Lumbosacral Disease

Anatomy of the Lumbosacral Spine

The structural anatomy of the lumbosacral spine includes the intervertebral disk, the articular facet joints, and spinal ligaments. Detailed descriptions of these structures are discussed in Intervertebral Disk Disease. The last lumbar vertebra (L7) attaches to the fused three-segment (S1-3) sacrum rather than to another individual vertebra (Fig. 50-1). The range of motion of the sacrum is limited by its attachment to the pelvis. This anatomic difference leads to different biomechanical loading forces at this level in comparison with the other lumbar vertebra [1,2]. The characteristic that renders the lumbosacral joint unique clinically is that, in most instances in the dog and cat, the spinal cord itself does not extend into this region. At birth the spinal cord extends through the vertebral canal of the sacrum [3]. Differential growth rates between the vertebral column and spinal cord result in termination of the spinal cord around the L6 vertebra in medium-to-large breed dogs and at L7 cats (Fig. 50-2) [3]. Small-breed dogs (< 7 kg) may have longer spinal cords in which case the spinal cord may terminate as far caudally as the lumbosacral junction [3]. Caudal to the lumbar intumescence, the spinal cord tapers to an elongated cone, or the conus medullaris, which contains the sacral and caudal spinal cord segments [3]. The spinal cord itself terminates as the filum terminale, which is a band composed of glial and ependymal cells [3]. The filum terminale unites with the dura and arachnoid layers to form the caudal ligament around the S1 vertebra. The caudal ligament anchors the spinal cord by fusing with the periosteum and dorsal longitudinal ligament in the caudal vertebrae. The thecal sac and arachnoid space, which contain cerebrospinal fluid, extend beyond the filum terminale up to 2 cm and may extend as far as the S1 vertebra [2]. Because the spinal cord terminates before the lumbosacral junction, the caudal lumbar, sacral, and caudal nerve roots travel the remaining distance to exit the spinal canal at the appropriate intervertebral foramen. This collection of the nerve roots is referred to as the cauda equina [3]. The dorsal and ventral nerve roots fuse distal to the dorsal root ganglion near the intervertebral foramen to form the spinal nerves. The spinal nerves carry the axonal fibers that form the sciatic, pudendal, parasympathetic, and caudal nerves (Fig. 50-3) [3]. Throughout the thoracic and lumbar spinal cord, spinal nerves exit caudally to the vertebra of the same number (e.g., L7 nerve root exits at the L7-S1 intervertebral foramen) just cranially to the intervertebral disk. Therefore, with lumbosacral disease, typically the last lumbar and sacral roots are most clinically affected. However, degenerative disease at L6-7 can occur independently or in conjunction with lumbosacral disease. The nerve roots, like the spinal cord, are covered by epidural fat which acts as a protective layer against injury. Paired venous sinuses, present the length of the spinal canal, diverge laterally over the disk spaces and converge toward the midline as they reach the mid-body of the vertebrae.

Figure 50.1. A schematic drawing of the anatomy of the lumbosacral junction. (From Miller ME, Christensen GC, and Evans HE: Anatomy of the Dog. Philadelphia: WB Saunders, 1964. Drawn by M. Newsom).

Figure 50.2. Spinal cord segmental relationship to the vertebral bodies. From T11 through the caudal segments the spinal cord, roots, ganglia, and nerves have been exposed by removal of the vertebral arches. The dura mater has been removed except on the right side. The numbers on the right represent the levels of the vertebral bodies. (From Miller ME, Christensen GC, and Evans HE: Anatomy of the Dog. Philadelphia: WB Saunders, 1964. Drawn by M. Newsom).

Figure 50.3. Peripheral nerve formation from the spinal cord segments of the lumbar intumescence in the dog. (Modified from De Lahunta A: Veterinary Neuroanatomy and Clinical Neurology, 2nd ed. Philadelphia: WB Saunders, 1983).

Two adjacent vertebral bodies and the intervening intervertebral disk constitute a motion segment. In the vertebral motion segment are three joints: two articular facets and the intervertebral disk. Motion is constrained by associated joint capsules and ligaments [4]. As these structures reach the limits of their elasticity and tensile forces are encountered, flexibility is reduced [1]. Movement in the joint is primarily flexion with more restricted degrees of rotation, lateral bending, and extension [1,5]. Flexion in the sagittal plane is limited by the supraspinous, intraspinous, interarcuate, and dorsal longitudinal ligaments and the articular facet joint capsules [1,5]. Extension is restricted by the ventral longitudinal ligament and articular facet joint capsules [1,5]. Lateral bending is controlled by the articular facet joint capsules, the intertransverse ligaments, the annulus fibrosus of the intervertebral disk, and to some degree, by the pelvis [1,5]. Finally, axial rotation is controlled by the articular facets and annulus fibrosus of the intervertebral disk [1,4].

Gross Pathology of Lumbosacral Disease

The most common cause of cauda equina pathology is stenosis, or narrowing of the spinal canal, resulting in compression of the cauda equina (Fig. 50-4) [2,4-8]. Although stenosis can be congenital in nature, it is often secondary to degenerative lumbosacral disease, which encompasses multiple changes in the structural anatomy at this location. Most often, these changes affect the intervertebral disk, the articular facets, and the ligaments of the spine. Congenital abnormalities, such as instability, transitional vertebrae, and osteochondrosis of the sacral or L7 endplate [8,9], as well as trauma are also associated with lumbosacral disease. These abnormalities usually result in acquired secondary degenerative changes, which ultimately contribute to clinical signs at a later time [2,4,5,7,8,10]. True congenital lumbosacral spinal stenosis may present early with clinical signs prior to development of degenerative changes if the canal diameter is narrowed severely enough. However, congenital abnormalities and prior trauma are not required for degenerative changes to occur at this level and are certainly not as common as aging and degenerative changes. Other diseases that affect the lumbosacral spine and cauda equina are neoplasia, trauma (fracture), inflammation, infection, and vascular injuries. In this chapter, the focus is on degenerative lumbosacral disease. The anatomic pathology associated with degenerative lumbosacral disease is found in Table 50-1.

Figure 50.4. MRI images of a dog with a normal lumbosacral junction (A and B) and from a dog with degenerative lumbosacral disease (C and D). A. T2-weighted sagittal plane MRI. All of the intervertebral disks are well hydrated. The neural tissue and fat are easily identified in the lower lumbar and sacral spinal canal. White arrow = L7-S1 disk B. T2-weighted transverse plane MRI through the L7-S1 joint. Thin arrow = spinal canal containing the dural tube, nerve roots and epidural fat. Block arrows = L7 nerve roots exiting the L7-S1 intervertebral foramen. C. T2-weighted, sagittal plane MRI. There is a large disk extrusion at the L7-S1 disk and the canal is severely stenotic. (black arrow) Multiple degenerative intervertebral disks (noted by loss of signal owing to dehydration) are at L3-4, L4-5, and L7-S1. Ventral spondylosis is noted at the lumbosacral joint (white arrow) and other disk spaces as well. D. T2-weighted, transverse plane MRI through the L7-S1 disk space. The spinal canal is obliterated with extruded disk material (white arrow), and the neural anatomy is difficult to appreciate. IL = wing of the ilium; IVD = intervertebral disk; * = L7-S1 facet joint.

The lumbosacral joint is one that transitions between a more freely flexible region of the spine (lumbar segments) and one that is static in position (sacrum) because of the attachment to the pelvis. The range of motion is higher at this location than at any of the other lower lumbar motion segments [11]. It is not surprising that this joint can undergo excessive biomechanical stresses. It is postulated that the disease is most prevalent in large breed dogs owing to an imbalance between body weight and dimension of the lumbosacral contact area not seen in smaller breeds and resulting in higher loading forces around the joint [12]. Spondylosis and osteophyte production around the articular facet joints commonly observed with degenerative lumbosacral disease may be one way of reducing the loading forces [12,13]. German shepherds, which have a high reported incidence of the disease, have been shown to have hypermobility and increased longitudinal extension [8,12].

One of the earliest changes occurs in the intervertebral disk and is related to type II degeneration. An extensive review of type II intervertebral disk disease can be found in Intervertebral Disk Disease. As the disk degenerates at the lumbosacral joint, derangements occur in the normal load-bearing functions. As the disk begins to herniate or collapse, the normal disk height is lost, the facet joints begin to share more than their normal biomechanical load [8,14], and subluxate to some extent [15]. Although collapse of the disk results in foramen narrowing [5,16], the significance of this is not known. The clinical significance of foraminal narrowing without spinal canal stenosis has been challenged in the human literature [17]. As in other diarthrodial joints, development of periarticular osteophytes occurs in response to excessive biomechanical stressors [18]. Osteophytes often form on the ventromedial side of the facet and the dorsolateral endplate of L7 [18-20]. They may severely restrict the foramen and L7 root, especially if the joint is extended [14]. In addition, changes in the function of the motion segment may cause interarcuate ligament hypertrophy and infolding into the spinal canal, which may contribute to stenosis and compression of the nerve roots dorsally [4,5,21,22]. Ventral spondylosis is a typical finding; however, it is most often an indicator of degenerative changes rather than a true cause of clinical signs.

Congenital anomalies tend to accentuate the effects of excessive biomechanical loads and contribute to lumbosacral degenerative changes. Congenital instability is typically related to malformation and malarticulation of the facets and abnormal vertebral alignment. The response to instability at this joint can be likened to that seen with caudal cervical malformation/malarticulation ("Wobbler's disease"). Osteochondrosis is the most recently reported congenital anomaly that contributes to degenerative changes at the lumbosacral joint. The cartilaginous endplate defect is more commonly seen in the cranial sacral endplate than the caudal lumbar endplate [9]. Not surprisingly, this anomaly has been most commonly observed in German shepherd dogs, with a higher male: female ratio [9]. It is often seen clinically in dogs younger (as early as 18 months) than the typical patient presenting with lumbosacral disease [9]. Degenerative changes at the lumbosacral joint may be attributable to an increase in abnormal stresses and possibly impairment of intervertebral disk nutrition at an age earlier than that of typical type II intervertebral disk degeneration.

The term transitional vertebrae encompasses a variety of congenital osseous abnormalities seen in vertebrae between the spinal segments. Most commonly, these changes have been seen in the lumbosacral motion segment [23]. This abnormality displays morphologic characteristics of both lumbar and sacral vertebrae [24-26]. Normally, the development and alignment of the last lumbar vertebrae are such that the transverse processes are not in contact with the ilia [23]. Abnormal development arises through abnormal mechanisms during growth in the ossification centers of the sacrum and/or L7 vertebrae [23]. The most distinct abnormalities usually occur in the transverse processes, which may or may not be attached to the ilium or sacrum [23,24,26]. Variations in the morphology of the vertebral body are less common [23-26]. These abnormalities may result in abnormal sacroiliac attachments (usually shortening of the attachment on the affected side) and abnormal rotation of the pelvis [23]. The abnormal anatomic alignment likely results in altered biomechanics and excessive loading on the intervertebral disk and facet joints. Transitional vertebrae may, therefore, be clinically relevant to degenerative lumbosacral disease, especially in German shepherd dogs, which have a high reported incidence of degenerative lumbosacral disease [23,24,26-28].

Transitional vertebrae in people are more likely to be associated with degenerative disk disease and facet joint degeneration at the joint above the transitional vertebra, and this malformation is somewhat protective against degenerative changes to the disk below [29]. The joint between the transverse process of the transitional vertebra and the sacrum may be restrictive to rotation and bending, thus protecting the lumbosacral disk (L5-S1 disk in people) from excessive biomechanical forces [29,30]. The disk height below the transitional vertebra is found to be shorter than disc heights of the upper levels in young men, but the disk height is similar in people with this condition in comparison with that in normal men in the middle-aged population [29]. The shortened disk height in the unaffected older population is presumably a result of degeneration [29,31]. The protective effect of the transitional vertebra on the disk below is stronger for the annulus fibrosus than the endplate and nucleus pulposus but it is postulated that the degenerative changes seen in these structures may be induced by another mechanism, i.e., natural aging of the disk [29].

Clinical Signs of Lumbosacral Disease

Clinical signs of lumbosacral spinal disease are differentiated from injury to the lumbosacral spinal cord segments. Because of the neurologic anatomy at the lumbosacral joint, the L7 and the sacral and caudal nerve roots are typically affected rather than the spinal cord by pathology at the lumbosacral junction. Compression of the cauda equina is most often associated with pain, gait changes, and mild-to-moderate degrees of neurologic dysfunction compared with injury to the lumbosacral spinal cord segments that may be more complete or severe. Table 50-2 is a list of the possible findings during the neurologic examination in animals with cauda equina syndrome. It is important to note that not all of the deficits listed in the table are necessarily present in any one patient.

The most common disease entity affecting the cauda equina is stenosis, or narrowing of the spinal canal, which results in compression of the nerve roots [2,4-7]. Dogs often display signs of lumbosacral pain observed on deep lumbosacral palpation with or without hip extension, dorsal elevation of the tail base, rectal palpation (pressing dorsally), and pelvic limb wheelbarrowing. Clients often comment on lower tail carriage or lack of tail wagging. Pain associated with lumbosacral disease may originate from the disk (discogenic pain), nerve root and inflamed articular facet capsules, and spinal ligaments. Radicular, or nerve root, pain may be manifested as allodynia, dysesthesia, hyperesthesia, or paresthesia [32]. Definitions of these pain syndromes are shown in Table 50-3. The pathogenesis of pain associated with lumbosacral disease will be discussed further.

It is critical to rule out the presence of coxofemoral osteoarthritis as a source of lumbar area pain, which can also be present concurrently. Lumbosacral pain often leads to lameness, which is either intermittent or persistent, and a bunny hopping gait. These gait changes are also seen with coxofemoral disease. Difficulty rising, exercise intolerance, and inability to perform tasks that were once normal activities may be attributed to both pain and/or weakness.

Neurologic examination may be normal in an animal with only lumbosacral pain, or, if the results are abnormal, the examination may reveal pain and mild deficits in postural reactions and spinal reflexes. Earlier recognition of signs, improved health care for pets, and finally, improvements in diagnostic modalities and therapies may contribute to the lower incidence of significant neurologic deficits seen by the author. However, the disease may result in variable degrees of dysfunction in the postural reactions and reflexes innervated by the sciatic, sacral, and caudal nerves. These include deficits in conscious proprioception, hopping reactions, decreased-to-absent spinal reflexes, urinary and/or fecal incontinence, and abnormal tail and perineal function or sensation.

The patellar reflex is spared because the nerve roots that supply the femoral nerve (L4-L6) have exited the spine rostral to the lumbosacral joint. Paraplegia is rare because when femoral nerve function is intact, the pelvic limb can extend and bear weight. In cases of severe sciatic nerve dysfunction, the animal may be very weak, but not usually non ambulatory. When the sciatic nerve is extremely dysfunctional, the patellar reflex may appear to be exaggerated. This is called "pseudohyperreflexia" because it is not a true "upper motor neuron" sign, but occurs secondary to loss of antagonist muscle action in the biceps femoris, semitendinosus, semimembranosus, and gastrocnemius muscles. Typically, this "false" reflex is observed as a brisk extension of the limb with a poor rebound of flexion of the hock. True hyperreflexia is associated with clonus of the limb.

Occasionally, the single or most severe clinical sign of lumbosacral disease is loss of autonomic function, such as urinary and/or fecal incontinence. The function of the bladder is impaired by disruption of the S1-3 parasympathetic nerve fibers that form the pelvic nerve and control function of the detrusor muscle of the urinary bladder. The somatic fibers that are present in the S1-3 spinal nerves control the external urinary sphincter, or urethralis muscle. In addition, the sensory afferent fibers are also affected. The sensory afferent fibers send information to the sacral spinal cord that facilitates local reflex activity. Ascending information from these fibers also converges on the sympathetic preganglionic nerve cell bodies in the thoracolumbar spinal cord (around L1-4) which form the hypogastric nerve that innervates the internal urinary sphincter. The varying degree of loss of ascending information to these cell bodies may result in inefficient internal urinary sphincter tone as well. However, this sphincter may be somewhat more functional than the urethralis muscle and may contribute to any retention of urine present. Presence of significant autonomic dysfunction typically results in a guarded prognosis even with therapy [5,33].

An animal sometimes displays nerve root signature with compression of the cauda equina. This is characterized by lifting the pelvic limb on ceasing movement. It is probably related to an attempt to improve comfort by decreasing stretch of the nerve root that has been compressed. Nerve root signature appears to be much more common with cervical radiculopathy in which the animal will hold up a thoracic limb.

Pathophysiology of Pain in Lumbosacral Disease

Pain is one of the key clinical features of lumbosacral stenosis and degenerative disease. Although the causes of pain might seem somewhat obvious, the variable nature of the degree of pain and the lack of resolution of pain following decompressive surgery among individual human and animal patients remain an enigma [34]. Direct mechanical compression of the nerve roots plays a role in the pathogenesis of pain, but it is not the sole cause. A tremendous amount of research has been conducted in both experimental and clinical settings to elucidate the underlying causes of low back pain with lumbosacral degenerative disease. Pain results from compression and inflammation of sensory nerve roots (radicular pain) and from degenerative and inflammatory changes in the articular facet joints, other supporting structures, and the intervertebral disk (discogenic pain) [35].

Many of the neuropathic pain models that have been developed experimentally involve the manipulation of the sciatic nerve, the cauda equina, or an isolated nerve root. Injury to the nerve, nerve root(s), or cauda equina has been induced by many mechanisms: manipulation, transaction, ligation, cryoneurolysis, crush, and chronic constriction [36-39]. Local pathophysiology at the site of nerve injury has implicated a plethora of potential mediators of pain. Compression of the nerve or nerve roots results in mechanical deformation with spontaneous (ectopic) depolarization of the nerve, vasogenic edema, infiltration of immune cells with subsequent up-regulation of immune mediators in both the nerve roots and spinal cord, and central sensitization of the spinal cord to up-regulate pain perception [35,40]. These factors are not independent entities and are interrelated in a complex pathway (Fig. 50-5).

Figure 50-5. A schematic overview of the interplay of factors in acute and chronic radicular pain with nerve root compression. Abbreviations: Pgs = prostaglandins; NO = nitric oxide; mΦ = macrophages; TNF-α = Tumor Necrosis Factor-α; IL-1β = Interleukin-1β; MAF = Macrophage Activating Factor; INF-γ = Gamma-interferon; NR = nociceptive receptor; T-h = T-helper cells; APCs = antigen presenting cells; ICAM-1 = Intercellular Adhesion Molecule 1; PECAM = Platelet Endothelial Cell Adhesion Molecule-1; CGRP = Calcitonin Gene Related Peptide; SubP = Substance P; NMDA-R = N-Methyl-D-Aspartate Receptor; IVD = intervertebral disk.

Nociception in the Anatomic Structures of the Spine

On a basic nociceptive level, sensory innervation is present to all of the structures of the spine [41]. Innervation to these structures primarily arises from the primary dorsal nerve root distal to the dorsal root ganglion [42,43]. Recurrent articular nerves innervate the vertebral periosteum, articular facet joint capsules, and ligamentous tissues of the dorsal neural arch (e.g., interarcuate ligament) [42]. Another branch forms the sinuvertebral nerve, which enters back into the intervertebral foramen and gives rise to cranial and caudal branches that supply rich innervation to the dorsal longitudinal ligament and dorsal annulus fibrosus of the intervertebral disk [42]. These fibers were recently shown to possess sensory-specific sodium-gated pain channels (SNS/PN3 and NaN/SNS2), which may provide a new therapeutic mechanism for analgesia [41]. The extension of these fibers over one to two vertebral segments may limit precise localization of pain to a specific intervertebral joint [41]. These two sensory nerves also have an intimate association with the autonomic plexus [41]. Painful or pathologic conditions of these nerves or of structures innervated by them may result in autonomic manifestations.

Because the articular facet is a true arthrodial joint, facet-associated pain can be attributed to mechanical stresses, osteoarthritis, and synovial inflammation [44-47]. Similar to other load bearing joints, degenerative changes in the articular facet include hyaline cartilage degradation, remodeling and osteosclerosis of subchondral bone, and osteophyte formation [47]. Although inflammatory processes in typical joints (i.e., knee and hip) have been well documented, studies evaluating the facet joint are few [47]. Pain secondary to osteoarthritis and joint degeneration is associated with cytokines produced by the arachidonic acid cascade. In one study, multiple prostaglandins and leukotrienes, both products of the arachidonic acid cascade, were produced in the facet joint, disk tissues, and subchondral bone of human patients with degenerative lumbosacral spinal disorders. The cytokines produced were similar to those obtained from cartilage and subchondral bone tissues of other osteoarthritic joints [48]. In another human clinical study [47], the synovium and joint cartilage of 40 patients undergoing lumbar spinal surgery were harvested for evaluation of cytokine production. Although the focus of the study was to identify the presence of interleukin-Iβ (IL- Iβ) and tumor necrosis factor-α (TNF-α), the authors were surprised to find that these cytokines were not found in a high concentration in patients with either lumbar disk disease or spinal canal stenosis. However, interleukin 6 (IL-6) concentration was elevated in both groups of patients. IL-6 production has been found to be induced by both IL-Iβ and TNF-α. Both of these cytokines are believed to be present early in acute inflammation and subsequent production of IL-6, further contributing to the inflammatory cascade [49-54]. The presence of elevated IL-6 without increased IL-Iβ and TNF-α may be attributed to the chronic nature of degenerative lumbar disease.

Less attention has been given to pain generated from compression and inflammation of the dura mater. Sensory innervation of the dura mater is provided by the sinuvertebral nerve which carries both sensory and sympathetic fibers [55]. Many peptidergic (pain-associated) nerve fibers are distributed throughout the dura mater [55] and likely play a role in generation of pain.

Mechanical Deformation of the Nerve Roots and Cauda Equina

In the normal state, the nerves of the cauda equina are somewhat moveable within the spinal canal and intervertebral foramen to accommodate changes in spinal extension and flexion and movement of the pelvic limbs. When the nerve roots become compressed or entrapped by extruded intervertebral disk, hypertrophied tissues, or canal stenosis, the movement of the nerve root becomes limited and traction and compression of the nerve root induce pain mechanisms and morphologic changes in the nerve. Chronic compression can be thought of as repeated episodes or persistence of acute compression [40]. The basic underlying pathophysiology of pain in degenerative lumbar spinal disease has been extensively studied. It begins with nerve root compression within the entire cauda equina or with compression of select nerve roots. Direct mechanical, chemical injury or mechanical injury to lumbar roots in the rat produces mechanical allodynia and thermal hyperalgesia [56]. Pain, allodynia, and numbness are subjective symptoms of lumbar radiculopathy that may be induced by ectopic or spontaneous firing of nociceptive sensory neurons [43,57]. Excitability and spontaneous firing in the dorsal root ganglion have been shown to be linked to sensory nerve sodium-gated channels (SNS/PN3 and NaN/SNS2) recently implicated in neuropathic pain [41]. Mechanical compression has additionally been shown to lead to intraneural tissue reaction, including edema, demyelination, and fibrosis [40,58,59]. Various morphologic and physiologic changes occur with separate phases (acute, subacute, and chronic) of nerve root compression [37,40,59-62]. A study by Kobayashi et al. [40] evaluated the morphologic derangements induced by clip compression of the L7 nerve root in a dog model. The clip was retained postoperatively and the nerve roots were evaluated at 1 and 3 weeks postoperatively. Three characteristic processes were observed: Wallerian degeneration, influx of macrophages, and radiculitis. Wallerian degeneration was observed in the nerve root proximal and ventral to the dorsal root ganglion. In the regions of Wallerian degeneration, macrophages were observed metabolizing necrotic debris. Macrophages are involved in the inflammatory cascade by enhancement of vascular permeability, production of chemotactic signals for other inflammatory cells, and modulation of these cells' activity, the exact origin of the macrophages is unknown [63,65]. Nervous tissue does not possess a lymphatic system; however, the cerebrospinal fluid may play a role similar to the lymphatic system because the flow of the fluid can remove metabolites and waste products from the spinal cord and nerve roots [63-65].

Many studies have been performed elucidating the pathogenesis of radiculitis associated with nerve root compression. Chemically induced inflammation of the nerve root by the presence of ruptured disk tissue disseminated along the root sheath has been proposed [35,63,66]. Disk substances thought to play a role in inducing chemical radiculitis include glycoproteins, immunoglobulin G, phospholipase A2, cytokines, and hydrogen ions [35,63,64,66-69]. Although these findings do not explain the production of inflammation with lumbar canal stenosis, compression of a nerve root results in intraneural edema and macrophage influx, which is associated with production of inflammatory cytokines, such as IL-1β, IL-6, and TNF-α, nitric oxide, and proteases [49-54,67].

In a study paralleling the dog nerve-clipping model [63], changes in the dorsal root ganglia were evaluated at 24 hours, 1 week, or 3 weeks after clipping. Compression of the dorsal roots proximal to the dorsal root ganglion resulted in an impairment to axonal flow in the root and morphologic changes including central chromatolysis in cell bodies of the dorsal root ganglion. Axonal transport plays a role in both the retrograde and anterograde movement of neurotransmitters, nutrients, and neurotrophic factors that are synthesized by dorsal root ganglion cells. Disturbance of axonal flow would be expected, then, to cause neurologic dysfunction. Central chromatolysis and other morphologic changes seen in the ganglion were considered a reflection of decreased synthesis of neurotransmitters and other neuronal homeostatic factors in favor of structural proteins such as the cytoskeleton and neurotrophic factors required for repair [63,70,71]. Nerve growth factors and other neurotrophic factors are thought to be involved in the maintenance of sensory neurons and may actually inhibit cell death after nerve transection [63,72-76]. When compression results in mild central chromatolysis, the neuron can usually recover fully once the compression is relieved. However, sustained compression could result in irreversible damage to the dorsal root ganglion, which may, in part, explain those patients that do not recover neurologic function after decompressive surgery.

Intraneural Blood Flow and Edema

Decreased blood flow within nerve roots or the cauda equina may play a role in the symptoms seen in patients with degenerative or stenotic lumbosacral disease [77]. Blood flow is shown to be negatively affected by experimental nerve root constriction [61,78-81], and hypoxic stress was shown to induce ectopic firing in the dorsal root ganglion and increased sensitivity to mechanical stimuli [78,82]. A large number of sensory roots pass through the dorsal root ganglion, which has an abundant vascular network and no blood-nerve barrier [80,83]. Compression likely occurs first in thin-walled venules, which results in impaired perfusion of the capillary system that feeds the nerve roots, producing ischemia and augmenting radicular edema [58,62,82,84,85]. Increased vascular permeability and subsequent edema in the remainder of the root are attributable to a breakdown in the blood-nerve barrier [40,63,84,86]. Endoneurial edema, which has been shown experimentally to be temporally associated with the onset of nerve root pain and neurologic dysfunction, is observed in the dorsal root ganglion during compression [57,80]. Edema results in high pressures being exerted on the ganglion cells because the perineurium prevents leakage into the epineurium and the fluid is retained [87,88] (Fig. 50-6). The edema may then increase the functional compression in addition to the mechanical stress and contribute to the impaired cell function [80]. In the aforementioned study of canine L7 clipping by Kobayashi et al. [40], increases in vascular permeability were most marked at 1 week postoperatively and began to resolve by 3 weeks. The early marked edema was attributed to loss of vasomotor control induced by vasoactive amines and neuropeptides produced in response to injury at this site. The persistence of increased vascular permeability beyond the acute injury phase was attributed to a greater demand for blood flow to provide nutrients and metabolic building blocks during regeneration [40,89]. Support of the contribution of vascular compromise in nerve root compression was offered by Hida [90], who showed that nerve-root blood flow measured intraoperatively during spinal lumbar decompressive surgery was increased in those patients whose neurologic deficits or pain resolved shortly after operation compared with those patients whose neurologic deficits or pain did not resolve after surgery.

Figure 50-6. A. T2-weighted, transverse plane MRI image at the level of the lumbosacral joint of a dog with left pelvic limb muscle atrophy and lumbar and limb pain. The dog had mild conscious proprioceptive deficits and weakness in the left pelvic limb. Severe swelling of the nerve root was identified (white arrow) intraoperatively and confirmed by biopsy. A foraminotomy was performed; the dog improved and was normal at 6 months postoperative. IL = wing of the ilium B. T2-weighted, coronal plane MRI image of the same dog. White arrow = edematous nerve root. SC = spinal cord.

Neuroimmune Responses, Inflammatory Mediators, and Central Sensitization

The process of nociception involves the generation of nerve impulses in small-diameter sensory neurons and propagation of those impulses to the spinal cord. A complex interplay exists between the nociceptors and the inflammatory cascade. Inflammatory mediators cause sensitization or enhanced responsiveness to stimuli, which are also accentuated by neuropeptides released in the inflammatory cascade [34]. Leukocyte trafficking into spinal cord has been observed in nerve root compression [91].

The nervous system is considered immunologically privileged, meaning that it is not typically surveyed by circulating lymphocytes as are other organs. Immune-membrane glycoproteins (CD4 and major histocompatibility complex, or MHC class II) and cell adhesion molecules (intracellular adhesion molecule-1, or ICAM-1, and platelet endothelial cellular adhesion molecule-1, or PECAM-1), which are integral to immunologic activation, and influx of immune cells were found to increase in response to nerve root injury [91]. MHC class II molecules are usually expressed on antigen-presenting cells (APCs) for interaction with T-cells in immune recognition. The antigen-presenting cells in the nervous system are perivascular cells and microglia [92,93]. Glial cells do not usually express MHC class II molecules, but expression can be induced by cytokines [94]. CD4 membrane glycoprotein, expressed on T-helper cells, macrophages, and microglia [95], plays the role of antigen recognition by APCs in association with MHC class II molecules. Lumbar nerve root injury resulted in enhanced CD4 expression in the grey matter of the spinal cord compared with sham operated and normal rats [91]. The role of the CD4 molecules is not completely known, but contribution to immunocompetence and response in the CNS after nerve root injury is likely. The nervous system constitutively expresses very low levels of these membrane proteins, which contribute to the nervous system's immunoprivileged status [91].

Nerve root injury may produce a cascade of events that upregulates ICAM-1, which enables entry of the hematogenous cells into the CNS, contributing to neuroinflammation and the development of central sensitization [91,96-98]. The upregulation of these molecules has been temporally associated with the onset of mechanical allodynia after nerve root injury [91]. Additionally, these changes are associated with the dorsal horn laminae in which nociceptive sensory fibers terminate [91]. Increased expression of cell adhesion molecules and membrane glycoproteins has been demonstrated with various disease processes of the CNS and PNS, including infection, autoimmune inflammation, and nerve root compression.

Experimental models have produced evidence for central neuroinflammation from nerve root compression, which includes astrocyte and glial activation and increased expression of proinflammatory cytokines [98-100]. Local production of proinflammatory cytokines via immune-cell activation has been implicated in enhanced nociceptor activity [101]. The CNS becomes infiltrated by immune cells from the peripheral circulation [91,102]. Microglia and astrocytes also become activated in response to peripheral nerve or nerve root injury [37] and release pro- and anti-inflammatory molecules, chemokines, and cellular adhesion molecules [96,97,102].

T cells recognize myelin sheath breakdown products and release macrophage-activating factors such as γ-interferon [103]. Activated macrophages begin to infiltrate locally to remove cellular debris and release inflammatory cytokines such as IL1-&beat;, nitric oxide, and TNF-α. When applied epineurally, TNF-α was shown to produce hypersensitivity that was blocked when antibodies against TNF-α were co-administered.101 TNF-α and nitric oxide may potentiate demyelination by Schwann cell injury. Nitric oxide also contributes to increased vascular permeability [104,105] and pain [106-108]. Prostaglandins, which also play an important role in inflammation, have not been shown to have a direct nociceptive effect but may lower the pain threshold at sensory nerve endings [109]. In addition, algesic spinal neuropeptide expression (calcitonin gene-related peptide, or CGRP and substance P, or SP) was shown in models of nerve root compression [110]. Mechanical compression of the dorsal root ganglion in rats resulted in an increase of substance P concentration in the ipsilateral dorsal root ganglion and spinal dorsal horn cells, which was linked to the onset of pain [111]. Accumulation of SP inside the axons of the central branches increased distal to the dorsal root ganglion as the strength and duration of compression increased, suggesting neurotransmitter dynamics are involved in the appearance of radicular pain [112].

The amino acid glutamate is one of the most ubiquitous neurotransmitters and has been shown to be associated with pain processing throughout the nervous system [113-117]. Glutamate antagonists that work at NMDA, AMPA, and kainite receptors have been shown to attenuate pain responses in rat models [113-120]. Although the actions of glutamate in the dorsal root ganglion are not completely known, the cell bodies have been shown to possess a high density of glutamate receptors [121-124] colocalized with nociceptive neurons [114,125,126], and mechanisms for reuptake of glutamate from the neuromuscular junction have been identified [122,123,127-129]. Consequently, glutamate is also found in high concentrations in aggrecan, the most prevalent proteoglycan in the extracellular matrix of the intervertebral disk [117,130-133]. Harrington et al. [134] hypothesized that glutamate released by proteoglycan degradation during disk degeneration may be a potential source of biologically active neurotransmitter by diffusing across the extradural space to affect glutamate receptors in the dorsal root ganglion [117,133,135,136]. Human herniated and non-herniated intervertebral disks were evaluated using both high-performance liquid gas chromatography and immunohistochemistry to test for the presence of glutamate. Glutamate was found to be abundant in the extracellular matrix of the intervertebral disk with concentrations significantly higher in the herniated disk material, suggesting release of glutamate from the proteoglycan structure. In addition to identification and semi-quantification of glutamate from the intervertebral disk, they evaluated the ability of tritiated-glutamate to diffuse across the epidural space and bind with glutamate receptors on dorsal root ganglion cells in a rat model. Tritiated glutamate was found to be bound to dorsal root ganglion receptors when infused into the epidural space at levels two magnitudes lower than that measured in the herniated disk material. These results suggested that the dorsal root ganglion cells were capable of glutamate uptake from the epidural space and that the concentration of glutamate normally present in the epidural space was relatively low, resulting in a steep concentration gradient diffusing from the herniated intervertebral disk material into the dorsal root ganglion. Further studies are required to understand how free glutamate affects nociceptive pathways in the dorsal root ganglion.

Pain from lumbosacral disease is characteristically chronic in its nature. Nerve injury may lead to potentiation of central sensitization and the development of chronic lumbar radiculopathy [34]. It has been shown that long-term potentiation occurs in the spinal cord in response to noxious stimulation or injury to peripheral nerves [137]. Repeated nerve root injury in a rat model produced significant increases in the magnitude and duration of mechanical allodynia in the paw of the rat [34]. Studies have shown the involvement of chemical factors (cytokines) around the nerve root associated with radicular pain [138-141]. Moreover, early on, pronounced spinal neuroimmune activation and inflammation induce central sensitization by directly or indirectly inhibiting interneuron inhibitory activity at the dorsal horn in the spinal cord [37]. Glial and neuronal proinflammatory cytokines can sensitize the peripheral nociceptive fields and dorsal root ganglia [142,143]. Glial cells synthesize proinflammatory cytokines, proteases, inducible-NO, excess glutamate, oxygen free radicals, eicosanoids, and other toxins that act at the NMDA (N-methyl-D-aspartate) receptors that are implicated in central nervous system sensitization [56,144-149]. Finally, supporting the concept of central sensitization, nerve root compression exhibits the features of mirror pain [34]. Mirror pain is the presence of pain sensation in the opposite limb when only unilateral nerve root compression is present.