Abstract
Discogenic pain, or pain arising from the intervertebral discs, is a common cause of chronic spinal pain, constituting approximately 40% of cases of low back pain. Discs invariably degenerate as we age, such that most people have MRI evidence of disc degeneration by their 4th decade of life, and by their mid-60’s, over 90% of people have significant disc degeneration. The high prevalence of spinal pain, coupled with the high rate of disc degeneration in asymptomatic volunteers in all regions of the spine, indicate the need for a test that is capable of connecting symptoms to pathology. Yet, discography is associated with a high false-positive rate in select patients (e.g., those with psychopathology and prior surgery), and in certain individuals, may predispose to disc herniations and accelerated degeneration. In the lumbar spine, there is weak, positive evidence to support discography as preoperative test to select people for spinal fusion, while the evidence to support cervical discography is very weak and based on old, uncontrolled studies. Discitis is the most feared complication of discography, with an incidence of less than 0.5%. Steps that can be taken to prevent discitis include using a double-needle technique, adhering to strict aseptic protocol, and though not proven to definitively reduce disc infection, administering prophylactic systemic or intradiscal antibiotics.
Keywords
degenerated disc, discogenic pain, discogram, discography, intervertebral disc
Discography has been called a “test in search of an indication” and a “solution in need of a problem.” Originally employed as a diagnostic tool for herniated discs in the era prior to the advent of advanced imaging, its use in this capacity has been almost completely supplanted by safer, cheaper, and more sensitive modalities such as magnetic resonance imaging (MRI). Over the intervening years, discography continued to be used, evolving from a defunct imaging tool to the only ostensible means to correlate imaging with symptoms. As a diagnostic and prognostic tool, disc stimulation remains one of the most controversial interventional pain procedures, with many physicians and, more importantly, third-party payers refusing to perform it or to authorize its use. Yet new research published since the last edition of this book has demonstrated that at least in some cases, discography may improve surgical outcomes.
Overview of Spinal Pain
Pain originating from the spine commonly manifests as pain in the low back and neck and less frequently as pain in the midback. Although many components of the spine are capable of generating pain, its exact source is often elusive. Several factors make the identification of spinal pain generators challenging. First, back pain can originate not only from various spinal column components but can also be referred from structures adjacent to the spine, such as abdominal or pelvic viscera, sacroiliac joints, and so on. Second, pain can be difficult to localize due to multisegmental innervation with resultant convergence in the spinal cord. The diagnosis of spinal pain is further complicated by the concurrent presence and overlapping clinical features of various spinal disorders, especially degenerative conditions. The lack of diagnostic tests that can reliably identify a spinal pain generator further adds to these challenges. Currently available tests, often based on high-resolution imaging, frequently show abnormal findings in asymptomatic individuals at all spinal levels. Because of the frequent spontaneous resolution of symptoms, the high incidence of benign abnormal findings, and the rarity of serious spinal disorders, indiscriminate diagnostic testing of patients with spinal pain can lead to inappropriate diagnoses and poor treatment results.
Mechanisms of Discogenic Pain
Although the role played by a herniated nucleus pulposus (NP) in causing spinal pain is well known, the concept of pain originating from the disc itself is less understood. The term internal disc disruption (IDD) has been used since the early 1970s to describe a disc that is considered the main source of a patient’s pain but appears functionally intact on spinal imaging. However, degenerative disc changes seen on spinal imaging are nearly ubiquitous, especially with advanced age. These myriad changes are collectively referred to as degenerated disc disease (DDD) and may represent normal age-related phenomena.
Isolated degenerative disc pathology, where one or two discs show profound degeneration in the presence of other relatively normal-appearing discs, is less common and more frequently encountered in younger individuals. Whether IDD and DDD are distinct pathologic entities or represent pathologic progression of the same disease entity is unknown. The terminology surrounding pain stemming from intervertebral discs can be confusing, with one systematic review finding that DDD is most commonly used in studies evaluating surgical treatments. The term discogenic pain (DP) describes a clinical state where the disc is considered the main source of spinal pain. For the present discussion, this term seems most appropriate, as it emphasizes the disc as the primary source of a patient’s pain irrespective of pathology.
A basic understanding of normal disc physiology is imperative to understand the mechanisms responsible for DP. A normal disc is grossly compartmentalized into its NP and annulus fibrosus (AF). Interspersed in abundant intercellular matrix in the two disc compartments are sparsely present cells. The cells populating the NP are chondrocyte-like, while those comprising the AF more closely resemble fibrocytes. The intercellular matrix in the NP is a “jelly-like” substance containing high concentrations of water and proteoglycans, while the matrix in the AF consists predominantly of type I and II collagen fibers. These fibers are arranged as 10–20 interlacing, concentric lamellae, which are firmly attached to the adjacent vertebral bodies. The compressive forces applied to the disc are directly borne by the NP and are distributed as a tensile force to the annular collagen. The incompressibility exhibited by a normal NP is due to its high water content, which in turn is maintained by the hydrostatic pressure generated by proteoglycans. The normal NP proteoglycan content is a function of the delicate balance between anabolic and catabolic enzymatic activities.
The vascularity of a normal intervertebral disc is limited to the outer one-third of the AF. In addition, the disc is separated from the vascular vertebral body by avascular cartilaginous end plates. Consequently the metabolic needs of the NP and inner AF are met almost entirely by diffusion from the capillary plexuses in the adjacent vertebral bodies and outer AF. This process is facilitated by circadian changes in intradiscal pressure; lower nighttime pressure facilitates the flow of fluids into the disc, while higher daytime pressure forces the fluids out of the disc. The end products of the NP cellular metabolic activities are also removed by diffusion. However, the disc lacks scavenger cells, so that degradative products tend to accumulate over time, which can interfere with normal homeostatic functions.
The presence of nerve fibers in the normal disc is predominantly limited to the outer one-third of the AF. Disc innervation is mostly in the form of mechanoreceptors, which originate from plexuses along the anterior and posterior longitudinal ligaments. The posterior plexus receives its input from the sinuvertebral nerve and gray rami communicans, while the anterior plexus receives contributions mainly from gray rami communicans. These rich autonomic connections may contribute to the vague, poorly localized pain characteristic of IDD.
As discs begin to lose their water content and degenerate, nerve ingrowth occurs, which increases the density of nociceptive nerve fibers and their penetration toward the NP. The mechanisms by which this occurs may include angiogenesis and alterations in the cellular matrix (e.g., influx of inflammatory cytokines) and function. Whether or not axial pain arising from degenerated intervertebral discs could be a form of neuropathic pain is a subject of controversy, but one study suggests that 12% of patients with axial low back pain (LBP) had neuropathic components, with higher numbers reported in those patients who had previously undergone disc surgery.
DDD has been associated with both genetic and acquired factors, such as vascular disease, smoking, lifestyle, and obesity. It is most likely a consequence of a combination of factors, including a decline in the number and function of viable disc cells, enhanced matrix metalloproteinase (MMP) activity, and increased activity of discal cytokines and other proinflammatory mediators. These metabolic derangements can result in a reduction of nuclear proteoglycans and loss of discal water content. The diminished NP hydrostatic pressure leads to increased NP compressibility, which exposes the AF to direct compressive forces. In addition to mechanical stress, the AF also undergoes degenerative changes similar to those of the NP. In young patients with nondegenerated spines, most of the load bearing is borne anteriorly by the intervertebral discs; but as degeneration occurs, the posterior neural arch assumes an increased burden. Collectively these changes result in the loss of annular collagen, mechanical failure, and the development of annular fissures that spread outward toward the periphery.
Annular fissures are a hallmark of DP. These tears are zones of highly vascularized and richly innervated granulation tissue. On T2-weighted MRI these may be seen as “high-intensity zones.” The two different types of nerve fibers found in these granulation zones are vasoregulatory nerves, which accompany neovascularization, and free nerve endings high in substance-P concentration. In addition, annular tears are abundant in mononuclear cell infiltrates, which release nerve growth factors that contribute to nerve ingrowth and accelerated degeneration. Disrupted discs also contain high concentrations of proinflammatory mediators, which serve to sensitize nerve endings and maintain a state of hyperalgesia. This state has been linked to the painful response associated with minimal pressure elevation during discography, a term denoted as “chemically sensitized.” Owing to limited repair capacity, a painful disrupted disc may remain a long-standing source of disability.
In the long term, changes in disc morphology may alter spinal mechanics, increase stress on adjacent spinal structures, and result in osteophyte formation, sclerosis, and autofusion. This can, in turn, lead to further degeneration of the disc and vertebral end plate, sacroiliac and facet joint pathology, and spinal stenosis ( Fig. 69.1 ).
Prevalence
Epidemiologic studies evaluating the incidence of spinal pain vary greatly, as the conditions producing back and neck pain are often poorly defined. This is especially true for DP. The lifetime prevalence rate for LBP varies between 50% and 80%. For neck pain, the annual prevalence rate generally varies between 15% and 50%, with one systematic review reporting a mean prevalence rate of 37.2%.
Epidemiologic studies for DP are uncommon. In an oft-cited study by Schwarzer et al. conducted in 92 patients with chronic, nonradicular LBP and no previous surgery, the authors reported a 39% prevalence of IDD using exact pain reproduction, abnormal computed tomography (CT)–discography imaging, and a negative adjacent control disc as the criteria. In a large-scale study performed in 127 patients with axial LBP who failed facet interventions, Cohen et al. reported a prevalence rate of 65%. More recently, a systematic review that included three prevalence studies determined that the prevalence of DP, as assessed by radiographic abnormalities and concordant pain provocation, varied between 39% and 42%.
Studies conducted in the cervical spine tend to yield widely disparate results. In a prospective observational study evaluating 173 cervical discograms, Grubb and Kelly reported at least one positive level in 86% of patients. In a smaller ( n = 31) retrospective study, Connor and Darden found that 84% of patients experienced provocative concordant symptomatology; these were considered positive. Neither study required a control disc. Another retrospective study performed in 143 individuals with chronic neck pain reported the prevalence rate of DP to be 16%. A systematic review published in 2013 evaluating the accuracy and utility of cervical discography found prevalence rates ranging between 16% and 53%.
The Controversy Surrounding Discography
Rationale
The rationale for discography is based on three factors: the high prevalence of spine pain, the high prevalence rate of abnormal MRI findings at asymptomatic levels, and the low success rate of surgical interventions for degenerative spondylosis. The lifetime prevalence of serious LBP episodes ranges from 50% to 80% ; for neck pain, the annual prevalence rates range from 16% to as high as 50%. Confounding matters is that MRI studies conducted in asymptomatic volunteers have consistently demonstrated that a majority of people have abnormalities in the lumbar, thoracic, and cervical spine regions, with the proportion increasing with age.
Having an inexpensive, safe, and reliably effective bridge to natural recovery is therefore paramount, yet none exists. Based on systematic reviews, it is clear that surgery performed for axial spine pain is associated with a high rate of failure and significant complications and that most patients recover without procedural interventions. The high prevalence rates for spine pain and coincidental imaging abnormalities, coupled with the absence of any reliable interventional treatment for IDD, augur in favor of an accurate means of correlating symptoms with imaging results.
False-Positive and False-Negative Results
The principal criticism surrounding provocation discography is the high rate of false-positive (FP) results. The first study to quantitatively question the validity of lumbar discography was performed by Holt over 40 years ago ; he reported an FP rate of 37% in 30 asymptomatic prisoners. Over 20 years later, Walsh et al. performed CT-discography in 10 asymptomatic male volunteers and 7 “control” patients with chronic LBP. In the asymptomatic subjects, CT-discograms were interpreted as abnormal in 17% of the 35 discs injected and half of the 10 subjects. However, none of these patients experienced concordant pain associated with pain-related behavior during the injections.
The bulk of the work on FP discography was done by Carragee and colleagues in the late 1990s and early 2000s. In general, these studies have found high FP rates for lumbar discography in patients who had undergone previous iliac crest bone grafting, those with failed back and neck surgery, individuals with psychopathology (especially somatization disorder), and those with secondary gain issues.
The Carragee studies have been criticized on several accounts. The first is that modern provocation discography requires that the evoked pain must be concordant with a patient’s baseline pain, which is not possible in asymptomatic subjects. Another flaw is that pressure readings were not a determining factor in the designation of a positive disc.
To control for some of these factors, Derby et al. performed 43 discograms in 13 volunteers with either no history or infrequent episodes of LBP. In the subjects with occasional back pain, 35% of injected discs were painful, versus 52% in volunteers without LBP. Most discs required high pressures before pain was provoked. No relationship was noted between painful disc injections and radiologic or discographic abnormalities. Controlling for the intensity of response and the pressures at which pain was elicited, the authors concluded that the incidence of FP discograms was less than 10%.
Wolfer et al. performed a systematic review reanalyzing data from five previous studies including those of Carragee et al. based on the criteria of the International Association for the Study of Pain (IASP) and International Spinal Intervention Society (ISIS) for a positive lumbar discogram. The authors found an overall FP rate of 9.3% per patient and 6.0% per disc. In patients without back pain or confounding factors, the FP rates declined to 3.0% per patient and 2.1% per disc. Patients with chronic pain were found to have FP rates per patient and disc of 5.6% and 3.9%, respectively. The highest FP rates per patient and disc were for postdiscectomy patients (15% and 9.1%, respectively) and those with somatization disorder (50% and 22.2%, respectively).
Fewer studies have examined the incidence of FP discograms in the cervical and thoracic regions. In a study by Schellhas et al., none of 40 cervical discograms done in 10 asymptomatic volunteers elicited reported pain or facial expressions indicative of pain. In a later study done in the thoracic region, the same group of investigators reached slightly different conclusions ; that is, 3 of 40 discs injected in 10 asymptomatic volunteers provoked intense (≥7/10) pain, with 2 positive responses occurring in one subject. One possible cause of inaccurate or FP discography results is that pressurization of a disc during fluid injection results in increased pressure in adjacent discs, which may undermine the specificity.
The issue of “false-negative” discograms has received far less attention but can lead to inaccurate diagnoses, unnecessary interventions, and the withholding of beneficial treatment from otherwise good candidates. There are several reasons for this phenomenon, 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 contrast extravasation in severely degenerated discs. The failure to elicit pain in a degenerated, ostensibly painful disc may be more likely to occur in elderly patients. In a review by Cohen et al., the authors estimated that between 15% and 25% of degenerated discs fail to elicit concordant pain provocation during stimulation. The proportion of these occurrences that represent false-negative versus true negative responses is a question that remains to be answered.
In summary, FP discograms can occur in all regions of the spine but are infrequent in unoperated individuals with no confounding factors. A careful consideration of the risks and benefits of discography should be done when discography is being considered for individuals at high risk for FP results since many of these factors are also associated with treatment failure. If discography is conducted in high-risk individuals, one should consider obtaining two adjacent control discs and correlating reported pain with more objective measures such as an increase in heart rate and/or change in facial expression. Other factors that may increase the risk of FP discograms include extreme anxiety, performing disc stimulation before allowing previously provoked pain to return to baseline, inadvertent annular injection, contrast-induced irritation of nervous tissue, end-plate deflection resulting from suboptimal needle placement, and rapid or excessive pressurization ( Table 69.1 ).
| Study, Year | Region | Subjects | Criteria | Results |
|---|---|---|---|---|
| Holt, 1964 | Cervical | 50 male volunteer inmates, 148 discs | Pain provocation + contrast extravasation | All injections provoked severe pain; contrast extravasation noted in all pts, 93% of discs |
| Massie and Stevens, 1967 | Lumbar | 52 male subjects, 156 discs | NR | FP rate not reported but stated “injection only occasionally produced symptoms” |
| Holt, 1968 | Lumbar | 30 male volunteer inmates, 70 discs (20 failed injections) | Pain provocation | 60% FP rate per subject, 37% per disc |
| Walsh, 1990 | Lumbar | 10 male volunteers, 30 discs | 3/5 pain provocation + 2/5 pain-related behaviors | 0% FP rate per subject and disc |
| Schellhas, 1996 | Cervical | 10 volunteers, 40 discs | 7/10 pain provocation + facial expressions | 0% FP rate per subject and disc |
| Wood, 1999 | Thoracic | 10 volunteers, 40 discs | 7/10 pain provocation + facial expressions | 20% FP rate per subject, 7.5% per disc |
| Carragee, 1999 | Lumbar | 8 males who had undergone recent iliac crest bone grafting for problems unrelated to low back pain, 24 discs | 3/5 “concordant” pain provocation (to previous iliac crest pain), + 2/5 pain-related behaviors | 50% FP rate per subject, 38% per disc |
| Carragee, 2000 | Lumbar | 6 subjects with somatization disorder, 10 with failed neck surgery, and 10 control pts with no pain after successful cervical spine surgery; 78 discs | 3/5 “concordant” pain provocation (to previous iliac crest pain), + 2/5 pain-related behaviors | FP rate per subject: 83% for somatization, 40% for failed neck surgery, and 10% for “control” group; FP rate per disc: 33% for somatization, 23% for failed neck surgery, and 3% for control group |
| Carragee, 2000 | Lumbar | 47 subjects who underwent a single-level discectomy. 20 subjects were “symptom-free,” while 27 pts continued to have back and/or leg pain; 138 discs | 3/5 pain provocation + 2/5 pain-related behaviors | FP rate per subject: 40% for asymptomatic subjects and 56% for pts with failed back surgery; FP rate per disc: 15% in asymptomatic group |
| Derby, 2005 | Lumbar | 13 volunteers, 43 discs | Criteria not noted; used 0–10 pain rating and 0–4 pain behavior scales along with manometry | Using 6/10 as criterion for a (+) disc, 0% FP rate; using 4/10 pain at ≤ 50 psi, FP rate 23% per subject and 9% per disc |
Correlation Between Magnetic Resonance Imaging And Discography
Several attempts have been made to correlate imaging with discography results. In one of the earliest studies comparing MRI, the most sensitive test for disc pathology, with lumbar discographic findings, Gibson et al. found agreement in 88% of 50 discograms. Among the six discs in which a discrepancy was observed, evidence of IDD was missed in five discograms and one MRI. Correlation in this study was based solely on radiographic findings and not provocation results. Collins et al. reported similar results. The authors found that discographic and MRI imaging characteristics correlated in 89% of 73 lumbar discograms. Of the eight discordant discs, four revealed early evidence of disc degeneration on discography but were normal on MRI and four were discographically normal but demonstrated mild degeneration on MRI. All discs that provoked concordant symptoms were degenerate on both discography and MRI. In a study by Schneiderman et al., the correlation between MRI and discographic morphology was 99%, with the only discrepancy being noted in a 13-year-old.
However, the more relevant question is whether provocation results can be predicted by radiologic imaging. Yoshida et al. investigated the relationship between provocation discography and MRI in 56 discograms from 23 patients. The authors found the sensitivity, specificity, positive predictive value, and negative predictive value of T2-weighted, gadolinium-enhanced studies in detecting symptomatic discs to be 94%, 71%, 59%, and 97%, respectively. These findings compared favorably with T1-weighted images. In another study, Aprill and Bogduk found that the presence or absence of a high-intensity zone in 118 discograms had a sensitivity, specificity, and positive predictive value for concordant pain provocation in 97%, 63%, and 95% of levels, respectively. Even stronger correlations have been found by others.
Not all studies have demonstrated positive results. Zucherman et al. reported a case series of 18 patients with normal MRI and positive discography. In a retrospective study, Sandhu et al. found a poor correlation between changes in vertebral end-plate signal observed on MRI and the results of provocation discography. In an observational study conducted in 25 patients, Horton and Daftari concluded that discrepancies between findings on MRI and discography necessitated that both be used in surgical planning. Finally, in a small study ( n = 26) that correlated the results of provocation discography, anesthetic discography, and MRI, Putzier et al. found no correlation between the three parameters and the presence of high-intensity zones, or Dallas and Pfirmann scores, and only a weak correlation between anesthetic discography and Modic changes.
To date, there have been few correlative observational studies performed in the cervical spine. In a study performed in 52 patients (104 discs), Parfenchuk and Janssen found that the sensitivity, specificity, FP and false-negative rates between MRI and pain provocation to be 73%, 67%, 33%, and 27%, respectively. Schellhas et al. later sought to correlate MRI with disc provocation results in 10 asymptomatic volunteers and 10 patients with chronic neck pain. In the asymptomatic cohort, half of the 40 discs were morphologically abnormal on MRI versus 88% that exhibited abnormalities on discography. However, none of the abnormal discs provoked concordant pain during stimulation. In the symptomatic patients, 29 of the 40 discs exhibited some degree of abnormality on MRI. Among the 11 normal discs, 10 were found to have annular tears discographically, with 8 of these shown to be painful when injected. In summary, whereas a significant correlation between concordant pain provocation and MRI findings has been demonstrated, the high FP and false-negative rates suggest the need for a reliable means to ascertain which abnormalities are pain generators.
Effect of Surgical Outcomes
Spinal Arthrodesis
Although uncontrolled studies evaluating the impact of preoperative discography on surgical outcomes have been mixed, the largest, most recent, and best-designed study suggests that discography may improve arthrodesis when used to screen patients before arthrodesis.
In one of the earliest studies to examine this question, Colhoun et al. found a strong association between disc stimulation findings and fusion results. Among the 137 patients with nonradicular LBP in whom disc stimulation provoked concordant pain, 89% had a favorable outcome at the mean follow-up period of 3.6 years. In contrast, only 52% of the 25 patients whose morphologically abnormal discs failed to elicit symptoms experienced significant benefit. Whereas some investigators cite this as evidence supporting provocation discography as a surgical screening tool, the absence of a comparator group that received MRI and the relatively high success rate in the negative provocation group limit the generalizability of the findings.
Later, uncontrolled studies failed to replicate even these equivocal results. Esses et al. retrospectively examined the influence of discography in predicting external fixation and fusion results in 32 patients with refractory LBP, finding that neither concordant pain provocation nor morphologic abnormalities predicted pain relief. The main flaw in this study is that it was not designed to assess the influence of discography on surgical outcomes. The next attempt to correlate discographic findings with spinal fusion outcomes was by Madan et al. who performed a retrospective analysis in 73 patients with chronic LBP. At the minimum 2-year follow-up, no difference in any outcome measure was noted between the two matched groups. In another retrospective study evaluating the value of discography as a preoperative screening tool, Derby et al. found that patients with chemically sensitized discs experienced better outcomes following interbody/combined fusion than after other treatments.
A recent randomized controlled trial supports the use of discography as a surgical screening tool. Margetic et al. randomized 310 patients with chronic LBP being evaluated for surgery in a 2:1 ratio to receive provocation discography with fusion done only in those with a positive test or fusion performed without discographic screening. In the discography group, 158 subjects had a positive discogram and negative screening for depression and somatization; they proceeded to surgery. At 1-year follow-up, the change in Oswestry disability index scores in the discography group was 29.7%, which compared favorably with the 24.6% improvement in the control group ( P = .12). When only those with degenerative disc disease were considered ( n = 127), the difference favoring the treatment group became statistically significant (35% vs. 22% improvement in function; P < .001).
In the cervical spine, only one study has evaluated the predictive value of discography in selecting surgical candidates. Kikuchi et al. performed a retrospective study in the pre-MRI era evaluating surgical outcomes on 138 patients with either mechanical ( n = 41) or radicular ( n = 97) neck pain who underwent anterior discectomy and fusion based on discography results. One year postprocedure, 80% were either pain-free or experienced only mild discomfort that did not interfere with work. In a control group who underwent cervical fusion without the benefit of discography, 60% had favorable outcomes. Similar to the findings of Colhoun et al., the lack of MR imaging and the high success rate in the control cohort limit the relevance of these findings.
Determining Operative Levels
Two studies have examined the effect of discography in identifying treatment levels in patients already selected for spinal fusion. In a prospective study conducted in 193 patients with neck pain and neurologic symptoms, Hubach evaluated the use of intraoperative discography to select operative levels. In the initial group of patients ( n = 23) who were fused without discography, 35% developed juxtafusional pain at long-term follow-up. In the ensuing 156 patients in whom the operative levels were based on discography, only 12% developed pain at an adjacent segment. Yet in a later prospective study conducted in the lumbar spine, Willems et al. failed to demonstrate a benefit for discography in determining fusion levels. In summary, the evidence favors the use of discography as a screening tool to identify candidates for lumbar spinal fusion, but results are equivocal concerning the cervical spine and regarding whether discography should be used to identify the levels of arthrodesis ( Table 69.2 ).
