INTRODUCTION

Promoting an algorithmic approach to the patient with lumbar axial pain may at first appear paradoxical. Algorithms imply a defined if not mathematical process, yet lumbar pain generators are often multifactorial and elusive. It is actually the clandestine nature of the lumbar axial pain generator and the often speculative diagnoses assigned to patients with persistent lumbar pain which warrant a more methodological diagnostic and therapeutic approach. This chapter is written in the context of an interventional spine text. The minority of patients with lumbar pain will require an interventional approach. While remarkably prevalent, most episodes of debilitating axial pain will prove short lived. The majority of patients will respond to a period of activity modification. Some will participate in a structured rehabilitation program and utilize oral analgesics or antiinflammatory agents in order to realize relief. It is the relative minority of patients whose axial pain persists in a debilitating fashion who are candidates for an interventional approach. In such cases, a more meticulous history, examination, and image review can offer guidelines for the judicious use of diagnostic and therapeutic spinal injection procedures. Spinal pain generators are typically not definitively identified by history, highlighted by imaging, or reliably provoked during physical examination. While it is always the goal to treat in a less interventional fashion, in the case of persistent lumbar axial pain, diagnostic injection techniques offer an opportunity to identify pain generators in this patient population where diagnoses might otherwise remain ill defined. Offering these challenging patients further therapeutic interventions before exhausting a more algorithmic diagnostic approach might lead the patient along a less appropriate and ineffective treatment pathway.

A general algorithmic approach to the patient with axial pain can be proposed, but diagnostic and therapeutic pathways will vary depending upon the pain generator in question. Each interventional algorithm assumes that symptoms have first persisted for a reasonable period of time, i.e. 4–6 weeks. The algorithm employed should be logical and evidence based but needs to remain practical and feasible in the daily clinical setting and current medical environment. At each step in the assessment and treatment pathway, the patient should remain center stage, informed, and encouraged to play an active role in the decision-making process. Prior to diagnostic or therapeutic interventions, appropriate advanced imaging, and blood work-up if necessary, is performed to rule out a more concerning and underlying pathologic process. In this author’s practice, a positive response to a diagnostic injection requires a quantitative pain relief response of at least 80%. During the postinjection assessment, the patient’s symptom response to typically provocative postures and maneuvers is assessed. In the patient with a suspected painful intervertebral disc, lumbar discography serves as a provocative diagnostic measure but is more invasive than the commonly employed therapeutic injection approach. In the disc algorithm, more definitive diagnostic testing is therefore uniquely deferred in favor of a more presumptive therapeutic approach in an effort to minimize the extent of spinal interventions.

Once a pain generator is confirmed through completion of the diagnostic component of the algorithm, therapeutic approaches should also be considered in a logical fashion. Arguably, one’s ability to identify the origin of pain through an interventional approach, which in itself remains inexact and incomplete, remains superior and less controversial than the efficacy of the therapeutic options available to the interventionist. Less interventional treatments, defined as those which strive to reduce pain and inflammation, but maintain tissue integrity, should be offered prior to those that denervate, ablate, modulate tissue, resect or alter regional anatomy and function. When therapeutic corticosteroid injections are performed, it is the author’s practice to schedule two injections 2 weeks apart. If the first injection offers complete relief, the second is deferred. If no relief is offered after two injections, injection therapy is determined likely to be ineffective. If a component of lasting, but incomplete relief is described 2 weeks after the second injection, a third is offered. A fourth and final injection is only considered for those patients who describe an incremental and sustained response to the initial three. Injection series are ideally not repeated, but if offered in the future, this does not transpire until 6–12 months after the initial series. This number of injection procedures described finds its basis in part in a review of the literature describing the role of therapeutic epidural injections in the treatment of lumbosacral radiculopathy.¹ With an increasing number of injections performed, declining therapeutic efficacy and potential local and systemic adverse corticosteroid effect should raise greater concerns.² As more interventional and less reversible therapeutic approaches are entertained, additional and confirmatory diagnostics can be utilized to assure accurate identification of the pain generator. For those patients who ultimately fail to realize relief following an algorithmic approach or who are determined to be poor surgical candidates, a paradigm better related to chronic pain modulation is proposed. The chronic pain program can employ the use of analgesics, alternative maintenance therapies such as acupuncture and biofeedback, pyschological care, and, in select cases, a consideration of implantable pain modulating devices (Fig. 89.1).

Fig. 89.1 General algorithmic approach to chronic axial lumbar pain.

Ultimately, the patient with persistent and debilitating axial pain presents the spine specialist with unique and formidable bookend challenges. The diagnostic algorithm can prove multifaceted with initially suspected pain generators ultimately found to be inert. The therapeutic arm of the axial pain algorithm remains hindered by a general lack of more conclusively supportive and well-designed outcome studies. The common disorder of persistent axial pain continues to challenge the many disciplines of the spine care community, and the multitude of current studies examining therapeutic approaches reflects the ongoing search for more definitive treatments. This chapter will address several syndromes, which are the topic of other dedicated chapters in this text. In these dedicated chapters, meticulous descriptions of procedural techniques and a more complete review of the pertinent literature have been included. In each of the following sections, background information pertaining to the pain generator of interest will be provided first. The significance of radiographic studies, patient history, and physical examination findings will then be discussed with an inclusion of supportive literature and the author’s practice in each realm. A diagnostic and therapeutic algorithm will then be presented in a logical and sequential fashion and summarized in figure form. The chapter will conclude with a mention of axial pain of potential radicular origin, more malignant processes which can lead to axial pain, and finally, two exemplary cases in which an algorithmic approach is employed.

SACROILIAC JOINT

The sacroiliac joint (SIJ) will be addressed first, as sacroiliac joint syndrome (SIJS) remains a commonly assigned diagnosis for the patient with axial pain. Prior to Mixter and Barr’s 1934 study highlighting pathology of the intervertebral disc,³ the SIJ was believed to represent the most likely pain generator in patients with low back pain. As with the other anatomic structures addressed in this chapter, the SIJ remains a viable candidate for a primary pain generator as it satisfies several criteria outlined elsewhere: first, the structure should have a nerve supply; second, it should be susceptible to diseases or injuries known to be painful; and third, it should be capable of causing pain similar to that seen clinically.⁴ The complex and variable innervation of the SIJ has been described in many anatomic studies.^4–6 Painful SIJ conditions are known to arise from spondylarthropathies,⁷ infection,⁸ trauma,^9,10 and malignancy.¹¹ Intra-articular injections, performed both provocatively in asymptomatics¹² and diagnostically in patients with chronic lumbar pain,^13,14 have demonstrated the SIJ to be a potential source of pain. The prevalence of SIJS in the population with chronic low back pain has been estimated at 13–30%.^13,14 SIJS, which remains a controversial diagnosis among spine practitioners, has been hypothesized to develop from degenerative change affecting the joint or altered joint mobility. The SIJ is mobile, albeit limited to 3° of rotation and only a few millimeters of glide.^15,16

Radiologic or surgical pathology has not been identified in patients with SIJS, further raising pathophysiologic speculation. Bone scan has demonstrated a poor sensitivity (12.9%) in patients demonstrating a positive response to diagnostic intra-articular injections.¹⁷ Bone scan has demonstrated a sensitivity in patients with SIJS which is significantly lower than that observed when utilized to detect sacroiliitis in patients with rheumatological disease.^17–20 These findings do not exclude the possibility that SIJS does not involve more mild synovial irritation which remains undetected by radionuclide imaging which relies upon an acceleration of osteoblastic activity.¹⁷ While magnetic resonance imaging (MRI) has demonstrated a high sensitivity in cases of true sacroiliitis,^18,19 the role of MRI in diagnosing SIJS has also yet to be defined. At least 24.5% of asymptomatics >50 years of age demonstrate degenerative SIJ changes on plain radiographs,^21,22 and similar degenerative findings in aging asymptomatics have been demonstrated utilizing computed tomography (CT) imaging.²³ Similarly, no specific pathology has been described in patients undergoing surgical intervention for chronic sacroiliac joint pain.²⁴

Sacroiliac joint pain referral patterns often localize to the sacral sulcus and buttock,¹² but patients can also describe symptom referral to the medial thigh and groin,^13,25–28 posterior thigh, and calf.^9,13,29–31 Physical exam maneuvers utilized to detect motion irregularities have demonstrated poor intertester reliability^32,33 and have been observed to be positive in 20% of asymptomatics.³⁴ The detection of joint motion abnormalities by physical examination,²⁹ response to pain provocation tests such as Faber’s and Gaenslen’s maneuvers,^17,29,35 and historical findings^13,29 have all correlated poorly with the response to fluoroscopically guided diagnostic intra-articular injections, which have come to be recognized as the gold standard for diagnosing SIJ pain. As specific history, examination, and imaging findings have all correlated poorly with a positive response to a diagnostic intra-articular injection, it is more likely a combination of presenting factors which might lead the treating clinician to suspect the SIJ as an active pain generator and warrant introduction into the SIJ diagnostic and therapeutic algorithm. Suggestive symptoms include pain predominantly below the L5 level in the region of the sacral sulcus, with or without referral to the groin¹³ or more distal lower limb, and sacral sulcus tenderness to palpation.^29,35,36 Additionally, patients with unilateral symptoms and a positive response to multiple provocative maneuvers may be more likely to demonstrate a positive diagnostic injection response and subsequently benefit from targeted therapeutic approaches.^29,35,36 While a previous history of a fall upon the affected buttock, recent pregnancy, pelvic trauma, or an antecedent gait alteration³⁷ also prompt this author to consider the SIJ as a pain generator, these historical points have not been demonstrated to be predictive by the available literature.

For those patients who are believed to be reasonable candidates, an algorithmic approach to the SIJ (Fig. 89.2) can be initiated and a confirmatory diagnostic SIJ injection performed. Diagnostic intra-articular injections were initially described in 1938,³⁸ and the utilization of fluoroscopic guidance was first introduced in 1979.³⁹ It has been estimated that successful intra-articular SIJ entry is achieved in only 22% of injections performed without image guidance,⁴⁰ and such blind injections performed for either diagnostic or therapeutic purposes are not included in an algorithmic approach to SIJS. Given the complexity of the anatomy and configuration of the SIJ, it would seem that this 22% figure may even be an overestimate, further amplifying the requirement of using fluoroscopic guidance. If a positive response to a guided diagnostic injection is realized, therapeutic injections should follow. Therapeutic SIJ corticosteroid injections have been advocated as an appropriate treatment for patients with persistent SIJ pain.⁴¹ A retrospective and uncontrolled study of SIJ injections in 31 patients with chronic pain of SIJ origin has suggested a lasting resultant improvement in pain, work status, and disability.⁴² The efficacy of these injections has been more clearly demonstrated, both prospectively and retrospectively and in a controlled fashion, in the seronegative spondyloarthropathy population.^43,44 The mechanism of action of injections in patients with presumed SIJS outside of the spondyloarthropathy population remains less clear, but is presumed to arise from the well-known antiinflammatory properties of corticosteroids.^28,45 The role of inflammation in SIJS remains unconfirmed. The role of injection therapy in this patient population therefore remains poorly defined. Injection therapy can be combined with a targeted physical therapy regimen which emphasizes pelvic stabilization techniques and SIJ-specific therapies.

Fig. 89.2 General algorithmic approach to the sacroiliac joint.

For those patients who fail to respond to therapeutic injections, a more recently described and evolving radiofrequency denervation approach to the SIJ might then be considered. The nerves which can be targeted for this approach include the L5 dorsal ramus and the lateral branches of S1–3.⁴⁶ The ability to reliably target and anesthetize these nerves with fluoroscopic guidance remains less clear.⁴⁷ In those patients who demonstrate a positive diagnostic response to anesthetization of these innervating branches or who demonstrate a concordant response⁴⁸ with lateral branch stimulation, a denervation procedure can be trialed. The ability of such a denervation approach to successfully anesthetize the SIJ remains less clear with only small and uncontrolled studies suggesting a significant therapeutic response.^49–52

For those patients with persistently debilitating SIJ pain who either fail to respond to denervation techniques or who demonstrate an initial negative response to L5 dorsal ramus and sacral lateral branch anesthetization, surgical arthrodesis of the SIJ can be considered as a final treatment option.^50,53 In selecting candidates for SIJ fusion, those patients who demonstrated both a positive lateral branch response and an initial diagnostic intra-articular injection may be considered less likely to be false-positive SIJ patients. In these cases, it would appear that a confirmatory diagnostic injection has been employed to lessen the likelihood of an initial placebo response. As the approach to and efficacy of lateral branch anesthetization remains less well defined, any patient contemplating SIJ fusion should first demonstrate a second positive response to a confirmatory diagnostic intra-articular injection utilizing an anesthetic of different duration or a blind and placebo-controlled diagnostic injection protocol. As well, the candidacy of such individuals for surgical intervention might be further strengthened by a negative response to provocation discography.

ZYGAPOPHYSEAL JOINT

The zygapophyseal joints (Z-joints) have been recognized as a potential source of lumbar pain since 1911.⁵⁴ The lumbar Z-joints are paired synovial joints with an intra-articular volume capacity of 1–2 mm.⁵⁵ While the more cephalad lumbar Z-joint orientation tends to be in the sagittal plane, the lower joints are more coronally situated.⁵⁶ Each lumbar Z-joint is innervated by the medial branch of the dorsal ramus at the level of the joint and by the medial branch arising from the dorsal ramus of the next cephalad level.⁵⁷ In 1976, Mooney and Robertson⁵⁸ demonstrated that lumbar axial pain and symptoms referred to the extremity could arise from intra-articular injections of normal saline in asymptomatic volunteers. McCall et al.⁵⁹ later corroborated these findings and observed a more intense pain response with injection into the capsular tissues when compared with injection into the joint’s intra-articular space. Utilizing progressive local anesthesia during surgery, Kuslich et al.⁶⁰ demonstrated a pain response with stimulation of the Z-joint capsule, but the pain described by patients was often not concordant with their more debilitating axial pain. Z-joint pain, or ‘facet syndrome,’ is presumed to arise from osteoarthritic change, chondromalacia, or occult fractures.^61–66 Less common conditions affecting the Z-joints including infection, ankylosing spondylitis, and villonodular synovitis have also been reported.^67,68 Lumbar Z-joint fractures, capsular tears, hemorrhage, and cartilaginous injury have been observed in postmortem studies of trauma patients with normal radiographs.⁶⁶ A study of 176 consecutive patients presenting with chronic lumbar pain, in which lidocaine and confirmatory bupivacaine medial branch or intra-articular diagnostic injections were employed, suggests that the Z-joint represents the primary pain generator in 15% of cases.⁶⁹ Plain radiographs often reveal degenerative arthrosis of the Z-joints which strongly correlates with age, but not with symptoms of axial pain.^70,71 Utilizing diagnostic intra-articular injections, degenerative Z-joint change detected by CT imaging has also demonstrated a poor correlation with pain arising from the lumbar Z-joints.⁷²

Several studies^73–75 have specifically investigated the historical and physical examination findings which might predict the Z-joint as a pain generator. In one of these studies⁷⁴ a second and confirmatory diagnostic Z-joint injection was included to establish a diagnosis, and in two^73,75 a positive response to only a single diagnostic injection was utilized. Patients with confirmed primary Z-joint pain were noted to describe lumbar discomfort, but pain could similarly be referred to the lower limb. Patients generally did not describe central lumbar pain. While no particular symptoms or historical findings demonstrated a significant correlation with a positive response to diagnostic injections, patients with Z-joint pain tended to be older, without exacerbations during coughing, and without described provocation with forward flexed postures. The L5–S1 Z-joints were found to be more likely symptomatic than L4–5, with pain arising from the L3–4 and L2–3 joints much less commonly observed.⁶⁹ While eliciting a concordant pain response by direct articular palpation has been described as a potential screening mechanism in patients with suspected Z-joint pain,⁶¹ the utility of this examination technique was not clearly substantiated by the aforementioned studies.

Deciding which patients to introduce into the Z-joint algorithm presents a challenge to the treating clinician quite similar to that described in patients with SIJS. Radiographic evidence of degenerative change is typically not a reliable indicator, and historical and physical examination findings do not clearly correlate. As in the case of SIJS, it is neither practical or economical to subject each patient with chronic axial pain to a confirmatory diagnostic injection without a reasonable clinical suspicion. Older patients, i.e. greater than 60 years of age, presenting with uni- or bilateral axial lumbosacral pain, and with an absence of provocation during flexion and coughing, may be more likely to demonstrate pain of Z-joint origin. In the author’s practice, a Z-joint pain generator is more suspect in older patients who describe pain which is most reliably provoked by standing, ambulating, or ipsilateral extension, and relieved by sitting. These latter historical features are not supported as clearly reliable by the available literature.

In the patient with suspected Z-joint pain, an algorithmic approach can be initiated (Fig. 89.3) and a confirmatory diagnostic intra-articular injection performed.^69,74 There is no standard protocol for the selection of joints to anesthetize. It has been suggested that the joints with maximal tenderness during palpation can be marked and identified under fluoroscopic inspection. Alternatively, in patients with unilateral lumbosacral pain, the L4–5 and L5–S1 joints can be investigated. If more isolated posterior element arthrosis is observed radiographically, a single joint can be studied. In patients with bilateral lumbosacral pain, these joints can be bilaterally addressed in a diagnostic fashion. If the diagnostic response from the L4–5 and L5–S1 injections is negative and a midlumbar pain complaint is reported, the L2–3 and L3–4 joints can then be studied in a similar coupled and unilateral or bilateral fashion. If the patient presents with midlumbar rather than lumbosacral pain, the L2–3 and L3–4 joints might be studied first. In the author’s practice, while a radiographic correlate to Z-joint pain has yet to be defined, MRI or CT scan evidence of more advanced arthrosis is also considered in selecting joints for further study when these imaging findings correspond to the patient’s pain location. Following a positive diagnostic response, therapeutic intra-articular injections could follow.

Fig. 89.3 General algorithmic approach to the zygapophyseal joint.

Similar to the treatment of SIJS, the therapeutic response arising from intra-articular injection with corticosteroid has not been more definitively revealed by controlled studies, and an inflammatory injury component has only been theorized. While intra-articular effusions can be observed on MRI, histologic studies have not revealed inflammatory cells in patients with spondylotic joints.⁶¹ Similarly, clinical studies have yet to identify a therapeutic effect from intra-articular injections in patients with inflammatory rheumatologic spinal conditions.⁶¹ A randomized, controlled study investigating the therapeutic benefits of intra-articular methylprednisolone, utilizing an initial single diagnostic injection screen and an intra-articular saline control, have suggested up to 46% of patients can realize significant pain relief at 6 months’ follow-up.⁷⁶ Open and uncontrolled studies of fluoroscopically guided intra-articular injections suggest an 18–63% therapeutic effect for more than 6 months following injection.^61,77–80 Pain relief realized following intra-articular injection might provide a window of opportunity to progress the patient in a spine rehabilitation program and graduate mechanical and manual therapies. If the patient failed to realize relief following therapeutic intra-articular injections or spondylotic anatomy prevented the performance of a diagnostic intra-articular injection, a diagnostic medial branch block⁸¹ could be utilized to confirm a diagnosis of Z-joint-mediated pain.

In the setting of a positive response to medial branch block, radiofrequency neurotomy can be considered. In the patient who initially demonstrated a positive response to an intra-articular injection, the medial branch block can serve as a confirmatory diagnostic measure. For patients in whom a diagnostic intra-articular injection was not previously performed, a positive response to medial branch block should be confirmed with an additional diagnostic screen to address the estimated 38% false-positive response rate with single uncontrolled blocks.⁷⁴ A second and confirmatory medial branch block can be performed with a local anesthetic of different duration of action. In this scenario, the patient’s analgesic response with each test should correlate with the duration of action of the anesthetic utilized. Alternatively, the confirmatory injection can be performed with an informed patient in a blind fashion with either a saline placebo or anesthetic injected and the patient monitored for an appropriate response. The difficulty with employing local anesthetics of different duration is that the patient is required to accurately record and communicate their pain response over a period of hours which extends beyond the time spent in the office of the evaluating team. Additional response interpretation difficulties can arise in the patient who describes marked symptomatic relief after the injection of the longer-acting anesthetic, but whose response does not last for as long as one would predict based upon the agent’s half-life. To avoid these confounding factors, only shorter-acting anesthetics can be utilized and the response assessed in the office setting by an independent observer.

Patients who demonstrate a positive response to either variant of this double diagnostic screen can then be considered for medial branch radiofrequency denervation. While radiofrequency denervation represents a minimally invasive approach to the patient with chronic posterior element pain and localization of the innervating medial branches appears to be reliable, outcome studies are limited. Modest therapeutic responses have been described, uncontrolled small patient populations have been studied,⁸² and such procedures likely need to be repeated to offer sustained relief as reinervation to the painful joints can occur. A double-blind, controlled study of radiofrequency denervation versus sham lesions which employed a single diagnostic injection screen revealed only a two-point reduction in visual analog pain scores with the minority of patients realizing complete relief.⁸³ A meticulously designed but small and uncontrolled study of medial branch neurotomy, which included electromyographic assessment of the paraspinal musculature to confirm denervation, demonstrated pronounced relief in 60% (n=9 of 460 initially screened) of patients at 12 month follow-up evaluation.⁸² If the denervation procedure proves helpful but the symptomatic response wanes, neurotomy can be repeated. A significant response should first be realized for at least 6–12 months.⁸²

For those individuals with persistent axial pain of Z-joint origin which fails to improve despite the aforementioned therapies, surgical intervention in the form of posterolateral fusion should be considered only with great caution. While studies have yet to address the success of this approach in patients with Z-joint pain confirmed with a double diagnostic screen, the available literature does not suggest a correlation between posterior element pain and successful outcomes following fusion.^84–86 While it may be tempting to assume that posterior element pain can be relieved by fusion, this has yet to be demonstrated. The potential remains for the fused joint to serve as an ongoing pain generator, as its painful tissue components remain after a typical posterior fusion approach. The possibility also remains that forces continue to be borne, albeit reduced, by the painful posterior elements following posterior stabilization. Finally, as mentioned in the discussion of SIJS, for any patient considered for fusion for Z-joint pain, preoperative lumbar discography, which will be further addressed in the next section of this chapter, should also be considered to rule out a contributing discogenic pain source. Discography also allows the clinician to better evaluate the anatomy and symptom response from adjacent levels. While a positive response to a double diagnostic screen would appear confirmatory for a primary Z-joint pain generator in these chronic cases, the possibility remains, potentially in a minority, that a concordant pain response will similarly be demonstrated during discography.⁸⁷

INTERVERTEBRAL DISC

While the sacroiliac joint and Z-joints are suspected as primary pain generators in a significant but relative minority of patients with persistent axial lumbosacral pain, the intervertebral disc has stood center stage as the most commonly suspected pain generator in patients with both debilitating chronic and more acute axial pain. Mixter and Barr’s 1934 publication described herniated lumbar discs and their relationship to nerve root compressive syndromes.³ In addition to inducing radicular complaints affecting the lower limb, the degenerative disc is also believed to be a common primary pain generator in patients with axial pain. In a study of 92 consecutive patients presenting with chronic lumbar pain, provocative discography revealed a concordant pain response in 39%, most commonly at L4–5 and L5–S1.⁸⁸ In a unique study utilizing progressive local anesthesia and selective tissue stimulation intraoperatively, Kuslich et al. demonstrated that while stimulation of the nerve root typically resulted in buttock and lower limb pain, stimulation of the disc, and in particular the anulus, most commonly resulted in a reproduction of the patient’s lumbar pain in a concordant fashion.⁶⁰

The innervation of the intervertebral disc has been well defined. Abundant nerve endings with a variety of free and complex terminals have been identified in the outer third to half of the anulus fibrosus.^89–91 The posterior plexus of nerves responsible for innervating the posterior anulus and posterior longitudinal ligament (PLL) is derived predominantly from the sinuvertebral nerve. The sinuvertebral nerve arises from both the somatic ventral ramus and autonomic gray ramus communicans and also supplies innervation to the ventral dura mater.⁹²