Lumbosacral pain with leg radiculopathy following nerve injury from a prolapsed intervertebral disc is one of the most frequent causes of neuropathic pain.¹ Conventional patient management of neuropathic back and leg pain secondary to disc herniation includes surgery. However, when pain persists and no further surgical target exists, the patient is often described as suffering from failed back surgery syndrome (FBSS), defined as “persistent or recurrent pain, mainly in the region of the lower back and legs, even after technically, anatomically successful lumbosacral spine surgeries.”² In these patients neuropathic pain is a main pain-generating mechanism characterized by predominant leg pain (radiculopathy). FBSS is common, affecting approximately 10% to 40% of patients who have undergone lumbar spinal surgery³; among the sources of the FBSS pain are recurrence of disc herniation, arachnoiditis, epidural fibrosis, and various radiculopathies.⁴

The most painful component in FBSS is the radiculopathy, as shown by the characteristics of an FBSS population compared to other chronic pain conditions.⁵ The severity of pain in FBSS patients was on average moderate in the back (mean visual analog scale [VAS] 4.9) but severe in the legs (mean VAS 7.4); patients were severely disabled, as indicated by several validated scores (Oswestry Disability Index [ODI], SF-36, EQ-5D), and their health-related quality of life was very poor.⁵ All patients had tried at least one class of drug or nondrug treatment before undergoing a trial of neuromodulation, and 87% had tried four or more types of drug and nondrug treatments.

Patients with FBSS typically have failed to respond to multiple therapies. Few options remain; these most prominently include a new surgical procedure, multidisciplinary rehabilitation, or neuromodulation. Neuromodulation, the reversible and adjustable blockade or manipulation of pain pathways to modify physiological function, may be applied to spinal cord, deep brain structures, motor cortex, and peripheral nerves.⁶ Among all the available neuromodulation techniques, spinal cord stimulation (SCS) and intrathecal drug infusion are the most common and efficacious.

Spinal Cord Stimulation

SCS is an evidence-based therapy that has been used for many years in the treatment of several types of refractory neuropathic pain,⁷ including chronic radiculopathy. As outlined in recent reviews,^7,⁸ the use of SCS in patients suffering from FBSS has been shown to provide (1) a sustained, long-term pain relief, with a reduction in concomitant pain medication; (2) an improvement in quality of life and functional status; (3) an increased ability to return to work; (4) an increased patient satisfaction; (5) minimal side effects; (6) a cost-effective alternative to conventional therapies; and (7) the opportunity for maximized combination therapy with oral and intrathecal methods.

Mechanism of Action

The introduction of SCS was inspired by Melzack and Wall’s gate control theory in 1965.⁹ This theory proposed that painful “nociceptive” information in the periphery is transmitted to the spinal cord in small-diameter, unmyelinated C-fibers and lightly myelinated A-delta fibers, which end in the superficial laminae of the dorsal horn (i.e., the gate) of the spinal cord. Other sensory information such as touch or vibration is carried in large, myelinated A-beta fibers that pass through this gate. As they do, they give off small branches that terminate in the dorsal horn, where they have an inhibitory effect on the nociceptive conduction. The basic premise of the gate control theory was that stimulation of large, low-threshold fibers would close the gate to the reception of small-fiber information. In the first volume of Pain more than 35 years ago, Lindblom and Meyerson¹⁰ reported that SCS increased vibration thresholds and tactile thresholds but did not change the perception of cutaneous pain induced by pinching in five patients responding to SCS, suggesting that SCS affects A-beta fiber function but not C-fiber function.

The general mechanism of pain relief by SCS still is understood in these gating terms, even if there are several substantial problems with this understanding.⁷ For example, both acute and chronic pain should be suppressed by SCS, but this is not the case. It became obvious that not all types of pain are modulated uniformly; whereas SCS primarily affects neuropathic pain and nonnociceptive pain, activation of large afferent fibers can still signal pain,¹¹ and only chronic pain appears to be affected. Thus the mechanism of action of SCS must involve more than a direct inhibition of pain transmission in the dorsal horn of the spinal cord.

Pain modulation by SCS may involve supraspinal activity via the posterior columns of the spinal cord (i.e., down-modulation involving loops or feedback mechanisms that influence rostral transmission of pain from higher centers to the spinal cord). In humans suffering from several types of neuropathic pain, electrical stimulation of the dorsal columns in the thoracic spinal cord generated antidromic action potentials, activating large myelinated fibers in pure sensory peripheral nerves ¹²; and late cellular activation in the dorsal horn after SCS has been also hypothesized, as suggested by an increase in c-Fos immunoreactive cells in the dorsal horn after SCS.¹³ Recent results in an animal model of neuropathic pain showed that the spinal serotonergic system plays a crucial role in the pain-relieving effect of SCS, suggesting an activation of serotonergic descending pathways that may inhibit spinal nociceptive transmission.¹⁴ Moreover, evidences of SCS-induced activation of specific areas of the brain (primary and secondary sensorimotor cortex, cingulate cortex, insula, thalamus, and premotor cortex) have recently been published¹⁵; and regional cerebral blood flow measured by positron emission tomography in patients undergoing SCS therapy for intractable pain was increased in the right thalamus, superior parietal lobule, and left inferior parietal lobule after application of SCS. Thus SCS controls pain cognition by modulating the thalamus and parietal association area, and it also controls the emotional aspects of pain by modulating the prefrontal region.

However, spinal segmental inhibition still seems to be crucial. Animal studies show that second-order afferent nerves and interneurons can be activated by SCS,¹⁶ and a proportion of these may manifest delayed inhibitory activity following brief stimulation.¹⁷ SCS also selectively inhibits abnormal hypersensitivity in dorsal horn neurons¹⁸ and significantly modulates neuronal activity in dorsal column nuclei.¹⁹ It does not directly activate the neurons of the pars gelatinosa, but elicits inhibitory responses of these neurons via A-fibers by reducing excitatory neurotransmitters.²⁰ At the cellular level SCS was shown to stimulate the neurons of the dorsal horn of the spinal cord to release increased amounts of acetylcholine, substance P, serotonin, noradrenaline, glycine, and γ-aminobutyric acid (GABA).²¹^–²³ In a series of acute experiments using microdialysis in the dorsal horn of nerve-lesioned rats, SCS reduces the release of excitatory amino acids (glutamate, aspartate) and at the same time increases the GABA release²⁴; the reduction in excitatory amino acids is prevented by blockade of GABA receptors.²⁵ The state of central hyperexcitability manifested in the development of allodynia after peripheral nerve injury seems to be related to dysfunction of the spinal GABA systems, and it appears that SCS may act by restoring normal GABA levels in the dorsal horn.^25,²⁶ Moreover, rats that were nonresponders to SCS (their mechanical allodynia was not attenuated by SCS) could be converted to responders with intrathecal administration of low doses of baclofen.²⁵ Similar results were obtained in a pilot clinical study in humans in which the SCS effect was enhanced by simultaneous intrathecal administration of low dose baclofen, and the obtained pain relief results appeared to be long lasting.^27,²⁸ Thus experimental and clinical evidences point toward a significant role played by the activation of the GABA system within the spinal cord in the analgesic mechanism of SCS.

SCS is effective on mechanical allodynia in animal models of neuropathic pain.²⁹ The antiallodynic effect was observed after 30 minutes of stimulation, with a complete return to preneuropathy levels, and it lasted up to 60 minutes after SCS cessation. However, a differential antiallodynic effect seems to be related to the degree of severity of the allodynia itself. SCS leads to faster and better pain relief in mildly allodynic rats compared to the more severely allodynic ones.³⁰

Finally, SCS yields a marked sympatholytic effect. This effect is considered responsible for the effectiveness of SCS in peripheral ischemia, cardiac ischemia, and at least some cases of complex regional pain syndrome (CRPS) and FBSS.⁷

Spinal Cord Stimulation: An Evidence-Based Therapy?

Systematic Reviews

In the last 20 years several systematic reviews have been published on the efficacy and cost-effectiveness of SCS in FBSS patients; just two randomized controlled trials (RCTs) have been completed so far. Thus, despite all the efforts to perform a systematic review as accurately as possible, most of the reviews are redundant because of the paucity of the available data.^31,³²

According to one of the last published systematic reviews ³² and based on Guyatt criteria,³³ the recommendation for SCS in an FBSS patient should be rated as 1B/1C (strong recommendation with moderate-to-low quality of evidence), with a caveat that it may change when higher-quality evidence becomes available. The American Pain Society found fair evidence that SCS is moderately effective for FBSS with persistent radiculopathy, although device-related complications are common³⁴; and both the European Federation of Neurological Society and the National Institute for Health and Clinical Excellence (UK) (NICE) guidelines support the effect of SCS in patients with FBSS.^35,³⁶

The last update (January 2009) of the Cochrane systematic review of SCS for chronic pain, the first that included only RCTs for the analysis, still found only two trials that assessed the effects of spinal cord stimulators for CRPS type I and FBSS. Despite the conclusions pointing toward the existence of limited evidence in favor of SCS, authors recognized the paucity of the existing trials and the need for debate about trial designs that will provide the best evidence for assessing this type of intervention.³⁷ So far SCS-induced paresthesia virtually abolishes the possibility of designing a placebo-controlled randomized trial.

SCS is an expensive technique. List prices for SCS systems are not publicly available, but the Association of British Healthcare Industries provided indicative SCS equipment costs: a midrange price based on the average cost of each manufacturer’s best-selling product, a lower cost based on the average cost of each manufacturer’s least expensive product, and an upper cost based on the average cost of the most expensive product. The prices supplied were: SCS system, including neurostimulator, controller, and charger, if applicable, but excluding leads £9282 ($13,445.35), range £6858 to £13,289 ($9,934.09 to $19,249.66); and leads £1544 ($2,236.55), range £928 to £1804 ($1,344.25 to $2,613.17); or £1136 ($1,654.54), range £1065 to £1158 ($1,542.70 to $1,677.41) for surgical or percutaneous implantation, respectively. However, although initially expensive because of the upfront implant costs, SCS proffers improvements in generic health-related quality of life. Of the total mean additional cost of SCS, 15% is offset in 6-months’ time by reducing the use of drugs for pain relief and other non-drug pain treatment.³⁸ At 6 months SCS increases health-related quality of life in patients with chronic back and leg pain with a neuropathic component after one or multiple surgeries at an additional mean health care cost of £11,373 ($16,474.25) per patient.³⁸ Moreover, a cost-effectiveness analysis based on data from another trial indicates that, at a mean follow-up period of 3.1 years, SCS is more cost-effective than reoperation in selected FBSS patients and should be the initial therapy of choice.³⁹ A recent systematic review assessing the cost-effectiveness of SCS in FBSS patients pooled all the available data and confirmed the hypothesis that SCS is both more effective and less costly in the long term.⁴⁰

Clinical Trials

Two RCTs investigated the effect of SCS on the treatment of FBSS. One trial (PROCESS) compared SCS in combination with conventional medical management (CMM) with CMM alone ;⁴¹ interestingly, most of the patients recruited had persistent radicular pain following anatomically successful surgery for herniated disc. The second trial compared SCS in combination with CMM with repeat operation in combination with CMM.³⁹ Follow-up in the PROCESS trial was at 6 and 12 months, with a further analysis published 24 months later⁴²; in the second trial follow-up was at 6 months and after a mean of 2.9 years. The primary outcome in both studies was the proportion of people who had 50% or greater pain relief.

The PROCESS trial reported that SCS had a greater effect than CMM in terms of the proportion of people experiencing 50% pain relief at 6 months (48% and 9% in the SCS and CMM groups, respectively), 12 months (34% and 7% in the SCS and CMM groups, respectively) and 24 months (37% vs. 2%, respectively). The second trial also reported a statistically significant benefit in terms of those experiencing 50% pain relief, favoring SCS over repeat operation (39% and 12% in the SCS and repeat operation groups, respectively). In the PROCESS trial opioid use did not differ significantly between the two groups (56% and 70% using opioids in the SCS and CMM groups, respectively). However, the second trial reported that SCS resulted in a significantly greater number of people reducing or maintaining the same dose of opioids when compared with repeat operation (87% and 58% in the SCS and repeat operation groups, respectively). In the PROCESS trial the SCS group showed a significantly greater improvement in function compared with the CMM group for mean change in functional ability (ODI) and significant benefits in health-related quality of life (SF-36). The second trial reported no statistically significant differences between SCS and repeat operation for pain related to daily activities or neurological function.

However, not all of the evidence is pointing in the same direction. First, an SCS device manufacturer funded both trials, and in the large international trial this manufacturer managed all study logistics and collected and analyzed the data. We should remember that industry-sponsored studies of drugs and devices yield more favorable results than do nonindustry-funded studies.⁴³ Second, in the PROCESS trial the comparator was CMM, even though patients had already failed such treatment and would probably continue to do so.⁴⁴ Third, independent analysis showing little or no evidence that SCS is superior to alternative treatment.⁴⁵

In summary, despite the fact that two quality RCTs have already been performed, a multicenter, independently funded definite trial should still be advocated.

Patient Selection

Patient selection is the key to having positive outcomes.⁷ As with any procedure that has risk, choosing a patient who wants to get better and has an etiology that has been shown to benefit from SCS is a must. Although not all patients are suitable for treatment with SCS, careful patient selection and implantation after evaluation by a multidisciplinary pain management team can offer safe and effective treatment of refractory neuropathic pain.⁴⁶ The search for objective criteria that predict an optimal result using implantable systems must include psychological criteria in the decision algorithm.³⁶ However, at present there is insufficient empirical evidence that psychological screening before surgery or device implantation helps to improve treatment outcomes, even if the current literature suggests that psychological factors such as somatization, depression, anxiety, and poor coping are important predictors of poor outcome.⁴⁷

One of the unique and inherent benefits of SCS is the ability to perform a trial before permanent implantation.⁴⁸ A 50% pain relief reported by the patient is the minimal yet usually the only efficacy measurement that needs to be met during the trial. Unfortunately, this 50% gold standard is purely subjective and is potentially biased by placebo effects. Therefore the need for valid objective test(s) that, along with the subjective report of pain relief, can help the implanter define a “successful” trial is obvious. In a small prospective pilot study of patients with FBSS or CRPS, Eisenberg and associates⁴⁹ showed that some quantitative sensory measures—vibration threshold and tolerance to electrical stimulation at 5 and 250 Hz—were changed with an SCS trial. These changes were also correlated with the decision regarding the permanent implantation, which was made independently of them.

The delay between the first back surgery and the trial of SCS seems to be a significant prognostic factor. In a study of 235 patients treated over 15 years, the success rate of SCS dropped from 93% in patients who had a 3-year delay between surgery and implantation to 9% for those who had a 12-year delay.⁸ Thus SCS should be considered early in the management of FBSS, before a second operation and before the use of high-dose opioid analgesics.⁵⁰

However, despite a proper patient selection some patients do not benefit adequately from SCS, in spite of the fact that stimulation-induced paresthesia covers the painful area.²⁴ In a recent retrospective analysis of a single-center experience with SCS over the past 22 years, Kumar, Hunter, and Demeria⁵¹ reported an 84% early–SCS trial success in FBSS patients, the success rate dropped to 60% in the long-term follow-up (mean 98 months).⁵¹ Interestingly, authors reported a lower failure rate (16%) in FBSS patients with a clear pain etiology and radicular distribution compared to a higher rate (33%) in FBSS patients who did not experience obvious neuropathic pain but had nonspecific pain in their arms or legs that was not in the distribution of any particular nerve root or roots and was not associated with radicular neurological deficits.

During the patient selection process some commonly accepted contraindications should be taken into consideration.²³ Absolute contraindications are (1) sepsis, coagulopathy, or other conditions associated with an unacceptable surgical risk; (2) previous surgery or trauma that obliterates the spinal canal; (3) localized infection at the implantation site; and (4) spinal bifida. Relative contraindications are (1) physical and/or cognitive/psychological disability, (2) unresolved major psychiatric disorder, (3) unmanaged substance abuse or cognitive disorders, (4) pregnancy, and (5) presence of a cardiac pacemaker or defibrillator.

Technical Issues

Briefly, SCS is applied through a subcutaneous electrical generator that delivers pulses by means of electrodes placed in the epidural space adjacent to a targeted spinal cord area presumed to be causing the pain. Definitive electrodes can be placed percutaneously or through a surgical procedure (most of the time, a laminectomy). Although electrodes placed via laminectomy in the thoracic region appear to be associated with significantly better long-term effectiveness,⁵² no definitive evidence has been published so far. However, a percutaneous trial is always required before proceeding to the full implant of the system.^48,^53,⁵⁴

Trial of Spinal Cord Stimulation

The patient is positioned in a comfortable prone position on a fluoroscopy table. A pillow underneath the abdomen may correct lordosis curve and open posterior interlaminar space, which can facilitate electrode insertion. The choice of level of electrode insertion is guided by several factors. A fundamental consideration is that several centimeters of the lead have to lie in the epidural space to ensure maximal stability of the electrode and minimize unwanted migration. To ensure this, insertion must take place at least two spine segments below the desired target.

The epidural space is accessed using median or paramedian approach and angling steeply up from below. It is recommended that the loss of resistance technique be used to access the epidural space since contrast may obscure lead placement and saline flush can decrease the consistency of paresthesias. Some physicians now believe that a single lead is sufficient, and the lead is manipulated up under fluoroscopic guidance to the target level. If dual leads are desired, most are placed parallel but can be staggered according the final electrical field elicited with the different configuration provided with the position of the electrodes. Manipulation of the leads up the epidural space can be a tricky and subtle technique. Several stylets are available with different tip configurations that aid in the placement of the lead to its desired location. Once satisfied with lead positioning (check for the position in the posterior epidural space), the next step is assessing paresthesia coverage.

Ultimately the patient must determine if the paresthesia that occurs in his or her back and/or legs is better than the back and/or leg pain that drove him or her to attempt the trial. During the trial period the patient should be encouraged to log activities, activity durations, and medications and possibly measure pain levels on a VAS equivalent. Pain treatment with SCS therapy is considered effective when a patient experiences a clinically significant (50% or greater) reduction in pain. When this occurs, the patient is then recommended for permanent SCS placement.

Full Implant of Spinal Cord Stimulation

If a two-electrode implant technique is chosen, usually two needles are inserted either on one side of the midline at two different levels or on both sides of midline. The first technique is preferred to perform leads anchoring through a single paramedian incision. The needles are inserted, and the leads are placed under fluoroscopy guidance using techniques similar to those described in the trial procedures.

The skin incision, starting at the stab wound around the needle shaft and extending 5 cm caudally, should be deep enough to expose either the supraspinous ligament or the fascia of the paravertebral muscles.

Several anchors to secure the lead to the ligament or fascia are marketed by different companies. A silicone anchor with percutaneous leads improved average time to mechanical failure compared to a rigid plastic anchor; moreover, supporting the lead with the tip of the anchor as it entered the lumbodorsal fascia improved average time to failure (fracture) by 60-fold compared to the nonsupported condition.⁵⁵ Finally, bonding the lead to the silicone anchor with silicone adhesive was shown to be the only method that reliably prevented slippage of the lead through the anchor during cyclical loading.⁵⁵

The positions of the leads are then documented again by fluoroscopy, and they are tunneled subcutaneously from the midline incision to the exit side where the pulse generator is going to be implanted. The skin incision is then irrigated (with or without antibiotic solution) and closed in layers using absorbable sutures for the subcutaneous tissue and nylon sutures for the skin (e.g., Vicryl sutures to close the subcutaneous tissue and 3.0 nylon sutures to close the skin).

No consensus exists regarding the best site for implanting the pulse generator (buttock, lower abdomen, subclavicular area) or the need of interposing an extension with safety loops between the lead and the implantable pulse generator (IPG) to release tension generated by patient’s movements (especially bending). However, it seems that placement of the IPG in the buttock region may produce up to a fivefold increase in tensile loading compared with placement in the abdomen or midaxillary line, and that the configuration that places the least amount of mechanical strain on the anchor is a coil lead with strain relief loops.⁵⁵

The totally IPG contains a lithium battery, and activation and control occurs through an external transcutaneous telemetry device (for the physician) or through a small portable controller (daily carried by the patient). Life span of the battery varies with usage and with the parameters used (i.e., voltage, rate, and pulse width). Most patients can expect the battery to last, under average usage, between 2.5 to 4.5 years.

Electrical Issues

Understanding the somatotopy of the spinal cord is paramount in understanding the technical aspects of implantation. It is generally agreed that SCS relieves pain only if it induces paresthesias in the area of the patient’s pain, although scientific proof is lacking. To this end, correlation of the somatotopy and the level of the spinal cord is necessary. Barolat and associates ⁵⁶ have published extensively about the mapping of the human spinal structures. A database was created to suggest areas of sensory response to dorsal SCS: the lead contact typically is several levels above the desired area for concordant paresthesia. Moreover, the low threshold of the dorsal root sensory fibers makes it imperative that the lead position is sufficiently midline to avoid recruitment of the root. However, despite an optimally placed electrode, differences in paresthesia coverage between subjects should be expected because of the large interpatient variability of the intraspinal geometry.⁵⁷

It is clear that stimulation on the dorsal aspect of the epidural space creates complex electrical fields that affect a large number of structures.⁵⁴ Electrical stimulation depends on the conductivity of the intraspinal elements in relation to the lead position.¹¹ If a neuron is made more electrically positive or depolarized, it will produce an action potential. Thus the neuron is activated or caused to propagate an action potential. An external electrode that can produce this effect must be negatively charged or a cathode. These effects are called cathodal effects. When the neuron is hyperpolarized or its membrane made more negatively charged, its ability to propagate an action potential is reduced, or its threshold for propagation is raised. A positively charged external electrode or anode produces this anodal effect. Thus the active electrode for stimulation is the cathode, and the anode or positive electrode may shield neuronal structures from the effects of stimulation.

The relative positions of cathodes and anodes and their distance from the spinal cord were demonstrated to be major determinants of axonal activation and paresthesia.¹¹ As the distance between the cathode and anode increases, the influence of the anode diminishes. At larger distances the field about the cathode becomes a sphere, with little influence produced by the anode. Closer contact produces the greatest influence on the anode, significantly shaping the field by pulling it toward the positive contact.⁵⁸ Thus dorsal columns of the spinal cord are most efficiently stimulated with closely spaced longitudinal bipolar (+ −) or tripolar (+ − +) configurations placed on the physiological spinal cord midline because the main current component of the stimulation field corresponds with the orientation of these fibers.^59,⁶⁰

The deepest penetration of the cord without creation of a larger electrical field is produced by a technique termed guarding.¹¹ A tripole, in which there is an anode, is placed in close proximity to each side of the cathode. This prevents the cathodal field from expanding beyond the anode on either side. Thus the anode represents the ultimate boundary for the cathodal field. When a guarded cathode is placed in a longitudinal fashion, as in classic SCS, it produces a field with greater penetration of the spinal cord. On the other hand, a guarded cathode used to stimulate in a transverse fashion (using three parallel leads) has the effect of shielding the nerve root and allowing use of greater amplitude to produce better penetration of the cord, contributing to maximum dorsal column stimulation with minimal dorsal root stimulation and providing analgesia to the lower back.⁶⁰ Left and right anodes can be set at different voltages, and changes in their voltage ratio (“balance”) can steer the electric field from one side of the dorsal columns to the other, thus changing the body area covered with paresthesia.⁶¹ Results from a pilot clinical study showed that the best “steering” score was obtained when the central cathode was >3 mm dorsal to the spinal cord and centered <2 mm from the midline, whereas the electrical field “steering” was impossible when, because of the transverse geometry of the spinal canal, the electrode–spinal cord distance was small compared with the anode-cathode distance (≈3 mm).⁶¹

According to 22 years of experience of a single teaching center in treating patients with a predominance of axial pain combined with either unilateral or bilateral leg pain, staggered, parallel, multicontact electrodes seem to be more efficacious than parallel, nonstaggered configurations.⁵¹ At present it is difficult to be clear about the electrophysiological explanation for the higher success rate achieved with staggered configuration. A possible explanation is that, by staggering the electrodes, the contact separation distance is reduced, providing more extensive paresthesia coverage.

However, when the paresthesia area can be covered with several configurations, it is beneficial for the patient to program a configuration with one cathode and either no or multiple anodes to decrease energy consumption and increase battery lifetime.⁶²

Although how and where the SCS stimulus is applied determines what structures are activated or inactivated,¹¹ no evidence has been published about the best lead configuration for SCS, and little is known about the effects of frequency and pulse width on SCS efficacy. SCS at 4 and 60 Hz seems to be more effective in reducing hyperalgesia than higher frequencies (100 and 250 Hz).⁶³