Transforaminal epidural steroid injections and selective nerve root blocks provide an alternative intervention to interlaminar epidural injections to target radicular pain. The two former procedures differ regarding the final needle position, with transforaminal injections positioned in the intervertebral foramen, and selective nerve root blocks positioned adjacent to the spinal nerve root. Both interventions share many similarities regarding procedural technique, relevant anatomic structures, injectate composition, and safety considerations. Selective nerve root blocks require low injectate volumes (0.5–0.6 mL) to reduce inadvertent spread to other roots and appear to improve surgical outcomes as well as provide value distinguishing among complex pain syndromes. Transforaminal injections use higher volumes (1–4 mL) and may demonstrate better pain outcomes but greater incidence and risks of complications than interlaminar injections. A heterogeneous group of studies comparing transforaminal and interlaminar techniques precludes a simplified summary. While rare, cases of severe neurologic injury from transforaminal injections resulted in greater awareness of critical vascular structures along the needle route and the need to perfect technique when approaching nerve roots and the spinal cord. More recently, awareness of such risks incited the creation of multispecialty safety guidelines for transforaminal injections, which include the use of fluoroscopy and nonparticulate steroid for the first injection.
Keywordscervical, thoracic, lumbar, and sacral nerve injection, epidural steroid injection, selective nerve root block, selective nerve root injection, spinal or spine blocks, spinal or spine injections, transforaminal injection
Transforaminal epidural steroid injections (TFESI) and selective nerve blocks provide an alternative interventional approach to address radicular pain, in comparison with interlaminar epidural steroid injections (ILESI; as reviewed in Chapter 62 ). The rationale for using the former routes of injection compared with the latter is that the delivery of medications occurs directly onto the nerve root(s) using the transforaminal approach. This permits maximal concentration of medication delivered to the site of suspected pathology. The distinction between TFESI and selective nerve root blocks may involve some overlap given anatomical considerations. Upon exiting the epidural space, spinal nerves transition in their covering from the dura mater to the fascial sheath in a continuous fashion. Consequently, medication injected adjacent to the spinal nerve may enter the epidural space irrespective of whether the needle tip progresses through the intervertebral foramen or not. Both transforaminal injections and selective nerve root blocks describe injections of medication adjacent to spinal nerves, but differ in the final location of the needle tip. The term selective nerve root block typically describes an injection adjacent to the spinal nerve root, with the needle tip remaining outside of the intervertebral foramen. In contrast, the term transforaminal injection describes an injection with the needle tip that resides within the intervertebral foramen.
The role of epidural steroid injections (ESI) in the treatment of lumbosacral radicular pain and other disorders has generated significant discussion and debate since its first description. Historically, the first publication describing ESI reported on the transforaminal or selective nerve root approach in 1952. Robechhi and Capra published a case study on the periradicular injection of hydrocortisone to the first sacral nerve root via the S1 sacral foramen for the treatment of lumbosacral radicular pain, and additional reports detailing treatment outcomes for radicular pain using the transforaminal approach soon followed. Alternate approaches then supplanted transforaminal injections in the following decades, including the use of intrathecal injections of steroids in the 1960s, which was later abandoned due to complications such as arachnoiditis. Use of the interlaminar technique was more frequent in practice, likely as a result of the procedure being performed by anesthesiologists who were most accustomed to the “blind” loss-of-resistance technique for interlaminar injections. The transforaminal technique received renewed attention for the management of radicular pain in 1992, when Derby et al. described fluoroscopically guided selective nerve root blocks as a predictive tool before lumbar disc surgery. The following decade bore witness to a significant rise in the use of interventional pain procedures, including TFESI, which were associated with a concomitant rise in rare, catastrophic neurological complications. A renewed outlook on transforaminal and selective nerve root injections incorporates safeguards to prevent such complications.
Among the different approaches for ESI, the transforaminal injection shares many commonalties with the interlaminar approach, including similar use of medications composing the injectate, a shared mechanism of action (see Chapter 62 , Mechanism(s) of Action), overlapping indications for treatment, and improved accuracy with use of fluoroscopic guidance. In contrast to the interlaminar and caudal approaches, transforaminal injections require the navigation of more complex anatomy to place a needle within the neural foramen, mandate the use of fluoroscopic guidance including appropriate image interpretation, and involve greater risks. Yet transforaminal injections increase the possibility of ventral epidural spread of the injectate, carry a lower risk of inadvertent dural puncture, and appear more likely to provide positive patient outcomes relative to interlaminar (or caudal) injections. Consequently, every patient deserves a close examination of the clinical need for the specific type of epidural injection and an informed discussion of the potential benefits and risks of the planned approach.
Selective Nerve Root Block
The techniques of TFESI and selective nerve root blocks share similar anatomical considerations, given that the approach to the spinal nerve root is the same. The term selective nerve root block fails to acknowledge the continuous nature of the spinal nerve root sheath with the dura mater, which typically permits injectate volumes as low as 0.5 mL to spread to other nerve roots. A technically perfect selective nerve root block with the needle verified to be outside of the foramen as viewed fluoroscopically may nevertheless result in epidural spread of the injectate. Despite this shared space, a selective nerve root block more selectively targets the spinal nerve root than the interlaminar approach, which permits these injections to be used diagnostically to isolate symptoms, and prognostically to predict surgical outcomes, when multilevel vertebral pathology is present.
Evidence for the diagnostic and prognostic characteristics of selective nerve root injections to improve surgical outcomes is weak but positive. The ability of different nerve blocks to localize an anatomical source of radicular pain was examined by North et al. in a randomized controlled trial of 33 patients with lumbosacral radicular pain, which revealed that most patients noted significant pain relief with selective nerve root, sciatic nerve, and facet blocks. Injectate volumes delivered in the study consisted of a relatively larger volume of local anesthetic (3 mL), which likely resulted in blockade of other nerve roots besides those targeted. As a result, positive responses to diagnostic selective nerve root blocks were unable to identify those with true nerve root involvement beyond other types of nerve blocks, and positive responses to nerve blocks lacked specificity. On the other hand, negative responses to diagnostic blocks appeared clinically informative.
The lack of specificity of selective nerve root blocks may be partially attributable to the inadvertent spread of small injectate volumes through the epidural and nerve root space, as established in a study by Castro et al. that tracked the spread of injectate to nearby nerve roots with various injectate volumes during L4 selective nerve root blocks. Using the smallest 0.5-mL volume to block L4 nerve roots, epidural and L3/L5 nerve root spread was fluoroscopically verified in 48% and 27% of injections, respectively. Additionally, spread of local anesthetic to the psoas muscle, situated in close proximity to the lumbosacral plexus, occurred in 12% of 0.5-mL injections and 71% of 2-mL injections. The unrestrained flow of local anesthetic injectate in selective nerve root blocks may partially explain the lack of specificity.
As past studies suggest, higher volumes of injectate contribute to worse specificity of selective nerve root blocks. In a quantitative evaluation of lumbosacral contrast flow patterns, Furman et al. injected 0.5-mL increments of contrast and examined fluoroscopic images for evidence of contrast spread in areas other than the single, ipsilateral nerve root. Using 0.5 mL of contrast, 30% of injections were no longer specific to the targeted nerve root. Specificity was further reduced with higher volumes, as injections of 1 mL and 1.5 mL of contrast were no longer specific in 67% and 87% of injections, respectively. A study by Anderberg et al. that compared injection volumes of 0.6, 1.1, and 1.7 mL in the cervical region illustrated a similar lack of specificity in the distribution patterns. The authors found that distribution patterns using 0.6 mL of contrast demonstrated sufficient nerve root specificity, though epidural spread was still documented, even at this low volume. At higher volumes, spread was noted in the epidural space or adjacent nerve roots in more than 50% of cases. The authors concluded that the diagnostic selectivity of a selective nerve root block is unreliable with volumes in excess of 0.5–0.6 mL.
Other factors that can influence the interpretation of selective nerve root blocks besides volume of injection include patient characteristics, such as preexisting sensory changes. Using sensory examinations before and after injections, Wolff et al. documented significant variability in pain reduction experienced by patients, even when dermatomal changes in sensation were obtained. Low volume selective nerve root blocks elicit paresthesias and more consistently result in hypesthesias (i.e., diminished sensation) of the affected dermatomes than actual pain reduction. Consequently, the validity of selective nerve root blocks may be reduced in patients with long-standing sensory changes and sensory changes that occur in nondermatomal patterns. When preexisting sensory changes are present, eliciting pain and paresthesias in the target dermatome(s), and producing hypesthesias on examination, can best be achieved when overlapping dermatomes of the target nerve root(s) are taken into consideration.
Reports on the accuracy of selective nerve root blocks vary based on spinal region. Yeom et al. evaluated the accuracy of selective nerve root blocks in 47 patients with single-level lumbosacral radicular pain using a series of 1 mL local anesthetic injections to block the symptomatic level, as well as an adjacent level as a control. A threshold of 70% pain reduction demonstrated the best discriminative ability, with the overall accuracy being 73%. False-positive findings resulted from spread of the injectate to the nearby epidural or nerve root space, with the authors reporting a sensitivity of 57%. The specificity of the injections was 86%, with the injectate failing to spread to the nerve root, insufficient infiltration, and injection within the epineural sheath resulting in the majority of false-negative results. In the cervical region, selective nerve root blocks using 1 mL correctly identified 19 of 20 patients with single-level pathology on magnetic resonance imaging (MRI). Among the 18 patients who underwent surgery, all noted resolution of pain. With more diffuse, two-level pathology, selective nerve root blocks correlated with the level of worst degeneration identified via MRI in 60% of cases, with neurological examination in 27% of cases, and with dermatomal mapping in 23% of cases. Agreement among selective nerve root block, MRI imaging, and neurological examination was rare, and no randomized trials exclusively compared selective nerve root block to MRI imaging. Overall, the accuracy of selective nerve root blocks appears to be moderate.
Selective nerve root blocks may not always contribute as a cost-effective method to the standard diagnostic workup prior to lumbar decompression surgery, based on economic modeling that placed the incremental cost per quality-adjusted life-year in excess of £1.5 million. The overall conclusions, including finding low-specificity for the blocks, relied on few studies with varying methods and overall poor quality. In addition to the relative costs of blocks and surgeries, and how well patients are selected (i.e., poor surgical selection is likely to make prognostic blocks more cost-effective), technical factors and patient characteristics such as discordant clinical symptoms and imaging all influence the cost-effectiveness of selective nerve root blocks.
In summary, the question of whether selective nerve root blocks improve surgical outcomes is answered with weak, positive evidence suggesting they do. Selective nerve root blocks remain appropriate in a number of clinical situations. Discrepancies among the patient’s clinical presentation and/or various diagnostic examinations (i.e., electromyography, MRI) raise the utility derived from a selective nerve root block. Patients with unusual extremity pains, anatomical variations such as transitional vertebrae or conjoint nerve roots, or with previous back surgery and atypical extremity pain, may also benefit from information provided from selective nerve root blocks.
Transforaminal Epidural Injections
The approach to transforaminal injections varies based on region of the spine, with regional variations in vertebral, vascular, and neural elements influencing the desired placement of the needle and final location of the needle tip. Multiplanar fluoroscopic guidance with use of contrast dye to verify the trajectory and confirm the location of the needle tip is necessary prior to injecting particulate steroid. This helps minimize the risk of involving nearby critical elements not directly visualized via fluoroscopy, such as the arterial supply to the spinal nerve root and/or the spinal cord or the spinal nerve root itself. Digital subtraction angiography is the reference standard for detecting cases of inadvertent vascular access prior to injection, with one study showing that real-time fluoroscopy has a sensitivity of 71%.
The cervical transforaminal injection approach differs from that of the thoracic and lumbar because the angle of the vertebral foramen and exiting spinal nerve is anterior in the cervical region, and posterior in the thoracic and lumbar region. Consequently, patient positioning is supine or lateral, with the former position modifying the fluoroscopic image to a postero-anterior (PA) view instead of an antero-posterior (AP) view. The lateral position facilitates injection at C1–C4 levels, while the supine position facilitates views without shoulder tissue at C4–C8 levels. After identification of the appropriate vertebral level, rotation of the C-arm axis obliquely 45–65 degrees optimizes the cervical vertebral foramen to permit clear visualization of surrounding structures. From this perspective, various bony borders outline the cervical vertebral foramen: The superior vertebral body resides anteriorly, the pedicle from the same level resides superiorly, the facet column formed by the inferior and superior articular processes resides posteriorly, and the pedicle from the inferior vertebral body resides inferiorly.
Critical vascular structures include the vertebral artery, which courses within the periodic protection of the transverse foramen but remains mostly unprotected along the anterior aspect of the cervical vertebral foramen ( Fig. 63.1 ). The blood supply of the spinal cord consists of one anterior and two posterior spinal arteries. Spinal radicular arteries, which pass out of the foramen and continue as radicular arteries adjacent to the spinal nerve roots, demonstrate a variable anatomical location. Consequently, the most inferior and posterior portion of the foramen is desired as the initial target to minimize the risk of entering or injuring the vertebral artery anteriorly. Introducing the needle via a coaxial fluoroscopic technique followed by contact with the superior articular process posteriorly provides for an additional margin of safety. Upon contact with the posterior portion of the foramen, the needle may be advanced anteriorly into the foramen. The importance of obtaining a lateral image to ascertain the needle depth and avoid inadvertent penetration of the spinal cord should be highlighted. Injection of contrast medium via conventional fluoroscopic imaging in the PA view should then demonstrate an outline of the desired nerve root. If appropriate spread outlining the nerve root or epidural distribution is not noted, slight advancement may be made followed by reassessment with contrast medium. Typically advancement of the needle occurs up to the “redline” of the facet joint, which anatomically serves as an approximate safety limit, depending on the degree of oblique angulation of the image intensifier (i.e., how far the facet column is situated within the vertebral body). Once appropriate needle positioning is confirmed, slow injection of medications is performed ( Fig. 63.2 ).
The lumbar and thoracic transforaminal injection approaches start with the patient prone, which permits an appropriate fluoroscopic view of the vertebral foramina that project in a posterior direction. Although the lumbar and lower thoracic approaches may be performed in a lateral position, most practitioners prefer prone. After identification of the appropriate vertebral level, rotation of the C-arm axis obliquely optimizes the vertebral foramen to permit clear visualization of surrounding structures. In the lumbar region, the degree of angulation of the image intensifier affects the final needle position, and consequently the tendency for epidural versus nerve root spread, when a coaxial approach is used. For example, steeper angulation (>35 degrees) results in a more medial needle position and a greater proportion of epidural uptake, whereas less acute angles are used to obtain more nerve root spread. Caution to avoid overrotation beyond 20 degrees in the T1–T8 thoracic regions is needed, given the increased possibility of the needle trajectory causing a pneumothorax ( Fig. 63.3 ). An optimized oblique image arranges the various bony borders and relevant landmarks of the vertebral foramen in a silhouette that mimics a Scottish terrier in the “scotty dog” position: The superior articular process outlines the dog’s ear, the transverse process projects over the vertebral body as the nose, the inferior articular process serves as the dog’s front leg, the pedicle overlaps the region of the dog’s eye, and the spinous process mimics the dog’s feet.
Consensus regarding optimal needle positioning, which permits the delivery of medications without perturbation of the adjacent spinal nerve or injection into nearby vascular structures, continues to be deliberated. Traditionally, the “safe triangle” or subpedicular approach recommended needle placement within the fluoroscopic region posterior to the vertebral body, defined by the lateral aspect of the inferior pedicle (i.e., below the pedicle) and the superior border of the imagined nerve root that courses from the vertebral foramen inferiorly and laterally. Critics of the subpedicular approach note that the primary advantage of the technique, avoiding needle contact with the spinal nerve, fails to incorporate variability in nearby vascular anatomy (i.e., radicular arteries and the artery of Adamkiewicz), which increases the risk of inadvertent intravascular injection. An alternative approach targets Kambin’s triangle, a region described in the context of accessing intervertebral discs for surgery in 1972 by Kambin and Sampson. Kambin’s triangle describes an area overlying the posterolateral disc that is bounded by the inferior vertebral body at its base, the exiting spinal nerve root at the hypotenuse, and the traversing nerve root or dura at the vertical leg ( Fig. 63.4 ). Kambin’s triangle may offer advantages in comparison to the subpedicular approach regarding inadvertent vascular injection, spinal nerve contact, and intradiscal injection, but no one approach for transforaminal injections fully eliminates risk.
Critical vascular structures in the thoracic and lumbar regions include similar spinal arterial branches, which were described in the cervical region ( Fig. 63.5 ). However, the largest spinal arterial branch, the artery of Adamkiewicz, typically enters through the left intervertebral foramen in the thoracic or upper lumbar region but may be present on either side from T5 to S1. Besides vascular structures, the ribs, pleura, and mediastinum may be at risk of penetration during the thoracic approach. Using a similar coaxial fluoroscopic technique as described previously, the ideal needle position for the subpedicular approach is slightly inferior to the pedicle and lateral to the pars interarticularis, above the superior articular process inferiorly. The needle is advanced in the superior and posterior aspect of the vertebral foramen to avoid the spinal nerve. An AP fluoroscopic view provides context for the mediolateral location of the needle. Alternatively, the inferolateral aspect of the pars interarticularis may serve as a depth marker from which the needle may be walked off. Unexpected bony resistance along the path is likely from contact with the pars interarticularis, and redirection inferiorly, anteriorly, and medially should permit further advancement. Once the needle tip is located inferior to the medial aspect of the pedicle, a lateral fluoroscopic view permits visualization of the needle tip being advanced into the foramen. The ideal needle tip position resides in the anterior one-third of the foramen, slightly inferior to the pedicle. An ideal trajectory to Kambin’s triangle may target the area over the intervertebral disc, slightly lateral to the superior articular process. Contact with the lateral lower aspect of the superior articular process may serve as a depth marker, after which the needle may be directed laterally a few millimeters. Ideal final positioning demonstrates the needle located medially within the upper pedicle using an AP view and at the posteroinferior aspect of the vertebral foramen using a lateral view ( Fig. 63.6 ).
Injections of the L5 nerve root may require modifications to this technique in the presence of an overriding iliac crest that obstructs an oblique view of the vertebral foramen. Aligning the inferior vertebral end plate of L5 by positioning the C-arm in a more cephalad direction adjusts for this anatomical issue. This position creates a triangular area formed by the iliac crest, the superior articular process of S1, and the inferior border of the L5 transverse process.
For S1 nerve root injections, prone patient positioning permits access to the posterior sacral foramen. Slight C-arm axis rotation 5–15 degrees in an ipsilateral, oblique direction may optimize the relevant anatomy and reduce the incidence of intravascular injection. A randomized study comparing transforaminal injections at S1 randomized using two different angles of approach demonstrated intravascular uptake in 29% of injections using the AP view, in contrast to 11% with an oblique view. A small caudocranial rotation (toward the patient’s head) may result in an improved view of the S1 foramen given that the sacral foramina project in a slight cephalad direction. Needle advancement toward the sacral foramen may then proceed. Obtaining a lateral view is necessary to confirm the needle tip remains at or above the level of the caudal epidural space and avoids trespassing into the pelvic region.
During the transforaminal approach, the patient may experience paresthesias. Although pain may result from perturbation of the nerve root, other structures such as the facet joint, periosteum, and annulus fibrosus may cause referred pain to the leg. Regardless of the etiology, onset of paresthesias indicates an appropriate time to withdraw the needle slightly. Paresthesias must terminate prior to injection of contrast medium. Additionally, verifying needle tip placement in two views is critical during a transforaminal injection at any region of the spine. Inadvertent penetration of structures such as the spinal cord or vasculature cannot be excluded using only one view. After negative aspiration of blood and cerebrospinal fluid, contrast medium should be injected, which may confirm anterior epidural spread. A pattern of spread in the ventral epidural contrast flow is desirable, given the pattern’s association with increased pain relief. Careful examination of the images should ensure the absence of intravascular uptake. Vascular injection is reported to occur in up to 12%–14% of transforaminal injections, and may occur in all spinal regions but is most likely to happen in the cervical region. Inadvertent intravascular injection may be missed without the use of digital subtraction angiography in up to one-third of all cases. Because it significantly enhances the ability to detect vascular uptake of contrast medium, digital subtraction angiography was recently endorsed by a multispecialty workgroup, though it is not mandated before injection of steroid. To optimize the image quality of digital subtraction angiography, the patient may be requested to hold their breath. After intrathecal and intravascular injections have been excluded, the medication may be injected.