Neuraxial Injections

In review of the history of spinal injections for pain management, it is clear that the procedures have been developed and expanded from their original use in anesthesia (Fig. 1-1). After the publication of evidence in 1884 that cocaine could be used to render the cornea insensate for ophthalmologic procedures,¹ interest in the ability to anesthetize only the region to be operated upon grew. In 1885, the neurologist James Leonard Corning first described spinal anesthesia,² which was interestingly a result of a misunderstanding of the anatomy and physiology of the spine and its contents. His intention was to inject cocaine into the interspinal blood vessels, so that it could be delivered to the spine via the communicating vessels in the spinal cord. No such vessels exist, although the correct anatomy had been described in Gray’s Anatomy by 1870.³

Fig. 1-1 Timeline outlining history of spine injections (1884-2010). EBP, epidural blood patch; FDA, Food and Drug Administration; NEJM, New England Journal of Medicine; RF, radiofrequency; SI, sacroiliac; TB, tuberculosis.

Corning used a hypodermic needle with the goal of injecting cocaine into the interspinous vessels. He wrote: “I hoped to produce artificially a temporary condition of things analogous in its physiological consequences to the effect observed in transverse myelitis or after total section of the cord.”²

In his report, he describes injecting 120 mg of cocaine into a male human subject and 13 mg of cocaine into a dog. By description of onset and effects, it appears that the injections probably resulted in epidural anesthesia in the man and spinal anesthesia in the dog. The doses used far exceeded the potential toxic doses, but fortunately, there were no significant complications. Corning had been searching for treatment for neurological diseases but noted that the procedure certainly could have surgical implications.⁴

Documentation of intentional dural puncture was introduced by Dr. Essex Wynter in 1891. Using a Southey’s tube and trocar, he placed the tube between the lumbar vertebrae after making an incision in the skin for the purpose of draining the fluid in tuberculous meningitis. He noted temporary relief and no complications with the procedure, although none of the patients survived the tuberculous meningitis.⁵

Six months later, Heinrich Irenaeus Quincke wrote “Die Lumbalpunction des Hydrocephalus.”⁶ He based his approach on the knowledge of the lumbar anatomy of the continuous subarachnoid space and the end of the spinal cord at approximately L2, which allows for the introduction of a needle below that point, avoiding spinal cord injury. The procedure was introduced for the treatment of hydrocephalus. Quincke improved the technique by the use of needles that were 0.5 to 1.2 mm in diameter, including a stylet in the larger needles. The initial description is a paramedian approach, starting 5 to 10 mm from midline.

The application of the procedure for spinal anesthesia rather than a therapeutic option was developed by a surgeon, Dr. August Bier. He published his findings in 1899 with the title “Versuche uber Cocainisirung des Ruckenmarkes” (“Research on Cocainization of the Spinal Cord”).⁷ His goal was to use minimal amounts of medication to anesthetize a large region. News and promise of the technique spread quickly, and by October of 1899 Drs. Dudley Tait and Guido Caglieri had tried the approach in San Francisco, becoming the first to do so in the United States.⁸ By January 1901, a report in The Lancet stated that there were already almost 1000 published reports of spinal anesthesia.⁹

In 1901, Fernand Cathelin demonstrated the ability to gain access to the epidural space via the caudal approach. He noted that fluids rose in a fashion that was proportional to the volume and speed of the injection.¹⁰ Both Cathelin and Dr. John Sicard presented a paper on epidural injections the same year (1901), but the two physicians were working independently.

De Pasquier and Leri attempted intrathecal injections of 5 mg of cocaine at the lumbar level but noted in their results “toxic cocaine accidents . . . to the bulbar and cerebral centers.” Using a rubber band “gently tightened around the neck,” they tried to prevent the flow of cocaine to the brain but were unsuccessful. They claimed a better level of success with sacral epidural injections.¹¹

W. Stoeckel¹² published his experience in obstetrical care with the caudal epidural method in 1909 after modifying the method by using the less toxic procaine rather than cocaine. He was interested in the possible spread of medication in the epidural space from a caudal injection and used colored fluid in cadavers to document the extensive spread, including through the sacral foramina.

In 1925, Dr. Norman Viner¹³ published his experience in treating intractable sciatica with caudal epidural injections. He described his technique of injecting first 20 cc of 1% Novocain followed by 50 to 100 cc of sterile Ringer’s solution, normal saline, or liquid petrolatum. These injections were typically repeated three to four times at weekly intervals. He notes that “liquid petrolatum is frowned on by some on account of the remote possibility of fatty embolism” but goes on to note the overall low risk. He concludes his paper by suggesting the procedure be tried with many other conditions because he believed it could be very successful in the treatment of sciatica.

Cortisone (called Compound E) was discovered in 1936.^14,15 Hench et al^16,17 reported in a 1950 publication that it could treat rheumatoid arthritis, rheumatic fever, and other conditions as well. A longer acting steroid, Compound F (hydrocortisone) was noted by Hollander¹⁸ to reduce the synovial membrane inflammation histologically but even then the author was cautious to state that the action of the steroid was palliative, not curative. The use of steroids to treat many conditions became common during this period.

Claiming that patients’ sciatic pain was a result of inflammation, Robecchi and Capra¹⁹ reported using “periradicular” hydrocortisone to treat lumbar disc herniation in 1952. Lievre et al²⁰ described caudal epidural injections as being effective when five of 20 patients improved. No data were reported more than 3 weeks after the injections, and no control subjects were used, so placebo and the natural history of the problem were not addressed. The popularity of caudal epidural injections appears to have increased after this report.

“Pressure caudal anesthesia” was advocated by Brown,²¹ who used 50 to 70 cc of mixtures of lidocaine, normal saline, and steroid. He noted improved success (100% vs. 32 of 38) when a steroid was added to the normal saline and local anesthetic. Again, however, the lack of control subjects and structured follow-up is notable.

The first clinical description of the technique for a paramidline lumbar approach is credited to Pagés in 1921.²² The procedure then modified to include the loss of resistance technique, introduced by Dogliotti in 1933.²³ Gutierrez suggested the hanging drop technique by using the negative pressure of the epidural space in the same year.²⁴ Dogliotti was also the first to describe an epidural injection into the cervical region.²³

During the middle of the twentieth century, investigators experimented with treatments using both intrathecal and epidural injections. In the 1950s, there was interest in treating patients with multiple sclerosis with intrathecal steroids, but no control subjects or follow-up were included in these trials.^25–27 In later reports,^28,29 excitement about the procedure waned as persistent improvement was seen in only a limited portion of the patients.

In the early 1960s, Gardner et al³⁰ tried high-volume epidural injections (20 cc of 1% procaine and 125 mg hydrocortisone) in 239 patients with sciatica. About half of the subjects had failed to obtain relief with surgery. After 57% of the patients failed to get pain relief with the epidural injections, the investigators started using an intrathecal approach with 80 mg of methylprednisolone acetate and 40 mg of procaine. Sixty percent of the 75 subjects noted relief of the sciatic pain for more that 4 months. By 1963, Sehgal and Gardner³¹ and Sehgal et al³² had treated more than 1000 patients with intrathecal steroids for diverse conditions, but no improvement data or control group was reported.

The transition back to epidural injections began in response to data published by Winnie et al in 1972.³³ At the time, there continued to be controversy as to the aspect of the procedure that produced pain relief. The theories proposed were therapeutic benefit from the injections resulted from lysis of adhesions by large volumes of injectate, interruption of the sympathetic reflex mechanisms by the local anesthetic, or the antiinflammatory effect of the steroid. By demonstrating success of epidurals with low-volume injectate, Winnie et al³³ proposed that the effect seemed to be from the steroid itself. They further suggested that the success of an injection seemed to be related to the proximity of the injection to the pathology causing the patients’ complaint.

The more recent modifications of epidural injections have occurred as a result of the concern regarding accurate delivery of the medication to the site of pathology. Multiple studies^34,35 show that the loss of resistance technique in a lumbar epidural steroid injection results in inaccurate needle placement up to 30% to 40% of the time. The use of fluoroscopy has been encouraged by some to improve accuracy in epidural injections for chronic pain in recent years.^36,37 Fredman and Nun³⁸ reported a lower incidence of inaccurate placement into the epidural space during “blind” epidural injections (8.3% failure rate) than previous reports but noted that the intended level of the injection was missed in 53% of the cases. Interestingly, in the cases in which the needle placement was correct, the contrast reached the level of the pathology in only 26% of the patients, largely because of altered anatomy. This study was done on patients with failed back surgery syndrome and highlights the potential difficulty of injections in this population.

The other major modification of the procedure is the transforaminal approach to the epidural space. This technique mandates the use of fluoroscopy. This technique was developed with the recognition that in caudal and translaminar approaches, the medication is delivered into the dorsal aspect of the spinal canal. The dorsal median epidural septum can stop the spread of the medication to the contralateral side.³⁹ The translaminar technique delivers the medication to the ventral aspect of the nerve root sleeve and to the dorsal aspect of the disc herniation.^40,41 Although the transforaminal technique is commonly used, one complication that has been particularly concerning is inadvertent arterial injection of particulate steroid, which has resulted in devastating consequences. To decrease the risk, nonparticulate steroid and digital subtraction imaging can be used. Some practitioners have abandoned the procedure altogether because of this risk, particularly in the cervical region.

The epidural steroid injection is the most common spinal procedure performed in pain management today, but the effectiveness is unclear. Although many studies suggest pain relief in the short term, long-term effectiveness has been disappointing. The ability of epidural injections to decrease the rate of subsequent spinal surgery has also been questionable.

Epidural Blood Patch

It is interesting to note that the development of a dural puncture headache was described very early in the development of spinal procedures. Quincke,⁶ while noting some improvement of patients with hydrocephalus after lumbar puncture, also reported that some patients complained of a pattern of pain for several days that would seem consistent with a postdural puncture headache (PDPH). Multiple punctures with a large-bore needle had been used. The observation of edema in the surrounding tissues seems to be evidence of continued cerebrospinal fluid (CSF) leaks.

It was not, however, until 1898 that August Bier⁴² clearly made the association between dural punctures and subsequent headaches that appeared to have unique characteristics. He reported that three of his first six patients in whom he performed the procedure complained of a headache shortly after the procedure. As an experiment, Dr. Bier and his clinical assistant went on to perform spinal anesthesia on themselves and then developed classical symptoms of PDPH. They documented their personal experience in what makes both interesting and somewhat comical reading today.

By January 1901, in the almost 1000 published reports of spinal anesthesia, physicians continued to note concern about the common occurrence of PDPH. Investigation into the etiology and treatment of PDPH quickly followed the early reports.

Treatment of PDPH historically can be viewed as using one of several different approaches. One approach focused on replacing the lost CSF volume to restore the intracranial pressure. Infusions of normal saline into the intrathecal space were attempted, which tended to provide temporary relief but also produced a second dural puncture. The intracranial hypotension would return with the painful symptoms shortly after the infusion was stopped with redistribution of the fluid and pressure.^43–47 Such efforts were abandoned in the 1950s. Attempts to increase CSF production by using hypotonic intravenous saline infusions and intramuscular pituitary extract resulted in perhaps some relief for a portion of patients but again did not produce consistent or dramatic results.^48,49

In attempt to produce a “splint” type of effect, the second of the approaches, epidural infusions were used.⁵⁰ This technique avoided the second dural hole in theory but failed to produce long-term pain relief because again the pain would return with redistribution of the fluid shortly after the infusion was stopped.⁵¹

The third approach is the one that we are still working with today. The principle is to plug to hole in the dura that is allowing for escape of the CSF. Dr. James Gormley was a general surgeon in the truest sense in a time (1950s and 1960s) when surgeons were also directly involved in the anesthetic care of the patient. Spinal anesthesia was attractive because the surgeon could perform the block and allow the patient to maintain his or her own airway while performing surgery and supervising the nurse for management of vital signs. One of his observations was that bloody taps seemed to result in a lower incidence of PDPH. Also important was the idea that blood in the central neuraxis did not appear to result in disaster as previously believed.⁴ He published a report of seven cases in which 2 to 3 mL of autologous blood was injected into the epidural space for the treatment of PDPH in 1960.⁵² He was actually one of these subjects who presented with a PDPH after a myelogram. Although later studies have refuted the notion that bloody taps decrease the incidence of PDPH, it was a fortunate mistaken idea.

In 1960, Dr. Anthony DiGiovanni, having just read Gormley’s letter in Anesthesiology, was asked to help in the care of a woman on the obstetric ward who had a severe headache after a spinal anesthetic. Because the anesthesiologist who did the initial procedure had attempted the injection at multiple levels and could not remember the level of the successful block, Dr. DiGiovanni decided to use a volume of 10 mL of autologous blood, thinking that the higher volume could possibly cover several levels. This resulted in the resolution of the patient’s headache, and Dr. DiGiovanni continued to treat patients presenting with PDPH with this volume as a result of this initial success.⁴ In subsequent years, he trained many other anesthesiologists in this technique and published his experience with the procedure in 1970.⁵³

This procedure, quite understandably, was met with resistance by many in the field, particularly because of concerns about safety. Animal studies as well as prospective data accumulated with time and suggested that the technique was not only effective but also very safe.⁵⁴ This led to the general acceptance of the blood patch in the treatment of PDPH.

Progress has certainly been made in reducing the occurrence of PDPH with the use of smaller gauge needles and “pencil point” tips (versus the prior use of “cutting” needles). Some physicians have tried to prevent PDPH with prophylactic blood patches at the time of the dural puncture. The evidence does not, however, support this practice.

Given the long history and the well-accepted practice of performing blood patches for PDPH, it is interesting to note the relative lack of evidence from randomized, controlled clinical trials. Van Kooten et al⁵⁵ published such a study in 2007 that strongly supports the current practice and is interesting to review.

Differential Spinal Blockade

Clinically, the etiology of a patient’s pain is sometimes difficult to define. This has remained true despite recent advances in medicine. In the 1920s, Gasser and Erlanger^56,57 published some groundbreaking work in the area of neural conduction. Although incorrect about the site of conduction (mistaking it to be within the axoplasm), they established the idea that fiber size was related to conduction velocity and fiber function. They were able to define three classes (A, B, and C) of nerve fibers and subdivided class A fibers into 4 groups (α, β, γ, and δ). Working with cocaine, they were able to demonstrate that the fibers types appeared to have different sensitivities to local anesthetic.

This understanding was the basis for the differential spinal block developed by Sarnoff and Arrowood.^58,59 Noting prior animal experimentation suggesting that a low concentration of procaine could selectively abolish the carotid sinus reflex without affecting respiration or motor function, they proceeded to test this principle in patients with varying diagnoses (residual limb pain, herpes zoster, sciatic nerve pain, and inguinal hernia repair pain). The 1948 publication is focused on patients with stump pain or phantom limb pain, with the goal of the study to decipher if the pain was of a local origin or whether it was related to a projection from the sensory cortex. If it were found to be of local origin, the investigators wanted to know if interruption of the sympathetic nervous system would result in pain relief. An initial bolus of 0.2% procaine was injected into the subarachnoid space. This was followed by an infusion of the same concentration, and observations were made regarding pain relief and neurological examination results. The results of the procedure were intended to aid in surgical planning. Table 1-1 shows their results, demonstrating their ability to block some nerve fibers and spare others.

Table 1-1 Fibers That Are and Those That Are Not Blocked by the Introduction of 0.2% Procaine Hydrochloride in Large Amount into the Subarachnoid Space

If the smaller fibers were successfully blocked without relief of pain, full spinal anesthesia was induced to test if the pain had a local origin. This technique was further modified to the conventional technique as described by Winnie and Candido.⁶⁰ This technique involved four sequential injections (normal saline, 0.25% procaine, 0.5% procaine, and 5% procaine). If the patient responded to the normal saline, the pain was classified as “psychogenic.” Response to 0.25% procaine was interpreted to mean that the pain was sympathetically mediated because the concentration is usually sufficient to block B fibers but not A-δ and C fibers. No response to the first two injections but pain relief with 0.5% procaine was interpreted as consistent with a somatic pain diagnosis as such a concentration is usually able to block B, A-δ and C fibers without blocking A-α, A-β, and A-γ fibers. The solution of 5% procaine blocked all fiber types, and failure to respond to that solution was interpreted as having a “central mechanism,” the possibilities of which include a central lesion, psychogenic pain, encephalization, and malingering.

Because this type of investigation is clearly time consuming and made with the assumption of a “typical” minimum blocking concentration response for each patient when clinically there is variation, a modification was proposed.^61–64 The newer technique requires only two injections, the first with normal saline and the second with 2 cc of 5% procaine. The pain response and neurological examination are then followed with the return of function of the different nerve fibers. This decreases the time for the procedure and does not rely on an average minimal concentration response of the nerve fibers. After a patient recovers sensation, only the sympathetic fibers remain blocked. Pain relief that remains after recovery of sensation suggests a sympathetically mediated pain.

Raj⁶⁵ presented a similar differential block strategy using the epidural space in 1977. The technique is limited, however, because of the even slower onset of the blockade and even less clear distinctions of the appropriate dose and concentration of local anesthetic for any particular patient compared with the intrathecal approach. In theory, however, the technique has the advantage of avoiding dural puncture.

The theory behind the differential spinal block was challenged by other investigators, including Fink.⁶⁶ He found that the size of the fiber did not truly explain the differential blockade and proposed the “bathed length principle.” To block conduction of a nerve, at least three consecutive nodes must have adequate local anesthetic exposure. He reasoned that thicker axons have larger intermodal distances, and this decreases the likelihood of blocking the larger axons compared with the smaller fibers. He was also able to explain the differential block of the sympathetic nerves was a result of decremental block. Fink⁶⁶ explained some of the phenomenon noted during a spinal epidural differential block and contributed to a better understanding of the clinical observations noted during a differential spinal block.

The true utility of these blocks have been questioned in recent years, and the use of the technique has certainly declined. There is a significant range of conduction speed and fiber size within a fiber type. A lack of correlation of size and necessary anesthetic concentration for blockade within a group creates an overlap of the fiber types that seems to “negate any possibility of obtaining steady state differential interruption” by local anesthetics.^67,68 The vulnerability of the fiber type to diffusion of the local anesthetic also seems to play a significant role in explaining the timing of the neural blockade.⁶⁹ The clinical result of the overlap is that a partial block of the A fibers has already occurred by the time C fiber activity is blocked.⁷⁰

The complex nature of pain often makes interpretation of even well-designed techniques difficult. The differential spinal blocks are a good reminder of not only the complexity of the nervous system but also the important role of performer bias, reliable and valid measurement, placebo response, and patient expectations. Although some authors⁶⁰ continue to promote the use of this procedure to establishing more accurate diagnosis, others^70,71 suggest significant caution in their use and applications.

Injections and Procedures Targeting the Zygapophysial (Facet) Joints

Lumbar medial branch blocks (MBBs) were first described in the late 1970s and were supported by anatomical studies showing that these branches of the lumbar dorsal rami were a valid and accessible target.^72–75 The sole purpose of the lumbar MBB is to determine if the patient’s pain is relieved by anesthetizing the nerves targeted. Because the lumbar zygapophysial joints (Z-joints) account for 15% to 40% of low back pain⁷⁶ and because the lumbar medial branches send an intraarticular branch that supplies these joints, by convention a positive response to the MBB suggests the pain is arising from the Z-joints (facet joints). Although there was a flurry of literature describing intraarticular facet joint corticosteroid injections in the 1980s, the literature suggests that these did not provide lasting relief in the majority of studies. Studies by Dreyfuss et al⁷⁷ and Kaplan et al⁷⁸ showed that lumbar MBBs were target specific and a valid test of zygapophysial joint pain. Lumbar medial branch neurotomy has emerged as the treatment of choice for patients with pain arising from the lumbar zygapophysial joints.

Radiofrequency (RF) neurotomy has been used successfully for the treatment of trigeminal neuralgia since the pioneering work by White and Sweet in 1969.⁷⁹ Early studies by Rees^80,81 and Shealy^80–85 reporting neurotomy or rhizolysis of the “facets” sparked interest in the Z-joint as a source of pain and target for treatment. Subsequent analysis of their technique, however, lead to conclusions that they were unsuccessful in severing the nerves to the lumbar zygapophysial joints.^72,73,86 After this, a modified technique targeting the correct nerve locations was reported in 1980,⁸⁷and subsequent studies showed good benefit. Analysis of the lesion created with RF neurotomy⁸⁸ led to a modification in technique that placed the needle and therefore lesion parallel to the target nerve.⁸⁹ A study using controlled diagnostic blocks as a diagnostic step and lesions created parallel to the target nerve demonstrated significant benefit for patients with chronic lumbar zygapophysial joint pain.⁹⁰ Pulsed RF (PRF) treatment has also been proposed as an alternative to conventional RF, although overall, there is less supportive evidence.

In the cervical spine, the premise for cervical MBBs as a diagnostic test for cervical zygapophysial joint pain also appears justified. The technique of selective blockade of the cervical dorsal rami was first suggested in 1980.⁹¹ Anatomical studies again led to further refinement and description of blockade of the cervical medial branches⁹² rather than their parent nerve. Diagnostic utility of cervical MBBs was established with studies for head pain and neck pain beginning in 1985. Pain referral maps were created that enabled practitioners to better predict the segmental level of painful joints.^93,94 In the early 1990s, a series of papers argued the importance of comparative diagnostic blockade and the shortcomings of single diagnostic blocks. Epidemiological studies that followed reported a high prevalence of pain arising from the cervical zygapophysial joints, especially in patients with head or neck pain from whiplash injuries. Injecting cervical zygapophysial joints with corticosteroids did not provide any additional benefit over anesthetizing the joint.

The first descriptions of cervical medial branch neurotomy appeared in the 1970s in papers focused predominantly on low back pain, and the first studies focusing exclusively on neck pain appeared in the early 1980s. Over the following decade, several more studies appeared, but similar to the lumbar treatments, the selection criteria and techniques varied, resulting in only fair overall results. The publications between 1995 and 2003 on cervical medial branch RF neurotomy demonstrated significant and prolonged benefit, most notably with the publications by Lord et al⁸⁸ and Govind et al⁹⁵ on patients with neck and head pain, respectively.