Early Imaging Recommendations

Based on such data, there is broad international consensus on the lack of efficacy of imaging in acute pain syndromes of spinal origin. The American College of Radiology^11,12 considers that imaging is inappropriate in acute low back pain unless there are complicating (“red flag”) features including the following:

Similarly, the Agency for Healthcare Research and Policy (U.S.) guidelines advises no imaging in patients younger than 50 years of age in the absence of systemic disease, or until pain persists for at least 4 weeks.¹³ The Australian National Health and Medical Research Council recommends against imaging in nonspecific low back pain of less than 12 weeks’ duration in the absence of red flag features suggesting underlying systemic disease.¹⁴ The American College of Physicians and the American Pain Society released a joint guideline in 2007 stating that imaging should not be obtained in patients with nonspecific low back pain unless there is a severe or progressive neurologic deficit, or when serious underlying systemic disease is suspected.¹⁵ Furthermore, patients with signs or symptoms of radiculopathy or spinal stenosis should be imaged with MRI (preferred) or CT only if they are candidates for surgery or epidural steroid injection.¹⁵

The primary role of imaging in the back pain patient is to aid in the detection of systemic disease that is causal of back or back-related leg pain. In essence, imaging is undertaken to detect neoplasm, infection, or manifestations of unsuspected traumatic injury. In a patient presenting with back pain to a primary care physician, these conditions are highly uncommon. In this population, only 0.7% of patients will have metastatic neoplasm, 0.01% spine infection, 0.3% ankylosing spondylitis, and 4% will have osteoporotic compression fractures.¹ The differential diagnosis of back pain, and the relative frequency of causal conditions, are displayed in Table 43-1.¹ The imperfect sensitivity and specificity of all imaging modalities makes diagnostic accuracy problematic, given the low pretest probability of systemic disease. According to Jarvik and Deyo, there are three essential questions that must be answered for the patient with back pain: (1) is there evidence of underlying systemic disease? (2) Is there neurologic impairment requiring surgical intervention? (3) Are social/psychological factors exacerbating the pain? Imaging may help to identify systemic disease and anatomic neural impingement, but its role should not be overstated.¹ As Jarvik and colleagues note, depression is a more significant predictor of new low back pain than MRI imaging findings.¹⁶

Table 43-1 Differential Diagnosis of Low Back Pain^∗

∗ Diagnoses in italics are often associated with neurogenic leg pain. Figures in parentheses indicate estimated percentage of patients with these conditions among all adult patients with signs and symptoms of low back pain. Percentages may vary substantially according to demographic characteristics or referral patterns in a practice. For example, spinal stenosis and osteoporosis will be more common in geriatric practices and spinal infection will be more common in injection drug users.

† The term mechanical is used here to designate an anatomic or functional abnormality without an underlying malignant, neoplastic, or inflammatory disease.

‡ Strain and sprain are nonspecific terms with no pathoanatomic confirmation. Idiopathic low back pain may be a preferable term.

§ Because spondylolysis is equally common in asymptomatic persons and those with low back pain, its etiologic role remains ambiguous.

¶ Internal disc disruption is diagnosed by provocative discography (injection of contrast material into a degenerative disc, with assessment of pain at the time of injection). However, discography often generates pain in asymptomatic adults, and many patients with positive discogram results improve spontaneously. Thus, the significance and appropriate management of this disorder remain unclear. Discogenic back pain is often used synonymously with internal disc disruption.

∗∗ Presumed instability is loosely defined as >10 degrees of angulation or 4 mm of vertebral displacement on lateral flexion and extension radiographs. However, diagnostic criteria, natural history, and surgical indications remain controversial.

Data obtained from Deyo,⁴ Hart et al.,⁶ Deyo et al.,⁷ and Deyo et al.⁸ Reproduced with permission Deyo RA, Weinstein JN: Low back pain. N Engl J Med 2001;344:363-370. Copyright© 2001. Massachusetts Medical Society. All rights reserved.

Risk/Benefit Analysis

As with any medical procedure, a decision to proceed with imaging of the spine should involve a risk/benefit analysis. Certainly patients will benefit from imaging when it aids in the identification of previously undiagnosed systemic disease. In the setting of progressive neurologic deficit, imaging can identify the neural compressive lesion and assist in planning surgical or minimally invasive interventions. Pain physicians performing interventional procedures must scrutinize available imaging to best plan access routes for medication or energy delivery. There should be benefit to the patient through reassurance in confirming the absence of sinister pathology. This is dependent on the pain physician actively educating the patient regarding the significance, or inconsequence, of imaging findings. Benefit should also accrue from stopping any further work-up. In patients whose pain becomes chronic, imaging may help in identification of the cause of chronic pain and allow institution of therapy specific to that process.

There are, however, risks involved in imaging. These include the labeling effect, radiation exposure, monetary cost, and the provocation of intervention. The labeling effect refers to the inevitable identification of degenerative phenomenon on any spine imaging test. Degenerative findings on spine imaging are almost always inconsequential. If this is not made very clear to patients, they may identify themselves as suffering from a degenerative process from which there is no ultimate recovery. This can lead to fear avoidance behaviors, deconditioning, and depression. The pain practitioner is obligated to educate the patient that degenerative findings seen on imaging have little if any prognostic value, and actively contradict common misconceptions regarding back pain. A recent Cochrane review concluded that intensive patient education is effective in patients with acute and subacute low back pain; its value in chronic pain patients is less clear.¹⁷ The Australian media studies of Buchbinder also document the effectiveness of carefully planned educational efforts in changing patient and physician attitudes and reducing disability related to back pain.¹⁸

Radiation exposure from radiographs or CT must also be carefully considered. The effective radiation dose (biologic effect) is measured by the Sievert (Sv). The average natural background exposure is approximately 3.0 mSv per year.¹⁹ A three- view lumbar spine radiographic series exposes the patient to approximately 1.5 mSv; lumbar spine CT carries an exposure of approximate 6 mSv. An abdominal and pelvic CT delivers an average of 14 mSv; a technetium bone scan results in an average of 6.3 mSv exposure. Frontal and lateral chest radiographs, by contrast, expose the patient to 0.1 mSv.¹⁹ Radiation exposure is cumulative; repeated radiographic or CT examinations without clear expectation of patient benefit through improved decision-making cannot be condoned.

Imaging is costly. In the United Kingdom, imaging is known to comprise approximately 5% of the direct medical costs related to back pain.²⁰ This is likely significantly higher in the less regulated U.S. marketplace. The 2009 Medicare reimbursements for lumbar spine imaging in the United States are: lumbar radiographs, $36; lumbar spine CT (noncontrast), $241; CT myelography, $525; MRI (noncontrast), $402, and bone scan with SPECT, $239.²¹ Nominal fees may be three to five times Medicare reimbursements. These costs directly affect our patients and our society and should be incurred only through reasoned decision-making.

Finally, imaging is likely to provoke intervention. Jarvik noted that early MRI imaging of the spine leads to increased surgical interventions despite equivalent pain and disability profiles when compared with nonimaged patients.²² Similarly, a study by Lurie and associates found that the intensity of CT and MRI use can account for virtually all of the extensive (12-fold) regional variation in surgical rates for spinal stenosis in the United States.²³ When we image, we intervene. This occurs with standard of care interventions (many of which have little evidence basis) and totally unproven interventions with significant risks that have found their way into our highly imperfect medical marketplace. The single most important imaging precept in the back pain patient is this: treat the patient, do not treat the images.

Radiographs

In the setting of clinical red flags suggestive of systemic disease, pain unresponsive to conservative measures, or progressive neurologic deficit, imaging should commence with plain radiographs.¹¹ Radiographs are a modest sensitivity screening tool for sinister processes such as neoplasm, infection, or fracture. They also provide vertebral enumeration, and allow assessment of sagittal and coronal balance. To provide the physiologic information of balance, radiographs should be obtained while the patient is weight-bearing. Single frontal and lateral views of the lumbar spine are adequate. Spot images of the lumbosacral junction, oblique views, and flexion-extension views should not be obtained on a routine basis. They provide little additional information, and significantly increase radiation exposure.⁴³ The significance of vertebral enumeration should not be underestimated. There are anomalies of segmentation at the lumbosacral junction in 7% to 12% of the back pain population.⁴⁴ Radiographs should establish the convention of vertebral numbering for all subsequent advanced imaging in the individual patient, preventing wrong level minimally invasive or surgical interventions.

Radiographic identification of morphologic changes of disc degeneration or osteoarthritis in lumbar zygapophyseal or sacroiliac joints has no value in identifying the cause of an individual patient’s pain.⁴⁵ Radiographic manifestations of disc degeneration include loss of disc space height, nitrogen gas within the disc, and end plate hypertrophs; these findings have no predictive value for discogenic pain. Similarly, radiographic findings of osteoarthritis, including hypertrophic change, sclerosis, joint space narrowing, gas within the joint space and subchondral cyst formation do not correlate with response to intraarticular anesthetic injections. Morphologically degenerated joints are most commonly asymptomatic, and joints that appear normal radiographically may be painful.

Radiographs provide a modest sensitivity screen for systemic disease causal of back pain, which may require advanced imaging to characterize and establish extent of disease (Fig. 43-1). The most common neoplastic condition is metastatic disease. Skeletal metastases are primarily blastic (bone forming), lytic (bone destroying), or of mixed type. Radiographs primarily image cortical bone. Metastases most commonly involve trabecular bone; at least 50% of trabecular bone mass must be destroyed to be visible on a radiograph. Radiographs therefore have only modest sensitivity (60%) in the detection of metastatic disease.¹ Myeloma may also be initially suggested by radiographic evaluation; myeloma typically manifests as small discrete lytic lesions without sclerotic margination. When diffuse, myeloma may be indistinguishable from metabolic osteoporosis. Primary bone tumors are rare causes of spine-related pain. Osteoid osteoma deserves mention because it commonly presents with nocturnal pain.⁴⁶ This neoplasm of adolescents and young adults has a posterior element predilection, and it generates a lesion consisting of a central lytic nidus with considerable surrounding sclerotic reaction. The diagnosis is seldom made with plain films; CT imaging, with image directed biopsy and radio frequency ablation constitute standard of care. Osteoid osteomas may also be identified on MRI examinations or bone scans as focal intensely enhancing (or metabolically active) posterior element lesions.⁴⁷

Figure 43-1 Radiographic detection of underlying conditions causal of spine or limb pain. Radiograph (A) and CT image (B) of patient with absent left T12 pedicle, shown to be aneurysmal bone cyst on CT. Radiograph (C) and T1 weighted MRI (D) of patient with spotty demineralization on radiograph, prompting the MRI revealing the innumerable lesions of multiple myeloma. Radiograph (E) and CT (F) of patient with L5 pars defects. Lateral radiograph (G) and CT image (H) of patient with axial pain. Radiograph suggests destruction of the anterior-inferior aspect of L3 body, confirmed on CT with associated paraspinal mass compatible with pyogenic spondylodiscitis.

Radiographs may also suggest infectious or inflammatory disease. Radiographic findings of infectious spondylodiscitis (disc space infection) only become apparent several weeks after the onset of disease.⁴⁸ Because disc space infection in the adult begins as a hematologically seeded vertebral osteomyelitis, radiographs initially demonstrate rarefaction in the anterior subchondral bone adjacent to the end plates within a single disc space. Over time, there will be frank end-plate destruction with loss of disc space height.⁴⁸ These findings should provoke an urgent MRI examination with gadolinium enhancement. Granulomatous disease (TB, brucellosis) may be indistinguishable from pyogenic disc space infection. Granulomatous discitis more commonly spares the disc space with vertebral collapse and kyphotic deformity at the site of infection.⁴⁸

Inflammatory spondyloarthropathies may be suggested or diagnosed with plain radiographs. Rheumatoid arthritis primarily involves the cervical spine, with inflammatory destruction of the transverse ligament binding the dens to C1 with consequent atlantoaxial subluxation.⁴⁸ There is frequently multilevel cervical disc space narrowing with low grade anterior subluxations resulting in a stair-step appearance to the cervical spine. The seronegative spondyloarthropathies are more commonly apparent in the sacroiliac joints, lumbar and thoracic regions (Fig. 43-2). Narrowing of the sacroiliac joints with associated indistinct cortical surfaces and adjacent sclerosis may herald ankylosing spondylitis, psoriatic arthritis, or Reiter syndrome. The ankylosing spondylitis patient may also have fusion of portions of the thoracolumbar or cervical spine with vertically oriented, gracile syndesmophytes bridging adjacent vertebral bodies. The long-segment spine rigidity in these patients makes them susceptible to catastrophic fracture dislocation, often with neurologic compromise, even with modest trauma. Such fractures may be difficult to visualize with plain radiographs, and the threshold for use of advanced imaging should be very low in ankylosing spondylitis patients suffering trauma.⁴⁹

Figure 43-2 Ankylosing spondylitis (AS). Pelvic radiograph (A), axial CT (B), and coronal T1 weighted MRI image (C) all show manifestations of AS, with indistinct sacroiliac joints with adjacent sclerosis dominant on the iliac side of the joint. MRI shows patchy mixed signal abnormality about the sacroiliac joints. In another patient, sagittal CT images (D and E) show gracile syndesmophytes bridging the lumbar vertebral bodies. The paramedian image (E) demonstrates fusion of the zygapophyseal joints.

Compression fractures can be identified on plain radiographs, but the acuity of a fracture may not be apparent in the absence of serial images. MRI or technetium bone scan can characterize fractures as acute or subacute, metabolically active, and therefore more likely to respond to bone augmentation procedures such as vertebroplasty (Fig. 43-3). However, recent randomized controlled trials have raised questions regarding the efficacy of vertebroplasty.^50,51 Similarly, pelvic insufficiency fractures may be apparent on radiographs within the sacral ala, pubic rami, pubic symphysis, and periacetabular regions of the pelvis. The plain film finding of a vertical sclerotic band paralleling the sacroiliac joint in the sacral ala strongly suggests insufficiency fracture. CT, technetium bone scan, or MRI are all more sensitive than radiographs in detection of insufficiency fractures.^52,53

Figure 43-3 Fracture Chronicity. Elderly osteoporotic patient presents with acute axial upper lumbar pain. Radiograph (A) demonstrates compression fractures of L1 and L2 vertebral bodies. No prior films were available to assess chronicity of fractures. On technetium bone scan (B), only the L2 fracture shows increased uptake of tracer. The L2 body was targeted for bone augmentation (C) with good clinical response.

Another traumatic entity in which plain films play a role is spondylolysis. Defects in the pars interarticularis may be detected on lateral plain films, with a prevalence of approximately 3% to −10% of the population.⁵⁴ Most cases are asymptomatic. If pars defects are suspected as the cause of an axial pain syndrome, a localized CT will confirm or refute the presence of such a defect and aid in characterizing the defect as acute or chronic. There is no role for oblique radiographs, which double the gonadal radiation dose over the standard images, and do not advance the diagnosis.⁴³

Computed Tomography

Computed tomography has undergone a technologic renaissance in the last decade with the evolution of multidetector scanners. Even the most uncomfortable patient can tolerate the few seconds necessary to obtain a high quality data set, which can then be reconstructed in any plane without loss of spatial resolution. CT provides superior imaging of cortical and trabecular bone. It has reasonable contrast resolution, allowing the identification of most root compressive lesions (Fig. 43-4). It is less costly than MRI, more widely available, and is not subject to imaging disqualification due to indwelling magnetically susceptible devices.

Figure 43-4 CT demonstration of disc herniation. Middle-aged male with acute L3 radicular pain pattern, unable to tolerate MRI. CT images (A, B, and C in descending sequence) well demonstrate the left foraminal disc extrusion displacing the L3 root posteriorly. In another patient, CT images D and E show a large left central and subarticular L5 herniation deforming the thecal sac. There is an associated osseous fragment of end plate accompanying the herniation.

CT remains inferior to MRI in contrast resolution and cannot identify intrathecal pathology. It is less sensitive than MRI to early inflammatory or infectious processes, and to marrow infiltrative disorders, specifically primary or secondary malignancies. The significant radiation dose incurred by CT must also be weighed in decision making.

As with plain radiographs, morphologic changes of disc degeneration, zygapophyseal joint arthrosis and sacroiliac joint osteoarthritis seen on CT do not correlate with pain syndromes, and are of no significance.^55,56 Disc herniations are readily identifiable with computed tomography. There is no literature comparing current generation CT scanners with MRI in the detection of root-compressive lesions. A study by Thornbury in 1993 showed little difference in MRI, CT myelography, or CT alone in detecting root-compressive disease.⁵⁷ Spine MRI has advanced modestly in the intervening 15 years; CT has been reborn. A more recent study from 2006 showed no evidence that CT is inferior to MRI in detecting disc herniation, although there was more interobserver variability with CT.⁵⁸ This study used two-slice CT scanners; current scanners in general practice are 64-slice scanners with 128-slice machines being introduced. It can be argued that CT is underused in evaluating the younger patient with root-compressive symptoms and no clinical likelihood of systemic disease. In this setting, CT can provide adequate imaging at a lesser cost than MRI. This is, of course, dependent on the local medical ecosystem; if the patient is a surgical candidate and the surgeon involved demands an MRI for planning purposes, then CT becomes an additive rather than a lower cost substitute test.

CT is more sensitive than plain radiographs, but less sensitive than MRI, in the detection of infectious or inflammatory spinal lesions.¹ Discitis may be suggested when end-plate destructive changes are seen, often accompanied by a paraspinal inflammatory mass.⁵⁹ CT alone may be insufficient to assess central canal compromise, and CT myelography or MRI may be necessary to assess the need for surgical intervention. Noninfectious inflammatory spondyloarthropathies are more conspicuous on CT than on plain films.⁶⁰ Like radiographs, however, CT is limited to displaying static morphology and cannot assess physiologic inflammation.

Traumatic or insufficiency fractures are more apparent on CT than on radiographs, particularly with the assistance of multiplanar reconstructions. Assessment of fracture acuity with CT requires evaluation of the fracture line and fragments. Very angular fracture lines and fragments with no marginal sclerosis suggest acuity, whereas smooth remodeling of bone implies chronicity. MRI or technetium bone scans provide more specific, physiologic assessment of fracture acuity. When fatigue fractures of the pars interarticularis are suspected on plain radiographs, CT can assist in further characterization of the fractures and the prognosis for healing (Fig. 43-5). If sclerosis is present within the pars without a discrete defect, this represents a stress reaction with healing potential. An ill-defined lucent zone with marginal sclerosis is an incomplete fracture—also with healing potential.⁶¹ A chronic defect with well-defined rounded and sclerotic margins has no healing potential. The combination of SPECT technetium scan with CT is generally considered the standard of care for complete characterization and treatment planning of spondylotic defects.⁶¹

Figure 43-5 CT characterization of pars interarticularis insufficiency fractures. CT image (A) demonstrates sclerosis in the pars interarticularis consistent with stress response and healing potential. CT image (B) shows incomplete pars defects with ill-defined linear lucent zones bilaterally surrounded by modest sclerosis. This lesion also has healing potential. In contrast, CT image (C) is that of a chronic insufficiency fracture with a broad well-defined lucent zone. There may be dense marginal sclerosis. This appearance lacks healing potential.

Bony destruction or blastic response as a manifestation of tumor is much more evident on CT than on radiographs; MRI, however, is even more sensitive.¹ CT exceeds MRI in its ability to assess the degree of structural compromise caused by neoplastic spinal involvement and, therefore, is critical in determining if surgical stabilization is necessary. Neoplasms that infiltrate marrow without trabecular or cortical destruction, such as lymphoma, may be difficult to identify with CT alone. Neoplastic soft tissue lesions involving the central canal or paraspinal regions may be detected by CT, but will be far more conspicuous on MRI.

CT Myelography

The demise of CT myelography has been predicted for more than 2 decades; it nonetheless remains a viable imaging adjunct with a problem-solving role in the lumbar region and a dominant role in the cervical spine (Fig. 43-6).⁴⁸ The addition of radio-opaque contrast material to the subarachnoid space allows detection of subtle epidural lesions of benign or malignant etiology, which encroach on the thecal sac and root sleeves. Intrathecal lesions may also be identified, although they are less well characterized than with MRI. The spatial resolution of CT myelography exceeds that of MRI, a critical parameter in the tight confines of the cervical spine. CT myelography excels at the discrimination between soft disc and bony osteophyte, which may alter the surgical approach in the cervical spine. Myelography is minimally invasive, and it adds considerable cost above CT alone.

Figure 43-6 CT myelography. T2 weighted MRI image (A) at L4 disc in patient with right L5 radicular pain pattern suggests lateral recess compromise, but is not definitive. Sequential CT myelography images (B) and (C) at L4 disc clearly show effacement of traversing L5 root sleeve by disc herniation. In a patient with left C6 radicular pain, T2 weighted MRI image (D) at C5-6 interspace suggests foraminal narrowing, but cannot determine if this is due to bone or soft disc. CT myelography images (E) and (F) at the C5-6 interspace reveals both soft disc herniation and uncovertebral joint hypertrophy compromising the foramen.

MRI

MRI remains the preeminent spine imaging modality, a status it has enjoyed over much of the last 2 decades, owing to its superior contrast resolution. MRI remains unchallenged in its ability to discriminate between soft tissues of varying character. Alterations of disc contour (herniations) are evaluated with substantial interobserver reliability.⁶² Lesions compressing the thecal sac, root sleeves, and exiting nerve roots can be directly visualized and the cross-sectional dimension of the thecal sac can be quantified without invasion. Interobserver reliability for these parameters is moderate.⁶³ The soft tissue contrast characteristics of MRI allow detection of neoplastic or inflammatory marrow infiltrative lesions that do not destroy trabecular or cortical bone. MRI is more sensitive than CT or CT myelography in the detection of neoplastic or infectious processes.¹ With the use of heavily T2 weighted imaging sequences (short tau inversion recovery, fat-saturated fast spin echo) or contrast enhancement, MRI can image an inflammatory response, allowing potential identification of degenerative or noninfectious inflammatory pain generators.

MRI: Degenerative Disease

MRI will identify disc degeneration as loss of the normally high T2 signal in the nuclear compartment of the disc before there is detectable loss of disc space height on either CT or radiographs. This finding alone is not predictive of a painful disc. The disc and the adjacent vertebral cartilaginous end plate are a functional unit; end-plate inflammatory changes (Modic changes) frequently accompany disc degeneration.⁶⁴ Modic type 1 changes are an inflammatory state consisting of vascularized granulation tissue within the sub–end-plate marrow. MRI demonstrates low T1 signal and high T2 signal with enhancement if gadolinium is administered. Modic 1 change may progress to Modic type 2, in which there is fatty infiltration of sub–end-plate marrow, evidenced by elevated T1 and T2 signal (Fig. 43-7). Immunohistochemistry studies suggest that end-plate abnormalities are related to inflammation and axon growth induced by tumor necrosis factor (TNF); such nerve in-growth in abnormal end plates may be a cause of low back pain.⁶⁵ This is more prominent in Modic 1 change than Modic 2 change. Modic 3 changes represent sclerosis and are seen as low T1 and low T2 signal on MRI. It is a postinflammatory state.⁶⁶ Inflammatory end-plate changes have been correlated with a painful disc, with a positive predictive value of 91% in a single small study.⁶⁷

Figure 43-7 Modic end-plate changes. Type 1 end-plate changes, consisting of vascularized granulation tissue, will have diminished T1 signal (A) and elevated T2 signal (B) This imaging finding is associated with the presence of inflammatory mediators in subchondral bone. Type 2 end plate change is fatty infiltration of marrow, likely a less inflammatory state. It is visible as elevated T1 (C) and T2 (D) signal in subchondral bone.

MRI may also identify annular fissures. As the nuclear-annular junction deteriorates, radial fissures may extend to the outer aspect of the disc annulus. Such fissures, called high intensity zones (HIZs), may be visualized as a focal zone of elevated T2 signal, which may enhance due to the presence of vascularized granulation tissue (Fig. 43-8).⁶⁸ The initial study showed that an HIZ increases the odds that provocation discography (PD) will be positive in a patient at that level by a factor of 6.5.⁶⁸ Studies on HIZs since that time have all shown that an HIZ has high specificity (range 0.74 to 0.93)^69–73 but variable sensitivity, from as low as 0.09⁷³ to as high as 0.78⁷¹ for positive PD. Thus, it is uncommon for an HIZ to occur in a disc that is not painful to PD, and the calculated likelihood ratios (ranging from 1.3 to 6.5 and averaging 4.1) indicate that an HIZ increases the odds that PD will be positive by at least 50%.⁷⁴ Additionally, although HIZs are present in asymptomatic and symptomatic subjects, they are found significantly more in symptomatic subjects (prevalence 60% ±15%) than asymptomatic subjects (24% ± 11%).⁷⁵ The interobserver reliability in identifying a HIZ is fair to good,^72,76 with reported kappa values of 0.44⁷⁶ and 0.57.⁷²

Figure 43-8 High intensity zone. Focal zone of elevated T2 signal is seen in the L4 posterior annulus on both sagittal (A) and axial (B) images. Note the L4 disc is of generalized low T2 signal and a normal internuclear cleft is not visible. On a fat-saturated T1 weighted postgadolinium image (C), the L4 high-intensity zone enhances, implying it has acquired vascular supply owing to the attempted reparative response.

In summary, a “dark disc” on MRI suggests physiologic degeneration but is an imperfect predictor of a painful disc (discogenic pain). The presence of an annular fissure (HIZ) or end-plate inflammatory change (especially Modic 1) enhances the positive predictive value of MRI but still falls short of discography in the identification of a painful disc. Discography also remains controversial as a diagnostic test and is not universally accepted. Attempts to identify a painful disc by imaging ultimately fall prey to the gold standard dilemma—there is no pathologic, surgical, or imaging standard against which measurement can occur.

Degenerative discs may undergo herniation, which is the displacement of disc material beyond the normal confines of the intervertebral disc space. The accepted nomenclature for disc herniation has recently been codified (Fig. 43-9).⁷⁷ Disc herniation by definition involves less than 50% of the circumference of the disc; displacement of greater than 50% of the circumference of the disc may be termed a diffuse bulge. A broad-based herniation encompasses 25% to 50% of disc circumference, whereas a focal herniation involves less than 25% of the disc circumference. Disc herniations may be subdivided into protrusions and extrusions, which are shape definitions but carry implications regarding the integrity of the outer annulus. A protrusion remains bounded by an intact outer annulus; herniated material never exceeds the dimension of the parent disc in any plane (Fig. 43-10). An extrusion is uncontained by an intact annulus and may migrate cephalad or caudal to the disc space level, attaining a dimension greater than the annular defect in the disc from which it originated. A sequestration is a free disc fragment which has lost continuity with its parent disc and undergone migration away from it (Fig. 43-11). The location of herniated disc material in the axial plane may be described as central, subarticular (lateral recess), foraminal, or extraforaminal (far lateral), in progression from the midline laterally.

Figure 43-9 Herniation nomenclature. Axial plane (A) and sagittal plane (B) illustrations of disc herniation nomenclature. Protrusion and extrusion are shape definitions, but carry the implication that in an extrusion there is complete disruption of the outer annulus.

Figure 43-10 Disc protrusion. Sagittal T2 weighted MRI image (A) and consecutive axial T2 weighted images (B) and (C) demonstrate left L4 central and subarticular disc protrusion impinging on traversing left L5 root. Left L5 radicular pain syndrome treated with left L5 transforaminal epidural steroid injection (D).

Figure 43-11 Disc extrusion, peripheral enhancement. Patient with left “hip” pain, experienced in inguinal crease and medial thigh. Sagittal T2 weighted MRI image (A) reveals extruded L3 fragment that has migrated cephalad into left L3 root axilla, confirmed on axial T2 (B) and T1 (C) images. Postcontrast fat-saturated T1 sagittal (D) and (E) axial images show the intense inflammatory epidural enhancement about the disc fragment. The disc material itself does not enhance.

In the patient with radicular pain or radiculopathy, in whom a disc herniation is suspected, imaging should begin with radiographs but may progress to MRI if conservative measures fail and epidural steroid injections or surgical intervention are contemplated.^12,15 Ninety percent of lumbar disc herniations occur at the L4-5 and L5-S1 interspaces.⁷¹ Disc herniations most commonly have a posterolateral vector, impinging on the nerve root (traversing root) which will exit under the next lowest pedicle (L4 disc herniation causing L5 radicular pain syndrome). For a lumbar herniation to affect the like-numbered nerve root, herniated material must migrate cephalad into the neural foramen (Fig. 43-12). In the cervical region, the nerve root exits low in the foramen, directly over the subadjacent pedicle, and is vulnerable to disc herniation from the intervertebral disc at the same level as its exit. Remembering that in the cervical spine the numbering of the exiting root matches the subadjacent vertebral body, most cervical herniations will affect the next lower numbered root (e.g., disc herniation at the C6-7 interspace will affect the exiting C7 root [Fig. 43-13]). Herniations are most evident on T2 weighted images, where the mass effect of the herniation can be directly visualized on the thecal sac or nerve root. The herniated disc material is often of higher T2 signal than the parent disc from which it arose. The imaging natural history of disc herniation in both the lumbar and cervical regions is resolution. Numerous longitudinal imaging studies demonstrate that most disc herniations resolve spontaneously, with larger herniations and extrusions most likely to undergo resorption.^78–80 The functional disability incurred by disc herniations is better predicted by psychosocial factors than the size or imaging characteristics of the herniation.⁸¹

Figure 43-12 Foraminal (lateral) disc extrusion. Patient with left L4 radicular pain pattern. Sagittal T1 (A) and T2 (B) weighted MRI images show foraminal disc extrusion. Note that on the T2 weighted image (B) the extruded fragment is of higher T2 signal than the parent disc. The exiting L4 root is displaced laterally on the axial T2 weighted image (C). Epidural contrast injection preceding steroid injection (D) and (E) outlines the disc herniation.

Figure 43-13 Cervical disc extrusions. Patient presents with right C7 radicular pain syndrome. Sagittal T2 weighted images (A) and (B) show congenital fusion of the C5-6 bodies. Disc extrusion at C6 interspace. Consecutive T2 weighted axial images (C, D, E) at C6-7 interspace demonstrate right-sided extrusion deforming exiting C7 root and mildly deforming cord caudal to root entry zone.

Gadolinium is not typically used in the nonoperated cervical or lumbar spine; when used, it will often show that the volume of herniated disc material (which does not enhance) is relatively small compared to accompanying enhancing granulation tissue. In the postoperative spine, contrast enhancement is mandatory. Imaging is essentially blind to the detection of recurrent disc herniation in the 6 weeks immediately following surgical intervention. Beyond this perioperative period, the use of gadolinium enhancement should allow accuracy of up to 96% in distinguishing recurrent herniation from epidural fibrosis (scar).^82,83 Epidural fibrosis enhances uniformly and rapidly following gadolinium administration, whereas herniated disc material enhances minimally and slowly (Fig. 43-14). Extensive epidural fibrosis is in itself a negative prognostic sign in the postoperative patient and is associated with an increased incidence of persistent postoperative radiculopathy.⁸⁴ In the postoperative setting, the thecal sac should also be examined for evidence of arachnoiditis, usually attributed to hemorrhage within the thecal sac. In this condition, the roots of the cauda equina either clump or are adherent to the neural tube (Fig. 43-15). The dural tube may appear empty of roots which are smoothly scarred to its wall. Clumped roots in arachnoiditis may exhibit enhancement. In severe cases, the subarachnoid space is completely obliterated by scarring.

Figure 43-14 Recurrent disc herniation. Patient with history of left L5 discectomy presents with new left S1 radicular pain. T2 weighted sagittal (A) and axial (B) MRI images reveal apparent left L5 disc extrusion. In the postoperative setting, this cannot be known without use of gadolinium contrast material. Pre- (C) and postgadolinium (D) T1 weighted axial images demonstrate multiple extruded fragments, which do not enhance, separated by thin enhancing septa of granulation tissue. The left S1 root sleeve is effaced.

Figure 43-15 Arachnoiditis. T2 weighted MRI images show a normal distribution of roots in the dorsal thecal sac in the upper lumbar region (A), but thickened, clumped roots in the lower lumbar canal (B and C). In another patient, there is a paucity of roots evident in the thecal sac on a sagittal T2 image (D). Axial images in the lower lumbar region (E and F) reveal that all lumbar roots are adherent to the dural tube, giving the impression of an “empty sac.”

Spinal stenosis describes a clinical syndrome characterized by neurogenic claudication with exertion. Imaging identifies the morphologic substrate of central canal narrowing. Patients who exhibit a narrowed central canal on imaging frequently are asymptomatic and cannot carry the diagnosis of spinal stenosis. Central canal compromise is well displayed on T2 weighted MRI (Fig. 43-16), although CT can also make this diagnosis; CT myelography provides the best spatial resolution in this regard. Central canal narrowing in the lumbar region is usually primarily the result of posterior element impingement: central buckling of the ligamentum flavum, zygapophyseal joint arthropathy, and synovial cysts. Disc herniation and disc-osteophyte complexes may also contribute from the ventral aspect of the canal. Sagittal T2 weighted MRI or CT myelography images also often demonstrate serpiginous redundant nerve roots or dilated veins within the subarachnoid space in the stenotic central canal. Multilevel central canal stenosis is more likely to provoke clinical symptoms. It must be remembered that advanced imaging obtained on a recumbent patient may be insensitive to dynamic lesions.⁴⁰

Figure 43-16 Central canal stenosis. Patient with pseudoclaudicatory pain syndrome exhibits multilevel central canal compromise on sagittal T2 MRI image (A). Stenosis is moderate at L3 interspace (B), severe at L4 interspace level (C), with a widely patent central canal at L2 (D). In another patient with central canal compromise due to epidural lipomatosis, the prominent ventral, dorsal, and lateral epidural fat is well displayed on T1 weighted sagittal (E) and axial (F) and (G) images.

MRI, like radiographs and CT, can identify the morphologic features of zygapophyseal joint osteoarthritis, which have no predictive value for pain generation.⁵⁶ MRI can, however, also identify inflammatory change using heavily T2 weighted imaging sequences such as short tau inversion recovery (STIR) or fat saturated fast spin echo. In these sequences, T2 signal abnormality is white against a dark background and is highly conspicuous (Fig. 43-17). Gadolinium enhancement, best seen with fat saturated sequences, is equivalent in its identification of inflammation, but more costly. A recent paper by Czervionke and associates suggests that such T2 signal abnormality may be useful in identifying painful joints.⁸⁵ This paradigm may apply to the sacroiliac joint as well, although published work to date describes such MRI findings in inflammatory spondyloarthropathy such as ankylosing spondylitis rather than osteoarthritis.

Figure 43-17 Zygapophyseal joint. Radiograph (A) in patient with right-sided axial pain shows sclerosis in right L2-3 joint at junctional scoliotic curve above fusion construct. The morphologic change alone does not correlate with a painful joint. Increased uptake on technetium bone scan (B) at this joint does predict favorable response to intervention. In a different patient, severe structural change of osteoarthritis on CT scan(C) at L4-5 joint is not, itself, predictive of pain generation. Enhancement in and about this joint on fat-saturated T1 weighted MRI images (D) and (E) is predictive of a pain generator.

Synovial cysts, a manifestation of zygapophyseal joint osteoarthritis, may be a cause of radicular pain syndromes and are readily identified on MRI (Fig. 43-18).⁸⁶ A rent in the fibrous joint capsule allows the synovial membrane and its contained fluid to form a juxtaarticular cyst, most common at the L4 level. Synovial cysts are present in approximately 10% of patients undergoing MRI imaging for lumbar pain issues; most of the cysts are dorsal, arising from the inferior recess of the joint and project harmlessly into the multifidus musculature.⁸⁷ Cysts, which project ventrally into the spinal canal, have a prevalence of slightly greater than 2% and may compress neural elements in the central canal, lateral recess, or neural foramen.⁸⁶ On MRI, synovial cysts may signal as CSF does, or be heterogeneous owing to hemorrhage or inspissated proteinaceous material. Ultimately, synovial cysts may calcify. If the cyst wall is very thin, this suggests the cyst may collapse with aspiration under image guidance; cysts with a thick fibrous wall of low T2 signal are less amenable to minimally invasive treatment.

Figure 43-18 Synovial cyst. Patient presented in 2003 with right L5 radicular pain syndrome. T2 weighted MRI images (A) and (B) show synovial cyst arising from right L4-5 zygapophyseal joint deforming thecal sac, impinging on traversing right L5 root. Right L5 transforaminal epidural steroid injection (C) combined with intraarticular administration of betamethasone (D) provided durable pain relief. The same patient presented in 2007 with new axial lumbar pain. T2 weighted MRI images (E and F) show new L5 compression fracture; the synovial cyst is no longer present.

MRI: Systemic Disease

MRI is the most sensitive modality for the detection of systemic disease, which presents as back or leg pain. This broadly consists of infection, unsuspected trauma, or neoplasm. As noted previously, MRI has superior sensitivity in the early detection of pyogenic or granulomatous spondylitis.^59,88–90 Early MRI findings consist of marrow edema (low T1 signal, high T2 signal) about a disc which will likely have zones of elevated T2 signal within it. Gadolinium administration is essential in cases of spondylodiscitis; it will typically reveal enhancement both within the index disc and its adjacent vertebral bodies as well as in the epidural space and paraspinal soft tissues (Fig. 43-19).^59,88 The combination of gallium and technetium nuclear medicine studies has a sensitivity equal to that of MRI in the detection of spondylodiscitis, but will not provide the anatomic information necessary to make a clinical decision regarding medical versus surgical therapy.⁹¹ The presence of an epidural abscess in the setting of neurologic compromise may provoke emergent surgical decompression. Only MRI can provide this anatomic information, as well as provide high sensitivity in the detection of less fulminant cases of spondylodiscitis.^59,88 The imaging appearance of discitis on MRI lags behind the clinical state. Imaging findings may continue to be dramatic when the patient has clearly demonstrated clinical response to therapy. In the author’s clinical experience, therapy should be followed with biochemical markers such as C-reactive protein, not MRI; repeat imaging should be reserved for situations where there is concern for failure of therapy and compromise of the central spinal canal.

Figure 43-19 Pyogenic spondylodiscitis. Patient presents with axial neck pain, myelopathy. Sagittal T1(A) and T2 (B) weighted MRI images demonstrate diminished T1 signal in the C3 and C4 vertebral bodies, narrowed C3 disc space and ventral epidural tissue deforming the cord. Fat-saturated postgadolinium sagittal (C) and axial (D) T1 images show enhancement in the C3 and C4 bodies, ventral epidural space, C3 disc space, and prevertebral tissue consistent with pyogenic spondylodiscitis. Small ventral epidural abscess is present to the right of midline on axial image (D).

Granulomatous spondylodiscitis may be indistinguishable from pyogenic disease, but is more likely to spare the disc with vertebral collapse and kyphotic deformity (Fig. 43-20).⁴⁸ Up to 50% of cases of granulomatous spondylitis have no disc space involvement, and the isolated vertebral lesions can mimic metastatic disease.⁹² CT may provide a clue to the infectious origin of such lesions by demonstrating a broad sclerotic margin about the lytic lesions, implying a more indolent course then is seen with most neoplasms. Diagnosis frequently depends on image-guided biopsy and culture. Assay for tuberculous RNA can provide rapid (24 hour) diagnosis of tuberculous infection while awaiting cultures to provide sensitivity information.

Figure 43-20 Granulomatous spondylodiscitis (Mycobacterium tuberculosis). Young female with axial thoracolumbar junction pain. Vertebral body and disc space destruction with kyphotic deformity is demonstrated on sagittal T2 weighted MRI (A). Coronal (B) and axial (C) postgadolinium T1 weighted images reveal extensive paraspinal abscesses, vertebral body destruction, and epidural abscess. Patient responded well to medical therapy and did not require surgical débridement. A second patient presented with L5 radicular pain. Fat-saturated sagittal T2 image (D) shows multiple vertebral body and posterior element lesions; discs are spared. Epidural disease is present at L5 level. CT images at T10 (E) and L4 (F) show destructive vertebral lesions extending into paraspinal tissues. Imaging had suggested metastatic disease. Note, however, the sclerotic bony reaction suggesting indolent lesions. Biopsy revealed M. tuberculosis.

MRI can detect the characteristic inflammatory response about the sacroiliac joints in the seronegative spondyloarthropathies before morphologic changes are detectable on radiographs or CT. Technetium bone scan is equally sensitive in this regard. In rheumatoid arthritis, MRI can directly evaluate the integrity of the transverse ligament at C1-2, the presence of pannus formation, and potential compromise of the cervical spinal canal.⁴⁸

MRI is not typically used in the acute traumatic setting, but has a role in evaluation of traumatic lesions as subacute and chronic pain generators. Osteoporotic compression fractures are a relatively common cause of back pain in the elderly.¹ Many patients respond well to analgesics and perhaps bracing. Vertebroplasty has been shown to be no better than placebo,^50,51 and although these fractures can be severely painful initially, the natural history is relatively good.⁹³ Unfortunately, a small percentage of vertebral compression fractures are the result of metastatic disease or hematologic malignancy; vertebral body expansion with associated paraspinal or epidural enhancing mass suggests neoplasm. Image guided biopsy can then establish a diagnosis.

Insufficiency fractures are also common in the pelvis, most often seen in the sacrum (Fig. 43-21) but also in the pubic rami, body of the pubic bone, or periacetabular regions. Linear zones of elevated T2 signal or diminished T1 signal in characteristic locations (Honda sign in sacral lesions) suggest insufficiency fracture; MRI signal abnormality will be apparent before reactive sclerosis is detectable by CT or radiographs.⁹⁴ Technetium bone scan is similar in sensitivity to MRI for insufficiency fracture detection.^95,96

Figure 43-21 Insufficiency fracture. Elderly female with axial pelvic pain. T1 weighted coronal MRI images (A and B) show linear low signal medial to sacroiliac joints. Coronal fat-saturated T2 weighted MRI images (C) and (D) show vertically oriented band of high signal in the sacral ala jointed by a horizontal band traversing S1 segment, completing the “H” or Honda sign. (Imagination helps.)

Neoplasm, whether extradural or intradural, is best detected and characterized by MRI.¹ Most extradural lesions will be metastases and manifest as vertebral lesions with low T1, high T2 signal enhancement with gadolinium, perhaps with extension into the epidural space. Hematologic malignancies may diffusely replace marrow fat, with uniformly low T1 signal that may not be readily apparent; marrow should always be brighter than disc on T1 images. Intradural lesions include meningiomas, schwannomas, leptomeningeal metastases; the intramedullary neoplasms include ependymoma, astrocytoma, and hemangioblastoma. These will be detected as enhancing mass lesions.

Nuclear Medicine

Nuclear medicine techniques, including technetium bone scans, gallium scanning, and positron emission tomography (PET) hold niche roles in evaluating the patient with pain of spinal origin. With the addition of single photon emission computed tomography (SPECT), technetium bone scanning acquires greater sensitivity and significantly improved anatomic resolution. By imaging metabolic activity (bone turnover), technetium scans are sensitive, but relatively nonspecific, in the detection of inflammatory, infectious, traumatic, and neoplastic lesions. In degenerative disease, the presence of increased activity on a bone scan may identify active inflammation in zygapophyseal and sacroiliac joints, providing a target for minimally invasive intervention. It is questionable whether it is efficacious in terms of cost or radiation dose to use bone scanning as an additive test for such purposes.

Characterization of compression fractures in terms of chronicity can be readily accomplished by bone scan; this is particularly useful in the non–MRI-compatible patient. Bone scanning is highly sensitive in the detection of insufficiency fractures of the pelvis. Technetium scans also play a role in the detection of spondylolysis; the presence of increased activity in a pars interarticularis implies an actively metabolic lesion, a likely source of pain (Fig. 43-22).⁶¹ Limited CT may be necessary for further characterization of healing potential and therapy planning. SPECT bone scans are very useful in the detection of pseudarthroses in patients who present with pain following fusion surgery.⁹⁷ The combination of gallium plus technetium scans is highly sensitive and specific for the detection of spine infection.⁹⁸ It suffers from its lack of anatomic information, which will likely be necessary for planning therapy and result in additive tests; MRI is generally preferred when spine infection is suspected. Technetium scanning room retains its longstanding role in assessing tumor burden.

Figure 43-22 Imaging of spondylolysis. Fourteen-year-old female volleyball player complains of left-sided low lumbar axial pain. Lateral radiograph (A) is unremarkable. Planar images from technetium bone scan (B) suggest increased uptake in left L5 posterior elements. SPECT images (C) confirm foc al uptake in region of left L5 pars interarticularis. Sagittal reconstruction of limited CT demonstrates subtle lucent line traversing left L5 pars (D). Patient responded well to conservative care.

The assessment of tumor burden is now being assumed by PET scanning. Integrated PET/CT scanners provide metabolic and anatomic information, hopefully obviating the need for additional imaging in following the patient with neoplastic disease.

Imaging Principles and Strategies

The foregoing discussion of the literature which bears on imaging efficacy, sensitivity, and specificity shortcomings of all spine imaging and the characteristics of multiple imaging modalities may be distilled into the following basic imaging principles and strategies:

The best understanding of the relevance of imaging findings will occur when there is close communication between the spine clinician and the imaging professional, and their skill sets overlap. The spine clinician must understand the physical principles underlying imaging and its inherent strengths and weaknesses. The imaging professional must be knowledgeable of clinical pain syndromes, their pathophysiology, and imaging manifestations.