Conventional Spinal Cord Stimulation

Treatment of chronic pain remains a large unmet medical need. The seminal “gate control” theory of pain postulates that stimulation of low-threshold afferents (Aβ-fibers) would result in an inhibition of spinal nociceptive transmission ( ). This theory led to the development of viable pain therapies including conventional spinal cord stimulation (SCS) and transcutaneous electrical nerve stimulation (TENS) ( ). A contemporary view of gate control now includes the broadened “neuromatrix” concept that encompasses multiple interacting neuronal circuits ( ). Conventional SCS is produced by delivering mild electrical pulses through electrodes placed in the epidural space overlaying the dorsal columns (DCs), typically at an intensity that elicits paresthesia, at a frequency of 30–60 Hz.

SCS has been shown to provide pain relief and improve quality of life in patients with refractory pain ( ), the most common indications having been neuropathic pain of peripheral origin or “mixed pain conditions” with a dominant neuropathic component such as complex regional pain syndrome (CRPS) and failed back surgery syndrome (FBSS) ( ). Conventional SCS remains the mainstay and most used neurostimulation pain therapy ( ), and has, in many published studies, performed well in comparison with other therapies including “conventional medical management” (CMM) when investigating safety, appropriateness, fiscal neutrality, and effectiveness, the SAFE principles recently introduced into relevant literature by . Based on the SAFE principles, systematic analysis of clinical evidence suggested that SCS should be considered before submitting a patient to either long-term systemic opioid therapy or repeat spinal surgery for chronic pain resulting from FBSS and CRPS but not as a last resort in pain therapeutic algorithms ( ).

High-Frequency Spinal Cord Stimulation

Recently, a new, high-frequency SCS (HF-SCS, e.g., 10 kHz) paradigm has been introduced for pain treatment that is paresthesia free ( ). Like conventional SCS, HF-SCS (1 and 10 kHz) inhibits mechanical hypersensitivity in an intensity-dependent fashion in a rat model of neuropathic pain, and certain types of HF-SCS may have a greater effect or require a lower intensity for pain inhibition than conventional SCS ( ). A randomized and controlled clinical trial also suggests that subliminal/paresthesia-free HF-SCS may be superior to conventional SCS in the long-term treatment of back and leg pain ( ).

Burst Spinal Cord Stimulation

In the 1970s, burst TENS was launched as a variant to tonic TENS with very low, acupuncture-like frequency or normal TENS at about 100 Hz ( ). Hypotheses about different mechanisms for burst and high-frequency TENS were also discussed at that time, with burst stimulation seen as more appropriate for inhibiting nociceptive pain involving a release of endogenous opioids. The concepts regarding burst SCS (bursts of five pulses with internal frequency of 500- and 40-Hz interburst frequency), introduced by De Ridder et al., stems from his work with cortical stimulation ( ). In several clinical studies, burst SCS, administered with subsensory amplitudes, was initially presented as a stimulation mode that was more effective than conventional SCS for the lumbosacral pain and nociceptive back pain component of FBSS ( ). It is reported that 63% of nonresponders to conventional SCS respond to burst stimulation. In addition, 95% of the responders to tonic SCS experienced further improvement with burst stimulation ( ). Thus, burst SCS may significantly rescue nonresponders to conventional SCS and further improve pain inhibition in responders to conventional SCS. Recent studies suggest that burst SCS may also be more efficacious than tonic 50-Hz SCS in inhibition of neuropathic pain and visceral nociception and the improvement of physical activity ( ). Nevertheless, a recent systematic review suggests that the clinical evidence for burst SCS, in managing chronic intractable pain, remains insufficient ( ).

Conventional Tonic Spinal Cord Stimulation

Pain inhibition by conventional SCS is intricately, though not exclusively, linked with suppression of DH neuronal activity ( ). Electrophysiological studies have revealed important features of neuronal inhibition by SCS that may correlate with clinical pain inhibition. Conventional SCS, using either surface ball electrodes or bipolar, needle electrodes inserted into the DCs of rats, decreases WDR neuronal activity ( ). SCS, in rats after nerve injury, inhibits spontaneous discharges and evoked responses to peripheral mechanical and electrical stimuli in WDR neurons ( ). These cellular mechanisms ( Fig. 15.1A ) may contribute to conventional SCS’s inhibition of ongoing pain and tactile allodynia ( ), two characteristic features of neuropathic pain ( ).

There is dynamic change of spinal neuronal circuits in response to different, environmental cues. Windup and long-term potentiation (LTP) in DH neurons may share certain mechanisms with central sensitization that underlies persistent pain ( ). Conventional SCS blocks windup ( ) and normalizes LTP in WDR neurons ( ). These physiologic data and computer model simulations suggest that inhibition of neuronal sensitization may be an important mechanism underlying neuropathic pain inhibition by SCS in humans ( ). A windup-like phenomenon, as from repetitive heat stimuli and repeated noxious mechanical stimuli, also occurs in humans during natural stimulation of C-fibers ( ).

It has been difficult to find predictors of conventional SCS analgesia in patients ( ). Recently, inhibition of the temporal summation of pain (e.g., windup) during trial SCS was shown to be helpful when selecting patients for the therapy and predicting the long-term efficacy of SCS ( ). This work presents a good example of how preclinical findings provide a conceptual framework and important rationale for improving clinical applications of SCS. Intriguingly, when used earlier after nerve injury (1 day), SCS produces more responders and longer pain reduction than when given later after nerve injury (16 days) ( ). This suggests that earlier intervention for neuronal sensitization with SCS may be more beneficial than when the therapy is introduced at a later period of time. Other spinal mechanisms may also underlie conventional SCS pain inhibition and include activation of descending pain-inhibitory pathways, induction of inhibitory postsynaptic potentials in DH neurons ( ), and/or facilitation of primary afferent depolarization, which elicits presynaptic inhibition of incoming afferent inputs ( ).

High-Frequency Spinal Cord Stimulation

Despite increasing clinical use, the neurophysiologic basis for pain inhibition by HF-SCS remains unclear. Using the antidromic, sciatic, compound action potential as a way to titrate stimulation intensity, investigators recently have shown that 50-Hz SCS inhibits WDR neurons in an intensity-dependent fashion ( ). At the high stimulation intensity that maximally activates Aβ-fibers in the DC (suprasensory threshold), and without activating Aδ/C-fibers, 1-kHz HF-SCS was less effective than 50-Hz SCS in suppressing windup, a short-term neuronal sensitization to repetitive noxious inputs, in WDR neurons ( ). This suggests that pain inhibition from kilohertz levels of frequency and 50-Hz SCS at suprathreshold intensities may involve different, spinal, neuronal mechanisms. Although, an earlier study suggests that 10-kHz HF simulation of the DR or at the dorsal root entry zone (DREZ) may depress hyperactivity in WDR neurons after nerve injury ( ), the stimulation site and intensity used in this study may not mimic those of clinical HF-SCS.

The onset of pain relief is often quick (minutes) following the introduction of conventional SCS in both patients and animal models ( ), but it takes much longer, sometime days, before pain inhibition occurs after paresthesia-free HF-SCS ( ). Accordingly, an important aspect to consider when studying the MOA differences between conventional and paresthesia-free HF-SCS may be the length of the stimulation period. Examining the neurophysiologic and neurochemical changes that may occur slowly and progressively at spinal levels after a longer period, subliminal HF-SCS may reveal new knowledge about its MOA.

Burst Spinal Cord Stimulation

Burst SCS attenuates DH, neuronal hyperactivity in a rat model of cervical radiculopathy ( ), and the attenuation of responses to noxious, somatic stimuli from burst stimulation is greater than attenuation to visceral stimuli in lumbosacral neurons ( ). Burst stimulation was thought to be more effective than tonic stimulation, because less temporal integration is needed to activate neurons, making it more effective as a “wake-up call to the brain.” An earlier functional magnetic resonance imaging study showed that conventional SCS predominantly modulates the lateral pain pathways and inhibits the somatosensory component of pain, as judged by changes in blood oxygen levels in the somatosensory cortices ( ). Burst SCS, comparatively, may primarily modulate the medial pain pathways and alleviate the affective component of pain ( ). Another hypothesis presented here relates, in a way, to an HF-SCS hypothesis: burst stimulation disrupts synchronous burst firing of high-threshold fibers, thus inhibiting an activation directly related to pain perception.

GABA

SCS increases extracellular GABA content in the spinal cord and is associated with decreases in glutamate and aspartate that exceed the duration of stimulation ( ). Inhibition of pain and decreased WDR neuronal hyperexcitability are closely associated with the time course of elevated GABA levels and decreased levels of glutamate and aspartate after SCS. The GABA _B receptors may play a more important role than the GABA _A receptors in mediating these inhibitory effects ( ). Furthermore, the intrathecal administration of a low dose of baclofen transforms SCS nonresponder rats into responders to SCS and enhances SCS-induced pain relief in human subjects ( ). However, effects of burst SCS do not rely on activation of GABA _B receptors ( ) because GABA levels do not change after burst SCS and pain inhibition is not blocked by CGP35348, a GABA _B receptor antagonist ( ). In SG neurons, blocking fast inhibitory neurotransmission mediated by GABA _A and glycine receptors does not preclude the inhibition of C-fiber eEPSCs by 50-Hz Aβ-fiber stimulation ( ). The neurochemical mechanisms by which SCS inhibits excitatory and inhibitory interneurons are still incompletely known. It will be important to develop cell type-selective modulation and rationally choose adjuvant pharmaceutical agents to enhance SCS analgesia (e.g., agents to prevent the suppression of inhibitory neurons by SCS).

Serotonergic, Cholinergic, and Adrenergic Mechanisms

SCS induces release of serotonin into the spinal cord ( ), which decreases neuronal excitability and pain transmission by activating 5-HT _2A , 5-HT ₃ , and 5-HT ₄ receptors ( ). Serotonin may also increase the spinal expression and synthesis of dynorphin, enkephalin, and GABA ( ), which prolongs pain inhibition by SCS. SCS also induces the release of acetylcholine, adenosine, and noradrenaline ( ). Spinal acetylcholine content is elevated by SCS only in responder rats and not in nonresponder rats ( ). Acetylcholine and noradrenaline may excite spinal, GABAergic interneurons by binding to muscarinic M4 receptors and α ₁ -adrenoceptors, respectively ( ). Intrathecally administered, subeffective doses of clonidine or a muscarinic receptor agonist were shown to transform SCS nonresponders into SCS responders ( ).

A recent study suggests that activation of spinal cannabinoid CB1 receptors may contribute to synaptic depression from high-threshold afferent inputs in SG neurons after electrical stimulation of Aβ-fibers in the DR. It may also be involved in conventional SCS-induced inhibition of spinal nociceptive transmission after nerve injury ( ).

Collectively, SCS may initiate feedforward activation of various spinal neurochemical mechanisms, some of which (e.g., GABA) may be compromised by nerve injury ( ). It remains to be determined if different neurochemical mechanisms exhibit synergistic functional interactions. Better understanding of the neurochemical basis for SCS analgesia will provide rationales for developing new treatment strategies that include the addition of adjuvant drugs to potentiate SCS analgesia ( ).

Other Mechanisms

In addition to the mechanisms just mentioned, SCS may induce transcriptional and posttranslational changes in neurons. Examples include activation of extracellular signal–regulated kinase (ERK) and AKT pathways ( ) and expression of the immediate early gene c- fos ( ). These changes may alter gene expression and exert long-lasting impact on cellular function, but the physiologic implications of SCS analgesia remain to be determined. High throughput approaches such as full-genome microarray analyses begin to reveal SCS-induced changes in a large number of genes in the spinal cord and DRG and uncover the highly significant interconnected genes and pathways that may be responsive to pain inhibition by SCS ( ).

Computational Modeling

In the spinal cord, computational models have explored the response of DC fibers to kilohertz stimulation frequencies. One study explored whether “depolarization blockade” of DC and/or DR fibers occurs with 1-kHz epidural SCS. These authors concluded that to achieve any blockade of the largest fibers, the stimulation amplitude must be approximately twice the activation threshold, similar to observations in peripheral nerve, and that the clinically reported amplitudes for 10-kHz paresthesia-free SCS were likely below the block threshold ( ). A different computational modeling investigation of kilohertz stimulation in SCS posed that DC blockade may be fiber selective ( ). Under the assumption that the tissue intervening between the SCS contact surface and the DC axons alters the shape of the stimulating pulse from symmetric biphasic to quasi-monophasic, Arle et al. calculated in a computational model of SCS that large-fiber blockade could be achieved at much lower amplitudes while still activating medium and smaller-diameter myelinated DC fibers. In these authors’ view, this might explain the lack of paresthesia, because the larger-diameter DC fibers would primarily communicate touch and pressure sensations from the periphery to higher centers (i.e., yielding perceived paresthesia), while the smaller-diameter fibers would be primarily proprioceptive and generate no sensation. Given that there are far more medium-sied and small fibers in the DC than large fibers, these proprioceptive fibers might then provide a more profound spinal gating of pain than activation of the relatively fewer large-diameter DC fibers ( ). However, there have been no clinically reported observations of altered sensorimotor function with 10-kHz paresthesia-free SCS, so significant blockade of sensory fibers and/or massive activation of proprioceptive fibers appears unlikely. Thus, this hypothesis remains to be validated in both preclinical and clinical settings.

Animal Studies

Preclinical investigations of kilohertz stimulation frequency to date have predominantly focused on the response of peripheral nerves, primarily exploring the phenomenon of “depolarization blockade.” For example, 20-kHz sinusoidal currents block activity in peripheral nerves ( ) and biphasic rectangular pulses induce block on motor nerve conduction ex vivo when high amplitudes are used at frequencies of 4 kHz and above ( ). Kilgore and Bhadra investigated neural blockade as produced by voltage-controlled and current-controlled stimulation by pulsatile and sinusoidal waveforms (among others) and monopolar versus bipolar electrode configurations and produced complete sciatic nerve conduction block over the frequency range of 2–20 kHz in adult bullfrogs ( ). Later, in studying motor fiber blockade in the rodent by using high-frequency sinusoidal alternating current, Bhadra and Kilgore demonstrated that stimulation between 10 and 30 kHz, at intensities between 2 and 10 V, produced complete repeatable and reversible motor block that showed a linear relationship to frequencies used ( ). Joseph and Butera also observed both A- and C-fiber blockade at kilohertz frequencies up to 50 kHz but stated that unmyelinated C-fiber blockade had a nonmonotonic relationship to frequency ( ). Cuellar et al. explored the effects on DH neurons on stimulation frequencies ranging from 2 to 100 kHz applied to DRs and DREZ and appeared to also have observed depolarization block of primary afferent neurons. They saw that responses to innocuous as well as nociceptive inputs were reduced at kilohertz frequencies with increasing stimulation amplitude ( ). Notable in the studies of kilohertz frequency depolarization blockade is the relatively high amplitude of stimulation required to achieve the block. Both computational modeling and preclinical studies in peripheral nerves have shown that the block threshold (i.e., the voltage or current amplitude of the stimulation) is approximately 2.5–5 times the activation threshold for the neuron ( ). Indeed, the axon to be blocked must generate at least one AP for the gating dynamics of the membrane ion channels to enter the state for a possible depolarization block ( ). It is believed that activation of K ⁺ channels or inactivation of Na ⁺ channels is responsible for the blocking effects of kilohertz-frequency nerve stimulation ( ). Clinically, attempts at peripheral nerve blockade have indicated that sensation tends to precede or accompany an assumed blockade of activity ( ).

Recent preclinical work looking into mechanisms of HF-SCS has yielded variable results. In preparations where the DC was targeted for stimulation, variations in frequency demonstrated behavioral changes only when relatively large amplitudes of stimulation were used. Shechter et al., using spinal nerve ligation rodent pain models, observed that only stimulation intensities of 40% or greater of motor threshold (MT) yielded significant increases in paw withdrawal threshold (PWT) for 50 Hz, 1 kHz, and 10 kHz when compared with sham stimulation ( ). In these series of experiments, 40% of MT was considered to be just below sensory threshold; the highest amplitude, 80% of MT, yielded even larger effects but may be considered too strong a stimulation intensity for clinical use. These behavioral effects were shown to be more pronounced with repeated 30-min applications of SCS over 3 days and by use of kilohertz frequency (compared with 50 Hz). Additionally, at very strong intensity, 50 Hz reduced the C-fiber component of WDR neuronal wind-up response compared with 1 kHz. , in both spared nerve injury and inflammatory pain model rats, explored behavioral effects of 50 Hz, 500 Hz, 1 kHz, and 10 kHz and made efforts to program the stimulation to intensities believed reflective of clinical practice: 80% of MT for 50 Hz and 40–50% of MT (“subparesthetic intensity,” determined by behavioral observations and corroborated by companion, electrophysiologic measurements in nucleus gracilis) for the higher frequencies ( ). Stimulating for 2 h, they saw large early changes for 50 Hz compared with other frequencies, but by the end of the stimulation period, all frequencies yielded similarly, significant improvements in mechanical (but not thermal) PWT in the affected limb compared with the contralateral limb. In a follow-up study, Song et al. explored the importance of pulse shape and pulse width for 50-Hz and 1-kHz frequencies ( ). They found similarly that 50 Hz, at PW = 200 μsec at 80% MT provided reductions in PWT earlier in the stimulation protocol, but after 60 min of stimulation, monophasic and biphasic 1-kHz stimulation with a negative phase of 24 μs (but not 12 μsec) at 40–50% MT showed similar reductions in PWT. The lack of a distinct response to kilohertz frequency in these DC-focused preclinical studies suggests that different models and neural targets may be necessary to uncover HF-SCS mechanisms, as discussed in “Future Directions” later.