Chapter 92. Interventional Management of Chronic Pain

1. The best pain medicine practitioners strike a reasonable balance between interventional and noninterventional management. This practice pattern is sustainable, and those adopting a balanced style of practice will be able to adapt to evolving scientific evidence that appears to support pain treatment, regardless of the type of treatment.

2. Although the evidence supporting the need for routine radiographic guidance is still evolving, the intuitive appeal of this more precise approach has evolved to the point where most practitioners now perform at least a portion of their injections using fluoroscopic guidance.

3. The key to safety and success of any interventional pain technique is a clear understanding of normal anatomy. The procedures described in this chapter require an understanding of the normal anatomy of the spine, including the epidural and subarachnoid spaces, zygapophyseal joints, intervertebral disks, and, most importantly, the spinal cord with its somatic and sympathetic components.

4. Epidural steroid injections are efficacious in the treatment of acute lumbosacral radicular pain or in radiculopathies secondary to herniated or bulging disks. Epidural steroid injections also have been used to treat back pain secondary to degenerative disk disease, spinal stenosis, trauma, spondylolysis or spondylolisthesis, and in pain following laminectomy. The epidural space can be approached through the interlaminar space (median or paramedian), intervertebral foramen (transforaminal), or sacral hiatus (caudal). The approach selected depends on patient selection, indication for injection, practitioner's experience, and availability of imaging. We are still lacking large-scale studies comparing clinical outcomes with the transforaminal versus the interlaminar approach.

5. Many practitioners continue to use sympathetic blockade as a part of a multidisciplinary approach to treating complex regional pain syndrome. Sympathetic blocks are one tool that can reduce pain and facilitate functional recovery.

6. Intra-articular facet injection has been largely supplanted by radiofrequency treatment techniques for facet-related pain. Clinical experience and a limited number of published observational studies suggest that intra-articular injection of local anesthetic and steroid leads to relief of facet-related pain that is of limited duration. In contrast, radiofrequency treatment is safe and effective in producing longer-term pain relief in the same group of patients.

7. Discography is a diagnostic test in which radiographic contrast is injected into the nucleus pulposus of the intervertebral disk. Although originally developed for the study of disk herniation, discography now is used most commonly to identify symptomatic disk degeneration. The usefulness of this diagnostic test remains controversial.

8. Intrathecal morphine and other opioids are now widely used as adjuncts in the treatment of acute and chronic pain, and a number of agents show promise as analgesic agents with spinal selectivity. Patient selection for intraspinal pain therapy is empiric and remains the subject of debate. In general, intrathecal drug delivery is reserved for patients with severe pain that does not respond to conservative treatment.

9. Direct electrical stimulation of the dorsal columns, known as spinal cord stimulation or dorsal column stimulation, has proven effective, particularly for treatment of chronic radicular pain. Patient selection for spinal cord stimulation is empiric and remains the subject of some debate. In general, spinal cord stimulation is reserved for patients with severe pain that does not respond to conservative treatment. The pain responds best when relatively well localized because success of spinal cord stimulation depends on the ability to cover the entire painful region with electrical stimulation.

The subspecialty of interventional pain medicine involves minimally invasive treatments and minor surgery as part of pain care, including neural blockade and implantable analgesic devices. Despite the paucity of scientific evidence to guide pain practitioners, a number of techniques appear to have efficacy based on limited observational data and have been adopted into widespread use. As practitioners we are left to choose among available treatment modalities, often with only anecdotal evidence and personal experience to guide us in treating a group of desperate patients with intractable pain who are willing to accept almost any treatment, even those that remain unproven. There is no single practice pattern that a pain specialist can point toward as the correct way to treat patients with chronic pain. Training programs vary widely in the scope of what they train practitioners to do. The best pain medicine practitioners strike a balance between interventional and noninterventional management. This practice pattern is sustainable, and those adopting a balanced style of practice will be able to adapt to evolving scientific evidence that appears in support of pain treatment, regardless of the type of treatment. A balance between treatment modalities also allows practitioners to switch from one mode to another or to incorporate multiple treatment approaches simultaneously. Although this chapter focuses on interventional treatments used to treat chronic pain, use of these interventional modalities is just a small part of the armamentarium of the skilled pain practitioner.

A little more than a decade ago, radiographic guidance was used infrequently by pain practitioners; it was reserved for use in major procedures such as neurolytic celiac plexus block. During the last several years, 2 forces have been at work. First, pain practitioners are now being called on to serve as diagnosticians. Patients and referring practitioners expect pain physicians to have familiarity with imaging modalities and their usefulness in diagnosing pain conditions. At the same time, pain practitioners have come to realize the usefulness of radiographic guidance in achieving precise anatomic placement of needles and catheters. Although the evidence supporting the need for routine radiographic guidance is still evolving, the intuitive appeal of this more precise approach has evolved to the point where most practitioners now perform at least a portion of their injections using fluoroscopic guidance.1 In some cases (eg, patients with intractable pain associated with metastatic cancer), radiographic guidance has proven invaluable in the planning and implementation of therapy directed toward pain relief.

Some years ago, our research group examined the distribution of injectate in a series of patients who received epidural steroid injections for radicular pain associated with new herniated disks.2 We found that the injectate often spread to the side opposite the disk herniation. This finding is not at all surprising. A disk herniation present on one side might well obstruct the flow of fluid through the relatively confined epidural space. The fluid would follow the path of least resistance, spreading preferentially to the contralateral unaffected side and exiting the contralateral intervertebral foramina. This study and others3,4 have challenged the conventional wisdom that suspending the steroid in a modest volume is sufficient to consistently produce spread of the injectate to the affected levels, regardless of where the solution is placed within the epidural space. Using this line of reasoning, investigators have concluded that the blind loss-of-resistance technique is not the best way to deliver steroid to the site of inflammation.3,4 Using radiographic guidance, bony structures can be visualized directly and in real time. The needle can be seen within the radiographic field, and simple geometry can be used to guide the needle directly from the skin's surface to its destination.

More recently, ultrasound-guided nerve blocks have become commonplace for providing regional anesthesia for surgery, and multiple techniques have been described.5-8 Growing experience with ultrasound among anesthesiologists, as well as improvement in image resolution and processing, have sparked the interest of pain physicians to learn more about the usefulness of ultrasound for pain procedures.9,10 Proponents of ultrasound-guided pain procedures point to advantages such as reduced radiation exposure for patients and providers and improved visualization of soft tissues, vessels, and nerves. Most of the experimental work published to date consists of cadaver studies and case reports describing various techniques.11-14 In the only available randomized controlled trial, Galiano et al randomized 40 patients with low back pain to undergo facet joint injections guided by ultrasound versus computed tomography.15 The duration of the procedure and radiation dose was significantly less in the ultrasound group; however there was no difference in pain relief between the groups in this small study.15 The data available to date are scant, and although the technology is appealing, clinical trials are needed to investigate the efficacy and safety of ultrasound-guided pain procedures.16,17 For now, radiographic guidance remains the standard means for conducting most interventions in the pain clinic.

This chapter provides an overview of the most common techniques used in interventional pain medicine. The section begins with a discussion of the anatomy relevant to image-guided interventions for treating chronic pain. Then the clinical usefulness of each technique and the technical aspects of conducting each intervention are described. Illustrations for the most common techniques are provided. We have published detailed technical descriptions of these and other pain treatment techniques elsewhere.18

Anatomy Relevant to Image-Guided Intervention for Chronic Pain

The key to safety and success of any interventional pain technique is a clear understanding of the normal anatomy. The procedures described in this chapter require understanding of the normal anatomy of the spine, including the epidural and subarachnoid spaces, the zygapophyseal (facet) joints, intervertebral disks, and, most importantly, the spinal cord with its somatic and sympathetic components. The basic anatomy relevant to common interventions used in the treatment of chronic pain is reviewed here.

Anatomy of the Spine

The spinal cord is protected within the vertebral canal and extends from the foramen magnum to the first or second lumbar vertebra in human adults. Its is covered by 3 meninges. The most internal pia mater lies in close apposition to the cord. It is separated from the thin arachnoid mater by the free-flowing cerebrospinal fluid (CSF). The dense dura mater lies most external, surrounding the arachnoid and accompanying the segmental nerve roots well into the vertebral foramen. The epidural space lies within the bony vertebral canal surrounding the dura matter and spreads from the base of the skull to the sacrococcygeal membrane.

Normally, the vertebral canal is nearly triangular, surrounded by the bony components of the vertebrae. There are 33 vertebrae: 7 cervical, 12 thoracic, 5 lumbar, 5 fused elements that make up the sacrum, and 4 to 5 fused ossicles that form the coccyx (Fig. 92-1).19 A typical vertebra consists of a vertebral body and 2 pedicles that extend posteriorly surrounding the spinal canal and epidural space to join a pair of arched laminae (Fig. 92-2). The laminae fuse in the midline to form a dorsal projection called the spinous process. Near the junction of the pedicles and the laminae are found the lateral projecting transverse processes, the superior and the inferior articular processes (zygapophyses or facets). The pedicles and their articulating processes form the superior and inferior vertebral notches. In the articulated spine, these notches form the intervertebral foramina.19

The zygapophyseal, or "facet," joints are paired structures that lie posterolaterally on the bony vertebrae at the junction of the lamina and pedicle medially and the base of the transverse process laterally. The facet joints are true joints, with opposing cartilaginous surfaces and a true synovial lining. They are subject to the same inflammatory and degenerative processes that affect other synovial joints throughout the body.19 Two opposing articular surfaces compose each facet joint. The facet joint articular processes are named for the vertebra to which they belong. Thus each vertebra has a superior articular process and an inferior articular process. This nomenclature can be confusing because the superior articular process of a given vertebra actually forms the inferior portion of each facet joint.

The intervertebral disk is composed of glucosaminoglycans with a relatively fluid inner nucleus pulposus surrounded by a stiff, lamellar outer annulus fibrosus.19 With aging, hydration of the intervertebral disks declines, leading to loss of disk height and fissure formation in the annulus fibrosus. These fissures begin centrally near the border between the nucleus pulposus and the annulus fibrosus and can extend to the periphery of the disk space. This process of degradation is called internal disk derangement and is believed responsible for producing discogenic pain. The annulus contains neural elements from the sinuvertebral nerve, which are believed to be responsible for pain transmission.20 These same radial fissures within the annulus represent paths through which nuclear material can pass and extrude as a herniated nucleus pulposus. When this extruded material is adjacent to a spinal nerve posterolateral to the intervertebral disk, it can lead to intense inflammation, nerve root compression, and radicular pain with or without radiculopathy. Radicular pain is common; radiculopathy of nerve root dysfunction in the form of numbness, weakness, and/or loss of deep tendon reflexes is less common. The paired facet joints, along with the vertebral bodies and intervertebral disks, form the 3 weightbearing support columns that distribute the axial load on the vertebral column while allowing for movement. The structure of the vertebrae varies from cephalad to caudad and should be thoroughly familiar to the practitioner, especially when using imaging (see Figs. 92-1 and 92-2). Of importance when performing injections, the spinous processes of the cervical and lumbar regions approach the lamina in a nearly perpendicular fashion, which facilitates a midline approach when performing epidural or subarachnoid injections. The cervical facet joints are oriented nearly parallel to the axial plane where the atlas (C1) articulates with the occiput and become gradually more steeply angulated in a cephalad to caudad direction at lower cervical levels. The orientation of the cervical facet joints in a plane close to the axial plane allows for a great degree of rotation of the neck as well as flexion and extension.

The midthoracic (T5-T9) spinous processes are acutely angled caudad, making the midline approach to the epidural space more difficult than the paramedian approach. The thoracic facet joints are so steeply angulated that they approach the frontal plane, which makes intra-articular injection difficult or impossible. At midthoracic levels, the inferior articular process of the vertebra forming the superior portion of each thoracic facet joint lies directly posterior to the superior articular process forming the inferior portion of each joint. This allows for some degree of flexion and extension but limited rotation of the spinal column in the thorax region.

The spinous processes of the lumbar vertebrae approach the lamina in a nearly perpendicular fashion. The lumbar facet joints are angled with a somewhat oblique orientation, allowing for flexion, extension, and rotation that is greater than that in the thorax but less than in the cervical region. The sacral hiatus is the area where the fifth sacral vertebra lacks both the laminae and the spinous process posteriorly (Fig. 92-3). The 2 sacral cornua lie on either side of the sacral hiatus and cephalad to the coccyx and are useful landmarks when performing an epidural from a caudal approach.

The midline distance from the ligamentum flavum to the dural sac varies considerably, depending on the vertebral level, increasing from 2 mm at C3-C6 to 5 to 6 mm in the midlumbar region.21 The epidural space narrows posterior and laterally toward the intervertebral foramina. The anterior boundary of the epidural space is provided by the posterior longitudinal ligament covering the vertebral bodies and the intervertebral disks. Posteriorly, the epidural space is limited by the periosteum of the anterior surfaces of the laminae, the articulating facet processes (zygapophyses), and the ligamentum flavum. Laterally, the pedicles and the intervertebral foramina limit the epidural space.

Knowledge of surface anatomy enhances safety when performing procedures under image guidance and is an absolute necessity when imaging is not available. Important surface landmarks include the spinous process at C7 (vertebra prominens), which is the most prominent cervical spinous process palpable when the neck is flexed. The spinous process at T7 lies opposite the inferior angle of the scapula when the arm is at the side. A line joining the superior aspects of the iliac crests passes through the spinous process of the fourth lumbar vertebrae. The spinal cord generally terminates at the L2 level and the dural sac ends at S2, which corresponds to the level of the posterior superior iliac spines. The tip of an equilateral triangle drawn between the posterior superior iliac spines and directed caudally overlies the sacral cornua and sacral hiatus.

Additionally, the articulated spine is supported by the anterior and posterior longitudinal ligaments, the supraspinous and interspinous ligaments, and the ligamentum flavum.21 Many of the procedures discussed here make reference to the ligamentum flavum. The ligamentum flavum, or "yellow ligament," is a structure of variable thickness and completeness, composed of elastic fibers, that defines the posterolateral soft tissue boundaries of the epidural space. Because its leather-like consistency resists active expulsion of fluid from a syringe, loss of this resistance as a needle passes through the ligamentum flavum is valuable in signaling entry into the epidural space. The ligament's structure is steeply arched and tent-like, so much so that the lateral reflection may be up to 1 cm deeper than at the midline. In the cervical and thoracic epidural space, the ligamentum flavum often does not fuse in the midline, which can become problematic during loss-of-resistance techniques. When the dense ligamentum flavum is absent in the midline, it is possible to enter the epidural space without ever sensing significant resistance to injection. The ligamentum flavum is thickest at the lumbar and thoracic levels and thinnest at the cervical level. Its thickness also diminishes at the cephalad aspect of each interlaminar space and as the ligamentum flavum tapers off laterally. In patients who have undergone spinal surgery, scarring of the posterior epidural space is common, such that the loss of resistance and the flow of injected solutions are less predictable.

The spinal cord is a cylindrical structure composed of an external white matter and internal gray matter that is protected by the bony vertebral column. White matter represents the myelinated ascending and descending tracts of the spinal cord, which conduct information to and from the brain. The gray matter contains axons, dendrites, and synaptic terminals arranged grossly in the shape of butterfly wings when viewed in cross section. The spinal cord receives its vascular supply from arteries of the brain and from segmental spinal arteries arising from the subclavian artery, aorta, and iliac arteries.19 The posterior spinal cord receives its blood supply from a paired system of posterior spinal arteries arising from the posterior inferior cerebellar arteries. The anterior spinal artery is a single discontinuous vessel formed by the union of a terminal branch from each vertebral artery that descends along the anterior midline of the spinal cord. In all regions of the cord, the anterior spinal artery provides nutrition to approximately 75% of the cord tissue, including all of the gray matter,21 which makes this territory most vulnerable to ischemia.

The segmental spinal arteries reach the spinal cord by way of the intervertebral foramina and the epidural space, to reach the spinal nerve roots. Although the main purpose of these segmental arteries is to provide blood supply to the nerve roots (and thus they have been termed radicular arteries), some send branches that penetrate the dura to join the anterior and posterior spinal arteries, providing blood supply to the spinal cord. The largest of the segmental arteries is the artery radicularis magna (artery of Adamkiewicz), which supplies the anterior spinal artery for the thoracolumbar region of the cord. In 78% of cases, it enters by way of a left intervertebral foramen between T8 and L3; in 15% of cases, the artery of Adamkiewicz takes off more cephalad, at about the level of T5.21 Many of our pain procedures are performed in this region, and damage to or embolization caused by particulate steroid injected into the artery of Adamkiewicz or other spinal segmental arteries can result in ischemia and infarction of the anterior two-thirds of the lumbar spinal cord, with its resultant predominantly motor lesion.22

Venous drainage of the spinal cord occurs via large epidural veins, which are more prominent in the anterior and lateral portions of the epidural space. These veins are valveless and connect to the systemic circulation via the basivertebral venous plexus (Batson plexus), intracranial veins, and azygos veins. Inferiorly, this plexus communicates with the sacral venous plexus, which drains into the uterine and iliac veins. At each level, the vertebral plexus sends branches through the intervertebral foramina that anastomose with the thoracic and abdominal veins. The extensive communication between the epidural venous plexus and the systemic circulation is responsible for the distension of the epidural veins that occurs with increases in intra-abdominal pressure. Thus significant obstruction of the inferior vena cava results in rerouting of the venous return through the epidural venous plexus to the azygos vein.

The spinal nerve at each level traverses the intervertebral foramen and divides into anterior and posterior primary rami. The anterior ramus contains most of the sensory and motor fibers at each vertebral level. Of importance, a small branch of the anterior ramus, the sinuvertebral nerve, provides neural branches to the posterior outer layers of the annulus of the disk.20 The posterior primary ramus, in turn, divides into a lateral branch, which provides innervation to the paraspinous musculature, and a small variable sensory distribution to the skin overlying the spinous processes; the medial branch passes over the base of the transverse process where it joins with the superior articular process of the facet joint and courses along the articular process to supply sensation to the joint. Each facet joint receives sensory innervation from the medial branch nerve at the same vertebral level as well as from a descending branch from the vertebral level above; thus 2 medial branch nerves must be blocked to anesthetize each facet joint (eg, medial branch blocks at the base of the L4 and L5 transverse processes are needed to anesthetize the L4-L5 facet joint). The specific course of the medial branch nerves and cannula position for radiofrequency treatment at specific spinal levels is discussed in the following sections.

Anatomy of the Sympathetic Nervous System

Because we cover sympathetic nerve block techniques, a brief review of the anatomy is warranted. The sympathetic nervous system is part of the autonomic nervous system that innervates cardiac, smooth muscle, and glandular tissues, mediating a variety of reflexes. In certain pathologic pain states, neuronal activity in the sympathetic nervous system may be involved in the maintenance of chronic pain.23 Anatomically, the peripheral sympathetic system arises as efferent preganglionic fibers that leave their cell bodies in the intermediolateral column of the spinal cord, which extends from the first thoracic spinal segment to approximately the second lumbar segment. These axons leave the spinal cord within the ventral root and initially proceed as part of the spinal nerve. They separate from the somatic motor neuron, forming the white rami communicantes, and project to the sympathetic chain, located along the anterolateral surface of the vertebral bodies. Within the sympathetic nerve trunk, the preganglionic fibers can transverse variable distances cephalad and caudad, forming synapses with many postganglionic neurons in different ganglia at other levels in the chain. The ratio of preganglionic sympathetic fibers to postganglionic fibers is estimated to be 1:10, permitting coordinated activity at several spinal levels.23 The axons of the postganglionic neurons are predominantly unmyelinated and exit the ganglia as the gray rami communicantes, which join the spinal nerves en route to their peripheral targets. In addition to the multiple ganglia located in the thoracolumbar chain, there are classically 3 cervical ganglia, 4 to 5 lumbar ganglia, 4 sacral ganglia, and 1 coccygeal ganglion (the ganglion impar).

In the cervical region, the sympathetic chain lies at the anterolateral aspect of the vertebral bodies. Most sympathetic fibers traversing to and from the head, neck, and upper extremities pass through the stellate ganglion. In most individuals, the stellate ganglion is formed by fusion of the inferior cervical and first thoracic sympathetic ganglia; in some individuals the 2 ganglia remain separate. The ganglion is commonly found just lateral to the longus colli muscle, anterior to the neck of the first rib and the transverse process of the seventh cervical vertebra. In this position, the ganglion lies posterior to the first portion of the subclavian artery at the origin of the vertebral artery posterior to the dome of the lung. Although several approaches to stellate ganglion block have been described, the most common is the anterior paratracheal approach at C6 using surface landmarks. Performing the block at C6 reduces the likelihood of pneumothorax, which is more likely when the block is carried out close to the dome of the lung at C7. The anterior tubercle of the transverse process of C6 (Chassaignac tubercle) is palpable in most individuals.

Sympathetic innervation to the abdominal viscera arises from the anterolateral horn of the spinal cord between the T5 and T12 levels. Nociceptive information from the abdominal viscera is carried by afferents that accompany the sympathetic nerves. Pain transmitted via the celiac plexus originates from the upper abdominal structures, including the pancreas, diaphragm, liver, spleen, stomach, small bowel, ascending and proximal transverse colon, adrenal glands, kidney, abdominal aorta, and mesentery. The celiac plexus is composed of a diffuse network of nerve fibers and individual ganglia that lie over the anterolateral surface of the aorta at the T12-L1 vertebral level. Presynaptic sympathetic fibers travel from the thoracic sympathetic chain toward the ganglion, traversing over the anterolateral aspect of the inferior thoracic vertebrae as the greater (T5-T9), lesser (T10-T11), and least (T12) splanchnic nerves. Presynaptic fibers traveling via the splanchnic nerves synapse within the celiac ganglia, over the anterolateral surface of the aorta surrounding the origin of the celiac and superior mesenteric arteries at approximately the L1 vertebral level. Postsynaptic fibers from the celiac ganglia innervate all of the abdominal viscera with the exception of the descending colon, sigmoid colon, rectum, and pelvic viscera.

The lumbar sympathetic chain consists of 4 to 5 paired ganglia that lie over the anterolateral surface of the second through fourth lumbar vertebrae. The cell bodies that send projections to the lumbar sympathetic ganglia lie in the anterolateral region of the spinal cord from T11 to L2, with variable contributions from T10 and L3. As in the thoracic area, the lumbar preganglionic fibers leave the spinal canal with the corresponding spinal nerve, join the sympathetic chain as white communicating rami, and then synapse within the appropriate ganglion. Postganglionic fibers exit the chain to join the diffuse perivascular plexus around the iliac and femoral arteries or via the gray communicating rami to join the nerves that form the lumbar and lumbosacral plexus. Sympathetic fibers accompany all of the major nerves to the lower extremities. Most of the sympathetic innervation to the lower extremities passes through the second and third lumbar sympathetic ganglia, and blockade of these ganglia results in near-complete sympathetic denervation of the lower extremities.

The superior hypogastric plexus is composed of a flattened band of intercommunicating nerve fibers that descend over the aortic bifurcation. The plexus carries sympathetic afferents and postganglionic efferent fibers from the lumbar sympathetic chain as well as parasympathetic fibers that arise from S2-S4. The plexus is retroperitoneal in location and lies over the anterior surface of the fourth and fifth lumbar and the first sacral vertebrae. Sympathetic nerves passing through the plexus innervate the pelvic viscera, including the bladder, uterus, rectum, vagina, and prostate.

Interlaminar Epidural Injection Steroids and Transforaminal Injection of Steroids

Overview

The theoretical background supporting the use of epidural steroids is based on the existence of inflammation as the basic pathophysiologic process. Nerve root edema has been demonstrated surgically and with computed tomography (CT) in patients with herniated disks.24 Inflammation of nerves in the presence of a herniated disk has further been confirmed during surgery,25 myelography,26 and histologic examinations.27,28 More recently, phospholipase A2 (PLA2), the rate-limiting enzyme in the conversion of arachidonic acid to prostaglandins and leukotrienes, has been found in high levels in the extruded disk material of patients undergoing discectomy for herniated disk.29 Other inflammatory mediators such as prostaglandins have also been shown to produce hyperalgesia.30 Clinically, improvement of pain has been shown to coincide with resolution or decrease in nerve root edema, despite a persistent herniated disk.20

In laboratory animals, PLA2 has been implicated as a primary inflammatory mediator. Its administration adjacent to lumbar spinal nerves produces motor weakness and decreased mechanical withdrawal thresholds.31 Histologic examination shows reversible demyelination, vacuolar degeneration in the nerve roots, and unclear axonal margins.27 In another animal model of radiculopathy, Hayashi et al32 ligated the left L4 and L5 nerve roots while surgically placing an epidural catheter with the tip between the ligated nerve roots. All rats demonstrated reversible motor weakness that resolved completely in 4 weeks, and all of the animals exhibited thermal hyperalgesia. The rats were separated into groups and treated with epidural betamethasone alone, normal saline, or epidural betamethasone plus bupivacaine. All treatment groups demonstrated a transient reduction in thermal hyperalgesia, but both betamethasone groups showed prolonged benefit lasting 3 to 4 weeks.

Steroids decrease inflammation by inducing the biosynthesis of a PLA2 inhibitor, preventing prostaglandin generation.33,34 Steroids also suppress ongoing discharge in chronic neuromas and prevent the development of ectopic neural discharges from experimental neuromas, which has been attributed to direct action on the membrane.35 Steroids may block nociceptive input. Local application of methylprednisolone was found to reversibly block transmission in the unmyelinated C-fibers but not in Aβ-fibers.36

Patient Selection

Many investigators have attempted to identify which patients are most likely to benefit from epidural steroid injections. In 1960, Goebert et al37 administered 3 consecutive epidural hydrocortisone and procaine injections in 239 patients with low back pain. Fifty-eight percent of the patients had more than 60% relief for 3 months or longer, and 8% of patients claimed 40% to 60% pain relief.

In a prospective study, White et al38 followed the response of 304 patients with low back pain to epidural steroid injection. They found longer pain relief in patients with less than weeks of pain and in those patients without "psychologic overlay." Eighty-seven percent of the patients reported good short-term success, without significant differences between the acute and chronic pain groups, whereas 34% of patients in the acute pain group still reported pain relief at 6 months compared with 12% in the chronic pain group. Twenty-four percent of patients without psychologic overlay were still relieved of their pain at 6 months compared with only 1.5% of patients with psychologic overlay.

One prospective randomized study of lumbar epidural steroid injections followed 36 patients younger than 50 years with radicular pain, positive straight-leg raising test, and a prolapsed disk at L4-L5 or L5-S1 confirmed by magnetic resonance imaging (MRI).39 These patients were randomized to 2 groups; both received identical conservative therapy, and one of the groups also received a series of 3 epidural steroid injections. At 2-week follow-up, patients in the epidural steroid injection group had a significantly greater increase in straight-leg raise measurements compared with the control group. In addition, a trend in pain relief favoring the epidural steroid group was noted but did not reach significance. This study was not blinded, and all injections were given within 14 days of randomization, so the interval from onset to injection was very brief.

The effectiveness of epidural steroids in the relief of radicular pain attributable to herniated disk has been documented by many studies.40-47 In 2 fairly recent systematic reviews, national organizations have assembled practice guidelines based on the best available scientific evidence; both groups concluded the epidural steroid injections are useful in speeding the resolution of acute radicular pain associated with disk herniations.48,49 Although their usefulness has not been clearly documented by controlled studies, epidural steroid injections also have been used to treat back pain secondary to degenerative disk disease,40,44,45 spinal stenosis,50 trauma,37 spondylolysis or spondylolisthesis,38,40,42 and in pain following laminectomy.37,51

There has been some recent data suggesting that the transforaminal and interlaminar routes of injection may vary in their effectiveness. In a 2009 study, Lee et al studied a cohort of patients with disk herniations and spinal stenosis.52 Patients were randomized to receive an interlaminar epidural steroid injection versus bilateral transforaminal epidural steroid injections. There was no difference in pain scores or patient satisfaction in the disk herniation group. However, in patients with spinal stenosis, there was significant improvement in pain scores and satisfaction in those that received bilateral transforaminal epidural when compared with the group receiving interlaminar injections.

In a large observational study on the frequency of epidural steroid injections and patient characteristics, Fanciullo et al53 collected information on 25479 patients seen at 23 specialty spine centers in the United States from 1995 to 1998. Epidural steroid injection formed part of the treatment plan for only 2002 patients (7.9%). Lumbar problems were far more likely in patients who had an epidural steroid injection (12.6%) than in patients with cervical (3.7%), thoracic (1.8%), or sacral problems (2.4%). The most frequent diagnoses in patients who were offered epidural steroid injection were spinal stenosis (40.1%) and herniated disk (34.3%), followed by spondylolisthesis (7.8%) and compression fractures (1.6%). Evidence supports the use of epidural steroids in patients with disk herniations and other presumed inflammatory conditions that cause acute radicular pain, such as spinal stenosis or spondylolisthesis, but the efficacy of epidural steroids is undocumented in patients with compression fractures. Thus patient selection for epidural steroid injections is not always consistent with existing data, and concordance between clinical practice and reported evidence of efficacy is variable. Indeed, epidural injection of steroids is one of the most overused treatments for back pain in the United States.

Drug Selection for Epidural Steroid Injection

Most studies reported in the literature used a mixture of local anesthetic and steroid, saline with steroid, or steroid alone. The steroids most commonly used are either methylprednisolone acetate (Depo-Medrol) or triamcinolone diacetate (Aristocort). The doses of methylprednisolone most widely used vary from 80 to 120 mg, and the doses of triamcinolone most commonly used vary from 50 to 75 mg.44,45,54,55

Methylprednisolone acetate has been approved for intramuscular, intrasynovial, soft tissue, and intralesional injection. It is a glucocorticoid with an elimination half-life of 139 hours and a range from 58 to 866 hours.56 Triamcinolone diacetate, which has an elimination half-life of 18 to 36 hours, possesses glucocorticoid properties while being essentially devoid of mineralocorticoid activity, thus causing little or no sodium retention. It has been approved for administration by intramuscular, intra-articular, and intrasynovial routes.57

Serious injury to the central nervous system has been reported following transforaminal epidural steroid injections.58-60 The likely etiology is embolization of steroid particles into the vertebral artery or a segmental spinal artery causing end-arteriolar occlusion, resulting in posterior circulation strokes or spinal cord infarction. The resulting injuries are catastrophic: quadriplegia, paraplegia, cortical blindness, or death. The most recent and convincing evidence that embolization of particulate steroid can result in stroke comes from a large animal study. Okubadejo et al61 administered particulate (methylprednisolone) and nonparticulate (dexamethasone) directly in to the vertebral arteries of anesthetized swine; those animals receiving particulate steroid never regained spontaneous respiratory function, and MRI demonstrated changes consistent with acute stroke, whereas those receiving dexamethasone had no changes on MRI and recovered fully with no apparent harm. In an effort to avoid microembolic complications, practitioners have started using dexamethasone for transforaminal injections, specifically in the cervical spine. Dexamethasone sodium phosphate is a synthetic glucocorticoid that has an elimination half-life of only 1.8 to 3.5 hours in patients with normal renal function but a biologic half-life of 36 to 54 hours. Because dexamethasone is water soluble and has been approved for intravenous administration, it is thought that the risk of vascular microembolization is diminished. Derby et al62 evaluated particle size of various steroid preparations of dexamethasone, triamcinolone, betamethasone, and methylprednisolone. They found that dexamethasone particles measured about 0.5 μm, smaller in diameter than a red blood cell, and no aggregation was observed when mixed with local anesthetic or contrast medium. There are no randomized studies comparing clinical efficacy with different formulations of corticosteroids; however, observational studies did not find any statistical differences between particulate and nonparticulate steroids in cervical transforaminal epidural injections.63,64 We have recently changed our practice to use dexamethasone in cervical transforaminal epidural steroid injection as detailed later.

In a review of the literature, Kepes and Duncalf65 found that methylprednisolone was the least irritating, the most beneficial, and the longest acting, although Delaney et al66 prefer triamcinolone because of its excellent anti-inflammatory effect and low potential for sodium retention. No study has compared the effectiveness of triamcinolone and methylprednisolone, which likely are equally effective. Both of these preparations contain polyethylene glycol, which has been found to impair nerve transmission in rabbit vagus nerve and cause degenerative lesions in rat sciatic nerves.67,68 Both preparations also contain benzyl alcohol, which is potentially toxic when administered locally to neural tissue.44 Neither of these preparations should be used intrathecally.

Most practitioners dilute the steroid with local anesthetic or sterile saline, and the results are apparently comparable with either diluent.43 Some authors have recommended use of local anesthetics "in the presence of muscle spasms."39,69 However small the dose, use of local anesthetic carries some risks, including hypotension, arrhythmias, and seizures from intravascular injections. Brown44 suggests that because the results are comparable, use of saline probably is sufficient. Some investigators have combined epidural methylprednisolone with morphine.70,71 In an initial study, Cohn et al70 showed encouraging results in postlaminectomy patients; however, a subsequent study was unable to repeat such beneficial effects. Epidural morphine carries significant risk of respiratory depression and should be used with caution in the outpatient setting.

Traditionally, the volume of diluent has depended on the site of injection: lumbar, caudal, or transforaminal. When using a caudal approach, 20 to 25 mL of a solution has been recommended to ensure epidural spread cephalad to the desired level.40,43 When using a lumbar interlaminar approach, a volume of 5 to 10 mL has been recommended to reach the areas most commonly involved in the lumbar region.38 Other practitioners use smaller volumes (2-3 mL), especially when using the transforaminal approach. Some authors have suggested that several nerve roots may be inflamed in addition to those adjacent to the herniated or bulging disk, and they recommend against using a small volume of diluent.72 Wood et al68 suggested diluting the steroid after their study showed degenerative lesions in rat sciatic nerves attributable to the polyethylene glycol vehicle in the steroid preparation. The optimal volume of injectate and site of epidural placement remain unresolved.

Epidural steroids have been associated with glucose intolerance and pituitary-adrenal axis suppression for up to 3 weeks after repeated administration.73-75 Ward et al measured insulin sensitivity and fasting blood glucose, fasting plasma insulin, and fasting serum cortisol levels in 10 healthy individuals 24 hours before and 1 week after epidural administration of triamcinolone 80 mg via a caudal route.76 They found that 24 hours after epidural steroid injection, insulin sensitivity had decreased to nearly half the baseline, fasting insulin levels increased 1.4-fold, and fasting glucose levels had increased 1.1-fold. All of these values normalized by 1 week after injection, and they demonstrate the marked changes in insulin sensitivity occurring in nondiabetic healthy individuals after epidural steroid injection.

Technique

The epidural space can be approached through the interlaminar space (median or paramedian), intervertebral foramen (transforaminal), or sacral hiatus (caudal). The approach selected depends on patient selection, indication for injection, practitioner's experience, and availability of imaging. The patient can be positioned in the lateral decubitus position, sitting or prone, depending on the technique to be used.

Interlaminar Technique for Epidural Steroid Injection

The term interlaminar has been given to the traditional posterior approach to the epidural space to easily differentiate it from the transforaminal or caudal approach. The interlaminar approach can be either midline or paramedian. Correct epidural placement with this technique is facilitated by the use of image guidance. As reported by Sharrock et al,77 anesthesiologists have a high success rate of epidural anesthesia using loss of resistance; however, use of image guidance can confirm proper placement in all cases. When using the midline approach, the point of insertion depends on the level of entry and angle of the corresponding spinous processes. For the cervical and lumbar midline approach, the needle should be almost perpendicular to the neuraxis, in line with the corresponding spinous process. This is also true with the low thoracic approach below T9. In the midthoracic region, the spinous processes are sharply angled caudad, such that the tip of the spinous processes lies opposite the lamina of the inferior vertebral body. In this region, the midline approach requires that the needle be inserted with a steep cephalad angle, often approaching 130°. Many practitioners prefer the paramedian approach in the midthoracic region. The choice of needles when using the interlaminar approach is similar to those available for anesthesia; the most common type is the 18- or 20-gauge Tuohy needle.

For cervical interlaminar epidural injections, most practitioners place the patient in a sitting or prone position. When in the sitting position, the patient is asked to sit comfortably and flex the neck anteriorly. The forehead can be leaned against a sturdy, padded horizontal surface to minimize involuntary movement. This position avoids rotation of the spine and widens the lower cervical epidural space. Because of its large interlaminar distance, the most common approach is C6-C7 or C7-T1, especially because the long spinous process of C7 (the vertebra prominens) serves as a reliable surface marker.

When the prone position is used for the cervical interlaminar approach, most practitioners prefer to use image guidance. This allows for good visualization of the interlaminar space and needle advancement between adjacent spinous processes.

Cervical Epidural Steroid Injection Technique

Classically, the epidural space has been identified using loss of resistance to saline. Regardless of the approach, sterile technique must be strictly observed. The skin and subcutaneous tissues overlying the interspace where the block is to be performed are anesthetized with local anesthetic. The cervical interspaces with the largest interlaminar distance typically are found at C6-C7 and C7-T1. Because of the ease of entry, many practitioners place the needle via one of these larger interspaces, regardless of the level of pathology, and they rely on the flow of steroid in the epidural space to reach the level of pathology. The same technique can be used in all cervical interspaces below C3-C4. Interlaminar injections above the C2-C3 level have not been described.

An 18- or 20-gauge Tuohy needle is placed through the skin and advanced several centimeters until the needle is firmly seated in the interspinous ligament. An anteroposterior (AP) image is taken, and the needle is redirected toward midline. We use the loss-of-resistance technique with saline to find the epidural space. Repeat images taken after every few millimeters of needle advancement will ensure that the needle direction does not stray from midline. A firm grasp of the adjacent structures and the proximity of the spinal cord is essential during a cervical interlaminar epidural injection. After the needle tip enters the epidural space, the position is confirmed by injecting nonionic radiographic contrast, and adequate spread is verified in the AP and lateral planes. If lateral imaging of the cervicothoracic junction and low cervical spine is hindered by the adjacent structures of the torso and arms, a second lateral image taken just above the shoulders often can be much simpler to interpret when trying to confirm epidural contrast flow. Once epidural needle position has been confirmed, a solution containing steroid diluted in preservative-free saline is injected. In our practice, we routinely use 80 mg of methylprednisolone acetate or the equivalent diluted in 3 to 5 mL of total volume.

Midthoracic Epidural Steroid Injection

Because of the steep angulation of the spinous processes at this level, image guidance is often used. The patient lies prone, with the head turned to one side. The C-arm is rotated 40° to 50° caudally from the axial plane without any oblique angulation. This allows for good visualization of the interlaminar space and needle advancement between adjacent spinous processes.

The skin and subcutaneous tissues approximately 1 cm lateral and 2 to 3 cm caudad to the interspace where the block is to be performed are anesthetized. An 18- or 20-gauge Tuohy needle is placed through the skin and advanced several centimeters until the needle is firmly seated in tissue. An AP image is taken, and the needle is directed toward the superior margin of the lamina below the interspace that is to be entered, near the junction of the spinous process and the lamina. Although a midline approach can be used at low thoracic levels, the spinous processes are angled too steeply to allow for true coaxial needle placement at the midthoracic levels. Thus the needle is best directed toward the superior margin of the lamina. While advancing, repeat images are taken and care must be taken to keep the needle tip over the lamina until bone is gently contacted. The periosteum is then anesthetized, and the needle is slowly advanced over the superior margin of the lamina until loss of resistance occurs. Because the needle is unlikely to lie within the interspinous ligament when using a paramedian approach, there will be little resistance to injection until the needle enters the interlaminar space and traverses the ligamentum flavum. A firm understanding of the adjacent structures and the proximity of the spinal cord is essential for thoracic interlaminar epidural injection. After the needle tip enters the epidural space, the position is confirmed by injecting nonionic radiographic contrast, and spread is verified in the AP and lateral planes. Once epidural needle position has been confirmed, a solution containing steroid diluted in preservative-free saline is injected, and the needle is removed.

Lumbar Epidural Steroid Injection

Epidural injection at the lumbar level can be administered with the patient in a sitting or prone position. The patient is asked to sit comfortably with his or her back to the practitioner, curving the spine posteriorly and pushing the lumbar region against the examiner's fingers in an attempt to separate the spinous processes. If the procedure is to be performed in the prone position under fluoroscopic imaging, a pillow is placed under the mid and lower abdomen to reduce the lumbar lordosis and increase the separation between adjacent spinous processes. The C-arm is rotated 15° to 20° caudally from the axial plane without any oblique angulation. This allows for good visualization of the interlaminar space and needle advancement between adjacent spinous processes.

The skin and subcutaneous tissues overlying the interspace where the block is to be performed are anesthetized. An 18- or 20-gauge Tuohy needle is placed through the skin and advanced 1 to 2 cm until the needle is firmly seated in the interspinous ligament. If the patient is prone, a lateral image should be taken as the needle is advanced, until loss of resistance occurs. If the patient is sitting, adequate positioning will allow a proper midline approach. A firm knowledge of the adjacent structures and the proximity of the thecal sac and cauda equina are essential for lumbar interlaminar epidural injection (Fig. 92-4). After the needle tip enters the epidural space and aspiration is negative for CSF or blood, the position is confirmed by injecting nonionic radiographic contrast, and spread is verified in the AP and lateral planes. Lateral imaging of the lower lumbar spine is hindered by the overlying iliac crests, and visualization can be difficult in the obese patient. Once epidural needle position has been confirmed, a solution containing steroid diluted in preservative-free saline is injected, and the needle is removed. In the lumbar region, we usually use 80 mg of methylprednisolone acetate or the equivalent diluted in 5 to 10 mL of total volume. When larger injectate volumes are used, the solution spreads extensively in both the anterior and posterior aspects of the epidural space. In patients with significant lumbar pathology, the injectate tends to follow the path of least resistance, often flowing toward the side opposite the pathology.2

Caudal Epidural Steroid Injection

Anesthesiologists will be familiar with caudal injections administered to pediatric patients in the operating room. Because of difficulty identifying the sacral hiatus clinically in adults, these procedures usually are performed under fluoroscopy. The patient lies prone with the head turned to one side. The C-arm is rotated 20° to 30° caudally from the axial plane without any oblique angulation. This allows for good visualization of the sacrum, sacral hiatus, and coccyx.

Once the sacral hiatus is identified radiographically, the overlying skin and subcutaneous tissues are anesthetized. The sacral hiatus can be difficult to visualize radiographically. The approximate location can be identified by palpating the paired sacral cornua in the midline, near the superior extent of the gluteal cleft. An 18- or 20-gauge Tuohy needle can be used, but a smaller 22-gauge, 3.5-in spinal needle is adequate. The needle is placed through the skin and advanced directly through the sacrococcygeal ligament. Once the needle has passed through the sacrococcygeal ligament and is within the caudal spinal canal, the angle of the needle is decreased to lie closer to the plane of the sacrum, and the needle is advanced into the spinal canal an additional 1 to 2 cm. AP and lateral imaging confirm the needle's position within the caudal epidural space. The caudal epidural space is generously supplied with veins, and intravascular needle placement is ruled out by injecting nonionic radiographic contrast under live fluoroscopy. Once caudal epidural needle position has been confirmed, a solution containing steroid diluted in preservative-free saline is injected, and the needle is removed. The caudal epidural space is distant from the usual sites of nerve root inflammation near the lumbosacral junction; thus a significant volume of injectate usually is required to effect spread to the level of the lumbosacral junction. For this approach, we use 80 mg of methylprednisolone acetate or the equivalent diluted in at least 10 mL of total volume.

Complications of the interlaminar approach include dural puncture with subsequent postdural puncture headache, which can occur when this technique is used at any level of the spine. Although in a different patient population, the incidence of dural puncture in parturients undergoing lumbar epidurals ranges between 0% and 2.6%.78 Postdural puncture headaches occur frequently after unintended dural puncture with large epidural needles. The incidence of headache following cervical dural puncture is lower than that following lumbar puncture, likely because of the diminished column of CSF cephalad to the point of dural puncture.

Although cervical and thoracic epidural blood patches using small volumes of autologous blood have been described, most practitioners manage postdural puncture headaches following cervical or thoracic epidural injection conservatively with fluids and oral analgesics. Dural puncture also can occur during caudal epidural injection but usually only if the needle is advanced several centimeters cephalad within the caudal spinal canal. The thecal sac extends to the level of approximately S2, and the position can be approximated by palpating the adjacent posterior superior iliac spines, which lie at the same level. Epidural blood patch using autologous blood is a safe and effective treatment that relieves headache symptoms promptly in 70% to 98% of patients who fail to improve after 24 to 48 hours of conservative treatment and oral analgesics.79 The incidence of unintentional dural puncture may be higher in patients with previous lumbar surgery because of scarring within the epidural space and adhesion of the dura to the posterior elements.

Regardless of the level of injection, epidural bleeding or infection can occur. Epidural hematoma or abscess can lead to significant compression of the spinal cord or cauda equina. Interlaminar epidural injection should be avoided or postponed in patients receiving anticoagulants.85

Transforaminal Technique for Epidural Steroid Injection

Transforaminal epidural steroid injection and selective nerve root injection can be performed using similar techniques. The distinction between the 2 techniques is questionable because the fascial sheath surrounding the spinal nerves is contiguous with the dura mater within the epidural space. A solution injected around a spinal nerve may well enter the epidural space, whether or not the needle tip is advanced through the intervertebral foramen before injection. Nonetheless, many practitioners reserve the term selective nerve root injection for injections performed with the needle tip adjacent to the spinal nerve, outside of the intervertebral foramen and the term transforaminal injection for injections performed with the needle tip within the intervertebral foramen. Unlike the interlaminar technique, the transforaminal approach requires the use of radiographic imaging to proceed with safety.86

Several important anatomic aspects are unique to transforaminal epidural and selective nerve root injections. At cervical levels, the ventral and dorsal roots converge in the vertebral canal to form the spinal nerve in the intervertebral foramen. The foramen faces obliquely anterior and laterally. Its roof and floor are formed by the pedicles of the articulated vertebrae. Its posterolateral wall is formed largely by the superior articular process of the lower vertebra and in part by the inferior articular process of the upper vertebra and the capsule of the zygapophysial joint. The anteromedial wall is formed by the lower end of the upper vertebral body, the uncinate process of the lower vertebra, and the posterolateral corner of the intervertebral disk. Immediately lateral to the external opening of the foramen, the vertebral artery ascends within the foramen transversarium in close proximity to the spinal nerve and anterior to the articular pillars of the zygapophysial joint. The spinal nerve, in its dural sleeve, lies in the lower half of the foramen, whereas the upper half is occupied by epiradicular veins. Arterial branches arise from the vertebral arteries to supply the nerve roots (radicular arteries) or the spinal cord via the anterior and posterior spinal arteries (spinal segmental or medullary arteries). Medullary and radicular arterial branches also may arise from the deep or ascending cervical arteries and traverse through the entire length of the foramen adjacent to the spinal nerve. It is these spinal segmental arteries that are at risk for penetration during cervical transforaminal injection.

At the lumbar levels, the ventral and dorsal roots of the spinal nerves descend within the vertebral canal to form the spinal nerve in the intervertebral foramen. Its roof and floor are formed by the pedicles of consecutive vertebrae. Its posterior wall is formed largely by the superior articular process of the lower vertebra and in part by the inferior articular process of the superior vertebra and the capsule of the zygapophysial joint. The anterior wall is formed by the vertebral body and the intervertebral disk. The spinal nerve, in its dural sleeve, lies in the anterior and superior portions of the foramen, just inferior to the pedicle.

The most common indication for a transforaminal approach or selective nerve root injection is for placing the corticosteroid immediately adjacent to the inflamed nerve root causing the radicular symptoms. Nerve root inflammation may stem from an acutely herniated intervertebral disk causing nerve root irritation or other causes of nerve root impingement, such as isolated foraminal stenosis due to spondylitic spurring of the bony margins of the foramen. Although some studies87,88 have shown better ventral dye spread with transforaminal and parasagittal approaches versus interlaminar injection, there is scant evidence that better clinical outcomes result from one or the other approach.86 Selective nerve root injection with local anesthetic has been used diagnostically to determine which nerve root is causing symptoms when pathology exists at multiple vertebral levels. This information can prove invaluable in planning surgical intervention.

Cervical Transforaminal Steroid Injection

The patient lies supine, facing directly forward. The C-arm is rotated 45° to 55° lateral oblique until the neural foramina are clearly visualized. The patient may also be asked to rotate the head away from the side of injection. Although this position facilitates access to the side of the neck, the neural foramina and bony elements of the cervical spine will no longer be aligned, which may prove confusing to the inexperienced practitioner.

A 25-gauge, 1.5- to 2.5-in blunt-tipped needle is sufficient in length for all but the most obese patients. To avoid the vertebral artery and the spinal nerve, the needle is advanced toward the posterior and middle aspect of the intervertebral foramen. Care is taken ensure that the needle tip remains superimposed on the bone of the facet column during advancement. In this way, the superior articular process of the facet just posterior to the foramen is contacted first, preventing needle advancement through the foramen and into the spinal canal. Once the needle contacts the facet, it is walked anteriorly into the foramen and advanced no more than another 2 to 3 mm. The depth is assessed by obtaining an image in the direct AP plane. To avoid direct trauma to the spinal cord and intrathecal injection, the needle should be advanced no further than halfway across the facet column. Many practitioners place a small length of flexible tubing between the needle hub and the syringe, so that once the needle is in final position, small changes in needle position do not occur as the syringe is taken on and off of the needle itself. Nonionic radiographic contrast is injected under "live" or real-time fluoroscopy (or digital subtraction cineradiography) to ensure that the needle tip lies in close proximity to the nerve root without any intravascular or intrathecal spread. The solution containing the steroid can then be injected safely. We have stopped using particulate steroid in cervical transforaminals and currently use four 4 to 12 mg of dexamethasone sodium phosphate (in 1-2 mL).

Lumbar Transforaminal Steroid Injection

The patient lies prone on the fluoroscopy table. The C-arm is rotated 20° to 30° lateral oblique to allow direction of the needle toward the superolateral aspect of the intervertebral foramen. A somewhat less oblique approach will result in a final needle position slightly lateral to the intervertebral foramen, which some practitioners advocate as a means of limiting spread of the injectate to a single nerve root. However, even small volumes of injectate are often seen to track along the exiting nerve root to enter the lateral epidural space.

A 22- or 25-gauge, 3.5-in spinal needle is sufficient in length for patients of average build, whereas a 5-in needle may be needed in obese patients. To avoid the spinal nerve, the needle is advanced coaxially toward the superior aspect of the intervertebral foramen, just inferior to the pedicle and inferolateral to the pars interarticularis. This serves as an effective depth marker. Once this bony margin is contacted, the C-arm is rotated to a lateral view and the needle is slowly advanced toward the anterior and superior aspect of the foramen. If the patient reports paresthesia at any time during needle advancement, the needle should be withdrawn slightly and the position confirmed with radiographic contrast. With the needle in final position, nonionic radiographic contrast is injected under real-time fluoroscopy (or digital subtraction cineradiography) in the AP position to ensure that the needle tip lies in close proximity to the nerve root without any intravascular or intrathecal spread. Obtaining a final lateral image will allow assessment of the extent of spread of the injectate.

Complications of the transforaminal technique can be catastrophic. A firm grasp of the anatomy of adjacent vascular and neural structures is essential to avoid complications during cervical and lumbar approaches (Fig. 92-5). Direct intravascular injection into the vertebral artery may produce generalized seizures when local anesthetic is used or cerebral ischemia when particulate steroid solutions are used.86,89 Direct injection of particulate steroid into a spinal segmental artery supplying the spinal cord at the cervical or lumbar level can lead to catastrophic infarction of the spinal cord. Needle positioning toward the posterior aspect of the foramen and advancing the needle in a plane parallel to the nerve root reduces the risk of entering a vascular structure. Again, particular care should be taken when performing transforaminal injection on the left between T8 and L3 because the artery of Adamkiewicz lies between these levels. However, use of radiographic contrast injected during "live" or "real-time" fluoroscopy (or digital subtraction cineradiography) to visualize final needle position and detect any hint of intravascular injection is the only means to verify accurately that injectate is not located within an artery.

Subarachnoid injection may occur if the needle is advanced too far medially and pierces the dural cuff as it extends laterally onto the exiting nerve root. Direct trauma to the spinal nerve or the spinal cord itself also may occur. Intradiscal placement of the needle during attempted transforaminal epidural steroid injection has been reported but usually is without sequelae.90

Sympathetic Blocks

Overview

The sympathetic nervous system is involved in the pathophysiology of a number of different chronic pain conditions, including complex regional pain syndrome (CRPS) and ischemic pain. These chronic pain states often are referred to as sympathetically maintained pain because they share the characteristic of pain relief following blockade of the regional sympathetic ganglia.91,92 In an extensive review, Cepeda et al93 questioned the efficacy of sympathetic blockade in the diagnosis and treatment of CRPS. In a review of 29 studies that included 1144 patients, only 29% of patients reported "full positive response," and 41% had a "partial positive response." According to these authors, most of the studies were retrospective and of poor quality. Sympathetic blockade is still widely used despite the scant evidence supporting its diagnostic and therapeutic role.92,93 Many practitioners continue to use sympathetic blockade as a part of a multidisciplinary approach to treating CRPS; sympathetic blocks are one tool that can reduce pain and facilitate functional recovery. Blockade of the stellate ganglion has been used in the diagnosis and treatment of sympathetically maintained pain of the head, neck, and upper extremity, and a second lumbar sympathetic block is used for diagnosis and treatment of sympathetically maintained pain of the lower extremities. Celiac plexus block has been used for malignant and nonmalignant pain involving the upper abdominal viscera, and several techniques have been described.95 Celiac plexus neurolysis has been used successfully to treat pain from pancreatic cancer.96,97 Superior hypogastric block for treatment of chronic pain arising from the bladder, uterus, rectum, vagina, and prostate has been well described.98

Stellate Ganglion Block

Stellate ganglion block has long been the standard approach to diagnosis and treatment of sympathetically maintained pain syndromes involving the upper extremity, such as CRPS. Other neuropathic pain syndromes, including ischemic neuropathies, herpes zoster (shingles), early postherpetic neuralgia, and postradiation neuritis, also may respond to stellate ganglion block. Blockade of the stellate ganglion has proven successful in reducing pain and improving blood flow in vascular insufficiency conditions such as intractable angina pectoris,100 Raynaud disease,101 frostbite,102 vasospasm, and occlusive and embolic vascular disease.103 The sympathetic fibers control sweating; thus stellate ganglion block can be quite effective in controlling hyperhidrosis.104

Patients with signs and symptoms of CRPS of the upper extremities may gain significant pain relief from stellate ganglion block.99 Unfortunately, the duration and magnitude of the pain relief are unpredictable.93 This led to the use of repeated sympathetic blocks, sometimes as often as daily or weekly over an extended period of time in attempt to improve the duration of pain relief. No controlled studies have ever verified this practice, and experts currently agree that repeated sympathetic blocks alone rarely eliminate the pain and disability associated with CRPS. Nevertheless, incorporating sympathetic blockade into a coordinated multidisciplinary rehabilitation plan is essential for effective treatment of patients with CRPS. This treatment plan typically includes physical therapy, oral neuropathic pain medications, and supportive psychotherapy. Neuroablation has been used to destroy the sympathetic chain in patients who attain excellent pain relief of temporary duration with local anesthetic blocks. Few data evaluating the success of sympathetic ablation are available, and expert opinion regarding the usefulness of this approach in the long-term treatment of CRPS is varied.

To perform a stellate ganglion block, the patient is placed supine, facing directly forward with a pillow under the upper back and lower neck to hold the neck in a slight extension. To perform the block without radiographic guidance, the operator palpates the cricoid cartilage and then slides a finger laterally into the groove between the trachea and the sternocleidomastoid muscle, retracting the muscle and adjacent carotid and jugular vessels laterally. Chassaignac tubercle typically is palpable in this groove at the C6 level. Once the tubercle has been identified, a needle is advanced through the skin and seated on the tubercle, where local anesthetic is injected. The local anesthetic spreads along the prevertebral fascia, ideally in a caudal direction to anesthetize the stellate ganglion, which lies just inferior to the point of injection in the same plane. In practice, marked variations in the size and shape of the Chassaignac tubercle reduce the rate of successful block. The adjacent vertebral artery and C6 spinal nerve must be avoided to optimize conduct of this block. A simple modification of technique in which the needle is directed medially toward the base of the transverse process using radiographic guidance is a simple means of improving the reliability of stellate ganglion block and is described here.

When using fluoroscopic imaging, the C-arm is centered over the lower cervical spine without angulation. The position of the vertebral bodies and transverse processes of C6 and C7 are identified. The skin and subcutaneous tissues overlying the base of the transverse process of C6 or C7 on the affected side are anesthetized. The transverse processes are often difficult to distinguish from the underlying facet columns, but the transverse process joins the vertebral body just inferior to the uncinate process of the vertebral body, a structure that is easy to identify on the posteroanterior (PA) radiograph. The block can be carried out at either the C6 or C7 level when using radiographic guidance. However, it is important to realize that the vertebral artery passes near the base of the transverse process at C7, and many individuals lack a bony foramen transversarium at this level. Thus, at C7, care must be taken to keep the needle tip in line or medial to a line connecting the uncinate process of C7 and T1. Straying more lateral will risk penetration of the vertebral artery. The overlying carotid artery must be retracted laterally to perform the classic technique for stellate ganglion block over the C6 transverse process, but this is unnecessary when the needle is directed toward the base of the transverse process because the needle passes medial to the great vessels of the neck.

A 25-gauge, 2.5-in needle is placed through the skin and advanced coaxially until it is seated in the tissues. The needle is adjusted to remain coaxial as it is directed toward the base of the transverse process, just inferior to the uncinate process using repeat PA images after every 2 to 4 mm of needle advancement. Once the surface of the vertebral body is contacted, the needle is in its final position. Intravascular placement is ruled out and proper position is assured by injecting radiographic contrast. The contrast should spread along the anterolateral margin of the vertebral bodies in both PA and lateral radiographs. Thereafter, 10 mL of local anesthetic (0.25% bupivacaine) is injected incrementally. Repeat radiographs following local anesthetic injection should show dilution of the contrast and spread of the solution inferiorly to the T1 level where the stellate ganglion lies. Sympathetic block should ensue within 20 minutes following injection and is assured by seeing a 1°C or greater rise in temperature of the ipsilateral hand. Table 92-1 lists the signs of successful stellate ganglion block.

Many important structures lie within the immediate vicinity of the needle's tip once it is properly positioned for stellate ganglion block. Commonly, diffusion of local anesthetic blocks the adjacent recurrent laryngeal nerve. This often leads to hoarseness, a feeling of having a lump in the throat, and subjective feelings of shortness of breath and difficulty swallowing. Bilateral stellate ganglion block should not be performed because bilateral recurrent laryngeal nerve blocks may lead to loss of laryngeal reflexes and respiratory compromise. The phrenic nerve is also commonly blocked by direct spread of local anesthetic and leads to unilateral diaphragmatic paresis. Diffusion of local anesthetic as well as direct placement of local anesthetic adjacent to the posterior tubercle results in somatic block of the upper extremity. This may take the form of a small area of sensory loss due to diffusion of local anesthetic or a complete brachial plexus block when the local anesthetic is placed within the nerve sheath.

Major complications associated with stellate ganglion block include neuraxial block (spinal or epidural) and seizures. Extreme medial angulation of the needle from a relatively lateral skin entry point may lead to needle placement into the spinal canal through the anterolaterally oriented intervertebral foramen. In this manner, local anesthetic can be deposited in the epidural space, or, if the needle is advanced far enough, it may penetrate the dural cuff surrounding the exiting nerve root and lie within the intrathecal space. Medial angulation will direct the needle toward the trachea and esophagus and risk penetration of these structures. More likely is placement of the needle tip on the posterior tubercle and spread of local anesthetic proximally along the nerve root to enter the epidural space. In this case, partial or profound neuraxial block, including high spinal or epidural block with loss of consciousness and apnea, may ensue. Airway protection, ventilation, and intravenous sedation should be promptly administered and continued until the patient regains airway reflexes and consciousness. Because the maximal effect of epidural local anesthetic may require 15 to 20 minutes to develop when longer-acting local anesthetics are used, it is imperative that patients be monitored for at least 30 minutes after stellate ganglion block.

Intravascular injection during stellate ganglion block likely will result in immediate onset of generalized seizures. The carotid artery lies just anteromedial to the Chassaignac tubercle, whereas the vertebral artery lies within the bony transverse foramen just posteromedial to the tubercle. If injection occurs into either structure, the local anesthetic injected enters the arterial supply traveling directly to the brain, and generalized seizures begin rapidly and after small amounts of local anesthetic. However, because the local anesthetic rapidly redistributes, the seizures typically are brief and do not require treatment. In the event of seizure, halt the injection, remove the needle, and begin supportive care.

Celiac Plexus Block

Neurolytic celiac plexus block is among the most widely applicable of all neurolytic blocks. Neurolytic celiac plexus block has a long-lasting benefit for 70% to 90% of patients with pancreatic and other intra-abdominal malignancies. A meta-analysis by Eisenberg et al105 concluded that adequate-to-excellent pain relief can be achieved in 89% of patients in the first few weeks following the block. From 70% to 90% of patients still had complete pain relief during the 3-month interval before their death.105 Although encouraging, interpretation of these data requires caution because this meta-analysis is based on retrospective studies.

Several techniques for localizing the celiac plexus have been described. In the classic technique, a percutaneous posterior approach uses surface and bony landmarks to position needles in the vicinity of the plexus. Numerous reports have described new approaches for celiac plexus block using guidance from plain radiographs, fluoroscopy, CT, or ultrasound.106-108 No single methodology has proven clearly superior with regard to safety or success rate. In recent years, it has been generally agreed that radiographic guidance is necessary to perform celiac plexus block. Many practitioners have turned to routine use of CT, taking advantage of the ability to visualize adjacent structures when performing this technique.109

The transcrural approach to the celiac plexus (Fig. 92-6) places the local anesthetic or neurolytic solution directly on the celiac ganglion, anterolateral to the aorta. The needles pass directly through the crura of the diaphragm en route to the celiac plexus. In contrast, splanchnic nerve block (see Fig. 92-6) avoids the risk of penetrating the aorta and uses smaller volumes of solution, and the success is unlikely to be affected by anatomic distortion caused by extensive tumor or adenopathy near the pancreas. Because the needles remain posterior to the diaphragmatic crura in close apposition to the T12 vertebral body, this has been termed the retrocrural technique. Splanchnic nerve block is a minor modification of the classic retrocrural celiac plexus block; the only difference is that, for splanchnic block, the needles are placed more cephalad. In most cases, celiac plexus (transcrural or retrocrural) and splanchnic nerve block can be used interchangeably, with the same results. Although some strongly advocate one approach or the other, there is no evidence that either approach results in superior clinical outcomes.95

Celiac plexus and splanchnic nerve block are used to control pain arising from intra-abdominal structures. These structures include the pancreas, liver, gallbladder, omentum, mesentery, and alimentary tract from the stomach to the transverse colon. The most common application of neurolytic celiac plexus block is treatment of pain associated with intra-abdominal malignancy, particularly pain associated with pancreatic cancer.96,110 Neurolysis of the splanchnic nerves or celiac plexus can produce dramatic pain relief, reduce or eliminate the need for supplemental analgesics, and improve quality of life in patients with pancreatic cancer and other intra-abdominal malignancies.110 The long-term benefit of neurolytic celiac plexus block in patients with chronic nonmalignant pain, particularly those with chronic pancreatitis, is debatable.

Transcrural Technique

Once the C-arm is aligned over the thoracolumbar junction, the skin and subcutaneous tissues overlying the superior margin of the L1 vertebral body are anesthetized. The aorta lies to the left of midline over the vertebral bodies. By routinely placing the left-sided needle first, a single needle can often be used for the block. If the aorta is penetrated en route, a transaortic technique is used. A 22-gauge, 5-in spinal needle is advanced just caudal to the margin of the 12th rib and cephalad to the transverse process of L1 to contact the anterolateral surface of the L1 vertebral body. The C-arm is then rotated to a lateral projection and the needle advanced to lie 2 to 3 cm anterior to the anterior margin of L1 in the lateral view. Continuous aspiration should be applied as the needle is advanced past the anterior border of L1. If blood appears, the needle has penetrated the aorta and should be advanced through the anterior wall of the aorta until blood can no longer be aspirated. The needle tip should be medial to the lateral border of the L1 vertebral body in the AP view. Final needle position is confirmed by injecting radiographic contrast under live fluoroscopy. The contrast should layer over the anterior surface of the aorta and appear pulsatile. If the contrast spreads to both sides of midline over the anterior surface of the aorta, then only a single needle is necessary for the block. If the contrast remains to the left of midline over the anterolateral surface of the aorta, a second needle is placed from the contralateral side using the same technique described for the left-sided block. Diagnostic celiac plexus block before neurolysis is performed using 10 to 15 mL of 0.25% bupivacaine per side. The dose should be given in increments of 5 mL, aspirating periodically to ensure that the needle has not moved to an intravascular location.

Splanchnic Nerve Block or Retrocrural Technique

Once the C-arm is aligned, the skin and subcutaneous tissues overlying the anterolateral margin of the midportion of the T12 vertebral body are anesthetized. For splanchnic nerve block and neurolysis, needles must be placed on both sides. A 22-gauge, 5-in spinal needle is advanced just caudal to the margin of the 12th rib and cephalad to the transverse process of L1 to contact the anterolateral surface of the T12 vertebral body. This requires 20° to 30° of cephalad angulation of the C-arm. The C-arm is then rotated to a lateral projection and the needle advanced 1 to 2 cm to align with the anterior third of the T12 vertebral body in the lateral view. The needle tip should be just medial to the lateral border of the T12 vertebral body in the AP view. Final needle position is confirmed by injecting radiographic contrast under live fluoroscopy. The contrast should layer over the anterolateral surface of the T12 vertebral body. A second needle is placed from the contralateral side using the same technique described for the left-sided block. Diagnostic splanchnic nerve block prior to neurolysis is performed using 5 to 8 mL of 0.25% bupivacaine per side. The dose should be given in increments of 5 mL, aspirating periodically to ensure that the needle has not moved to an intravascular location.

Celiac Plexus and Splanchnic Neurolysis

The technique for needle placement is identical for diagnostic local anesthetic block of the celiac plexus or splanchnic nerves and for neurolysis. The 2 commonly used neurolytic solutions are ethyl alcohol and phenol. Phenol is a combination of carbolic acid, phenic acid, phenylic acid, phenyl hydroxide, hydroxybenzene, and oxybenzene. There is no commercially available phenol preparation, but a solution can be prepared by a compounding pharmacist from anhydrous phenol crystals.

Phenol is highly soluble in glycerin and radiographic contrast solutions. Phenol has local anesthetic properties at lower concentrations and is neurolytic at higher concentrations. Concentrations lower than 5% cause protein denaturation, whereas higher concentrations produce protein coagulation and segmental demyelination. Poorly myelinated and unmyelinated nociceptive fibers are destroyed at concentrations of 5% to 6%. Higher concentrations can produce axonal damage, spinal cord infarction, arachnoiditis, and meningitis. Large systemic doses of phenol cause effects similar to those seen with local anesthetic overdose, such as seizures and cardiovascular collapse. A 10% to 12% solution of phenol can be prepared in radiographic contrast. This allows radiographic monitoring of the spread of neurolytic solution as it is injected.

For celiac plexus neurolysis, 10 to 15 mL per side is injected. If the neurolytic solution spreads to both sides of midline over the anterior surface of the aorta, then only a single needle is necessary for the block. If the neurolytic solution begins to spread posteriorly toward the intervertebral foramen, the injection should be halted to avoid nerve root injury. During splanchnic neurolysis, the contrast should layer over the anterolateral surface of the T12 vertebral body. A second needle is placed from the contralateral side using the same technique described for the left-sided block. For splanchnic neurolysis, 5 to 8 mL per side is injected. The needles should be flushed with saline or local anesthetic before they are removed to avoid depositing the neurolytic solution along the needle track.

Neurolysis also can be performed using 50% to 100% ethyl alcohol in similar volumes. Alcohol in excess of 33% results in extraction of cholesterol, phospholipids, and cerebrosides from neural tissues. It produces nonselective destruction of neural tissue. Unlike phenol, the degree of neural blockade increases over the first several days following neurolysis; however, it is intensely inflammatory and has been associated with persistent or worsened pain and neuritis. Phenol has a direct local anesthetic effect and is associated with minimal pain on injection. Because of the intense burning pain on injection, ethyl alcohol is best diluted with local anesthetic prior to injection or injected after placing a small volume of local anesthetic.

Although most cases can be carried out using fluoroscopic guidance alone, CT allows excellent visualization of the anatomic structures that lie in close proximity to the target site during neurolytic celiac plexus block.111 To directly ablate the celiac plexus, the needles must be advanced through the diaphragm until they lie adjacent to the anterolateral surface of the aorta. This can be accomplished by advancing 2 separate needles adjacent to the anterolateral surface of the aorta or using a single needle advanced through the aorta.

CT-guided celiac plexus block is carried out with the patient positioned prone in the CT scanner gantry. A radiographic marker is placed on the skin surface 1 cm inferior to the inferior margin of the 12th rib and 7 cm from midline, and axial CT images extending from T12 through L1 are taken in 3-mm intervals. In this way, the position of the needle entry site on the skin's surface can be adjusted to form a direct path to the anterolateral surface of the aorta, without passing through adjacent structures. The skin is anesthetized, and a 22-gauge, 5-in spinal needle is seated in a plane that corresponds to the axis seen on CT. With the needle seated in the subcutaneous tissue, but still superficial, a repeat CT image is obtained through the tip of the needle, and the angle of the needle is redirected toward the anterolateral surface of the aorta. The needle is advanced, and repeat CT images are obtained after every 1 to 2 cm of needle advancement. Once the needle is in position, a small volume of radiographic contrast is injected to confirm needle position. A solution of neurolytic drug in radiographic contrast allows radiographic monitoring of the spread of neurolytic solution as it is injected. A repeat CT image is obtained after every 5 mL of injection. If the neurolytic solution spreads to both sides of midline over the anterior surface of the aorta, then only a single needle is necessary for the block.

Following celiac plexus block, several physiologic side effects are expected, including diarrhea and orthostatic hypotension. Blockade of the sympathetic innervation to the abdominal viscera results in unopposed parasympathetic innervation of the alimentary tract and may produce abdominal cramping and diarrhea. Likewise, the vasodilatation that ensues often results in orthostatic hypotension. These effects are invariably transient but may persist for several days after neurolytic block. The hypotension seldom requires treatment other than intravenous hydration.

Complications of celiac plexus and splanchnic nerve block include hematuria, intravascular injection, and pneumothorax. CT allows visualization of the structures that lie adjacent to the celiac ganglion as the block is being performed and may help to avoid these complications.111 The kidneys extend between T12 and L3, with the left kidney slightly more cephalad than the right. The aorta lies over the left anterolateral border of the vertebral column. The celiac arterial trunk arises from the anterior surface of the aorta at the T12 level and divides into the hepatic, left gastric, and splenic arteries. Using the transaortic technique, caution must be used to avoid needle placement directly through the axis of the celiac trunk as it exits anteriorly. The inferior vena cava lies just to the right of the aorta over the anterolateral surface of the vertebral column.

Neurolytic celiac plexus block carries small but significant additional risks. Intravascular injection of 30 mL of 100% ethanol will result in a blood ethanol level well above the legal limit for intoxication but below the danger limit of severe alcohol toxicity. Intravascular injection of phenol is associated with clinical manifestations similar to those of local anesthetic toxicity: CNS excitation, followed by seizures and, in extreme toxicity, cardiovascular collapse. The most devastating complication associated with neurolytic celiac plexus block is paraplegia. The actual incidence of this complication is unknown but appears to be less than 1:1000. The theoretical mechanism is spread of the neurolytic solution toward the posterior surface of the aorta to surround the spinal segmental arteries, specifically the artery of Adamkiewicz.

Lumbar Sympathetic Block

Lumbar sympathetic blockade has been used extensively in the diagnosis and treatment of sympathetically maintained pain syndromes involving the lower extremities.99 Patients with peripheral vascular insufficiency due to small vessel occlusion also can be treated effectively with lumbar sympathetic blockade. If local anesthetic block improves blood flow and reduces pain, these patients often benefit from surgical or chemical sympathectomy.112

Other patients with neuropathic pain involving the lower extremities have shown variable response to lumbar sympathetic block. In those with acute herpes zoster and early postherpetic neuralgia, sympathetic block may reduce pain.113 However, after 3 to 6 months, once postherpetic neuralgia is well established, sympathetic blockade is rarely helpful. Likewise, deafferentation syndromes such as phantom limb pain and neuropathic lower extremity pain following spinal cord injury have shown variable and largely disappointing responses to sympathetic blockade.114

Lumbar sympathetic block typically is carried out using a single-needle technique and a large volume of local anesthetic to spread cephalad and caudad to bathe adjacent ganglia. The patient lies prone with a pillow under the lower abdomen and iliac crest to reduce lumbar lordosis. The C-arm is centered over the midlumbar region. The final needle position for lumbar sympathetic block is over the anterolateral surface of the lumbar vertebral body. The C-arm is rotated obliquely 20° to 30° until the tip of the transverse process of L3 overlies the anterolateral margin of the L3 vertebral body.

The ganglia of the lumbar sympathetic chain vary in number and location among individuals. The ganglia lie between L2 and L4, and in most humans the ganglia lie over the inferior portion of L2 and the superior portion of L3.115 Thus the optimal location for a single needle is over the anterolateral margin of the inferior portion of L2, the L2-L3 interspace, or the superior margin of L3.

After the skin and subcutaneous tissues are anesthetized, a 22-gauge, 5-in spinal needle is advanced using a coaxial technique toward the anterolateral surface of the L3 vertebral body to gently contact bone. The needle is then walked laterally off the bony margin. The C-arm is rotated to a lateral projection, and the needle is advanced until the tip lies over the anterior third of the vertebral body. Proper needle position is verified in the AP projection where the needle tip should lie medial to the lateral margin of the vertebral body.

Once the needle is in position, aspiration to detect intravascular needle placement is carried out, followed by administration of nonionic radiographic contrast. Contrast spread should be seen anterior to the L3 vertebral body. From 15 to 20 mL of 0.25% bupivacaine is injected incrementally. Signs of successful sympathetic blockade in the lower extremities include venodilation and temperature rise. Skin temperature should be monitored in the contralateral foot to assess for changes unrelated to the block. A rise in temperature of at least 1°C without a rise in temperature of the contralateral limb should occur with successful sympathetic block.

Lumbar Sympathetic Neurolysis

Neurolytic lumbar sympathetic block has been used in efforts to provide long-term sympathetic blockade in patients who receive only short-term pain relief with local anesthetic blocks. Lumbar sympathetic neurolysis can be accomplished using either injection of a neurolytic solution or radiofrequency lesioning. Because the locations of the lumbar sympathetic ganglia are variable, injection of neurolytic solution that spreads to encompass an area beyond the needle tip may produce more reliable neurolysis than radiofrequency treatment. Nonetheless, when the needle tips are positioned accurately, the discrete lesions resulting from radiofrequency treatment can produce effective neurolysis.116 Although the techniques are well described, few data are available to guide the choice among chemical neurolysis, radiofrequency neurolysis, and open surgical sympathectomy.

Chemical Neurolysis

Chemical neurolysis of the lumbar sympathetic chain is carried out by placing 3 separate needles at the L2, L3, and L4 levels, as described for local anesthetic block. The needles should be directed to the mid or inferior aspect of L2, the superior aspects of L3, and L4 to correlate with the most frequent anatomic locations of the lumbar sympathetic ganglia. Three needles are placed so that the smallest volume of neurolytic solution can be injected to treat the ganglia at each level. Once proper needle position has been confirmed in the AP and lateral projections, a small volume of radiographic contrast is placed through each needle to ensure that the needles are not located intravascularly and that the injectate will layer in close apposition to the anterolateral margin of the vertebral bodies. Thereafter, 2 to 3 mL of neurolytic solution is placed through each needle.

Radiofrequency Neurolysis

Significant and potentially toxic levels of local anesthetic can result from direct needle placement into a blood vessel and intravascular injection during lumbar sympathetic block. Hematuria can follow direct needle placement through the kidney but usually is self-limited. Nerve root, epidural, or intrathecal injection can arise when the needle is advanced through the intervertebral foramen and usually is avoided entirely with proper use of radiographic guidance. Following neurolytic lumbar sympathetic block, significant postsympathectomy pain arises in the L1 and L2 nerve root distribution over the anterior thigh in as many as 10% of treated patients. This observation stems from results following open surgical sympathectomy, but such postsympathectomy neuralgia also has been reported after both chemical and radiofrequency sympathectomy. Postsympathectomy neuralgic pain in the anterior thigh has been postulated to result from partial neurolysis of adjacent sensory fibers, most often the genitofemoral nerve.117

Facet Joint Injections: Intra-Articular Injections, Medial Branch Blocks, and Radiofrequency Treatment

Overview

Intra-articular facet injection has been largely supplanted by radiofrequency techniques for treatment of facet-related pain. Clinical experience and a limited number of published observational studies suggest that intra-articular injection of local anesthetic and steroid leads to relief of facet-related pain that is limited in duration.118,119 In contrast, radiofrequency treatment is safe and effective in producing longer-term pain relief in the same group of patients. Nonetheless, an understanding of facet-related pain syndromes and the methods for placing medication directly within the facet joint may still prove useful for those practitioners who are unable to provide radiofrequency treatment.

Osteoarthritis of the spine is ubiquitous and an inevitable part of aging. The degenerative cascade that leads to degeneration of the intervertebral disks causes progressive disk dehydration and loss of disk height. Typically starting in the third decade of life, disk degeneration leads to increased mobility of adjacent vertebrae and increased shear forces on the facet joints themselves. This can lead to a pattern of pain over the axis of the spine that increases with movement, particularly with flexion and extension, but produces little or no pain radiating toward the extremities. In the past, the only treatment available for those with debilitating facet-related pain was segmental fusion of the spine to completely arrest motion within the painful portion of the spine.120

Most patients have pain that is gradual in onset and can be localized only to a general region of the spinal axis.121 However, a subgroup of patients present with sudden onset of pain, often associated with trauma in the form of sudden flexion or hyperextension of the spine in the affected region. Diagnostic studies are invariably unrevealing, showing either no abnormalities or facet arthropathy at multiple levels. In those with pain of sudden onset, it may be possible to isolate one or more facets that are causing the pain. It is in these patients with sudden onset of well-localized pain that intra-articular facet injection with local anesthetic and steroid can prove most beneficial.

Patients with facet-related pain are difficult to distinguish from those with other causes of axial spinal pain. Some patients present with sudden onset of pain following a significant flexion-extension (whiplash) injury, but more commonly the onset is insidious over months to years. Patients with myofascial or discogenic pain as well as those suffering from sacroiliac dysfunction present with similar symptoms. Nonetheless, certain features can be helpful in differentiating facet-related pain from other causes of spinal pain. The pain caused by facet arthropathy is most pronounced over the axis of the spine itself and typically is maximal directly in the region of the most affected joints. The pain tends to be exacerbated by movement, particularly extension of the spine, which forces the inflamed articular surfaces of the facet joints together. However, axial spinal pain at rest or worsening with forward flexion or rotation of the spine is another common feature. The most important historical feature is a predominance of axial spinal pain; patients who report that the predominance of their pain is in the extremities are more likely to have acute or chronic radicular pain than facet-related pain. The quality of the pain typically is deep and aching, with the pain waxing and waning with activity. Burning or stabbing qualities suggest neuropathic pain rather than facet arthropathy.

Diagnostic studies often are unrevealing. Patients with significant facet-related pain may have unremarkable plain x-ray film and/or imaging studies of the spine, or they may show facet arthropathy at multiple levels. Although there is no reported correlation of facetogenic symptoms with plain x-rays or CT, the latter is currently considered to be the optimal imaging modality to visualize facet arthropathy. In recent years, some practitioners have begun using single-photon emission computed tomography (SPECT) to aid in the diagnosis of facet-related pain. Unlike planar views of regular bone scans, SPECT scanning integrates the use of a radioactive isotope and CT by rotating a gamma camera that creates a topographical image to better localize a point of abnormal tracer uptake.123,124 Radionuclide bone scintigraphy depicts bone areas with increased bone turnover, including synovial changes caused by inflammation and degenerative changes undergoing remodeling. With SPECT, the sensitivity to depict these changes is increased.125

Several studies have shown that SPECT results correlate with response to facet joint steroid injections. Dolan et al looked at 58 patients with clinical signs of facet-generated pain. Twenty-two patients had a positive SPECT scan; 36 had negative scans. All patients underwent facet joint injections with steroid and local anesthetic. The group of patients with positive SPECT showed significant improvement in pain relief and the McGill Pain Questionnaire.123 In a prospective randomized study, Pneumaticos et al identified a group of 47 patients with clinical criteria of facet-generated pain and treated the joints based on SPECT results. In patients with negative SPECT scans, the facets to be injected were determined based on clinical criteria. Patients who underwent facet joint injections based on the results of the SPECT scan had significantly lower pain scores, at 1 and 3 months, than patients whose level had been determined clinically.126 Although these results suggest that SPECT may prove a useful adjunct in the diagnosis of facet-generated pain, an analysis of cost, risks associated with radiation and use of radioactive isotopes, and potential benefits is needed before this test can be recommended.

The patterns of pain caused by abnormalities in specific facet joints have been established by injecting a mild irritant into a specific facet joints in healthy volunteers and then recording the pattern of pain produced.127-129 The levels treated are chosen by correlating the patient's report of pain to these pain diagrams. Occasionally a patient presents with evidence of facet arthropathy and a pattern of pain that corresponds to a single level, but this is uncommon. Most patients have more diffuse pain that can only be narrowed to a specific region. Treatment should be directed toward the joint(s) that most closely matches the pattern of referred pain that has been established for each joint and that typically requires treatment at more than one level.

Intra-Articular Facet Joint Injections versus Radiofrequency Treatment

Choosing between intra-articular facet injection and diagnostic medial branch blocks followed by radiofrequency treatment is a common clinical scenario. Limited outcome studies of intra-articular injection, particularly at the cervical level, have demonstrated only transient pain relief lasting from days to weeks in most patients.118,119,130 In a randomized trial, Marks et al131 showed equal pain relief in patients receiving facet joint injections and medial branch blocks. Patients who obtain significant pain relief from diagnostic blocks of the medial branch nerves may attain significant pain reduction from radiofrequency treatment that is longer lasting. Two randomized controlled trials have shown that radiofrequency ablation provides prolonged pain relief up to 6 months.132,133 Based on this improved efficacy and a long track record of safety, more practitioners are moving immediately to radiofrequency treatment rather than intra-articular injection. Indeed, a recent study by Cohen et al suggests that omitting diagnostic blocks altogether and proceeding directly to radiofrequency treatment can produce dramatic cost savings.134 Intra-articular injection remains valuable in patients with recent-onset pain that is discrete in location and suggests involvement of a single facet joint. Intra-articular injection also is a reasonable alternative when the expertise or equipment for radiofrequency treatment is not available, but it will provide only transient symptomatic relief in patients with facet-related pain who have not responded to conservative treatment.

Intra-Articular Facet Injection

Cervical Intra-Articular Facet Injection

The patient lies prone, facing directly toward the table with a small headrest under the forehead to allow for air flow between the table and the patient's nose and mouth. The C-arm is rotated 25° to 35° caudally from the axial plane without oblique angulation. This brings the axis of the x-ray path in line with the axis of the facet joints and allows for good visualization of the joints. Although the cervical facet joints also can be entered from a lateral approach with the patient lying on his or her side, advancing a needle using radiographic guidance in the AP plane allows the operator to directly see the position of the spinal canal at all times and avoid medial needle direction that could lead to spinal cord injury.

The skin and subcutaneous tissues overlying the facet joint where the block is to be carried out are anesthetized. The cervical level is easily identified by counting upward from the T1 level, where the T1 vertebra is easily distinguished by the presence of a large transverse process that articulates with the first rib. A 22-gauge, 3.5-in spinal needle is placed through the skin and advanced until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The needle is then advanced in increments of 0.5 to 1 cm using repeat images to redirect the needle toward the facet joint. Once the surface of the joint space is contacted, a lateral radiograph is obtained, and the needle is advanced slightly to penetrate the posterior joint capsule. The needle should not be advanced into the joint between articular surfaces; this serves no purpose and is likely to abrade the articular surfaces and lead to worsened pain once the local anesthetic block subsides. Although intra-articular location of the needle tip can be confirmed with radiographic contrast, this is unnecessary if the needle location is correct in both AP and lateral planes.

Thoracic Intra-Articular Facet Injection

Thoracic intra-articular facet injection is not commonly used. The plane of the thoracic facet joints is steeply angled, nearing the frontal plane. Even with steep angulation of the C-arm, the joint space cannot be visualized directly but must be inferred from the position of adjacent structures. The patient is positioned prone with the head turned to one side. The C-arm is angled 50° to 60° in a caudad direction from the axial plane. The plane of the mid and lower thoracic facet joints lies at an angle of 60° to 70° from the axial plane, but further angulation of the C-arm is impractical without the image intensifier resting against the patient's back. This angle allows visualization of structures adjacent to the facet joint from which the position of the joint can be inferred. The inferior articular process (superior aspect of the joint) lies posteriorly, directly over the superior articular process (inferior aspect of the joint). The needle tip is advanced toward the inferior aspect of the joint.

The thoracic level is easily identified by counting upward from the T12 level where the 12th and lowest rib joins the T12 vertebra. A 22-gauge, 3.5-in spinal needle is placed through the skin and advanced until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The needle is adjusted to remain coaxial and advanced toward the inferior margin of the joint space. Because of the joint's steep angle, the needle can be advanced only into the inferior and posterior most extent of the joint. Lateral radiography is difficult to interpret because of the overlying structures of the thorax.

Lumbar Intra-Articular Facet Injection

The anatomy of the lumbar facet joint and surrounding structures is illustrated in Fig. 92-7. The patient is positioned prone with the head turned to one side. The C-arm is angled obliquely 25° to 35° from the sagittal plane and without caudal angulation. This angle allows direct visualization of the facet joint. The skin and subcutaneous tissues overlying the facet joint where the block is to be carried out are anesthetized. The lumbar level is easily identified by counting upward from the sacrum. A 22-gauge, 3.5-in spinal needle is placed through the skin and advanced until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The needle is adjusted to remain coaxial and advanced toward the joint space.

Whether the intra-articular injection is done at the cervical, thoracic, or lumbar level, the facet joint itself holds only limited volume, typically less than 1.5 mL. Placing contrast in the joint limits the ability to place local anesthetic and steroid within the joint. Intra-articular injections are commonly carried out at the lumbar levels. At this level, the articular space is Z-shaped, with the superior recess extending slightly lateral to the axis of the articular surfaces and the inferior recess extending slightly medial to the axis of the articular surfaces. Once needle position has been confirmed, a solution containing steroid and local anesthetic is placed. A total dose of 80 mg of methylprednisolone acetate or the equivalent should be divided over all of the joints to be injected, but more than 40 mg per joint probably is unnecessary. Using concentrated steroid (40 or 80 mg/mL) allows a 1:1 mixture with local anesthetic (0.5% bupivacaine) to provide some immediate pain relief.

Complications associated with intra-articular facet injection are uncommon. The most likely adverse effect is an exacerbation of pain. This is frequent when intra-articular cervical facet injection is carried out and the needle is advanced within the joint space. The joint space is narrow, and advancing the needle within the joint can abrade the articular surfaces, causing increased pain. This exacerbation usually is self-limited. Infection can occur, leading to abscess within the paraspinous musculature, but the incidence is exceedingly low.135 Bleeding complications have not been associated with intra-articular facet injection.

Facet Medial Branch Blocks and Radiofrequency Treatment

In patients who receive only temporary relief from therapeutic intra-articular facet injections or have pain that is more diffuse, requiring treatment at numerous levels, radiofrequency treatment can produce significant, enduring pain relief. Many investigators have pointed to the need for controlled diagnostic injections to determine who will respond to radiofrequency treatment. Despite the value of placebo-controlled injections, they are impractical in most clinical settings. Most practitioners rely on a single set of diagnostic local anesthetic blocks to the medial branch nerves at the levels of suspected pathology to determine who should receive radiofrequency treatment. Patients who report significant pain relief, usually defined as 50% or more pain reduction lasting the average duration of the local anesthetic, go on to radiofrequency treatment. Similar transient pain relief with intra-articular injection of local anesthetic can be used as a reasonable prognostic test before proceeding with radiofrequency treatment.

Conventional radiofrequency treatment produces a small area of tissue coagulation surrounding the active tip of an insulated cannula. When the tip of the radiofrequency cannula is placed in close proximity to a neural structure, the lesion encompasses the nerve, causing denervation. The most commonly used cannulas for facet treatment are 22-gauge SMK (Sluijter-Mehta) cannulas, which come in lengths of 5, 10, and 15 cm. These radiofrequency cannulas have a noninsulated area where coagulation occurs, called the active tip, which may be 4, 5, or 10 mm in length. Conventional radiofrequency damages neural tissue by creating an electrical field between the active tip of the needle connected to a voltage generator and an inactive or dispersion electrode at a distance. This induces the movement of a tissue ionic current that follows the alternating current, generating friction and therefore creating heat surrounding the needle tip. Radiofrequency power produces heat by current flow and not through heat transfer from the tip.136 The lesion, although variable, is well circumscribed and reproducible when the physical parameters are properly controlled. The size mostly depends on needle diameter, length of the active tip, tissue vascularization, tip temperature, and time of exposure. Lesions are characterized by a central core filled with blood related to electrode placement surrounded by an area of coagulation necrosis and separated by a wall of neuroglial proliferation from a zone of liquefaction necrosis. The lesion is surrounded by an area of demyelination.137,138 For all but the most obese patients, the 10-cm cannulas with 5-mm active tips are used.

In recent years, pulsed radiofrequency treatment has come into frequent use. Studies that investigated the effects of nonheated tissues exposed to the electromagnetic field showed that the so-called isothermal (42°C-45°C) radiofrequency procedure induced physiologic changes in tissues.139 Van Zundert et al140 showed increase in expression of c-fos in the dorsal horn of experimental animals up to 7 days after pulsed radiofrequency treatment, which suggests sustained activation of a pain-inhibiting process. Although the concept of long-lasting pain reduction without neural destruction is appealing, as yet little clinical evidence supports the efficacy of this new technique.141

The key concept when using conventional versus pulsed radiofrequency is understanding where the lesion or pulse radiofrequency energy will occur relative to the active tip. The lesion produced by conventional radiofrequency is along the shaft of the needle surrounding the active tip. There is scant tissue destruction at the tip of the needle; thus the active tip of the cannula must be placed along the course of the nerve. In contrast, the highest density of voltage change during pulsed radiofrequency emanates directly from the tip of the radiofrequency cannula; thus the tip of the needle should be directed perpendicular to the course of the nerve to be treated. Techniques for both conventional and pulsed radiofrequency treatment are discussed here.

Cervical Facet Medial Branch Block and Radiofrequency Treatment

The medial branch nerves to the cervical facets course across the articular pillar, midway between the superior and inferior articular processes. The nerves can be anesthetized by placing a needle from a posterior or lateral approach. For the patient, the lateral approach is more comfortable because he or she can lie on one side rather than face down, and the needle must traverse less tissue en route to the target. However, when the needles are inserted from a lateral approach, they are directed toward the spinal cord; even slight rotation of the neck can lead to confusing the left and right articular pillars and result in needle entry into the spinal canal. For performing diagnostic medial branch blocks, either approach is adequate because the local anesthetic will be deposited in the same location in both approaches. For conventional radiofrequency treatment, the cannulas should be placed using a posterior approach, which will allow the entire length of the 5-mm active tip to be placed along the course of the nerve on the articular pillar. For pulsed radiofrequency treatment, the cannulas can be placed from a lateral approach because the voltage fluctuations are maximal at the tip of the cannula.

Posterior Approach

The patient lies prone, facing directly toward the table with a small headrest under the forehead to allow for air flow between the table and the patient's nose and mouth. The C-arm is rotated 25° to 35° caudally from the axial plane without any oblique angulation. This brings the axis of the x-ray path in line with the axis of the facet joints and allows for good visualization of the articular pillars.

Lateral Approach

The patient lies in the lateral decubitus position with a pillow under the head to keep the neck horizontal and minimizes lateral flexion of the neck to either side. The C-arm is placed directly over the patient's neck without rotation or angulation. Care must be taken to ensure that the left and right articular pillars are aligned directly over one another. This is a point of great confusion among practitioners who are inexperienced with radiographic anatomy of the cervical spine. Even small degrees of rotation can place the left and right facet joints in significantly different locations on lateral radiographs. It is difficult to discern the left side from the right, and if a needle is advanced toward the contralateral facet target in error, the needle can easily penetrate into the spinal canal.

Diagnostic Medial Branch Blocks

The cervical level can be identified by counting upward from T1 or downward from C2. Radiographically, T1 is identified in the AP view by its large transverse process that articulates with the head of the first rib, and C2 can be identified by its odontoid process in the AP view and its large spinous process in the lateral view. The skin and subcutaneous tissues overlying the facet target where the block is to be carried out are anesthetized. A 22-gauge, 3.5-in spinal needle is placed through the skin and advanced just until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The needle is adjusted to remain coaxial and advanced toward the facet target in the middle of the articular pillar, midway between superior and inferior articular surfaces of the vertebra. This appears as an invagination, or "waist," on AP radiographs and as a trapezoid on lateral radiographs. From the posterior approach, the needle is gently seated on the lateral margin of the facet column in the middle of the "waist"; from a lateral approach, the needle tip is seated in the middle of the trapezoid. Needle position is confirmed with AP and lateral radiographs. Once needle position has been confirmed, a small volume of local anesthetic is placed at each level and the needles are removed. The patient is instructed to assess the degree of pain relief in the hours immediately after the diagnostic blocks.

Radiofrequency Treatment

Radiofrequency cannulas are placed using a technique identical to that described for medial branch blocks. For conventional radiofrequency treatment, 5-cm SMK cannulas with 5-mm active tips are used and placed from a posterior approach. Once the lateral margin of the facet column is contacted, the needle is walked laterally off the facet and advanced 2 to 3 mm to position the active tip along the course of the medial branch nerve. Proper testing for sensorimotor dissociation is conducted. For sensory testing, the patient is asked to report pain or tingling during stimulation at 50 Hz at output less than 0.5 V. Motor testing is carried out at 2 Hz, slowly increasing the output to 3 times the sensory threshold. There should be no motor stimulation to the affected myotome throughout the testing period. We routinely increase the output to 3 V to rule out stimulation of the nerve root before we proceed with the radiofrequency procedure. Thereafter, great care must be taken to prevent any movement of the cannulas. Each level is anesthetized before ablation, and lesions are created at 80°C for 60 to 90 seconds.

For pulsed radiofrequency treatment, 5-cm cannulas with 5-mm active tips are inserted from a lateral approach. The tip is placed in the center of the trapezoid of the target facet, midway between articular surfaces and midway between the anterior and posterior extents of the facet column. Proper testing for sensory thresholds is conducted as for conventional radiofrequency treatment. Each level then is treated with pulsed radiofrequency adequate to maintain voltage fluctuations of 40 to 45 V for 120 seconds, without exceeding a tip temperature of 42°C. Local anesthesia is not needed for pulsed radiofrequency treatment but can be placed before the cannulas are removed.

Thoracic Facet Medial Branch Block and Radiofrequency Treatment

The medial branch nerves to the thoracic facets course over the base of the transverse processes where they join with the superior articular processes. The patient lies prone, with the head turned to one side. The C-arm is positioned over the thoracic spine, rotated 25° to 35° caudally from the axial plane without any oblique angulation. The transverse processes of the thoracic vertebrae are best seen from this angle at both high and low thoracic levels.

Diagnostic Medial Branch Blocks

The thoracic level can be identified by counting downward from T1 or upward from T12. The skin and subcutaneous tissues overlying the facet target where the block is to be carried out are anesthetized. A 22-gauge, 3.5-in spinal needle is placed through the skin and advanced just until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The needle is adjusted to remain coaxial and advanced toward the base of the transverse process where it joins the superior articular process and seated just on the bony margin. Once the needle is in position, a small volume of local anesthetic is placed at each level and the needles are removed. The patient is instructed to assess the degree of pain relief in the hours immediately after the diagnostic blocks.

Radiofrequency Treatment

Radiofrequency cannulas are placed using a technique identical to that described for medial branch blocks. For conventional radiofrequency treatment, 5- or 10-cm SMK cannulas with 5-mm active tips are used. Once the needle is seated against the superior margin of the transverse process where it joins the superior articular process of the facet, the cannula is walked superolaterally off the transverse process and advanced 2 to 3 mm to position the active tip along the course of the medial branch nerve. Proper testing for sensorimotor dissociation is conducted as previously described. Thereafter, great care must be taken to prevent any movement of the cannulas. Each level is anesthetized and lesions are created at 80°C for 60 to 90 seconds. Cannula placement for thoracic pulsed radiofrequency treatment is carried out in the same manner.

Lumbar Facet Medial Branch Blocks and Radiofrequency Treatment

The medial branch nerves to the lumbar facets course over the base of the transverse process where they join with the superior articular processes (Fig. 92-8). The medial branch nerve lies in the groove between the transverse process and the superior articular process, which slopes inferolaterally. The patient lies prone, with the head turned to one side. A pillow is placed under the lower abdomen in an effort to tilt the pelvis backward and swing the iliac crests posteriorly away from the lumbosacral junction. The C-arm is positioned over the lumbar spine with 25° to 35° of oblique angulation so that the facet joints themselves and the junction between the transverse process and the superior articular process are clearly seen. For medial branch blocks, the needle can be advanced in the axial plane without caudal angulation. However, for radiofrequency treatment, the C-arm should be angled 25° to 30° caudal to the axial plane so that the active tip of the radiofrequency cannulas will be parallel to the medial branch nerve within the groove between the transverse process and the superior articular process as it slopes inferomedially.

Diagnostic Medial Branch Blocks

The lumbar level can be identified by counting upward from the sacrum. The skin and subcutaneous tissues overlying the facet target where the block is to be carried out are anesthetized. A 22-gauge, 3.5-in spinal needle is placed through the skin and advanced until it is gently seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The needle is adjusted to remain coaxial and advanced toward the base of the transverse process where it joins the superior articular process and seated just on the bony margin. Once the needle is in position, a small volume of local anesthetic is placed at each level and the needles are removed. The patient is instructed to assess the degree of pain relief in the hours immediately after the diagnostic blocks.

Radiofrequency Treatment

Radiofrequency cannulas are placed using a technique identical to that described for medial branch blocks; however, the C-arm is angled 25° to 30° caudal to the axial plane so that the active tip of the radiofrequency cannulas will be parallel to the medial branch. For conventional radiofrequency treatment, 10-cm SMK cannulas with 5-mm active tips are used. Once the needle is seated against the superior margin of the transverse process where it joins the superior articular process of the facet, the cannula is walked off the superior margin of the transverse process and advanced 2 to 3 mm to position the active tip along the course of the medial branch nerve. Proper testing for sensory-motor dissociation is conducted as previously described, assuring there is no stimulation of the motor nerves to the lower extremities. Thereafter, great care must be taken to prevent any movement of the cannulas. Each level is anesthetized and lesions are created at 80°C for 60 to 90 seconds. Cannula placement for lumbar pulsed radiofrequency treatment is carried out in the same manner, except that the active tip need not be parallel to the medial branch nerve.

Complications of Medial Branch Block and Radiofrequency Treatment

Complications associated with diagnostic medial branch nerve blocks are uncommon and similar to those following intra-articular facet injections. Unlike intra-articular injection, it is unusual for medial branch blocks to cause an exacerbation of pain. Patients should be warned to expect mild pain at the injection site lasting 1 or 2 days after the procedure. Radiofrequency treatment of the facets also is associated with few complications. Despite the fact that conventional radiofrequency produces actual tissue destruction, injury to the spinal nerve roots is uncommon, perhaps because of sensory and motor testing. Injury to the spinal nerve root has been reported following radiofrequency treatment and could present with new-onset radicular pain with or without radiculopathy. The importance of physiologic testing before each lesion should be emphasized because this will reduce the chance that the active tip of the cannula is close enough to the anterior nerve root to cause injury.

Exacerbation of pain following conventional radiofrequency treatment is common, and patients should be instructed to expect an increase in pain, similar in character to their usual pain, that will last from several days to a week or more. A smaller group of patients report uncomfortable dysesthesia, usually in the form of a sunburn-like feeling of the skin overlying the spinous processes and often accompanied by allodynia. This adverse effect is more common following cervical radiofrequency treatment and usually subsides over several weeks. These dysesthesias likely stem from partial denervation of the lateral branch of the posterior primary ramus, which supplies a variable region of cutaneous innervation overlying the spinous processes. Likewise, some patients report a small patch of complete sensory loss in this same region.

Pulsed radiofrequency treatment does not produce tissue destruction, so it is not surprising that most patients have no worsening of their pain following treatment or have a transient, mild exacerbation that is short lived. Painful dysesthesia and other consequences of nerve injury do not occur with pulsed radiofrequency treatment. It is precisely because of this lack of neural destruction and associated adverse effects that pulsed radiofrequency treatment has become popular among practitioners. If controlled trials emerge to support the efficacy of pulsed radiofrequency treatment, it may rapidly replace conventional radiofrequency.141

Lumbar Discography and Intradiscal Treatments

Overview

Discography is a diagnostic test in which radiographic contrast is injected into the nucleus pulposus of the intervertebral disk. Although originally developed for the study of disk herniation, discography now is used most commonly to identify symptomatic disk degeneration. The 2 components of discography are (1) the anatomic appearance of contrast spread within the disk (using plain radiographs and/or CT) and (2) the presence or absence of typical pain during contrast injection within the disk (pain provocation). The usefulness of discography remains controversial. Some clinicians routinely use discography to identify symptomatic disks before surgical fusion or intradiscal thermal annuloplasty, whereas others believe the test is of unproven benefit in identifying symptomatic disks.142,143 Discography remains the only test available that attempts to correlate pain response from the patient during provocation with abnormal disks discovered on imaging studies. Improved surgical outcomes following lumbar fusion have been reported when guided by the use of discography.143-146 Intradiscal electrothermal therapy (IDET) is a minimally invasive procedure that offers an alternative treatment to a subset of those patients with discogenic low back pain. Much like its use prior to fusion, discography is used to identify symptomatic intervertebral disks prior to IDET.147

The patient with low back or neck pain originating from the vertebral disk often presents with deep, aching axial midline pain. Pain can be referred to the buttocks and posterior thigh from lumbar disks but does not extend to the distal extremities. Patients with discogenic pain often are young and otherwise healthy. Discogenic pain is common in those with jobs that require repetitive motion of the affected spine segment (eg, package handlers) or expose the spine to excessive vibration (eg, long-distance truck drivers, helicopter pilots, and jackhammer operators). Onset of symptoms usually is gradual. Pain is experienced with prolonged sitting (sitting intolerance), standing, and bending forward. The referred pain usually remains in the proximal part of the extremity. Results of physical examination usually are nonspecific, with limited range of motion at the affected segment or pain with movement, particularly on flexion.

Proponents of discography argue that MRI and CT reveal only nonspecific findings, such as loss of disk height and/or hydration. Jensen et al148 showed that more than 50% of asymptomatic patients have abnormal findings on MRI scans, in at least one intervertebral disk. The presence of a high-intensity zone on MRI (an area of increased T2-signal intensity at the posterior aspect of the disk) indicates that a radial tear of fissure may be present in the annulus fibrosis, again a nonspecific finding common in individuals without back pain.

Treatment for discogenic pain starts with conservative therapy, including physical therapy and oral nonsteroidal anti-inflammatory drugs. In those with prolonged or disabling pain that is suspected to be of discogenic origin, provocative discography can help to identify the affected level and guide targeted therapy.

Diagnostic Lumbar Discography

Lumbar discography is a painful procedure, even when performed by the most skilled practitioners. Intravenous sedation can facilitate the procedure; however, caution must be used to avoid oversedation, which could impede ongoing communication with the patient. The patient must be able to report paresthesias before neural injury occurs. Discography relies on the patient reporting the location and severity of symptoms during provocation, and excessive sedation can make interpretation of the results difficult.149

The patient lies prone, with the head turned to one side. A pillow is placed under the lower abdomen, above the iliac crest in an effort to reduce the lumbar lordosis. Having the patient rotate the inferior aspect of the pelvis anteriorly toward the table tips the iliac crest posteriorly and often is key to performing discography successfully at the L5-S1 level. The C-arm is rotated 25° to 35° obliquely centered on the disk space to be studied. The C-arm is then angled in a caudad-cephalad direction; the degree varies among patients, depending on the disk to be studied and each patient's degree of lumbar lordosis. In general, the L3-L4 disk lies close to the axial plane and requires no cephalad angulation to align the vertebral end plates; the L4-L5 disk requires 0° to 15° of cephalad angulation; and the L5-S1 disk requires 25° to 35° of cephalad angulation (Fig. 92-9). Proper alignment of the C-arm is critical to the safety and success of discography.

The skin and subcutaneous tissues overlying the disk space where discography is to be carried out are anesthetized, and additional local anesthetic is instilled liberally as the needle is advanced. A 22-gauge, 5-in spinal needle is placed through the skin and advanced just until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. A 7- or 8-in spinal needle is often required in obese patients and often needed at the L5-S1 level because of the long and oblique trajectory to the disk space. Without careful use of a coaxial technique throughout the entire course of needle advancement, discography will require multiple repositionings of the needle, if it can be done successfully at all. The direction of the needle should be rechecked after every 1 to 1.5 cm of needle advancement and adjusted to remain coaxial. The position of the exiting nerve root beneath the pedicle should be kept in mind at all times, and efforts to ensure that the needle does not stray cephalad or lateral to the intended point over the middle of the disk will reduce the likelihood of striking the nerve root en route to the disk.

Once the needle is in contact with the surface of the disk, there will be a notable increase in resistance to needle placement. At this point, the C-arm should be rotated to a lateral position and the needles advanced halfway from the anterior to the posterior margin of the disk. Proper final placement is then checked in the AP plane, where again the needle should be in the midportion of the disk space. The nucleus pulposus occupies the central third of the disk space, and placement of the needle tip anywhere within the nucleus should suffice. The final needle path lies just inferior to the exiting nerve root, and in many patients it is difficult or impossible to position the needle exactly in the center of the disk.

Once the needles are in final position at all levels to be tested, provocative testing is carried out. A small volume of radiographic contrast containing antibiotic is placed at each level (<1.5 mL of iohexol 180 mg/mL containing 1 mg/mL of cefazolin). The contrast material is injected under live fluoroscopy to observe the pattern of contrast spread within the disk. As the contrast is injected, the resistance to injection is noted and the patient is questioned about his or her symptoms. Some practitioners use an inline pressure monitoring device to ensure that excess pressure is not delivered during the provocative test. Some evidence indicates that pain reproduction using small volumes without excessive pressure during injection correlates most closely with symptomatic diskogenic pain; injection under high pressure or with large volumes may well produce pain even in normal disks.150

A concordant diskogram result occurs when the patient reports his or her typical pattern of severe pain during injection at the level of suspected pathology and the same patient reports no pain on injection of an adjacent disk that is normal in appearance.149,151 After injection of all levels, final AP and lateral radiographs should be obtained to document the levels tested and the patterns of contrast spread during injection. Some practitioners advocate for subsequent CT to assess the patterns of disk disruption using axial imaging, but the usefulness of CT discography in planning subsequent therapy is unclear.

Complications of Lumbar Discography

Most patients experience a marked exacerbation of their typical back pain in the days following discography. They should be warned to expect this pain and given a short course of oral analgesics for treatment of the exacerbation. Less commonly, injury to the exiting nerve roots occurs. The position of the nerve roots is in close proximity to the needle's path. Care must be taken to advance the needle slowly as it passes over the transverse process en route to the posterolateral margin of the disk. If the patient reports a paresthesia to the lower extremity, the needle should be withdrawn and redirected. Paresthesia occurs in a small proportion of patients, even with good technique. Persistent paresthesias are uncommon and typically ensue only after repeated paresthesias occur during the procedure. In a recent 10-year retrospective case-control study, Caragee et al found that discography resulted in accelerated disk degeneration compared with matched controls, and they call on practitioners to carefully weigh the risks and benefits before recommending disk injection.152

Infection can occur, leading to abscess within the presacral musculature, but the incidence is exceedingly low. Infection within the disk space (discitis) is the most feared complication of discography, with an incidence less than 1:1000. Treatment of discitis may require long-term administration of intravenous antibiotics and/or surgical removal of the infection. No cases of discitis occurring in patients who received intradiscal antibiotics during discography have been reported. Bleeding complications have not been associated with intradiscal injection.

Intradiscal Electrothermal Therapy

Spinal fusion has been reserved for patients with advanced disk degeneration, with clinical results varying between 46% and 82%.95,114 Patients who have early degenerative disk disease with preservation of near-normal disk height (>75% of normal disk height remaining) but severe ongoing back pain that does not improve with conservative therapy may be adequate candidates for intradiscal thermal coagulation (intradiscal electrothermal therapy [IDET]).153 The mechanism of action of IDET is unclear, but thermal energy has been shown to coagulate neural tissue154 and induce collagen denaturation,155 thereby addressing both nociceptive and mechanical aspects of discogenic pain.156 Early prospective studies demonstrated significant pain reduction and improvement in physical function in 30% to 50% of patients treated with IDET.153,156 However, 2 randomized controlled trials comparing IDET to placebo have reached conflicting conclusions.157,158 A recent systematic review by Helm et al that examined the 2 randomized controlled trials and 16 observational studies found that IDET produced significant relief in only half of the patients. Currently, strong evidence in support of IDET is lacking, and there has been a dramatic decline in the use of this technique.159

Patients with diskogenic pain present with concordant pain on discography at one or 2 spinal levels and no pain during provocation of an adjacent control disk. IDET makes use of a navigable thermal resistance wire that is placed percutaneously and positioned along the posterior aspect of the of the annulus fibrosis. Once in position, the disk is heated using a standardized protocol. Like discography, IDET is a painful procedure, even when performed by the most skilled practitioners. Intravenous sedation can facilitate the procedure, but a level of sedation that allows for ongoing communication with the patient is essential. The patient must be able to report paresthesias or excess discomfort during intradiscal treatment before neural injury occurs. Placement of cannulas for IDET is identical to that for needle placement during discography. The patient lies prone, with the head turned to one side. A pillow is placed under the lower abdomen above the iliac crest in an effort to reduce the lumbar lordosis. The C-arm is rotated 25° to 35° degrees obliquely centered on the disk space to be studies. The C-arm is then angled in a caudad-cephalad direction that varies among patients, depending on the disk to be studied and the degree of lumbar lordosis.

Like discography, the L3-L4 disk lies close to the axial plane and requires no cephalad angulation to align the vertebral end plates; the L4-L5 disk requires 0° to 15° of cephalad angulation; and the L5-S1 disk requires 25° to 35° of cephalad angulation. Proper alignment of the C-arm is critical to the safety and success of IDET. The technique for placing the cannulas through which the IDET catheter is introduced into the disk is similar to that for needle placement for discography. However, the best final position of the introducer is in the anterolateral aspect of the nucleus rather than the central portion of the nucleus. This allows for a more gradual angle as the IDET catheter exits the introducer and curves around the inner aspect of the annulus. The skin and subcutaneous tissues overlying the disk space where IDET is to be carried out are anesthetized, and additional local anesthetic is instilled liberally as the cannulas are advanced. A 17-gauge introducer supplied by the manufacturer is placed through the skin and advanced just until it is seated in the tissues in a plane that is coaxial with the axis of the x-ray path. The IDET introducer is stiff and easy to redirect and advance. The direction of the cannula should be rechecked after every 1 to 1.5 cm of needle advancement and adjusted to remain coaxial. The position of the exiting nerve root beneath the pedicle should be kept in mind at all times, and efforts to ensure that the needle does not stray cephalad or lateral to the intended point over the middle of the disk will reduce the likelihood of striking the nerve root en route to the disk.

Once the needle is in contact with the surface of the disk, there will be a notable increase in resistance to needle placement. At this point, the C-arm should be rotated to a lateral position and the needles advanced halfway from the anterior to the posterior margin of the disk. Proper final placement is then checked in the AP plane, where again the needle should be in the midportion of the disk space.

Once the IDET introducer is in a satisfactory position, the navigable thermal resistance wire (SPINECATH, Smith & Nephew, Andover, MA) is introduced. The tip of the wire slides along the medial circumference of the annulus and can be guided by gently rotating the proximal end of the catheter. The catheter is first advanced beyond the tip of the introducer and into the disk space using lateral radiography. When the tip of the catheter passes to the posterior aspect of the annulus and begins to traverse along the posterior annulus, the C-arm is rotated to the AP view, and the catheter is advanced to final position across the entire posterior annulus. The catheter has 2 radiopaque guides that indicate the active treatment portion of the catheter. These markers should be positioned to either side of the disk to indicate that the entire posterior annulus will be treated.

This brief description is simplistic; guiding the IDET catheter to final position can be quite challenging and requires delicate manipulation of the catheter to keep the tip from advancing into radial tears within the annulus. Overaggressive handling of the catheter causes it to kink, and, once kinked, the catheter is difficult or impossible to steer.

Once the catheter is in final position, heat is introduced using a specific protocol designed to gradually raise the temperature within the disk to 80° to 90°C and maintain that temperature for a minimum treatment period, typically 14 to 16 minutes. It is important that the patient is not overly sedated during the actual heat treatment so that he or she can report discomfort due to excess heat before neural injury occurs.

Complications of IDET

Patients should be warned of the typical postprocedural flare-up in pain symptoms that occurs after IDET. This results in an exacerbation of typical axial back pain, often lasting several days to weeks. Less commonly, injury to the exiting nerve roots occurs. The position of the nerve roots is in close proximity to the needle's path. Care must be taken to advance the needle slowly as it passes over the transverse process en route to the posterolateral margin of the disk. If the patient reports a paresthesia to the lower extremity, the needle should be withdrawn and redirected. Paresthesia occurs in a small proportion of patients, even with good technique. Persistent paresthesias are uncommon and typically ensue only after repeated paresthesias occur during the procedure.

Cauda equina syndrome, with severe neuropathic pain in the lower extremities as well as bowel and bladder dysfunction, has been reported to occur from IDET.160,161 Injury to the cauda equina is more likely to occur when there is an insufficient posterior annulus and the thermal catheter lies in close proximity to the thecal sac. The catheter can exit the disk space to enter the epidural space; however, this should be evident before treatment on lateral radiographs. Assuring that the patient is sufficiently awake to report excessive discomfort during the IDET treatment should reduce the chances of significant neural injury. Finally, overaggressive handling of the IDET catheter leads to kinking of the catheter near the point where it exits the tip of the introducer within the intervertebral disk. Repeated attempts to reposition the catheter once it is kinked can lead to shearing of the catheter tip. Catheter breakage and migration of the tip have been described.162

A key to successful outcome following IDET is strict adherence to a structure rehabilitation program that guides the patient through gradual increases in physical activity over a 6-week to 3-month time period. Rehabilitation following IDET is similar to the programs used following lumbar fusion.

Plasma Disk Decompression (Nucleoplasty)

Discectomy is one of the most common spine surgeries performed and the standard treatment for persistent radicular pain from disk herniation. With current trends toward less invasive techniques, various percutaneous disk decompression techniques have been introduced into clinical practice.

Plasma-mediated disk decompression, or nucleoplasty, uses a device that is inserted into the nucleus pulposus and creates channels within the nucleus, thereby reducing the intradiscal tissue volume. The device uses radiofrequency energy to excite electrolytes in a conductive medium. The energized particles break down molecular bonds dissolving soft tissue at temperatures ranging from 40°C to 70°C.163,164 In 2002, Sharps and Isaac165 prospectively followed 49 consecutive patients after undergoing nucleoplasty. They reported significant improvement in pain scores 1 year after the procedure. In a similar study, Mirzai et al166 studied 52 consecutive patients and also found decreased pain scores and decrease in analgesic consumption up to 1 year after nucleoplasty. In a recent multicenter prospective trial, Gerszten et al167 randomized 90 patients with radiculopathy and contained herniations to receive nucleoplasty versus transforaminal epidural steroid injections. At the 2-year follow-up, patients in the nucleoplasty group had significantly better pain scores and quality of life.

Patients who are suitable for nucleoplasty present with persistent radicular pain that has not responded to conservative therapies and a concordant, contained herniation of the nucleus pulposus. Some practitioners have suggested that MRI imaging should reveal a contained disk protrusion less than 6 mm and at least 50% of disk height maintenance.168 Opinions vary on the use of a discogram before nucleoplasty. Singh and Derby recommend performing discography to recreate the patient symptoms.169 Other authors have suggested performing discograms to confirm that the herniation is contained within the annular fibers and not as a provocative test.170,171

Nucleoplasty can be facilitated by intravenous sedation, but a level of sedation that allows for ongoing communication with the patient is essential. The patient must be able to report paresthesias or excess discomfort during intradiscal treatment before neural injury occurs. Placement of the 19-gauge introducer is identical to that for needle placement during discography. The patient lies prone, with the head turned to one side. A pillow is placed under the lower abdomen above the iliac crest in an effort to reduce the lumbar lordosis. As in discography, the C-arm is rotated 25° to 35° degrees obliquely centered on the disk space to be studied. The C-arm is then angled in a caudad-cephalad direction that varies among patients, depending on the disk to be studied and the degree of lumbar lordosis.

Once the introducer needle is placed within the nucleus pulposus under fluoroscopic guidance, the stylet is removed and the nucleoplasty device is introduced in the cannula and advanced slightly into the nucleus. The proximal limit is determined by using a circumferential reference mark on the shaft. The nucleoplasty device is then advanced until it comes into contact with the annulus on the opposite side. Here the depth stop marker on the shaft is advanced to the hub of the cannula to designate the distal limit for advancing the device. The device is then withdrawn to the proximal marker, and the reference line on the device is oriented in the 12 o'clock position. The ablation mode is activated and the device is advanced until the anterior annular border is reached. The coagulation mode is then activated and the device is withdrawn to the proximal marker. The device should be moved at approximately 2 mm/second. The reference line on the device is then rotated clockwise, and the procedure is repeated 6 times at the 2, 4, 6, 8, and 10 o'clock position, creating a total of 6 channels.172

Complications of Nucleoplasty

Nucleoplasty is a relatively new technique for disk decompression, and complications are thought to be rare. Aside from the well-known potential complications such as bleeding, infection, and nerve damage, 2 complications specific to this procedure have been reported in the literature. Smuck et al173 reported the case of a patient who developed a radiculopathy several months after nucleoplasty. A repeat MRI revealed epidural fibrosis, and the radicular pain spontaneously resolved. Li et al reported a case of a breakage of the device within the disk. Although the fragment could not be retrieved, the patient had a good clinical outcome.174

Implantable Drug Delivery Systems and Spinal Cord Simulation

Implantable Drug Delivery Systems

Intrathecal morphine and other opioids are now widely used as useful adjuncts in the treatment of acute and chronic pain. A number of agents show promise as analgesic agents with spinal selectivity. Continuous delivery of analgesic agents at the spinal level can be carried out using percutaneous epidural or intrathecal catheters, but vulnerability to infection and the cost of external systems typically limit them to short-term use (<6 wk). Reliable implanted drug delivery systems are available that make long-term delivery of medications to the intrathecal space feasible. These systems consist of a drug reservoir/pump implanted within the subcutaneous tissue of the abdominal wall, which is refilled periodically through an access port. The pump may be a fixed-rate, constant-flow device or a variable-rate pump that can be programmed using a wireless radiofrequency transmitter similar to those used for implanted cardiac pacemakers.

The intrathecal catheter is placed directly within the CSF of the lumbar cistern by advancing a needle between vertebral laminae at the L2-L3 level or below. Direct delivery of the opioid at the spinal level corresponding to the dermatome in which the patient is experiencing pain may improve analgesia, particularly when local anesthetics or lipophilic opioids (eg, fentanyl or sufentanil) are used. Thus, in the past, some practitioners advocated threading the catheter cephalad to the appropriate dermatome. Unfortunately, cases of inflammatory mass formation surrounding the catheter tip of some indwelling intrathecal catheters have been reported.175-179 These inflammatory masses often present with gradual neurologic deterioration caused by spinal cord compression. Currently, many physicians recommend that implanted intrathecal catheters be placed only within the lumbar cistern below the conus medullaris, where the appearance of an inflammatory mass is less likely to directly impinge on the spinal cord.180

Patient selection for intraspinal pain therapy is empiric and remains the subject of debate. In general, intrathecal drug delivery is reserved for patients with severe pain that does not respond to conservative treatment.180,181 Most patients with cancer have ongoing pain despite appropriate oral opioid therapy, or they develop intolerable side effects related to these medications. Randomized controlled trials comparing maximal medical therapy with intrathecal drug delivery for cancer-related pain have demonstrated improved pain control and reduction in opioid-related side effects in patients who received intrathecal pain therapy.182 Intrathecal drug delivery has been widely used for noncancer pain, particularly for treatment of chronic low back pain.183 However, use of this therapy in noncancer pain has not been subject to controlled trials and remains controversial.184

Once a patient is selected for intrathecal therapy, a trial is performed. Most physicians now conduct trials by placing a temporary percutaneous intrathecal catheter and infusing the analgesic agent over several days to judge the effectiveness of this therapy before a permanent system is implanted. Some carry out the trial of intrathecal therapy using a single dose or a continuous epidural infusion. The most common analgesic agent used for spinal delivery is morphine, which remains the only opioid approved by the Food and Drug Administration (FDA) for intrathecal use.

It is important to discuss the benefits and risks involved in the procedure with the patient and his or her family. Before the procedure, discuss with the patient the location of the pocket for the intrathecal pump. Most devices are large, and the only region suitable for placement is the left or right lower quadrant of the abdomen. Once the site is determined, mark the proposed skin incision with a permanent marker while the patient is in the sitting position. The position of the pocket on the abdominal wall is deceptively difficult to determine once the patient is lying on his or her side. If the location is not marked, the pocket is often placed too far lateral within the abdominal wall.

Implantation of an intrathecal drug delivery system is a minor surgical procedure that is performed in the operating room using aseptic precautions, including skin preparation, sterile draping, and use of full surgical attire.185 The procedure can be conducted under either local anesthesia or general anesthesia using dedicated anesthesia personnel. Performing the initial spinal catheter placement under general anesthesia is controversial, and concerns about neural injury are similar to those associated with any neuraxial technique performed under general anesthesia.186

The patient is positioned on a radiolucent table in the lateral decubitus position with the patient's side for the pump pocket nondependent. The arms are extended at the shoulders and secured in position so they are well away from the surgical field. The skin is prepared, and sterile drapes are applied. The radiographic C-arm is positioned across the lumbar region to provide a cross-table AP view of the lumbar spine. Care must be taken to ensure that the x-ray view is not rotated by observing that the spinous processes are indeed in the midline, halfway between the vertebral pedicles.

Surgical Technique

The L4-L5 or L5-S1 interspace is identified using fluoroscopy. The spinal needle supplied by the intrathecal device manufacturer must be used to ensure that the catheter will advance through the needle without damage. The needle is advanced using a paramedian approach starting 1 to 1.5 cm lateral to the spinous processes and just inferior to the superior margin of the lamina that forms the inferior border of the interspace you plan to enter. The needle is directed to enter the spinal space in the midline. After dural penetration, the stylette is removed to ensure adequate flow of CSF. Using fluoroscopic guidance, the spinal catheter is advanced through the needle until the tip is well into the spinal space but below L2. Position of the catheter tip is verified using fluoroscopy in the AP and lateral planes. The needle is then withdrawn slightly (∼1-2 cm) but left in place around the catheter within the subcutaneous tissues to protect the catheter during the subsequent dissection. The catheter is secured to the surgical field using a small clamp to ensure that it does not fall outside the sterile field.

A 5- to 8-cm incision parallel to the axis of the spine is extended from just cephalad to just caudad to the needle, extending directly through the needle's skin entry point. The subcutaneous tissues are divided using blunt dissection until the lumbar paraspinous fascia is visible surrounding the needle shaft. A purse string suture is created within the fascia surrounding the needle shaft. This suture is used to tighten the fascia around the catheter and prevent backflow of CSF, which may lead to a chronic subcutaneous CSF collection. The needle is removed, taking care not to dislodge the spinal catheter. Free flow of CSF from the catheter should be evident after the needle is removed; if no CSF flows from the catheter, a blunt needle can be inserted within the end of the catheter and gentle aspiration used to ensure that the catheter remains within the thecal sac. If CSF cannot be aspirated from the catheter, it should be removed and replaced. Once the provider is certain that the catheter is adequately placed, it is then secured to the paraspinous fascia using a specific anchoring device supplied by the manufacturer.

Attention is now turned to creating the pocket within the patient's abdominal wall. A 10- to 12-cm transverse incision is made along the previously marked line, and a subcutaneous pocket is created using blunt dissection. The pocket should always be created caudad to the incision. If the pocket is placed cephalad to the incision, the weight of the device on the suture line is likely to cause wound dehiscence. In many patients, the blunt dissection can be accomplished using gentle but firm pressure with the fingers. It is simpler and less traumatic to use a small pair of surgical scissors to perform the blunt dissection. After the pocket is created, the pump is placed in the pocket to ensure that the pocket is large enough. The pump should fit completely within the pocket without any part of the device extending beneath the incision.185 With the device in place, the wound margins must fall into close apposition. There should be no tension on the sutures during closure of the incision or the wound is likely to dehisce.

After pocket creation is completed, a tunneling device is extended within the subcutaneous tissues between the paraspinous incision and the pocket. The catheter is then advanced through the tunnel (most tunneling devices place a hollow plastic sleeve in the subcutaneous tissue through which the catheter can be advanced from the patient's back to pump pocket). The catheter is trimmed to a length that allows for a small loop of catheter to remain deep to the pump and attached to the pump. The pump is placed in the pocket, with a loop of catheter deep to the device. This loop allows for patient movement without placing tension on the distal catheter and causing it to be pulled from the thecal sac. Two or more sutures should be placed through the suture loops or mesh enclosure surrounding the pump and used to secure the pump to the abdominal fascia. These simple retaining sutures prevent the pump from rotating or flipping within the pocket. The skin incisions are then closed in 2 layers: a series of interrupted subcutaneous sutures to securely close the fascia overlying the pump and the catheter, followed by a skin closure using suture or staples.

Surgical Technique for Permanent Epidural Catheter Placement

For placement of a permanent epidural catheter, patient positioning and use of fluoroscopy are similar to those described for intrathecal catheter placement. The interspace of entry varies with the dermatomes that are to be covered, particularly if local anesthetic solution is to be used. A typical loss-of-resistance technique is used to identify the epidural space, and a Silastic catheter is threaded into the epidural space. A paraspinous incision is created, and the catheter is secured to the paraspinous fascia as described for intrathecal catheter placement.

Two types of permanent epidural systems are available: (1) a totally implanted system using a subcutaneous port accessed by a needle placed into the port through the skin and (2) a percutaneous catheter that is tunneled subcutaneously but exits the skin to be connected directly to an external infusion device.

To place a permanent epidural with a subcutaneous port, a 6- to 8-cm transverse incision is made overlying the costal margin halfway between the xiphoid process and the anterior axillary line. A pocket is created overlying the rib cage using blunt dissection. The catheter is then tunneled from the paraspinous region to the pocket as described for intrathecal catheter placement and secured to the port. The port must then be sutured securely to the fascia over the rib cage. Care must be taken to ensure that the port is secured firmly in a region that overlies the rib cage. If the port migrates inferiorly to lie over the abdomen, it becomes difficult to access. The rigid support of the rib cage holds the port firmly from behind, allowing for easier access to the port. The skin incisions are then closed in 2 layers: a series of interrupted subcutaneous sutures to securely close the fascia overlying the catheter, followed by a skin closure using suture or staples.

To place a permanent epidural without a subcutaneous port, a tunneling device is extended from the paraspinous incision to the right upper abdominal quadrant, just inferior to the costal margin. A small incision (∼0.5 cm) is made to allow the tunneling device to exit the skin. Percutaneous epidural catheters are supplied in 2 parts: the distal portion of the catheter that is placed within the epidural space, and the proximal portion of the catheter that enters the abdominal wall and connects with the distal portion of the catheter. The proximal portion of the catheter is secured to the tunneling device and pulled through the incision in the abdominal wall subcutaneously to emerge from the paraspinous incision. Many catheters are supplied with an antibiotic-impregnated cuff designed to arrest entry of bacteria along the track of the catheter. This cuff should be placed approximately 1 cm from the catheter's exit site along the subcutaneous catheter track. The proximal and distal portions of the catheter are trimmed, leaving enough catheter length to ensure there is no traction on the catheter with movement. The 2 ends of the catheter are connected using a stainless-steel union supplied by the manufacturer and sutured securely. The paraspinous skin incision is closed in 2 layers with a series of interrupted subcutaneous sutures to securely close the fascia overlying the catheter, followed by a skin closure using suture or staples. The skin incision at the epidural catheter's exit site in the right upper quadrant is closed around the base of the catheter using 1 or 2 simple interrupted sutures.

Complications

Bleeding and infection are risks inherent to all open surgical procedures. Bleeding within the pump pocket can lead to a hematoma surrounding the pump that may require surgical drainage. Bleeding along the subcutaneous tunneling track often causes significant bruising in the region but rarely requires treatment. Similar to other neuraxial techniques, bleeding within the epidural space can lead to significant neural compression. Infection of drug delivery devices is infrequent (range: 2.5%-9%), and most infections involve the pump pocket site.185 Signs of infection within the pump pocket typically occur within 10 to 14 days of implantation but may occur at any time. Some practitioners have reported successful treatment of superficial infections of the area overlying the pocket with oral antibiotics aimed at the offending organism and close observation alone. However, infections within the pocket or along the catheter's subcutaneous course almost universally require removal of all implanted hardware and treatment with parenteral antibiotics to eradicate infection. Catheter and deep tissue infections can extend to involve the neuraxis and result in epidural abscess formation and/or meningitis. Although meningitis is rare, each time a pump is refilled provides an opportunity for the drug to become contaminated during preparation or for the pump reservoir to become contaminated during needle penetration.185 Permanent epidural catheters without subcutaneous ports have a higher infection rate than those with ports in the first weeks after placement, but both systems have a similar high rate of infection when left in place for more than 6 to 8 weeks.187

Spinal cord injury during initial catheter placement has been reported.188,189 Most practitioners recommend placing the catheter only in the awake patient so that the patient can report paresthesia during needle placement. However, this is a topic of some debate, and placement of the intrathecal catheter under general anesthesia using radiographic guidance below the level of the conus medullaris (∼L2) is considered appropriate by some physicians. The catheter can be placed incorrectly within the subdural compartment or the epidural space. In both cases, free flow of CSF will not follow, indicating improper location of the catheter tip.

Wound dehiscence and pump migration are infrequent problems. Assuring that the size of the pocket is sufficient to prevent tension on the suture line at the time of wound closure is essential to minimize the risk of dehiscence. Pump migration usually occurs because retaining sutures were omitted at the time of pump placement. Placing 2 or more sutures through the suture loops or mesh on the pump and securely fastening them to the abdominal fascia will minimize the risk of pump migration.

Subcutaneous collection of fluid surrounding the pump (seroma) can be problematic and typically follows pump replacement. Percutaneous drainage of the sterile fluid collection often is successful in resolving the problem. Subcutaneous collection of spinal fluid, particularly in the paraspinous region, can develop, even many months after pump placement. This complication can be managed with observation alone unless the fluid collection is large or painful; in these instances, neurosurgical exposure of the spinal catheter as it enters the dura and placement of a pursestring suture around the catheter to eliminate the spinal fluid leak may be needed.

Based on a series of deaths reported through the FDA's MedWatch system, the manufacturer of the most widely used intrathecal drug delivery pump conducted a large-scale population study and concluded there is a small increase in unexplained mortality in patients with noncancer pain receiving chronic intrathecal drug delivery.190

Spinal Cord Stimulation

The idea that direct stimulation of the ascending sensory tracts within the spinal cord might interfere with the perception of chronic pain is founded in everyday observations. We all are familiar with the fact that rubbing an area that has just been injured seemingly reduces the amount of pain coming from that injured region. The advent of transcutaneous electrical stimulation (TENS) whereby a light, pleasant electrical current is passed through surface electrodes in the region of ongoing pain reinforced the observation that stimulation of sensory pathways reduces pain perception in chronic pain states. In 1965, Patrick Wall, a neurophysiologist exploring the basic physiologic mechanisms of pain transmission, and Ronald Melzack, a psychologist working with patients who had chronic pain, together proposed the gate control theory to explain how non-noxious stimulation can reduce pain perception.191 In their theory, they proposed that second-order neurons at the level of the spinal cord dorsal horn act as a "gate" through which noxious stimuli must pass to reach higher centers in the brain and be perceived as pain. If these same neurons receive input from other sensory fibers entering via the same set of neurons within the spinal cord, the non-noxious input can effectively close the gate, preventing simultaneous transmission of noxious input. Thus the light touch of rubbing an injured region or the pleasant electrical stimulation of TENS closes the gate to the noxious input of chronic pain. From this theory, investigators developed the concept of direct activation of the ascending fibers within the dorsal columns that transmit nonpainful cutaneous stimuli as a means of treating chronic pain.

We have learned much about the anatomy and physiology of pain perception since the gate control theory was first proposed. It is unlikely that the simplistic notion of a gate within the dorsal horn is responsible for our observations, but the theory served as a useful concept in the development of spinal cord stimulation. Both the peripheral nerve fibers and second-order neurons within the dorsal horn transmitting pain signals become sensitized after injury, and anatomic changes, cell death, and altered gene expression all likely have a role leading to chronic pain.192,193 Direct electrical stimulation of the dorsal columns, known as spinal cord stimulation (SCS) or dorsal column stimulation, has proved effective, particularly in the treatment of chronic radicular pain.193 North et al193 reported a prospective, randomized, crossover trial comparing SCS with reoperation in patients with persistent radicular pain after lumbosacral spine surgery. Patients randomized to SCS were considered more "successful" and less likely to cross over than those randomized to surgery. Although the sample was small, there was no difference in activities of daily living and work status. The mechanisms behind SCS remain unclear, but direct electrical stimulation within the dorsal columns may produce retrograde changes within the ascending sensory fibers that modulate the intensity of incoming noxious stimuli.

The epidural SCS lead is placed directly within the dorsal epidural space just to one side of midline using a paramedian, interlaminar approach. Entry into the epidural space is performed several levels below the final intended level of lead placement. Typically, leads for stimulation of the low back and lower extremities are placed via the L1-L2 interspace and those for upper extremity stimulation are placed via the C7-T1 interspace. Investigators have mapped the patterns of electrical stimulation of the dorsal columns and the corresponding patterns of coverage reported by patients with leads in various locations.194 In general, the epidural lead must be positioned just 2 to 3 mm to the left or right of midline on the same side as the painful region to be covered.

For lower extremity stimulation, successful coverage usually is achieved by placing the lead between the T8 and T10 vertebral levels, although upper extremity stimulation usually requires lead placement between the occiput and C3 vertebral levels. If the lead ventures too far from midline, uncomfortable stimulation of the exiting nerve roots will result. If the lead is placed too low, overlying the conus medullaris (at or below L1-L2), unpredictable patterns of stimulation may result. In the region of the conus, the fibers of the dorsal columns do not lie parallel to the midline; rather, they arc from the corresponding nerve root entering the spinal cord toward their eventual paramedian location several levels cephalad.

Patient selection for SCS is empiric and remains a subject of debate. In general, SCS is reserved for patients with severe pain who do not respond to conservative treatment. The pain responds best when relatively well localized because success of SCS depends on the ability to cover the entire painful region with electrical stimulation Attaining adequate coverage is more difficult when pain is bilateral, often requiring 2 leads, 1 to each side of midline.195,196 When the pain is diffuse, it may be impossible to get effective coverage with stimulation using SCS. Among the best established indications for SCS is chronic radicular pain with or without radiculopathy in either the upper or lower extremities. Use of SCS for treatment of chronic axial low back pain has been less satisfactory, but results seem to be improving with the advent of dual-lead systems and electrode arrays that allow for a broad area of stimulation.197 Randomized controlled trials comparing SCS with repeat surgery for patients with failed back surgery syndrome have demonstrated greater success in attaining satisfactory pain relief in those treated with SCS.193,198 In a more recent randomized controlled trial, Kumar et al demonstrated sustained improvements over 24 months in pain relief, functional capacity, and health-related quality of life in patients with persistent pain after prior lumbar surgery who received SCS.199 A small randomized controlled trial by Kemler et al200 also suggested significant improvement of pain relief in patients with CRPS who were treated with SCS in conjunction with physical therapy compared with physical therapy alone. Functional status did not improve in either group.200 The usefulness of psychologic screening before SCS remains controversial; some investigators have suggested that screening for patients with personality disorders, somatoform disorder, or hypochondriasis may improve the success rate of SCS.197,201

Once a patient is selected for therapy with SCS, a trial is carried out by placing a temporary percutaneous epidural lead. The screening is conducted using an external device as an outpatient procedure to judge the effectiveness of this therapy before a permanent system is implanted. Some carry out the trial of SCS using a surgically implanted lead that is tunneled using a lead extension that exits percutaneously. The strictly percutaneous trial lead is simpler to place and does not require a full operating room setup for placement. The surgically implanted trial lead requires placement in the operating room and surgical removal if the trial is unsuccessful. If the trial is successful, the implanted trial lead can remain, and the second procedure to place the impulse generator is brief, not requiring placement of a new epidural lead. In either case, after successful trial stimulation, a permanent system is placed and the lead is positioned to produce the same pattern of stimulation that afforded pain relief during the period of trial stimulation.

Placement of a percutaneous trial spinal cord stimulator lead can be carried out in any location that is suitable for epidural catheter placement. This may be done in the operating room but also can easily and safely be carried out in any location that allows for adequate sterile preparation of the skin and draping of the operative field. Fluoroscopy must be available to guide anatomic placement. Using a strictly percutaneous trial, the trial lead is placed in the same fashion used for permanent lead placement, but the lead is secured to the skin without any incision for the trial period.

Before permanent spinal cord stimulator implantation, discuss with the patient the location of the pocket for the impulse generator. The regions most suitable for placement are the lower quadrant of the abdomen and the lateral aspect of the buttock. Once the site is determined, mark the proposed skin incision with a permanent marker while the patient is in the sitting position. As with the pump reservoir, the position of the pocket is deceptively difficult to determine once the patient is lying on his or her side. If the location is not marked, the pocket is often placed too far lateral within the abdominal wall. Placement of the impulse generator within the buttock allows for the entire procedure to be carried out with the patient in the prone position and simplifies the operation by obviating the need to turn from the prone to lateral position halfway through implantation. If the impulse generator is placed in a pocket overlying the buttock, it must remain well below the superior margins of the iliac crest.

Implantation of a spinal cord stimulator lead and impulse generator is a minor surgical procedure that is carried out in the operating room using aseptic precautions, including skin preparation, sterile draping, and use of full surgical attire. The procedure must be conducted using local anesthesia and light enough sedation that the patient can report where he or she feels the electrical stimulation during lead placement.

The patient is positioned on a radiolucent table in the prone position. Initial lead placement can be carried out with the patient in a lateral decubitus position, but even small degrees of rotation along the spinal axis can make positioning of the lead difficult. The arms are extended upward so they are in a position of comfort well away from the surgical field. The skin is prepared and sterile drapes are applied. For stimulation in the low back and lower extremities, the radiographic C-arm is positioned directly over the thoracolumbar junction to provide an AP view of the spine. Care must be taken to ensure that the x-ray view is not rotated by observing that the spinous processes are in the midline, halfway between the vertebral pedicles.

Surgical Technique

The L1-L2 interspace is identified using fluoroscopy. The epidural needle supplied by the device manufacturer must be used to ensure that the lead will advance through the needle without damage. The needle is advanced using a paramedian approach starting 1 to 1.5 cm lateral to the spinous processes and somewhat caudad to the interspace to be entered. The needle is directed to enter the epidural space in the midline with an angle of entry not more than 45° from the plane of the epidural space. If the angle of attack of the needle on initial entry into to the epidural space is too great, the epidural lead will be difficult to thread as it negotiates the steep angle between the needle and the plane of the epidural space. The epidural space is identified using a loss-of-resistance technique. The electrode is then advanced through the needle and directed to remain just to one side of midline in the dorsal epidural space as it is threaded cephalad under fluoroscopic guidance. The electrode contains a wire stylette with a slight angulation at the tip. Gentle rotation of the electrode as it is advanced allows the operator to direct the electrode's path within the epidural space. For stimulation in the low back and lower extremities, the electrode initially is positioned 2 to 3 mm from the midline on the same side as the patient's pain between the T8 and T10 vertebral levels.

Final electrode position is attained by connecting the electrode with an external impulse generator and asking the patient where the pattern of stimulation is felt. In general, cephalad advancement results in stimulation higher in the extremity and caudad movement leads to stimulation lower in the extremity. However, if the lead is angled even slightly from medial to lateral, the pattern of stimulation may change less predictably with movement of the electrode (eg, cephalad advancement can lead to stimulation lower in the extremity under these circumstances). Final electrode position should be recorded using radiography so that a permanent lead can be placed in the same position. For trial stimulation, the needle is then removed, the electrode is secured to the back, and a sterile occlusive dressing is applied. The patient is instructed on the use of the external pulse generator and scheduled to return in 5 to 7 days for assessment of his or her response and for removal of the trial lead.

During permanent implantation, the procedure for initial lead placement is identical to that for trial stimulation. Once final lead position is attained and the optimal pattern of stimulation is confirmed, the lead must be secured, a pocket for the impulse generator created, and the lead tunneled beneath the skin to connect with the impulse generator. Following initial lead placement, the epidural needle is withdrawn slightly (∼1-2 cm) but left in place around the lead within the subcutaneous tissues to protect the lead during the subsequent incision and dissection. A 5- to 8-cm incision parallel to the axis of the spine is extended from cephalad to caudad to the needle, extending directly through the needle's skin entry point. The subcutaneous tissues are divided using blunt dissection until the lumbar paraspinous fascia is visible surrounding the needle shaft. The stylette is then removed from the lead and needle is withdrawn, taking care not to dislodge the electrode. The lead is secured to the paraspinous fascia using a specific anchoring device supplied by the manufacturer.

If lead placement has been carried out in the prone position and the impulse generator is to be placed over the buttock, this site is included in the initial skin preparation and draping, ensuring that it is below the superior margins of the iliac crest. If the generator is to be placed in the abdominal wall, the lead must be coiled beneath the skin, the paraspinous incision temporarily closed using staples, and a sterile occlusive dressing applied. The sterile drapes are then removed, and the patient is repositioned in the lateral decubitus position with the side where the abdominal pocket will lie upward. After repeat preparation of the skin and application of sterile drapes, attention is turned to creating the pocket within the patient's abdominal wall or overlying the buttock. An 8- to 10-cm transverse incision is made along the previously marked line, and a subcutaneous pocket is created using blunt dissection. The pocket should always be created caudad to the incision. If the pocket is placed cephalad to the incision, the weight of the impulse generator on the suture line is likely to cause wound dehiscence. Blunt dissection is accomplished using gentle but firm pressure with the fingers or using a small pair of surgical scissors. After the pocket is created, the impulse generator is placed in the pocket to ensure that the pocket is large enough. With the device in place, the wound margins must fall into close apposition. As with the pump reservoir, there should be no tension on the sutures during closure of the incision or the wound is more likely to dehisce.

After the pocket creation is completed, a tunneling device is extended within the subcutaneous tissues between the paraspinous incision and the pocket. The electrode is advanced through the tunnel (tunneling devices vary and are specific to each manufacturer). The means with which the electrode is connected to the impulse generator also varies by manufacturer. Some devices use a lead extension that connects the impulse generator and the lead; others use a one-piece lead that is connected directly to the impulse generator. After tunneling, the lead and/or lead extension are connected with the impulse generator. Any excess lead is coiled and placed behind the impulse generator within the pocket. This loop allows for patient movement without placing tension on the distal electrode and causing it to be pulled from the epidural space.

The skin incisions are closed in 2 layers: a series of interrupted subcutaneous sutures to securely close the fascia overlying the impulse generator within the pocket and the electrode over the paraspinous fascia, followed by a skin closure using suture or staples.

Complications

Bleeding and infection are risks inherent to all open surgical procedures. Bleeding within the impulse generator pocket can lead to a hematoma surrounding the device that may require surgical drainage. Bleeding along the subcutaneous tunneling track often causes significant bruising in the region but rarely requires treatment. Similar to other neuraxial techniques, bleeding within the epidural space can lead to significant neural compression. According to Follett et al,185 the complexity of lead implantation, including the trial period and a second-stage implant operation, may contribute to the higher multiple-site infection rate observed for SCS patients in postmarket surveillance data. Signs of infection within the impulse generator pocket also appear within 10 to 14 days after implantation but may occur at any time. Infections within the pocket or along the lead's subcutaneous course almost universally require removal of all implanted hardware and treatment with parenteral antibiotics to eradicate infection. Lead and deep tissue infections can extend to involve the neuraxis and result in epidural abscess formation and/or meningitis.

There is a significant risk of dural puncture during initial localization of the epidural space using the loss-of-resistance technique. The epidural needle used for electrode placement is a Tuohy needle that has been modified by extending the orifice to allow the electrode to pass easily. This long bevel often results in equivocal loss of resistance; it is not uncommon to have minimal resistance to injection along the entire course of needle placement. To minimize the risk of dural puncture, the needle tip can be advanced under fluoroscopic guidance and first seated on the margin of the vertebral lamina. In this way, the depth of the lamina is certain and the needle need be advanced only a small distance over the lamina, through the ligamentum flavum, and into the epidural space. Loss of resistance is used only during the final few millimeters of needle advancement over the lamina. If dural puncture does occur, there is no clear consensus on how to proceed. Some practitioners abandon the lead placement and allow 1 to 2 weeks before reattempting placement; this approach allows the practitioner to watch and treat postdural puncture headache, which is nearly certain to occur. Other practitioners proceed with lead placement through a more cephalad interspace; if postdural puncture headache ensues and does not respond to conservative treatment, an epidural blood patch is placed at the level of the dural puncture. Spinal cord and nerve root injury during initial lead placement have been reported. Placing the epidural needle and lead in the awake, lightly sedated patient who is able to report paresthesias should minimize the risk of direct neural injury.

The most frequent complication following spinal cord stimulator placement is lead migration. The first line of defense is ensuring that the lead is firmly secured to the paraspinous fascia. Suturing the lead to loose subcutaneous tissue or fat is not adequate. Postoperatively, the patient must be clearly instructed to avoid bending and twisting at the waist (lumbar leads) or bending and twisting the neck (cervical leads) for at least 4 weeks after lead placement. Placing a soft cervical collar on those who had a cervical lead placed provides an easy reminder to avoid movement. Lead fracture may occur, often months or years after placement. Avoiding midline placement or tunneling the lead across the midline will reduce the incidence of fracture caused by compression of the lead on bone. Lead fracture is signaled by a sudden loss of stimulation and is diagnosed by checking lead impedance using the spinal cord stimulator programmer.

In the rapidly changing world of modern health care, new technologies are appearing at a dizzying rate. Many of these new treatments require physicians to acquire detailed new knowledge and technical skills. Interventional pain medicine is evolving as a distinct discipline that requires detailed new knowledge and expertise. Familiarity with radiographic anatomy for the conduct of image-guided injection and the minor surgical skills needed to place implanted devices such as spinal cord stimulators and implanted drug delivery systems are just a few of the techniques that practitioners must master. As we set out to introduce new interventional techniques to our own pain practices, we must be properly trained to conduct these techniques to ensure safety and success.

The field of evidence-based medicine has emerged as a new paradigm to guide practicing physicians.202 This field endeavors to educate practitioners about how to frame specific questions based on the clinical problems they are faced with every day. They then venture to the published scientific literature with focused questions about prevention, treatment, and diagnosis of specific clinical conditions. Many evidence-based medicine centers offer concise and periodically updated summaries about specific clinical conditions. The idea is to get the best information available to the practicing clinician. It describes the best available evidence, and if there is no good evidence, it says so. In pain medicine, we are faced with an expanding array of treatment options that strike us as logical developments that should provide pain relief for our patients. However, there is a dearth of clinical evidence to guide rational choice and application of most of these emerging treatments.

Merrill203 has presented a detailed analysis of the current state of evidence guiding the use of interventional treatments in the field of pain medicine. He points out the frequent flaws in existing studies including the lack of valid comparators (eg, no treatment) and concludes that "the practice of invasive pain medicine teeters at a particularly critical juncture … crippled by a lack of vigorous self-evaluation of its role in the treatment of chronic pain." Merrill goes on to detail the means by which we, as scientists and clinicians, can proceed to build a better body of evidence for the treatments we are using.

The field of pain medicine is young and early in development, and it is perhaps unreasonable to expect an accumulation of randomized clinical trials just yet. The evidence-based medicine movement gives little guidance to practitioners whose tools are still under development. It simply reminds us that no evidence regarding many of our techniques exists. As individual practitioners, we must monitor our own outcomes using valid measures, be more reflective and systematic in studying our own outcomes and patterns of care, and provide this information to our patients as part of the decision-making process. As pain practitioners we have an expanding range of treatment options available to us, few with convincing evidence of efficacy superior to alternate treatments. We must evaluate each patient and use the limited evidence available to us today to guide compassionate and rational, if not evidence-based, use of therapy for our patients.

• Buchner M, Zeifang F, Brocai DRC, et al. Epidural corticosteroid injection in the conservative management of sciatica. Clin Orthop. 2000;375:149-156.
• Cepeda MS, Lau J, Carr DB. Defining the therapeutic role of local anesthetic sympathetic blockade in complex regional pain syndrome: a narrative and systematic review. Clin J Pain. 2002;18:216-233.
• Cohen SP, Larkin TM, Barna SA, et al. Lumbar discography: a comprehensive review of outcome studies, diagnostic accuracy and principles. Reg Anesth Pain Med. 2005;30:163-183.
• Horlocker TT, Wedel DJ, Benzon H, et al. Regional anesthesia in the anticoagulated patient: defining the risks. (The second ASRA consensus conference on neuraxial anesthesia and anticoagulation). Reg Anesth Pain Med. 2003;28:172-197.
• North RB, Wetzel FT. Spinal cord stimulation for chronic pain of spinal origin. Spine. 2002;27:2584-2591.
• Pauza KJ, Howell S, Dreyfuss P, Peloza JH, Dawson K, Bogduk N. A randomized, placebo controlled trial of intradiscal electrothermal therapy for the treatment of discogenic low back pain. Spine J. 2004;4:27-35.
• Rathmell JP. Atlas of Image-Guided Intervention in Regional Anesthesia and Pain Medicine. Philadelphia, PA: Lippincott Williams & Wilkins; 2006.
• Smitt PS, Tsafka A, Teng-van de Zande F, et al. Outcome and complications of epidural analgesia in patients with chronic cancer pain. Cancer. 1998;83:2015-2022.
• Tehranzadeh J. Discography 2000. Radiol Clin North Am. 1998;36:463-495.
• Van Kleef M, Barendse GAM, Kessels A, et al. Randomized trial of radiofrequency lumbar facet denervation for chronic low back pain. Spine. 1999;24:1937-1942.
• Ward A, Watson P, Wood C, et al. Glucocorticoid epidural for sciatica: metabolic and endocrine sequelae. Rheumatology. 2002;41:68-71.
• Wong GY, Schroeder DR, Carns PE, et al. Effect of neurolytic celiac plexus block on pain relief, quality of life, and survival in patients with unresectable pancreatic cancer. A randomized controlled trial. JAMA. 2004;291:1092-1099.