The Institute of Medicine’s report on chronic pain in America estimated that more than 100 million citizens are living with chronic pain. Of this group, chronic low back and often associated radicular pain is the most prevalent type of pain and results in significant health care expenditure and adverse social, emotional, and financial consequences in the populace. Radiculopathy is a syndrome of neurologic conductive loss (sensory or motor) arising from any compressive force (e.g., impingement of a disk on the foraminal spinal nerve, spondylosis, vertebral subluxation, ligamentum flavum cyst, infection). Experimentally, displacement of the spinal nerve with a silk loop can mimic the effects of lumbar disk protrusion and produce deficits in sensory and motor conduction. Radiculopathy is not uniformly associated with radicular pain. Radicular pain, or pain in the distribution of a spinal nerve, probably requires both pathologic compression of the nerve and inflammation. The resulting radiculitis is largely driven by glial production of cytokines.
Surgical treatment of acute radiculopathy is widely accepted by patients, but very few studies comparing surgery with comprehensive nonsurgical care have been reported. The efficacy of surgical procedures for many spinal conditions has come under greater scrutiny, though, because of disparate geographic practice patterns and inconsistent outcomes. Previous studies suggested that radicular pain can be treated effectively by diskectomy. These studies also demonstrated a favorable natural history of lumbar radicular pain syndromes, which compares well with the previous epidemiologic study of cervical radicular pain. Recent data from the Spine Pain Outcomes Research Trial (SPORT) regarding nonsurgical versus surgical care of lumbar disk herniation, however, demonstrated no statistically significant superiority of surgical diskectomy over conservative care for primary study outcomes, and thus conservative treatments such as injections for radicular pain will probably continue to increase ( Fig. 60.1 A to C ). In addition, SPORT agreed with previous trials that the natural history of discogenic radicular pain consists of slow improvement over time. Recent 5-year follow-up data from an earlier controlled trial of transforaminal epidural steroid injections suggest that conservative care may be effective in decreasing the need for disk surgery over time, with potentially lower morbidity.
During the period 1993 to 1999, the number of injections billed to the U.S. Medicare system steadily increased to 680,000 epidural steroid procedures per year, with a large number of other spinal injections increasing as well. Cervical and thoracic epidural procedures increased the most, from 10,105 to 48,210 procedures (377%). Because large studies suggest that the natural history of radiculopathy secondary to disk herniation is favorable, the challenge of successful treatment of radicular pain is to achieve a reduction in pain and restoration of function while minimizing risk or harm to the patient. The uncertainty regarding nonsurgical best practice with a seemingly increasing number of serious complications poses a significant challenge to interventional pain physicians.
History
Mixter and Barr were the first to suggest that intervertebral disk herniation might be responsible for radicular pain via mechanical compression on the exiting spinal nerve root. Four years before the idea that the disk was somehow involved, Evans had proposed treating the syndrome of primary or idiopathic sciatica with injections of 120 mL of 2% procaine (Novocaine) via the trans-sacral canal. Kelly later suggested that inflammatory changes on the nerve might be the cause of the pathophysiologic injury. Building on these ideas, Lindahl and Rexed demonstrated that inflammation, as evidenced by lymphocyte infiltration and edema, could be seen on histologic section of dorsal root nerve biopsy specimens in 7 of their 10 patients at the time of surgical diskectomy. Kelly later suggested that inflammatory changes on the nerve might be the cause of the pathophysiologic injury.
Early work on injectable corticosteroids by Hench, Kendall, and colleagues was a Nobel Prize–winning effort that culminated in synthesis of the drug hydrocortisone. Soon thereafter, corticosteroids were used in injections, initially reported as a series of knee joint injections by Hollander and associates. Theorizing that inflammation might be amenable to injection in the epidural space, the first injection of epidural corticosteroids took place in Europe in 1952 by Robecchi and Capra. Then in 1953, Li’evre and associates described their results in 20 patients in whom hydrocortisone was injected into the epidural space.
In 1961, Goebert and colleagues discussed their results of 121 injections in 113 patients over the preceding 5-year period. Of the total, all but 27 of these injections were caudal and only 3 were cervical. The injections consisted of a mixture of 1% procaine in 30-mL volumes with 125 mg hydrocortisone acetate, usually on 3 consecutive days. Between 1960 and the late 1980s, interlaminar and caudal approaches to the epidural space predominated, largely via non–image-guided techniques using surface landmarks and “loss-of-resistance” techniques. In 1988, el-Khoury and coworkers published a study suggesting that a large percentage of these “blind” epidural injections were not placed appropriately, either not being in the epidural space or not properly communicating with the target area of the injection. In a subsequent large review, Johnson and colleagues described their experience with 5489 consecutive injections that resulted in only four complications. They proposed that use of a epidurogram allowed one to provide accurate localization within the epidural space while also demonstrating that the injectant reached the target area. Furthermore, only 10 patients in the entire series required sedation, thus suggesting that the procedure was exceedingly safe on an outpatient basis.
Epidural steroid injections are now most widely used for radicular pain caused by herniated disk material. Guidelines from organizations such as the International Spinal Intervention Society, which are based on evidence, have become accepted by some practices.
Pathophysiology
Inflammation of the spinal nerve root caused by proinflammatory mediators such as prostaglandins and cytokines and by mechanical compression are now thought to be key inciting events of radicular symptoms. Corticosteroids, through their anti-inflammatory effects, may reduce these inflammatory changes in or around the nerve and thereby decrease pain and improve function.
Until recently the pathophysiology of nociceptive pain emanating from the disk has been poorly understood. Nociceptive neural structures have been demonstrated in the outer third of the annulus fibrosus with modern immunohistochemistry techniques. The nociceptive neural fibers are small unmyelinated C fibers that are activated by peptidergic neurotransmitters such as calcitonin gene–related peptide or substance P. In diskography-proven pathologic disks, nerve ingrowth into the disk may extend farther into the disk matrix, usually accompanied by neovascularization and further degradation of the disk matrix. These nerves appear to be originating from the vertebral end-plate region, which may be the progenitor of the pathologic pain attributed to the disk itself. The disk is relatively avascular, and homeostatic attempts to improve disk nutrition may explain why new blood vessels (and associated nerve fibers) invade the disk from the vertebral end-plate region. Blood flow is predominantly via passive diffusion. Circulation in the end plate is probably controlled by local neurotransmitters, which suggests that neural fibers are present as well.
The processes that lead to disk degeneration and disease may ultimately lead to herniation of the disk. Herniated nucleus pulposus results in local release of cytokines and other inflammatory mediators that cause a chemical radiculitis. Burke and associates found increased levels of the inflammatory cytokines interleukin-6 (IL-6) and IL-8 in disk material taken from patients with known disk disease. Olmarker and coworkers found that the application of disk material onto spinal nerve roots induced functional and morphologic changes in these nerves. In addition, disk cells express tumor necrosis factor-α (TNF-α), which when applied to spinal nerve roots, causes similar changes as those seen after the application of disk material, and selective inhibition of TNF-α may reduce the intraneural edema seen in this context.
Corticosteroid Alternatives
Because corticosteroids have specific dose limitations and side effects, many researchers are interested in finding alternative agents for therapeutic injection in and around the epidural space and dorsal root ganglion. Systemic delivery of specific anti-inflammatory agents for the treatment of radicular syndromes in humans has shown early promise. Korhonen and colleagues used intravenous infliximab, a TNF-α inhibitor, to treat patients with disk herniation and radicular pain and found that pain scores were reduced at the 1-year follow-up when compared with controls. Etanercept, another TNF-α inhibitor, has also shown promising but preliminary efficacy as systemic treatment of lumbosacral radicular syndromes. Cohen and associates performed a novel double-blind trial involving varying doses of etanercept in patients with subacute lumbosacral radiculopathy. The results revealed significant improvement in pain scores at 1 month when compared with saline. Unfortunately, in a follow-up, randomized, controlled, parallel-group design of epidural etanercept, saline, or steroid in which the patients and investigators were blinded, no significant difference was found between the groups. Potential shortcomings of this study were that the doses of corticosteroid were somewhat lower than in previous studies and that all patients received a small dose of bupivacaine before injection of the study drug. Previously, both human and animal data have suggested that local anesthetics may not be inferior to corticosteroids in their effects. Manchikanti and colleagues performed a prospective, randomized, double-blind trial comparing local anesthetic only with local anesthetic mixed with steroid (nonparticulate betamethasone) in fluoroscopically guided cervical interlaminar epidural injections. The study group included 60 patients with central cervical spinal stenosis and upper extremity pain. Thirty patients were randomized to each group. Multiple outcome measures were studied, including the numeric rating scale (NRS), Neck Disability Index, employment status, and opioid intake at 3, 6, and 12 months after treatment. Significant pain relief was noted in 73% in group 1 (local anesthetic only) and 70% in group 2 (local anesthetic with betamethasone); however, the preliminary results showed no significant difference in pain relief or functional status whether patients received injections with or without steroid. One of the significant limitations of the study was the small sample size ( N = 60).
Ohtori and colleagues conducted a more recent trial to test the effect of epidural etanercept on lumbar radicular pain in patients with spinal stenosis and found that etanercept was more effective than dexamethasone for leg and low back symptoms without any adverse events noted.
In another study evaluating the use of alternatives to corticosteroids, Burgher and coworkers studied the difference between clonidine and corticosteroid (triamcinolone) in transforaminal epidural injections for the indication of acute lumbosacral radiculopathy. Thirty-three patients were screened and randomized; ultimately, 26 patients were divided into two groups, clonidine (11) and steroid (15). Although no difference was seen between the groups in the primary outcome (11-point pain intensity NRS at 1 month), both groups showed significant improvement in pain scores at 2 weeks and 1 month when compared with baseline ( P < 0.05). The corticosteroid group showed additional functional improvement at 1 month in comparison to clonidine ( P = 0.022). Even though adverse effects did occur, none resulted in serious complications.
Prolonged continuous delivery of anticytokine agents to the site of inflammation has not yet been studied.
Pathophysiology
Inflammation of the spinal nerve root caused by proinflammatory mediators such as prostaglandins and cytokines and by mechanical compression are now thought to be key inciting events of radicular symptoms. Corticosteroids, through their anti-inflammatory effects, may reduce these inflammatory changes in or around the nerve and thereby decrease pain and improve function.
Until recently the pathophysiology of nociceptive pain emanating from the disk has been poorly understood. Nociceptive neural structures have been demonstrated in the outer third of the annulus fibrosus with modern immunohistochemistry techniques. The nociceptive neural fibers are small unmyelinated C fibers that are activated by peptidergic neurotransmitters such as calcitonin gene–related peptide or substance P. In diskography-proven pathologic disks, nerve ingrowth into the disk may extend farther into the disk matrix, usually accompanied by neovascularization and further degradation of the disk matrix. These nerves appear to be originating from the vertebral end-plate region, which may be the progenitor of the pathologic pain attributed to the disk itself. The disk is relatively avascular, and homeostatic attempts to improve disk nutrition may explain why new blood vessels (and associated nerve fibers) invade the disk from the vertebral end-plate region. Blood flow is predominantly via passive diffusion. Circulation in the end plate is probably controlled by local neurotransmitters, which suggests that neural fibers are present as well.
The processes that lead to disk degeneration and disease may ultimately lead to herniation of the disk. Herniated nucleus pulposus results in local release of cytokines and other inflammatory mediators that cause a chemical radiculitis. Burke and associates found increased levels of the inflammatory cytokines interleukin-6 (IL-6) and IL-8 in disk material taken from patients with known disk disease. Olmarker and coworkers found that the application of disk material onto spinal nerve roots induced functional and morphologic changes in these nerves. In addition, disk cells express tumor necrosis factor-α (TNF-α), which when applied to spinal nerve roots, causes similar changes as those seen after the application of disk material, and selective inhibition of TNF-α may reduce the intraneural edema seen in this context.
Corticosteroid Alternatives
Because corticosteroids have specific dose limitations and side effects, many researchers are interested in finding alternative agents for therapeutic injection in and around the epidural space and dorsal root ganglion. Systemic delivery of specific anti-inflammatory agents for the treatment of radicular syndromes in humans has shown early promise. Korhonen and colleagues used intravenous infliximab, a TNF-α inhibitor, to treat patients with disk herniation and radicular pain and found that pain scores were reduced at the 1-year follow-up when compared with controls. Etanercept, another TNF-α inhibitor, has also shown promising but preliminary efficacy as systemic treatment of lumbosacral radicular syndromes. Cohen and associates performed a novel double-blind trial involving varying doses of etanercept in patients with subacute lumbosacral radiculopathy. The results revealed significant improvement in pain scores at 1 month when compared with saline. Unfortunately, in a follow-up, randomized, controlled, parallel-group design of epidural etanercept, saline, or steroid in which the patients and investigators were blinded, no significant difference was found between the groups. Potential shortcomings of this study were that the doses of corticosteroid were somewhat lower than in previous studies and that all patients received a small dose of bupivacaine before injection of the study drug. Previously, both human and animal data have suggested that local anesthetics may not be inferior to corticosteroids in their effects. Manchikanti and colleagues performed a prospective, randomized, double-blind trial comparing local anesthetic only with local anesthetic mixed with steroid (nonparticulate betamethasone) in fluoroscopically guided cervical interlaminar epidural injections. The study group included 60 patients with central cervical spinal stenosis and upper extremity pain. Thirty patients were randomized to each group. Multiple outcome measures were studied, including the numeric rating scale (NRS), Neck Disability Index, employment status, and opioid intake at 3, 6, and 12 months after treatment. Significant pain relief was noted in 73% in group 1 (local anesthetic only) and 70% in group 2 (local anesthetic with betamethasone); however, the preliminary results showed no significant difference in pain relief or functional status whether patients received injections with or without steroid. One of the significant limitations of the study was the small sample size ( N = 60).
Ohtori and colleagues conducted a more recent trial to test the effect of epidural etanercept on lumbar radicular pain in patients with spinal stenosis and found that etanercept was more effective than dexamethasone for leg and low back symptoms without any adverse events noted.
In another study evaluating the use of alternatives to corticosteroids, Burgher and coworkers studied the difference between clonidine and corticosteroid (triamcinolone) in transforaminal epidural injections for the indication of acute lumbosacral radiculopathy. Thirty-three patients were screened and randomized; ultimately, 26 patients were divided into two groups, clonidine (11) and steroid (15). Although no difference was seen between the groups in the primary outcome (11-point pain intensity NRS at 1 month), both groups showed significant improvement in pain scores at 2 weeks and 1 month when compared with baseline ( P < 0.05). The corticosteroid group showed additional functional improvement at 1 month in comparison to clonidine ( P = 0.022). Even though adverse effects did occur, none resulted in serious complications.
Prolonged continuous delivery of anticytokine agents to the site of inflammation has not yet been studied.
Evidence-Based Therapy
Many reviews of trials of epidural steroid injections suggested that the early studies either were not well controlled or had methodologic deficiencies. The effect of epidural corticosteroid injections in treating radicular pain syndromes overall appears to be a transient improvement. There are multiple examples of controlled trials of procedural therapies that demonstrate little significant clinical benefit when previous noncontrolled observational studies and case series had suggested benefit. Furthermore, “epidural” describes an anatomic space that can be accessed by injection via different approaches to reach the epidural space: interlaminar, transforaminal, and caudal. The different technical approaches to the epidural space additionally complicate the study because different approaches may have different efficacy and risks. Further complicating the clinical scenario is whether the pathology is cervical, thoracic, or lumbar, given the significant pathologic differences between different areas of the spine.
Several well-designed studies have corroborated a short-term benefit with epidural steroid injections. These short-term benefits must be recognized when selecting a technique for the injection, as discussed further in the section “Complications.” Given the favorable natural history of the majority of patients with radicular pain syndromes, especially those related to disk protrusion, a thorough understanding of benefits and risks associated with epidural steroid injections for radicular pain syndromes is needed. A selection of these studies with respect to interlaminar, transforaminal, and caudal epidural steroid injections is reviewed in this section.
Carette and colleagues performed a randomized, placebo-controlled trial to investigate the efficacy of lumbar interlaminar epidural steroid injections (up to three) for sciatica ( Table 60.1 ). The study group included 158 patients with unilateral or bilateral lower extremity pain, signs of nerve root irritation or compression, and computed tomography (CT) evidence of nerve root compression at the appropriate levels. A total of 78 patients were allocated to the methylprednisolone group (80 mg) and 80 to the placebo group (saline). The primary outcome was patient function as measured by the Oswestry Disability Index. The epidural injections were not fluoroscopically guided, which may have led to a significant frequency of misplaced injections. The treatment group received 80 mg of methylprednisolone acetate mixed with 8 mL of isotonic saline or 1 mL of isotonic saline in the epidural space. The study did not show a statistically significant change in functional improvement as assessed by Oswestry scores, but it did find a reduction in leg pain as assessed by the visual analog scale (VAS) at 6 weeks in the corticosteroid treatment group (difference of −0.11 in mean change; 95% confidence interval, −21.1 to −0.9; P = 0.03); however, this improvement was no longer significant at 3 months.
| Authors | Patients | Design/Technique | Outcome Measure | Conclusion |
|---|---|---|---|---|
| Arden et al. | C = 108 T = 120 | RA, DB, PC No fluoroscopic guidance | 1—Oswestry 2—VAS, others | Improvement at 3 wk with ESI but not thereafter |
| Wilson-MacDonald et al. | C = 48 T = 44 | RA, DB, PC No fluoroscopic guidance | 1—Oswestry 2—Oxford pain chart | Improvement at 35 days with ESI but not thereafter |
| Carette et al. | C = 80 T = 78 | RA, DB, PC No fluoroscopic guidance | 1—Oswestry 2—VAS, others | Improvement in leg pain at 6 wk but not thereafter |
| Cuckler et al. | C = 31 T = 42 | RA, DB, PC No fluoroscopic guidance | 1—subjective improvement >75% | No significant improvement |
Arden and colleagues performed a randomized, placebo-controlled trial to examine the efficacy of lumbar interlaminar epidural steroid injections (up to three) for sciatica as well. The study group included 228 patients with unilateral lower extremity pain and signs of nerve root irritation. A total of 120 patients were allocated to the triamcinolone group (80 mg) and 108 allocated to the placebo (saline) group. The primary outcome was patient function as measured by the Oswestry Disability Questionnaire (ODQ). A number of secondary outcomes were studied, including a 100-mm VAS for leg and back pain. The epidural injections were not fluoroscopically guided. The treatment group received 80 mg of triamcinolone and 10 mL of 0.25% bupivacaine at weeks 0, 3, and 6. The placebo group received injections of 2 mL of normal saline into the interspinous ligament. The treatment group reported a statistically significant improvement in self-reported function as compared with placebo at 3 weeks (improvement in the ODQ adjusted for baseline, 10.3 [14.8] vs. 6.6 [15.6]; P = 0.017). The number of patients achieving a 75% improvement in the ODQ was greater in the active group than in the placebo group (15 [12.5%] vs. 4 [3.7%]; P = 0.016). At 3 weeks the number needed to treat to achieve a 75% improvement in the ODQ over and beyond placebo injection was 11.4. There were no statistically significant differences between the groups at 6 weeks or beyond on any outcome measure.
In response to the review of Koes and coworkers, Karppinen and colleagues hypothesized that the poor efficacy reported may be a result of insufficient penetration of corticosteroid into the locus of nerve root irritation when administered via an interlaminar approach ( Table 60.2 ). They conducted a randomized, double-blind trial to test the efficacy of periradicular corticosteroid injection for sciatica. The study group included 163 patients with unilateral lower extremity pain. Magnetic resonance imaging (MRI) was performed, but nerve root compression was not part of the inclusion criteria. Eighty patients were randomized to a single injection of methylprednisolone-bupivacaine and 80 were allocated to a saline injection, for a total of 160 patients. Three patients were not randomized because of an inability to produce a neurogram with fluoroscopic guidance. The primary outcome measure was back and leg pain on a 100-mm VAS. Transforaminal epidural injections were performed under fluoroscopic guidance, with injection of contrast dye to produce a neurogram. The injectant consisted of 2 to 3 mL, depending on the level, of either methylprednisolone (40 mg/mL) plus bupivacaine (5 mg/mL) or isotonic (0.9%) saline. The results showed that the effect of treatment on leg pain was significantly better in the steroid-bupivacaine group at 2 weeks, with a 45% reduction in pain versus 24% for placebo ( P < 0.01), but not thereafter. Back pain was better in the steroid group at 3 months. Back and leg pain was better in the saline group than in the steroid group at 6 months (difference of −16.2 [−26.8 to −5.6], P = 0.003 for leg pain). No difference in groups was noted at 1 year. However, only one injection was used in the study. Interestingly, they noted that a single transforaminal corticosteroid injection seemed to be associated with a rebound phenomenon at 3 and 6 months.
| Authors | Patients | Design/Technique | Outcome Measure | Conclusion |
|---|---|---|---|---|
| Karppinen et al. | C = 80 T = 80 | RA, DB, PC Fluoroscopic guidance, 1 injection | 1—VAS 2—Oswestry, others | Improved leg pain at 2 wk, no difference at 4 wk Increased back and leg pain at 6 mo in the steroid group |
| Riew et al. (both RCT and 5-yr follow-up data) | C = 27 T = 28 | RA, DB, PC Fluoroscopic guidance, up to 4 injections | 1—rate of operative intervention | Reduced surgical rates in the group receiving corticosteroid plus LA |
| Ng et al. | C = 43 T = 43 | RA, DB, PC Fluoroscopic guidance, 1 injection | 1—Oswestry 2—VAS | No significant difference between groups |
Ng and colleagues performed a randomized, double-blind controlled trial of transforaminal epidural steroid injections for sciatica as well. The study group included 88 patients with unilateral leg pain. Symptoms had to be consistent with the MRI diagnosis of nerve root compression secondary to either lumbar disk herniation or foraminal stenosis. Two patients were withdrawn because of blinding failure. A total of 43 patients were allocated to treatment with a transforaminal injection of 40 mg methylprednisolone and bupivacaine, along with 43 patients allocated to the placebo group (local anesthetic injection only). The primary outcome was a 10% change in the Oswestry Disability Index. Secondary outcomes assessed included the VAS (100 mm, with a 20% change being regarded as significant), change in walking distance, and the patient’s level of satisfaction. The study demonstrated no statistically significant difference in outcome measures between the groups assessed at 3 months.
Riew and colleagues published a prospective randomized trial of lumbar transforaminal epidural steroid injection versus local anesthetic injection in which lumbar spine surgery was used as a primary outcome. The study group included 55 patients with lower extremity pain and MRI- or CT-confirmed disk herniation or spinal stenosis. A total of 28 patients were allocated to a bupivacaine-betamethasone injection and 27 to a transforaminal bupivacaine injection only. Injections were performed under fluoroscopic guidance. They found a significant reduction in the frequency of lumbar spine surgery in patients treated with up to four transforaminal epidural steroid injections. Recently, Riew and colleagues reported 5-year follow-up data on the patients who avoided operative treatment in their 2000 randomized trial. The majority of these patients still had not undergone operative treatment. There was no statistically significant difference in the number of patients in the bupivacaine versus bupivacaine-betamethasone groups with regard to lumbar spine surgery in the interval period since the original trial.
Bush and Hillier studied caudal epidural corticosteroid injections in patients with intractable sciatica in a double-blind, placebo-controlled fashion ( Table 60.3 ). Patients with lower extremity pain and a positive straight-leg raise test were included in the study. Patients with suspected cauda equina syndrome or symptoms of less than 4 weeks’ duration were excluded. The study group included 23 patients with lower extremity pain and signs of lumbar nerve root irritation. A total of 12 patients were allocated to the steroid group and 11 to the placebo group. The caudal epidural injection was performed without fluoroscopic guidance, with 25 mL of injectant being administered, including 80 mg triamcinolone and 0.5% procaine hydrochloride. The placebo group received 25 mL of saline. Outcome measures studied included a symptomatology questionnaire and a VAS. The treatment group showed statistically significant improvement in lifestyle and reduction of pain at 4 weeks (VAS score of 16 vs. 45, P = 0.02). At 1 year, both groups showed statistically significant resolution of their symptoms. No major side effects were reported.
