Ablative Neurosurgical Procedures for Chronic Pain
Benjamin L. Grannan
Muhamed Hadzipasic
Emad N. Eskandar
Consideration of an ablative procedure for chronic pain typically follows numerous evaluations for other therapeutic options that have either failed or were deemed not suitable for the patient. The decision tree for deciding a patient’s candidacy for surgical ablation hinges on careful consideration of the type of pain present, its mechanism, the anatomic distribution, the desired durability of a cure, and the overall state of health and life expectancy of the patient. This clinical information is then merged with our current framework and understanding of pain signaling pathways to develop a treatment plan with the highest probability of achieving the patient-specific metric for success. The efficacy of ablative procedures depends heavily on our understanding of the physiology of the human pain experience incorporating all elements beginning with the nature of the peripheral stimulus end ending with behavioral and emotional perception of pain. Current theory has evolved from concepts of serial processing defining the “labelled line” theory of pain transmission to more sophisticated, distributed models of a “pain matrix,” in which peripheral nociception feeds into multilayered central circuitry that not only localizes painful stimuli but also incorporates emotional and temporal contexts, elicits top-down control, and changes the way future stimuli are perceived.1 Ablative procedures seek to alter pain processing by removing part of this complex circuitry (Fig. 105.1). In doing so, it is possible to disrupt pain transmission but also to change how it is processed—sometimes leading to initial pain relief followed by recurrence. Because neuroaugmentative procedures incorporating high-frequency stimulation and local analgesic administration have become routine parts of clinical practice, use of neuroablative techniques has decreased. However, several procedures still provide significant value in certain clinical scenarios. In this chapter, we review the indications, anatomy, techniques, and outcomes in dorsal root entry zone lesioning, cordotomy, stereotactic cingulotomy, thalamotomy, and tractotomies within the brainstem.
Dorsal Root Entry Zone Lesioning
INDICATIONS
Lesioning of the dorsal root entry zone of the spinal cord (DREZ procedure) is performed in patients with refractory neuropathic pain of peripheral origin, most commonly resulting from nerve root avulsion or injury within the brachial plexus. The procedure has also been performed in neuropathic pain resulting from spinal cord injury (SCI), spasticity, and postherpetic neuralgia.2,3 Although less typical, it has also been performed in patients suffering cancer-related pain. Because the DREZ procedure disrupts afferent pain signaling at a specific level, patients with pain distributions that are focal, only spanning a few dermatomal levels, are considered better candidates for this procedure compared to individuals with more diffuse pain syndromes.4
ANATOMY AND PHYSIOLOGY
The DREZ procedure is built on the hypothesis that following injury of primary peripheral neurons and subsequent deafferentation, central second-order neurons in the spinal cord become hypersensitive and produce aberrant afferent pain signals. This occurs in the absence of a real painful stimulus and is therefore neuropathic in nature. The cell bodies of the second-order sensory neurons carrying afferent pain information reside in the
substantia gelatinosa (Rexed layers 1 and 2) of the dorsal horn throughout the spinal cord.3 Primary neurons, whose cell bodies reside in the dorsal root ganglion, enter the spinal cord via the DREZ and either immediately synapse in the substantia gelatinosa or travel within Lissauer’s tract 1 to 2 levels above or below the level of entry prior to synapsing.5 The area of synapse lies immediately deep and medial to the entry zone where nerve rootlets enter the spinal cord (Fig. 105.2). Lesioning this area disrupts the abnormal pain signaling of second-order neurons at the associated dermatomal level but does not affect transmission of painful stimuli arising from levels below the lesion that travel within the spinothalamic tract. This is in contrast to cordotomy which specifically targets the ascending spinothalamic tract, thus severing pain transmission of all dermatomal levels whose second-order neurons have already entered and are traveling in the spinothalamic tract. It is important to recognize that the corticospinal tract lies lateral to the dorsal horn and ablations targeted too laterally can result in ipsilateral hemiparesis. Additionally, in cervical levels of the spinal cord, proprioceptive information from the ipsilateral arm runs in the dorsal column just medial to the dorsal horn. Therefore, lesioning aimed too far medially may result in loss of ipsilateral upper extremity proprioception.5,6
substantia gelatinosa (Rexed layers 1 and 2) of the dorsal horn throughout the spinal cord.3 Primary neurons, whose cell bodies reside in the dorsal root ganglion, enter the spinal cord via the DREZ and either immediately synapse in the substantia gelatinosa or travel within Lissauer’s tract 1 to 2 levels above or below the level of entry prior to synapsing.5 The area of synapse lies immediately deep and medial to the entry zone where nerve rootlets enter the spinal cord (Fig. 105.2). Lesioning this area disrupts the abnormal pain signaling of second-order neurons at the associated dermatomal level but does not affect transmission of painful stimuli arising from levels below the lesion that travel within the spinothalamic tract. This is in contrast to cordotomy which specifically targets the ascending spinothalamic tract, thus severing pain transmission of all dermatomal levels whose second-order neurons have already entered and are traveling in the spinothalamic tract. It is important to recognize that the corticospinal tract lies lateral to the dorsal horn and ablations targeted too laterally can result in ipsilateral hemiparesis. Additionally, in cervical levels of the spinal cord, proprioceptive information from the ipsilateral arm runs in the dorsal column just medial to the dorsal horn. Therefore, lesioning aimed too far medially may result in loss of ipsilateral upper extremity proprioception.5,6
TECHNIQUE
Under general anesthesia, the patient is placed in the prone position with rigid head fixation. Multiple laminotomies are made to provide adequate access to the spinal cord level of interest and one to two spinal cord levels above and below. A midline durotomy is performed, and the posterolateral sulcus on the affected side is identified. In instances of nerve root or brachial plexus injury, nerve root atrophy can often be observed and used to identify the appropriate level for lesioning. This can be confirmed after placing the radiofrequency ablation (RFA) probe (e.g., Nashold electrode) in the DREZ and measuring an impedance value between the DREZ of the spinal cord and a peripherally placed grounding electrode. Injured areas have markedly decreased impedance values. This measurement will also serve as a preablation baseline (typically 900 to 1,200 ohms) because the impedance will also decrease after lesioning is performed. To perform the ablation, the Nashold electrode is placed 1 to 2 mm into the area of the entry zone at an angle 20 to 30 degrees medial to lateral relative to the plane perpendicular to the spinal cord surface.7 Each RFA is then carried out by powering to a temperature of 75° C for 15 to 20 seconds. A total of 40 to 60 lesions centered around the level of interest are performed.8 Microsurgical technique as well as laser ablation serve as alternatives to RFA.
OUTCOMES
Patients experiencing the most favorable outcomes following the DREZ procedure are those suffering from neuropathic pain related to brachial plexus avulsion injury. A review of the published clinical studies that reported greater than 20 patients identified 62% to 91% patients suffering from brachial plexus avulsion pain to have “good improvement” in symptoms after the surgery,9,10,11,12,13 where “good improvement” was defined as the need for only minimal ongoing medical management to maintain satisfactory pain relief.5 The knowledge surrounding the durability of pain control is limited by the length of follow-up in the published studies but has been confirmed to last, in some instances, up to 3 to 5 years. One particular study performed a subgroup analysis to evaluate the effect of intraoperative electrophysiology mapping of the DREZ on outcome and found improved pain control in the group that received intraoperative mapping.12 Patients undergoing DREZ for SCI-related pain can expect pain relief of localized, end-zone pain occurring just at or above the level of sensory loss. More diffuse and distal pain phenomenon below the level of injury, however, is not as well managed, with efficacy rates falling
to 20% or lower.5 To improve efficacy of the DREZ procedure in SCI patients, Falci et al.14 have suggested using intramedullary electrophysiologic guidance to identify regions of spontaneous DREZ hyperactivity to identify optimal lesion placement. In a group of 32 patients who underwent this approach, 84% achieved 100% pain reduction. Pain control following the DREZ procedure in patients suffering from postherpetic neuralgia is felt to be only transient, with only approximately 25% of patients experiencing relief at 1 year.2 Outcomes are largely comparable for RFA versus laser or microsurgical technique.15 However, many surgeons utilize the RFA technique because of the reproducible lesions provided by the Nashold electrode.5
to 20% or lower.5 To improve efficacy of the DREZ procedure in SCI patients, Falci et al.14 have suggested using intramedullary electrophysiologic guidance to identify regions of spontaneous DREZ hyperactivity to identify optimal lesion placement. In a group of 32 patients who underwent this approach, 84% achieved 100% pain reduction. Pain control following the DREZ procedure in patients suffering from postherpetic neuralgia is felt to be only transient, with only approximately 25% of patients experiencing relief at 1 year.2 Outcomes are largely comparable for RFA versus laser or microsurgical technique.15 However, many surgeons utilize the RFA technique because of the reproducible lesions provided by the Nashold electrode.5
The most common adverse effect of the DREZ procedure is paresthesia, which is usually transient but can be permanent in a minority of cases. This is estimated to incidence of transient or permanent paresthesias, that is, 15% to 30%.4 Transient or temporary motor weakness is also a known risk of the procedure which results from an ablation field that extends laterally beyond the DREZ to involve the corticospinal tract. Incidence rates of permanent weakness vary widely, with the majority of studies citing less than 10% risk, with some studies citing up to 40% to 60% according to a review by Konrad.5
Cordotomy
INDICATIONS
Destruction of the anterolateral column, or spinothalamic tract, for treatment of chronic pain, was first reported in 1912 by Spiller and Martin.16 The procedure, referred to as cordotomy, has evolved since that time but exists today in its same conceptual form, with the goal of eliminating the transmission of afferent pain signal by ablating the spinothalamic tract. The most appropriate candidates are those who have localized, unilateral pain that is nociceptive in origin.17 The most typical, current-day indication would be in the setting of intractable unilateral pelvic, flank, or chest wall pain from metastatic cancer. Cordotomy has become less commonly performed due to the rise of opioid therapies and intrathecal pain pumps for palliation but should continue to be considered in situations in which these initial therapies are ineffective or inappropriate in the setting of the patient’s wishes or life expectancy. Additionally, it is important to note that chronic cancer-related pain can develop a neuropathic component due to actual injury to nerve fibers, and it is important to counsel patients regarding the likely persistence of the neuropathic component after cordotomy.17 The surgery is either performed percutaneously at the C1/2 level with the patient awake to provide sensory feedback or via an open surgical approach at an upper thoracic level with the patient under general anesthesia in the prone position. For percutaneous cervical technique, the patient must be able to tolerate lying flat and still for approximately 45 minutes. For thoracic open cordotomy, the patient must not have other active cardiac, pulmonary, or hematologic medical problems that would contraindicate an open surgery in the prone position.18
ANATOMY AND PHYSIOLOGY
The anatomic target of the cordotomy is the spinothalamic tract within the anterolateral quadrant of spinal cord at the high cervical level (C1/2) or mid-to-upper thoracic level (T4) on the side contralateral to the patient’s pain.19,20 Pain-mediating afferent fibers enter the spinal cord through the DREZ on the side of the painful stimulus before synapsing in the substantia gelatinosa of the dorsal horn and decussating via the anterior commissure and joining the ascending fibers of the spinothalamic tract contralateral to the side of pain. The decussation of fibers for a given nerve root entry level can occur over 2 to 5 spinal levels.17 Therefore, the cordotomy lesion ought to be sufficiently rostral relative to the level of pain in order to fully disrupt nociceptive transmission. As general rule of thumb, allowing for per-patient variation, a C1/2 cordotomy tends to result in analgesia at C5 and below, whereas a T4 cordotomy will achieve analgesia at T10 and below.20 The somatotopic arrangement of the spinothalamic tract lends itself to targeted lesioning when percutaneous RFA is performed. The ascending spinothalamic tract accrues fibers decussating from their contralateral entry at its medial-most aspect. This leads to the sacral representation being most dorsal, and lateral and the cervical representation being most ventral and medial, with lumbar and thoracic information arranged accordingly in between. This somatotopy is demonstrated in Figure 105.2.
As a result, in patients with predominant hemipelvis or unilateral lower extremity pain, it is possible to selectively lesion sacral and lumbar fibers while preserving the cervical and thoracic domains of the spinothalamic tract.17
The operating surgeon must be keenly aware of critical structures in near proximity to the lesion target. Anterior horn cells controlling level-specific motor function lie deep and medial to the spinothalamic tract throughout the spinal cord and are at risk of being involved in the field of ablation. Therefore, C1/2 and midthoracic levels are preferred because loss of anterior horn cells at these levels is not associated with significant morbidity. However, high cervical lesions near the cervicomedullary junction can involve decussating fibers of the cortical spinal tract and pose the risk of causing contralateral weakness of the lower extremity. Importantly, interneurons involved in respiratory control reside within the upper cervical levels and are located within the ventromedial cord in near proximity to the spinothalamic tract. Because of this, bilateral cervical cordotomies must be judiciously considered in order to avoid the life-threatening sleep-related apnea phenomenon, commonly referred to as Ondine’s curse.17
TECHNIQUE
Percutaneous cervical cordotomy is performed under either fluoroscopic21 or computed tomography (CT)-guidance.17 The patient is placed in the supine position with the head maintained in slight flexion either with a fixation band or rigid frame fixation depending on surgeon preference. A combination of local anesthetic and light sedation is provided for anesthesia. The C1/2 interspace is identified using lateral x-ray or CT imaging. In the case of fluoroscopy or x-ray technique, a 20-gauge spinal needle must first be introduced into the subarachnoid space in order to inject contrast to aid in identification of the dentate ligament, which is the critical landmark in the anterior-posterior dimension for safe localization of the spinothalamic tract. CT myelogram with contrast injection through either lumbar or cervical site provides multiplanar acquisition of imaging to more confidently identify the needle or cannula trajectory with respect to the dentate ligament and spinothalamic tract. Once the dentate ligament is identified, the electrode cannula with stylet is introduced into the spinal cord with a goal target of 1 to 2 mm anterior to the dentate ligament for sacral and lumbar fibers and 2 to 3 mm anterior to the ligament for thoracic and cervical fibers. Repeat CT imaging may be obtained to confirm appropriate cannula positioning. The stylet is then removed, and the noninsulated electrode (2.5 mm in length by 0.25 mm in diameter) is introduced into the spinal cord through the cannula. The electrode impedance is tested to confirm intraparenchymal placement (greater than 800 ohms).17,21 Stimulation of 0.2 to 1.5 mV can be provided at 100 Hz to obtain neurophysiologic confirmation of electrode placement. The patient will likely experience dysesthesias in the form of inappropriate temperature sensation. Once there is satisfactory confirmation of correct electrode positioning, ablation is performed by heating the tissue to 70° C to 80° C for 60 seconds. Depending on the extent of the patient’s pain symptoms, multiple lesions may be performed. Postprocedurally, the patient is monitored closely either in perioperative recovery unit or in the intensive care unit for at least 4 to 6 hours prior to less intensive monitoring or discharge.
Open thoracic cordotomy is performed in the operating room with the patient in the prone position. A bilateral or hemilaminectomy is performed and a durotomy is made to provide access and direct visualization of the spinal cord. The dentate ligament is identified and separated from the dura to facilitate gentle upward rotation of the spinal cord to provide the surgeon access to the anterior spinal cord. An angled cordotomy knife is then inserted anterior to the dentate ligament to a depth of 3 to 4 mm and swept anteriorly to ensure complete disruption of the spinothalamic tract. It is essential that the surgeon respect the dentate ligament as the posterior boundary in order to avoid injury to the corticospinal tract.5 The dura is then closed in a watertight fashion in order to prevent cerebrospinal fluid (CSF) leak. Figures 105.2b and 105.2c demonstrate the differences in approach and the expected lesion size for percutaneous and open cordotomy. Of note, the ventral spinocerebellar tract is more likely to be involved in the open lesion which may contribute to postoperative ataxia.
OUTCOMES
Compared to other ablative procedures for pain, cordotomy is well studied with over 3,600 patients reported in the literature, including one prospective study.22 In general, as reviewed by Konrad,5 immediate pain relief following percutaneous cordotomy occurs in approximately 90% of patients with rates of pain relief falling to 50% to 60%, 1 year after procedure. Similarly, open cordotomy provides high rates of immediate pain relief (up to 93% reported by Cowie and Hitchcock23) which fall to 54% to 65% after 1 year follow-up.23,24,25 Therefore, cordotomy remains an impactful option for patients with severe pain due to advanced metastatic disease. Its efficacy has also been studied in cases of chronic pain from nonmalignant etiologies. In a study of 122 patients who underwent percutaneous cervical cordotomy, 27 suffered from nonmalignant sources of pain. In this subgroup, only 20% experienced complete pain relief compared to a rate of 66% in cancer patients. Cordotomy, therefore, is felt to be most effective for cancer-related pain but still remains an option for patients suffering from intractable pain due to SCI or other etiologies, in which all other options have been exhausted.
The most common adverse outcomes reported following cordotomy include spinal headache from dural puncture, short duration of “mirror pain” during which time symptoms similar to preoperative pain are experienced on the contralateral side of the body, dysesthesias, ataxia, temporary leg weakness, and temporary bowel and/or bladder dysfunction.18 Estimated incidences of these side effects are reported to be in the range of 10% to 34%, with ataxia and bladder dysfunction being most commonly reported. Serious adverse events including respiratory impairment and even associated death have been reported even in unilateral procedures. Ischia et al.26 cites a 0% to 9% risk of mortality rate in unilateral procedures and 11% risk of death in bilateral cervical cordotomy. It is generally felt that the majority of complications arise from poor electrode placement and that the rise of CT guidance has led to decreased rates of severe complication.17 Open cordotomy comes with the increased risk of CSF leak due to larger durotomy, but the ablation-related neurologic side effects discussed are largely similar in both percutaneous and open approaches.5
Cingulotomy
INDICATIONS
Although cingulotomy was first performed for psychiatric disease by Sir Hugh Cairns in 1948, it was not until over a decade later in that the first series on cingulotomy for intractable pain was reported in 1962 by Foltz and White at the University of Washington.27,28 Motivated by significant reduction of intractable pain in patients who had undergone prefrontal lobotomy in combination with animal studies implicating the frontal cingulum fasciculus as the critical white matter tract involved generating the emotional valence of pain,29,30,31 Foltz and White28 sought to employ stereotactic cingulotomy as a refined ablative approach for pain control. Rather than enrolling patients based on a specific etiology of intractable pain, they worked with psychiatry colleagues to select patients who “showed prominent emotional factors” such that the cingulotomy might “modify the patient’s emotional response … so that his expressions of fear and anxiety no longer augmented critically whatever pattern of organic pain was present.”28 Following their initial success with their report describing 11 of 16 patients experiencing good pain relief, many other groups explored the role of anterior cingulotomy for chronic medically refractory pain, including applications for both neoplastic and nonneoplastic sources of pain. In recent years, however, the rise of implantable intrathecal pain pumps and neuromodulatory devices, anterior cingulotomy has become less popular.32 However, anterior cingulotomy continues to offer a widely applicable approach for severe, intractable pain, especially in patients predominantly featuring an affective component to their pain. It may also be most appropriate in patients who are not candidates for implantable devices due to barriers to frequent follow-up care or due to short life expectancy in the setting of end-stage malignancy.
ANATOMY AND PHYSIOLOGY
General pain perception has been described as being divided into two components: a lateral pain system involved in spatial discrimination of pain and a medial pain system involved in the affective and autonomic response to pain.33 The anterior cingulate is a main component of the medial system. It is located mesial, ventral, rostral, and dorsal relative to the genu of the corpus callosum and has broad connections to many different sites involved in pain processing. These include the medial frontal cortex, thalamic nuclei (anterior, midline, interlaminar, and others), septum, amygdala, basal ganglia, nucleus accumbens, and brain stem sites such as the periaqueductal gray.34 The anterior cingulate receives nociceptive input and has connections with several sites involved in the affective and attentional component of pain which is supported by a broad range of data. Foltz and White,28 for example, noted changes in comfort and affect in several patients immediately following lesioning in the operating room. Additionally, functional imaging studies have shown bilateral response within the anterior cingulate in response to pain.35 Furthermore, intraoperative single neuron recordings in humans undergoing anterior cingulotomy for psychiatric disease have demonstrated modulated activity in response to thermal and mechanical noxious stimuli. Those same neurons did not respond to nonnoxious stimuli.36 Furthermore, stimulation studies in monkeys and humans have been able to induce responses of fear or happiness in response to anterior cingulate stimulation.33