35 Epidural Steroid Injections

Cervical, Thoracic, and Lumbar: Transforaminal, Interlaminar, and Caudal

Milton H. Landers, DO, PhD, Charles N. Aprill, MD

Drink deep or taste not from the Pierian spring: There shallow draughts intoxicate the brain, And drinking largely sobers us again.

Alexander Pope

History

The term epidural merely denotes a space within the spinal canal but is applied to a variety of injection procedures. They may be classified as to region of interest: cervical, thoracic, or lumbar, and route of administration—caudal, interlaminar, or transforaminal. Each procedure involves directing a needle to a specific target employing an imaging modality to guide and document final needle position. This chapter limits discussion to fluoroscopic guidance. All of these procedures require injection of a contrast material to demonstrate the dispersal pattern of the injectates and documentation of that pattern with spot films. Finally, a report detailing the procedure, including specific technique, details of injected solutions (contrast material, local anesthetic, and corticosteroids), and recording the initial outcome is prepared for the permanent record.

There is a report in English literature that refers to the caudal injection approach for pain management in 1930.¹ The first recorded use of corticosteroids injected into the epidural space for the treatment of “sciatica” was in 1952.² The route of administration was one of the dorsal foramina of the sacrum, an early, blind, transforaminal technique. This technique was employed in Europe in the 1950s and 60s.³

Interlaminar approaches to the lumbar epidural space were reported in the 1960s.⁴^–⁶ Caudal and interlaminar techniques became standard practice in the United States, Britain, and most of continental Europe for the next quarter century. Interest in the transforaminal approach reemerged in the last quarter of the 20th century, in an attempt to deliver more target-specific injections.⁷^–⁹ The cervical literature is sparse by comparison to that of the lumbar spine. The interlaminar approach for epidural injections was generally used in the 1980s.¹⁰^–¹² Cervical transforaminal injection was first reported in 1988.¹³ A descriptive outcome study by Bush and Hillier employing this approach followed in 1996.¹⁴ The transforaminal approach was adopted to deliver therapeutic agents to more specific targets influenced by observations that lumbar transforaminal injections appeared to be effective in treating lumbar radicular pain.^15,¹⁶

There are few reports of the origin of thoracic epidural steroid injections (ESIs) for treatment of spinal pain.^17,¹⁸ A specific technique for thoracic transforaminal epidural injection was described in 2001.¹⁹

A recent systematic review revealed an increase in the number of lumbosacral spine injection procedures performed in the last decade of the twentieth century. There was an increase of more than 270% noted between 1994 and 2001.²⁰ Forty percent of these procedures were associated with the diagnosis codes for “sciatica,” “radiculopathy” or “herniated disc.” “Axial low back pain” diagnoses accounted for 36% and “spinal stenosis” 23%. A study of the Medicare population between 1993 and 1999 revealed that the annual number of cervical, thoracic, and lumbar epidural steroid injections peaked at more than 680,000 in 1998.²¹ These studies focused on the Medicare population, but it is likely that similar increases would be found for the general adult pain population.

A review of conservative care for sciatica published in 2000 surveyed 19 randomized controlled trials and pooled odds ratios calculated for several treatment types. The data indicated, “Epidural steroids may be beneficial for subgroups of nerve root compression.”²²

In 2009, Lavin published results of an extensive MEDLINE/PubMed search and searches of large review articles on the major interventional spine topics. This review was performed to find all prospective, double-blind, randomized, placebo-controlled trials in the English language interventional spine literature. He concluded “Fluoroscopically guided lumbosacral transforaminal epidural corticosteroid injections are effective in the treatment of acute/subacute lumbosacral radicular pain, and in preventing future surgeries. No firm conclusions were drawn about cervical epidural (CESI) injections or lumbosacral epidural (LESI) injections for the treatment of chronic radicular pain.²³ A study of the American Society of Anesthesiologists (ASA) Closed Claims Project database revealed that epidural steroid injections account for more complications than any other interventional pain procedure.²⁴

Relevant Anatomy

The spinal epidural space is a continuous potential space in the vertebral canal extending from the foramen magnum to the sacral hiatus, lying between the dural sac and the osseoligamentous wall of the vertebral canal. The epidural space is bounded posteriorly by the lamina and ligamentum flavum. The anterior boundary is formed by the posterior disc margin, the posterior longitudinal ligament and the peridural membrane. This membrane attached to the upper and lower ring apophysis of each vertebra, blends with the disc annulus. It spans the waist of the vertebral body.²⁵ Together these tissues form the floor of the vertebral canal. The lateral boundary of the epidural space is fenestrated by paired root canals or foramina at each segment. The vertebral pedicles lie between the root canals.

The dural sac nearly fills the vertebral canal and extends from the foramen magnum terminating in the sacral canal at the level of the S1 to S2 segments. There is some variability and the thecal sac may terminate as high as the lumbosacral interspace or may extend to S3. At each vertebral segment paired dural root sleeves extend toward the root canals. The cervical and thoracic dural root sleeves are short and nearly horizontal in orientation, traversing the center of their respective root canals. The lumbar root sleeves are longer, increasing slightly in length and downward angle from L1 to the sacrum. Lumbar root sleeves turn laterally into grooves in the vertebral bodies and pedicles and are in intimate proximity to the medial/inferior border of the pedicles. This is the entrance zone of the root canal, or lateral recess of the vertebral canal. The dural root sleeve contains the dorsal and ventral rootlets of the segmental nerve. The dorsal ganglia lie generally in the midzone of the root canals, between the pedicles of adjacent vertebra. The dura and arachnoid membranes form a “water-tight” seal at the proximal end of the ganglia. Cerebrospinal fluid (CSF) fills the root sleeve but does not extend beyond the ganglia. Occasionally ectasia of a dural root sleeve will occur, resulting in extension of the subarachnoid space beyond usual boundaries. Referred to as cystic root sleeve dilatation or Tarlov cysts, they usually communicate freely with the subarachnoid space.²⁶ They may occur at any level of the spine but are most frequent in the sacral region. Usually asymptomatic, they are rarely associated with symptoms.²⁷

There are accumulations of fat in the posterior vertebral canal, deep to the ligamentum flavum and posterior to the dural sac in the thoracic and lumbar spine.²⁸ This posterior fat pad may be small or absent at the L5-S1 level when the dural sac is full size and is applied close to the lamina and ligamentum flavum, as it frequently is. Additional accumulations of fat are found laterally at the level of the discs within the root canals. There is little fat in the cervical epidural space. Midline sagittal spin echo (SE) T1 images of the spine demonstrate the lumbar and thoracic fat pads quite well. Fat is rarely visualized in the vertebral canal above the C7-T1 level. Ample fat fills the vertebral canal below the termination of the dural sac in the sacral vertebral canal.

A thin anterior membrane extends from the posterior longitudinal ligament to the anterior surface of the dural sac. This ventral meningovertebral ligament forms a nearly continuous septum, that functionally divides the ventral epidural space into left and right compartments in the lumbar region.²⁹ A midline dorsal connective tissue band fixes the dural sac to the flaval ligaments. The appearance of the band varies from strands of connective tissue to a complete membrane.³⁰

A plexus of valveless veins occupies the anterior epidural space.³¹ Paired anterior internal epidural veins run the length of the vertebral canal. There are numerous interconnecting veins crossing the floor of the canal, the anterior internal vertebral venous plexus (AIVV).³² A smaller posterior internal venous plexus lines the roof of the epidural space in the lumbar and thoracic regions. Extending laterally around the vertebra, the peridural membrane encloses veins that course around the body of the vertebra.²⁵ These veins communicate freely with the basivertebral veins, at the center of the posterior wall of each vertebral body and the anterior internal vertebral venous plexus. Several radicular veins traverse each root canal. They lie predominantly in the ventral root canal and connect the epidural venous plexus with the ascending lumbar veins and the vena cava.

Segmental arteries traverse the root canals at all spinal levels.³³ They course in the anterior superior root canals along the anterior superior aspect of the segmental dural root sleeve. Some arterial branches perfuse the dorsal ganglia and segmental neural elements and, as such, are radicular arteries. Other branches penetrate the dura running along the segmental rootlets and may anastomose with the anterior spinal artery. They are, therefore, medullary arteries. Varying numbers of medullary arteries of differing sizes are found in the cervical, thoracic, and lumbar region, and contribute to the blood supply of the spinal cord.

The radicular and medullary arteries of the cervical spine arise primarily from the vertebral artery.^34,³⁵ Some arise from the ascending cervical and posterior cervical arteries. Medullary arteries penetrate the dura and divide into anterior and posterior branches, eventually reaching the anterior and paired posterior spinal arteries. Cervical medullary arteries are variable in size and distribution. There are usually two or three arteries of moderate size in the adult. They may be present at any segment and on either side.^36,³⁷

The upper and midthoracic cord is supplied by medullary arteries arising from branches of the costocervical trunk and spinal branches of the aorta. There is usually a single large medullary artery arising from the aorta and supplying blood to the lower half of the thoracic cord and conus. This large medullary artery, the artery of Adamkiewicz, has a variable origin but usually originates on the left side (80% of patients), between T7 to L4, most often between T9 and T11.³⁸ In a study of 4000 spinal cord angiograms, there were three instances in which the major medullary artery of the distal thoracic cord originated at the fourth lumbar artery.³⁹

Paired lumbar arteries originate from the aorta at each lumbar segment. These give rise to spinal branches that include radicular and medullary branches.⁴⁰ Most lumbar spinal branches are small. Some may be 1.0 to 2.0 mm in diameter. There is potential for a medullary artery of substantial size to be present in any root canal.⁴¹

Spondylogenic Pain

Spondylogenic pain arises from a component of the spine or its supporting structures, and is synonymous with other terms such as nonspecific or idiopathic spinal pain. The International Association for Study of Pain (IASP) published a taxonomy, defining clinical terms used to describe pain, in an effort to standardize the use of terms.⁴²

Somatic pain results from noxious stimulation of a musculoskeletal component of the body and arises as a result of stimulation of nociceptive nerve endings in bone, ligament, joint, muscle or tendon. Visceral pain occurs with noxious stimulation of a body organ or its capsule. Both somatic and visceral derangements cause pain in the location of the source structure, but it is often vague and ill defined.

Referred pain is pain perceived in areas adjacent to or remote from the source of pain. Convergence is the physiologic basis for referred pain. Sensory neurons from different peripheral sites converge on common neurons in the spinal cord and thalamus that relay to higher centers in the brain. Without additional sensory input, the brain may not be able to determine the specific site of the initial input. Referred pain is a misperception of the origin of the signal that reaches the brain.

Somatic spinal pain syndromes may be expressed topographically as neck, midback, or low back pain. Local pain in these areas is often accompanied by referred pain. Somatic neck pain may be perceived with pain referred to the shoulder girdle or upper extremity.

Lesions of the midback refer to the chest wall and low back pain is commonly associated with pain in the buttocks and lower extremity. Somatic referred pain is typically diffuse, deep, and ill defined, aching in quality and aggravating.

Radicular pain occurs when there is irritation or injury of a spinal nerve or its roots. In the lumbar spine, simple compression of a normal nerve or nerve root does not provoke pain.⁴³ Alone, pressure on a spinal nerve or its roots may produce a conduction block of axons. Sensory axon conduction block results in loss of sensation or “numbness.” Motor axon block results in weakness. Compression alone can result in radiculopathy, without radicular pain. Pressure on a previously injured or damaged nerve or root provokes pain. Local ischemia or inflammation can induce ectopic impulses in a dorsal root canal ganglion resulting in radicular pain. Lumbar radicular pain is discretely localized, “bandlike” in distribution, and shooting, lancinating, or shocking in quality⁴⁴ and can have a cutaneous component.

Preprocedure Evaluation

A practitioner of spinal interventions may be a managing physician directing the care of a patient with a painful spinal condition or a consultant to whom a patient has been referred for a procedure. With each encounter, the responsibility for assessing the clinical condition and the appropriateness and risks of a given procedure, rests with the physician injectionist. The practitioner must attempt to understand the pain and associated disability. The patient is the primary source of that information.

What is the primary complaint? Is this an acute or chronic problem; a constant or recurring problem? Is local pain somatic or visceral? Is there a radicular or somatic referred component? One must localize the somatic component and the location of radiating pattern. The character and intensity are as important as the location. What are the mechanical aspects, that is, what provokes and relieves the pain? An example: A 35-year-old female presents with recent onset of left posterior and lateral aching neck pain for no apparent reason. Constantly present, the aching varies from moderate to severe on occasion. There is diffuse aching of left posterior shoulder girdle muscles. She has recurring episodes of severe, “sharp, stabbing” left upper extremity pain, extending along the posterior arm, medial forearm, and into the left long finger. No arm weakness, frequent “tingling” of the whole upper extremity and normal strength and reflexes are observed. Prolonged sitting, particularly at a console or working at a desk, aggravates the neck pain. Neck extension and left side bend induces a painful “shocking” sensation in the left arm. Though intermittent, the arm pain is the dominant complaint. She has axial somatic neck pain, with somatic referred pain to the left shoulder girdle and left upper extremity radicular pain, without radiculopathy.

Self-pain rating, employing a specific method such as the visual analog scale (VAS) is essential. What is the intensity rating at the time of assessment? Range of intensity is determined by noting the pain rating at its best and worst. Patients can express their pain on pain drawings, which should be correlated with the description of symptoms. These instruments are very useful in assessing the immediate response to an intervention.

Simple psyche screening questionnaires to assess anxiety, depression, self-perception of disability, and somatization are useful in assessing pain behavior. What has been the response to any previous interventions or procedures? Finally, what medications, at what doses have been used to treat and what has been the response? Psychological factors contribute to pain and influence the outcome of interventions. There is no simple screening tool. Therefore, one must know his or her patient and history.

Medical history includes review of all allergies and idiosyncratic reactions, any current illness or infection and chronic conditions such as cardiovascular disease, diabetes, seizures, respiratory and renal disorders. All these factors affect how procedures will be performed.

The history of all surgical procedures and their outcomes should be explored. A complete list of current medications is mandatory. A general review of systems should include information about sleeping pattern. Social habits such as tobacco and alcohol use or abuse are important. Patient’s work and/or avocations may pertain to complaints and should be explored. A family history may contribute to recognition of an underlying hereditary disorder.

The physical examination on a first encounter is general but with focus on primary and secondary complaints. Developmental status, any deformities or postural aberrations should be recorded. The level of distress during interview and examination, patient’s willingness and ability to cooperate with interview/examination and the patient’s demeanor and orientation should be observed and recorded along with routine vital signs. Examination may include assessment of gait and both sitting and standing postures. Passive or active movements of the cervical, thoracic, or lumbar spine through their range of motion are observed. Palpation of regional musculature and bony prominences helps to localize somatic pain. Epidural procedures are injections into the spinal neuraxis and require a basic neurologic examination prior to the procedure. On a case-by-case basis, examination of core and extremity strength, weakness, range of motion, and reflexes contribute to formulating a working hypothesis.

Spine imaging plays a definite role in the evaluation of spine pain complaints both in demonstrating relevant pathology and aiding in planning a safe and efficient approach for procedures. Modern imaging modalities provide excellent depictions of spine anatomy. However, many lesions, evident in the maturing spine, are clinically insignificant.⁴⁵ Imaging studies often demonstrate pathology that is irrelevant. Computed tomography (CT) or magnetic resonance imaging (MRI) may demonstrate a left C3-4 disc protrusion in a patient complaining of right arm pain and altered sensation radiating to the right forearm, thumb, and index finger. It cannot be overstated that there is a significant incidence of pathology in asymptomatic subjects.⁴⁶ A focal disc protrusion on the right at L3-4 may be clearly defined by MRI, in a patient complaining of left buttock pain extending to the left calf.⁴⁷ Scans should be reviewed overall, but with particular attention to areas that might be related to the patient’s underlying complaints, as gleaned from the interview and examination. One should carefully study the left L4-5 and L5-S1 root canals in a patient with pain extending from the left buttock to the left anterolateral leg and dorsum of the foot. Recently 150 patients with unilateral leg pain, and MRI confirming neural compression by prolapsed disc or foraminal stenosis had lumbar transforaminal epidural injections (LTFE) at appropriate sites. These were effective in relieving their pain initially, and only18% of these patients required surgery at a minimum 1-year follow up.⁴⁸ Not all abnormalities need be targeted but rather those most likely to be associated with the expressed symptoms and physical findings.

Indications: This clinical process should provide a basis for a hypothesis as to the nature of the underlying spinal disorder. Epidural injections are indicated when a patient expresses constant or frequently recurring episodes of radicular pain, but also patients with somatic spinal pain associated with mixed somatic referred and radicular pain referred to one or both extremities or the chest wall.

Contraindications: Absolute contraindications include a bleeding diathesis, systemic infection or local infection in the procedure field, pregnancy, and inability or unwillingness to cooperate and participate in the procedure. Relative contraindications include significant competing disease—that is, uncontrolled hypertension, diabetes, congestive heart failure, and anticoagulation therapy.

There is risk of a major depressive disorder in patients with chronic pain. Pure psychogenic pain is rare and used pejoratively. Almost all chronic pain can be altered by psychological adjustments and issues. Without these being addressed, successful diagnosis and treatment is unlikely.

Absolute contraindications include active psychosis and homicidal or suicidal ideation. Studies have shown a relationship between physical and sexual abuse and various types of pain. Correlation of failed low back surgery with childhood trauma has revealed definite risk factors. These include physical, psychological, and/or sexual abuse, emotional neglect, abandonment, and chemical dependence and/or loss of a primary caregiver.⁴⁹^–⁵¹ Relative contraindications include personality disorders, bi-polar disorder, severe depression, lack of social support, and drug dependence/abuse.

Negative predictors are external locus of control, “just fix me,” unstable relationships, poor vocational adjustment, neurocognitive deficits, inability or unwillingness to participate in active rehabilitation, unreasonable expectation of procedures, ongoing litigation, symptoms inconsistent with pathology, high fear avoidance, inability to actively exercise, and catastrophizing cognitive style.

Positive predictors are history of adherence to treatment regimen, positive support system, compliance (currently and historically), and appropriate expectations.

A Few Medical Issues

Vasovagal

Vasovagal responses are not rare. About 1 in 14 patients undergoing cervical interlaminar epidural injection will experience a vagal response. Incidence varies by gender, women are at greater risk than men and with procedures, vagal response is five times more common in cervical than in lumbar procedures. Although they are usually a minor complication, vasovagal events may be quite severe. Vigilant screening is necessary to discover previous episodes; suspect “fainters.” Pretreat with atropine when suspicious and recognize the early signs (i.e., “I feel sick,” nausea, and slowing of pulse rate). However, these doses are much in excess of those required for any interventional pain procedure.

Contrast Material

Allergic reactions to modern contrast materials are extremely rare. Some may be related to preservatives such as methylparaben. Wang reviewed 84,928 intravenous contrast injections noting “allergic” reactions in 545 (0.6%).⁵² Seventy seven percent of the reactions were “mild,” primarily urticaria, successfully treated with diphenhydramine. Twenty one percent had moderate reactions with respiratory symptoms, facial edema, or both. Corticosteroids intravenously or by inhaler were effective in controlling these reactions. Two percent, eight women and three men, had severe reactions (0.0012%). Three were unresponsive; frank cardiorespiratory arrest occurred in one. There were no deaths and no long-term sequelae in 10 of the 11 patients with severe reactions.

Contrast material is employed to document where the injectate is spreading and to determine where it is not spreading. Modern nonionic contrast agents are safe for all epidural injections. Unintentional vascular or even subarachnoid spread will cause no problem. In those very rare instances of documented history of severe allergic reaction to nonionic contrast material, gadolinium may be employed as a substitute radiography contrast agent.⁵³

Diabetics taking metformin are at risk for contrast material nephropathy (CMN) if exposed to nonionic contrast material, during radiographic procedures. Solomon found no change in renal function in those subjects who had normal renal function prior to testing.⁵⁴ However, some patients with normal renal function exhibited increases of serum creatinine with high-contrast loads.⁵⁵ This should not be a problem with epidural injections because the contrast doses employed are limited and much lower than the volumes delivered in the radiologic procedures. Doses in excess of 100 mL are not unusual for enhanced CT or angiograms. However, diabetic patients with renal impairment should discontinue metformin, with approval of their primary physician.

Local Anesthetics

The local anesthetics employed in the epidural injections are lidocaine and bupivacaine. These amides are primarily metabolized in the liver and excreted in the urine. Their allergic potential is extremely low. Ester-based local anesthetics (procaine) have a higher incidence of allergic reaction. A large survey attests to the overall safety of the amides as spinal anesthetics.⁵⁶ For epidural injections, lidocaine may be employed in concentrations of 0.5% to 4.0% and bupivacaine at 0.25% to 0.75%. The percentage concentration of the anesthetic is reduced with increasing volume delivered. Although central nervous system and cardiovascular system toxicity can occur following the epidural, intravenous, or subarachnoid injection of local anesthetics, the toxic doses are far greater than the volumes employed in epidural injections—40 mL of 1.0% lidocaine (400 mg) delivered to the epidural space of test subjects resulted in average peak plasma concentrations safely below neurotoxic levels.⁵⁷ One author recommended the maximum epidural dose of lidocaine to be 500 mg and 225 mg for bupivacaine.⁵⁸

Acute cardiac complications, decreased cardiac output being one, can occur from intravascular injection of lidocaine or bupivacaine. Study has shown that normal cardiac function is maintained in subjects with lidocaine plasma levels of 4 to 8 mcg/mL. Plasma levels this high require doses of >400 mg of lidocaine.⁵⁹ In human volunteers slow intravenous (IV) injection of bupivacaine to produce plasma concentrations equivalent to those achieved during epidural injections did not cause significant cardiovascular change.⁶⁰ Between 3.0 and 7.0 mL of low concentrations of anesthetics for interlaminar injections and 0.75 to 2.0 mL of higher concentrations for transforaminal injections are not toxic in the epidural space. These doses are not toxic if technical misadventure results in delivery to the vascular system or subarachnoid space. Because of the longer half-life of bupivacaine, this anesthetic should be avoided in interlaminar epidural injections, particularly in the cervical spine. Inadvertent delivery into the subarachnoid or subdural space results in prolonged effects, specifically affecting respiratory function with misplaced cervical injection. Decreased concentration of anesthetics must be considered for interlaminar epidural injections in the cervical and thoracic regions in patients with respiratory or cardiovascular health problems because decreased cardiac output and heart rate can occur with sympathetic blockade, which occurs to some degree with all interlaminar epidural injections.

Corticosteroids

The steroid preparations most frequently employed in ESI are methylprednisolone acetate, triamcinolone acetonide, betamethasone sodium phosphate, betamethasone acetate, and dexamethasone sodium phosphate. All corticosteroids have systemic effects when injected in the epidural space. Side effects may include: fever, myalgia, malaise, fluid and electrolyte imbalance, hypertension, hyperglycemia, myopathy, ulcers, immunosuppression, behavioral changes, allergic reaction, and pituitary-adrenal suppression. Abrupt withdrawal after prolonged oral use may precipitate acute adrenal insufficiency. Most of these side effects are clearly associated with the systemic absorption of corticosteroid and subsequent transient hypercorticism, and likely account for many of the “minor” complications associated with epidural injections.

The cardiovascular system is affected directly and indirectly. Direct effects are due to steroid receptors on heart and smooth muscle. Indirectly, steroids increase sodium uptake and vascular tone, raising blood pressure. Judicious use is prudent in patients with cardiac disease, hypertension, and congestive failure. Dose and frequency of administration must be carefully monitored in this group.

In the central nervous system, nerve tissues exhibit increased excitability, EEG abnormalities are seen, subjects experience euphoria and behavioral changes. Rarely, these effects can be severe. A case report describes 67-year-old male who developed psychotic episodes within a week of a cervical ESI, along with other steroid injections in one treatment session. The symptoms spontaneously resolved in approximately 7 to 10 days.⁶¹ This case serves to remind us that we must always be cognizant of the doses we deliver, and remember patients may be receiving “steroid shots” from other physicians for unrelated problems.

Endocrine system alterations include decreased adrenocorticotropic hormone (ACTH), thyroid-stimulating hormone (TSH), and follicle-stimulating hormore (FSH), and testosterone. The changes are largely dose related. Bizarre responses do occur rarely. A 24-year-old man underwent a second cervical ESI of 60 mg of methylprednisolone for ongoing severe arm pain. A cushingoid appearance developed 1 month later. Serum cortisol was undetectable, there was no adrenal response to synthetic ACTH, and urinary-free cortisol was below normal at 12 weeks. Cortisol normalized in 4 months, however, the patient’s cushingoid appearance persisted for 12 months.⁶²

An epidural steroid injection of 160 mg of methylprednisolone will suppress adrenal function for weeks.⁶³ Does injection of corticosteroids increase the risk of infection? Infections do occur in patients following ESIs.⁶⁴ Postoperative infections are increased in hip replacement patients who received preoperative steroid injections.⁶⁵ No studies have been done on outcomes of spine surgery after ESI, but suppression of the immune system suggests there is inherent risk, especially in the first 4 weeks post injection.

Epidural administration of glucocorticoids results in potent suppression of insulin action. This should be taken into account when patients with diabetes receive ESIs.⁶⁶ Lumbosacral transforaminal and caudal epidural betamethasone injections are associated with statistically significant elevations in blood glucose levels in diabetic subjects. This effect peaks on the day of the injection and lasts approximately 2 days.⁶⁷ The administration of three epidural injections (250 mg prednisone equivalent) was followed by suppression of the corticotropic axis that persisted beyond 21 days. Plasma cortisol and ACTH and urinary free cortisol were markedly reduced at the day 1 and day 7 posttreatment visits, compared to baseline. At 21 days, these variables were still diminished.⁶⁸ We must be vigilant in follow-up of our diabetic patients.

The musculoskeletal system suffers from some bone mineral density loss, muscle weakness and wasting, and predisposition toward avascular necrosis with long-term oral steroid use. Although systemic corticosteroids may be associated with loss of bone mineral density, a prospective study of 204 patients followed for 1 year after standard doses of epidural steroids revealed no change in bone mineral density.⁶⁹

The currently used steroids contain preservatives. Bernat thought that seizures following intrathecal cortisone acetate were due to the preservative, 0.9% benzyl alcohol. Benzyl alcohol is common in depot steroids and is neurotoxic.⁷⁰ Hodgson reported that there was no concrete evidence of neurotoxicity of intrathecal steroids, but did not recommend purposeful delivery of these agents into the subarachnoid space.⁵⁶

Major Complications

Directing a needle deep into the tissues in and about the spine approaches sensitive spaces and structures of the neural axis. There is a spectrum of possible major complications.

Infection

Infection following epidural injection is a serious issue. It has been described as a rare complication.⁷¹ However, the possibility of procedure-related infection must be a concern for the injectionist. Analysis of the American Society of Anesthesiologists (ASA) Closed Claims database from 1970 to 1999 revealed that infection was the third most common complication of chronic pain procedures, accounting for 13% of all complications.²⁴ Any penetration of the skin surface with a needle then placed in deep spaces in and about the spine has the potential of introducing organisms with subsequent infection.⁷² Standard aseptic preparation and draping of the local injection site and good aseptic technique will minimize this complication. Diabetic and immunocompromised patients are a subgroup that should be recognized prior to procedures. Antibiotic prophylaxis should be considered for immunocompromised patients undergoing ESI.⁷³ Antibiotics should not be mixed with the contrast during any epidural injection. Accidental delivery of antibiotics into the subarachnoid space must be avoided.

Bleeding/Hematoma

Many veins lie in the needle path or near endpoint targets of transforaminal epidural injections (TFESI). Furman and colleagues reported the incidence of fluoroscopically confirmed “vascular uptake” occurring during lumbar ⁷⁴ and cervical⁷⁵ transforaminal injections. Observing blood at the needle hub reliably predicts intravascular injection, (97% specific). But this finding has low sensitivity, (45% overall). The absence of blood in the needle hub despite aspiration is not reliable. The incidence of vascular uptake with lumbar procedures is about 11%. There is a higher incidence in the cervical spine where confirmed intravascular injection was observed in nearly 20% of patients. Kim and coworkers have recently reported a much higher rate of vascular uptake for cervical transforaminal injections, more than 50%.⁷⁶ Although neither author specified, most of the vascular uptake seemed to be venous. No complications occurred. Venous uptake during epidural injections is common and innocuous as long as it is recognized and needle adjustment is made before injecting physiologic solutions. Veins are less abundant in the posterior epidural space and venous uptake is less frequently seen during interlaminar epidural injections.

Three surveys of cervical interlaminar epidural injection on more than 2200 patients claim no major complications.⁷⁶^–⁷⁸ Minor adverse responses, such as increased headache, insomnia, vasovagal episodes, facial flushing, dural puncture, and short-term nocturnal fever were encountered. Similar response profiles occur with thoracic⁷⁹ and lumbar interlaminar injections and caudal epidurals as well. Most of these are likely a response to the corticosteroid and are unavoidable for the most part.

The incidence of epidural hematoma is approximated to be less than 1 in 150,000 epidural injections.⁸⁰ Clinically relevant hematoma may occur at any level, the majority of reported cases were associated with injections in the cervicothoracic area. Spinal epidural hematoma causing acute myelopathy is a rare complication. In a case report, Stoll and colleagues noted epidural hematoma after ESI in a young man with no predisposing factors.⁸¹ An older patient with apparently normal coagulation received multiple cervical ESIs for palliative pain control over several years. A hematoma that required surgical decompression occurred following the seventh procedure.⁸²

The potential for bleeding and hematoma formation is increased in patients with a coagulopathy, liver disease, or in patients taking anticoagulant medications. It is crucial that the injectionist be familiar with a patient’s coagulation status and with common anticoagulant and antiplatelet medications.⁸³ Anticoagulant therapy should not be interrupted until there is full understanding of the reason for the therapy. Actual discontinuation of anticoagulant therapy is usually best left to the physician managing and prescribing the patient’s anticoagulation. Epidural injections are elective procedures. Interrupting anticoagulant therapy may increase the risk of serious complications such as stroke, myocardial infarct, or pulmonary embolus.

The following guidelines for performing spinal procedures in anticoagulated patients are based on the second American Society of Regional Anesthesia and Pain Medicine (ASRA) Consensus Conference on Neuraxial Anesthesia and Anticoagulation in 2003.⁸⁰ Warfarin therapy should be discontinued 4 to 5 days before spinal procedures and the international normalized ratio (INR) should be within normal range at the time of the procedure. Thienopyridine derivatives, (e.g., clopidogrel, ticlopidine) should be suspended 7 days and 14 days, respectively, prior to spinal procedures to allow for recovery of primary and secondary platelet aggregation and platelet-fibrinogen binding.

Low-molecular weight heparin should be held for at least 12 hours before the procedure in thromboprophylactic dosing and at least 24 hours in therapeutic dosing. Aspirin and nonsteroidal antiinflammatory drugs have not been found to have any contraindications for spinal procedures. It may be prudent to discontinue full dose aspirin therapy for 10 days in patients undergoing interlaminar procedures.

Dural Puncture

During interlaminar epidural injections, the dural sac is at risk. Transforaminal injections place the dural root sleeves at risk. Nonfluoroscopically guided interlaminar injections have increased incidence of dural puncture with prevalence up to 5.0% in cervical epidural injections.⁸⁴ Dural puncture is reportedly rare when the procedure is performed by “expert” interventionalists.⁸⁵ Minor complications, including headache secondary to dural puncture occurred in less than 0.5% in more than 4300 procedures. Dural puncture may occur with more frequency in the cervical spine above C7 and at L5-S1. In both areas there may be a paucity of epidural fat with the dural sac juxtaposed against the flaval ligament. Failure to recognize subarachnoid spread results in injection of anesthetic and steroid into the subarachnoid or subdural space. Delivery of anesthetics in sufficient quantity and concentration may produce significant spinal block, most serious at cervical levels where the block may impair respiration. These misadventures are avoided by selection of appropriate target, proper needle, good visualization, a cooperative, comfortable—but minimally sedated—patient and precise technique.

Subdural injections between the dural and arachnoid layers are rare but can and do occur. The injectionist must be able to recognize contrast injection into this space as noted by a cystlike appearance. Contrast injected into the subdural space can often be completely aspirated, which is not possible with subarachnoid or epidural injection.

Spinal Cord Puncture

Minimal direct cord injury occurs if there is puncture of the spinal cord. Injection of any solution into the substance of the cord is a major problem. Injury is proportionate to the volume of injectate. Direct puncture of the cord provokes pain and altered sensation in most subjects. Injection of solution into the substance of the cord adds to the trauma with ischemic pressure injury at the site of the injection in the cord. Hodges reported two cases of cervical cord injury secondary to cord injection. Both patients were heavily sedated an unable to respond.⁸⁶ Severe thoracic cord injury occurred when a low thoracic epidural injection was performed under general anesthesia.⁸⁷ Some cord punctures may not provoke pain.⁸⁸ Mayall reported permanent paraplegia resulting from multiple attempts at thoracic epidural injection in an awake patient. No lateral view was used during the procedure.⁸⁹ Surgery subsequently revealed multiple punctures of the cord. These complications are technical misadventures. They are minimized by careful attention and precise technique.

Brain and Spinal Cord Infarct

Arterial filling during transforaminal injections has been rarely reported. Only two cases of cervical medullary artery filling during transforaminal injection have been reported.^90,⁹¹ In both instances medullary artery filling was recognized during real time fluoroscopy and confirmed, in one instance, with digital subtraction angiography (DSA). Both procedures were aborted without further incident. A single case of lumbar radicular/medullary artery filling during transforaminal injection has been published.⁹² Filling of a small artery in the central canal during a right S1 transforaminal injection was recognized during real time fluoroscopy, but demonstrated more dramatically with DSA.

Transforaminal injections are hazardous because of important arterial structures in the target field. Delivery of a particulate steroid into an artery may result in embolization of particles, larger than capillaries, with subsequent ischemia and/or infarct of the perfused tissue. There were nonpublished, anecdotal reports of catastrophic neurologic complications associated with epidural injections circulating during the last decade. Brouwers and colleagues reported the first such occurrence in 2001.⁹³ The central cervical spinal cord infarct occurred in conjunction with a fluoroscopically guided C6 transforaminal injection. This was likely the result of embolization of the anterior spinal artery watershed by way of a reinforcing cervical medullary artery.³⁷

The next year Houten and colleagues reported the first cases of lower thoracic and conus infarct occurring as a result of lumbar transforaminal injections.⁹⁴ Two injections were performed at the L3-4 level, one on the left, the other on the right. The third was an injection at S1. The authors implied probable injection into an aberrant artery of Adamkiewicz.

A more likely hypothesis is injection into lumbar or sacral radiculomedullary arteries. In 1971, Lazorthes reported observing a well-developed anastomotic circulation to the conus and distal thoracic cord by way of lumbar and sacral arteries. In an experimental study, colloid material was injected into the aortic circulation, below the origin of the artery of Adamkiewicz and was subsequently identified in the distal cord and conus of the experimental animals.⁴¹ He postulated that these colloid emboli reached the anterior spinal artery circulation by way of reinforcing lumbar and sacral radiculomedullary arteries.

Glaser reported paraplegia following a thoracolumbar transforaminal injection in 2005.⁹⁵ Yin and Bogduk reported retrograde flow in a radicular arterial branch and subsequent filling of the thoracic anterior spinal artery. Retrograde flow into medullary arteries or the vertebral artery without direct cannulation can occur, providing an alternative mechanism of potential injury to the spinal cord or brain during transforaminal injections.⁹⁶

Since the first report in 2001, there has been a frightening number of case reports of catastrophic neurologic complications in the literature.⁹⁷ In a mail-in survey, Scanlon collected 287 responses from physician members of American Pain Society. There were 78 major neurologic complications associated with cervical TFE. All procedures involved injection of particulate steroid preparations. There were vertebrobasilar brain infarcts and cervical spinal cord infarcts.

The survey did not explore thoracic spinal cord and conus infarcts. There are more events than case reports; many are encountered in the context of medicolegal proceedings.⁹² One of the authors (CA) has reviewed six distal thoracic/conus infarcts in the last 2 years, none of which has been reported in the literature. Kennedy and colleagues reported two cases of distal thoracic cord and conus infarcts that occurred in conjunction with L3 TFE injections. A left L3 TFE was performed with fluoroscopic guidance, using betamethasone (Celestone) as injectate. A right L3 TFE was performed with CT guidance, using methylprednisolone acetate (Depo-Medrol). Paraplegia occurred almost immediately in both cases. Follow-up MRI scans demonstrated typical distal thoracic spinal cord and conus infarcts, with increased signal intensity in the central cord from about T9 to the tip of the conus.⁹⁸ Prevention of these complications begins with awareness that this can happen.

First, recognize intraarterial injection when it occurs.⁹¹ Careful real time fluoroscopy, in the frontal plane, during slow injection of an adequate volume of contrast to opacify vessels is mandated. Visualization of vessels can be enhanced by DSA.

Second, injection of a small test dose of concentrated local anesthetic before injecting steroids, especially with particulate varieties, to test for onset of neurologic signs may allow recognition of unexpected flow in the vertebral or medullary arteries.⁹⁹

Currently the presumed mechanism of spinal cord injury is an embolic shower of particles of the steroid preparation into the circulation of the anterior spinal artery.^100,¹⁰¹ A third safety measure is to avoid particulate steroids entirely for all transforaminal epidural injections.

Dexamethasone sodium phosphate is a nonparticulate steroid. Derby and associates found the particle size in this preparation to be approximately 10 times smaller than red blood cells and they do not appear to aggregate. They have lower density than the particles and aggregates of methylprednisolone acetate, triamcinolone acetonide, and betamethasone sodium phosphate/acetate, the most commonly used agents.¹⁰⁰ In a comparison study, Dreyfuss and colleagues reported that dexamethasone was slightly less effective than triamcinolone, but the difference was not statistically significant.¹⁰² There have been no reported infarcts when dexamethasone has been used for transforaminal injections. Two animal studies have demonstrated severe neurologic injury to the animals when particulate steroid preparations were injected into the carotid¹⁰³ and vertebral¹⁰⁴ arteries. There was no evident injury to the animals receiving intraarterial injections of the nonparticulate preparation, dexamethasone. A theoretically safer nonparticulate agent appears to be a valid alternative to the particulate agents that have been used to date.

Epidural Injection Techniques

Equipment

Fluoroscopy

Fluoroscopy is mandated for all interventional spinal procedures, including injections performed for the diagnosis or treatment of pain of spinal origin. Although “blind,” nonfluoroscopically guided, interlaminar injections for peri- and postoperative analgesia are the norm in the anesthesia arena, the specialty of interventional pain or non-surgical spine, demands precise needle placement, selective administration of medications, validation of the procedure, and a historical record. The literature ¹⁰⁵^–¹⁰⁷documents that in experienced hands, successful epidural needle placement is to be expected in only 60% to 70% of cases where fluoroscopy is not used. A 30% to 40% failure rate does not provide a sufficient probability of success on which to base a diagnostic or therapeutic modality. Therefore, interventional spinal pain procedures performed using nonfluoroscopically guided (i.e., blind) techniques should be considered as highly inappropriate and possibly sham, and fraudulent procedures whether being performed for diagnostic or therapeutic objectives. Rationalizations that the cost of fluoroscopy is prohibitive, equipment unavailable, historical precedent, or the “vast experience and skill” of the practitioner, usually belies economic impairment or poor, insufficient training and cannot be used as an excuse for the substandard practice of medicine.

It is imperative that all physicians performing fluoroscopically guided spinal injections have training in the interpretation of real time fluoroscopic images whether cervical, thoracic, lumbar, or sacral. This training must be beyond the level of a residency and most current “pain” fellowships. Expertise in radiologic interpretation is far beyond the training and proficiency of nurse anesthetists (CRNAs), physician assistants (PAs), and other so-called “mid level nonphysician providers.” Performance of interventional pain procedures constitutes the practice of medicine as noted by at least one judicial body within the United States,¹⁰⁸ and numerous medical specialty societies including the American Medical Association,¹⁰⁹ the International Spine Intervention Society,¹¹⁰ and the American Society of Anesthesiologists.¹¹¹

The fluoroscopic instrument of choice is the C-arm, which can be moved in any plane to provide optimum imaging to direct needle insertion and to verify, and document, final needle position and contrast spread. Radiation-saving modalities such as low dose, pulsed mode, collimation, and saving of last image are essential and ensure safe practice. Digital subtraction imaging (DSI), although not at present standard of care, can provide significant useful information when used correctly.

The majority of interventional pain procedures are performed where the C-arm fluoroscope is used to align the skin entry point with the target. A needle can then be advanced parallel to the beam. Using this down-the-beam technique, the correct angle of needle insertion is unmistakable, and when the injectionist is familiar with the anatomy lying between the skin and target, it offers the safest approach. For precise work, parallax demands that the target is placed in the center of the image.

To use the C-arm, an x-ray-compatible procedure table is required. Although some operating room tables will suffice, most have metal incorporated into their structure, which limits certain fluoroscopic projections. Carbon fiber tables, with the pedestal at the foot, are now quite affordable and provide unlimited access and unrestricted imaging possibilities.

Radiation safety precautions, including the use of lead aprons, thyroid shields, and radiation exposure monitors must be provided for all personnel in the procedure room. Some have advocated the use of lead glasses.

One often overlooked aspect of radiation safety is the radiologic technician, RT, actually controlling the C-arm. A competent RT will not only assist in obtaining the correct images for safe needle placement, but should also control collimation and the power output of the fluoroscopy equipment to ensure that the physician performing the procedure is acquiring the least radiation while maintaining adequate imaging quality throughout the procedure. However, obtaining the correct image required for an injection procedure is the responsibility of the physician alone, and the essential “fine tuning” to provide this image is best handled personally.

Regarding radiographic guidance, although the use of CT guidance has been advocated by a small minority of radiologists,¹¹² most radiologists within the interventional pain community reject the use of this imaging modality for the placement and validation of safe needle position. Using CT allows one to accurately delineate the skin entry point; however, the advancement of the needle toward the target is accomplished crudely, without imaging assistance. This adds to the physical discomfort of the patient, the time required to complete the procedure, and the chance for needle misadventures. CT necessitates the injection of contrast for verification of needle placement without the benefit of active, real time imaging. Vascular injections will, in all likelihood, escape detection owing to the rapid “washout” of contrast secondary to blood flow within the vein or artery. In addition, the marked increase in total radiation exposure to the patient, as inherent in CT, should be of concern. No literature exists in which evidence indicates any benefit to the use of CT for routine spinal injections. Specialty group guidelines published by the International Spine Intervention Society (ISIS)¹¹³ and the Physiatric Association of Spine, Sports, and Occupational Rehabilitation (PASSOR),¹¹⁴ specifically do not mention CT use for epidural injections, whether transforaminal or interlaminar. As noted previously, severe permanent neurologic sequelae following the use of CT for cervical and lumbar transforaminal epidural injections have been reported in the literature and medical legal documents. These cases involved paraplegia, quadriplegia, and death secondary to cord infarct.

A historical record of all procedures must be archived either with hard or digital images. Multiple images must clearly identify that the needle is in safe position prior to injection of contrast, and that the contrast pattern represents a safe flow of the injectate.