INTRODUCTION

Low back pain is ubiquitous to mankind by virtue of our upright posture. Traditionally, it has been believed that most episodes of spinal pain are short lived and that at least 90% of patients with low back pain recover in about 6 weeks with or without treatment.^1–3

Contrary to this prior assumption of only 10–20% patients experiencing recurrent or chronic symptoms after an initial episode of low back pain, recent literature puts that number as high as 35–79%.^4–7

The challenge lies, in this subgroup of subjects, to identify the specific pain generator so that appropriate interventional therapy may be initiated as and when indicated. The study by Kuslich et al.⁸ identified along with intervertebral disc, dura, ligaments, fascia, and muscles, zygapophyseal or facet joints as capable of transmitting pain in the low back region. Even prior to this anatomical study, Goldthwait⁹ in 1911 first recognized lumbar zygapophyseal joints as potential source of low back pain. He believed that joint asymmetry could result in pain secondary from nerve root pressure. Subsequently, in 1933, Ghormley¹⁰ first coined the term ‘facet syndrome’ as lumbosacral pain with or without sciatic pain, occurring after a twisting or rotary strain of the lumbosacral region. Badgley¹¹ in 1941 suggested that facet joints by themselves could be a primary source of low back pain separate from the nerve compression component. However, it was not until 1963 when Hirsch et al.¹² demonstrated that the low back pain distributed along the sacroiliac and gluteal areas with referral to the greater trochanter could be induced by injecting hypertonic saline in the region of the lumbar facet joints. In 1976, Mooney and Robertson,¹³ and three years later McCall et al.¹⁴ used fluoroscopy to confirm the location of intra-articular lumbar facet joint injections in asymptomatic individuals, demonstrating reproduction of back and leg pain after injection of hypertonic saline. Eventually, Marks¹⁵ in 1989, followed by Fukui et al.¹⁶ in 1997, described the distribution of pain patterns after stimulating the lumbar facet joints.

Recent literature¹⁷ indicates that the prevalence of chronic lumbar zygapophyseal joint-mediated pain has a wide range that varies from 15% in the relatively younger patients, with confidence limits of 10–20%. This prevalence can be as high as 40% among the elderly population with confidence limits of 27–53%.¹⁸

ANATOMY

The lumbar facet or zygapophyseal joints are paired, true synovial, diarthrodial joints that form the posterior aspect of the respective intervertebral foramina. The joint consists of hyaline cartilage, a synovial membrane, a fibrous capsule, and nociceptive fibers transmitted via the medial branches of the dorsal rami.^19,20 These facet joints are innervated by the medial branches of the dorsal rami (posterior primary rami) that exit the intervertebral foramina. The posterior primary rami travel posteriorly over the base of the transverse process. The dorsal ramus divides into a medial, an intermediate, and a lateral branch. The lateral branch ascends from the dorsal ramus just before it reaches the transverse process. The medial branch passes under the mammilloaccessory ligament and sends branches to the adjacent facet joint, the facet joint below, and the more medial erector spinae muscles. Thus, the joint is typically innervated from a branch at the same level and a branch originating from the foramen above. In contrast, the dorsal ramus at the fifth lumbar vertebral level (L5) travels between the ala of the sacrum and its superior articular process, which divides into the medial and lateral branches at the caudal edge of the process, the medial branch continuing medially, where it innervates the lumbosacral joint.^21–23

Neuroanatomical studies^12,24–26 indicate that the facet joint capsule is richly innervated, containing both free and encapsulated endings, and can undergo extensive stretch under physiological loading.²⁷ It has been reported that protein gene product (PGP) 9.5, substance P (SP), calcitonin gene-related peptide (CGRP), dopamine β-hydroxylase (DBH), vasoactive intestinal polypeptide (VIP), neural peptide Y (NPY), and choline acetyl transferase (chAT) immunoreactive (IR) fibers are present within the human facet joint capsule.^{21,26,28–30} Ashton et al.³¹ confirmed immunoreactivity for SP, CGRP, and VIP in surgically removed human facet joint capsule. A subsequent anatomical study confirmed higher concentration of inflammatory cytokines in facet joint tissue from patients with lumbar spinal stenosis than in lumbar disc herniations, suggesting that these cytokines may have some contribution to the etiology of symptoms in degenerative spinal stenosis.³²

PATHOPHYSIOLOGY

With aging, the changes that occur in the zygapophyseal joints are the same as those seen in any diarthrodial joint. The earliest change is synovitis, which may persist with the formation of a synovial fold which projects into the joint space itself. Later, degenerative changes set in gradually, and eventually become more marked. Eventually, the capsular laxity allows for subluxation of the joint surfaces. Continued degeneration mainly due to repeated torsional strains results in the formation of subperiosteal osteophytes, and possibly subchondral synovial cysts. The end result is that of gross articular degeneration, with almost complete loss of articular cartilage, and formation of bulbous zygapophyseal joints and marked periarticular fibrosis. The pathogenesis of a degenerative cascade in the context of a three-joint complex involving the intervertebral disc and the adjacent zygapophyseal joint also leads to the assumption that the degenerative changes within the disc are also believed to lead to associated facet degeneration.^33–35

PAIN PATTERNS

The clinical presentation of lumbar zygapophyseal joint-mediated low back pain appears to overlap considerably with the presentation of low back pain due to various other etiologies. Lumbar facet joints have been shown to be capable of producing pain in the low back and referred pain in the lower extremity in normal volunteers.^13–16 McCall et al.¹⁴ concluded that pain referral patterns overlap between the upper and lower lumbar spine. Marks¹⁵ also studied patterns of pain induced from lumbar facet joints, from the posterior primary rami of L5, and from the medial articular branches of the posterior primary rami from T11 to L4. He observed no consistent segmental or somatic referral pattern. Marks also concluded that pain referred to the buttocks or trochanteric region occurred mostly from the L4 and L5 levels, while inguinal or groin pain was produced from L2 to L5. This latter finding led to the conclusion that the nerves innervating the joints gave rise to distal referral of pain more commonly than the facet joint itself. Fukui et al.¹⁶ concluded that the major site of referral pain from the L1–2 to L4–5 joints was the lumbar region itself. However, stimulation of the L5–S1 joint caused referred spinal pain as well as gluteal pain. Referred pain into the lower extremities was not observed by Fukui et al.¹⁶ as was reported by Mooney and Robertson.¹³ However, it must not be overlooked that in the study by Mooney and Robertson, the relatively large volume of the injectate (3 cc) used may very well have inadvertently anesthetized structures other than the facet joint intended. In essence, these studies indicate that there is no clear dermatomal or somatic pattern that is diagnostic of lumbar facet joint-mediated pain and an overlap with a varied pattern is the norm.

DIAGNOSIS

All this being said, there is no identified correlation between a clinical picture or any imaging study used, such as the magnetic resonance imaging (MRI), computed tomography (CT) scan, single photon emission computed tomography (SPECT), or radionuclide bone scan in diagnosing pain mediated via the lumbar zygapophyseal joint.^36–40 Simple static radiographs of the lumbar spine, and also dynamic images including standing flexion–extension films, may reveal evidence of arthritis involving the lumbar zygapophyseal joint as commonly in asymptomatic individuals as in patients with axial low back pain. What remains contentious is how lumbar zygapophyseal joint-mediated pain should be diagnosed.

Another point not to be overlooked is the fact that psychological factors can be involved in cases of chronic low back pain. Long before the monumental work of Freud, it was recognized that emotional states could be associated with physical symptoms in the absence of an organic pathology. Consequently, it behooves the primary spine care provider to learn how to identify these patients, how to diagnose their conditions, how to treat them, and how, when, and from whom to seek consultation. Unless chronic spinal pain is simultaneously understood and treated from a musculoskeletal and psychological perspective, treatment failure will be the norm. Treatment modalities directed only at a physical or mechanical problem will not be sufficient or helpful if there is an associated unrecognized or ignored psychological problem. Diagnosis is the cornerstone of a rational treatment approach to any kind of illness. Spine physicians confronted with a patient describing chronic or ongoing back pain are pressed to establish a diagnosis. Appropriately, any clues derived from the history and from the physical examination are pursued toward that end. Schemata describing the great classifications of disease illustrate the breadth and scope of physical pathologic conditions that can be associated with low back pain. Thus, the differential diagnosis in a patient presenting with chronic low back encompasses numerous entities. Clinical problems tend to become complex when there is difficulty in establishing a diagnosis or when then the treatment options are unclear or difficult to implement. In approximately a quarter of the patient population there is a coexisting lesion responsible for the patient’s symptoms. Up-to-date laboratory and imaging techniques notwithstanding, nothing supplants and nothing exceeds the diagnostic accuracy of a thorough history and a careful physical examination. Because the causes of back pain are legion and involve virtually every organ system as well as the psyche, a truncated history, focused only on the spine should be avoided. A tunnel vision, essentially blinding oneself, will only lead to diagnostic followed by therapeutic pitfalls.

REVIEW OF LITERATURE SUPPORTING CLINICAL EFFECTIVENESS OF LUMBAR RADIOFREQUENCY

Historically, interventional management of lumbar zygapophyseal joint pain started with the use of a knife-cut in the region of the lumbar facet joint for denervation. Rees⁵⁷ reported immediate relief of pain in 998 of 1000 patients suffering from ‘intervertebral disc syndrome’ using this technique. Shealy^58,59 introduced the procedure in North America but, after a large number of operations, changed to a radiofrequency coagulation lesion through a probe placed in the region of the nerve supply of the facet joint.

Many investigators have studied the effectiveness of radiofrequency denervation of medial branches of the lumbar zygapophyseal joint, of which only four are prospective studies. These include two randomized, controlled trials by van Kleef et al.⁶⁰ and Leclaire et al.,⁶¹ one double-blind, controlled study by Gallagher et al.,⁶² and a case series by Dreyfuss et al.⁵⁵ In each of these studies, patients were included when they had failed to respond to traditional conservative treatment, such as a trial of bed rest, medications, and physical therapy. However, the inclusion or exclusion criteria were inconsistent across the studies.

Gallagher et al.⁶² included patients with low back pain for at least 3 months and used diagnostic injections of 0.5 cc of 0.5% bupivacaine in and around the lumbar zygapophyseal joint. Patients with good or equivocal response to this injection within 12 hours were randomized to undergo either a lumbar facet joint denervation or placebo treatment. Radiofrequency lesions were performed for 90 seconds at a temperature of 80°C. McGill pain questionnaire⁶³ and the visual analog scale (VAS)⁶⁴ were used at 1 month but only the VAS at 6 months. Statistical comparison of pain scores was conducted using the Kruskal-Wallis and Mann Whitney U tests. Reduction in pain scores was approximately 50% at 1 month and was sustained at 6 months. The only difference in pain scores were seen when comparing people, all of whom met the inclusion criteria of a good response to an initial injection using 0.5 cc of 0.5% bupivacaine. The only statistical differences in pain scores in the post-treatment group occurred during comparing patients with a good initial response to local anesthetic injection but proceeding to a sham procedure.

The randomized, controlled trial of van Kleef et al.⁶⁰ included 31 patients with low back pain of at least 12 months. These were injected with 0.75 cc of 1% lidocaine about the two medial branches innervating a particular zygapophyseal joint. Patients were evaluated 30 minutes after this injection. Those with at least a 50% relief on a four-point Likert scale (0–30% pain relief was considered as no relief; 30–50% was defined as moderate; 50–80% pain relief was good, and 80% or more was considered as pain free) were randomized. Radiofrequency probes were placed in both groups but only one group underwent the actual radiofrequency lesioning for 60 seconds. Patients were followed for at least 12 months and the outcome measures included the VAS instrument, Oswestry disability index, and global perceived effect. At 8 weeks, a higher rate of success in the treatment group compared to the control group was noted. Subsequent log rank test showed a statistically significant difference (p=0.02) at 3, 6, and 12 months.

Leclaire et al.⁶¹ performed a diagnostic intra-articular facet joint block with 0.5 cc of lidocaine 2% and 0.5 cc of 40 mg of triamcinolone acetonide. Significant relief of low back pain for at least 24 hours in the week after this injection was considered as positive. For each nerve, radiofrequency neurotomy was performed at two locations, one at the proximal portion and another at the distal portion of the dorsal ramus with the temperature controlled at 80°C for 90 seconds. For patients in the control group the temperature of the probe was maintained at 37°C. Follow-up assessments were undertaken at 4 and 12 weeks when the disability questionnaire, the visual analog scale, the triaxial dynamometry, and the return to work assessment was repeated. The primary analysis was based on the intention to treat principle. At the end of 4 weeks, there was an improvement in the Roland-Morris score⁶⁵ by 8.4% and 2.2% in the radiofrequency and control groups, respectively. However, at 12 weeks there was no significant difference in the Roland-Morris,⁶⁵ Oswestry disability,⁶⁶ VAS,⁶⁴ strength/mobility, and return-to-work status in both groups.

Dreyfuss et al.⁵⁵ used 0.5 cc of lidocaine 2% for the initial zygapophyseal joint injection. A positive response was defined as at least 80% relief of pain lasting longer than 1 hour. At the end of 1 week, this group of patients underwent another diagnostic injection using bupivacaine 0.5%. Subjects experiencing greater than 80% relief, lasting longer than 2 hours, were included in the study. The technique used by Dreyfuss et al.⁵⁵ was different in that the radiofrequency ablation was performed for 8–10 mm along the length of the target nerve with the temperature being maintained at 85°C. The outcome measures included VAS,⁶⁴ Roland-Morris questionnaire,⁶⁵ prescription analgesic medication, McGill pain questionnaire,⁶³ Short-Form (SF)-36 general health questionnaire,^67,68 North American Spine Society (NASS)⁶⁹ treatment expectations, isometric push and pull, dynamic floor-to-waist lift, and isometric above-shoulder lifting tasks. In addition, a needle electromyography of the L2–L5 bands of the multifidus muscle was performed before and at 8 weeks after the radiofrequency denervation. Patients were evaluated with at least a 12 month follow-up. Statistical analysis was performed using median scores and interquartile ranges of all outcome measures, Friedman two-way analysis of variance, Wilcoxon paired test, and Spearman correlation coefficients. Sixty percent of patients experienced at least a 90% pain reduction, whereas 87% of patients experienced at least a 60% reduction in pain. This effect lasted for 12 months. Forty-seven percent of patients obtained absolute reduction of at least 5 in the VAS,⁶⁴ and 60% obtained relative reduction of at least 3. They noted that, even for patients who experienced pain for more than 5 years, radiofrequency denervation of the medial branch was helpful, cost effective, and less time consuming than other interventions such as exercise-based physical therapy or manipulative care. The results of needle electromyography warrants a special mention, with 11 of 15 patients sustaining 100% denervation, whereas 4 achieved partial denervation.

A critical review of each of these studies is of prime importance if one is to derive a deep understanding of this therapeutic intervention for the treatment of lumbar zygapophyseal joint mediated pain. Each study used a small sample size, with van Kleef et al.⁶⁰ and Dreyfuss et al.⁵⁵ using only 15 subjects each in the treatment group. Gallagher et al.⁶² and Leclaire et al.⁶¹ did not specify how a successful outcome to the diagnostic injections was objectively rated. Dreyfuss et al.⁵⁵ and Leclaire et al.⁶¹ were the only two sets of authors who followed their patients for at least 12 months. There were also discrepancies noted in reporting the diagnostic injection. Gallagher et al.,⁶² in their abstract, mention that the solution used was a combination of local anesthetic agent with steroid, whereas the main text of the paper states that the solution used was only local anesthetic. Van Kleef et al.⁶⁰ used a double-blind, placebo-controlled injection paradigm. Unfortunately, patients were selected on the basis of a single positive diagnostic injection which, as discussed previously, is known to have a false-positive rate of approximately 38%.⁵¹ The technique used by van Kleef et al.⁶¹ was different, such that the electrodes were placed at an angle to the target nerve. Laboratory studies have shown that the electrode must lie parallel to the nerve if the nerve is to be optimally and maximally coagulated.⁵⁶ This may explain why van Kleef et al.⁶⁰ obtained only modest results. Similarly, van Kleef et al.⁶⁰ and Dreyfuss et al.⁵⁵ included subjects with low initial VAS scores. Although Dreyfuss et al.⁵⁵ used the strictest inclusion criteria in their prospective audit, they unfortunately treated only a small sample size. By performing two different diagnostic injections with two anesthetic agents, each with a different duration of action, they tried to eliminate the false-positive responders. Also, during the radiofrequency denervation, the electrodes were placed meticulously to optimize coagulation of the nerve. As previously stated, these authors conducted a needle electromyography of L2–L5 bands of multifidus before and at 8 weeks after the denervation. Although electromyographic evidence of denervation indicates that the radiofrequency ablation successfully coagulated the assessed medial branch, such an outcome did not correlate with symptom improvement. However, this study did not include a control group. In the Leclaire et al.⁶¹ study, the inclusion criteria was significant relief of low back pain for more than 24 hours after the facet injection. Roland-Morris score⁶⁵ was used as an important outcome measure, which is a measurement of functional limitation and not of pain relief. A more in-depth questionnaire on pain, such as the McGill Pain Questionnaire,⁶³ possibly could better identify the therapeutic response in these patients.

These conflicting trials, discussing the use of radiofrequency denervation in the management of lumbar zygapophyseal joint-mediated pain, highlight the use of varied inclusion criteria, varied treatment, lack of uniform outcome measures, use of concurrent interventions, and lack of a sufficient follow-up. A detailed review of these deficiencies and critical analysis of the relevant peer-reviewed publications can be found elsewhere.^70,71 A more recent anatomical study undertaken by Lau et al.⁷² validates the prime importance of placement of electrodes parallel to the target nerve. This underscores the fact that a meticulous technique will lead to superior results.

Because of the insufficient quantity of high-quality studies, it is evident that prospective, randomized, controlled trials with uniform inclusion and exclusion criteria and assessing an appropriate number of subjects in each are necessary, as dictated by a power analysis. Additionally, such studies should use a standardized treatment, uniform outcome measures, and have an adequate follow-up duration of at least 12 months.

The Agency for Health Care and Policy Research describes evidence rating for management of low back pain in adults.⁷³ For the development of these guidelines, a rating schema was used to assess the strength or efficacy of a treatment or procedure. They offered five levels of strength: level I (conclusive): research-based evidence with multiple relevant and high-quality scientific studies; level II (strong): research-based evidence from at least one properly designed randomized, controlled trial of appropriate size (with at least 60 patients in each arm) and high-quality or multiple adequate scientific studies; level III (moderate): evidence from well-designed trials without randomization, single group pre-post cohort, time series, or matched case-controlled studies; level IV (limited): evidence from well-designed nonexperimental studies from more than one center or research group; and level V (intermediate): opinions of respected authorities, based on clinical evidence, descriptive studies, or reports of expert committees. Critical review of the literature reveals that the type and strength of efficacy evidence for radiofrequency neurotomy in managing lumbar zygapophyseal joint-mediated pain is level III evidence.

CONCLUSION

In summary, the process of establishing a diagnosis of lumbar zygapophyseal joint-mediated pain, the techniques used to make this diagnosis and implementing an effective treatment algorithm has evolved extensively over the past three decades since Mooney and Robertson proposed the posterior joints of the lumbar spine as a potential source of low back pain (Fig. 93.1). The question is not whether the syndrome of lumbar zygapophyseal joint pain exists but how to come to an accurate diagnosis and how to treat it effectively. The challenge lies in the accurate diagnosis of this entity so that appropriate interventional therapy may be instituted. Regardless of the approach used, the diagnosis can only be affirmed using, at a minimum, a double-block paradigm due to the low positive predictive value of a single diagnostic injection. If attractive, compelling results are to be achieved following a lumbar radiofrequency denervation, the clinician needs to follow two basic but often overlooked steps: establish an accurate diagnosis of lumbar facet syndrome and then, during treatment, adhere to a meticulous placement of electrodes.

Fig. 93.1 Effective treatment algorithm.