Description

Anatomy. Pain transmission in the nervous system begins at the level of nociceptive receptors in the peripheral nervous system. These receptors may respond to only mechanical stimulation (high-threshold mechanoreceptors) or to both mechanical receptors and intense temperature. Additional nociceptive receptors may be activated by mediators of tissue damage, such as adenosine triphosphate or glutamate as well as toxins or venoms. Most of these receptors result from small, thinly myelinated (A-gamma) or unmyelinated C fibers with cell bodies in the dorsal root (or trigeminal) ganglia. Approximately half of all receptive neurons in a dorsal root ganglion respond to pain or temperature (05). Central projections from the dorsal root ganglion enter the spinal cord via the dorsal root. Most of these thinly myelinated and unmyelinated fibers enter the tract of Lissauer until they encounter the neurons of the dorsal horn. Unmyelinated fibers traverse fewer segments before synapsing on the dorsal horn nuclei. The unmyelinated C fibers synapse chiefly in lamina II, whereas the thinly myelinated A-gamma fibers may synapse in lamina I, II, and V (05). In lamina I, the connections appear to be more or less specific and well localized, whereas the lamina V fibers tend to be more diffuse and dynamic. Neurotransmitters at this level are thought to include excitatory amino acids, such as glutamate, as well as neuropeptides, such as Substance P, with amino acids thought to regulate faster synapses and with neuropeptides handling slower synaptic transmission (05).

From the dorsal horn cells, several distinct ascending pathways emerge. The best known is the spinothalamic tract, which can be further divided into the neospinothalamic tract and the paleospinothalamic tract. The neospinothalamic tract arises from laminae I and V and projects to the ventral posterior lateral and ventral posterior inferior nuclei of the thalamus. The paleospinothalamic tract arises mainly from lamina V and projects more diffusely to the ventral medial, medial dorsal, parafascicular, and centromedian nuclei among others (05). Other tracts project to the mesencephalon, including the perimesencephalic grey and reticular nuclei. Trigeminal projections go chiefly into the ventral posterior medial nucleus. One thalamic area appears to be dedicated to receiving nociceptive relays from lamina I via the spinothalamic tract. The posterolateral thalamus is dedicated to pain fibers, and microstimulation produces discrete pain and cold sensations that project to the insular cortex and primary sensory cortex (07).

Cortical connections from the thalamus and brainstem are also diffuse. The primary sensory cortex is situated in the postcentral gyrus and receives input from the posteroventral lateral and posteroventral medial thalamic nucleus. Other cortical areas receiving presumed nociceptive input include the insular cortex, secondary sensory area, and anterior cingulate cortex. Shibasaki has shown different roles of these areas in various stages of pain. Although each area is involved in pain perception, basic pain processing is distinct in the sensory area I, higher pain processing involves sensory area II and the insular cortex, and emotional aspects of pain are processed through the anterior cingulated limbic cortex as well as the parietal insula and parietal operculum (101).

Etiology. Prior to Melzack and Wall’s description of the gate control theory of pain, the commonly held notion was that pain receptor signals traveled relatively unabated through the dorsal root ganglia to the dorsal horn and into the thalamus. Melzick and Wall postulated that myelinated (A-alpha and A-beta) ascending fibers were capable of regulating transmission from the unmyelinated C fiber pain and temperature pathways either directly or through inhibitory interneurons. This hypothesis allowed for regulation of pain by non-pain pathways (73). This concept for interactions with the pain pathways resulted in expansion of the theory to include descending cortical influences. This concept was cemented by findings of anesthesia induced by brainstem stimulation and suppressible by lesions of the dorsolateral fasciculus (52).

Breakthrough cancer pain is certainly multifactorial. Svendsen and colleagues suggest that this complex process involves at its genesis mechanical stimuli such as microfractures, leading to changes in local chemical environments and further influenced by tumor-induced release of proinflammatory cytokines or peptides such as tumor necrosis factor. In turn, this chronic change in the neural microenvironment leads to chronic long-term sensitization and hyperexcitability (111).

Tissue injury itself is complex with release of proinflammatory, pain-producing peptides such as cytokines, tumor necrosis factor, bradykinin, adhesion molecules, prostaglandins counterbalanced by the production of endogenous opioids, somatostatin, and anti-inflammatory cytokines (93).

Chessell and associates have suggested a primary role in pain transmission for the proinflammatory cytokine interleukin 1beta (IL1-beta). They specifically looked at the P2X (7) purinoreceptor whose activation releases this cytokine. Knockout mice lacking this receptor showed a limited response in models of chronic pain generation. The receptor was also found to be upregulated in injured nerve and dorsal root ganglia (14).

Central pain syndromes can occur as the result of ischemic insults to the thalamus or brainstem. Kumral and colleagues documented a 9% incidence of central pain in patients with thalamic hemorrhage (63). Brainstem stroke, specifically Wallenberg lateral medullary syndrome, was associated with a 25% risk of long-term chronic pain (67).

Endogenous opioids and extracted or designed opiate alkaloids provide pain relief in the central nervous system. Three subtypes of opioid receptors mediate central anesthesia: delta, kappa, and mu. Delta receptors appear to modulate cutaneous pain; kappa receptors mediate visceral pain; and mu receptors affect both subtypes. In general, opioid binding activates G proteins that, in turn, may open calcium or potassium channels or activate adenylate cyclase, turning adenosine triphosphate into cAMP. The highest concentration of opioid receptors in the spinal cord resides in laminas I, II, and III, although they are present throughout the lamina (42).

The etiology of chronic nerve pain remains somewhat obscure as well. Investigators disagree about the prominence of sympathetic dysfunction in what was once known as “reflex sympathetic dystrophy.” Possible mechanisms for chronic nerve pain include peripheral nerve inflammation via cytokines, changes in the sympathetic reactivity or sodium channel permeability in the dorsal root ganglion, long-term potentiation of dorsal horn cells, and increased expression of alpha2 adrenoreceptors; these mechanisms have all been implicated in the etiology of chronic regional pain syndrome (08). Ramer and Bisby found a significant increase in sympathetic sprouting in the dorsal root ganglion within 4 days of onset in a chronic sciatic constriction model (86).

Several competing hypotheses exist to explain central pain from spinal cord disease or injury. After experimental spinal cord injury, local increases in the concentration of excitatory neurotransmitters, as well as increased tissue levels of multiple cytokines, are associated with the extent of tissue injury. In fact, both ischemic and excitotoxic injuries evoke similar neuronal firing patterns that can be blocked by glutamate antagonists (22). Local spinal inhibitory interneurons are damaged in spinal cord injury. Loss of this inhibitory influence may cause more thalamic excitation as may the effects of long-term depolarization due to tissue injury (128).

Brooks and colleagues have demonstrated changes in the blood-brain barrier in rats with chronic inflammatory pain. The investigators demonstrated edema formation and blood-brain barrier breakdown in animals exposed to chronic pain with downregulation of occludin and upregulation of claudin-3 and claudin-5 (10).

Diagnostic workup. Diagnosis of chronic pain may be relatively straightforward in cases with defined injury and relatively difficult in cases with central or obscure pain patterns. Phantom limb pain may be mistaken for pain arising from poorly fitting prosthetics and vice versa. Change in orthotics may provide a diagnostic as well as therapeutic evaluation. Sympathetic blockade has been hypothesized to be of benefit in the diagnosis of complex regional pain syndrome, but the topic remains controversial (08). For peripheral nerve pain, EMG and nerve conduction studies with synaptic nerve action potentials may be of value in both identifying and localizing the level of an injury. For facial pain, such as trigeminal neuralgia and other facial pain syndromes, MRI is indicated to look specifically for compressive vessels or lesions at the trigeminal nerve root exit nerve. In general, MRI of the brain is helpful in central pain syndromes to look for infarction, tumors, multiple sclerosis plaques, or other causative lesions. Functional imaging has also been explored as a diagnostic tool in individuals with chronic pain. Positron emission tomography and functional MRI have detailed blood flow changes in acute pain in both thalami as well as in the cerebellum, cortex, basal ganglia, and brainstem (19). Functional imaging studies show that in chronic pain, thalamic contralateral blood flow is decreased rather than increased (49).

Clinical applications

Pharmacological treatments

Opioid analgesics. Exogenous opioids exploit endogenous endorphin and enkephalin receptors both peripherally and centrally.

Antidepressants. Most antidepressants work as selective serotonin and norepinephrine reuptake inhibitors in the CNS. Pain relief is obtained at lower dosages than those required for antidepressant effects. There is no evidence that any individual antidepressant or class of antidepressant has superior efficacy (72). Newer agents such as venlafaxine work at both serotonin and noradrenergic synapses with at least comparable efficacy (129).

Anticonvulsants. Almost every drug that has been employed as an anticonvulsant has garnered interest as an agent for managing chronic pain. For many of these agents, mechanisms are more putative than understood. Gabapentin is perhaps the most widely used anticonvulsant agent in current clinical practice for chronic neuropathic pain. Gabapentin exerts excitatory effects at gamma-aminobutyric (GABA) inhibitory synapses in the central nervous system. Gabapentin is also capable of either directly or indirectly inhibiting spinal release of excitatory neurotransmitters aspartate and glutamate (17). Tiagabine is a GABA-selective reuptake inhibitor—thereby, also potentiating GABAergic synaptic activity. Todorov and colleagues found similar effects of these two agents on chronic neuropathic pain and sleep quality (114).

Antiarrhythmics. Antiarrhythmics that block sodium channels modulating NMDA and NK (among other) synapses are postulated to affect spontaneously firing nerves. These agents have been found to be efficacious in both peripheral diabetic neuropathy and peripheral nerve injury (13; 110).

Surgical treatments. Surgical treatments are summarized in Table 1.

Table 1. Surgical Procedures for Chronic Pain

Surgical management of pain. Surgical management of chronic pain presents a confusing and inconsistent literature. The reasons for this difficulty are numerous and well summarized in a comprehensive evaluation of the chronic pain surgical literature by Coffey and Lozano (18). They conclude that the majority of the pain literature is either retrospective or prospective but methodologically flawed. They suggest the need for better-designed prospective trials with better use of randomization, sham stimulation, carefully designed endpoints, longer-term follow-up, and blinding of patients and examiners. With their cogent comments in mind, we have reviewed the literature exploring the evidence and rationale for the various surgical options for the management of chronic pain.

Spinal cord and peripheral nerve stimulation. The gate theory of pain provided a rationale for modulating pain transmission not only by interrupting the firing of unmyelinated C fibers but also by upregulating the firing in inhibitory interneurons and myelinated fibers. Electric stimulation of the dorsal columns was believed to be capable of activating these inhibitory pathways and influencing pain transmission (64). Unfortunately, no single unified hypothesis has satisfactorily explained the mechanism of action of spinal cord stimulation. Linderoth and Myerson provide a comprehensive discussion of potential mechanisms for the efficacy of spinal cord stimulation (64). These include:

Spinal cord stimulation has been studied extensively in terms of clinical utility. Kemler and colleagues performed a randomized trial of spinal cord stimulation and physical therapy versus physical therapy alone for patients with chronic reflex sympathetic dystrophy. The spinal cord stimulation group demonstrated improved pain scores and global perceived effects with no improvement in functional status (58). Allegri and associates looked prospectively at spinal cord stimulation in 170 patients with neuropathic pain, failed back surgery syndrome, and vascular ischemic pain and found an efficacy of 69.9%, with no difference between the three etiologies of pain (02). North and colleagues examined outcomes in patients undergoing laminectomy versus percutaneous placement of electrodes and found initial better success in the patients undergoing direct placement of electrodes after laminectomy. However, over time, successful outcomes were limited in both groups (8 of 24 successful at 2.9 years) (77). The same group documented better results with single as opposed to dual electrodes (76).

North and associates also presented an economic analysis of spinal cord stimulation versus reoperation in patients with failed back surgery syndrome. This group randomized patients with failed back surgery to the two treatment modalities and found a statistically significantly higher rate of crossovers in the surgical patients. Spinal cord stimulation was both more effective and less expensive in their patient cohort (78).

Kemler and colleagues followed a series of patients receiving physical therapy versus spinal cord stimulation plus physical therapy and followed them over 5 years. Interestingly, although the pain-relief advantage afforded by stimulation seemed to recede over time, the patients stated that they were happy with the procedures (59). Similarly, Sears and associates noted that patient satisfaction rates with paddle lead spinal cord stimulation for both complex regional pain syndrome and failed back surgery syndrome (70.6% and 77.8%) were higher than the percentage who reported greater than 50% pain relief (98).

Bala and colleagues performed a meta-analysis of spinal cord stimulator reports and concluded that the procedure was efficacious in terms of both pain control and quality of life. They could not clearly establish an economic benefit although they noted that the papers directly addressing this issue were supportive (04).

Conversely, Hollingworth and associates performed a cost-effectiveness analysis in spinal cord stimulation in a workers’ compensation setting and found that the expense of spinal cord stimulation did not balance lower costs of subsequent care (45).

The efficacy of spinal cord stimulation is not limited to failed back surgery syndrome. De Vos and colleagues reported improvement in pain scores in nine patients receiving spinal cord stimulation for painful diabetic neuropathy (21).

A multicenter, randomized trial comparing spinal cord stimulation to reoperation is ongoing (79).

Spinal cord stimulation is often performed as an outpatient procedure performed with either introduction of percutaneous leads into the epidural space under fluoroscopic guidance or with the placement of paddle electrodes through a single-level laminotomy in an open surgical approach. Complications can include loss of efficacy from lead failure or migration, and infection, which is catastrophic because of the need to remove the hardware, to the extreme of spinal cord injury. Hayak and associates reported on complications in 345 individuals undergoing percutaneous placement of spinal cord stimulators and noted an overall complication rate of 34.6% with hardware failure or migration accounting for 74.1% of complications (39). Twenty-four percent of patients required revision, and 24% had their systems explanted for lack of efficacy or infection. Shamji and colleagues examined 11 studies involving 542 patients and found a lower (12%) rate of revision for lead migration (99). Finally, Petraglia and associates looked specifically at the risk of spinal cord injury with spinal cord stimulation in a database of 8326 patients (84). Percutaneous stimulation was associated with spinal cord injury in 2.35% of patients compared to 1.71% in open placement. Spinal hematoma rates were similar (0.63% to 0.71%) with the two procedures.

Rauck and associates performed a long-term review of spinal cord stimulation in 1289 patients in 79 centers. In this large multicenter trial, 7.6% of devices were explanted with inadequate pain relief in only 2.5% of individuals. Infection rate was less than 1% (89).

Spinal cord stimulation is recognized as a valid and successful treatment for post-laminectomy radicular pain, complex regional pain syndrome, and neuropathic pain. Efficacy has also been claimed for axial low back pain as part of failed back surgery syndrome. Rigoard and colleagues examined the efficacy for back pain individually and found modest success rates, with 13.6% showing greater than 50% improvement at 6 months, although this was still a statistically significant benefit over medical care alone (92). Hara and colleagues performed a trial or burst stimulation versus placebo in patients with failed back surgery syndrome (36). They alternated a burst stimulation protocol with a placebo protocol and found no difference in mean Oswestry disability index reduction. Both groups had improvement, indicating a significant placebo effect. Mean reduction in the treatment group was 10.6 and mean reduction in the placebo group was 9.3.

Although spinal cord stimulator leads remain largely unchanged, newer battery iterations have allowed for different treatment parameters. Fishman and associates compared a differential multiplexed spinal cord stimulation protocol to traditional treatment parameters. Response rates were 80.1% with the differential multiplexed treatment compared to 51% with traditional treatment parameters (25).

The vast majority of spinal cord stimulation procedures have involved lower thoracic placement of either percutaneous or paddle electrodes from T8 through T10 or T11. However, the placement of cervical or cervicomedullary stimulators for chronic skull base or neck pain has also been employed. Concerns exist over the potential for significant spinal cord injury as opposed to the thoracic spine. Deer and colleagues reported on 38 patients at multiple centers with 92% patient satisfaction and clinical success rates characterized as excellent or good by 61.6% of patients over the course of multiple time points up to a year (20). Chivukula and associates reported on 121 patients evaluated for a cervical spinal cord stimulator. Twenty-one failed trial, and 100 underwent permanent implantation (15). Mean pain reduction was noted to be 56% and was greater for extremity pain than neck pain and greater for cervical pain than occipital pain. Twenty-four revisions were necessary. A prospective multicenter trial enrolled 45 individuals with neck and arm pain (03). Epidural leads spanned from C2 to C6, with 86.7% achievement of at least 50% pain relief at 3 months. One epidural hematoma and one infection were noted in the 45 subjects. Pain relief was maintained to the 12-month endpoint.

Farber and colleagues examined spinal cord stimulation in terms of short-term and long-term cost-effectiveness (24). They found that only 4.3% of all patients with failed back surgery syndrome underwent spinal cord stimulator implantation. Overall costs for those patients undergoing stimulator implantation were increased at the time of implantation, but there was a sustained decrease of 68% in costs after initial placement, with an average cost reduction of 40% annually with follow-up out to 9 years.

Peripheral nerve stimulation provides an alternative, but theoretically less invasive, method for treating chronic pain using leads subcutaneously in the area of chronic pain. Yakovlev and associates reported a retrospective study of 18 patients with post-laminectomy syndrome who had benefit from a peripheral nerve stimulation trial and then underwent permanent implantation, with pain reduction sustained over 12 months in 100% (125). Verrills and colleagues reported the outcomes of 100 patients receiving peripheral nerve stimulation for chronic pain, with a 72% rate of sustained improvement demonstrated by a reduction in analgesic use (118). Burgher and associates found average pain relief of 45% in 10 patients with ultrasound-guided placement of leads for both axial neck and back pain (11).

As an offshoot of the utility of peripheral nerve stimulation, these devices have been used percutaneously at the base of the skull to treat occipital neuralgia and, in some instances, such diverse disease as cluster and migraine headache. Huang and colleagues described a series of 102 patients undergoing occipital nerve stimulation for refractory occipital neuralgia and found at least 50% pain reduction in half of the patients (46). However, the endpoint was just at 3 months and long-term revision rates for lead migration were not addressed. Brewer and associates found fair long-term results for chronic migraine and good long-term results for cluster patients in a small study (09). Silberstein and colleagues reported on a class 1 randomized trial of occipital stimulation for migraine, which showed safety but failed to meet efficacy targets (102). Cochrane review of occipital nerve stimulation opined that existing trials were no better than class IV anecdotal data with no demonstrable benefit to date (51).

Intrathecal opioids. Direct titration of opioid peptides into the spinal fluid via implantable programmable pumps represents a sensible strategy to present needed concentrations of exogenous opioids while avoiding systemic toxicity. Eisenach and colleagues have suggested a rationale to explain why intrathecal opioids might engender superior pain control to systemic application. They found that intrathecal but not intravenous administration of opioid peptides induces adenosine formation in the cerebrospinal fluid. Adenosine is a known analgesic agent in the spine, and intrathecal opioid analgesic effects can be reversed by the administration of adenosine antagonists (23).

Multiple studies have addressed the efficacy of intrathecal opioids, either intraspinal or intraventricular. Sallerin-Caute and colleagues treated 159 patients with spinal intrathecal spinal morphine and found good to excellent pain control in 80% at an average of 3 months (95). Karavelis and associates treated 90 patients with intraventricular morphine with a 3-month follow-up and found good or excellent pain control in 82% (56). The utility of intrathecal opiates in nonmalignant pain remains controversial from both ethical and therapeutic issues. A multicenter large study found pain control in 61% of patients with failed back surgery, reflex sympathetic dystrophy, postherpetic neuralgia, and peripheral nerve injuries (82).

Given the current opioid crisis of addiction, overdosing, and illicit prescribing and distribution, the use of direct intrathecal distribution has received renewed traction. Review articles have attempted to address long-term success and safety. Hamza and associates reported on 61 patients referred for trial, 58 of whom received pumps (35). At 3 years, mean pain scores were reduced from 7.47 at baseline to 3.41, with a statistically significant reduction in oral morphine equivalents from 128.9 mg/day to 3.8 mg/day with no escalation in intrathecal dosing (35). Kleinmann and Wolter reported 36 patients with pumps placed for an average of 11 years and found that pain scales stayed well controlled at long-term follow-up with no life-threatening complications (61). Currently, morphine and ziconotide are FDA-approved for this purpose, although multiple other agents and mixtures of opioids and nonopioids have been employed intrathecally with varying success.

The ultimate success of these systems can only be demonstrated in the ability to reduce oral intake of morphine equivalents. Hatheway and associates looked at the short- and long-term cost of oral opioid prescriptions in 389 patients undergoing implantation of intrathecal pain pumps. Fifty-one percent were able to entirely eliminate the use of oral opioids within 12 months, and a 10% to 17% over reduction in annual drug expenditures was demonstrated (38). Another report of 99 individuals undergoing intrathecal pump placement for back and limb pain primarily found total opioid elimination was possible in 68% at 1 month, 84% at 1 year, and 92% at 5 years with successful implantation (12).

Ablative procedures--spinal cord and peripheral. Ablative procedures have been tried at almost every level of pain transmission from the peripheral nerve to the dorsal ganglion, dorsal root, dorsal entry zone, dorsal grey, and the spinothalamic tract. Peripheral neurectomy is technically simple but seldom used because of high recurrence rates and the potential for neuroma formation.

Dorsal rhizotomy has also largely been abandoned because of poor long-term success (65; 81). Van Wijk and associates examined the efficacy of radiofrequency lesions to the dorsal root ganglion as a treatment for chronic pain. Fifty-one percent of patients had pain control at 2 months, with a mean duration of pain control of 44 months (117).

Dorsal root entry zone lesions involve ablation of the dorsal root, zone of Lissaur, and the most dorsal laminae of the dorsal horn. Spaic and associates reported 69% long-term pain relief in patients with central pain (106). Sindou and colleagues reported 66% long-term improvement after dorsal root entry zone lesions in patients with brachial plexus injuries (105).

Spinal cordotomy is designed to selectively ablate the spinothalamic tract contralateral to a painful lesion. Nagaro and colleagues reported results of 45 patients undergoing cordotomy for unilateral cancer pain. Although initial pain control was good, appearance of new pain occurred in 73% of patients. However, the new pain was usually less intense than the preoperative pain (75). Cordotomy can be performed in a traditional open fashion or stereotactically using CT or MRI guidance with comparable results (71).

Cordotomy has been suggested to be a viable option, primarily for pain associated with terminal cancer. Viswanathan and colleagues conducted a small clinical trial of 16 patients with chronic cancer pain, with a greater than 33% reduction in pain seen in six of seven patients after cordotomy and none of nine patients undergoing palliative care (120). A larger trial was subsequently initiated (121). Aljuboori and associates compared cordotomy to intrathecal pump placement and found considerable cost savings from cordotomy (01). Zomers and colleagues reviewed results in 52 patients undergoing percutaneous cervical cordotomy for long-term intractable cancer pain. There was 83% patient satisfaction in the short term maintained over the long term, with median opioid use decreasing from 240 mg down to 55 mg a day after the procedure. Dysesthesias occurred 40% of the time in the short term with mild muscle weakness occurring 11% of the time (131).

Sympathetic pain has been addressed via sympathectomy by various approaches, such as percutaneous blocks, open lumbar or thoracic sympathectomy, and thorascopic sympathectomy. Schwartzman and colleagues reported a 65% improvement in open lumbar sympathectomy (96). Johnson and colleagues reported 67% improvement in patients following thoracic sympathectomy (53).

Motor cortex stimulation. Hyperactivity of neurons in the sensory cortex has been postulated as a rationale for sensory cortex stimulation for chronic pain. However, motor cortex stimulation was found to result in superior control of chronic pain (115). Katayama and colleagues found superior effects of motor cortex stimulation over thalamic stimulation in patients with Wallenberg syndrome (57). The same group demonstrated that morphine resistance and ketamine and thiamylal sensitivity predicted a positive response to chronic motor stimulation in patients with thalamic and suprathalamic poststroke pain syndromes (126). Fregni and colleagues investigated motor cortex stimulation in patients with central pain after spinal cord injury and found an improvement with stimulation compared to sham treatment over a 5-day treatment period. Long-term results were not presented (30). Saitoh and colleagues assessed motor cortex stimulation in various frequencies in patients with cerebral and noncerebral lesions. They concluded that 5 and 10 Hz stimulation was effective in both groups, with greater effects seen in patients with extracerebral lesions (94).

Deep brain stimulation. Although deep brain stimulation has been tried for chronic pain since the 1950s at various locations, subsequent studies have focused on the thalamus and the periaqueductal grey. The thalamus is felt to be a target for pain control in an analogous fashion to spinal cord stimulation for central pain. Namely, blocking or overriding of chronic pain signals either arising centrally or peripherally. Mazars first reported success treating chronic pain, with deep brain stimulation targeted at the posteroventral lateral nucleus of the thalamus in 1960 (70). Stimulation on either the thalamic ventralis caudalis nucleus or the periaqueductal gray results in long-term pain relief in up to 44% of patients with neuropathic pain (thalamic stimulation) and 71% of patients with nociceptive non-neuropathic pain (62). Hamani and colleagues evaluated 21 patients after deep brain stimulation for chronic neuropathic pain. Thirteen had electrodes placed in the ventrocaudalis thalamic nucleus alone, and eight had electrodes placed at both the thalamus and the periaqueductal gray. Out of these 21 patients, nine showed substantial improvement in pain scores from the insertional effect alone without stimulation. One patient even obtained permanent pain relief without stimulation. However, the long-term value of the procedure was limited in that only 13 of the 21 patients had enough benefit to merit pulse generator implantation, and only five patients showed long-term benefits of the procedure (34).

Pereira and colleagues reported a 1-year prospectively accrued trial on 12 patients undergoing thalamic stimulation procedures for either phantom limb pain or brachial plexus avulsion, and they reported persistent pain relief in 11 or 12 patients at 1 year (83). The posteroventral lateral nucleus was used exclusively as the target.

Some studies have attempted to treat refractory cluster headaches with deep brain stimulation. Franzini and colleagues reported using stimulation of the posteromedial hypothalamus to treat patients with cluster headache, trigeminal neuralgia, and aggressive behavior associated with mental retardation (28). They found encouraging results for patients with cluster headache, multiple sclerosis-induced neuralgia but not idiopathic trigeminal neuralgia, and behavioral disorders. Fontaine and associates conducted a randomized trial of hypothalamic deep brain stimulation for cluster headache (26). Interestingly, in a randomized phase with active and sham treatments, there was no benefit of deep brain stimulation; however, in a subsequent open phase in which the stimulators were turned on for a year in all patients, 6 of 11 patients derived long-term benefit from stimulation.

Ablative procedures—central. During attempts to produce disease control with pituitary ablation, multiple investigators discovered a greater effect on central pain than on tumor progression (85). Fracchia and associates examined pain control in over 200 patients undergoing hypophysectomy for metastatic breast cancer and found pain control in about 90% of patients (27). Ramirez and Levin postulated that the paraventricular nucleus is the anatomic basis of this effect with projections from the paraventricular nucleus to lamina I of the spinal cord, the perimesencephalic grey matter, and the nucleus raphe magnus (87).

Radiosurgery has allowed noninvasive selective pituitary ablation in patients with chronic pain. Hayashi and associates initiated a multicenter trial of stereotactic radiosurgery for intractable pain. All six patients with cancer pain improved after a high dose (at least 140 Gy) pituitary radiation, and seven of eight patients with thalamic pain after stroke improved as well. However, by 6 months posttreatment, many patients had recurrent chronic pain (40). Hayashi followed these patients longer and determined that persistent pain relief was seen in 38.5% of patients with thalamic pain syndrome (41).

Radiosurgery has also been employed for chronic pain syndromes. Franzini and associates used a target in the medial thalamus, specifically the central lateral nucleus of the medial thalamus, as a target for radiosurgery in 21 individuals with central pain syndromes (29). They found meaningful pain reduction in 57% of subjects at 3 months and in 33% at last follow up. Overall, 48% pain reduction was seen at 1 and 2 years post-treatment, although this effect was diminished to 19% at 5 years. The treatment worked best for trigeminal pain. Yang and colleagues reviewed the use of radiosurgery for chronic pain in a systematic review (127). They found 12 studies documenting 101 patients with chronic pain syndrome with widely different results, although the overall success rates were similar to deep brain stimulation. Nüssel and associates compared deep brain stimulation and radiosurgery to a newer treatment of high-intensity focused ultrasound to treat central pain and found similar success rates in small numbers of patients (80). They postulated that thermal ultrasound lesions may be safer and more efficacious, although more research and larger trials are necessary.

Viswanathan and colleagues published a review of cingulotomy for refractory cancer pain (119). They revealed eight studies describing stereotactic cingulotomy with success rates ranging from 32% to 83% of significant pain relief, although outcome measures were not standardized. The authors suggested a target 2 cm posterior to the tip of the frontal horn; however, targets among the studies were not standardized. They suggest that cingulotomy is a viable alternative when other procedures have failed or are not viable.

Surgical management of patients with “nontraditional” pain. Most of the work to date with the technologies discussed in this article was derived from patients with pain stemming from neurosurgical procedures or neurologic evaluation. Patients that have failed back surgery syndrome or chronic pain stemming from multiple sclerosis and spinal cord injuries dominate the clinical material. However, additional reports have assessed the techniques described in this article for the management of angina (113), mediastinal pain (33), pelvic pain (54), pancreatitis (55; 60), pain from breast cancer surgery (116), and pain from groin hernia surgery (130).

Specifically, Taylor and associates did a systematic review of clinical trials of spinal cord stimulation for angina and reported similar efficacy and lower cost in relieving angina as compared to coronary bypass surgery (113). Zacest and colleagues showed a long-term benefit in 66.7% of patients undergoing neurectomy for groin pain after hernia operations (130). Many of the other articles cited were single case reports or small anecdotal series. However, application of the technology known to be effective for failed back surgery syndrome and complex regional pain syndrome appears to be promising.

Raslan and associates published a systematic review of neuroablative procedures for cancer pain. The specific goal of the study group was to look at both spinal cord and central procedures for cancer pain. They specifically focused on four types of cancer pain: unilateral somatic nociceptive or neuropathic body cancer pain, craniofacial cancer pain, midline visceral cancer pain, and disseminated cancer pain. All of their recommendations were considered level III, indicating a lack of evidence-based justification, with the exception of percutaneous image-guided cordotomy for patients with unilateral nociceptive somatic cancer pain, which was given a class II recommendation. All other procedures such as DREZ lesioning, rhizotomy, thalamotomy, mesencephalotomy, cranial nerve rhizotomy, nucleus caudalis DREZ lesioning, trigeminal tractotomy, midline myelotomy, and cingulotomy could not reach this level of recommendation. Multicenter randomized trials were advised, with case-control trials as an alternative. The investigators also advised the reporting of morphine equivalents, satisfaction scores, pain scale scores, and other standardized outcome measurements, including cost-effectiveness (88).