Role of Early Imaging

The use of diagnostic imaging earlier in the course of a patient’s symptoms is supported by the following evidence: (1) patients with herniated discs treated surgically have better short-term outcomes than those treated conservatively^10,11; (2) the earlier the surgery the better the outcome¹²; and (3) surgery is cost-effective compared with conservative management (though this varies with the type of surgery).^13,14 However, the potential reduction in morbidity from early intervention requires the identification of accurate prognostic indicators, be they clinical or imaging based. Certainly, more than morphology is a factor. As has been suggested by Postacchini,¹⁵ two needs likely require consideration: (1) protracted conservative treatment may lengthen the time off from work and reduce the chances of successful surgical treatment; and (2) surgery should be avoided in patients with a herniated disc that may become asymptomatic or even disappear within a few months of onset. Thus, the possible benefits of imaging for degenerative diseases of the spine, other than as a presurgical tool, have never been carefully documented and quantified. Predicting which patients will benefit from imaging is a complicated and controversial subject. Its use as a preintervention planning tool is critical, but its employment for diagnostic information per se is less well accepted and the subject of much debate. Nevertheless, if imaging could be employed to prospectively determine which patients will not do well with surgery or other types of therapy, it would be of great value. Conversely, its use to identify patients undergoing prolonged conservative treatment who require more aggressive therapy (e.g., surgery) might be equally beneficial. This would save the cost of lost work, medical expenses, and personal discomfort. Clearly, there is little consensus, either within or among specialties, on the use of diagnostic tests for patients with back pain.¹⁶ The diagnostic evaluation depends heavily on individual physicians, their specialties, and patient socioeconomics, in addition to the patient’s symptoms.^16,17 On the other hand, patients with underlying risk factors (“red flags”), mainly, infection, trauma, and malignancy, warrant early imaging for timely intervention.

Imaging Considerations

The major benefit of imaging is its ability to depict the presence or absence of anatomic derangement and, in the former case, its effect on adjacent structures. The contrast sensitivity and multiplanar imaging capability of proton magnetic resonance (MR) provides a unique noninvasive means of imaging the intervertebral disc, adjacent osseous structures, and associated soft tissues. From an anatomic perspective, it is important to be able to demonstrate the osseous, fluid, and soft tissue interfaces within the three traditional compartments: the intramedullary, extramedullary intradural, and extradural spaces. The potential combinations of pulse sequence parameters, imaging planes, and postprocessing techniques available are almost limitless, and they can be used to highlight different aspects of the discovertebral complex and adjacent spaces. The methods include, but are not limited to, surface coil technology, cardiac gating, gradient refocusing, paramagnetic contrast agents, saturation pulses, gradient echo volume imaging, Turbo (fast) T2-weighted spin echo, fat suppression, magnetization transfer, short-tau inversion recovery (STIR), and diffusion techniques (including diffusion tensor imaging). Other modalities certainly do have a role based on the cost-effectiveness, accessibility, individual unique characteristics of some modalities, and, more importantly, contraindications to MRI. The technical aspects, advantages, and disadvantages of conventional and new modalities are briefly discussed in the following sections.

Counting Reference

The topic is discussed in detail in Chapter 175. It is extremely important that counting reference is clearly stated in the spine imaging report. This is to avoid wrong level surgeries with potential medico-legal implications. The two major factors responsible for discrepancy are (1) a transitional vertebra, which can be a sacralized L5 or lumbarized S1, and (2) variations in the number of rib-bearing vertebrae. Lumbosacral transitional vertebra (LSTV) is a common finding in the general population. Identification of LSTV on imaging is the essential first step in the interpretation of lumbar and thoracic imaging.¹⁸ The coronal MRI localizer image of the thoracic and lumbar spine can be useful because L4-5 is generally at the level of the iliac crest. Some prefer to use a vitamin E back-skin marker at an overlapping level in the thoracic spine, which, however, requires imaging of the entire spine. The iliolumbar ligament is easily identified on axial MRI and can be used as a marker of L5.¹⁹ There is a greater tendency of counting the lowest lumbar-type intervertebral disc space as L5-S1. Nevertheless, it is crucial that these approaches and variations be specified and clearly communicated in the report.

Plain Radiography

Plain radiography (PR) is still a very useful and cost-effective diagnostic tool by virtue of its availability and efficiency.²⁰ PR is generally used in trauma, for screening of degenerative disease, and for evaluation of sagittal translation using flexion-extension views. PR can aid MRI in characterization of disc disease by better demonstration of vacuum phenomena, disc calcification, bridging osteophytosis, and dorsal longitudinal calcification (though less common in the thoracolumbar spine). Conventional screen-film imaging has been nearly completely replaced by digital acquisition modalities, using either imaging plates (computed radiography, CR) or flat-panel detectors (direct radiography, DR).

Routine Magnetic Resonance Imaging

Unlike CT and conventional radiography, which depend on information related to electron density, proton MR signals are influenced by the T1 and T2 relaxation time, proton density, and motion (both macro and micro) to provide greater tissue contrast. Thus, its role may go beyond gross anatomic appraisal to actual tissue characterization of pathology and biochemical change. The routine sequences and their significance are as follows: (1) sagittal T1 to evaluate bone marrow infiltration and end-plate bone marrow degenerative changes; (2) sagittal T2 to evaluate disc and end-plate degeneration, thecal sac extrinsic impression, the conus medullaris, and the cauda equina morphology; and (3) sagittal STIR to detect bone marrow and paraspinal soft tissue edema. Axial T1 and T2 basically confirm the findings on sagittal imaging and provide a better assessment of central canal/thecal sac narrowing, cauda equina abnormalities, facet disease, and lateral disc herniations. The major degrading factors are patients’ gross motion, positioning, body habitus, and instrumentation.

Postcontrast Magnetic Resonance Imaging

The use of contrast agents is important for the improved depiction of reactive or inflammatory changes. Surgery results in reactive granulation tissue and scarring, typically in the epidural space. Noncontrast MRI is both sensitive and specific for the diagnosis of vertebral osteomyelitis. Contrast is helpful in the evaluation of paraspinal and epidural soft tissue involvement, including abscess formation. Contrast is not usually needed in the evaluation of bony neoplastic disease, unless epidural disease definition and diagnosis are required in selected cases. Contrast is essential in cases of malignant leptomeningeal disease.

Noncontrast Computed Tomography

Despite the preference for MRI in the evaluation of spinal canal stenosis, CT is essentially as accurate as MRI for the assessment of degenerative disease. Using bony and soft tissue algorithms and multiplanar reconstruction, CT is frequently used for preoperative planning.

Computed Tomography Myelography

Though still performed, conventional myelography has been significantly overshadowed by CT myelography. Because it is an invasive procedure, the CT myelogram is generally reserved for the following circumstances: (1) for instrumented spines, (2) as a problem-solving tool in clinically challenging cases supplementing or complementing MRI, and (3) for patients with contraindications to MRI. CT myelographic evaluation can be suboptimal if there is poor contrast mixing with the cerebrospinal fluid (CSF) or artifact from some of the instrumentations or the patient’s gross motion. Multidetector-enabled CT acquisitions with thinner slices facilitate 3D multiplanar orthogonal and angular reconstructions of the CT myelogram.

Magnetic Resonance Myelography

MR myelography (MRM) is currently not practiced routinely in the United States for two major reasons: (1) Intrathecal gadolinium is not yet approved by the FDA, and (2) MRM has no significant added advantage compared with a CT myelogram or conventional MRI.^21,22 MRM is, however, comparably accurate in detecting CSF leaks in patients with intracranial hypotension compared with radioisotope cisternography.²³

Open and Large-Bore Magnetic Resonance Imaging

The open-design MRI systems enable imaging of those who are claustrophobic and overweight and offer the potential of dynamic imaging. The major penalty is some compromise of the image quality due to decreased signal-to-noise ratio. However, the diagnostic information is satisfactory in most cases.^24,25 Closed high-magnetic-field scanners with short magnets and wide-bore tubes offer patients improved comfort and lessen anxiety.²⁶

Dynamic Imaging and Upright Magnetic Resonance Imaging

Sagittal segmental lumbar motion is traditionally assessed with lateral flexion-extension radiographs. These dynamic studies often demonstrate a decrease in the slip percentage between the vertebral segments with extension and an increase with forward flexion. Ironically, the degree of spondylolisthesis can actually improve on the lateral film taken on the operating table in an anesthetized patient.²⁷ Somewhat analogous to the flexion-extension imaging, supine-prone imaging of the lumbar spine can also be more sensitive by showing a higher degree of listhesis in the prone position.²⁸ Position-dependent stenosis can be demonstrated on the lateral myelogram in flexion-extension postures, which would otherwise have been underestimated on the follow-up CT myelogram²⁹ (Fig. 172-1). Ben-Galim and Reitman in a study of a small group of patients showed position-dependent severe spinal canal stenosis on CT myelography with grade I or II spondylolisthesis. Additionally, they showed fluid-filled distended facet joints, a marker of positional translation.³⁰ Imaging findings on supine MRI may correlate poorly with clinical findings. This may be related to the positional dependence of spinal stenosis, which reflects dynamic changes in soft tissue structures. Therefore, it is potentially valuable to evaluate spinal disorders under mechanical loading. Hydraulic axial-loading devices have also been tried.³¹ Upright MRI in different postures can reproduce the positions that bring about their symptoms and may unmask findings that are not visible with routine supine imaging. Small disc herniations and ventral spondylolisthesis can be unmasked, and disc herniations can be increased in size on upright MRI compared with recumbent MRI.³² Assessment of the degree of spinal stability in the degenerative and postoperative lumbar spine can also be performed.^33,34 Kinetic magnetic resonance images (kMRIs) in axially loaded, upright-neutral, flexion-extension positions can also accentuate the size of the disc herniations and play a potential problem-solving role in radiculopathy with negative supine static imaging.³⁵ Clearly, axial loading changes anatomic relations. The significance of the findings in terms of therapeutic decision making is not very clear.³⁶

FIGURE 172-1 Dynamic imaging. Erect lateral views of a lumbar myelogram in the extended (A) and flexed (B) positions showing a subtotal block to the contrast column at the L4-5 level in extension that is relieved in flexion. C, Routine supine CT myelogram shows mild diffuse disc bulge without significant compression of the thecal sac.

(From Saifuddin A: The imaging of lumbar spinal stenosis. Clin Radiol 55[8]:581–594, 2000, with permission.)

Disc Degeneration

The term disc degeneration is used for a variety of changes, including any or all of the following: real or apparent desiccation, fibrosis, narrowing of the disc space, bulging, fissuring or mucinous degeneration of the anulus, osteophytes of the vertebral apophyses, and end-plate/adjacent marrow changes. MRI and CT show disc space narrowing, T2-weighted signal intensity loss from the intervertebral disc, presence of fissures, fluid, vacuum changes and calcification within the intervertebral disc, ligamentous signal changes, marrow signal changes, and osteophytosis. These alterations result in disc herniation, malalignment, and spinal canal stenosis. Conventional theory suggests that aging and degeneration are very similar processes, albeit occurring at different rates. Historically, the two processes have been characterized morphologically and termed as spondylosis deformans and spondylosis osteochondrosis (deteriorated disc), respectively. The former affects the anulus fibrosus and adjacent apophyses, whereas the latter mainly affects the nucleus pulposus and the vertebral body end plates, particularly fissuring (numerous tears) of the anulus fibrosus. Scientific evidence suggests that spondylosis deformans is the consequence of normal aging, whereas intervertebral osteochondrosis represents a pathologic, though not necessarily symptomatic, process.³⁹ Anteriolateral osteophytosis is a reliable marker of spondylosis deformans, with a 100% incidence above 40 years.³⁹ Disc degeneration and aging are multifactorial processes that encompass a wide spectrum of changes and sequelae, of which the radial tear is but one. Disc degeneration may be explained primarily by genetic influences and complex unpredictable interactions of unidentified factors.

The nucleus pulposus is eccentrically located and more closely related to the dorsal surface of the intervertebral disc. With degeneration and aging, type II collagen increases outwardly in the anulus, and water loss from the nucleus pulposus is greater than from the anulus. This results in a loss of the hydrostatic properties of the disc, with an overall reduction of hydration in both areas to about 70%. In addition to water and collagen, the other important biochemical constituents of the intervertebral disc are the proteoglycans. It is not the individual composition but the relative composition of these constituents that in turn affects the hydrostatic properties and tensile strength of the intervertebral disc. The ratio of keratin sulfate to chondroitin sulfate increases. The disc becomes progressively more fibrous and disorganized, with the end stage represented by amorphous fibrocartilage and no clear distinction between nucleus and anulus. The degenerating disc manifests as diminishing T2 signal and loss of internal lamellation. T2 measurements also correlate with glycoaminoglycan concentration rather than absolute water content. So, it is not just the absolute concentration of water but also the state of water involved in the disc degeneration. The status of the proteoglycans is thus a major determinant of disc space T2 and T1 signal^39–41 (Fig. 172-2).

FIGURE 172-2 Degenerative disc changes. Anular disc bulges at L1-2 and L2-3. Disc space vacuum phenomena at L2-3 to L4-5, extending to the epidural space along the bulging disc at L2-3 (arrow). Mild degenerative spondylolisthesis, anterolisthesis at L4-5, and retrolisthesis at L2-3.

Anular Tears

Also called anular fissures, anular tears are separations between anular fibers, avulsion of fibers from their vertebral body insertions, or breaks through fibers that extend radially, transversely, or concentrically and involve one or many layers of the anular lamellae. The term tear, or fissure, describes the spectrum of such lesions and does not imply that the lesion is a consequence of trauma. Although it has certainly been verified that anular disruption is a sequela of degeneration and is often associated with it, its role as the causal agent of disc degeneration has certainly not been proved. Anular disruption initiates a reparative process as it is replaced by dense fibrous tissue and cystic spaces.^42–45 Fissuring, chondrocyte generation, and granulation tissue formation have been noted within the end plate, anulus fibrosus, and nucleus pulposus of degenerative discs, indicating attempts at healing.³⁹ Anular tears are commonly seen in the dorsal bulging discs as focal T2 hyperintensity on T2 and STIR imaging; there may be associated small disc protrusion. Anular tears consistently show focal contrast enhancement⁴⁶ (Fig. 172-3).

FIGURE 172-3 Anular tears. Sagittal (A) and axial (B) T2 images show typical posterior anular tear with hyperintense T2 signal (arrows) and associated small disc protrusion.

Disc Calcification and T1 Hyperintensity

Patchy, partial, or complete calcification of the disc space can be seen either as an isolated finding without degenerative stigmata or as a chronic dystrophic change in a degenerated disc. The clinical significance of a calcified disc is unknown. The disc space on CT and plain films can appears as fused, perhaps “autofused.” Calcification, depending on its biochemical status, can appear as hyperintense or hypointense on sagittal T1 imaging. Furthermore, the hyperintense T1 disc may represent an ossified disc, as can be proven by loss of signal on fat-suppressed MRIs due to lipid marrow. A calcified disc is easily demonstrable on CT (Fig. 172-4).

FIGURE 172-4 Disc space calcifications: Sagittal T1-, T2-, and fat-suppressed T1-weighted images of the upper lumbar spine. On the T1-weighted images, there is a focus of high signal (arrow). On the T2-weighted image, this same area has a decreased signal (arrow). On the fat-suppressed image, the region maintains its high signal, while areas which contain fat have a decreased signal (arrow). The high signal on T1, decreased signal on T2, and unaffected by fat suppression are indicative of calcification producing T1 shortening.

Degenerative Marrow Changes

The anulus fibrosus, vertebral end plate, and vertebral body have an interreactive relationship in the degenerative process, as is also evident from the model of chymopapain-treated discs.^48,49 Paradiscal MR signal changes in the end plate and vertebral bodies take three forms.

Type I Changes

Type I changes demonstrate decreased signal intensity on T1-weighted images and increased signal intensity on T2-weighted images. Type I change is found in about 4% of nonsurgical lumbar spines, 8% in postdiscectomy discs, and 40% to 50% of chymopapain-treated discs.^48–50 The last group can be used as a model of acute disc degeneration. Histopathology of type I changes demonstrates disruption and fissuring of the end plate and vascularized fibrous tissues within the adjacent marrow, accounting for prolonged T1 and T2 MR signal. Enhancement of type I vertebral body marrow changes is seen with administration of IV gadolinium. The enhancement may involve a portion of the disc itself and is presumably related to the vascularized fibrous change. The type I signal is expectedly accentuated on STIR, mimicking bone marrow edema, and may be very difficult in some cases to differentiate from discitis osteomyelitis (discussed later under the differential diagnosis of vertebral osteomyelitis; Fig. 172-5).

FIGURE 172-5 A–C, Type I end-plate changes. T1 hypointense and T2 hyperintense end-plate changes at L3-4 (curved arrows); the end-plate signal is accentuated on short-tau inversion recovery (STIR) image due to suppression of the fat in the surrounding marrow. Note that the same bone marrow changes also surround a large Schmorl node at L3 (straight arrows). Note a small Schmorl node at L2; however, there is no surrounding reactive change or hypointense rim.

Type II Changes

Type II changes are represented by increased signal intensity on T1-weighted images and isointense or slightly hyperintense signal on T2-weighted images. Type II changes are identified in approximately 16% of patients on MRI.^48–50 Discs with type II changes, like type I, show evidence of end-plate disruption, with fat marrow in the end plate and adjacent vertebral body resulting in T1 hyperintensity due to T1 shortening (Fig. 172-6).

FIGURE 172-6 A–C, Type II end-plate changes. T1 and T2 hyperintense end-plate changes at L4-5 (arrows). Like the surrounding fatty marrow, the end-plate signal is suppressed on the short-tau inversion recovery (STIR) image.

Type III Changes

Type III changes are represented by a decreased signal intensity on both T1- and T2-weighted images due to extensive bony sclerosis, as evident on plain radiographs and CT. The first two types show no definite correlation with sclerosis seen at radiography, which is not surprising when one considers the histology. The sclerosis on plain radiographs is a reflection of dense woven bone within the vertebral body rather than of the marrow elements. The lack of signal in the type III change represents the relative absence of marrow in areas of advanced sclerosis (see Fig. 172-17B).

Facet Joint Effusions

Pencil-thin homogeneous bilateral facet joint effusion is a common and normal finding in the lower lumbar spine. In an arthritic facet joint, synovial villi may become entrapped within the joint, with resulting joint effusions. STIR is the best method for demonstrating joint effusions. Large joint effusions (>1.5 mm) are highly predictive of degenerative spondylolisthesis on dynamic imaging, particularly at L4-5⁵⁷ (Fig. 172-7).

FIGURE 172-7 Facet effusions. Sagittal short-tau inversion recovery (A) and axial T2 (B) images showing facet joint effusions at L4-5 (arrows) and common associated finding of mild anterolisthesis at L4-5 (C).

Dorsal Elements and Paraspinal Edema

Routine use of STIR and similar sequences has resulted in increased detection of dorsal elements and paraspinal soft tissue edema.^59,60 Facet joint degenerative arthritis can result in para-articular and pedicular edema, joint effusion, and para-articular soft tissue edema. In patients at high risk, it may be difficult to differentiate from infected facet joints. The signal changes and sclerosis in the pedicles can be seen secondary to facet joint arthritis and underlying pedicle or pars fractures. The greatest mechanical stress is on the pars interarticularis, followed by pedicles as shown by Sairyo et al.⁶¹ These stresses can result in a spectrum of reactive bone marrow changes in the pars and pedicles. T2 hyperintense signal in the pedicles has been shown as an early marker of spondylolysis.⁶¹ There is evidence that categories of changes in MR signal intensity, similar to those described adjacent to degenerating discs, can be seen in lumbar pedicles adjacent to a spondylolytic defect of the pars interarticularis⁶² (Figs. 172-8 and 172-9).

FIGURE 172-8 Facet edema. A and B, Hypertrophic facet arthropathy at L2-3 (arrows). C, Associated marrow and soft tissue edema (arrows). Also note L2 and L3 pedicles edema (anterior arrows).

FIGURE 172-9 Pedicle edema. Hypointense T1 (A) and hyperintense T2 (B) pedicle signal (arrow) at L5. C, The T2 signal is accentuated on short-tau inversion recovery (STIR) image. D, CT shows pars interarticularis defect.

Segmental Instability

Each level in the thoracolumbar spine consists of three joints: intervertebral disc space and bilateral facet joints. Degeneration of these joints results in segmental instability replacing the normal spinal motion, which can be irregular, excessive, restricted, translational, and/or angular. Segmental instability, which can cause static and dynamic stenosis, is considered a cause of LBP but is poorly defined.⁶³ On flexion-extension plain radiography, generally about 3 mm or more of sagittal translation is equated with instability. Spondylolisthesis is a term used in static imaging and refers to intervertebral displacement—anterolisthesis, retrolisthesis, or lateral listhesis. The different types, based on etiology, include congenital, degenerative, isthmic, iatrogenic, pathologic, and traumatic (Fig. 172-10).

FIGURE 172-10 A and B, Isthmic spondylolisthesis. Sagittal T2 imaging showing grade I anterolisthesis due to L5 spondylolysis (arrow).

Grading of Spondylolisthesis

Spondylolisthesis in the sagittal plane can be graded by one of two methods, independent of the imaging modality and type of spondylolisthesis.⁶⁴ The first is the method described by Meyerding.⁶⁵ The anteroposterior (AP) diameter of the superior surface of the lower vertebral body is divided into quarters, and a grade of I to IV is assigned to slips of one, two, three, or four quarters of the superior vertebra, respectively. The second method, first described by Taillard,⁶⁶ expresses the degree of slip as a percentage of the AP diameter of the top of the lower vertebra. Complete slip of L5 on S1 is termed spondyloptosis. The second method is favored by most authors as it is more accurately reproducible.⁶⁷ Measurement of the slip and its apparent progression, however, should be viewed with caution. Studies have shown that inter- and intraobserver error of up to 15% is possible. This variation can increase if an element of rotation is present. Therefore, only a progression of greater than 20% slip can be reliably assessed.^68,69

Degenerative Spondylolisthesis

Degenerative spondylolisthesis (DS), which is usually seen with an intact pars interarticularis, is related to degenerative changes of the apophyseal joints and/or intervertebral discs and is most common at the L4-5 and L5-S1 vertebral levels. Lumbar DS is a major cause of spinal canal stenosis and is often related to low back and leg pain.⁷⁰ Long-standing DS is almost invariably accompanied by disc and facet degenerative changes, which obscure the origin of the process. Discogenic DS and facetic DS cause secondary stress and stretch on the facet joints and disc spaces, respectively. The degree of sagittal listhesis is generally mild in DS relative to other types, particularly isthmic spondylolisthesis. There may be associated degenerative scoliosis and listhesis in nonsagittal planes. Type I end-plate changes, which may be more extensive and quite prominent on a STIR sequence, even without spondylolisthesis on static imaging, may represent segmental instability⁷¹ and can mimic discitis osteomyelitis, as discussed later (Fig. 172-11).

FIGURE 172-11 Degenerative spondylolisthesis and scoliosis. A, Grade I anterolisthesis at L3-4 and L4-5. Note high-grade stenosis at L4-5 with redundant nerve roots (arrow). B, Typical degenerative scoliosis with lateral listhesis.

Pseudarthrosis

Pseudarthrosis is more common in a postoperative spine and can occur at the fusion level (nonunion) or adjacent levels (transitional). The latter can also have noniatrogenic causes.

Nonunion of Bone Fusion

Nonunion is the result of failed spinal fusion. The failed fusion typically manifests with axial or radicular pain months to years after surgery. The common causes of this complication are inadequate surgical technique, excessive stresses across the fusion site, insufficient internal or external stabilization, and unrecognized metabolic abnormalities.⁷² Diagnosis is based on clinical presentation and imaging studies, after other causes of persistent pain are excluded.⁷³ The degree of fusion and associated motion seen on flexion-extension radiographs that is indicative of solid or failed fusion remains a point of controversy due to significant interobserver variations.⁷⁴ The choice of bone graft material can affect the radiographic assessment of interbody fusion. CT has higher contrast resolution and is therefore better than radiography for assessment of bone fusion. One of the main reasons for CT’s superiority is that premineralized osteoid may be functionally fused but radiographically lucent, therefore leading to underestimation on plain radiography. Bony arthrodesis is usually evident by 6 months, with evidence of bridging trabecular bone. Bridging bone is usually seen lateral to the implant and may also be noted within the implant itself. Mature solid fusion is generally present at between 12 and 24 months^75,76 (Fig. 172-12).

FIGURE 172-12 Pseudarthrosis. A, Postoperative spine with well-formed L4-5 interbody fusion (black arrows) but pseudarthrosis at L5-S1 interbody fusion (white arrow). Note anti-kickout screw at L4-5 (curved arrow). B, Posterolateral fusion mass at L5-S1 with gross defect due to pseudarthrosis (arrow). C, Segmental pseudarthrosis through anterior and dorsal columns at a level just above the L2-3 and L3-4 interbody fusion (arrows). D, Similar segmental pseudarthrosis due to neuroarthropathy (Charcot spine, arrows).

Disc Space Failed Fusion

Failed fusion at the disc space manifests as lucency at the device margins or lucent lines through the fusion mass by about 18 to 24 months. Lucency around the fusion devices (e.g., cage or pedicular screws) represents abnormal motion and predicts delayed or failed fusion. Cystic change at the end plates adjacent to the implant is a reliable marker of failed fusion. Other signs include malpositioning and subsidence of the fusion devices^74–76 (see Fig. 172-12).

Posterolateral Failed Fusion

Posterolateral failed fusion is divided into three types: Type I shows atrophy and resorption of the grafted bone; type II shows a lack in trabecular continuity of the fusion mass; and type III shows a gap either cranially or caudally between the fusion mass and lumbar matrix (which includes the transverse process and superior facet). CT is better than PR for diagnosing all types of posterolateral failed fusion, particularly type III⁷⁷ (see Fig. 172-12).

Intervertebral Disc Displacement (Disc Herniation)

Degeneration allows the disc to displace focally or diffusely in any direction beyond the end-plate margins of the disc space. The anulus fibrosus in a displaced disc tries to contain or limit the herniating nucleus pulposus. A variety of terms are used to describe disc displacement. See Figure 172-13 for diagrammatic representation of disc displacements.

FIGURE 172-13 Schematics of disc herniations. Diagrammatic representation of disc herniations. A, Sagittal illustration of normal disc, anular tear, and herniation of the nucleus pulposus. B, Focal disc herniation (<25% of the disc margin) and broad-based herniation (>25% of the disc margin). C, Difference between protrusion and extrusion in the axial plane. Note that the base of the protrusion is broader than the extent of the disc herniation beyond the margins of the disc. The extrusion, on the other hand, shows a greater extension beyond the disc margins than the width of the base.

Protrusions

Protrusions are disc displacements in which the greatest distance in any plane between the edges of disc material beyond the disc space is less than the distance between the edges of the base in the same plane (neck is broader than dome). Protrusion is still contained by the anulus fibrosus fibers and therefore is not a true herniation. There may be an associated anular tear, which is usually evident on MRI as T2 hyperintensity, typically in the dorsal or dorsolateral disc protrusions. If the protrusion is less than 25% of the disc circumference, it is referred to as focal, and if between 25% and 50%, it is referred to as broad-based (Fig. 172-14).

FIGURE 172-14 Disc protrusion. Mild broad-based disc protrusion at L2-3 with moderate thecal sac compression; there is underlying developmental shortened pedicles (arrows, A and C). Right foraminal L3-4 disc protrusion (arrows, B and D).

Extrusions

Extrusions are disc displacements that are true herniations in which in at least one plane, any one distance between the edges of the disc material beyond the disc space, is greater than the distance between the edges of the base in the same plane (dome is broader than neck). Extrusions usually extend above or below into the suprapedicular or infrapedicular zone, resulting in further narrowing of the neck but still abutting the native disc. The signal intensity of the extruded portion may be increased on T2-weighted images. There may be associated curvilinear areas of decreased signal intensity on the T2-weighted images, which are related to torn portions of the anulus and posterior longitudinal ligament (Fig. 172-15).

FIGURE 172-15 Disc extrusion. Sagittal (A) and axial (B) T2 imaging showing small left central disc extrusion (arrows) with compression of left S1 nerve root, which cannot be discretely identified (grade III compression). Note intact right S1 nerve root and its sheath (arrow on the right in B).

Sequestration

Sequestration is displaced disc material that has completely lost any continuity with the parent disc. The extrusion is referred to as “migrated” if it has displaced from the site of extrusion, regardless of whether it is sequestrated. Sequestrated fragments can lie ventral to the posterior longitudinal ligament (especially if they have migrated behind the vertebral bodies where the posterior longitudinal ligament is not in direct opposition), dorsal to the ligament, rarely in the intradural or intraradicular space, and even in the dorsal extradural space.⁷⁸ Nevertheless, penetration almost invariably occurs through the posterior longitudinal ligament; either dorsal, where it is fused with the anulus, or rostrally or caudally, where it fuses with the vertebral body margin. In the majority of patients in whom sequestrated fragments migrate behind the vertebral body, the sequestration usually lateralizes, with disc material pushing across the midline and the leading edge being smoothly capped. It has been postulated that this shape is imposed by a midline septum in the ventral epidural space. This space is largest in the lower lumbar region and is delineated dorsally by the posterior longitudinal ligament and laterally attached membranes and ventrally by the vertebral body. It is divided into two compartments by a sagittally aligned septum. Sequestrated fragments within the lateral recess and the neural foramen have been shown to produce eroded cortical bone and expansion of those spaces and, thus, should be considered in the differential diagnosis of a mass arising and expanding the neural foramen and lateral recess. Intradural disc herniation is very rare, with few cases reported.^79,80 It is most frequent in the lower lumbar spine. The mechanism is thought to be the development of chronic inflammation leading to adhesions between the dura mater and posterior longitudinal ligament. The appearance on axial MRI is described as a “hawk-beak sign.”⁸¹ As the herniated disc penetrates, the ligament extends through the dura instead of pushing it away. Other possible causes are congenital connections between the ligament and the dura or previous surgery (Fig. 172-16).

FIGURE 172-16 Sequestered disc. Sagittal T1 (A) and T2 (B) imaging shows left-sided sequestered disc with mild thecal sac compression. Note that extruded disc has lost contact with the bulging disc at L4-5 (arrows). C, Also note increased T2 signal in the herniated disc resulting in lesser conspicuity on axial T2 slice due to adjacent hyperintense fat (arrows).

FIGURE 172-17 A, Calcified herniated disc. Large disc extrusion at L2-3 with severe central spinal canal stenosis. Note isointense T1 and T2 signal to the native disc and expected mild rim enhancement. Surgery revealed a calcified rocklike herniated disc material. MRI was not helpful in predicting the calcified nature of the herniated disc. B, Calcified herniated disc and type III end-plate change. Advanced disc degeneration at L5-S1 with type III end-plate changes, shown as hypointense T1 and T2 signal and sclerosis on CT (anterior arrows). CT also shows a calcified extruded disc as also predicted on MRI due to T1 hyperintense signal (posterior arrow). Surgery revealed an epidural perineural mass–like structure. Biopsy was negative for neoplasm.

Imaging Characterization of a Disc Herniation

Disc herniation is not always clear-cut on imaging, and it may appear as a protrusion in one plane and an extrusion in another. If there is displacement away from the disc space in craniocaudal plane, it should be referred to as an extrusion. In the transverse plane, the disc abnormality is usually described as central, right, or left central, lateral recess, foraminal, or extraforaminal (lateral and far lateral). In the sagittal plane, the terms discal, infrapedicular, and suprapedicular are most commonly employed.

Containment refers to the integrity of the outer anulus covering the disc herniation. Imaging is not accurate in delineating containment of the nucleus pulposus. One may view the continuum of herniated disc disease as starting with anular disruption, proceeding on to small focal herniation that is not broken completely through the anulus-ligamentous complex to frank herniation (extrusion), which has dissected through the anulus and dorsal ligamentous complex completely. Disc herniations may show variable degrees of containment, and a line of decreased signal intensity has been reported around sequestrated fragments and large extruded discs where disruption of the anulus and ligament is clearly evident. This is thought to be secondary to anular and ligamentous fibers, which are carried away with the disc herniation. The anulus fibrosus and posterior longitudinal ligament are so intertwined at the level of the disc that a distinction between the two structures may be impossible or, for that matter, irrelevant. Technical limitations of CT and MRI usually preclude the distinction of a contained from an uncontained disc herniation.

Contrast enhancement is commonly seen as rim enhancement along the dorsal margin of disc herniation. Rarely, the disc herniation is diffusely enhanced, and it may be difficult to differentiate from epidural scarring in a postoperative spine.

Hyperintense zone (HIZ) in a disc has been used in correlative studies of MR and provocative discography. This nonstandard term generally represents anular tears.^82,83

Compression of the nerve root within its sleeve may not correlate with patient symptoms. A grading system by Pfirrmann et al.⁸⁴ had significant reliability and surgical correlation:

A herniated disc may be densely calcified, which may not be easily predictable on preoperative MRI. An isointense to hyperintense T1 signal can be helpful in such cases. CT clearly demonstrates calcification of disc herniation and ossification of the posterior longitudinal ligament (Fig. 172-17).

Intravertebral Disc Herniations (Also Known as Schmorl Nodes)

Nonacute Schmorl node (SN) intrabody herniations are common spinal abnormalities regarded as incidental observations. They have been reported in 38% to 75% of the population.^85,86 Most intrabody herniations probably form after axial loading trauma, with extrusion of nuclear material through the vertebral end plate, resulting in reactive bone marrow edema. Another theory is that Schmorl nodes are the end result of ischemic necrosis beneath the cartilaginous end plate, with secondary herniation into the body of the vertebra.⁸⁷ SNs are mostly well defined and have an isointense intrinsic T1/T2 signal to the disc.⁸⁸ SNs have been traced to episodes of significant, sudden-onset, localized, nonradiating back pain and tenderness. Wagner et al.⁸⁹ showed in symptomatic patients that 57% of cases on MRI had SNs surrounded by vertebral body marrow edema. The remaining nodes (43%) were not immediately apparent as SNs and manifested only as vertebral body edema, suggesting end-plate fracture, but did evolve into classical chronic SNs as revealed on follow-up imaging.⁸⁹ Contrast enhancement demonstrated vascularized SNs, with a higher incidence of surrounding edema. They are more common in symptomatic patients.^90,91 A concentric hypointense ring appearance has a high negative predictive value for the absence of underlying fracture, infection, or malignancy.⁸⁸ Giant cystic SNs have been described by Hauger et al.⁹² as a clinicoimaging entity in a small series of six cases (Fig. 172-18; and see also Fig. 172-5).

FIGURE 172-18 Schmorl node. Ill-defined hypointense T1 (A) and hyperintense T2 (B) reactive marrow changes (arrows) around Schmorl node at L4 inferior end plate. The signal alteration is similar to type I end-plate change. C, Note sclerosis along the margin of the Schmorl node. Also note hypointense rim of the node (B). D, Another patient with well-defined hyperintense T2 signal around the Schmorl node (arrow).

Although consistent terminology is important for communication, it is not at all clear whether or not these categories of descriptive findings of disc herniations are clinically relevant. Although it has been proposed that it is critical to differentiate between various degrees of herniation, the reality is that disc herniation most likely represents a spectrum or continuum rather than discrete entities with specific clinical relationships. Therefore, the term herniation has been interchangeably used in the literature with protrusion, extrusion, and sequestration.

Spinal Canal and Foraminal Stenosis

Spinal stenosis refers to any type of narrowing of the spinal canal, nerve root canals, or intervertebral foramina.⁹³ Two broad groups have been defined: (1) acquired (usually related to degenerative changes) and (2) congenital or developmental. Developmental stenosis can be exacerbated by superimposed acquired degenerative changes. In the acquired type, there has been no association between the severity of pain and the degree of stenosis. The most common symptoms are sensory disturbances in the legs, LBP, neurogenic claudication, weakness, and relief of pain by bending forward. Clinical stenosis is perhaps a better term than spinal stenosis (as diagnosed on imaging studies). The imaging changes are in general more severe than expected from the clinical findings.⁹⁴ Although there does appear to be a correlation between cross-sectional area and midsagittal measurements in patients with symptomatic spinal stenosis, measurement of canal stenosis is not generally used.³⁹

Acquired Lumbar Spinal Stenosis

Acquired lumbar spine stenosis (LSS) is typically a sequela of hypertrophic degenerative disc and facet/ligamentous changes that can be accentuated by static or dynamic degenerative DS. Anatomically, spinal stenosis is divided into central canal, lateral canal (subarticular and lateral recess), and foraminal stenosis. Subjective visual grading of the spinal stenosis is a routine practice in interpretation of the spine imaging. Quantitative evaluation by measuring the AP diameter or cross-sectional area of the central canal is used in research studies but is not commonly done in clinical practice.⁹⁵ Recent Framingham study data used a cut-off AP diameter of 12 mm for relative LSS, and 10 mm for absolute LSS.⁹⁶ There is tendency to subjectively grade central spinal canal stenosis as mild, moderate, and severe based on the effacement of the thecal sac due to a combination of anular disc bulging, end-plate osteophytosis, and dorsal ligamentum laxity. Nevertheless, spinal stenosis, even of a severe grade, is meaningless if it lacks a correlation to patient symptoms (Fig. 172-19).

FIGURE 172-19 Degenerative spinal canal stenosis. A, Plain CT showing L4-5 disc degeneration and anterolisthesis. B, CT myelogram reveals severe central canal stenosis with large ventral extrinsic compression due to extruded disc (arrow). Note near-complete myelographic block with fainter opacification of the thecal sac caudal to the stenosis.

Congenital Lumbar Stenosis

Congenital lumbar stenosis manifests at a relatively younger age with multiple levels of involvement. Relative milder degenerative changes can cause significant spinal and foraminal stenosis by virtue of developmentally compromised reserve. Radiographically, Dewing et al.¹⁰¹ and Singh et al.¹⁰² showed that these patients have a shorter pedicular length, resulting in a smaller cross-sectional spinal canal area (mean critical values of 6.5 mm and 213 mm², respectively). The AP pedicle:vertebral body length ratio was decreased (e.g., 0.36 in patients vs. 0.48 in controls at the L4 level). Recognizing congenital stenosis alerts the treating surgeon to the possibility of multilevel treatment. Subjective MR signs of congenital stenosis include elongated teardrop foramina in the sagittal plane and shortened pedicles with stubby dorsal elements and decreased AP diameter in the axial plane (Fig. 172-20).

FIGURE 172-20 Congenital spinal stenosis. Sagittal (A) and axial (C) imaging shows developmentally shortened pedicles (arrows). Note slitlike narrowing of the foramina (A). Mild disc bulging causing relatively moderate central canal stenosis due to underlying limited reserve (B).

Spinal Epidural Lipomatosis

A common incidental finding in the sacral and lower lumbar canal, spinal epidural lipomatosis (SEL) is typically of no clinical significance. Epidural lipomatosis is usually idiopathic and related to the patient’s body habitus. Exogenous steroids can also result in SEL. There are reports of symptomatic epidural lipomatosis treated with decompression surgery.^103,104 SEL does reduce the reserve of the central canal in patients who have hypertrophic degenerative disease. Both CT and MRI demonstrate epidural lipomatosis well. Severe epidural lipomatosis compresses the thecal sac, resulting in a Y-shaped configuration^103,105 (Fig. 172-21).

FIGURE 172-21 Epidural lipomatosis. Sagittal (A and B) and axial (C) imaging shows concentric epidural lipomatosis (arrows) with relatively small thecal sac; lipomatosis more prominent at L5-S1 and sacral canal.

Asymptomatic Subjects with Positive Imaging

There is a high prevalence of degenerative disc changes in asymptomatic patients. Irrespective of age, Wiesel et al.¹¹³ found about 35% of spines to be abnormal in asymptomatic patients. Degenerative disc and/or facet disease, including spinal canal stenosis, was present in about 20% of patients younger than 40 years, whereas the prevalence rose to about 50% in patients older than 40 years. Other authors showed similar observations. The prevalence of morphologic alterations increases with advancing chronologic age. Boden et al.¹¹⁴ showed that 57% of the scans were abnormal in a group of individuals older than 60 years of age (37% had disc herniations and 21% had spinal canal stenosis), whereas in the group younger than 60 years, 20% had disc herniations. Jensen et al.¹¹⁵ evaluated MRI examinations of 98 asymptomatic patients and showed that 64% had disc space abnormalities, 52% anular disc bulges, 27% protrusions, 1% extrusions, 19% SNs, 14% anular defects, 7% central canal stenosis, and 38% had a disc abnormality at more than one level.

Boos et al.¹¹⁶ showed a very high prevalence of disc herniation (i.e., 76% in age-, gender-, and occupation-matched asymptomatic cases), which was comparable to that of symptomatic patients (i.e., 96%); there was no statistical significance. Patients had more severe disc herniations (disc extrusions) than asymptomatic volunteers (35% vs. 13%). There were no significant differences regarding disc degeneration between both groups (96% vs. 85%). The only substantial imaging morphologic difference between the groups was the presence of a neural compromise (83% vs. 22%).

The preceding data emphasize the importance of correlating the clinical picture with the MRI examination when major surgical decisions are made. In fact, the findings on lumbar spine MRI may be meaningless when considered in isolation. In addition, with only 1% of the sample showing an extrusion in one of the larger studies by Jensen et al.,¹¹⁵ subcategorizing herniations into protrusion and extrusion may be helpful for better characterizing asymptomatic versus symptomatic discs.

Symptomatic Patients with Negative Imaging

Back pain is thought to occur in some patients without evidence of such morphologic abnormalities as herniation or stenosis. The perplexing clinical scenario of patients who complain of incapacitating back pain but have no overt morphologic abnormality has given rise to the concept of the disc as a pain generator. This was classically described by Crock¹¹⁷ as “chronic internal disc disruption syndrome.” Many different names have been given to this idea, which becomes more confusing when combined with the various diagnostic tests that are used in an attempt to diagnose this confusing syndrome. Additional terms in the literature include internal anular tear, internal disc disruption, black disc disease, and discogenic pain. In a normal human lumbar disc, nerve endings can be found only in the periphery of the anulus, and the pain fibers are part of the sympathetic chain via the sinuvertebral nerve.^118,119 This innervates the outer layer of the anulus fibrosus. However, in very degenerated discs, nerves may even penetrate into the nucleus pulposus.¹²⁰ Potentially, stimulation of these fibers can occur not only from direct disruption and mechanical pressure on the anulus but also from various breakdown products of the nucleus pulposus or secondarily up-regulated inflammatory mediators. The normal disc acting as a source of pain is still a mystery. Provocative discography, as discussed in the following section under symptomatic patients, has been used in morphologically normal discs with an expected lack of pain reproduction.

Disc material leakage is hypothesized to cause back pain and radiculopathy through a chemoinflammatory mechanism. The nucleus pulposus can cause an inflammatory reaction with leukotaxis and increased vascular permeability.¹²¹ A simple incision of the anulus fibrosus can produce morphologic and functional changes in the adjacent nerves, such as increased capillaries and reduced nerve conduction velocities,¹²² with the presumed mechanism of disc material leakage into the epidural space. McCarron et al.,¹²³ using a dog model, demonstrated that autogenous placement of the nucleus pulposus into the epidural space caused acute and chronic inflammatory reactions, with influx of histiocytes and fibroblasts. Kayama et al.¹²² and Olmarker et al.¹²⁴ demonstrated that the contents of the nucleus pulposus applied to spinal nerves induces a wide variety of functional, vascular, and morphologic abnormalities, often followed by intraradicular fibrosis and neural atrophy.

Symptomatic Patients with Positive Imaging

This group of patients may have positive scans, but the significance of these findings is rather confounding. The approach to positive findings from diagnostic imaging is threefold: (1) to determine the etiologic correlation to the patient’s symptoms, (2) to stratify patients for conservative versus interventional treatment, and (3) to provide a patient prognosis. Unfortunately, diagnostic imaging produces no convincing, considerable, or consistent results on these three fronts.

Disc Space Degeneration

Treating the sequelae of disc degeneration (i.e., disc herniations, stenosis, and instability) is the goal of therapy and therefore the major focus in clinical and imaging trials rather than the disc degeneration itself. The intervertebral disc is considered to be the source of discogenic pain. Provocative discography has been used to identify one or more disc spaces as the source of pain. Weishaupt et al.,¹²⁵ however, showed a lower specificity (59%) and positive predictive value (PPV) (63%) for disc degeneration as a correlate of provocative pain, with mildly increased PPV for degenerated discs with HIZs. Their high negative predictive value (98%) in the normal-appearing discs at least indirectly suggests the origin of pain from other levels or causes. Recently, Kang et al.⁸³ demonstrated very high specificity (97.8%) and PPV (87%) for disc protrusion with HIZ in reproduction of pain. They classified the MR findings of the disc into four classes: (1) normal or bulging disc without HIZ; (2) normal or bulging disc with HIZ; (3) disc protrusion without HIZ; and (4) disc protrusion with HIZ. The first two classes had a less significant correlation to provocative pain. Disc protrusion without HIZ had a moderate correlation (specificity, 80.6%; PPV, 53.6%) to pain reproduction. As a note of caution, Carragee et al.¹²⁶ recently showed accelerated disc degeneration over 10 years in the discs that underwent discography using modern techniques.

Disc Herniations

Anular disc bulges and larger disc herniations contribute to spinal canal stenosis. Smaller or lateralized herniations, on the other hand, can cause radiculopathy. Disc extrusions are more likely to correlate with patient symptoms than disc bulges and disc protrusions.¹²⁷ There is a high incidence of radiculopathy in patients with imaging findings of central canal stenosis and nerve root compression by disc herniation.^127,128 The presence of a disc herniation in a patient with radiculopathy is a positive prognostic marker.^127,129 Most of the available studies have shown that the size of disc herniation can reduce dramatically on conservative management; the larger herniation shows a more prominent decrease in size. Extrusion and sequestered discs are also more likely to show a reduction in size; protrusions are less likely to change. There was, however, no clear correlation of these alterations with patient symptoms.^{9,8,130–132} The higher reduction and even disappearance of the extruded and sequestered discs is thought to be related to more vascular exposure.¹³¹ As far as surgery is concerned, most surgeons believe that larger disc herniations do better after surgery than smaller disc herniations. By the same token, larger discectomy defects predict lower rates of reherniation.¹³³ Recently, Dewing et al.¹⁰¹ showed significantly better outcomes in symptomatic patients after microdiscectomy in the following group of patients: younger, active, herniations at L5-S1 (compared with L4-5), and extrusions or sequestrations (compared with contained herniations). As far as patient preference for surgery is concerned, Lurie et al.¹³⁴ showed that younger patients having a lower level of education and higher disability/unemployment preferred the surgical option (Fig. 172-22).

FIGURE 172-22 Spontaneous resolution of disc herniation. A, A large inferior disc extrusion at L1-2 (arrow). B, Follow-up imaging at about 1 year afterward shows complete spontaneous resolution of the herniated disc. Also note redundant nerve roots caudal to the degenerative central canal stenosis at L2-3 (lower arrow).

End-Plate Marrow Changes

Type I and II changes are more common, whereas type III changes are relatively rare. The clinical significance is threefold: (1) correlation to back pain, (2) stability of end-plate changes, and (3) marker of segmental instability and fusion.

Segmental Instability

Type I changes have a higher correlation to segmental instability than type II changes do.¹³⁷ However, there is always a question about the methods for measuring instability. More convincing evidence of this correlation is shown by postfusion studies, which demonstrated a very high prevalence rather conversion into type II change in cases with solid fusion, whereas there was a higher type I change persistence in those with nonunion (pseudarthrosis)¹⁴⁵ (Fig. 172-23).

FIGURE 172-23 Postfusion type I to II end-plate conversion. Upper images demonstrate type I end-plate changes at L4-5. Patient underwent dorsal fusion (not shown) and decompression (arrows, lower images). Note conversion to type II end-plate change after surgery on follow-up imaging.

FIGURE 172-24 Facet joint instability and synovial cysts. A, Preoperative axial T2 image with moderate central canal stenosis at L4-5. Note bilateral facet joint effusions and small left ventral synovial cyst (arrow). B and C, Patient underwent decompression laminectomies at L4-5 without any fusion. Postoperative imaging showing progression of left-sided synovial cyst and development of a small ventral synovial cyst on the right side (arrows).

Facet Joint and Dorsal Elements

It is clinically important to identify pain originating from facet joints. Facet joint arthritis on imaging is a nonspecific finding usually accompanying disc degeneration. Dominant facet joint arthritis with or without effusion needs to be addressed in surgical planning. The relationship of joint effusions to spondylolisthesis and instability has already been discussed. Degenerative changes in the dorsal paraspinal structures on STIR are found in a higher percentage of subjects with LBP than in controls.⁵⁹ Lakadamyali et al.⁵⁹ showed a significantly increased incidence of facet joint effusions, para-articular soft tissue edema, and interspinous ligamentous edema in patients with LBP than in controls. Pedicle marrow changes are related to spondylolysis, as discussed earlier, but they are also commonly seen with facet degenerative disease and reflect abnormal stresses related to abnormal motion or loading caused by the degenerative changes in the spinal segment¹⁴⁶ (Fig. 171-24; see also Figs. 172-8 and 172-9).

Degenerative Spondylolisthesis

The natural history of degenerative spondylolisthesis (DS) is generally favorable. About 15% to 25% of patients eventually need surgery.^147,148 Up to 35% of patients show progression of DS, but there is no clear relationship between DS progression and change in patient symptoms. Matsunaga et al.¹⁴⁸ studied the long-term (>10 years) nonsurgical natural course of DS. The intervertebral spaces of the slipped segments decreased significantly in size during follow-up examination in patients in whom no progression was found. LBP improved following a decrease in the total intervertebral space size. The development of osteoarthritic spurs, hypertrophy and ossification of the intervertebral ligaments, and facet arthrosis are thought to cause stabilization that prevents slip progression. About 75% of patients remained asymptomatic after 10 years of follow-up. However, most of the patients (83%) who had significant initial symptoms such as intermittent claudication or vesicorectal disorder and who refused surgery showed progressive worsening. Johnsson et al.,¹⁴⁹ in a study of symptomatic patients over 4 years, showed that about 75% of the patients had stable spinal claudication. Overall, symptoms in 70% of the cases were unchanged, 15% showed improvement, and 15% worsened. No proof of severe deterioration was found after 4 years. Therefore, DS is clinically significant in a small portion of the cases, and the majority of DS on imaging remains stable over a long period of time. DS is a relative indication for instrumented fusion in the setting of spinal stenosis. Other indications are correction of deformity, recurrent spinal stenosis with instability, adjacent segment stenosis with instability, and multiple-level fusions.¹⁰⁹

Lumbar Spinal Stenosis

Lumbar canal stenosis is associated with a threefold higher risk of experiencing LBP.⁹⁶ However, the degree of lumbar spinal canal stenosis (LSS) on imaging is not strongly correlated with patient symptoms.^150,151 Also, imaging provides no clear means of management stratification and plays no prognostic role except in severe degrees of central canal stenosis, which has a fair correlation to functional disability and a favorable postoperative outcome.^150,152 Jonsson et al.¹⁵⁰ concluded that the “ideal patient for surgery” should have a pronounced constriction of the spinal canal, insignificant LBP, no concomitant disease affecting walking ability, and a symptom duration of less than 4 years. Regarding RNR in spinal canal stenosis, a recent study found a prevalence of about 33%; the longer the relative length of RNR, the better the outcome. Overall, the surgical outcomes in the RNR group were not statistically different from those in the non-RNR group.¹⁵³

Etiology and Risk Factors

The risk of vertebral osteomyelitis is increased in patients with IV drug abuse, endocarditis, sickle cell anemia, tuberculosis, malnutrition, alcoholism, diabetes mellitus, rheumatoid arthritis, as well as those undergoing hemodialysis and patients given steroids, due to compromise of host defenses.^157–159 The incidence of diabetes mellitus in patients with vertebral osteomyelitis is far out of proportion to that in the general population. Carragee et al. reported that up to 40% of patients had diabetes mellitus.¹⁵⁸

Hematogenous vertebral osteomyelitis is caused predominantly by Staphylococcus aureus. Up to half of these infections can be caused by methicillin-resistant S. aureus (MRSA).¹⁶⁰ S. aureus is also a major cause of bacteremia and endocarditis in adults. In a series of 37 adult patients with vertebral osteomyelitis from the Cleveland Clinic,¹⁶¹ S. aureus was the responsible organism in 68% of patients (25 of 37). Staphylococcus epidermidis, streptococci, gram-negative aerobic bacilli, and Mycobacterium tuberculosis were identified in 5% of the remaining patients. In IV drug abusers, Pseudomonas aeruginosa, other unusual gram-negative bacilli, and staphylococci cause spinal infection most frequently.^162,163 In elderly males with genitourinary tract infections, vertebral osteomyelitis is caused most often by Escherichia coli and Proteus spp.¹⁶² Furthermore, pyogenic vertebral osteomyelitis may be caused by a different organism than the one isolated from blood, urine, or sputum culture, as reported by Velan et al. in a series of three older patients.¹⁶⁴

Hematogenous

Both the venous and arterial vessels have been proposed as possible routes for the development of vertebral osteomyelitis.

In 1940, Batson¹⁶⁶ described a valveless, low-pressure vertebral venous system. This is composed of tributaries from the vertebral body metaphyses that drain into a large channel that exits the vertebral body via the nutrient foramen into the plexus of veins lining the spinal canal in the extradural space. There are also connections through the vertebral body cortex to a plexus of veins on the ventral and lateral aspects of the spine.¹⁶⁷ Batson¹⁶⁶ demonstrated rich anastomoses of this vertebral venous system with the thoracic and abdominal cavities and with pelvic veins. Slight increases in pressure in the vena caval system resulted in a diversion of blood into the vertebral venous system.

Septic embolization via the arterial system has also been proposed as a route for vertebral osteomyelitis. The arterial system of the spine forms a ladder-like anastomosis on its surface. The horizontal components are segmental arteries (1–3 mm) arising from the abdominal aorta that lie at the equator of the vertebral body, and two metaphyseal (200 mm) anastomosing arteries at each vertebral metaphysis. The vertical components join the segmental arteries to the metaphyseal anastomosing arteries, the metaphyseal anastomosing arteries of adjacent vertebral arteries, as well as metaphyseal anastomosing arteries of the same vertebral body. The intraosseous arteries arise from the segmental and metaphyseal anastomosing arteries and parallel the disc surface. Children have rich anastomoses of the intraosseous arteries, which involute by adulthood. Thus, in an adult, infection of the disc cannot occur directly via hematogenous means. It can occur via contiguous spread from an adjacent infected vertebral body. Ratcliffe¹⁶⁸ has proposed that septic emboli were more likely to enter the greater number of arteries at the metaphysis and that the smaller caliber of the vessels allowed for the development of thrombosis with resultant infarction—a requirement for osteomyelitis to develop. These two observations explain the propensity for vertebral osteomyelitis to involve the subchondral aspect of the body with relative sparing of the central portion of the vertebral body. The vertical anastomoses among the metaphyseal anastomosing arteries allowed for the spread of infection across disc spaces as well as within the same vertebral body while sparing the central portion. The intraosseous anastomoses in children would prevent large foci of bony infarction, thus resulting in smaller foci of osteomyelitis.¹⁶⁸

Whereas metastatic spread from the pelvic venous system to the vertebral bodies via the vertebral venous plexus has been shown in animals,^169,170 the spread of infection via the vertebral plexus could not be demonstrated experimentally. Further argument supporting arterial spread is that vertebral osteomyelitis is usually preceded by symptomatology compatible with bacteremia (sepsis). If spread is via the venous route, extradural thrombophlebitis with meningismus should be expected to represent a prominent feature of osteomyelitis.¹⁷¹ This has not proven to be the case.

Iatrogenic or Posttraumatic

Organisms can be directly implanted secondary to surgical procedures or penetrating spinal trauma. Discitis has been reported as a complication of cervical and lumbar discography,^172,173 chemonucleolysis (chymopapain injection),^174,175 and laminectomy.^176,177 Discitis as a complication of discography is uncommon. Its incidence has been reported as 0.05% to 1.4% in various series. An aggravation or increase of neck and back pain and fever are typical presenting complaints.^172,173

The diagnosis of disc space infection following discectomy is problematic. Patients typically present 2 to 6 weeks following surgery with recurrence of pain after initially having experienced the relief of symptoms. Patients are often febrile with muscle spasm and a positive straight-leg raising test.¹⁷⁶ The sedimentation rate is elevated. Roentgenographic findings appear several weeks after initial symptoms, with a decrease in the intervertebral disc space height and development of end-plate erosions.

Contiguous Spread

Vertebral osteomyelitis, or disc space infection, can result from contiguous spread from an adjacent infected source. It is the least common of the three mechanisms discussed here.¹⁷⁸ Contiguous spread requires the spread of a superlative soft tissue focus through the periosteum, cortex, and marrow of the vertebral body or through the longitudinal ligaments, anulus fibrosus, and nucleus pulposus of the disc. Contiguous spread of infection to the spine is encountered in the sacral region in association with infected decubitus ulcers, as well as genitourinary and intestinal processes.¹⁷⁸ Vertebral osteomyelitis and disc space infection have been reported as complications of colonic,¹⁷⁹ hypopharyngeal,¹⁸⁰ and esophageal^181,182 perforations or instrumentation, as well as pelvic abscesses.¹⁸³ Lesions of the aorta or of an aortic graft, such as mycotic aneurysm or pseudoaneurysm, may predispose to the development of vertebral osteomyelitis.¹⁸⁴ When considering vertebral osteomyelitis as a result of contiguous spread, the question often arises as to whether the infection was actually a result of hematogenous spread (either arterial or venous) rather than truly contiguous with direct invasion.^185–187 For example, the direct invasion of a vertebral body by a lung lesion is felt to be an unlikely sequence of events due to an effective barrier provided by the pleura.¹⁸⁵ Here it has been postulated that spread actually occurs via pleural veins communicating with the Batson plexus.

Conventional Radiography

Typically, the early findings of disc space infection on plain radiography are minimal disc space narrowing and erosions or indistinctness of the end plates. In addition to the bony changes, soft tissue swelling due to paraspinal phlegmon or abscess may also be evident. In the lumbar region, this presents as enlargement of the psoas shadows. In the thoracic region, disc space infection appears as a paraspinal soft tissue mass. In the cervical spine, prevertebral soft tissue swelling is evident. As the disease progresses, the disc space narrowing worsens. The destruction of the end plates becomes more obvious. In addition, sclerosis of the end plates and a periosteal reaction also become evident. In general, with healing, the disc space remains markedly narrowed or fuses, and there may be a concomitant loss of height of the vertebral body with resultant spinal deformity (see Fig. 172-31C).

Digby and Kersley¹⁸⁸ reported the following time sequence for the radiographic changes to correlate with the pathologic changes: at 2 weeks, decreased disc space height becomes evident; at 6 weeks, lytic vertebral body lesions are seen; reactive sclerosis is present at 8 weeks; and at 6 months, new bone formation and fusion are present. Malawski and Lukawski¹⁵⁶ also reported the time sequence of radiographic changes in a series of 150 cases of vertebral osteomyelitis. They found that radiologic changes became apparent after 4 to 6 weeks. In their series, an early sign was also a decrease in disc space height. After several weeks, lytic vertebral body lesions and end-plate disruption were evident.

Unique presentations of vertebral osteomyelitis have also been reported. McHenry et al.¹⁸⁹ reported a series of six patients with osteomyelitis who presented with underlying osteoporosis and compression fracture of a single vertebral body. On the initial plain radiographs, the end plates were intact. This unique presentation accounted for 13% of all hospitalized patients with vertebral osteomyelitis and 2.4% of inpatients with osteoporotic compression fractures over a 5-year period in our institution.

Three-Phase and Total-Body Bone Scan

In the setting of vertebral osteomyelitis, the three-phase bone scan demonstrates increased uptake in flow and blood pool phases and in delayed images. Intense uptake in two adjacent vertebral bodies with loss of the disc space is seen in patients with vertebral osteomyelitis. The bone scan is sensitive for vertebral osteomyelitis. However, it lacks specificity. In the spine, it can be difficult to separate increased radionuclide uptake due to degenerative disc disease, benign compression fracture, or metastatic disease from that occurring as a result of vertebral osteomyelitis. Moreover, some investigators report bacteriologically proven vertebral osteomyelitis in patients who had a negative bone scan at 2 weeks following the onset of symptoms.^192,193 In addition, the increased uptake of tracer does not differentiate active from inactive osteomyelitis, since increased uptake can persist for long periods following resolution of the infection.¹⁹⁴

Gallium-67

Gallium scan also demonstrates increased uptake in the setting of vertebral osteomyelitis. It also gives information about the surrounding soft tissues. The combination of gallium scanning and a three-phase bone scan may increase the specificity up to nearly 100%.¹⁹² The sensitivity is approximately 90%. The two tests provide complementary information.¹⁹⁵ It is generally accepted that the combination of bone and gallium imaging can identify vertebral osteomyelitis when (1) the distribution of the two tracers is spatially incongruent or (2) their distribution is spatially congruent and the relative uptake of gallium exceeds that of technetium.¹⁹¹ The combination of the two studies is currently the nuclear medicine study of choice for vertebral osteomyelitis. Gratz et al.¹⁹⁶ were able to demonstrate that gallium-67 (⁶⁷Ga)-citrate single-photon emission tomography (SPECT) activity correlated with the severity of infection in a small series of patients who had been receiving antibiotic treatment for an average of 7 weeks prior to scanning. The sensitivity in detecting vertebral osteomyelitis was 80% for individual foci and 100% for vertebral osteomyelitis. In addition, ⁶⁷Ga-citrate SPECT detected foci of infection outside of the spine before other imaging techniques demonstrated these¹⁹⁶ (Fig. 172-25).

FIGURE 172-25 Vertebral osteomyelitis on gallium and 2-[¹⁸F] fluoro-2-deoxy-d-glucose/positron emission tomography (FDG-PET). Coronal images from FDG-PET and ⁶⁷Ga single-photon emission computed tomography (SPECT). Increased radiotracer accumulation (white arrows) at L5-S1 due to discitis osteomyelitis.

(From Love C, Tomas MB, Tronco GG, et al: FDG PET of inflammation and infection. Radiographics 25(5):1357–1368, 2005, with permission.)

2-[¹⁸F] Fluoro-2-Deoxy-d-Glucose Positron Emission Tomography

With the exception of combined bone scan–gallium imaging, nuclear medicine imaging for vertebral osteomyelitis has been unsatisfactory. 2-[¹⁸F] fluoro-2-deoxy-d-glucose positron emission tomography (FDG-PET) has shown higher sensitivity and specificity in the diagnosis of vertebral osteomyelitis than has MRI.^197,198 FDG-PET has proven useful for differentiating degenerative and infectious end-plate abnormalities detected on MRI.¹⁹⁸ FDG imaging findings are likely due to increased glucose metabolism in inflammatory cells such as leukocytes, granulocytes, and macrophages. An additional advantage is that FDG-PET is not affected by metal implants. It may be useful in evaluating infection in patients with hardware. FDG-PET has better resolution than other nuclear medicine imaging techniques and can differentiate bone from soft tissue infection. Low FDG-PET uptake has been shown in fractures and pseudarthrosis. Hodges et al.¹⁹⁹ identified persistent disease on MRI as a sign of failure of conservative treatment. Recently, Kim et al.²⁰⁰ showed the utility of FDG-PET in differentiating between residual and nonresidual disease based on standard uptake value measurements. FDG-PET may therefore be used in cases of vertebral osteomyelitis as a problem-solving tool in proper clinical settings (see Fig. 172-25).

Type I Degenerative Marrow Changes

Type I degenerative marrow changes may mimic vertebral osteomyelitis. Type I marrow signal intensity change may enhance following the administration of gadolinium, and enhancement of the central portion of the cervical and lumbar intervertebral discs in patients who have not undergone surgery does occur in the setting of degenerative disc disease.²³⁷ In advanced disc degeneration, blood vessels once again can be seen within the disc. Usually, the signal intensity of the disc on T2-weighted images helps differentiate the two entities. The disc is usually decreased in signal on T2-weighted images in degenerative disc disease. On occasion in degenerative disc disease, the intervertebral discs may show increased T2 signal, as discussed in the following section. In infectious spondylitis, the distribution of the increased signal intensity on T2-weighted images does not conform to the anatomic configuration of the disc. The clinical history and the presence or absence of increased sedimentary rate help to differentiate the two entities. Only 4% of patients in a series of 474 patients with degenerative disc disease demonstrated the type I pattern²¹⁵ (Fig. 172-32).

FIGURE 172-32 A, T2 hyperintense degenerative disc. Patient presented with intractable back pain. B, Hyperintense T2 signal in the L5-S1 disc space (arrow), better demonstrated on short-tau inversion recovery imaging (STIR), represents disc hydration. Note type I end-plate changes and disc space narrowing (noncontrast T1 image not shown). C, The type I end plates enhance after contrast administration.

T2 Hyperintense Degenerated Disc

A T2 hyperintense degenerated disc is not an uncommon presentation. Typically, a diffuse T2 hyperintense disc, disc space narrowing, variable end-plate changes, and degenerative disc disease are evident at other levels. Generally, this does not represent a major challenge, but in a high-risk patient for whom there is strong clinical suspicion for vertebral osteomyelitis, contrast enhancement may be needed to further characterize the disc space. Lack of paraspinal soft tissue edema strongly argues against infection (see Fig. 172-32).

Metastatic Disease

Metastatic disease can be differentiated from vertebral osteomyelitis by its sparing of the intervertebral disc space. Disc involvement with metastatic disease is rare.^238,239 In addition, metastatic disease commonly involves the pedicles of the vertebral body, whereas osteomyelitic foci tend to be subchondral and do not affect the pedicle.

Primary Tumors of the Spine

Primary tumors of the spine are rare compared with metastatic tumors. There are, however, primary tumors of the spine that involve contiguous vertebral body levels and may therefore have disc involvement. Such tumors include plasmacytoma,²⁴⁰ aneurysmal bone cysts,²⁴¹ giant-cell tumors,²⁴² and chordoma.²⁴³ Eosinophilic granuloma presents with lytic vertebral body lesions that can progress to nearly complete destruction of the vertebral body or vertebral plana. Despite this, the adjacent intervertebral discs usually remain intact, allowing for differentiation from disc space infection. Aneurysmal bone cysts and giant-cell tumors of the spine can be characterized by lytic lesions that can span vertebral body levels. However, these tumors usually involve the dorsal elements. These tumors may be more difficult to distinguish from tuberculous spondylitis than from pyogenic spondylitis. Chordomas are uncommon tumors derived from notochordal elements. They can be found throughout the spinal column with a marked propensity for the clivus and sacrococcygeal regions. They are isointense or hypointense in signal on T1-weighted images and increased in signal on T2-weighted images. The intervertebral disc and adjacent vertebral bodies are commonly affected. A paraspinal soft tissue mass is often present.²⁴³

Spinal Neuroarthropathy (Charcot Spine)

Spinal neuropathy, or Charcot spine, is most commonly related to tabes dorsalis, syringomyelia, or diabetes mellitus and can be difficult to differentiate from infectious spondylitis on plain films and CT. Park et al.²⁴⁴ found MRI to be useful for distinguishing between the two entities. MRI in neuropathy demonstrated hypointensity in the abnormal disc and adjacent vertebral bodies on both T1- and T2-weighted sequences. In neuroarthropathy, the disc space and surrounding marrow were of lower signal intensity on T2-weighted images than that seen in spinal osteomyelitis. Wagner et al.²⁴⁵ evaluated the plain film, CT, and MRI findings in 14 patients with spinal neuroarthropathy versus 19 patients with disc space infection. They found that facet involvement with narrowing or erosions and vacuum disc were more common in spinal neuroarthropathy on imaging studies and rare in disc space infection. Vertebral body spondylolisthesis, osseous joint debris, joint disorganization, disc rim enhancement, and diffuse vertebral body enhancement were more frequent imaging findings in spinal neuroarthropathy (Fig. 172-33).

FIGURE 172-33 Charcot spine. Patient with long-standing spinal cord injury. L5-S1 erosive and hypertrophic sclerotic changes on plain film (A) with corresponding hypointense T1 (B) and T2 (C) spondylodiscal signal change.

Destructive Spondyloarthropathy

Destructive spondyloarthropathy (DSA) has been described in patients undergoing long-term hemodialysis and, more commonly, in the lower cervical spine.^246–248 Plain film changes include disc space narrowing and irregular end-plate destruction. CT images reveal osteolytic areas, with bone sclerosis of adjacent vertebral end plates and minimal osteophytosis.²⁴⁹ This entity typically demonstrates decreased signal in the disc on T2-weighted images and absence of an associated soft tissue mass that helps to distinguish it from vertebral osteomyelitis.²⁵⁰ With progression of the disease, collapse of a vertebral body and spinal instability may occur.²⁴⁹ Distinction between these two entities may be difficult since the sedimentation rate is typically elevated in both groups. Intervertebral disc biopsy may be required in some cases (Fig. 172-34).

FIGURE 172-34 Destructive spondyloarthropathy. Patient with chronic renal disease and long history of hemodialysis. There was no clinical evidence for spine infection. Sagittal reconstructed images from CT abdomen and pelvis; no further imaging was needed. Note osteolysis of the end plates at two levels in thoracic spine resulting in vertebral plana (arrows) with no evidence of hypertrophic sclerotic changes. Also note typical bandlike end-plate sclerosis (“rugger-jersey appearance”).

Ankylosing Spondylitis

Ankylosing spondylitis (AS) can mimic discitis osteomyelitis due to inflammatory and noninflammatory processes. Ventral and dorsal end-plate focal inflammatory changes (Romanus lesions) are multifocal and limited to the bones. Imaging shows focal changes at the ventral and dorsal end-plate margins. Andersson lesions represent rheumatic spondylodiscitis. The imaging features depend on the stage of the process. Early findings include end-plate erosion and hypointense T1 and hyperintense T2 signals. Later on, there is development of sclerosis and hyperintense T1 signal due to fatty marrow change. Differentiation from discitis osteomyelitis is usually not difficult in these lesions due to multilevel involvement, localized nature of the lesions, and lack of paraspinal soft tissue thickening.^251–253 Segmental pseudarthrosis can develop in AS following fracture of the spine and can better mimic disc space infection.²⁵⁴ It is usually seen at the thoracolumbar junction. On MRI, the segmental pseudarthrosis exhibits an increased signal on T2-weighted images. Shih et al.²⁵³ showed both T2 hyperintense and T2 hypointense discovertebral changes in fractures and pseudofractures in AS. Associated segmental dorsal element involvement and disruption of the thickened anterior longitudinal ligament (ALL) are specific features of AS (Fig. 172-35).

FIGURE 172-35 Ankylosing spondylitis. A, Ventral and dorsal margins spondylotic ankylosing changes on CT with ventral sclerotic (“shiny”) corner (arrow). B and C, Follow-up MRIs show anterior Romanus lesions of different stages: T1 hypointense (B; straight arrows), T2 hyperintense (C; straight arrows), and T1 hyperintense, which demonstrate fat-suppressed hypointense signal on short-tau inversion recovery (C; dashed arrows). Note ventral compression deformity (curved arrows). No Andersson lesions are present.