Theoretical Aspects

David G. Vivian, MM, BS, FAFMM

Discography is used to determine if an intervertebral disc is the likely source of pain in a person presenting with referred pain thought to derive from an innervated spinal structure. If positive, it leads either to a target-specific treatment or to cessation of the search for the source of the pain. However, its methodology and usefulness as a predictor of successful target-specific treatment remain contentious.

The superiority of discography over other imaging methods to demonstrate morphology is not disputed. However, this advantage is of no particular use because morphologic change is nonspecific. Disc degeneration (DD) as determined by discography, magnetic resonance imaging IMRI), and other imaging tests occurs with similar frequency in symptomatic and asymptomatic populations.

The controversy about discography relates mainly to its subjective component. During discography, attempts are made to reproduce or relieve the patient’s pain. Researchers, cognizant of the inaccuracies of discography, continue to refine this component; future developments may focus on whether or not analgesic discography (AD) and functional analgesic discography (FAD) add to its validity.

There are four important aspects to the concept of discography that should be implicitly understood ( Table 38-1 ). First, although there are substantial clinicopathologic correlates in the lumbar spine, it does not define a specific pathologic process that might have rendered the disc painful. It is, in particular, a procedural test that attempts to identify a cohort of otherwise labeled nonspecific spinal pain (NSSP) patients with a particular type of discogenic pain (DP), called internal disc disruption (IDD). Second, in IDD there is no breach of the outer annulus fibrosus (AF); thus, when the disc is pressurized for the subjective part of the discogram, it is possible to reproduce the patient’s pain in the case of provocation discography (PD) and reduce pain in the case of AD, as the injected material is confined to the disc. Third, discography cannot identify patients with putative DP in whom there is a breach of the external border of the AF, as consequent leakage of injected material is prone to both false-negative results (the pressure is insufficient to propagate intradiscal nociception or local anesthetic concentrations are insufficient to block nociception) and false-positive results (due to anesthetization subsequent to epidural spread of the local anesthetic). Fourth, there is no definitively proven treatment that has arisen for patients with IDD.

Table 38-1

Four Fundamental Concepts of Discography

Discography is a quasi-subjective test that attempts to define the disc as a source of pain: it does not establish the pathophysiology of the pain.
In internal disc disruption (IDD) there is, by definition, disruption of the normal internal disc architecture, often with radial or circumferential tears in the annulus fibrosis.
Discography cannot identify patients with putative discogenic pain in whom there is a breach of the external border of the AF.
There is no proven treatment that has arisen for patients with pain with IDD.

History and Development of Discography

After 1934, when Mixter and Barr published on disc prolapse and neural compression, most back pain was thought to arise from disc prolapse, yet it became apparent that neural compression was found only in a relatively small number of patients with spinal pain. Furthermore, disc prolapse has been shown to be common in the asymptomatic population (for example, a prevalence of 37% in the thoracic spine).

Discography, pioneered from 1947 by investigators such as Inman and Saunders, Falconer and colleagues and Wiberg, arose before the disc was considered or proven to be innervated and at the time when the cause of back pain and radicular pain was considered to be disc prolapse. The first clue to the possibility of intrinsic DP occurred when back pain was produced by probing the posterior AF during lumbar disc surgery performed under local anesthesia. Despite the prevailing view that the disc was aneural, various authors reported that there was a relationship between disc morphology and reproduction of back pain; and, more particularly, that morphologically normal discs were rarely painful, in contrast to degenerative discs with posterior bulges and discs with epidural leaks on discography which were virtually always painful.

Researchers then attempted to determine if abnormal discs occurred in patients without pain. Massie and Stevens, who reported their findings on discography in 52 normal subjects and 570 patients, found that abnormal discs were associated with age and far more common in patients with pain than in asymptomatic controls, and that in asymptomatic subjects, abnormal discs were rarely painful on PD. After this PD fell into disrepute when Holt reported a false-positive rate of 37% in a group of asymptomatic prisoners, but a later in-depth study refuted this report on methodologic grounds. Subsequent to this there was a period of debate about the role of discography, leading in 1988 to a position statement on discography by NASS that emphasized the pain-response as the most important part of the procedure, thereby restating the fact that abnormal morphology is of minimal relevance when the diagnostic methodology seeks to establish if the disc is the source of an individual’s pain.

Then, using a more refined technique, Walsh and colleagues studied a selection of normal volunteers and patients with low back pain. In this study, the discographer was blinded in all respects, and the patient responses were filmed and assessed by two external observers. This study revealed that normal discs in asymptomatic volunteers do not hurt, and only abnormal discs in patients are painful. At this time, therefore, PD performed in this manner was considered a highly specific procedure for symptomatic (painful) lumbar discs.

The next development was the finding that axial view CT discography revealed radial dispersal of contrast medium in distinct patterns of radial and circumferential fissures in the AF. From this arose the Dallas discogram scale, later modified by Aprill and Bogduk, where Grade 0 describes contrast medium contained wholly within a regular NP, grades 1 to 3 describe extension of contrast medium along radial fissures into the inner third, middle third, and outer third of the AF, respectively, grade 4, in which grade 3 fissures extend circumferentially around the AF by at least 30 degrees of arc, and grade 5 in which the fissures breach the outer lamellae of the AF. The morphologic patterns of disc disruption and the innervation of these different areas lead to a consideration that there was a greater chance of the more peripheral fissures being painful. By this time it was established that the inner third of the AF is never innervated, the outer third is regularly innervated, and the middle third may or may not be innervated.

Subsequently, Vanharanta and coworkers corroborated this clinicopathologic model by determining that on PD grade 0 and grade 1 disruptions are rarely painful, in contrast to grade 3 disruptions, which are associated with exact or similar pain reproduction in 75% of cases. Moneta reported that only grade 3 and 4 fissures reproduced pain that closely approximated the patient’s reporting of pain, and that annular fissures were not a feature of disc aging or degeneration.

By the 1990s it was determined that for NSLBP there were firm correlations between the innervation of the disc, pain reproduction from it, and a demonstrable lesion, namely, IDD.

Since then, a number of researches, particularly Carragee, have studied PD in depth. The first criticism leveled on PD is that it is nonspecific and fails because it produces excessive false-positive results. It should also be noted that use of pain provocation is questionable in other pain diagnosis areas, such as in the detection of zygapophysial joint pain.

Studies that question the validity of PD include:

  • Patients who prior to a bone graft procedure had no LBP, and who developed back or buttock pain after the procedure, were subjected to PD; in this series of 8 patients, 50% were considered to fulfil the criteria for IDD.

  • In patients with persistent LBP after laminotomy and discectomy for disc prolapse, the relative pain intensity response to PD in these symptomatic subjects was compared to that of those who were asymptomatic after the same operation. It was found that, independent of psychometric profiles, a high percentage (about 40%) of this asymptomatic group had a pain response equivalent to what would be considered positive pain provocation on PD; this contrasted to a 70% positive PD pain response in the symptomatic group. Additionally, in both groups, patients with abnormal psychologic profiles had significantly higher rates of positive disc injections than either asymptomatic volunteers or symptomatic subjects with normal psychological screening.

  • Subjects without chronic LBP but with clinical profiles similar to those of symptomatic LBP subjects were investigated with PD. Of this group, 10 had no pain anywhere, 10 had chronic neck and arm pain, but no LBP; and 6 were considered to have “primary somatization disorders” without low back symptoms. Significant positive pain response and pain-related behavior were found in 10% of the pain-free group, in 40% of the chronic cervical pain group, and in 83% of the somatization disorder group completing the injections. Twenty-four subjects had negative control discs. It was concluded that the rate of false-positive discography is low in subjects with normal psychometric profiles and without chronic pain, but it is moderate in subjects with tears of the AF and chronic pain or abnormal psychometric testing.

  • Subjects with NSLBP with disability and pain of insufficient degree for them to seek treatment for this symptom were subjected to PD, and the results were compared to a group with NSLBP being considered for surgery. In the control group 36% of subjects had a positive PD including the demonstration of AF disruption to or through to the outer AF and negative control discs. The comparison showed that more prominent NSLBP subjects were PD positive in 73% of cases.

  • From a morphologic perspective, discography is not predictive, and moderate-to-severe Modic changes are weakly predictive, of future development of serious NSLBP in subjects with either mild persistent low back pain or some other mild pain symptom, whereas abnormal psychosocial risk factor profiles are predictive.

As discussed earlier, the false-positive rates for PD in a carefully selected group are consid1ered to be 10%, but this figure increases to at least 25% in patients with abnormal psychologic profiles and in those with pain in other regions.

Discography as Criterion Standard Test

In medicine, a primary aim is to make a diagnosis based on solid criteria that definitively establish if a specific disease or condition exists. The criteria that underpin a reliable diagnosis are measured by diagnostic tests. Examples of the best criterion standards include pathologic specimens to diagnose cancer and infection, imaging for bone disease and fracture and angiography to detect vascular disease. Certain standards have been iterated for rating the tests used to confirm a diagnosis.

Some criterion standard tests are not routinely used because they are expensive, inaccessible, invasive, and risky. As a consequence, other diagnostic tests are used as surrogate criterion standard tests. Although surrogate tests might be cheaper, more available, less invasive, and safer; they might also be less accurate.

Robust diagnosis is not possible in many areas of pain medicine, and especially in the diagnosis of NSSP, because of the lack of a good criterion standard test. In putative DP even tissue biopsy is unhelpful in determining whether or not the disc is a source of pain. MRI may have a role in the detection of DP, but overall it has not been accepted as valid because it is not generally predictive of pain. Thus, in NSSP there is no criterion standard based on pathology, and no simple imaging test acting as an acceptable surrogate. Discography has been advocated as the best test, but it comes with costs of limited availability, some risks, limited reliability, and less than ideal validity.

The diagnostic dilemma encountered where there is no reasonable criterion standard is overcome by using a classification system that is based not on pathology, but on a manifesting symptom. Notable examples of such well recognized classification systems predicated on symptoms rather than on robust criterion-based tests include the classification system for psychiatric disorders, called the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV), and the classification system for headache, called the International Classification of Headache Disorders (ICHD). DSM-IV is a categoric classification system for psychiatric diagnosis in which the categories are prototypes, and a patient with a close approximation to the prototype is said to have that disorder. The ICHD classifies all headache disorders into major groups and each group is then subdivided one, two, or three times into headache types, subtypes, and subforms. Migraine is a diagnosis in this system. There is no test that confirms the diagnosis. It is a symptom-based diagnosis which has subtypes of migraine such as “migraine with aura,” further subgrouped into subforms such as “typical aura with migraine headache.” The ICHD uses all kinds of available evidence (clinical description, longitudinal studies of cohorts of patients, epidemiologic studies, treatment results, genetics) to establish such a diagnostic entity.

A diagnosis is required not only for individual patient management, but also so that diagnosis based management can be evaluated for all populations. Discography reveals the likely source of pain for some patients presenting with otherwise classified NSSP. It also provides a mechanism for the pain, namely IDD. The validity of the diagnosis IDD, and hence the assessment of treatment related to IDD, is predicated primarily on the validity of discography ( Table 38-2 ).

Table 38-2

Diagnostic Criteria for Discogenic Pain Secondary to Internal Disc Disruption (IDD)

Patient’s pain is reproduced by stimulation of the affected disc.
This pain must have an intensity of at least 7 on a 10-point scale.
This pain must be reproduced at a low to intermediate pressure of stimulation: 15-30 psi above opening pressure (1-3 kg/cm 2 )
Stimulation of adjacent disc(s) must not reproduce pain.
Postdiscography CT must demonstrate a grade 3 or 4 fissure.

From International Spine Intervention Society (ISIS, 2004). All categories must be satisfied. These criteria are being modified.

Disc Morphology


There is a gradual transition of disc morphology from the cervical spine, where movement is more relevant than loading, to the lower lumbar spine, where intervertebral movements are smaller, and load transfer is maximal.

The lumbar discs consist of a central NP surrounded by the AF which consists of concentric laminae of collagen fibers thickest anteriorly and laterally. In contrast, cervical discs consist of a crescent-shaped anterior interosseous ligament with thick anterior collagen fibers that taper laterally toward the uncinate processes; they are essentially deficient posterolaterally where there is only a thin layer of paramedian, vertically orientated fibers. The anterior longitudinal ligament covers the front of the cervical disc, and the posterior longitudinal ligament reinforces the deficient posterior AF with longitudinal and alar fibers.

Cervical discs have a different chemical morphology to thoracic and lumbar discs. In particular, they contain a higher collagen content in the NP and higher glycosaminoglycans content in the AF, both of which allow for the greater demands of bending and twisting movements in the cervical spine compared to other spinal regions.

Because discs are innervated, they have the capacity to be sources of pain. The normal disc is avascular and aneural except for the outer third of the AF. Lumbar discs are innervated by the lumbar sinuvertebral nerves, and branches of the lumbar ventral rami and grey rami comunicantes. There are five types of nerve terminations found in the lumbar disc: these have various morphologies and include simple and complex free nerve endings that concentrate particularly in the lateral disc, with a smaller amount posteriorly and the least amount anteriorly. The posterolateral portion of lumbar discs are innervated by the lumbar ventral rami that arise just lateral to the intervertebral foramen and by a branch of the grey rami communicans just before it connects with the ventral ramus. The sinuvertebral nerves also innervate the posterior longitudinal ligament, and the grey rami communicantes also innervate the lateral disc and the anterior longitudinal ligament. Additionally, the innervation by the grey rami communicantes is not a direct sympathetic innervation; it has been postulated that somatic and afferent fibers from lumbar structures use the grey rami as transmission pathways only.

Cervical discs are innervated laterally by branches of the vertebral nerve and more generally by branches of the cervical sinuvertebral nerves; these nerves penetrate as deeply as the outer third of the AF. The end-plate adjacent to the NP is also innervated.

In discogram positive discs, histologic studies have shown that there is there is a more extensive and deeper innervation of the disc, with free nerve endings expressing substance P reaching into the outer part of the NP, as well as the adjacent end plate and vertebral body. The ingrowth of nerves in the discogram positive disc is within a region of vascularized and innervated granulation tissue that accompanies fissures extending from the AF to the outer part of the NP. These nerves contain increased substance P (SP), neurofilament (NF), and vasointestinal peptide (VIP) immunoreac-tive nerve fibers, and are consistent with the concept of DP.

Disc Degeneration

Disc degeneration and similar morphologically descriptive terms including degenerative disc disease and zygapophysial joint spondylitis, are imaging-related labels that should not be used as diagnostic labels for a person presenting with NSSP. These terms describe mid- to late-stage degenerative change seen on imaging techniques. Although there may be a mild link between some imaging changes and NSSP, there is at present no reliable or valid method to detect this relationship in any individual case.

Degeneration is an inevitable process. The process of degeneration consists of gradual structural change, which at times is accompanied by altered homeostasis and changes in innervation. Although such a process might be painful in its own right, or might determine that the whole intervertebral segment is prone to pain, there does not appear to be an inevitability that pain will appear. Thus, DD can be observed, but at this stage of science, little if any pain attribution can be assigned to it. Put simply, DD is not DP!

The IVDs are mainly composed of extracellular matrix molecules and have only about 1% by volume cell content, with the content of cells decreasing from the periphery inward, reflecting the relative avascularity of the disc. These cells are of vital importance for disc homeostasis because they regulate the synthesis and metabolism of the extracellular matrix.

DD is represented at a molecular level by a relative rise in the quantity of senescent cell population on the background of an overall fall in the total number of cells within the disc. The percentage of senescent cells within a disc is also inversely related to the ability of other cells to proliferate, and the relative quantity of senescent and proliferating cells is independent of age. DD ends with structural failure, represented macroscopically by thickened vertebral end plates, increased cracks and fissures in the matrix, and delamination and tears in the AF. The radiologic endpoint of DD is disc space narrowing and osteophyte formation.

Cellular function within the disc is mediated by at least five major factors: genetics, nutrition (diffusion of nutrients and oxygen across the disc matrix), cell function regulation via IL-1, TNF-α and TNF-β, age and senescence, and mechanical loading. The contribution to DD by genetic factors is highly significant; it may be as high as 80% in the cervical spine, and general heritability for DD ranges from 34% to 61% in different regions of the spine. In the lumbar spine, the genetic contribution is between 29% and 54%, with environmental influences of about equal importance. For example, smoking has a moderate influence on the prevalence of DD, presumably due to its effects on disc nutrition. This emphasizes that DD is not primarily or significantly due to aging or to mechanically induced “wear and tear” processes.

Studies on symptomatic and asymptomatic populations with various imaging modalities emphasize that DD does not imply NSSP. In the cervical spine, radiologic DD is present in 13% of men and 5% of women during the third decade, in 85% to 90% of the population by the sixth decade, and nearly 100% by the age of 70. It occurs most commonly at C5-6, C6-7, and C4-5, respectively. In people aged 60 to 65 years without neck pain, about 95% of men and 70% of women have at least one degenerative change on their cervical spine plain x-rays. Although plain x-ray changes, including vertebral end-plate changes, disc space narrowing, spondylolisthesis, spondylolysis, sacral lumbarization, wedge vertebra, a sagittal diameter of less than 12 mm, and abnormal lumbar lordotic angle, have some predictive value for low back pain, the relationship is mild at best, and their detection is largely not helpful in the management of NSLBP because changes occur frequently in the asymptomatic population. As a consequence, plain x-rays should not be ordered unless there is suspicion of a red-flag condition.

The relevance of computed tomography (CT) scanning is similar to plain x-rays; it is an excellent test for some red-flag conditions, demonstrates DD well, but it is not helpful in the detection of DP. CT is better than MRI in detecting ZJ spondylitis, but this is of no particular clinical relevance. F-PET/CT (fluoride positron emission tomography with addition of CT) may have a role to play in that it is more likely to be positive in symptomatic patients, and in that it might identify the source of pain.

Plain x-rays combined with MRI provide the best method to detect DD. Plain x-rays are perhaps better than MRI at detecting early stages of DD; MRI detects later stages of DD. However, such information is only relevant to research because the detection of DD does not help in the diagnosis of NSSP. Studies of asymptomatic subjects have shown degenerative rates of more than 30% in the general population.

Discography is the best method to detect gross disc morphology. Fissures can be detected best on CT imaging postcontrast injection. However, although this component of discography detects degeneration, such a finding does not mean that the disc is a source of pain. For this reason, discography evolved from a purely morphologic test to one that attempts to detect whether or not the disc is painful. Discography, like other imaging modalities, should not be ordered in an effort to detect degeneration because degeneration is not a legitimate diagnosis for NSSP.

Discography and Pain

The relative uselessness of morphologic changes in the search for a source of pain has lead to the search for the painful source using two mechanisms; pain provocation and pain reduction.

Discography can theoretically cause pain from the disc itself by two fundamental mechanisms, annular distention and end-plate deflection. Annular distention during discography can initiate nociception because the peripheral AF is innervated. In vitro study has shown that in radiographically nondegenerated discs, discography causes measurable end-plate deflection ranging from 0.12 to 0.69 mm centrally and from 0.06 to 0.35 mm at locations between the center and the periphery of the end plate. Hsu investigated 692 discs in patients with back pain; he showed that during PD end-plate disruption is uncommon (2% of all discograms) but that when it occurs even with lightly induced pressure, the prevalence of moderate or severe fully concordant pain is very high (in this series, 13/14 or 93%); in contrast, 42.3% of the remaining discs without end-plate disruption had moderate-to-severe fully concordant pain on PD, and 57.7% had either mild or zero pain.

Pain reproduction during the disc stimulation component of PD is considered integral to its validity. All discs, if provoked enough, will produce pain. However, a study on asymptomatic subjects revealed that even if grade 3 annular tears are present, pain induced on PD is likely to be mild—even at high manometric pressures. From this study, pressures below which pain does not occur or it occurs at low intensity have been identified for asymptomatic subjects. The assumption is that this effect should be mirrored in studies of symptomatic patients, so that manometric pressure readings can be titrated against pain scores to set parameters above which the procedure can be called positive. For example, at a pressure of 50 psi any pain score above 6/10 is not seen in asymptomatic volunteers: at a pressure of 40 psi, the cut-off score is 5/10. Put another way, there is close to zero chance of an asymptomatic subject having any pain at 20 psi or less, and of having pain greater than 6/10 up to a pressure of 70 psi. When the criterion for a positive PD pain response is 6/10 or more at 50 psi, the false-positive rates in asymptomatic subjects falls to 10%.

Derby and colleagues then found that there was a correlation between postdiscography CT scan levels of annular disruption and the pain score; using VAS of 6/10 or more at less or equal to 50 psi above opening pressure, they found that these parameters revealed negligible positive results at annular disruption below grade 3, and 44% positive results with grade 3 or greater annular disruption, with severe pain being proportional to the pressure in grade 4 and 5 discs.

Subsequent to this study, Carragee and coworkers found the false-positive rate of low-pressure painful disc stimulation injections in subjects without chronic NSLBP to be about 25%. However, there were some drawbacks in this retrospective study including that some of these subjects had undergone previous back surgery. The conclusion from this study is that although the false-positive rate of PD may be as high as 25% in an imperfectly selected asymptomatic population, this figure is likely to be less in subjects who do not have a chronic pain state or who have not had previous spinal surgery.

In an effort to decrease the rate of false-positive findings, the methodology of discography is undergoing modification. The development currently is in AD, in which various local anesthetic agents are injected into the disc, and FAD, in which a balloon-tipped catheter is placed into the disc so that provocation maneuvers can be undertaken in the recovery room.

Clinical Application

By the time the specialist interventional clinician sees a patient with chronic severe recalcitrant NSSP, this patient will have usually seen numerous doctors, manual therapists and other health practitioners, had a variety of treatments, and be utterly confused about diagnosis and treatment. By this time, the patient often wants to understand the problem. Although different theories and models abound, an approach that is pathologically based and underpinned by anatomy has some logic and merit. Patients seem to understand this model when it is explained.

At this stage of management, there are two divergent pathways. The first is to cease further attempts at diagnosis and treatment and to continue to manage the problem using pain management principles. The second is to attempt to make a tissue-specific diagnosis and as a result, to deliver a target-specific interventional treatment, despite the fact that the assessment leads to a good outcome in less than 50% of cases.

The procession into this type of management should be understood by patient and clinician to be somewhat experimental. Patients should be alerted to the fact that the discography is a controversial test and that the management of IDD is contentious. Some basic issues that need to be understood by the patient at this stage include an understanding of referred pain and of the relative uselessness of special tests in the diagnosis of these referred pain problems.

Referred Pain is not the Same as Radicular Pain

In presentations of axial pain, one major hurdle for the patient to overcome is the differentiation of referred pain from radicular pain. Most patients seem to consider that if the pain travels into the leg from the back that it is sciatica caused by a “pinched” nerve. It is important to understand the concept of referred pain because most pain in the limbs from the spine is referred pain. Radicular pain is relatively uncommon. One simple example of referred pain is angina, and this is a good model to discuss to indicate to the patient that limb pain can come from structures other than nerves.

Special Tests Rarely Substantiate a Diagnosis

As discussed earlier, MRI, the most morphologically accurate test, can be used to identify profiles that render a patient more likely to have a positive discogram or at least an increased likelihood of NSLBP. MRI and other tests cannot provide a definitive diagnosis of pain pathophysiology or source in a patient with NSSP. Most patients have had or want MRI, and most have had plain x-rays and CT scans. Often they have been provided with an imaging-derived morphologic diagnosis, such as disc degeneration or facet joint degeneration, or the more pejorative and equally illegitimate label, “disc degenerative disease.” Perhaps even worse, they might have been labeled with the more personally challenging negative finding; “we have not found anything at all wrong on these tests,” which might have the implication that the pain is invented or at least “psychological.” These tests should be understood for what they achieve. First, they have a varied ability to exclude red-flag conditions. Second, they are particularly useful as a tool to establish the cause of radicular pain. Third, by and large they cannot establish the cause of NSSP.

Diagnosis is Rarely Tissue or Pathology Specific

Diagnosis is particularly vexing; it needs to be accepted by all that in pain practice, diagnosis is often generalized to descriptions of the pain itself, and that illegitimate labels abound. NSSP is one label that can be used for a presentation of spinal somatic referred pain that is not caused by any other process or named condition. NSSP means pain, be it local or referred, that appears to derive from one of the locally innervated structures of the spine and does not derive from any known pathologic process.

The prime indication for discography is NSSP of sufficient longevity, dimension, and character that the clinician and the patient consider that some further step needs to be taken to achieve significant pain relief. There is no consensus regarding acceptable parameters above which discography might be performed. It is generally stated that a patient should have exhausted all conservative measures including multidisciplinary management, pain management programs, and rehabilitation before being considered for intervention including discography, and that the likely medium- to long-term scenario is for persistent substantial disability and pain.

The problem for the clinician is what to recommend to a patient with chronic disabling NSLBP. Discography is useful only from an interventional perspective if it rules in or rules out some form of treatment; options are confined to intradiscal procedures and spinal surgery. But these treatments have not been shown to be effective for the general population of patients with DP established on PD. Conversely, rehabilitation and pain management programs also have not had significant impact on the management of NSSP, although exercise programs combined with fear-avoidance application have proven merit in the prevention of chronicity, and pain management programs are often recommended as being more effective than traditional medical conservative treatment for chronic pain in general. In any case, the decision to move into interventional management for any patient typically occurs after such attempts have failed.

Predicting long-term outcome is probably impossible in individual cases. Trends exist: but the course of back pain is not as optimistic as has been conveyed. Although about 76% of patients with a history of recent (within 6 months) low back pain recover or have low levels of pain, 14% have persistent, high disability with moderate-to-severe limitation of function.

Will the patient with severe disabling back pain ever get better? Many do not; however, some do. Smith performed a retrospective study of 25 NSLBP patients with positive PD in whom surgery was considered to be indicated yet was not undertaken. At a mean follow-up time of 5 years, 68% had improved, 8% were the same, and 24% had worsened. Of those who had become worse, 67% had psychiatric disease. As an aside, in this study 80% of those receiving workers’ compensation improved.

If the patient does go to surgery consequent upon PD, what will be the long-term result? Knox and Chapman performed a retrospective study on 22 patients who had anterior lumbar interbody fusion for discogram-concordant lower back pain and found that all double level fusions did poorly and in single level fusions, 35% had good, 18% fair, and 47% poor results, with previous surgery and workers’ compensation doing the worst. Parker prospectively studied 23 patients treated by a single surgeon with posterolateral fusion. All underwent preoperative discography and were monitored for an average of 4 years postoperatively, at which time 39% had a good-to-excellent result, 13% a fair result, and 48% a poor result, with 90% of those on compensation having a poor result. Wetzel performed a retrospective review of 48 patients with low back pain who had discogram/CT, then lumbar arthrodesis, with a success rate (good outcome) of 46%. Carragee performed total discectomy and fusion on patients with positive discography and achieved highly effective outcome in 27% of patients, and minimally acceptable or better outcome in 43%.

Is discography useful to detect pain possibly derived from discs adjacent to another disc that is to be fused? Willems found that PD adjacent to an intended to be fused segment (determined by temporary external transpedicular fixation) in patients with persistent back pain was not helpful and that patient outcomes after lumbar fusion were the same whether or not the adjacent disc was normal or degenerated.

Discography has also been used as a pretest for intradiscal procedures including intradiscal electrothermal therapy (IDET) and nucleoplasty. Results from these procedures suggest that there may be cohorts within the IDD subtype that do well.

What tests such as discography try to do is to develop a model that will predict which diagnostic subtype in which clinical scenario will do best with a particular treatment. It is a test that is not commonly performed; it is performed only on the most difficult presentations, and is done in interventional settings on a minimal number of patients presenting with spinal pain.

Can Other Tests Replace Discography?

How Useful is MRI in the Assessment of NSSP and IDD?

MRI changes occur in the asymptomatic population. In the lumbar spine, Boden and colleagues studied subjects with no past history of LBP and sciatica; in those younger than age 60, 20% had a herniated disc; in the 60 years and older group, 36% had a herniated disc and 21% had spinal stenosis. Disc degeneration or bulging at one or more levels occurred in 35% of subjects between 20 and 39 years, and in all but one of the subjects aged 60 to 80 years. MRI findings are common in asymptomatic individuals and the extent of changes is not predictive of the development or duration of future back pain.

In the evolution of PD, attempts have been made to ascertain whether or not MRI can be used to predict PD outcome because PD is not without morbidity. Although MRI is somewhat predictive of a positive PD test when a high intensity zone (HIZ) or Modic type I or II changes are found, it is somewhat insensitive.

The HIZ, defined on sagittal MRI as a very bright signal (equal to or brighter than CSF on T2 weighted scans) contained within the posterior AF, has been shown to have appreciable but variable correlation with positive PD in patients presenting with NSLBP. The initial study showed that an HIZ increases the odds that PD will be positive in that patient at that level by a factor of 6.5. Studies on the HIZ since that time have all shown that HIZ is highly specific (range 0.74 to 0.93) but variable sensitivity, from as low as 0.09, to as high as 0.78. 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 a PD will be positive by at least 50%. Additionally, although HIZs are present in both asymptomatic and symptomatic subjects, they are found significantly more in symptomatic (prevalence 60% ± 15%) than asymptomatic subjects (24% ± 11%).

Modic changes have also been studied in relationship to NSLBP and PD. The overall rate of vertebral end-plate signal changes (VEPSC) is about 43% in patients with NSLBP and 6% in those without. The most common association of VEPSC and NSLBP is extensive Modic type I changes at L5-S1.

Overall, VEPSCs are relatively insensitive, but quite specific, for the diagnosis of IDD established by discography, with sensitivity of 23.2%, specificity 96.8%, PPV 91.3%, and NPV 46.5% in one study. When Modic changes are studied according to type, the type I end plate has the highest PPV for positive PD at 81% (±7%).

It is possible for PD to be positive in the presence of a normal MRI. Zucherman reported on 18 symptomatic patients who had positive PD an average of 2.5 months after a normal MRI; positive meant production of concordant pain and abnormal disc morphology. This study was performed more than 20 years ago and whether or not this applies with current MRI technology is unknown.

Can Clinical Tests Predict IDD?

The prevalence of IDD in a population of NSLBP patients thought to have sufficient pain and disability to undergo PD is about 35%. The prevalence of IDD in a group of patients with positive clinical indicators increases to 52% to 69%. The centralization phenomenon, or pain centralization, which is the retreat of referred pain toward the spinal midline during specified clinical examination, a history of persistent pain between acute episodes, a significant loss of extension and a subjective report of so-named “vulnerability in the neutral zone” individually and in combination increase the likelihood of an eventual diagnosis of IDD in a group of highly disabled and psychosocially distressed patients with otherwise diagnosed NSLBP. Additionally, if vibration applied over individual spinous processes is considered a positive test, then the odds of a positive PD increases significantly.

Technical Performance of Discography

Optimal lumbar discography is based on consensus guidelines published by International Spine Intervention Society (ISIS). However, optimal technique is uncommon; in a survey of physicians performing discography, the adherence to patient safety guidelines were as follows: preoperative antibiotics (83.81%), intradiscal antibiotics (84.97%), postprocedure antibiotics (9.82%), double-needle technique (64.16%). Additionally, the adherence to technical guidelines were as follows: optional use of CT scan (64.78%), pain assessment sheet (66.47%), entering on the side opposite symptoms (48.55%), manometry for opening pressure (65.31%), manometry of pain reproduction pressure (72.25%), injecting a control disc first (78.61%), injecting discs adjacent to the painful disc (56.64%). It is also evident that there is considerable variation from one specialty to another in discography performance.

Interpretation of Discography

There are three cardinal components that need to be recorded separately but viewed collectively; (1) provocation/analgesia, (2) disc manometry, and (3) disc morphology. These three different aspects are required to write a discogram report, and all three elements are still evolving.

The International Spine Intervention Society (ISIS) describes positive PD as follows: (1) reproduction of the patient’s pain by stimulation of the affected disc; (2) such that the evoked pain has an intensity of at least 7 on a 10-point scale; and (3) pain is reproduced at a pressure of stimulation of <15 psi; (4) provided that stimulation of adjacent discs does not reproduce pain; and (5) postdiscography CT demonstrates a grade 3 or 4 fissure.

Provocation/Analgesia Assessment

Although this phase occurs as the last step in discography, it is the most important aspect to record because it, above all, is the component that separates discography from other tests. Because this phase requires the patient to report subjective pain responses, it is open to reporting bias and other error.

The optimal setting for the provocation component of discography is as follows: (1) the patient needs to be calm and comfortable, (2) the person recording the information from the patient should not be the injector, (3) that person and the patient should be blinded to the disc being injected.

The degree of pain production should be recorded. It is not possible in many instances to use a visual analog scale in the operating room during the procedure. If so, a verbal pain rating is used. The patient should understand beforehand that this is to be used and it should be ascertained that the patient understands how to communicate pain with 0 to 10 as being no pain and 10 of 10 being the most severe pain that he or she can imagine.

An example of the recording that can be used for PD is:

  • P0 No pain response is noted on injection or distention of the disc.

  • An equivocal response; vague, uncharacteristic, or discordant pain both by nature and location.

  • P+ Definite, convincing pain provocation that is familiar to the patient yet only reproduces part of the symptom complex.

  • P++ Exact pain reproduction, concordant with the symptom complex.

  • Analgesic response can be recorded as follows (R if for response):

  • R0 No change in pain after injection.

  • A vague, uncertain response, denoted as less than 2 on a 10-point VAS scale.

  • R+ A definite improvement of all or part of the presenting symptoms by 2/10 or more on the VAS.

  • R++ Complete relief of all or part of the manifesting symptoms.

Disc Manometry

Disc manometry is an estimate of the hydrodynamic competence of a disc. It is formally measured with manometric pressure systems. In IDD, NP pressure decreases irregularly, and AF stresses rise. The depressurization of the NP reflects the degradation of the nuclear matrix, which can no longer retain water efficiently, to sustain axial loading. This results in extra load having to be borne by the posterior AF. Both of these findings correlate with a positive provocation component.

Disc Morphology

The morphology of the disc is revealed during the contrast injection phase of discography. Under image intensification, AP and lateral views are used routinely. In a normal discogram, contrast remains contained within the NP, and it has smooth, regular margins. When abnormal, contrast is seen to flow from the NP into the AF, and when the disc is ruptured it flows into the epidural space.

Postdiscography CT scan performed within 3 hours of the procedure provides more information about disc morphology and is said to improve diagnostic accuracy, at least for disc prolapse. CT-discography importantly provides a better ability to subtype PD according to the Dallas system. Where CT is not available, caudad gantry views can show the discs close to AP.

Cervical and Thoracic Discography

It is not possible to determine if a cervical disc is a major source of pain using MRI or other imaging techniques, particularly as MRI findings do not correlate well even with advanced histologic DD. However, the cervical disc is likely to be a source of pain. If cervical DP is diagnosed, the ability of such knowledge to improve outcome in the management of neck pain is questionable. Cervical discography has hardly been studied, but has advocates. A review of cervical provocation discography, which includes disc stimulation and morphologic evaluation, found that there were few studies of good standard, and concluded the cervical discography is a useful tool for evaluating chronic cervical pain, without disc herniation or radiculitis, with the strength of evidence for diagnostic accuracy being at Level II-2. However, because the morphology of the cervical disc is substantially different from the lumbar disc, no comparative conclusions can be made from the studies on lumbar discography. The cervical AF does not consist of concentric laminae of collagen fibers as in lumbar discs; it is a crescentic mass of collagen that is thick anteriorly and tapers laterally toward the uncinate processes, and is deficient posterolaterally. There are no established postdiscography CT scan findings such as is seen in the Dallas description of these in the lumbar spine. There are no clinicopathologic correlations that underpin cervical, or thoracic, discography.

The technique of cervical provocation discography was first published in 1957, having been pioneered by Cloward, and subsequently developed. In the lumbar spine, abnormal morphology and precise pain reproduction are two of the more important components that contribute to a positive discogram. This is not the case in the cervical spine because fissuring of cervical discs is not reflective of trauma but of normal anatomy and aging, and, hence, extravasation of contrast from the disc in cervical discography cannot be considered to be indicative of an abnormal disc. The essential component of cervical discography is, therefore, not morphologic, but it is in what happens during the injection or “stimulation” phase because stimulation of cervical discs in asymptomatic volunteers is painless or only minimally painful, meaning that cervical disc stimulation has at least some negative predictive value: if the disc does not hurt when stimulated, it is unlikely to be a source of pain. Additionally, to diminish the possibility of false positives, subsequent opinion and study has discerned that adjacent structures should be similarly stimulated to ensure that pain is not apparent in multiple sites.

If cervical discography is used, it should be understood that it is in the very experimental stage of evolution. If performed, the recommended protocol is to perform multilevel procedures that stimulate pain: positive disc stimulation occurs when: (1) provocation of the target disc reproduces the patient’s (concordant) pain, and (2) adjacent site stimulation does not reproduce pain.

It is evident from the use of cervical facet joint blocks and disc stimulation that more than one source of pain can be detected at any one cervical intervertebral segment. As a consequence, when disc stimulation is used, a positive test requires: (1) reproduction of the patient’s concordant pain at a magnitude of at least 7 on a 10-point VAS, (2) exclusion of cervical facet joint pain at any segmental level where disc stimulation is positive because the false-positive rate without exclusion is 67%, and (3) not finding reproduction of the patient’s pain in adjacent structures.

Thoracic discography is even less studied. In a study on 10 asymptomatic subjects and 10 patients with chronic thoracic pain undergoing PD, it was found that: (1) even in asymptomatic people, thoracic discography could be intensely painful, especially when prominent Schmorl nodes were present, (2) MRI changes did not always correlate with disc morphology on discography, and (3) in the symptomatic group, 50% of studied discs were concordantly painful.

Complications of Discography


The use of sterile technique and single- or double-needle technique seem to be the minimum requirements for the prevention of infection. Prophylactic antibiotics may decrease the incidence of discitis.

Discitis is a nasty complication. A review of cases revealed the following:

  • It is accompanied by, at times, severe exacerbation of local pain.

  • There is an accompanying but delayed rise in ESR—averaging 20 days to reach a peak.

  • Bone scan changes peak at about 33 days, and at 18 days it is more common to have no bone scan changes.

  • MRI changes are also delayed, and if they occur in the first 2 to 3 weeks postdiscography they are likely to represent new pathology.

  • The course of lumbar discitis ranges from 8 to 11 weeks.

  • The course of cervical discitis ranges from 6 to 7 weeks—usually resulting in spontaneous fusion.

There are no randomized or controlled trials on the efficacy of prophylactic antibiotics. Based on animal experiments, the incidence of postdiscography discitis is lowered with the use of cefazolin given intravenously or in the intradiscal suspension injection. In high concentrations, contrast medium can itself inhibit bacterial growth at least to some degree, and it does not reduce the inhibitory effect of antibiotics. A follow-up of 127 consecutive patients having lumbar discography with contrast containing cefazolin 1 mg/mL revealed no cases of discitis. Willems and colleagues had no cases of discitis when they performed two-needle technique discography without preemptive antibiotics on 200 patients and followed them all for a minimum period of 3 months. The reported overall incidence of discitis without prophylactic antibiotics is in the region of 0.25% of all patients or 0.094% of all discograms.

It is of course important to know where the injectate is going. Initial injection of antibiotic containing material should not be made if the needle tip site is unknown. For example, intrathecal administration of cefazolin can be fatal.

Can Discography Cause Harm?

Experimental animal studies have revealed that disc puncture can induce accelerated disc degeneration. Carragee followed two matched cohorts of 75 each without serious LBP “illness.” Each subject was assessed with MRI and one group underwent L3-4, L4-5, and L5-S1 discography examination in 1997. Between 7 and 10 years after enrollment, each subject underwent a further MRI: 50 discography subjects and 52 control subjects were reassessed. The results were summarized as follows: (1) in all graded or measured parameters, discs that had been exposed to puncture and injection had greater progression of degenerative findings compared to control (noninjected) discs: progression of disc degeneration, 54 discs (35%) in the discography group compared to 21 (14%) in the control group; 55 new disc herniations in the discography group compared to 22 in the control group (2) new disc herniations were disproportionately found on the side of the AF puncture, (3) the quantitative measures of disc height and disc signal also showed significantly greater loss of disc height and signal intensity in the discography disc compared to the control disc.

Additionally, discography can produce long-term pain in subjects with abnormal psychological profiles. A 12-month follow up of subjects without LBP revealed that no subject with normal psychometric testing had persistent pain after discography, but pain of those with abnormal psychometric test results reported significant new low back pain.

These facts mean that discography should not be undertaken unless there are strong indications that the results will substantially alter the management and that the management changes have a reasonable chance of producing a far better outcome than might be obtained otherwise. The second part of this chapter discusses the technical and radiographic aspects of discography.

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Apr 13, 2019 | Posted by in PHYSICAL MEDICINE & REHABILITATION | Comments Off on Discography

Full access? Get Clinical Tree

Get Clinical Tree app for offline access