8-A860A-2018-Books-00091Rathmell5e_Ch096-NO CROP-ROUND1

C H A P T E R 9 6

Spinal Cord Stimulation

RICHARD B. NORTH and BENGT LINDEROTH

As shown in Figure 96.1, since 2010, frequencies as high as 10 kHz as well as “burst” (to be distinguished from con­ ventional monotonic) waveforms have been introduced. The therapeutic range for these new waveforms can begin below perceptual threshold, and this is fortuitous, as some patients prefer paresthesia-free stimulation, and as perceptual thresh­ old can coincide with discomfort threshold for frequencies . 800 Hz. 13 It also is now appreciated that even with conven­ tional (monotonic, , 1,500 per second) waveforms relief can be achieved at amplitudes below perceptual threshold. 14 Rel­ evant basic science and clinical experiences will be detailed in the following discussion. Today, SCS is a minimally invasive, reversible therapy im­ plemented with sophisticated techniques and implanted equip­ ment, including a variety of electrodes and multiple-output pulse generators. Unlike most surgical procedures undertaken to relieve pain, SCS does not ablate pain pathways or result in anatomic change. SCS also offers its candidates the advantage of undergoing a screening trial with a temporary SCS system before proceeding (or not) to implantation. Pain and its relief by SCS vary by condition and from pa­ tient to patient. When SCS is delivered to the right patient by the right (experienced) clinician in the right setting using the right equipment, pain relief is optimized and can be sustained for decades. It is important to remember, however, that SCS is expected to reduce, not eliminate, pain, particularly pathologic (neuropathic) pain which itself is a disease. SCS is not expected or intended to relieve nociceptive or biologically useful pain. Investigators are continually improving SCS patient care by refining techniques and equipment, and as we learn more about the mechanisms of action of SCS, we will be able to optimize its application.

History In antiquity, some healers successfully treated pain by placing electrogenic fish on or near the painful area of the patient’s body. 1 This crude form of electrotherapy enjoyed a measure of success but was limited in scope by the geographic and ecologic constraints associated with keeping the fish alive and avail­ able. Thus, electrotherapy was not widespread until creation of the Leyden jar in 1745 made it possible for physicians not only to deliver electrotherapy at will but also to exert a modi­ cum of control over how much electrical current the patient received. This was an exciting medical advance, and by the 19th century, physicians were considered well equipped only if they had a portable generator and could provide electrother­ apy for a wide range of indications. The advent of empirical medicine in the 20th century, however, caused most physi­ cians to abandon the application of electrical shocks to treat pain. In the mid-1960s, however, neurosurgeon C. Norman Shealy et al. 2 recognized that Melzack and Wall’s newly pub­ lished Gate Control Theory of Pain 3 provided a theoretical basis for a new form of electrotherapy that could be delivered with implanted electrodes. Shealy et al. 2 called the innovation “dorsal column stimulation”; today, we know it as spinal cord stimulation (SCS). Melzack and Wall’s 3 Gate Control Theory proposed that the balance of activity between large and small nerve fibers in the peripheral nervous system determines whether or not pain signals are transmitted centrally. According to the theory, when small fiber input is dominant, a pain “gate” opens in the dorsal horn (DH) of the entrance segment of the spinal cord; this gate closes when large fiber activity is dominant. Because electrical stimulation depolarizes large fibers before it affects small fibers, the theory suggests that it should be possible for stimulation to close the pain gate. This hypothesis inspired Shealy et al. 2 to deliver current directly to a terminal cancer patient’s spinal cord with an implanted electrode and external pulse generator in a successful bid to relieve the patient’s other­ wise intractable pain. Although the electrical stimulation therapies inspired by the Gate Control Theory have succeeded, the theory itself remains controversial. It predicts that all types of pain will be inhibited, but clinical experience has shown that SCS is more effective for neuropathic pain than for acute or nociceptive pain. 4,5 Further­ more, large fiber activity can itself signal pain (e.g., the pain of sunburn). 6 Despite the fact that the Gate Control Theory does not adequately explain all aspects of the therapies it inspired, it provides a useful description of the general concept of the transmission of pain signals and has stood the test of time to a remarkable degree. 7 Within a few years of the introduction of SCS, it was reported that coverage of each patient’s areas of pain by stimulation-evoked paresthesia was necessary for pain relief. 8,9 For many years thereafter, technical advances in SCS were directed at optimizing this, scaling amplitude to the range from percep­ tion of paresthesia to discomfort from the paresthesia. 10 More recently, automated methods have been developed to maintain amplitude below discomfort threshold based on either postural sensing or evoked potentials. 11,12

FIGURE 96.1  The presently used spinal cord stimulation (SCS) wave- forms. A: Traditional SCS frequency of about 30 to 80 Hz. B: Burst SCS with internal pulse frequency of 500 Hz and burst repetition rate of 40 Hz. C: High-frequency stimulation—usually 10 kHz. (Adapted from Pope JE, Falowski S, Deer TR. Advanced waveforms and frequency with spinal cord stimulation: burst and high-frequency energy delivery. Expert Rev Med De- vices 2015;12(4):431–437.)

2

3

CHAPTER 96  Spinal Cord Stimulation

Basic Science of Conventional Spinal Cord Stimulation INTRODUCTION The work of several investigative teams in the five decades since Shealy et al. 2 implanted the first “dorsal column stimu­ lation” electrode has helped us identify promising avenues of research, refine experimental techniques, and develop evidence in support of hypotheses that explain aspects of the mecha­ nisms of action of SCS. The challenge in experimental studies is to mimic the human painful condition and deliver stimulation, for example in rats, with parameters that would be clinically relevant in humans. Even in a homogeneous population of lab­ oratory animals, however, the yield of neuropathic pain models is uncertain, and clinical studies must grapple with even greater variability among human subjects and painful pathologies. Furthermore, as already mentioned earlier, the clinical applica­ tion of SCS elicits a discernible tingling, known as paresthesia, which confounds experimental blinding. An additional chal­ lenge is the fact that the universally agreed on measurement of the effectiveness of SCS (reduction in pain) is subjective and depends to an uncomfortable degree on a patient’s ability to remember the intensity of past pain in order to make a valid comparison with current pain. Pain assessment also relies on rather crude measurements, such as verbal rating scales and the Visual Analogue Scale. Thus, patient assessment has to be assisted by measures of medication use, physical activity, well-being, general perception of change, etc. Despite these obstacles, researchers have learned enough to propose distinct mechanisms of action for the therapeutic effects of SCS in the treatment of neuropathic, ischemic, and visceral pain. NEUROPHYSIOLOGY AND NEUROCHEMISTRY SCS affects both spinal and supraspinal circuits 15–21 and thus does not rely solely on antidromic activation of the dorsal columns. 17,22 Some SCS effects actually survive disruption of supraspinal circuits by transection of the dorsal columns, the dorsolateral funiculi, and even the entire spinal cord rostral to a lumbar electrode. 23–25 SCS modulates DH and/or supraspinal neurotransmitters; thus, the beneficial effects of SCS often outlast the period of active stimulation (see following discussion). Most studies indicate that endogenous opioids are not in­ volved in the pain-relieving effects of conventional SCS; for example, the effects of SCS are not reversed by the opioid an­ tagonist naloxone, 26 and the neuropathic pain of SCS patients is usually resistant to opioid therapy. SCS might 27 or might not 28 cause release of spinal opioids. There is some experimental evidence that only stimulation with very low frequencies (e.g., 4 Hz) might involve the release of endogenous opioids. 29 Additionally, in patients with angina, SCS during atrial pac­ ing and at rest has been demonstrated to release the opioid peptide b -endorphin into cardiac circulation, indicating that SCS affects “local myocardial turnover of the opioid peptides leu-enkephalin, b -endorphin, and calcitonin gene-related pep­ tide (CGRP), a powerful vasodilator.” 30 Nerve or nervous system injury can lead to neuropathic pain, which is often radiating, generally described as “burning,” and in some cases involves hyperalgesia (an extreme sensitivity to pain) and allodynia (pain from normally nonpainful stimuli). In contrast, dull, aching nociceptive pain, which is mediated by receptors in skin, muscle, bone, viscera, etc., is responsive to opioids. Because SCS, in most cases, is not thought to cause the release of endogenous opioids, 28 the clinical expectation is that SCS will be more effective as a treatment of neuropathic, ischemic, and visceral pain than of nociceptive pain. The fact

that SCS is effective in treating ischemic pain, which is con­ sidered a form of nociceptive pain, does not necessarily mean that SCS directly affects this type of pain. Instead, SCS seems to exert a beneficial effect on the underlying ischemic condition and pain relief in considered as secondary to that (see following discussion). Among the things rats and humans have in common are that nerve injury alone sometimes causes allodynia (in neuro­ pathic pain cases merely around 20% report allodynia 31 ) and that SCS is not universally therapeutic. Only in single clinical studies have the SCS effects on allodynia been monitored, but in selected material presented by Harke et al., 32 the therapeutic effect on allodynia was similar to that on continuous pain. Investigators also used models of painful neuropathy to ex­ plore the impact of SCS on the pain threshold in rats and found that during and after SCS, the threshold of withdrawal from innocuous mechanical stimuli increases and that SCS affects only the component of the flexor reflex mediated by A fibers (not the late component mediated by C fibers). 33 Thus, current thinking holds that, to a large extent, SCS acts at a segmental spinal level, 34 although additional inhibitory influences might be transmitted by descending serotonergic and noradrenergic pathways. A related study used the partial sciatic nerve ligation model to examine how SCS affects the response to innocuous stimuli in postligation rats with and without allodynia and in controls. Only in the allodynic rats did SCS significantly de­ press abnormal responses and spontaneous discharge of wide dynamic range (WDR) neurons. 35 The same research group introduced acetylcholine (ACH) into the list of transmitters possibly involved in the beneficial effects of SCS after a micro­ dialysis study demonstrated that ACH is released in the DHs of rats whose pain-related symptoms after nerve injury decreased in response to stimulation. 36 These effects seem to be caused by activation of muscarinic M4 and M2 receptors in the DH. 36,37 This finding might lead to new ways to enhance an otherwise inadequate effect of SCS in certain patients. Allodynia occurs when the activation of nerve fibers in the periphery causes hy­ peractivity of WDR neurons in the superficial laminae of cor­ responding DHs. 38 SCS relieves allodynia by suppressing this hyperactivity. 35 Treatment with g -aminobutyric acid (GABA) agonists also has this effect, and SCS induces GABA release in the DH of rats 18 and activates the GABA-B receptor. 39,40 SCS also decreases DH release of the excitatory amino acids glutamate and aspartate in rats. 40 One probable mechanism would be that release of GABA binding to presynaptic GABA-B receptors could inhibit the release of excitatory amino acids (e.g., glutamate). Investigators have also proposed that SCS might relieve pain by blocking conduction of primary afferents at the branch points of dorsal column fibers and collaterals. 6 This explana­ tion is insufficient, however, because the effect of SCS extends beyond the dorsal columns and electrical stimulation does not inhibit every type of pain. 21 Dorsal column activation, how­ ever, is more successful than ventral stimulation, 41 which might exert effects on the spinothalamic tract fibers that transmit nociception. SCS inhibition of the pathologic response (increased firing frequency of WDR neurons, afterdischarge in response to pres­ sure, etc.) of DH neurons in rats exhibiting symptoms of al­ lodynia after peripheral nerve injury continues for more than 10 minutes, 35 consistent with SCS-induced release of DH and ce­ rebral neurotransmitters, and changes the concentration of neu­ rotransmitters and their metabolites in cerebrospinal fluid. 41–44 Through the use of microdialysis and immunohistochemical techniques, investigators have examined the effects of SCS on the cerebral neurotransmitter serotonin. In decerebrate cats, SCS applied with clinical parameters evokes a DH release of serotonin. 42

4

PART FIVE  METHODS FOR SYMPTOMATIC CONTROL

It was recently demonstrated in neurophysiologic studies with microelectrode recordings from different brainstem re­ gions that both the “serotonergic cells” and the OFF-cells in the rostroventromedial medulla (RVM) are selectively acti­ vated by lumbar SCS in neuropathic rats, and this happens only in animals responding to the stimulation in previous behavioral studies. 45 In SCS-responding animals, the tissue concentration of 5 hydroxytryptamine (5-HT) is increased in lumbar DHs following stimulation. 46 The behavioral effects of the 5-HT release in the DHs seem to be mediated by 5-HT2A, 5-HT3, and the 5-HT4 receptors. 47,48 Similarly, a massive activation of neurons induced by SCS occurs in the locus coeruleus, another brainstem center, where many noradrenergic cell bodies are gathered. In contrast to the RVM where 5-HT release could be observed at the spinal level, there was no sign of direct projection of noradrenergic neu­ rons activated by SCS to the DHs (as demonstrated by several methods 49 ). Based on the finding that spinal extracellular GABA levels are lower in allodynic rats than in controls but increase in allo­ dynic rats that respond to SCS, 50 investigators have attempted to potentiate the therapeutic effect of SCS in nonresponding rats with concurrent intrathecal administration of normally sub­ therapeutic levels of GABA or the GABA-B agonist baclofen. 39 This strategy caused a marked increase in the rats’ threshold for paw withdrawal from innocuous mechanical stimulation. Intrathecal administration of the selective a 1 -adenosine recep­ tor agonist R-phenyl isopropyl adenosine produced similar re­ sults. 40 Administration of subtherapeutic doses of these agents as adjunct therapy with SCS causes nonresponding rats to re­ spond to SCS. 51 In the first clinical study based on this response, however, the addition of intrathecal baclofen and/or adenosine in small doses was effective, but the latter potentiated SCS in only 2 of 5 patients. 52 This result caused investigators to try instead other drugs administered orally in man for neuropathic pain. In rodents, the drugs were given intrathecally and intravenously instead: gabapentin and pregabalin in per se ineffective doses in non-SCS-responding rats with partial sciatic nerve lesion. In these rats, the drugs together with SCS reduced tactile allodynia in a dose-dependent manner and enhanced suppression of hyperex­ citability of WDR neurons. 53 Accordingly, investigators conducted a pilot study that of­ fered intrathecal baclofen (and, in a few cases, intrathecal ad­ enosine) to 48 patients who did not successfully respond to technically adequate SCS. 54 Although 20 patients achieved sat­ isfactory pain relief with baclofen plus SCS or baclofen alone, only 11 continued treatment: 7 with a combination SCS sys­ tem and intrathecal baclofen pump and 4 with only a baclofen pump. At mean 67 months postimplant, 2 SCS/intrathecal pa­ tients had their pumps removed and the remaining 9 in the group (5 SCS/intrathecal, 4 intrathecal) reported continued pain relief. The baclofen dose was increased 160% from base­ line, but 5 patients reduced their use of other analgesics. A similar dose-dependent effect occurred in rodents with intrathecal administration of an otherwise subtherapeutic dose of clonidine, which partially exerts its effects via release of ACH in the DH. 36,55 A small randomized, double-blind prospective clinical study reportedly indicated that clonidine could be equally useful as low-dose baclofen to enhance the effect of SCS. 56 We have yet to identify all of the neurotransmitters and neuromodulators affected by SCS, let alone to decipher their doubtless complicated interactions. 40,42,57 The mechanisms and neurotransmitters known or hypothesized (so far) to be in­ volved in the effects of conventional SCS in neuropathic pain are illustrated in Figure 96.2.

5-HT; NE

SCS

DLF

Dorsal roots

STT

DC

A

GABA

A

WDR

Ach

c

X

Skin (or organ)

Basic Science of New Spinal Cord Stimulation Waveforms HIGH-FREQUENCY SPINAL CORD STIMULATION In principle, high-frequency (HF) sinusoidal stimulation ap­ plied to a nerve or an axon provides a local conduction block (e.g., Kilgore and Bhadra 58,59 ). In contrast to HF current appli­ cation to a peripheral nerve, HF SCS applied to the dorsal spi­ nal cord of lightly anesthetized rats with the pulse width (PW) and the low amplitudes used clinically induced no block of transmission in the dorsal columns (DCs) nor any activation. 60 Clinical HF SCS is applied via bipolar stimulation at a pulse repetition rate of 10 kHz and a short PW of about 30 micro­ seconds at an amplitude that is below perceptual threshold (see Fig. 96.1). The “working hypotheses” for HF SCS so far have been (1) a transmission block or (2) activation of some pathways in the spinal cord. However, a recent computer simulation study 61 has demonstrated that both these hypotheses require high stimulation amplitudes that are at the upper end or outside of the ranges used clinically in HF SCS. A third hypothesis was based on observations from studies of the auditory system in cats and on some few patients with cochlear implants. HF stimulation seemed to be able to induce a desynchronization of neural signals from groups of neurons firing in synchrony. This interesting hypothesis has, to the best of our knowledge, never been studied in nociception (review Linderoth and Foreman 62 ). Furthermore, a study by Song et al, 60 has shown that trans­ mission in the DCs is not affected because there is neither fiber recruitment nor block with HF SCS. Another recent study 63 where recordings from rats with nerve injury were performed with tungsten electrodes on single DC fibers provided data sup­ porting the view that no blockade of the DCs is obtained with HF SCS at clinically relevant amplitudes. It seems that there is a therapeutic effect of HF SCS, but so far, it has not been found superior to that of traditional SCS in our FIGURE 96.2  Schematic illustration of mechanisms and neurotransmitters possibly involved in the effects of spinal cord stimulation (SCS) in neuro- pathic pain. SCS activation of dorsal column collaterals secondarily induces release of g -aminobutyric acid (GABA) from dorsal horn (DH) interneurons, activating mainly GABA-B receptors and decreasing the release of excit- atory amino acids from hyperexcited second-order DH wide dynamic range (WDR) neurons. SCS also causes cholinergic neurons to activate M4 and M2 muscarinic type receptors (Ach). Several other transmitters, adenosine, and hitherto unknown substances are also likely involved. Furthermore, the orthodromic SCS-induced activity in the dorsal columns might—via neuro- nal circuitry in the brainstem (or even more rostrally)—induce descending inhibition via serotonergic (5-HT) and noradrenergic (NE) pathways in the dorsolateral funiculus (DLF), which might contribute to inhibitory influences in the DHs. c, c fibers accompanying A d and A b fibers; DC, dorsal columns; STT, spinothalamic tract; X, unknown transmitters probably modulated by spinal stimulation.

5

CHAPTER 96  Spinal Cord Stimulation

own animal model using monophasic pulses. 60 Because the DCs are neither activated nor blocked by the HF SCS, the mecha­ nisms hypothesized may be segmental. In a rat study with bipha­ sic stimulation, Shechter et al. 64 compared 20%, 40%, and 80% of motor threshold (MT) as stimulation amplitudes. They used 50 Hz, 1 kHz, and 10 kHz applied for 30 minutes on 3 consecu­ tive days. The only amplitude producing relevant data was 40% of the MT. For this amplitude, some effects of SCS emerged over the 3 days, and actually, the frequency 1 kHz proved slightly better—or at least equally effective as the 10 kHz SCS. It must be remembered that clinical experiences demon­ strate that HF SCS ( . 800 Hz) might result in uncomfortable sensory experiences as soon as the amplitude is beyond the sen­ sory (paresthetic) threshold. 13 Very recent preclinical studies (unpublished data 65,66 ) using HF SCS (2 to10 kHz) where DC activation was also studied have yielded interesting results. Application of 10-kHz SCS in the rat through conductive agar from needle electrodes directly over the L5 segment demonstrated no evidence of a DC fiber conduction block nor activation. Stimulation for several hours did not induce asynchronous firing in myelinated primary sen­ sory neurons. In an inflammatory pain model producing more long-lasting pain, SCS using 20% MT for up to 135 minutes, which was verified to be subthreshold for activation of A b pro­ jection neurons, occurred after 45 to 90 minutes when com­ pared to control. The present view of HF SCS mechanisms may be summarized as follows: HF SCS must be applied with low amplitude—below paresthesia threshold; otherwise, it can be very uncomfortable. There seems to be neither activation nor block of the dorsal col­ umns with HF SCS. Although PW is short and amplitude low, HF delivers more energy, and thus, rechargeable or wireless de­ vices have to be used. The dorsal columns do not seem to be involved in the ef­ fect, 63,67 thus clearly distinguishing HF SCS from conventional SCS. As yet unpublished data indicate that HF SCS might

induce a slowly building up, inhibitory effect directly onto su­ perficial neurons in the DH. 65,66 BURST SPINAL CORD STIMULATION The use of irregular stimulation patterns including “burst stim­ ulation” originates from De Ridder’s work with cortical stim­ ulation, but “modulated stimulation” has been used earlier in the clinic. In the 1970s, burst transcutaneous electrical stimu­ lation (TENS) was launched as a variant to steady frequency TENS with very low “electro-acupuncture-like” frequency (1 to 5 Hz) as compared to normal TENS at about 100 Hz. 68 At this time, hypotheses about different mechanisms for burst and HF TENS were discussed, and the burst TENS was ap­ plied more for nociceptive pain types, and some data pointed to a possible mediation by release of endogenous opioids. As already mentioned, recent animal studies further indicate that the antinociceptive effect of low-frequency SCS may depend on opioid mechanisms. 29 One manufacturer marketed for a period during the 1980s an external stimulator that could give vari­ able stimulation patterns (“modulated SCS”), but it never be­ came popular, and the apparatus disappeared from the market. Burst SCS was presented as a stimulation mode, which would also be effective for the midline or axial low back pain com­ ponent of failed back surgery syndrome (FBSS). 69,70 De Ridder argues that bursts or irregular firing are similar to normal nerve activity and exert more prominent effects on supraspinal re­ lays (e.g., the thalamus, as “a wake-up call to the brain” to activate neurons). 71 De Ridder et al. 70 have, on the basis of “source localized electroencephalogram (EEG)” investigations of patients with burst SCS, advanced the idea that burst SCS could activate cortical areas involved in the modulation of pain perception (Fig. 96.3). In a very recent study, De Ridder and Vanneste 72 presented data from five patients undergoing tonic, burst, and sham stimulation. In a source-localized EEG subtraction and conjunction analysis, they showed that burst and tonic stimulation share activation of some cortical areas

Neuromodulation Nov 2015

FIGURE 96.3  Present hypotheses for mechanisms behind effects of burst stimulation of the spinal cord. Burst spinal cord stimulation (SCS) is hypothe- sized to especially modulate the activation of the medial (affective/attentional) pathway (right) . Conventional SCS more acts on the lateral spinothalamic tract which conveys information of nociception (strength; site). (Adapted from De Ridder D, Vanneste S. Burst and tonic spinal cord stimulation: different and common brain mechanisms. Neuromodulation 2016;19(1):47–59.)

6

PART FIVE  METHODS FOR SYMPTOMATIC CONTROL

such as the pregenual anterior cingulate cortex, whereas only burst SCS reduced the connectivity between the dorsal anterior cingulate and the parahippocampal cortices. The originators of the burst paradigm 72 claim that the burst type of SCS preferentially modulates the medial spinothalamic pathway leading to activity changes in subcortical centers involved in emotions, whereas conventional SCS acts more on the lateral tract leading to the lateral thalamic nuclei and eventually to the sensory cortex. They hypothesize that burst stimulation modulates the medial pain pathway by actions of C fibers synapsing onto lamina I neurons with projections to the dorsomedial nucleus of the thalamus and from there to the dorsal anterior cingulum. An earlier functional magnetic reso­ nance imaging (fMRI) study 73 demonstrated that conventional SCS predominantly modulated the lateral pain pathways as judged by changes in blood oxygen levels in the somatosensory cortices. Burst SCS, at least in rat models, does not seem to involve the DC system 74 —another difference to conventional pares­ thetic SCS. More recent data indicate that the effects of burst SCS do not rely on activation of GABA-B receptors 75 as has been claimed for conventional SCS (e.g., Cui et al., 39 Cui et al., 40 Meyerson et al., 44 Cui et al., 76 Ultenius et al. 77 ). Burst SCS is commonly administered at amplitudes that do not produce paresthesia. In spite of this, as the average fre­ quency is higher, the energy consumption is larger than that of conventional SCS. A recent finding is that the efficacy of burst SCS seems to relate to the electric charge per burst, at least as found in a rat model of neuropathic pain. 78 In conclusion, burst SCS as now most often used clinically (5 pulses at 500 Hz repeated with 40-Hz rate at subparesthetic amplitude) cannot be completely explained on the basis of physiologic mechanisms activated or inhibited by stimulation of the dorsal aspect of the spinal cord. Preclinical and clini­ cal studies are underway to shed more light on mechanisms of these novel algorithms. At present, almost all manufacturers produce stimulators that can generate stimulation frequencies up to and above 1 kHz and PWs up to 1 m sec. Higher frequencies than those typically used for SCS have been available for many years 79 but have been used and reported for pain only anecdotally. A few centers during the 80s applied high cervical SCS (around 1,000 Hz) for torticollis. 80 Because the higher frequencies and longer PWs were avail­ able already in the existing stimulators, many centers tried to adapt the stimulation parameters especially in cases with ther­ apy failure after some time with conventional SCS but eventu­ ally also in new cases during trials. It must be kept in mind that frequencies above 800 Hz with stimulation-induced paresthesias can be perceived as very unpleasant, 13 and so these parameters usually were used at subparesthetic amplitude. In this way, higher amounts of electric charge could be transmitted from the electrode poles to the neural tissue without discomfort or damage to the ner­ vous system. 81 Recent experiments in the rat have demonstrated that sub­ sensory SCS can exert clear effects on symptoms of neuropathic pain. 60,64,67 These studies also show that moderately increased frequencies of 1,000 Hz can be as effective at the subsensory level as 10-kHz SCS. Such changes in SCS parameters to in­ crease transfer of electric charge have been tried clinically in recent years referring to the fact that pulse density (i.e., the percentage active stimulation during a pulse cycle) can be in­ creased from just 1% to 2% up to 20% to 25% for the maxi­ mal available settings at a subsensory mode. MODERATE CHANGES OF CONVENTIONAL SPINAL CORD STIMULATION PARAMETERS

The few clinical reports on this type of stimulation now available 14,82–84 will be discussed below. Furthermore, some re­ cent studies also illustrate that for each patient, the SCS ther­ apy could be optimized by the use of “individualized” settings and that some patients actually like to keep their paresthetic SCS as part of their stimulation program. 14,84 In conclusion to this section, it must be admitted that we are just at the begin­ ning of exploration of the efficacy of slightly changed SCS pa­ rameters, and only short-term results are available so far. The mechanisms behind an increased effect has not been studied (besides using higher frequencies; e.g., 500 and 1,000 Hz), and the neurophysiologic/neurochemical correlates are not known. COMPUTER MODELING STUDIES Finite element computer models of SCS electrical fields in the spinal cord 85–87 have confirmed the current and voltage dis­ tribution measurements previously obtained in cadavers and primates. 88 Application of these models has also led to the pre­ diction that (1) an electrode’s longitudinal position governs its segmental effect; (2) bipolar stimulation with contacts sepa­ rated by 6 to 8 mm optimizes selection of midline, longitu­ dinally oriented fibers (a longer distance reduces therapeutic effect by favoring dorsal root stimulation); and (3) the elec­ trical field between two cathodes placed on either side of the physiologic midline does not sum constructively in the midline. Clinical experience confirms that the correct position and spac­ ing of SCS electrodes is essential for therapeutic success and that, instead of expanding the area of paresthesia, positioning electrodes cephalad to the involved spinal levels commonly elicits unwanted, excessive, local segmental effects. 10 The first computerized models developed to study the spi­ nal canal represented the meninges, cerebrospinal fluid, fiber tracts, and gray matter with geometric shapes. 89,90 The equally simplistic, two-dimensional initial computer models of dor­ sal epidural stimulation 91–93 were soon replaced by a three-dimensional model that took into account fiber tracts, their branch points, and dorsal roots. 85,94 In 1991, Holsheimer and Struijk 95 developed a new model that merged this three-dimensional construct with a McNeal-type 96 cable model of the electrical behavior of myelinated nerve fibers and data derived from mammalian myelinated fibers. 97 The resulting “University of Twente SCS Computer Model” allowed investigators to determine the impact of the location and configuration (anode/cathode, on/off) of SCS electrodes (e.g., to maximize recruitment of deep, midline, longitudinal axons rather than of the lateral, or dorsal root, fibers that can cause discomfort and motor responses) and to suggest ways to optimize SCS equipment design. The University of Twente model was further refined when the investigators were able to apply data from human sensory fibers. 98 This improved the accuracy of predictions about the impact of various SCS threshold voltages. 99 An additional improvement streamlined the mathematical techniques used to predict stimulation effects when computing the three-dimensional action potential field. 100 Model predictions for “transverse tripole” electrode performance have been validated in part by clinical studies. 101 Lempka et al. 61 recently applied a similar model to kilohertz frequency SCS and reported that at intensities used clinically, stimulation probably does not cause the direct activation or conduction block of dorsal column or dorsal root fibers. 61 Arle et al., 102 on the other hand, inferred from a similar model that because of an interaction between ion gate dynamics and the anodal current distribution over axons, larger fibers that cause paresthesia in low-frequency simulation are blocked at higher frequencies, whereas medium and smaller fibers are recruited, leading to pain relief by inhibiting centrally projecting WDR cells in the DH.

7

CHAPTER 96  Spinal Cord Stimulation

CONVENTIONAL SPINAL CORD STIMULATION MECHANISMS IN ISCHEMIC PAIN

SCS lead

SCS is thought to induce vasodilatation and improve micro­ perfusion in patients with ischemic pain, which is sharp, ach­ ing, heavy, and tiring 103 and a signal of local ischemia. Thus, SCS has a beneficial effect on the cause, not merely the symp­ toms, of ischemic pain. This might explain why ischemic pain is one of the few types of nociceptive pain known to respond to SCS and why the mechanisms seem to be fundamentally dis­ tinct from those that provide relief of neuropathic pain. 28,42,104 PERIPHERAL VASCULAR DISEASE To investigate the mechanism of SCS in the treatment of pe­ ripheral vascular disease (PVD), better renamed “peripheral arterial occlusive disease” (PAOD) (especially because if the vascular problem affects the venous side, standard SCS pro­ duces little or no effect), investigators developed a new ani­ mal model that involves applying mechanical pressure to an artery in the groin of rats. 105 Using this model, SCS delivered with clinically relevant stimulation parameters recovered nor­ mal microcirculation in 100% of treated rats versus 28% of controls. In addition, administering SCS preemptively reduced the amplitude of the invoked spasm and significantly shortened the time to recovery of microcirculation. In skin flaps with se­ verely compromised arterial blood supply, application of SCS could significantly increase the flap survival as judged 1 week after the provocation. If a CGRP receptor antagonist was given before the SCS treatment, the survival rate decreased consider­ ably, implicating this vasodilatory compound in the effect. 106 SCS also suppresses efferent sympathetic activity (maintained by nicotinic ganglionic receptors and a 1 -adrenoreceptors) 104 and might activate antidromic mechanisms at intensities far below the MT, 107–111 thus causing peripheral vasodilation by stimulating release of CGRP 106–108 from the terminals of sen­ sory fibers that contain transient receptor potential vanilloid-1 (TRPV1) receptors 112,113 and the release of nitric oxide from en­ dothelial cells. 113 The balance between these two mechanisms seems to de­ pend on the activity level of the sympathetic nervous system, the intensity of SCS, and individual patient factors (genetic dif­ ferences, diet, etc.). 111 In fact, antidromic activation dominated at low autonomic baseline activity, whereas the sympatholytic effects of SCS were clear with high baseline activity. 111,114 Later studies have indi­ cated that even small-diameter fibers are involved at SCS in­ tensities much below the MT 112,113 and have pointed toward additional mechanisms. The observation that SCS has a powerfully beneficial ef­ fect on vasospastic conditions, such as Raynaud’s syndrome, is consistent with theories that the cause of this syndrome is a combination of heightened sensitivity or increased density of a -adrenergic receptors 115 and CGRP-system dysfunction. 116 A stimulation-induced “normalization” of function in each sys­ tem could underlie the efficacy of SCS in treating this condition. Up to the present, most animal studies have utilized SCS frequencies that are routinely applied in the clinic (i.e., 40 to 80 Hz), but in a more recent study, higher SCS frequencies up to 500 Hz were tried at intensities 30% to 90% of MT. 89 This study showed that although the MTs for SCS at all frequencies were similar, SCS at 500 Hz induced a significantly larger blood flow elevation in the hind paw than did SCS at 50 Hz. The ef­ fects of these frequencies and intensities seem to depend on ac­ tivation of TRPV1-containing fibers and the release of CGRP. Thus, further trials with new stimulation parameters should be undertaken to increase benefits of SCS. A review of the mechanisms involved in SCS-induced vaso­ dilation is included in a report by Wu et al. 117 and by Foreman and Linderoth. 118

Dorsal roots

CGRP NO

A

A

DC

c

Artery

1 Adrenorec.

nicotinic

Sympathetic Efferent Fibers

The mechanisms and neurotransmitters known or hypoth­ esized to be involved in the effects of SCS in ischemic pain are depicted in Figure 96.4. SPINAL CORD STIMULATION FOR ANGINA PECTORIS AND CARDIAC DISEASE Investigators studying the mechanism of action of SCS in pa­ tients with otherwise refractory angina agree that SCS reduces ischemia 119 but disagree about how this occurs. Positron emis­ sion tomography has indicated that SCS causes a redistribution of coronary blood flow in patients with refractory angina 120,121 (even though other experimental studies have failed to demon­ strate this effect). 122 On the other hand, the decrease in the de­ pression extending from the end of the S wave to the beginning of the T wave (ST-segment depression) that appears on electro­ cardiograms (ECGs) during SCS treatment and the observed SCS-induced reversal of lactate production to extraction might indicate an accompanying decrease in cardiac myocyte oxygen demand. 123 The SCS-induced protective changes that increase myocar­ dial resistance to critical ischemia 124 are manifest by the im­ proved tolerance of patients to a deliberately paced increase in heart rate 119 and by increased time-to-angina in exercise tests. 119,123 This effect might signal an SCS-induced inhibition of the excitatory effect of ischemia on the intrinsic cardiac nervous system (such an excitatory effect might lead to dysrhythmia and increased ischemia). 124–126 The possibility that SCS modu­ lates cardiac neurons is supported by the finding that transcuta­ neous electrical nerve stimulation increases blood flow in intact human hearts but not in transplanted, denervated hearts. 127 There is no proper animal model of angina pectoris mimick­ ing the syndrome in humans. The animal studies discussed in the following text are instead focused onto various deleterious effects of experimentally induced chronic and/or acute coro­ nary ischemia. Because SCS reduces total body, but not cardiac-specific, norepinephrine spillover during pacing to moderate angina, 128 part of the anti-ischemic effect of SCS might owe its potency to an overall reduction in sympathetic activity. An experimental study using induced cardiac infarcts in a rabbit model, how­ ever, indicates that the decrease in infarct size with SCS therapy FIGURE 96.4  Schematic illustration of mechanisms and neurotransmit- ters possibly involved in effects of spinal cord stimulation (SCS) in ischemic pain. SCS probably indirectly exerts inhibition onto medullary neurons, thus perpetuating sympathetic efferent vasoconstriction via nicotinic gan- glionic receptors, mainly a 1 ( a 1 Adrenorec. 5 adrenoreceptors) peripheral receptors. In parallel, SCS activates an antidromic loop inducing peripheral release of calcitonin gene-related peptide (CGRP), probably also involving small-diameter fibers. An inhibition of nociceptive transmission has also been indicated in experimental studies but is clinically unlikely. c, small diameter unmyelinated “c-fibers”; DC, dorsal columns; NO, nitric oxide.

8

PART FIVE  METHODS FOR SYMPTOMATIC CONTROL

shares some mechanisms with “ischemic preconditioning” but, notably, does not cause cardiac ischemia. 129 In order to mimic the development of chronic ischemic heart disease in an animal model of myocardial ischemia, progressive occlusion of the arterial blood supply by a device or creation of cardiac infarction of moderate size could be used, after which a period of acute ischemia provocation with or without SCS could be programmed. In one early canine study, a slowly expanding material lining the inside of a metal constrictor ring was implanted around the proximal left circumflex coronary artery of a group of dogs. 124 This technique progressively reduces blood flow through the artery and facilitates the development of collaterals creating a collateral-dependent myocardial ischemia substrate. In sub­ sequent acute experiments, the exposed heart was paced at a basal rate of 150 beats per minute. An ECG plaque was used to record from multiple sites on the left ventricle supplied by the left coronary artery occluded by the constrictor. In order to stress the heart, either angiotensin II, administered via the local arterial supply to the right atrial ganglionated plexus, was used, or rapid ventricular pacing applied via a standard pace­ maker. Both these maneuvers produced an elevation of the ST segments that was markedly attenuated during SCS. In a similar way, ST-segment responses were largely unchanged when rapid ventricular pacing (240 beats per minute during 60 seconds) was induced during SCS. These experiments indicate that SCS may attenuate the deleterious effects that stressors, especially chemical activation of the intrinsic cardiac nervous system, exert on a myocardium with reduced reserve capacity. This ob­ servation led to the conclusion that SCS produces anti-ischemic effects that contribute to improved cardiac function. Further evidence to support the anti-ischemic effects of SCS on the heart is the observation that preemptive SCS appears to have a protective effect on the myocardium, which makes it more resistant to critical ischemia as demonstrated by rabbit experiments with left anterior descending (LAD) artery occlu­ sion lasting 30 minutes. In these studies, the infarct size was markedly reduced by preemptive SCS. However, the protective effects of SCS therapy were lost if SCS was begun after ischemia induction. 129 Patients with SCS therapy for chronic therapy-re­ sistant angina are recommended to use their stimulators at low amplitude most of the day or at least for 6 to 8 hours and to increase the amplitude when needed during an angina attack or when physical stress is expected to produce angina. Thus, the validity of this clinical recommendation is substantiated by experimental data. In experimental cardiology, there is a well-known phenome­ non that a short ischemic episode preceding a longer occlusion of a coronary vessel induces complicated protective processes in the myocardium that diminishes the resulting infarct size. This phenomenon is called ischemic preconditioning, and the details are still not completely known (e.g., Foreman 130 ). Re­ cent studies indicate that SCS-induced local release of cate­ cholamines in the myocardium may trigger protective changes related to mechanisms behind such ischemic preconditioning but without producing any signs of ischemic changes in the heart. There are also other signs indicating that SCS may in­ duce a state similar to that following a short ischemic period, for example, by activating protein kinase C, a substance which is pivotal in ischemic preconditioning. 129 An important part of the “general common pathways” in the communication between the CNS and the heart is the intrinsic cardiac nervous system (ICNS). The ICNS is located in the car­ diac ganglionated plexuses covered by epicardial fat pads situ­ ated on the myocardium. 131 The ICNS plexuses are composed of mixed somatosensory, sympathetic, and parasympathetic fibers. The ICNS plays a critical role in coordinating regional cardiac function and providing rapid reflex coordination of autonomic

neuronal inflow to the heart. In critical ischemia, the ICNS is vigorously activated. 126 The ICNS responds to ischemic stress by marked activity increase even if the ischemic region is situated far away from the neuron population. 125 If the increased activity persists, it may result in spreading dysrhythmias that may lead to more generalized ischemia and/or to ventricular fibrillation. Several experimental studies have clearly shown that SCS may potently inhibit and stabilize the activity of the ICNS especially during a critical ischemic challenge. In patients with angina, SCS can relieve the symptoms and signs of ischemia for long periods after the stimulation is ter­ minated which may relate to prolonged effects of SCS on ICNS activity observed at least up to 45 to 60 minutes after SCS stim­ ulation off in dogs. 131 Modulation of the ICNS may be one mechanism that pro­ tects the heart from more severe ischemic threats due to gen­ eralized arrhythmias. Others have confirmed the observation that experimental animals display less arrhythmia during isch­ emic provocation when being subjected to SCS. Experiments by Lopshire et al. 132 demonstrated that SCS might improve car­ diac function in canine heart failure following an experimental myocardial infarction and continued stress by HF pacing over 8 weeks. In addition, acute experiments with experimental oc­ clusion of the LAD carried out with or without SCS on land­ race pigs showed that the stimulation provided positive effects as displayed in the vectorized ECG. 133 In several of the studies mentioned earlier, the ischemic chal­ lenge induced arrhythmias, but in virtually all studies, these were less severe during SCS treatment. This observation was recently supported by a new study. 134 The use of SCS for angina reached a peak in the 1990s and early 2000s (when it was the best indication for SCS with out­ comes . 80% of patients clearly helped by the therapy), but thereafter, the use of this technique has diminished consider­ ably also in Europe due largely to increased use of stenting. It also should be noted that angina pectoris is presently not a U.S. Food and Drug Administration (FDA)-approved indica­ tion for SCS in the United States. Some of the pathways and mechanisms behind beneficial effects of SCS on cardiac function discussed earlier are sche­ matically summarized in Figure 96.5, which illustrates the mechanisms and neurotransmitters known or hypothesized to be involved in the effects of SCS in angina. The treatment of visceral pain is a relatively new application for SCS, and investigators have proposed that SCS might exert its positive effect on visceral pain (and dysfunction) by moderat­ ing the so-called “brain–gut” axis (the neural circuitry thought to control the interface among visceral afferent sensation, intes­ tinal motor function, and the brain). 130,135–137 In fact, moving an active SCS electrode along the neuraxis demonstrates that electric activation occurs at various levels; thus, in addition to its beneficial effects on pain and ischemia, SCS inhibits the viscerosomatic reflexes involved with the par­ ticular spinal segmental level being stimulated (Fig. 96.6). Some rodent studies in experimental colonic pain 137,138 demonstrated that SCS applied with conventional clinical stim­ ulation parameters significantly decreased the painful symp­ toms (measured by monitoring of abdominal contractions as response to balloon inflation in the distal colon). Based on these observations, first a case of irritable bowel symptom was successfully treated by SCS, 139 and thereafter, a prospective ran­ domized clinical study in a small series confirmed the beneficial findings in a pilot study. 140 In fact, as shown in Figure 96.6, SCS might have many as yet unexplored positive effects on visceral problems. MECHANISMS OF SPINAL CORD STIMULATION IN VISCERAL ABDOMINAL PAIN

9

CHAPTER 96  Spinal Cord Stimulation

Vagus nerve

STT

Dorsal roots

SCS Lead

A

A

FIGURE 96.5  Schematic illustration of some mechanisms possibly involved in the effects of spinal cord stimulation (SCS) on coronary isch- emic pain. SCS might exert indirect inhibitory effects on nociceptive transmission to higher centers and on the level of sympathetic activity; SCS might also have antidromically transmitted effects. Intrinsic cardiac nervous system (ICNS) are deeply involved in monitoring ischemic events in the heart, and this function is drastically influ- enced by SCS. The interplay between somatosen- sory and autonomic influences and the effects of SCS is presently largely unknown but is the sub- ject of intense investigation. DC, dorsal columns; STT, spinothalamic tract.

DC

c

ICN

?

?

Sympathetic Efferent Fibers

Organ Involved

SCS Effect

SCS

1. Bronchodilation

1

C2

Cervical

2

3

2. Peripheral vasodilation

T1

High thoracic

4

3. Stabilization of ICNS Reduction of ischemia and pain Decreased infarct size

Middle thoracic

5

4. Decreased colonic spasms Pain Reduction

L1

Low thoracic

6

5. Peripheral vasodilation

6. Decreased bladder Spasticity Increased volume tolerance

S1

Sacral

FIGURE 96.6  Spinal cord stimulation (SCS) applied at different levels of the neuraxis might, in addition to affecting pain and peripheral blood flow, induce changes in different target organs mediated via stimulation induced changes in local autonomic activity, dorsal root reflexes, or viscerosomatic eflexes. Some of these changes in target organ function might be beneficial. ICNS, intrinsic cardiac nervous system. (Redrawn after Linderoth B, Foreman RD. Mechanisms of spinal cord stimulation in painful syndromes: role of animal models. Pain Med 2006;7:514–526).