Overview of Low Back Pain
Back pain has plagued humanity since time immemorial. In most cases, the development of low back pain (LBP) is self-limited and does not require operative intervention. Fifty percent of cases of LBP resolve without medical attention within 1 week; 90% resolve within 4 months. However, the remaining cases exact an enormous burden on society in terms of personal suffering and economic impact. In recent years, the prevalence of disability from back pain has exploded in industrialized societies. In the United States, it was estimated that the direct health care expenditure for back pain was $90.7 billion in 1998. A recent systematic review of the global prevalence of LBP revealed that the mean “overall prevalence,” which was defined as all prevalence regardless of the prevalence period, was as high as 31.0%. The statistics are even more unsettling when viewed from a personal and economic perspective. In patients with LBP who have not worked in 6 months, the lifetime return-to-work rate is 50%. In those who have been off work for 1 year, only 25% will return to work. For patients whose injury has left them unable to work for 2 years, the return-to-work rate is less than 5%. This dramatic surge in the incidence and cost of chronic LBP has led to a concurrent rise in the use of diagnostic modalities and therapeutic interventions aimed at ameliorating this growing problem. Among the various types of LBP, internal disk disruption (IDD) is widely acknowledged to be the most common source of axial symptoms.
Diskography was first described in 1948 as a diagnostic tool for herniated nucleus pulposus (HNP). Since that time, simpler, safer, and more accurate imaging modalities have largely supplanted diskography as an investigative technique for nerve root compression. Yet in some circles provocative lumbar diskography continues to be a popular, albeit controversial, means of diagnosing axial LBP secondary to IDD. This is because unlike magnetic resonance imaging (MRI) or computed tomography (CT), diskography is not just an imaging modality but a provocative test purported to correlate symptoms with pathology. Although some studies have shown a high degree of correlation between the results of diskography and histologic findings and between diskography and surgical outcomes, others have failed to demonstrate such a relationship.
Provocative Tests in Context
Much of the criticism surrounding diskography stems from generalized disapproval about the diagnostic value of provocative procedures for other spinal disorders. In a study by Marks, no consistent segmental or sclerotomal referral patterns were found during 385 provocative lumbar facet blocks in 138 patients with chronic spinal pain. Bough and coworkers assessed the histologic findings of 127 facet joints surgically removed based on the results of preoperative provocative lumbar facet arthrography. The authors found the specificity of degenerative facet joint changes to be 75% but the sensitivity to be only 59%. They concluded that reproduction of symptoms during facet arthrography was of little value as a presurgical screening procedure. Schwarzer and associates conducted a prospective study of 90 patients (203 joints) to determine the relationship between pain provocation and the analgesic response to lumbar zygapophyseal joint blocks. Using a single analgesic block as the diagnostic criterion, reproduction of similar or exact pain was found to predict subsequent response to analgesic facet blocks. However, when the more stringent criterion of concordant analgesic response to confirmatory blocks with lidocaine and bupivacaine was used, no significant association was found.
In 1994, Fortin and colleagues conducted two studies designed to evaluate sacroiliac (SI) joint pain referral maps generated by the distention of joints in asymptomatic volunteers. In the first study, the authors designed a pain referral map by distending the SI joint capsule in 10 asymptomatic subjects by injecting radiopaque contrast material and lidocaine. Similar to a set of previous studies done on the cervical facet joints, the authors found that the pain generated during the initial joint injection corresponded with the hypoesthesia experienced after lidocaine was administered. In the second study, independent observers chose 16 patients with chronic LBP whose pain diagrams most closely resembled the pain referral maps generated in the first study and injected their joints with bupivacaine. Ten of the 16 patients obtained 50% or greater pain relief after the instillation of bupivacaine. Of note, the pain referral maps generated in the first Fortin study were significantly different from the SI joint referral zones described by other authors based solely on analgesic blocks. Finally, Schwarzer and associates conducted an SI joint prevalence study in 43 patients with LBP principally below L5-S1. Using the analgesic response to local anesthetic blocks as the sole criterion for diagnosis, 30% of patients were considered to have SI joint pain. Using pain relief combined with a ventral capsular tear on postarthrography CT imaging, 21% of the patients met the diagnostic criteria for SI joint pain. Using the three criteria of concordant pain provocation, abnormal findings on imaging, and analgesic response to injection of local anesthetic, just 16% were considered to have SI joint pain. Among the 27 patients with similar or exact pain reproduction during provocative testing, only 41% experienced “gratifying” pain relief following local anesthetic injection, thus suggesting that pain provocation may be associated with a significant false-positive rate. The lack of strong validity for provocative facet and SI joint injections as diagnostic tools is consistent with the findings of Slipman and coworkers, who demonstrated distinct differences between dynatomal and dermatomal maps during provocative cervical nerve root blocks. In summary, there seems to be little evidence to support the use of provocative injections to diagnose other sources of LBP.
Lumbar Intervertebral Disk Anatomy
The intervertebral disk complex is composed of the nucleus pulposus (NP), the annulus fibrosus (AF), and the vertebral end plates. Lying above and below the disk are the vertebral bodies or the sacrum below L5-S1. The disk is attached to the adjacent vertebral bodies via the vertebral end plates centrally and to the ligamentous attachments of the AF peripherally. Together, these components allow the principal movements exhibited by the lumbar spine, which include flexion, extension, axial rotation, and lateral flexion. Horizontal translation does not generally occur as an isolated movement but is involved in axial rotation. Posteriorly, the intervertebral disk is supported by the other two components of the three-part structure, the paired zygapophyseal joints. Working in concert, these structures function to support and stabilize the spine and prevent injury by limiting motion to specific planes of movement.
The NP consists primarily of water (70% to 90%) in a matrix composed of proteoglycan, a substance with high water-binding capacity, and type II collagen. The dry weight of the intervertebral disk consists of approximately 65% proteoglycan, with the remainder of the nucleus being composed largely of type II collagen. The high water content of the disk creates a broad, relatively noncompressible weight-bearing surface that serves to cushion the spine from the stress of the truncal load.
The AF is composed of primarily type I collagen arranged in highly organized, concentric lamellae. These lamellae form 10 to 20 sheets surrounding the NP and are thicker toward the center of the disk. The AF is thick and strong anteriorly and laterally but tends to be weaker posteriorly. This anatomic incongruity accounts for the disproportionate percentage of disk herniations occurring posteriorly. Posteriorly, the AF is concave in the lumbar spine. Like the NP, the AF has a high water content, in the range of 60% to 70% by weight. The annulus helps stabilize the vertebral bodies and limit excess motion.
The vertebral end plate is composed of hyaline cartilage close to the vertebral body and fibrocartilage near the NP. The fibrocartilage surrounds the NP and is formed as an extension of the annular fibers. The end plates completely envelop the NP centrally but taper off peripherally, where the outermost AF lamellae directly attach to the vertebral bodies. The composition of the end plates resembles that of the annulus at its attachments to the vertebral bodies and that of the NP centrally.
A healthy adult disk is basically avascular, with its nutrition supplied through the vertebral end plates and the AF via passive diffusion. Whereas the periphery of the annulus is completely permeable, the bone-disk interface is only partially permeable to substrates. The annulus contains blood vessels just in its most superficial lamellae. The nucleus itself contains no direct blood supply. The oxygen and nutrients that diffuse through the end plates come from branches of the lumbar arteries supplying the vertebral bodies.
The anabolic functions of a healthy disk are maintained by chondrocytes and fibroblasts, whereas its catabolic functions depend on the matrix metalloproteinase (MMP) enzymes collagenase (which degrades collagen) and stromelysin (which degrades proteoglycans). The metabolism of cells in the nucleus is exquisitely sensitive to changes in pH, and it is maximally active in the range of 6.9 to 7.2. Even in the presence of high oxygen concentrations, disk metabolism is mainly anaerobic. Any number of factors can lead to a breakdown in the delicate metabolic function of the disk, including changes in pH, inflammatory mediators, and nutritional deficiencies.
Functionally, innervation of the lumbar intervertebral disks stems from two extensive nerve plexuses that accompany the posterior and anterior longitudinal ligaments. These are known as the posterior and anterior plexuses. The anterior plexus consists of contributions from the anterior branches of the gray rami communicantes, small medioventral branches of the sympathetic trunk, and perivascular nerve plexuses. The posterior plexus is a diffuse network of interconnecting fibers receiving somatic and autonomic input from multiple spinal levels. Its visible components are derived mainly from the sinuvertebral nerves, which are formed from somatic roots arising from the ventral rami, and from autonomic contributions from the gray rami communicantes (which receive input from the sympathetic trunk), but the majority of its nerve fibers are actually microscopic. The posterior and anterior plexuses are connected via a less prominent conglomeration of nerves known as the lateral plexus, which is formed by branches of the gray rami communicantes. Together, these plexuses provide transmission of sensory information from the entire circumference of the intervertebral disk.
In newborns, innervation of the lumbar intervertebral disk is dense, most likely because of the extensive blood supply. This rich vascularization disappears around 4 years of age and is accompanied by a concomitant diminution in nerve density. However, in later years when degenerative processes set in, there is a recrudescence of blood vessels and nerve endings. Nerve fibers are typically sparse in the lumbar intervertebral disks, with only the outer third being innervated, but in patients with degenerative disk disease (DDD), the innervation becomes both denser and deeper and frequently penetrates into the inner AF and occasionally into the NP.
There is some evidence supporting the role that lumbar sympathetic afferent nerves play in the perception of LBP, including provocation of LBP by stimulation of the sympathetic trunk and relief of LBP following lumbar sympathetic block. The contribution of these sympathetic afferents appears to be transmitted mainly via the L2 nerve root, as evidenced by the work of Foerster, who demonstrated that L2 is the dermatome corresponding to LBP, and by the work of Nakamura and colleagues, who showed that LBP disappears or significantly decreases after selective blockade of the L2 nerve root. The observation that the lumbar intervertebral disks and their adjacent ligaments are innervated by branches of the sympathetic nervous system does not necessarily mean that sensory input from these structures returns to the spinal cord via the sympathetic trunk. Rather, it has been suggested that somatic afferent fibers from the disks and surrounding pain-generating structures course with the rami communicantes and return to the central nervous system via ventral rami. Several different types of nonvascular nerve endings have been described, including simple, cluster, and partially and fully encapsulated. Although the exact role of each type of nerve ending is unknown, it is speculated that under nonpathologic conditions, they function primarily as mechanoreceptors ( Fig. 65.1 ).
Pathogenesis of Discogenic Low Back Pain
Nerve Ingrowth
Whether the lumbar intervertebral disks receive sensory innervation continues to be the subject of controversy. In the early and mid-20th century, anatomic studies failed to demonstrate nerve endings within the lumbar intervertebral disks, and it was therefore believed that the disks could not be a principal source of pain generation. Subsequent studies have since disproved this concept. In normal human disks, sensory nerves extend into the outer third of the annulus. In degenerated and herniated disks, the innervation is deeper and more extensive, with some nerve fibers penetrating into the NP. It is now generally acknowledged that intervertebral disks do receive sensory innervation and indeed can be a significant cause of LBP.
In 1997, two groups described ingrowth of nociceptive nerve endings into degenerate intervertebral disks. Nerve fibers were identified by using a combination of histologic nerve stains. These nerves have the morphology of nociceptive nerves and express GAP43, a marker of nerve growth, and substance P, a nociceptive (and vasoregulatory) neurotransmitter. The mechanisms leading to this nerve ingrowth have been grouped into three categories.
The first of these categories is angiogenesis. During angiogenesis, endothelial cells of vessels growing into the intervertebral disk synthesize the neurogenic stimulator nerve growth factor, one of a family of neurotrophins. Importantly, nerves that are structurally nociceptive in nature are seen only in intervertebral disks that are classified clinically as “pain-level disks.” Hence, when these disks are stimulated (e.g., by diskography or direct probing), the patient’s symptoms of back pain or sciatica, or both, are reproduced. Intervertebral disks that show similar degrees of degeneration but do not provoke concordant pain with stimulation do not exhibit nerve ingrowth. The second mechanism proposed to lead to nerve ingrowth is altered intervertebral disk matrix biology. Johnson and colleagues showed that aggrecan from normal intervertebral disks inhibits neurite growth but that aggrecan from degenerate intervertebral disks has less of an inhibitory effect. This suggests that with degeneration, nerve ingrowth may occur as a consequence of changed aggrecan biology. Finally, the third possible reason for ingrowth is altered cell function. Even though aggrecan normally inhibits neurite outgrowth, this could be reversed by cells derived from degenerated intervertebral disks.
Genetic Predisposition
Over the last 25 years, several genetic associations have been implicated in the predisposition to intervertebral disk degeneration, but few have been replicated consistently. Only collagen IX and vitamin D receptor polymorphisms have been reliably associated with degeneration in reasonably sized populations. Other possible genes currently being investigated include those for collagen I α1, interleukin-6 (IL-6), aggrecan, MMP-3, thrombospondin, cyclooxygenase, tissue inhibitor of metalloproteinases 1 (TIMP-1), cartilage intermediate layer protein, and IL-1. Further studies are needed to examine the functional interaction of these genes within the framework of the molecular pathology associated with the degeneration of intervertebral disks.
Mechanical Changes
In a normal disk, mechanical interplay between the AF and NP distributes weight bearing uniformly across the entire disk surface. When a disk becomes physically stressed, such as by flexion of the spine, the nucleus acts as a noncompressible mass, with its gelatinous contents bulging in the axial, sagittal, and coronal planes. A competent annulus resists this outward bulging, thereby resulting in equal distribution of force. In these circumstances the annulus is not unduly stressed since its broad surface area translates into the nucleus bearing the greatest share of the load.
Over time, age-related changes or an acute injury can lead to a breakdown in normal load bearing. Histologic studies conducted on cadaver lumbar spine specimens have revealed that as early as the second decade of life, a reduction in blood flow leads to diminished nutritional supply to the end plate. This in turn results in tissue breakdown within the disk that commences in the NP and shortly thereafter in the vertebral end plates. This progressive, macroscopic degeneration, in conjunction with either low-level repetitive stress or an acute traumatic event, can lead to two possible sources of injury: microfracture of the vertebral end plate or an annular tear from torsional overload. When this occurs, the ability of the NP to evenly dissipate a compressive load becomes compromised. Unlike normal disks, a compressive load in degenerated disks is not uniformly distributed. Instead, the preponderance of the weight-bearing burden is borne by the richly innervated AF. Although this can be maintained for short periods, if the end-plate fracture does not heal, repetitive stress on the annulus can eventually lead to tearing of the fibers. These tears further decrease the load-bearing capacity of the disk since torn lamellae can no longer function as a support apparatus. This initiates a vicious circle that causes even more stress on the remaining lamellae and eventually leads to further tearing of the annular fibers, which can ultimately result in the complete loss of annular integrity. The net result of this sequence of events is that the disk is now predisposed to nuclear herniation. The loss of disk height that inevitably ensues may then deteriorate into continued narrowing of the disk, accompanied by the pathologic changes commonly seen in severe DDD, such as Modic changes, sclerosis of the end plates, and bridging osteophytes ( Table 65.1 ). In severe cases, this is manifested as autofusion and anklylosis of adjacent segments.
| Category of Change in Signal Intensity ∗ | T1-Weighted MRI | T2-Weighted MRI | Histopathologic Changes | Significance and Comments |
|---|---|---|---|---|
| Type I | Decreased signal intensity | Unchanged or increased signal intensity | Disruption and fissuring of the end plate and vascularized fibrous tissue within the adjacent marrow | Changes signify edema. Type I changes tend to convert to type II changes over time |
| Type II | Unchanged or increased signal intensity | Isointense or slightly hyperintense signal | End-plate disruption with yellow marrow replacement in the adjacent vertebral body | Signifies fatty degeneration. This is the most common type and tends to remain stable over time |
| Type III | Decreased signal intensity | Decreased signal intensity | Extensive bony sclerosis indicative of dense woven bone within the vertebral body rather than marrow | No marrow to produce MRI signal |
∗ Refers to changes in signal intensity in the vertebral body marrow adjacent to the end plates of degenerative intervertebral disks.
Chemical Changes
Several changes can occur in degenerated disks whose end result is a decrease in the threshold for nociception. First, a break in an end plate can lead to the introduction of inflammatory cytokines into the nucleus, which results in reduced oxygen diffusion, a rise in lactate levels, and a decrease in pH. This conglomeration of factors can result in a slowdown in metabolic and reparative processes and thus lead to increased degradative metalloproteinase activity and diminished chondrocyte activity, which can accelerate disk degradation. In certain contexts, proinflammatory cytokines can be a direct source of pain, but in the context of a functional annulus, no pain should be experienced since the inflammatory mediators cannot reach the nociceptors present in the outer part of the disk. However, when an annular tear develops, granulation tissue forms and nerve endings are able to extend through the annulus, with penetration sometimes as far centrally as the NP. Thus, in the presence of a compromised AF, chemical mediators are able to reach sensitized nerve endings, the by-product of which is LBP.
Sensitization and irritation of nerve endings in the end plate may also result in pain. This model of chemical nociception is supported by a number of studies showing disk immunoreactivity to a variety of substances in diseased and herniated intervertebral disks, including vasoactive intestinal polypeptide, substance P, and calcitonin gene–related peptide, as well as elevated levels of nitric oxide, prostaglandin E 2 , IL-2, IL-6, IL-8, phospholipase A 2 , leukotriene B 4 , thromboxane B 2 , and tumor necrosis factor-α.
Taken in concert, these factors provide a biochemical and mechanical rationale for performing diskography. The generation of pain at low intradiscal pressure is best explained by a preponderance of inflammatory mediators around the sensitized disk; in medical terminology, this is referred to as “chemically sensitized” disks. When disk degeneration is less severe or the disruption less acute, the disk may react to stimulation only at higher pressure, at which point the nerve fibers of the degraded annulus are stretched to the point of pain induction. This scenario describes a “mechanically sensitized” disk. Perhaps an easier way to conceptualize these models is that a chemically sensitized disk is analogous to the phenomenon of allodynia whereas a mechanically sensitive disk is akin to hyperalgesia. In the later stages of disk disease, the annulus may become functionally incompetent, in which case injection of contrast material may fail to generate intermediate to high pressure. This can result in an uninterpretable or even false-negative pain response if manometry is not used. On CT, severely degenerated disks are likely to show a diffuse pattern of spread of contrast material with extensive leakage into the epidural space, but it is possible to miss small leaks on plain fluoroscopy. Normal disks resist pain provocation because they lack both the chemical hypersensitivity and the apparatus for mechanical overloading that are present in diseased disks. In clinical practice, these examples represent an ideal diagnostic paradigm that fails to account for the multitude of genetic, social, cultural, and psychological factors that affect pain perception. To optimize diagnosis and treatment outcomes, these factors must be considered when performing any pain-provoking procedure.
Lumbar Disk Pain Referral Patterns
The premise on which diskography is based is that controlled pressurization of a painful disk will reproduce a patient’s symptoms. The limitations of this paradigm have been discussed previously. In addition to the flaws inherent in any provocative test, inaccuracies may also result from oversedation with anxiolytics or opioids (or both), excessive administration of local anesthetic, anxiety, procedure-related pain, the ephemeral nature of disk pressurization, and an inability to distinguish concordant from noncordant pain in the brief moments when disk pressure exceeds the nociceptive threshold. Nevertheless, several investigators have attempted to categorize the pain patterns with positive diskograms. In a prospective study conducted in 187 patients with LBP scheduled for diagnostic CT-diskography, Ohnmeiss and coworkers found that L3-4 diskograms were likely to be positive if patients described their pain as involving the lumbar region with radiation into the anterior but not the posterior aspect of the thigh and often into the anterior aspect of the leg. For L4-5 disks, the most common pain referral pattern was lumbar pain involving more equivalent proportions of the anterior and posterior thigh pain. In L5-S1 discogenic pain, the pain description generally encompassed the lumbar and posterior thigh regions, with fewer patients reporting anterior thigh or leg pain. Pain in the absence of disk pathology tended to be limited to the low back region and buttocks ( Fig. 65.2 ).
In the late 1980s, Vanharanta and colleagues performed a series of studies to evaluate the effects of various disk abnormalities on pain referral patterns. In the first of these studies, evaluation of CT-diskograms in 91 patients showed a positive relationship between the occurrence of pain and the presence of an annular rupture. The second study found that narrow disks were more likely to be associated with “exact pain reproduction” than were disks of normal height. In the same study it was also suggested that the degree of pain concordance is influenced by the spinal level. Specifically, severely degenerated L3-4 disks were less likely to result in concordant pain than were comparable disks at the L4-5 and L5-S1 spinal levels. The third paper, a prospective, multicenter study evaluating the results of 300 surgical candidates with a variety of clinical diagnoses, found no significant relationship between concordant pain provocation in four diagnostic groups: disk herniation (82%), DDD (81%), lumbar syndrome (56%), and radiculopathy (59%). Disks that were deteriorated tended to be painful regardless of a patient’s diagnostic classification. In the fourth study, based partly on the same patient population, the authors found that the percentage of degenerated disks that elicited either “no pain” or “noncordant pain” was higher in elderly patients. A later retrospective study found that outer annular ruptures were the only predictors of concordant pain in a reanalysis of 833 diskograms done in 306 patients with LBP.
Maezawa and Muro came to a different conclusion from Vanharanta and colleagues. In a retrospective analysis of 1477 diskograms performed in 523 patients with axial or radicular LBP, the authors found that elicitation of pain was only weakly associated with IDD whereas the presence of a herniated disk was strongly associated with pain provocation. Saifuddin and coworkers performed a retrospective review of 260 lumbar diskogram reports with the aim of correlating morphologic disk abnormalities with pain referral patterns. The authors demonstrated a highly significant association between annular tears and concordant back and radiating pain of any type. An association between isolated posterior tears and radiating pain was also found, but no relationship between anterior annular tears and any pain radiation was noted. No differences in pain referral patterns were identified when full-thickness and partial-thickness posterior tears were compared. There was an increased incidence of leg pain versus groin, hip, or buttock pain during L4-5 and L5-S1 disk provocation, but this trend did not reach statistical significance. Finally, a retrospective study by Slipman and colleagues did not find any correlation between the side of a patient’s concordantly painful annular tear on CT-diskography and the side with the pain complaints. To summarize, the presence of degenerative changes of any type is more likely to be associated with pain than are nondegenerated disks or disks with only minimal degradation. Pain provocation may also be more likely to occur in the presence of annular tears and nuclear herniations, with more evidence supporting the former assertion.
False-Positive Pain Provocation
In the past 30 years the use of diskography to detect disk pathology has largely been supplanted by more advanced radiologic studies such as MRI and CT. The evidence that these newer modalities are not only safer but also more accurate than plain diskography in detecting herniated nuclear material is indisputable. In a prospective study by Jackson and associates in which myelography, CT-myelography, plain diskography, and CT-diskography were compared with surgical findings in 231 disks, the authors found CT-diskography to be the most accurate test (87%) and plain diskography to be the least accurate (58%). CT-diskography ranked highest in sensitivity for HNP at 92%, as compared with 78% for CT-myelography, 72% for CT, 70% for myelography, and 81% for plain diskography. For specificity, CT-diskography was also the most accurate (81%), followed by CT (76%) and CT-myelography (76%), myelography (70%), and plain diskography (31%).
Since diskography is more invasive and unequivocally less sensitive in detecting most disk pathology than newer imaging techniques are, the primary justification for its continued use lies in its ability to correlate pathology with symptoms. This rationale seems reasonable given that close to two thirds of asymptomatic adults have abnormal findings on MRI of their lumbar spine, with the prevalence of these findings increasing with age. LBP is an epidemic of unprecedented proportion in industrial societies, and the estimated lifetime prevalence ranges from 60% to higher than 80%. Without a corroborative test to validate these abnormal MRI findings, it is likely that many of these patients would be conferred with incorrect diagnoses and subsequently undergo unnecessary surgical procedures. There is little doubt that unnecessary stabilization operations are performed far too frequently in the United States, but there is disagreement about whether diskography prevents or facilitates these unnecessary surgeries. Proponents of diskography believe that correlating pathology with symptoms prevents unnecessary surgical intervention, whereas opponents question both the significance of diskographic pathology and the validity of provoked symptoms. These criticisms are bolstered by the relative lack of specificity of diskography, the inherent difficulty in validating provoked symptomatology, the large number of studies showing false-positive pain provocation in patients without low back symptoms, and disparities with regard to histologic findings on surgical specimens.
In phase 1 of a two-part experiment, Yasuma and coworkers studied 181 lower thoracic and lumbar disks from 30 adult cadavers diskographically and histologically. Their findings revealed 32 true-positive, 15 false-positive, 122 true-negative, and 12 false-negative diskograms. Diskograms were designated false-positive when the injected contrast material was noted to extend beyond the peripheral vertebral margin but histologic sectioning of the disk was negative for protrusion. False negatives were defined as a negative diskogram despite histologically confirmed disk protrusion. In the 32 true-positive disks, both the diskograms and histologic sections showed 10 anterior protrusions, 17 posterior protrusions, and 5 posterior herniations. Among the 15 false-positive diskograms, 9 were misdiagnosed as a protrusion and 6 as a herniation. In part 2 of this study, the authors conducted a retrospective review of 77 diskography patients subsequently found to have a herniated disk during surgical exploration. The diskograms were falsely interpreted as negative in 32% of the 59 patients with a protruding disk and in 56% of the 18 patients with a prolapse. In a previous study the authors found that false-positive diskograms were more likely to occur when fissures or cysts were present in a degenerated annulus but did not establish continuity with the nuclear cavity.
The first study to question the validity of diskography was published in 1968 by Holt, who found false-positive results in 37% of 30 asymptomatic prisoners. More than 20 years later, Walsh and colleagues performed CT-diskography on 10 asymptomatic male volunteers and 7 “control” patients with chronic LBP. Sixty-five percent of the 20 disks injected in the patients with back pain showed radiologic abnormalities, with all 7 patients having at least 1 degenerated disk. In the patients without back pain, CT-diskograms were interpreted as abnormal in 17% of the 35 disks injected and in half of the 10 subjects. However none of these 5 patients experienced concordant pain associated with pain-related behavior during the injections. Thus, the false-positive rate in this study was 0%.
In 1996, Block and coworkers conducted a landmark prospective study in 90 patients with low back and leg pain who underwent CT-diskography at the three lowest lumbar levels. The Minnesota Multiphasic Personality Inventory was administered to each subject before diskography and scored independently. In the 72 patients with at least one nondisrupted disk, 34 reported discordant pain provocation at a normal level, with the remaining 38 patients reporting a “concordant negative response.” In the 34 patients who reported pain during pressurization of normal disks, the mean hypochondriasis, hysteria, and depression scores were significantly higher than in patients in whom stimulation of normal disks did not elicit pain.
In 1999 and 2000, Carragee and coauthors published a series of studies attempting to identify patients at high risk for false-positive diskograms. In the first study, eight patients with no history of back problems or structural spinal abnormalities who had recently undergone iliac crest bone grafting for reasons unrelated to lumbar spine, hip, or pelvic pathology were studied by provocative diskography of their three most caudal disks. Four of the eight study subjects experienced severe LBP similar to the postoperative pain at their bone graft site during injection of at least one disk. All symptomatic disks had an abnormal morphologic appearance.
In the second study, the authors performed lumbar diskography on 10 patients with neck and upper extremity pain but no lower back symptoms, 6 patients with somatization disorder, and 10 control patients devoid of pain symptoms. The three most caudad lumbar disks were injected in all patients, with five patients having the L2-3 disk studied as well. Eighty-three percent of patients in the somatization group, 40% of the chronic cervical pain patients, and 10% of the control patients experienced moderate or severe pain during injection of contrast material into at least one disk. Pain was provoked in 11% of disks with intermediate-grade disruption and in 37% of disks with annular tears. In contrast, injection did not result in pain in any of the 31 radiographically normal disks. In the last study, three-level diskography was performed on 47 patients who had undergone single-level diskectomy for sciatica. Twenty subjects with no recurrent symptoms were designated the “study” group, and 27 patients with persistent back or leg symptoms (or both) formed the “control” group. In the asymptomatic participants, positive injections occurred in 8 of 20 (40%) operative disks. In the “control” patients with symptoms, positive injections occurred in 17 of the 27 operative disks (63%), with the pain being concordant in 15. No significant differences were found in the diskography results between symptomatic and asymptomatic participants with normal psychometric scores. In contrast, patients with abnormal prediskography psychological test scores were more likely to rate their pain as unbearable during injection of both operative and nonoperative disks than were patients in both groups with no psychopathology. All positive disks were radiographically abnormal.
There are several flaws in the studies assessing “false-positive” diskograms in asymptomatic subjects, with the main one being inherent in the study design: in subjects without preexisting LBP, one cannot provoke true concordant pain, which has become a hallmark of modern-day diskography. The second major shortcoming is that although the Carragee studies used manometry to limit intradiscal pressure, pressure readings were not a determining factor in the designation of a positive disk. The guiding principle behind diskography is to identify disks that provoke pain during stimulation, but application of excessive pressure to any bodily structure that contains nerves, including components of the spine besides disks, will provoke pain in normal subjects.
In an effort to determine the effect that objective pressure readings have on the incidence of positive disk injections in asymptomatic volunteers, Derby and colleagues performed 43 diskograms on 13 subjects with either no history or infrequent episodes of LBP. In the patients with occasional back pain, 35% of the 20 injected disks were painful versus 52% in volunteers with no history of LBP. Most disks required that high pressure be reached before pain was elicited, and even then, the pain was mild. There was no relationship between painful disk injections and abnormalities on MRI or diskography. Controlling for the intensity of response and the pressures at which pain was elicited, the authors concluded that the incidence of false-positive diskograms was less than 10%. Some potential causes of false-positive diskograms include inadvertent annular injection ( Fig. 65.3 ), contrast-induced irritation of nervous tissue, end-plate deflection resulting from suboptimal needle placement, and stimulation of pressure receptors when excessively high pressure is generated.
Wolfer and coauthors published a systematic review in which data from five previous studies were reanalyzed by using guidelines from the International Association for the Study of Pain (IASP) and the International Spine Intervention Society (ISIS) for a positive lumbar diskogram ( Table 65.2 ). The authors found an overall false-positive rate of 9.3% per patient and 6.0% per disk. False-positive rates dropped to 3.0% per patient and 2.1% per disk in patients without back pain or confounding factors. Chronic pain patients were found to have false-positive rates of 5.6% per patient and 3.9% per disk. Finally, the highest false-positive rates per patient and disk occurred in postdiskectomy patients (15% and 9.1%, respectively) and those with somatization disorder (50% and 22.2%, respectively).
| NRS Pain Score | Pressure (psi) | Pain Behavior | Grade 3 Annular Tear | Control Disk NRS | |
|---|---|---|---|---|---|
| Walsh/Carragee | ≥6/10 | ≤100 | ≥2/5 | — | — |
| Derby | ≥6/10 | ≤50 | — | Y | <6/10 |
| ISIS/IASP (a) | ≥7/10 | <50 | — | Y | — |
| ISIS/IASP (b) | ≥7/10 | <50 | — | Y | <6/10 |
| ISIS/IASP (c) | ≥7/10 | <50 | — | Y | 0/10 |
| Low pressure, <22 psi (Carragee) | ≥6/10 | <22 | ≥2/5 | — | — |
| Low pressure, ≤15 psi (Derby) | ≥6/10 | ≥15 psi | — | Y | <6/10 |
One flaw in these criteria is that they do not consider a patient’s baseline pain rating, which can lead to flagrant inconsistencies. For example, a person with significant back pain disability who rates his baseline pain as 3/10 and in whom disk stimulation provokes 5/10 pain would be considered to have a negative diskogram by ISIS guidelines because the maximum pain score did not exceed the minimum cutoff threshold for designating a disk as positive. Yet a person with less functional impairment and 9/10 baseline pain in whom disk stimulation provokes 7/10 pain could theoretically be classified as having a positive diskogram.
Despite these limitations, given the high propensity for false-positive findings in patients with previous back surgery, psychopathology, or somatization symptoms, positive diskograms should be viewed with caution in these individuals. To optimize specificity, we suggest using two adjacent control disks, a recommendation previously endorsed by Endres and Bogduk in an attempt to improve accuracy in patients at high risk for false-positive diskography. Although it may seem intuitive that this would enhance specificity or improve outcomes, these issues have yet to be addressed.
The subject of “false-negative” diskograms has garnered far less attention but can result in inaccurate diagnoses, unnecessary interventions, and withholding of beneficial treatment from otherwise good candidates. There are several reasons that a person with discogenic pain may fail to experience pain with disk stimulation, including failure to detect an inadequate rise in intradiscal pressure because of the lack of pressure monitoring, injecting too slowly, excessive sedation, overzealous use of local anesthetic, and extensive extravasation of contrast material in severely degenerated disks. False-negative diskograms may be more likely to occur in elderly patients. In a review by Cohen and associates, the authors estimated that between 15% and 25% of degenerated disks fail to elicit concordant pain provocation during stimulation. The proportion of these occurrences that represent false-negative responses versus accurate reflection of a non–pain generator is a question that remains to be answered. The results of clinical studies evaluating false-positive lumbar diskography are presented in Table 65.3 .
| Study, Year | Subjects | Criteria | Results |
|---|---|---|---|
| Massie, 1967 | 52 male subjects, 156 disks | NR | FP rate not reported but stated “injection only occasionally produced symptoms” |
| Holt, 1968 | 30 male volunteer inmates, 70 disks (20 failed injections) | Pain provocation | 60% FP rate per subject, 37% per disk |
| Walsh, 1990 | 10 male volunteers, 30 disks | 3/5 pain provocation + 2/5 pain-related behavior | 0% FP rate per subject and disk |
| Carragee, 1999 | 8 males who had recently undergone iliac crest bone grafting for problems unrelated to low back pain, 24 disks | 3/5 “concordant” pain provocation (to previous iliac crest pain) + 2/5 pain-related behavior | 50% FP rate per subject, 38% per disk |
| Carragee, 2000 | 6 subjects with somatization disorder, 10 with failed neck surgery, and 10 control patients with no pain after successful cervical spine surgery; 78 disks | 3/5 “concordant” pain provocation (to previous iliac crest pain) + 2/5 pain-related behavior | FP rate per subject: 83% for somatization, 40% for failed neck surgery, and 10% for “control” group. FP rate per disk: 33% for somatization, 23% for failed neck surgery, and 3% for control group |
| Carragee, 2000 | 47 subjects who underwent single-level diskectomy (20 subjects were “symptom free,” whereas 27 continued to have back and/or leg pain); 138 disks | 3/5 pain provocation + 2/5 pain-related behavior | FP rate per subject: 40% for asymptomatic subjects and 56% for patients with failed back surgery. FP rate per disk: 15% in the asymptomatic group |
| Derby, 2005 | 13 volunteers, 43 disks | Criteria not noted. Used 0-10 pain rating and 0-4 pain behavior scales along with manometry | Using 6/10 as criterion for a positive disk, 0% FP rate. Using 4/10 pain at ≤50 psi, FP rate of 23% per subject and 9% per disk |
Prevalence of Discogenic Low Back Pain
Axial back pain is one of the most common, yet challenging, problems faced by pain physicians. Numerous structures besides degenerated disks can cause axial LBP, with two of the more common ones being the lumbar facet joints and muscles. The SI and facet joints are implicated as the primary cause of chronic axial pain in 15% to 30% and 10% to 15% of cases, respectively. In a porcine study by Indahl and coworkers, the authors determined that there is significant overlap between the neuromuscular connections of the intervertebral disks, zygapophyseal joints, and paraspinal muscles such that the relative contributions of each of these structures to LBP may be difficult to estimate. One may thus conclude that in many patients with chronic LBP, the cause of the pain is multifactorial.
The prevalence of discogenic LBP varies widely in the medical literature. In one of the most cited studies, Schwarzer and colleagues found the incidence of discogenic pain to be 39% in a prospective cross-sectional study using either zygapophyseal joint blocks or diskography in 92 consecutive patients with chronic LBP. The authors based their diagnoses on exact pain reproduction during provocative disk stimulation, coupled with abnormal findings on CT-diskography and the presence of an adjacent, negative control disk. Collins and associates conducted a prospective study comparing the use of MRI and diskography in the evaluation of 29 patients with unremitting axial LBP without focal neurologic deficits. The authors found exact reproduction of the symptoms to be present in 13 disks in 12 patients, for a prevalence rate of 41%. In all 13 symptomatic disks, both MRI and diskography showed degenerative changes. In another study comparing the results of MRI and diskography, Horton and Daftari performed 63 diskograms in 25 patients with nonradicular LBP. Diskography yielded moderate to severe pain at 26 levels, with 19 patients reporting similar or exact pain reproduction, for a prevalence rate of 76%. Only 1 of the 26 disks was morphologically normal. Long and colleagues conducted a very large epidemiologic study in 2374 patients with LBP seen by spine surgeons at seven academic medical centers. Final diagnoses were rendered after imaging studies were reviewed and treatment prescribed, with tests and therapies being performed at the discretion of the surgeon. Nonherniated degenerated disk was the final diagnosis in 6.1% of patients. Finally, in a retrospective analysis of clinical data from 127 patients with axial LBP who failed facet interventions, Cohen and coauthors reported a prevalence of 65%. Based on the conflicting evidence that does exist, discogenic LBP appears to be the major source of pain in more than a third of patients with chronic axial LBP. Whereas younger and middle-aged people with axial LBP are more likely to have disks as the principal source of their symptoms, the facet joints become more important as pain generators in the elderly.
Correlation with Radiologic Studies
There is compelling evidence that diskography, even without CT scanning, may overestimate the prevalence of clinically significant IDD, but it is equally clear that diskography may fail to detect the disk pathology seen with other radiologic studies. In a study by Gibson and associates in which MRI and diskography were compared in making the diagnosis of DDD, agreement between the two techniques was found in 44 of the 50 disks studied. In the six disks in which a discrepancy occurred, evidence of IDD was missed on five diskograms and one MRI study. In the five cases in which diskography failed to detect disk pathology, two were due to incorrect placement of the diskography needle in the annulus. Although disk stimulation symptoms were recorded in the study, the results of diskography were based only on radiographic findings.
Yoshida and coworkers sought to investigate the relationship between plain diskography and T2-weighted and gadolinium-enhanced T1-weighted MRI in 23 patients with chronic LBP. A posterior annular tear was detected in 16 of the 17 positive disks with T2-weighted MRI and in 10 of 14 positive disks with gadolinium-enhanced T1-weighted MRI. The T2-weighted study also detected 11 annular tears in 39 negative disks versus 8 annular tears in 32 negative disks with T1-weighted gadolinium-enhanced MRI. The sensitivity, specificity, positive predictive value, and negative predictive value of the T2-weighted study in detecting symptomatic disks were 94%, 71%, 59%, and 97%, respectively, which compared favorably with the T1-weighted images. The authors concluded that the high sensitivity and negative predictive value of T2-weighted MRI make it a useful screening tool to avoid unnecessary diskography in patients with chronic LBP.
Simmons and coauthors compared CT-diskography with MRI in 164 patients with chronic LBP, with or without radicular symptoms. Correlation between the two techniques was seen in 55% of cases. In the 371 disks in which MRI and diskography were concordant, 172 were normal and 199 were abnormal. In the disks classified as abnormal based on MRI, 37% were asymptomatic during injection. In 13% ( n = 60) of the disks, findings on MRI were abnormal but diskograms were normal. In 7% of the disks ( n = 34), MRI showed normal and diskography showed abnormal findings. In 21 of these 34 disks, injection of contrast material into the disk elicited exact pain reproduction.
In a comparative study evaluating MRI and diskography in patients with axial LBP, Collins and colleagues found that imaging characteristics for the two diagnostic procedures correlated in 65 of 73 disks (89%). In the other eight cases, four disk levels showed evidence of early degeneration on diskography but appeared normal on MRI, whereas four disks showed decreased signal intensity on T2-weighted MRI but were diskographically normal. In the 12 patients with concordant pain on diskography, spinal fusion was performed. At their 9-month follow-up, 9 of the 12 patients reported clinical improvement.
Aprill and Bogduk performed CT-diskography on 41 patients with chronic LBP who demonstrated a high-intensity zone (HIZ) on T2-weighted MRI. In all patients, CT-diskography revealed either a grade 3 or 4 annular disruption in the affected disk. The sensitivity and specificity of an HIZ for detecting similar pain reproduction during disk provocation were 63% and 97%, respectively. For detection of exact pain reproduction, the sensitivity was 82% and the specificity 89%. In the identification of a grade 4 annular disruption, the sensitivity of an HIZ was only 54%, but the specificity was 89% and the positive predictive value 90%.
Other studies have shown a much stronger correlation between MRI and diskography. Linson and Crowe found the two investigative modalites to be in agreement on 91 of 97 disks studied in 50 patients. In the six disks in which a discrepancy was present, five were read as normal by MRI but abnormal by diskography. In an earlier study by Schneiderman and coworkers, MRI and diskography positively correlated in 100 of 101 levels.
However, not all studies have demonstrated a high degree of correlation. Zucherman and associates reported positive diskography in the face of normal MRI findings in a case series involving 18 patients. Horton and Daftari conducted an observational study involving 25 consecutive patients and 63 disks studied by diskography and MRI. Abnormal diskographic morphology was noted in 42 of 59 cases, but only 23 disks were associated with significant pain. In dark disks, disk stimulation provoked pain in 7 of 11, but 0 of 5 “white” bulging disks were positive. Because of these discrepancies, the authors concluded that both MRI and diskography should be used for surgical planning. Finally, in a retrospective study of 53 consecutive patients who underwent both provocative diskography and MRI, Sandhu and colleagues found poor correlation between the vertebral end plate signal changes observed on MRI and the results of provocation diskography.
Overall, diskography appears to be comparable or slightly more sensitive for the detection of IDD than MRI or CT does, especially with regard to radial annular fissures. Approximately 25% of moderately degenerative disks identified on MRI will fail to have concordant pain on diskography. The main problem with correlative studies is that when an incongruity exists, it is impossible to determine whether the discrepancy is due to a lack of sensitivity (false negatives) or specificity (false positives) in one of the diagnostic procedures ( Table 65.4 ).
| Author, Year | Number of Subjects | Nature of Study | Results | Comments |
|---|---|---|---|---|
| Gibson, 1986 | 22 patients, 50 disks | Compared MRI and diskography in patients with mechanical LBP | Agreement between studies in 44 of 50 disks | Diskography results based on radiographic findings only as patients were sedated. In the 6 disks that did not correlate, MRI was superior to diskography |
| Schneiderman, 1987 | 36 patients with LBP with or without leg pain, 101 diskograms | Compared MRI and diskography | MRI accurate in assessing disk morphology in 100 of 101 levels. Of 52 disks with normal MRI findings, only 1 had a positive diskogram. Of 49 disks with decreased MRI signal, only 2 diskograms normal | Used only T2-weighted MRI. CT-diskography used on 39 levels |
| Zucherman, 1988 | 18 patients with LBP with or without radicular symptoms | Nonconsecutive clinical case series. In most cases diskography was followed by CT | All patients had normal MRI and abnormal diskography findings | Normal MRI and abnormal diskography findings were the basis for inclusion |
| Yu, 1989 | 8 cadavers, 36 disks | Compared MRI and diskography against cryomicrotomy anatomic sectioning for detecting annular tears | Diskography identified 15 radial fissures, 10 of which were seen on MRI. Two of the 15 annular fissures were missed on cryomicrotomy | Included a newborn and 2-year-old. Considered only radial tears of the annulus. Could not correlate findings with symptoms |
| Bernard, 1990 | 250 patients (725 disks) with chronic LBP who underwent CT-diskography | Retrospective study comparing the accuracy of MRI, intrathecally enhanced or nonenhanced CT, or plain radiography with CT-diskography. MRI was done before diskography in 67 patients (190 disks) | Normal T2-weighted MRI findings correctly predicted 64 normal disks by CT-diskography and incorrectly predicted 12 normal disks that were abnormal by CT-diskography. In 105 disks, abnormal T2 MRI correctly predicted abnormal disks by CT-diskography. 9 disks that were normal by CT-diskography had decreased signal intensity on T2-weighted MRI. Correlation between MRI and CT-diskography was 89% | CT-diskography provided additional information affecting patient management in 93% of cases. In 94% of the 180 operations, CT-diskography correctly predicted the type of disk herniation |
| Collins, 1990 | 29 patients, 73 diskograms | Compared MRI and diskography in patients with axial LBP | 57 disks were abnormal on diskography, with 13 producing concordant pain in 12 patients. Diskography findings correlated with those of MRI in 90% of cases. 4 disks showed degeneration on diskography with normal MRI, and 4 had abnormal MRI with normal diskography | The 12 patients with positive diskograms underwent spinal fusion, with 9 reporting clinical improvement at 9-mo follow-up |
| Linson, 1990 | 50 patients, 97 disks | Compared MRI and diskography in patients with chronic LBP | 91% correlation for disk degeneration between MRI and diskography | 5 of the 6 disks with negative correlation were read as normal by MRI and abnormal by diskography. No mention of control disks during diskography |
| Simmons, 1991 | 164 patients, 371 disks | Compared CT-diskography and MRI in patients with chronic LBP with or without radiculopathy | 55% correlation based on patients, 80% based on disks | MRI normal and diskography abnormal in 34 disks. Diskography normal and MRI abnormal in 60 disks. 37% of disks abnormal on MRI were asymptomatic on diskography. Did not include outcomes in 76 patients who underwent surgery |
| Birney, 1992 | 90 patients, 264 disks | Examined correlation between MRI and diskography for DDD and HNP. Compared surgical findings with diskography in 57 patients | Agreement between MRI and diskography in 86% of disks. MRI more accurate for HNP; diskography slightly superior to MRI for DDD (MRI missed 1 disk, diskography 100% sensitive) | Considered patients with LBP and radicular pain. Surgical findings correlated with diagnostic studies at 63 of 76 levels |
| Osti, 1992 | 33 patients, 114 disks | Compared MRI and diskography in patients with LBP | All 54 disks identified as abnormal on MRI showed abnormal diskogram patterns. 6 of the 60 disks identified as normal on MRI were abnormal on diskography. Of the 39 disks that provoked concordant pain on diskography, 27 were abnormal on MRI. 33 of the 39 asymptomatic disks by diskography had normal MRI signal, with 24 having normal diskographic patterns | 6 of 46 disks classified as degenerate on MRI were asymptomatic on diskography. Concluded that diskography is more accurate than MRI in detecting annular pathology. Patient population not well defined |
| Aprill, 1992 | 41 patients (105 disks) had chronic LBP with or without radicular symptoms | Compared HIZs on T2-weighted MRI with CT-diskography | In all patients who exhibited an HIZ on MRI, CT-diskography revealed either grade 3 or 4 annular disruption. Grade 3 or 4 disruption was also present in 34 patients without an HIZ | Concordant pain provocation with diskography was present in 38 of 40 disks with HIZs and 22 of 78 disks without an HIZ. CT-diskography performed in only 41 of 500 patients in whom MRI was performed |
| Horton, 1992 | 25 patients with nonradicular LBP involving 63 diskograms | Comparative study between MRI and diskography for discogenic LBP | 19 patients had positive diskograms. Of the different MRI patterns, only “dark/torn,” “dark/bulged,” or “speckled/flat” were more likely to be associated with positive rather than negative diskograms | MRI findings classified by pattern, not by presence or absence of pathology |
| Brightbill, 1994 | 7 patients with LBP | Clinical case series involving patients with discrepancy between diskography and MRI who underwent surgery and were found to have internal disk disruption | All 7 subjects had normal MRI findings and positive diskography | Did not consider surgical outcomes |
| Loneragan, 1994 | 18 patients with chronic LBP thought to be discogenic (43 disks) | Compared MRI and CT-diskography for the diagnosis of DDD and HNP | MRI missed 3 of 10 disks with early degenerative changes and 1 of 3 herniations | In no cases did MRI offer more information than CT-diskography |
| Schellhas, 1996 | 63 patients, 100 disks with HIZs on T2 MRI in patients with LBP and/or radicular pain | Retrospective analysis of the significance of HIZs in predicting positive diskography | All 100 disks with HIZs were abnormal on diskography, with 87 showing concordant pain. In 17 asymptomatic control patients, MRI revealed only 1 disk with an HIZ | 37 patients had back surgery previously. Also included patients with radiculopathy |
| Braithwaite, 1998 | 58 patients with chronic, nonradicular LBP | Retrospective study comparing vertebral end plate changes on MRI with pain provocation during diskography in 152 disks | Among 91 disks with degeneration on MRI, 78 elicited pain vs. only 12 of 61 disks without degeneration on MRI. Among 26 disks with vertebral end-plate changes on MRI, 24 were painful during diskography vs. 69 of 129 disks without end-plate changes | MRI revealed disk degeneration at 128 of 290 levels, and end-plate changes were identified at 31 levels. All patients were being investigated for discogenic LBP as a precursor to spinal fusion. 138 disks evaluated with MRI were not injected |
| Saifuddin, 1998 | 58 patients (152 disks) with chronic, nonradicular LBP | Retrospective study determining the sensitivity of T2-weighted MRI in detecting painful posterior annular tears | 86 annular tears on diskography, 54 of which were posterior and 26 anterior and posterior. Sensitivity, specificity, and positive and negative predictive values of MRI in diagnosing concordantly painful posterior annular tears were 27%, 95%, 89%, and 47%, respectively | Study evaluated the same patients as the Braithwaite study |
| Milette, 1999 | 45 patients, 132 disks | Evaluated MRI and diskography results in patients with chronic LBP | On MRI, 71% of disks showed a normal contour and 64% showed normal signal intensity. Only 40% of diskograms were radiographically normal. Diskography demonstrated stage 2 and 3 disk disruptions in 26% of disks with a normal contour on MRI and in 13% of disks with both a normal contour and signal | Used only T2-weighted MRI. Study was designed to assess differences between disk protrusions, bulges, and loss of signal intensity on MRI, not to compare imaging studies |
| Sandhu, 2000 | 53 patients with LBP, 133 diskograms | Retrospective analysis comparing diskography with vertebral end-plate signal changes on MRI | No significant correlation between diskography and end-plate signal changes | 41% of disks with positive end-plate changes had positive diskograms, compared to 27% of those without. Among positive diskograms, only 23% exhibited end-plate changes on T2-weighted MRI |
| Yoshida, 2002 | 23 patients, 56 disks | Examined correlation between MRI and pain response on diskography | Sensitivity, specificity, positive predictive value, and negative predictive value of T2- weighted MRI were 94%, 71%, 59%, and 97%, respectively | Did not specifically compare diskography and MRI. T2-weighted MRI superior to gadolinium-enhanced images |
| Kakitsubata, 2003 | 24 disks from 5 cadavers | Compared MRI and MR-diskography with anatomic correlation for detecting annular tears | Sensitivity of MR-diskography was 100%, 57%, and 21% for radial, transverse, and concentric tears in the annulus, respectively, vs. 67%, 71%, and 21% for conventional MRI | Could not correlate findings with symptoms |
| Lim, 2005 | 66 patients with chronic LBP and no neural compression | Retrospective study comparing T1- and T2-weighted MRI findings with CT-diskography results (97 disks) | Concordant pain was more common with grade 4 or 5 degeneration on MRI, in disks with HIZs, and when the disk was fissured and ruptured on CT-diskography or the contrast agent spread into or beyond the outer annulus | Concordant pain was not associated with decreased disk height, end-plate abnormalities, or facet joint arthritis |
| Yuan, 2006 | 265 patients, 298 disks | Comparison of MRI and CT-diskography in patients with LBP and leg pain | 96.4% of patients were accurately diagnosed via CT-diskography. MRI found to be of limited benefit in diseased disks with passive diskographic findings | MRI found to be inferior to CT-diskography, especially for contiguous disks |
| Kim, 2009 | 23 patients, 24 disks | Evaluated APCD vs. MRI and CT-diskography findings in patients with LBP | Positive correlation between APCD and MRI/CT-diskography in grades 2 and 4 disk degeneration, but not in grades 3 and 5 | Pain-provoking pressure was not statistically correlated with MRI grading |
Related posts:
Full access? Get Clinical Tree
