Spine pain arising from the facet joints is one of the leading causes of pain and disability. The facet joint is a common source of neck, shoulder, mid back, low back, leg pain, and headaches, increasing with age. Lumbar facet pain is usually due to chronic degeneration resulting from repetitive stress, while trauma is a frequent antecedent for cervical facetogenic pain. There are no historical or physical examination findings pathognomonic for diagnosis, but clinical assessment is important for ruling out other sources of pain and selecting candidates for diagnostic and therapeutic interventions. This chapter will cover the “reference standards” for diagnosis—intraarticular (IA) or medial branch (MBB) blocks—and review their sensitivity and specificity. This chapter will also explore potential reasons for treatment failure that should be taken into consideration for future clinical trials. For treatment of facet joint-related pain, study outcomes on radiofrequency (RF) denervation will be reviewed, which can provide significant relief ranging from 6 months to 1 year in carefully selected patients.
Keywordsfacet arthropathy, facet joints, facet syndrome, intraarticular injection, medial branches, neck and back pain, radiofrequency denervation
Spine pain is one of the leading causes of medical disability in the world. Most people will experience some type of neck and/or back pain throughout their life, with lifetime prevalence estimates as high as 84% for back pain and 67% for neck pain. Among all musculoskeletal disorders, low back pain (LBP) is the number-one reason patients seek medical attention and is the leading cause of disability; neck pain ranks as the fourth most common cause of years lost to disability. Between 2002 and 2004, the overall estimated financial costs for spine conditions, including lost wages, were estimated at over $200 billion. This ranks only behind joint pain as the most expensive musculoskeletal medical condition. ∗
∗ National Center for Health Statistics. ftp.cdc.gov/pub/Health_Statistics/NCHS/Dataset_Documentation/NHIS/2008/srvydesc.pdf ; Last accessed on November 29, 2009.
The causes of neck pain and LBP are complicated and often difficult to diagnose. The etiology is usually multifactorial, including muscles, ligaments, discs, nerve roots, and zygapophysial (facet) joints. The zygapophysial joint is a potential source of neck, shoulder, middle and lower back, and leg pain. It is also a potential source of headaches. Interventions for facet joint pain are second only to epidural steroid injections as the most commonly performed pain procedure in the United States. In 2006, interventions for facet pain represented approximately 37% of all pain interventions from Medicare, a 624% increase from 1997.
Anatomy and Function
The facet joints are paired structures that reside posterolaterally to the vertebral body; along with the intervertebral disc, they compose the three-joint complex. This complex works together to stabilize the joint and allow for different movements depending on the level. Facet joints are true synovial joints formed from the superior articular process of one vertebra and the inferior articular process of the vertebra above. The volume capacity of the joints is 1–1.5 mL and 0.5–1.0 mL in the lumbar and cervical regions, respectively.
The position of the joint relative to the sagittal and coronal planes helps determine the role that the joint plays in the restriction of motion. The lumbar facets vary in angle but are aligned lateral to the sagittal plane, with the inferior articular process facing anterolaterally and the superior articular process facing posteromedially. In the lumbar region, the higher joints tend to be oriented in more of a sagittal plane (26–34 degrees), providing resistance to axial rotation, while the lower lumbar facet joints are more coronally oriented, offering protection to injury incurred by flexion and shearing forces. The thoracic facets are the most vertically oriented joints, allowing for lateral flexion without axial rotation. In the cervical region, the shape and orientation of the joint also differ between the upper and lower joints. The C2–C3 joint, perhaps the most frequent cervical facet pain generator, is aligned approximately 70 degrees from the sagittal plane and 45 degrees from the axial plane, which inhibits rotation and anchors the C2 vertebra as a rotational pivot for the atlantoaxial joint (C1–C2). The area of greatest mobility in the cervical spine is at C5–C6, the second most affected cervical facet joint, which is where the cervical facets transition to their posterolateral position.
The medial branch is the terminal division of the posterior ramus, which provides sensory innervation to the facet joint ( Fig. 65.1 ). This smaller posterior division of the nerve root is divided into lateral, intermediate, and medial branches. The lateral branch in the lumbar region provides innervation to the paraspinous muscles, skin, and sacroiliac joint, while the small intermediate branch innervates the longissimus muscle. The medial branch is the largest of the divisions. It innervates the facet joint, multifidus muscle, interspinal muscle and ligament, and periosteum of the neural arch. Each facet joint is innervated by two medial branches, the medial branch at the same level and that at the level above (i.e., the L4–L5 facet joint is innervated by the L3 and L4 medial branches) ( Fig. 65.2 ).
The position of the medial branch in the lumbar spine does not vary significantly. It divides from the posterior primary ramus and wraps around the transverse process of the level below at the junction of the transverse process and superior articular process (i.e., the L3 medial branch lies on the transverse process of L4). The nerve traverses the dorsal leaf of the intertransverse ligament of the transverse process and courses underneath the mamilloaccessory ligament, splitting into multiple branches as it crosses the vertebral lamina (see Fig. 65.2 ). The mamilloaccessory ligament can become calcified and be a source of nerve entrapment, especially at L5. The main variation in the lumbar spine is at L5, where it is the primary dorsal ramus itself that is amenable to blockade.
The thoracic spine is generally similar to the lumbar spine in terms of innervation, with each joint being supplied by two medial branches, although some anatomic studies have shown that the dorsal ramus may send branches directly to the facet joint before dividing. In the upper and middle thoracic regions, two medial branches are present, including a cutaneous branch that is absent at lower thoracic levels. In the thoracic spine, medial branches assume different courses depending on the level. At lower thoracic levels, the optimal blockade site is similar for lumbar medial branch blocks. At higher levels, the nerve swings laterally to circumvent the multifidus muscle, thereby removing multifidus contraction as a means of needle confirmation prior to denervation. From T4 to T8 in some specimens, the medial branch courses somewhere in the intertransverse space without making contact with bone, making medial branch blocks challenging. At the middle and upper thoracic levels, the superolateral corner of the transverse process is the most consistent point for blockade ( Fig. 65.3A and B ).
The innervation of the cervical facets is more varied and complicated. There are eight cervical nerve roots, which exit above the corresponding vertebral bodies. Similar to the lumbar and thoracic regions, the C3–C4 through the C7–T1 joints receive innervation from the medial branches at the same level and the level above. The nerves curve around the waist of the articular pillars except at C7 and C8, where the anatomy is more variable. Medial branches at higher levels adhere tightly to the periosteum with tight fascia and the tendons of the semispinalis, which makes positioning more predictable ( Fig. 65.4A and B ). The majority of the innervation of the C2–C3 joint comes from the dorsal ramus of C3. The C3 dorsal ramus divides into two separate medial branches, the larger of which is known as the third occipital nerve. The C2 dorsal ramus divides into up to five branches, the largest of which is the greater occipital nerve. Pathology involving branches of the C2 and C3 dorsal rami is a common source of occipital headaches. In some people, dual medial branches are present at a single level, which is most common at C4 (27%) and C5 (15%).
The facet joints contain a rich supply of encapsulated, unencapsulated, and free nerve endings. Previous work has established the presence of substance P and calcitonin gene–related peptide reactive nerve fibers in cadaveric facets. Inflammatory mediators, including prostaglandins, interleukin-6, and tumor necrosis factor-α, have been demonstrated in the facet cartilage of patients undergoing surgical therapy for degenerative lumbar disease. Studies have demonstrated that leakage of these cytokines through the ventral joint capsule may be partially responsible for radicular symptoms in lumbar spinal stenosis. In addition, subchondral bone and intraarticular (IA) inclusions of facet joints have nerve endings, signifying that structures besides the joint capsule may be potential pain generators.
With the exception of whiplash injuries and major spine trauma, facet arthropathy and facet-mediated pain are infrequently due to acute injury. In a study performed in trauma patients, Levine et al. reported that bilateral facet dislocation occurred in 11% of cases requiring surgical stabilization, with the mechanism of action being “flexion-distraction” in 29 of 30 cases. Instead, years of repetitive strain, intervertebral disc degeneration, and minor trauma are more commonly implicated. Similar to other degenerative joint diseases, there is a poor correlation between pain and the degree of inflammation or degeneration. Facet arthropathy is known to occur more commonly in the elderly, which is consistent with the concept of a degenerative disorder.
Cadaveric and Animal Studies
Cadaveric studies have demonstrated that the greatest degree of motion and strain in the lumbar spine occurs in the lowest two facet joints (L4–L5 and L5–S1). At these joints, strain is maximized by forward flexion. In the middle joints (L3–L4), the greatest degree of strain is observed with contralateral bending, whereas the opposite was seen at the most cephalad joints (L1–L2 and L2–L3). Fusion of an intervertebral level has been shown to accelerate degeneration at adjacent levels.
Whereas facet joint pain is not normally considered an active inflammatory state, chronic strain and repetitive stimulation can lead to fluid collection and joint distention. If the intervertebral foramen is already narrowed from other pathology (intervertebral disc herniation, osteophyte formation, etc.), a hypertrophied facet joint may further compress the nerve root, thereby manifesting in radicular pain. In some cases, spasm of the paraspinous musculature can be superimposed.
The presence of facet arthropathy is more common in the elderly. The intervertebral disc and facets work in concert, such that degeneration of the intervertebral disc creates additional strain on the facet joints and vice versa. The two most caudal facet joints (L4–L5 and L5–S1) are associated with the greatest degree of degenerative disc disease, with L5–S1 being the most common clinically affected joint and L4–L5 being the most frequently radiologically affected joint. Intervertebral discs degenerate at an earlier age than facet joints, although one cadaveric study found that 93% of cadaver specimens between 40 and 49 years of age had evidence of facet arthrosis. Less common causes of facetogenic pain include inflammatory arthritis and pseudocysts.
As already noted, facetogenic pain can occasionally result from trauma, especially rapid deceleration injuries. In one study, capsular and articular damage was observed in 77% of facet joints in people who died in motor vehicle accidents. The most common presentation of trauma-induced facet pain is whiplash injury, which may account for over 50% of cases of chronic neck pain following motor vehicle accidents. However, trauma still only accounts for a relatively small portion of cervical facetogenic pain (13%–23%).
The prevalence of facet pain is a source of controversy. The lumbar facet joints are the most commonly affected due to the high frequency of LBP in the general population and the greater load borne by these joints. However, the cervical facet joints account for a higher percentage of chronic neck pain than the lumbar facet joints do in patients with chronic LBP. One limiting factor in determining the true incidence of facet pain is that the diagnosis cannot be made by history, physical examination, or radiologic findings; hence there is no reference standard according to which the sensitivity and specificity of diagnostic injections can be judged. The most reliable method for determining facetogenic pain is with image-guided medial branch (facet joint nerve) or IA facet joint blocks.
The prevalence of lumbar facet pain varies widely in the literature, with the best estimates ranging between 10% and 15%. Although comparative medial branch blocks (MBBs) have been endorsed as a diagnostic standard, the other branches of the dorsal primary ramus will also invariably be blocked when local anesthetic is injected, which may overestimate the prevalence. Another source of error is that most epidemiologic studies evaluating the prevalence of lumbar facet joint pain exclude patients with radicular symptoms, despite the fact that facet arthropathy can cause neuroforaminal stenosis and may occur concurrently with radicular pain.
Estimating the prevalence of cervical facet pain is equally challenging, with some of the most elegant clinical studies conducted exclusively in patients with whiplash injuries. However, the best studies using double blocks have generally reported prevalence rates ranging between 49% and 60% in patients with chronic, nonradicular neck pain. Among patients with chronic middle- and upper-back pain (BP), the estimated prevalence varies between 40% and 50%. Not surprisingly, reviews performed for lumbar facet joint and sacroiliac joint pain have generally found that studies using lower cutoffs (e.g., 50%) as the threshold for a positive response and those using single blocks report higher prevalence rates than studies that use higher cutoff values and employ double blocks.
History and Physical Examination
Many studies have investigated the ability to predict response to diagnostic facet blocks on the basis of history and/or physical examination. Although clinical symptoms and pain referral patterns can help guide physicians, the specificity is very poor. The terms lumbar facet syndrome and facet loading were coined from a small, poorly designed retrospective study of 22 patients done in 1988. Subsequent larger and methodologically sound studies failed to validate these findings. Yet, many pain physicians continue to rely on misguided signs and symptoms as being diagnostically significant. Tenderness to palpation in the cervical and lumbar paraspinous regions was found to be a positive predictor of radiofrequency (RF) denervation outcome in two large retrospective studies, but such findings need to be confirmed prospectively.
Pain referral patterns can provide clues to diagnosis. Studies have been conducted by provoking pain in healthy volunteers (i.e., distending the joint capsule and stimulating medial branches) and investigating pain patterns in patients whose symptoms are relieved by diagnostic blocks. Similar to other sources of spinal pain, the referral patterns associated with facet pain tend to be variable and overlapping.
Although there is a great deal of overlap between different spinal levels and different structures (i.e., facet joints and discs) at the same level, when the results of provocation and analgesic studies are combined, some patterns emerge ( Figs. 65.5 and 65.6 ). In the lumbar region, the upper facet joints tend to refer pain to the flank, hip, and upper lateral thigh. For lower levels, pain is generally experienced in the posterolateral thigh and occasionally the calf. In the cervical spine, upper facet arthropathy usually manifests as pain felt in the posterior upper neck and occipital region. Pathology involving middle cervical facet joints tends to radiate into the lower neck and supraclavicular region, while lower cervical facetogenic pain typically causes pain in the base of the neck and scapular region.
Although it is common for patients with chronic spinal pain to undergo multiple imaging studies, radiologic examination has limited utility in the diagnosis of facet-mediated pain. Whereas the lumbar facet joints account for a small percentage of chronic LBP cases, the prevalence of facet pathology on computed tomography (CT) scans is between 40% and 85%, with the rate increasing significantly with age. Similar rates of abnormal findings have been found in asymptomatic volunteers who undergo cervical and thoracic magnetic resonance imaging (MRI). Studies using MRI, CT, and other imaging studies to predict response to facet blocks have been decidedly mixed, with most showing a poor correlation.
Although fluoroscopic guidance for facet joint injections is the standard, use of ultrasound guidance is acknowledged as another modality for medial branch and facet joint injections. For RF denervation, ultrasound does not enable one to place electrodes parallel to the course of the target nerves, which limits its use in this context. It has also been shown to be less accurate in obese patients. Ultrasound was shown in a randomized feasibility clinical trial to provide significant differences in visual analog scale (VAS) pain scores and pain remission rate for facet joint injections when compared with blind injection without ultrasound. Ultrasound-guided facet joint injection in the middle and lower cervical spine has been shown to be comparable to CT guidance in VAS pain relief in addition to resulting in significant reduction in procedure duration without any exposure to radiation.
The inability to predict facet pain from history, physical examination, or radiologic study has led to the widespread use of medial branch and IA facet blocks for diagnosis. Although diagnostic MBBs have been used in multiple studies, some technical and anatomic considerations limit their diagnostic utility. Studies have demonstrated that volumes as small as 0.5 mL spread out over 6 cm 2 of tissue. Hence the intermediate and lateral branches are likely to be anesthetized with typical injection volumes, thereby blocking afferent transmission from portions of the paraspinous musculature and sacroiliac joint. A randomized study has demonstrated a clinically relevant improvement in specificity without undermining sensitivity when 0.25 mL of local anesthetic was used for cervical MBBs compared with 0.5 mL. The use of IA facet injections can reduce issues related to the inadvertent spread of local anesthetic, but they can be technically challenging, with a failure rate exceeding 33%.
Furthermore, excessive volumes of local anesthetic solution can rupture the joint capsule, leading to spread into the intervertebral foramen, epidural space and paraspinous musculature. Studies that have used MBBs have generally shown better results than those that used IA injections as prognostic procedures. A multicenter case-control study was performed to determine which prognostic facet block, IA injections or medial branch block, was a better predictive tool for analgesic response to RF denervation. The findings from this retrospective study suggest that MBBs are a superior prognostic tool for RF denervation outcomes compared to IA blocks. However, the major reason for treatment failure is poor patient selection (e.g., greater disease burden, psychosocial pathology), which should be taken into consideration when designing clinical trials.
False-Positive Diagnostic Blocks
Both MBBs and IA blocks are associated with high rates of false-positive results. False-positive rates have ranged from 25% to 40% in the lumbar spine and from 25% to 30% in the cervical spine. Although some experts have advocated comparative local anesthetic blocks as an alternative to “placebo controls,” this paradigm is not without limitations. A randomized, double-blind study of cervical MBBs in 50 whiplash patients with neck pain using normal saline, lidocaine, and bupivacaine in random order found comparative blocks (serial lidocaine and bupivacaine injections) to be highly sensitive (88%) but only marginally specific (54%).
Potential causes of false-positive blocks include placebo response, sedation, excessive superficial local anesthesia, and the spread of local anesthetic to other pain-generating structures. It is our belief that the use of sedation for diagnostic blocks should be limited, as even benzodiazepines can lead to muscle relaxation and interfere with a patient’s ability to assess pain relief. However, this assumption has been challenged by some, who argue that using a more stringent pain relief threshold (≥80%) mitigates against a higher false-positive rate when sedation is administered. Prospective and retrospective studies performed in the lumbar spine have found no difference in RF denervation outcomes between those patients who experience 50% or greater relief and higher thresholds. Moreover, a prospective crossover study evaluating the effect of sedation on diagnostic specificity found that even light sedation significantly increased the rate of positive blocks.
Several steps can be taken to negate or minimize the role of other factors in false-positive blocks. Dreyfuss et al. found that for lumbar MBB, targeting a lower point midway between the upper border of the transverse process and mamilloaccessory ligament significantly reduced epidural and foraminal spread compared with the conventional target point at the superomedial border of the transverse process. Cohen et al. showed that reducing the volume of injectate from 0.5 to 0.25 mL for cervical MBB resulted in a greater than 50% decrease in spread to adjacent pain-generating structures. In a randomized, double-blind study, Ackerman et al. found that injecting superficial lidocaine down to the facet joint or medial branch resulted in a greater than five-fold increase in positive blocks in the first of two blocks than when patients received superficial saline; this effect was not observed in the second block. One way to reduce the need for superficial injection is to use a single-needle technique, which was demonstrated in a randomized crossover study to decrease the amount of superficial lidocaine by 40% while providing comparable pain relief and contrast spread compared with the traditional multiple-needle technique. Recommendations to limit false-positive blocks are listed in Table 65.1 .
False-negative blocks have garnered less attention than false positives, but they can be a source of misdiagnosis and failure to select appropriate candidates for treatment. A study conducted in which 15 volunteers underwent MBBs with local anesthetic or saline followed by capsular distention estimated the incidence of false-negative blocks to be 11% based on the inability of local anesthetic infiltration to block the pain from facet joint capsular distention. A probable cause was deemed to be “aberrant innervation,” which would account for some cases of failed RF denervation in IA block responders. One of the principal causes of false-negative blocks is thought to be vascular uptake. This has been reported to range between 6% and 30% per level, with one study finding rates of 3.9% at cervical levels, 3.5% at lumbar levels, and 0.7% at thoracic levels. One study found that when vascular uptake occurs, even if the needle is repositioned, analgesia will be obtained only half the time. The most reliable means of detecting vascular uptake is with digital subtraction angiography, which has been shown to enhance the detection rate. Other potential causes of false-negative blocks are failure to discern between baseline and procedure-related pain and missing a target nerve or nerves.
Selection Criteria: 50% Versus 80% Relief, Single Versus Double Medial Branch Blocks
A great deal of debate has centered on the selection of patients for denervation, with the two primary arguments revolving around the percentage of pain relief and the number of blocks. The thresholds for pain relief studied are somewhat arbitrary, as previous research has determined a two-point or 30% decrease in pain to be clinically meaningful. Multiple retrospective analyses have found no difference in results between using 50% and 80% relief as the cutoff for a positive block. A multicenter prospective correlation study by Cohen et al. set out to evaluate the influence of pain relief after MBBs on RF outcome and determine an optimal “cutoff” threshold for diagnostic lumbar facet blocks. Their study showed no significant differences in RF outcomes based on any MBB with a pain relief cutoff threshold above 50%. No optimal threshold for designating a diagnostic block as positive, above 50% pain relief, could be calculated.
The issue of how many blocks should be performed is also the subject of considerable controversy. The argument in favor of double blocks is bolstered primarily by the high false-positive rate of uncontrolled blocks, whereas those who advocate single blocks point to time and cost constraints, the comparable complication rate between diagnostic blocks and RF denervation, and the fact that confirmatory diagnostic procedures are not used to select patients for intradiscal procedures, surgery, and other more invasive interventions. Multiple retrospective studies evaluating outcomes for lumbar facet, cervical facet, and sacroiliac (SI) joint RF denervation have failed to find a difference in success rates between patients selected with one and two diagnostic blocks. In a large, multicenter, randomized study comparing the cost effectiveness of zero, one, and two blocks before lumbar facet RF denervation, although the RF success rate was highest in the double-block group (64%), the overall success rate was 50% higher in the zero-block group (33% vs. 16% vs. 22%, respectively). In the cost-benefit analysis, the cost per effective treatment was $6054 in the zero-block group, $16,236 in the single-block group, and $14,238 in the double-block group.
However, this treatment paradigm remains controversial. A prospective, single-blind, triple-crossover study evaluating the diagnostic validity of facet joint injections in 60 patients with chronic LBP showed that a single IA facet block with local anesthetic provided pain relief comparable to that from a placebo injection or a sham procedure (needle placement outside the facet joint with no injection and was not useful as a diagnostic tool. The authors asserted that the only reliable means to identify a painful facet joint was to use double or triple controlled blocks with at least one placebo injection. However, the patients in this study were not prescreened by diagnostic blocks; it is therefore likely that only a small number of these patients suffered from facet arthropathy.