCS often has a limited ability to target focal painful areas that can be common in certain pain conditions, such as complex regional pain syndrome. SCS precision is often limited by current shunting in the cerebrospinal fluid, variations in stimulation levels due to postural changes, and lead migration [35]. Dorsal root stimulation is a low-frequency (~10–50 Hz), paresthesia-based alternative to SCS that can provide more focal stimulation effects and involves implantation of cylindrical electrodes implanted in the epidural space in close proximity to the dorsal root ganglia (DRG) [36]. DRGs contain the cell bodies of primary sensory neurons and play a key role in the development and maintenance of chronic pain [37]. Due to the small epidural space in the neural foramen in which the DRG lies, lead stability is increased and DRG stimulation demonstrates less postural variation in paresthesia and reduced lead migration [36]. Because stimulation is targeted at individual DRG and the corresponding dermatomes at that spinal level, extraneous stimulation in nonpainful areas is reduced with DRG stimulation. DRG stimulation also typically employs lower stimulation amplitudes (< 1 mA) because the electrodes are placed close to the DRG [36]. DRG stimulation was approved by the FDA in 2016 for the management of chronic intractable pain of the lower limbs in adults with CRPS. A large multicenter clinical trial showed a greater response rate (i.e. ≥ 50% pain relief at baseline) with DRG stimulation (81.2%) compared to conventional SCS (55.7%) at 3 months [36].

Innovations in pulse-generator capabilities

Voltage and current regulation

Some commercial SCS systems utilize voltage-controlled stimulation. However, large variations in electrode impedance can be observed across electrodes within an individual array or across patients [10]. For voltage-controlled stimulation, impedance variability can affect the extracellular voltages generated within the spinal cord and may require the adjustment of stimulation parameters to improve efficacy [38]. However, an increasing number of commercial systems apply current-controlled stimulation that may reduce the effects of electrode impedance variability on extracellular voltages generated within the central nervous system [3].

Multiple-source systems

Many commercial stimulators utilize a single-source system. While a user can select a variety of complex stimulation configurations, electrodes are either cathodes, anodes, or inactive. However, some commercial systems utilize multi-source systems in which the relative fractions of the total amplitude can be adjusted between individual electrodes to improve flexibility in shaping the electric field. This flexibility is believed to improve targeting of the desired areas within the spinal cord [10,39].

Reduced invasiveness

Several advancements have also reduced the invasiveness of SCS systems by reducing the footprint of the IPG. These technologies include the development of rechargeable systems that improve battery longevity (~9–25 years) and allow for a significant reduction in IPG size [35,40,41]. Wireless SCS systems have also been developed that include a passive electrode array implanted in the epidural space that contains a microprocessor receiver and an antenna. An external transmitting antenna and pulse generator are worn by the patient and transcutaneously transmit the stimulation parameters and power to the implanted array to stimulate the spinal cord [42].

Magnetic resonance imaging (MRI) compatibility

Chronic pain patients and candidates for SCS are likely to have comorbidities and ongoing pain issues that require diagnostic imaging. MRI is the gold standard for clinical evaluation and diagnosis of many disease states, including chronic pain. It is estimated that 82–84% of SCS patients are expected to need at least 1 MRI within 5 years post implant [43]. However, due to potential interactions between the MRI field and the implanted electronic devices, SCS patients have historically been excluded from MRI [43,44]. Over the last several years, there has been an improvement in MRI compatibility of SCS systems. There are now a number of commercial SCS systems that are conditionally safe for MRI [45].

Multiwave and updateable platforms

As discussed previously, in addition to standard SCS waveforms, several novel stimulation waveform paradigms (e.g. Burst-SCS, KHFSCS) are now used clinically. Some IPGs now have the capability to provide a wide range of stimulation parameters and paradigms that allow the user to select the waveform paradigms that provides optimal pain relief [46]. Multiple paradigms can be interleaved and/or applied at the same time to target different types of pain and/or painful areas. If novel SCS parameters are shown to potentially improve SCS-induced pain relief, some commercial systems can also be upgraded to implement these new stimulation parameters [47].

Closed-loop SCS

SCS is most often implemented as an open-loop system. Patients provide feedback with regard to stimulation-induced paresthesias, pain relief, and discomfort to help clinicians select optimal stimulation parameters. However, once optimal stimulation parameters are selected, these parameters largely remained unchanged. The clinical efficacy of a given set of parameters can vary with movement of the spinal cord due to changes in body position as well as respiration and heartbeat. It is estimated that the spinal cord can move ~2–3 mm in the anterior-posterior direction at the lower thoracic spinal levels and changes the perception threshold to SCS [48,49]. This movement alters the distance between the stimulating electrodes and the spinal cord and can lead to overstimulation or understimulation. Therefore, this movement can result in a given set of stimulation parameters being efficacious in one body position, but lead to mild or intense discomfort and/or reduced pain relief in another position. These movement-related side effects can also prevent patients from participating in routine daily activities, such as driving.

To improve the efficacy and stability of SCS, one commercial SCS system utilizes a 3-axis accelerometer contained within the body of the IPG to sense body position [15,50]. The IPG can be programmed for up to six body positions and can automatically adjust the stimulation parameters based on IPG orientation. In a multicenter randomized trial, position-adaptive SCS provided 86.5% of patients with improved pain relief with no loss of convenience or improved convenience with no loss of pain relief compared to manual programming only [50].

Another clinical SCS system under development uses inactive electrodes in the implanted arrays to record evoked compound action potentials (ECAPs) generated during SCS (Figure 4) [51]. These ECAPs reflect the summation of individual action potentials generated by an SCS pulse and provide a quantitative measure of neural recruitment in the spinal cord. A clinical study suggested that these ECAPs likely corresponded to the activation of somatosensory Aβ axons in the dorsal column of the spinal cord [52]. The ECAP amplitude was dependent on body position. Increased stimulation amplitude also produced an increase in ECAP amplitude that was correlated with the degree of paresthesia coverage. Therefore, this clinical SCS system uses ECAP amplitude as a control signal to continuously define the optimal stimulation parameters. To implement this closed-loop control, the user defines a reference ECAP amplitude that is recorded with stimulation parameters that the patient reports as comfortable and providing optimal pain relief. The IPG then utilizes a feedback controller to continuously adjust the stimulation amplitude so that the recorded ECAP amplitude says within a therapeutic window in spite of changes in body position and activity to maintain a constant level of neural recruitment (Figure 4B). The therapeutic potential of this device was demonstrated in a recent open-label uncontrolled clinical study [51].

Figure 4.

Closed-loop spinal cord stimulation (SCS). A) SCS pulses generate action potentials that travel both antidromically and orthodromically along the spinal cord. These evoked compound action potentials (ECAPs) can be recorded using inactive electrodes in the SCS array. The traces overlaid on the spinal cord illustrate an SCS-induced action potential traveling along the individual axon shown in black. Model-based bipolar ECAP recordings are shown on the right (the stimulus artifact is clipped). In this figure, a stimulus pulse was applied through C7 and bipolar recordings were performed with electrodes C6-C1 with respect to the reference electrode, C0. The stimulus onset is indicated by the dashed white lines. B) The ECAP amplitudes (e.g. P2N1) can serve as a control signal related to SCS efficacy. Movement of the spinal cord with respect to the stimulation electrode(s) can lead to sub-therapeutic stimulation or discomfort. The stimulation amplitude can be automatically adjusted using the ECAP amplitudes to maintain stimulation within a therapeutic range.

Discussion

It is an exciting time in the field of SCS and neurostimulation for pain. Over the last several years, there have been dramatic technological advancements to develop novel approaches that may improve the clinical efficacy of these neurostimulation therapies. These advancements include novel waveform paradigms, such as Burst-SCS and KHFSCS, innovations in lead design to increase the number of electrode contacts and improve the lead mechanical properties, improvements in IPG capabilities to increase battery lifetime, reduce invasiveness, and provide closed-loop stimulation. These innovations along with decades of clinical experience, have produced potentially powerful therapies to improve the lives of patients suffering from chronic pain. However, these advancements have outpaced our scientific understanding of these technologies. Remaining unanswered questions may limit the impact, optimization, and long-term reliability of these SCS therapies.

From the clinical side, we do not understand why SCS works well in some patients, but fails in others. We do not have clear indicators to reveal which patient may initially respond to SCS but will fail in the long term. The subjective nature of pain and corresponding pain ratings make it difficult to assess the true efficacy of these approaches. Along similar lines, we do not have objective biomarkers or surrogates of pain and pain relief to help optimize the clinical programming procedures, especially with regards to novel SCS waveform paradigms. Finally, how early should we intervene with SCS? SCS is typically only considered after several other treatment options have failed. However, evidence suggests that outcomes may be improved with earlier intervention with SCS [3,53].

From a scientific perspective, what are the mechanisms of action behind SCS-induced analgesia? What are the specific mechanisms of action for tonic SCS, Burst-SCS, and KHFSCS [12]? To answer these questions, we need to improve our knowledge of spinal cord anatomy and physiology, anatomical factors that affect the electric fields generated in the tissue, and how these electric fields translate into physiological and perceptual effects [54]. We also don’t have a clear understanding of how SCS modulates central pain processing or the degree to which placebo effects confound the results of scientific studies [55–57].

To better understand the mechanisms of action of SCS, it is critical to perform systematic, well-powered, randomized, double-blind, placebo-controlled, preclinical and clinical studies. While it is difficult (but possible) to perform placebo-controlled studies with paresthesia-based approaches, new paresthesia-free paradigms are well suited for placebo-controlled trials [27,55,56,58]. To further clarify the mechanisms of action of SCS and which patients might respond in the long term, objective measures characterizing the physiological effects of SCS (e.g. quantitative sensory testing, functional neuroimaging) could be combined with conventional patient-reported subjective outcome measures [57]. It is also important to establish a relationship between the mechanisms of action of SCS and the pathological mechanisms of a specific pain condition. This relationship is essential to predict the therapeutic efficacy of SCS and to improve patient selection.

Future research should also move towards developing patient-specific approaches to understand mechanisms of SCS. Due to limitations in animal models of chronic pain (e.g. inability of animals to report their pain, and unclear translation of animal results to humans) [59,60], it is important that we develop methods to systematically study SCS in human subjects. For example, patient-specific computational models that account for interpatient variability in anatomy and electrode locations are commonly used in other neurostimulation therapies, such as deep brain and transcranial electric stimulation, to investigate mechanisms of action and provide clinical decision support [61,62]. However, this type of patient-specific modeling approach is uncommon in SCS [63]. We believe that patient-specific approaches in SCS may be critical to understand the mixed clinical outcomes of SCS and will help optimize the clinical effectiveness of current and novel SCS technologies.