INTRODUCTION

Lumbar provocation discography is a commonly used diagnostic procedure utilized to determine the presence or absence of discogenic pain at a specific spinal segment. Although discography has been proposed as the criterion standard for identifying discogenic pain [NASS statement], the test is controversial. Skeptics argue that newer diagnostic tests such as MRI scans make discography obsolete and the test should be discontinued unless its utility can be validated.¹

This chapter will discuss the clinical relevance of lumbar provocation discography by correlating discography findings with the standard clinical work-up, and by its ability to help guide treatment. Additionally, we will describe controversies in technique and interpretation.

CORRELATION OF LUMBAR PROVOCATION DISCOGRAPHY WITH HISTORY AND PHYSICAL EXAMINATION FINDINGS

Various studies have attempted to correlate history and physical examination findings with the diagnosis of discogenic pain by discography and to compare these findings with respect to treatment outcome (Fig. 94.1). Simmons and Segil² reported a diagnostic accuracy of 44% for clinical examination and 82% for discography based on ability to predict symptomatic level as confirmed by successful clinical result after surgery. They did not specify any particular tests used preoperatively, but used the following in follow-up examinations: straight leg raise, neurologic examination including reflexes, sensory or motor, range of motion, and tenderness. Although they correlated discogram results with these physical examination techniques, no other outcome instruments were correlated.

Fig. 94.1 Pain drawings can be an important clinical tool in the prescreening evaluation. By having a patient diagram the location of their pain, a more streamlined approach to the patient can be pursued. (A) is an ideal discography candidate with axial back pain. (B and C) could potentially be screened out since patients with major components of radicular and whole body pain are suboptimal discography candidates.

Schwarzer et al.³ found that no historical or clinical examination finding that they studied in 92 patients could accurately identify patients with internal disc disruption as diagnosed by provocation discography. Historical findings sought were pain increased or was relieved by sitting, standing, or walking. Pain referral patterns studied were pain into the buttock, groin, thigh, calf, or foot and whether the pain was unilateral, bilateral, or midline. Physical examination findings performed included provocation of pain with forward flexion, extension, rotation, combined rotation with extension, or straight leg raising making either back or leg pain worse.³ There was a trend (lowest p values obtained) with historical finding of pain increased with sitting (p=0.13), pain increased with standing (p=0.13), but also pain relieved with sitting (p=0.16) and physical examination finding of pain increased with forward flexion (p = 0.16). There was a negative trend (highest p values obtained) with a historical finding of pain increased with walking (p=0.89) or physical examination finding of pain increased with extension.

Young et al.⁴ prospectively examined 81 patients with clinical examination and various diagnostic injections and found a weak but statistically significant correlation of discogenic pain as diagnosed by discography with centralization of pain with repeated end-range movements (p=0.025, Phi=0.5). Although localization of a specific symptomatic level was not studied, 47% of those with positive discograms had ‘retreat of referred symptoms from the periphery toward the midline of the spine’ (centralization) during the standard McKenzie evaluation. Centralization was not seen with zygapophyseal joint pain as diagnosed by single intra-articular injection. Furthermore, all patients with a positive discogram (15 of 24 total discograms) reported pain when rising from sitting. A positive correlation (p=0.02) was, however, also noted in patients with sacroiliac joint pain.

Donelson et al.⁵ also prospectively studied 63 patients with chronic low back pain to evaluate the ability of the McKenzie mechanical lumbar assessment to diagnose discogenic pain and assess annular competence as determined by provocation discography. Seventy-four percent of centralizers and 69% of peripheralizers had a positive discogram as defined as exact pain reproduction accompanied by an abnormal image (nucleogram/CT), provided no pain was reproduced at an adjacent control level. The disc was interpreted as having a complete annular disruption or noncontained pathology if there was poor resistance to injection and contrast spread through the anulus to the epidural/perineural or peridiscal space. The disc was interpreted as having an intact outer anulus or contained pathology if there was firm resistance to injection even if contrast leaked from the disc at peak injection pressure. Of the centralizers, 91% had a competent anulus. Of the peripheralizers, 54% had a competent anulus. Of the patients whose symptoms did not change with repeated end-range movement, only 12.5% had a positive discogram. These differences were significant. The localization of a specific symptomatic level was not addressed and, therefore, these maneuvers could only be used for screening but not for identifying specific symptomatic disc levels for targeted interventional treatment.

A ‘bony vibration stimulation test’ or ‘vibration pain provocation’ was described by Yrjama and Vanharanta and the results of bony vibration were compared to the results of provocation discography.⁶ The studies assumed provocation discography as the reference standard and compared the bony vibration stimulation test by itself or in combination with either ultrasound or MRI.

The bony vibration stimulation test described utilized a Braun 3-D electric toothbrush as a modified vibrator device with a frequency of vibration of 42–50 Hz. The patient was laid on the more symptomatic side. The lumbar spinous processes were sequentially compressed with the vibrator’s head in either the on or off position. If there was more pain with vibration than without vibration and this pain was described as concordant, the test was classified as ‘painful.’

Thirty-eight patients were studied with the bony vibration stimulation test, ultrasound, and provocation discography.⁷ Discs were graded by ultrasonographic findings. Grade 0 was a normal disc. Grade 1 discs showed a hyperechoic lesion in the inner anulus. Grade 2 discs demonstrated a hyperechoic lesion in the outer anulus. Grade 3 discs showed a hyperechoic area extending outside the disc. When used alone, the bony vibration stimulation test yielded a sensitivity of 65% and a specificity of 58%. In the patients with a grade 1 or grade 2 disc, the sensitivity was 90% and specificity 75%. In the patients who had a grade 3 disc and pain on bony vibration the sensitivity and specificity were 50%.

The authors also studied 33 patients with low back pain, correlating the results of bony vibration stimulation test and MRI with the results of provocation discography.⁸ When used alone, the bony vibration stimulation test yielded a sensitivity of 63% and specificity of 44%. If patients with history of previous lumbar surgery were excluded, the sensitivity was 61% and specificity 67%. In patients with or without previous history of surgery who had MRI findings of ‘partial annular rupture’ (as defined as irregular or absent intranuclear cleft images or bright-signal nuclear material into the outer anulus on T2-weighting), the sensitivity was 88% and specificity 50%. If the patients with previous surgery were excluded from this group, the sensitivity was 88% and the specificity 75%. The only false-positive finding in this group was a patient who was ‘hypersensitive’ and felt pain at all levels tested. If a ‘total annular rupture’ (defined as T2-weighted or proton density images showing discontinuity of the low-signal band representing the outer rim of the anulus) was seen, the sensitivity decreased to 47% and specificity to 50%.

These studies were, however, limited by their small sample sizes. Provocation discography was regarded as the reference standard for determining a symptomatic level without comparison to postprocedural outcome. Additionally, the findings of both bony vibration stimulation test and provocation discography were not recorded according to a specific spinal level. Finally, the protocol for discographic examination and interpretation was not fully elaborated.

UTILITY OF POSTDISCOGRAM COMPUTED TOMOGRAPHY

Discography was originally used as an adjunctive to myleography to visualize lateral herniations (Fig. 94.2). There are, however, fundamental limitations to using discograms exclusively as an imaging tool. The classic nucleogram patterns viewed with anteroposterior (AP) and lateral radiographs have been described as both normal and degenerative. Quinnell is credited with the original description of interpreting radiographic images of discography although he never proposed a classification scheme.⁹ He discussed the importance of monitoring contrast flow and volume to help improve the interpretation of the nucleograms.

Fig. 94.2 Discography showing a far lateral herniation.

In 1986, Adams et al. initially classified nucleogram patterns into five categories; cottonball, lobular, irregular, fissured, and ruptured (Fig. 94.3).¹⁰ Cottonball and lobular patterns are thought to be normal variations, while fissured and ruptured are pathological. An irregular pattern is considered intermediate between these normal and pathological presentations. Adams et al. hypothesized the pathological patterns were due to disc degeneration. Cottonball nuclear patterns show the contrast central to the disc with an ovoid appearance (Fig. 94.3A). Lobular nuclear patterns also show the contrast centralized within the disc; however, there are two distinct arcs that may or may not be contiguous (Fig. 94.3B). An irregular pattern shows some tracking of the contrast outside the central nucleus without extension to the outer anulus (Fig. 94.3C). Although the contrast is intranuclear with irregular nucleograms, the small crevices and clefts exhibit early evidence of degeneration. Fissured nucleograms extend to the posterior annular margin, while ruptured nucleograms demonstrate complete radial tears and show contrast spread into the epidural space (Figs 94.3D, 94.3E).¹⁰ Adams speculated that by viewing the first four types of nucleograms, a natural progression of disc degeneration can be seen.

Fig. 94.3 Illustrated examples of Adams’s nucleogram classifications.

Postdiscogram CT scans provided a more comprehensive view of the disrupted anulus (Figs 94.4–94.6). Axial views allowed a more detailed view of pathological contrast patterns than the AP and lateral radiographs (see Figs. 94.3C, 94.3D, 94.3E). The disorganized patterns seen on radiographs now coalesced into structured images, demonstrating organized annular tears with circumferential spreading.

Fig. 94.4 Postdiscography CT with left-sided inner annular injection: (A) axial and (B) coronal views.

Fig. 94.5 Illustrated is the annular description for the radial tear findings and circumferential degenerative findings for the original Dallas discogram scale described by BL Sachs et al. In (A), the radial extension of contrast moving away from the inner nucleus is shown. Grading of 1–3 is based upon the radial spread of the contrast towards the anulus (A–C). (D–F) The percentage quadrant distribution of circumferential spread displays the amount of degeneration within each disc. In addition to radial extension, grade 1–3 is also used to describe circumferential extension (D–F). Addition modifiers are meant to be employed to specify the areas of annular leakage as anterior, posterior, lateral, posterolateral, anterior, and posterior.

Fig. 94.6 Modified Dallas discogram scale. (A) The postdiscogram CT axial view demonstrates a contained normal centralized pattern without annular radial extension or circumferential degenerative changes. (B) The postdiscogram CT axial view demonstrates grade 1 of 5 with extension within the inner third of the anulus. Note: There is a needle tract seen on the left-posterior lateral aspect of the disc. (C) The postdiscogram CT axial view demonstrates a grade 2 of 5 tear with extension into the outer third of the anulus. (D) The postdiscogram CT axial view demonstrates a grade 3 of 5 tear with radial extension of contrast into the outer third of the anulus, with a limitation of circumferential spread less than 30 degrees. (E) The postdiscogram CT axial view demonstrates a grade 4 of 5 annular tear radial extension of contrast into the outer third of the anulus with greater than 30 degrees of circumferential spread into the outer anulus. (F) The postdiscogram CT axial view demonstrates a grade 5 of 5 full-thickness tear (focal in this example, although it may be circumferential) with extension of contrast freely into the epidural space.

In 1989, Thomas Bernard published a case series of 250 patients who underwent both discography and postdiscography CT.¹¹ He showed that computed tomography scanning after discography was not only able to define the type of herniation and disc architecture (protrusion, extrusion, sequestration, or internal disc disruption), but it could also be used to rationalize false-positive levels in the setting of non-nuclear injections (annular injections) (Fig. 94.4).

In 1987, Sachs et al. organized the previously inexactly described axial contrast patterns into the Dallas discogram scale (Fig. 94.5).¹² Sachs’s original description was based on the appearance of the anulus. This allowed objective categorization of the anulus, classifying both degeneration and annular disruption separately. Sachs’s grading of the Dallas discogram scale is based on a four-point scale of 0 to 3, with zero defined as normal.

Annular disruption defined as ‘leaking/protrusion/annular fissuring’ is based on the radial spread of the contrast away from the center of the nucleus to the periphery (Fig. 94.5A–C). Grade 1 is defined as spread within the inner third of the anulus; grade 2 is into the outer third of the anulus, while a grade 3 is defined as moving beyond the outer anulus.¹²

Degeneration findings were based on the circumferential distribution of contrast contained in quadrants of the anulus (Fig. 94.5D–94.5F). Grade 1 was defined as local spread in less than 10% of the anulus. Grade 2 was defined as partial spread or less than 50%, while grade 3 was defined as greater than 50% spread of contrast across the anulus.

The development of the Dallas discogram scale not only allowed a new perspective to classify annular degeneration and disruption, but it also increased the inter-rater reliability when interpreting these findings. Sachs et al. were able to show 91% reproducibility and 88% repeatability when using the Pearson correlation test.¹² Furthermore, they showed how standard AP and lateral discogram radiographs may appear normal until viewed axially. This reproducibility helped validate the Dallas discogram scale.

Aprill and Bogduk soon added a grade 4 to the disruption scale.¹³ Approximately 10 years after its original description, Schellhas et al. expanded upon the three-grade disruption classification system to propose the modified Dallas discogram scale. The modified Dallas discogram scale incorporates both aspects of degeneration described as circumferential involvement as well as annular disruption depicted as radial contrast extension. The definitions of grades 0–2 remained the same. Grade 3 was expanded slightly to include either focal or radial extension of contrast into the outer third of the anulus, with a limitation of circumferential spread less than 30 degrees. Grade 4 was defined the same as grade 3, but with greater than 30 degrees of circumferential spread into the outer anulus. Grade 5 is a full-thickness tear, either focal or circumferential, with extension of contrast outside the anulus (Fig. 94.6).¹⁴

CORRELATION OF LUMBAR PROVOCATION DISCOGRAPHY WITH LUMBAR MRI

Lumbar MRI is a sensitive means to investigate anatomic abnormalities of the low back. However, even asymptomatic subjects have been noted to have significant spinal pathology on imaging, including disc protrusion or extrusion with or without neural compromise.^15–17 Therefore, although MRI demonstrates pathology, it does not necessarily reveal whether the abnormality is causing a patient’s symptoms. The correlation between MRI findings and provocative discography results has been investigated. The common goal of these studies was to determine whether imaging findings could predict the presence of concordant discogenic pain. If imaging studies could reliably predict discography findings, discography was unnecessary. The accuracy of MRI findings such as the high-intensity zone (HIZ) and endplate changes in determining symptomatic levels have been compared to the results of discography.¹⁸

High-intensity zone

The high-intensity zone in the lumbar spine was first described by Aprill and Bogduk (Fig. 94.7)¹³ as a high-intensity zone (HIZ) seen on T2-weighted, sagittal images and defined by the authors as a high-intensity signal (bright white) located in the substance of the posterior anulus fibrosus. The HIZ must be clearly dissociated from the signal of the nucleus pulposus because it is surrounded superiorly, inferiorly, posteriorly, and anteriorly by the low-intensity (black) signal of the anulus fibrosus and because the HIZ has an appreciably brighter signal than the nucleus pulposus.

Fig. 94.7 T2-weighted image shows a high-intensity zone (HIZ) at L4–5. As originally described by Aprill and Bogduk in the British Journal of Radiology in 1992, an HIZ is an area of brightness or high signal intensity located in the posterior anulus fibrosus, distinctly separate from the nucleus pulposus and brighter than the nucleus on T2-weighted images.

Aprill and Bogduk investigated a subset of 41 patients who had both MRI and discography with postdiscography CT scan. The HIZ was identified in 28% of patients with low back pain sent for lumbar MRI. All discs with HIZ had abnormal imaging on postdiscography CT with either a modified Dallas discogram scale grade 3 or grade 4 annular tear, specifically a radial tear extending into the outer third of the anulus without (grade 3) or with (grade 4) circumferential contrast encompassing an arc of greater than 30 degrees. When compared to reproduction of pain by disc stimulation, the HIZ had a high correlation with exact or similar pain with a sensitivity of 63%, specificity of 97%, and positive predictive value of 95%. For exact pain reproduction, the sensitivity was 82% and specificity 89%.

Schellhas et al.¹⁴ sought to reproduce Aprill and Bogduk’s findings with a similar study. In their retrospective analysis of 100 HIZs in 63 symptomatic patients, all discs with HIZ were internally deranged with annular disruption, 87 of 100 HIZ discs were concordantly painful at discography, and all 87 of these concordantly painful HIZ-containing discs had grade 3 to grade 5 annular tears on their modified Dallas discogram scale. Their schema is identical to Aprill and Bogduk’s; however, they further stratified the Dallas discogram and added grade 5 to identify those discs which had a ‘full-thickness tear, either focal or circumferential, with extra-anular leakage of contrast.’ Individual discs could be classified into two categories, such as grade3/grade 5 or grade 4/grade 5.¹⁴

After the symptomatic group was analyzed, 17 asymptomatic volunteers underwent MRI without discography in an attempt to determine the prevalence of HIZ in patients not exhibiting low back pain (LBP). Only 1 of 17 (5.9%) of lifelong asymptomatic volunteers had a lumbar MRI with an HIZ. Those volunteers who never had prior LBP had a younger average age as compared to the symptomatic group (29.6 years old versus 37.5 years old).

Other similar studies correlating HIZ with positive provocation discography have shown poor sensitivity of 26%, high specificity of 90–95.2%, positive predictive values of 40–88.9%, and negative predictive value of 47–83%.¹⁹

The interobserver reliability of HIZ also varies. Aprill and Bogduk found that out of 67 images, the two observers agreed on the presence of HIZ in all but one (1.5%). Though both observers noted the abnormality, the two disagreed on its meeting the criterion of brightness due to poor-quality film. However, Smith et al.¹⁹ only found fair to good interobserver reliability with a kappa value of 0.57 with 95% confidence interval 0.44–0.70.

The original description by Aprill and Bogduk¹³ only included HIZ located in the posterior anulus and had a prevalence of 28%. The clinical significance of the HIZ was further studied by Rankine et al.²⁰ in a patient population without neural compression. When including HIZs in any aspect of the anulus, the prevalence in a specialty spine surgery clinic was 45.5%. Most of these were posterior (77%) followed by posterolateral (22%). The HIZ was associated with moderate disc degeneration as assessed by signal reduction on T2-weighted sagittal images. There was no correlation of presence of HIZ with clinical features such as age, duration of symptoms, Oswestry score, or Schober’s extension–flexion range of motion testing. There was also no correlation with patient history of employment, pain above or below the knee, or the positions or activities worsening pain. No physical examination findings studied predicted the presence of HIZ. The examination findings tested were paraspinal muscle spasm, spinal tenderness, straight leg raise, neurologic testing with reflexes, myotomal and dermatomal testing, and Waddell’s testing. Correlation of HIZ presence with positive provocative maneuvers producing concordant axial pain with forward bending or extension/quadrant loading was not measured.

LUMBAR PROVOCATION DISCOGRAPHY AND ITS PREDICTIVE VALUE FOR TREATMENT AND PROGNOSIS

Discography is a presurgical technique used to identify concordantly painful disc(s) and to verify that adjacent discs are negative for concordant pain prior to a therapeutic interventional procedure. Together with imaging studies, the provocation of concordant back pain during disc pressurization is used to confirm one’s clinical impression that a specific disc is a source of pain. The procedure is usually performed in patients with predominately axial pain who have failed conservative management and are considering more invasive treatments such as percutaneous intradiscal or open discal procedures based on the results of discography.

Because the results of discography are used to justify interventional treatments, the results should have predictive value for treatment success or prognosis. Although several studies compare the results of discography to decision formulation and outcome, most of the studies have methodological shortcomings such as the use of nonvalidated and/or subjective outcome scales. Additionally, many early studies were done before the routine use of MRI and CT and used discography to identify a disc herniation while today discography is used to ‘prove’ a disrupted disc is the source of the patient’s axial pain. The studies which correlate discography with clinical outcome after procedures assume that (1) the treatment chosen is appropriate for the given lesion, and (2) the procedure accomplished its intended goal (i.e. successful fusion, adequate decompression, etc.). This may have valid grounds and gives valuable information if there is a strong correlation of a discographically proven lesion with successful clinical outcome. If poor results are obtained from the surgical procedure, either the test (discography) or the treatment (specific procedure) is suspect.

In 1975, Simmons and Segil retrospectively reported the value of discography for the cervical, thoracic, and lumbar spine by accurately localizing a symptomatic spinal level based on postoperative results.² Lumbar discograms were performed at 995 levels in 393 patients. Discogram results were assessed by patient pain response, amount and resistance to injection, and discographic appearance by Collis criteria, which describes discs as normal, degenerate, or protruded. Various clinical examinations were compared to the results of surgery on a scale with poor, fair, good, or excellent categories based on patient rating, complaints, occupation, activities, examination findings, and radiographic findings. It was assumed that if the patient was relieved of preoperative symptoms or was significantly improved after surgery, that the level of surgery selected by a certain test was accurate. Specific diagnoses and numbers for each were not reported but lumbar surgeries performed were ‘discotomy,’ ‘discotomy and fusion’ (usually posterolateral and intertransverse), and fusion alone without laminectomy. Overall, 94% of these lumbar surgery patients had satisfactory (fair to excellent) results. Diagnostic accuracy for these surgical outcomes was 44% for clinical examination, 71.5% for routine radiography, 45.6% for myelography, and 82% for discography. Specificity, sensitivity, false-positive, and false-negative rates were not addressed.

In 1979, Brodsky and Binder²³ performed a retrospective study of patients who underwent discography. The authors performed discography only if the myelogram was negative but the patient was clinically thought to have discogenic pain. Specific indications included complaints which were atypical or without localizing neurologic signs, a myelogram which was equivocal or indeterminate, and a myelogram which was positive at one level while symptoms were suggestive of another. Discography was also performed to evaluate discs adjacent to a herniated disc. Decision-making was ‘significantly influenced’ by the discogram in 77.9% of cases. Decision-making would have been the same without discogram in 22.1%. Positive discography was confirmed with surgery in 55.8% of the cases.

In 1988, Calhoun et al. prospectively studied the predictive value of symptom reproduction during provocation discography as a guide to planning spine surgery for symptomatic intervertebral discs in the absence of nerve root compression.²⁴ All 195 patients had lumbar provocation discography at L4–5 and L5–S1, most at L3–4, and some at L2–3 and all underwent lumbar surgery including anterior or posterior fusion and/or laminectomy. With at least 2-year follow-up, surgical success was based on whether the surgical objective was achieved radiographically (successful fusion, etc.) and clinical result was based on complete or significant relief of symptoms, resumption of work and/or normal activities, and no intake of analgesics. Failure to achieve these results was ‘clinical failure.’

There were 137 patients who had a painful injection of a morphologically diseased disc in which 89% had significant clinical relief. There were 25 patients who had technically successful surgery for a morphologically abnormal disc but without symptom production on disc injection, and in these patients only 52% had significant symptomatic relief while most of the remaining 48% had no clinical benefit. Additionally, morphologically abnormal discs that did not cause symptoms during disc injection that were adjacent to symptomatic discs were found in 43 patients. These levels were not included in levels of surgical intervention and significant clinical benefit was 95% (41 of 43) in these patients. Of six patients who were surgically treated despite normal discography, only 50% had clinical benefit. Overall, comparing lumbar provocation discography to clinical outcome after technically successful surgery yielded an 82% accuracy rate, 90% sensitivity, and 9.7% false-negative rate. ‘Accuracy’ was defined by combining the 137 patients who had a positive discogram and did well with surgery (true positive) and the 25 patients who had morphologic abnormal discs but no symptoms on discogram.

Also in 1988, Blumenthal et al.²⁵ used provocation discography to diagnose internal disc disruption (IDD) with the main purpose of the study being to evaluate the efficacy of anterior lumbar fusion as a treatment for IDD. All 34 patients underwent provocation discography and anterior fusion with average follow-up of 29 months. The diagnosis of IDD required radiographic signs of degeneration and discography with concordant pain reproduction with ‘instillation of small amounts of contrast.’ Successful treatment was defined as return to work or normal activities and either no medications or use of NSAIDs only. Fusion success was judged by radiographs only. The successful fusion rate was 73 %, but successful clinical treatment rate was 74%. Four cases were indeterminate for fusion. Of those with healed grafts, 73% had clinical success while only 62.5% of those with nonunion had clinical success. Of those patients with a successful clinical result, 81% had evidence of fusion while only 56% of the clinical failures had evidence of fusion. The authors cite higher previous fusion rates of 91–96%. Wetzel et al. performed a retrospective review of 48 patients, with a minimum follow-up of 2 years, who had lumbar arthrodesis based on provocation discography results.²⁶ All were symptomatic for a mean time of 34.4 months prior to discography. Pain reproduction was graded as concordant or nonconcordant. The discography protocol did not, however, include a negative control disc. Fifty-four percent were single-level discographies, presumably without a control level. Two- and three-level discographies accounted for 31% and 14.5%, respectively. Sixty three-levels were graded as positive based on a concordant pain response. Seventy-five percent of patients had a single-level positive discogram, but 72% of these ‘single-level positive’ patients only had a single level studied. Two-level and three-level positive results were obtained in 18.9% and 6.2%, respectively. However, of the two-level positive discs, 77.8% only had a two-level discogram. None of the previously operated levels was included in discography. Fusion was performed to include all symptomatic levels determined by discography. Previously operated discs were included in fusion if they were adjacent to symptomatic discs. Anterior and posterior fusion techniques with and without varying instrumentation were performed. The clinical outcome was evaluated on the criteria of Zucherman which includes subjective symptom improvement, functional limitations, and amount of analgesic use. Categories were poor, fair, good, and excellent. Success of fusion was also assessed radiographically.

At mean 35-month (range 24–65 months) final follow-up, there was a 46% (n=22) satisfactory clinical outcome with solid arthrodesis in 47.9% (n=23). All patients who had satisfactory clinical outcome had a solid fusion. Conversely, 95.7% of patients who had a solid fusion had a satisfactory clinical outcome. None of the patients who had nonfusion had a satisfactory outcome. Therefore, if the surgical goal of fusion of the symptomatic levels as determined by provocation discography was obtained, the clinical success rate was 95.7%. This study was limited by its retrospective nature, small sample size, discography technique as judged by today’s standards, and varying operative techniques.

Derby et al. retrospectively evaluated pressure-controlled discography for its ability to predict surgical outcomes for interbody fusion, intertransverse fusion, combination fusion, or nonsurgical treatment.²⁷ The study’s premise is that pressure-controlled manometric discography may allow improved and more specific diagnosis and categorization of discogenic pain and may have the potential to predict outcome of surgery (Fig. 94.8). The positive discs found on provocation discography of 90 patients were classified as ‘chemically sensitive,’ ‘mechanically sensitive,’ or ‘indeterminate’ based on pain provocation at specific values above ‘static opening pressure.’ ‘Opening pressure’ is the manometric reading taken when injected contrast is first visualized entering the disc on fluoroscopy. Derby et al. classified a chemically sensitive disc as one which has positive concordant pain provocation of (1) immediate onset when less than 1 mL of contrast is visualized reaching the outer anulus, or (2) at less than 15 psi above opening pressure. These discs were considered the ‘most positive’ since the lowest pressures resulted in significant concordant pain. A disc was classified as ‘mechanically sensitive’ when concordant pain was noted between 15 psi and 50 psi above opening pressure. A disc was categorized as ‘indeterminate’ when concordant pain occurred between 51 and 90 psi above opening pressure. A disc was classified as normal if no pain occurred at pressures up to 90 psi above opening pressure. Thirty-six patients with chemically sensitive discs were identified. Clinical outcomes were based on the Patient Satisfaction Index adapted from the NASS low back pain outcome instrument, numerical rating scale for pain, and a modified ADL scale. ‘Favorable outcome’ was defined as at least two favorable outcomes of the three scales.

Fig. 94.8 Manometry has become an invaluable tool in the interrogation of painful discs. Several brands are currently available, with slight differences in the manometry meter and the ways to dispense the contrast while pressurizing the disc.

There was no significant difference in patient outcome between those who underwent interbody/combined fusion versus those who had intertransverse fusion if pressure-controlled discographic classification was not considered. However, those patients who had chemically sensitive discs were found to have an 89% rate of favorable outcome with interbody/combined fusion, 20% favorable outcome with intertransverse fusion, and only 12% favorable outcome with nonoperative treatment.

Although this study was limited by its small sample size and retrospective, nonrandomized nature, it demonstrates that low-pressure-sensitive discs may have a different pathobiology and respond differently to treatment than discs that are concordantly painful at higher pressures.²⁸

Derby et al. went on to perform a pilot study in which 32 patients were prospectively followed before and after undergoing IDET.²⁹ Percutaneous treatment was based on chronic low back pain for more than 6 months, axial pain comprising at least 60% of pain symptoms, negative neurologic and neural tension testing, and failure of conservative management. All had at least one positive disc (mean 2.04 symptomatic discs) on provocation discography as defined as provocation of at least 6/10 concordant pain accompanied by an abnormal nucleogram. The discs were classified as low-pressure discs if there was pain produced at minimal pressure less than 15 psi above opening pressure or high-pressure discs if pain was produced at more than 15 psi above opening pressure. Low-pressure discs comprised 37.5% of the discs and high-pressure discs were more common and comprised 62.5% of the discs. If there were more than two positive discs, a ‘clinical decision was made as to which discs were likely to be most symptomatic, and … included’ in the treatment levels.

Outcome was assessed using Roland-Morris Disability Questionnaire and Visual Analogue Scale, NASS Low Back Pain Outcome Assessment Instrument Patient Satisfaction Index, and a general activity questionnaire. Favorable outcome was defined as improvement of three or four of the tools while nonfavorable outcome was defined as worsening of three or four of the tools. No change was recorded for all other results.

Overall, 62.5% had a favorable outcome, 25% had no change, and 12.5% had a nonfavorable outcome. Among the low-pressure-sensitive discs, 75% had a favorable outcome while 55% of the high-pressure-sensitive discs had a favorable outcome. Thus, patients who had low-pressure-sensitive discs tended to do better that those patients whose discographic pain was reproduced at higher pressures.

This study was only a pilot study and was limited by small numbers of patients as well as the fact that the study population included patients with multilevel disc disease. There were also no data reported that compared pretreatment pain and disability scores between the low- and high-pressure disc groups. The heating protocol had been changed during the course of the study.

Rhyne et al.³⁰ retrospectively reviewed 25 cases in which patients who had single-level positive provocation discography did not undergo surgical treatment. The patients had been offered posterolateral fusion but refused due to fear of complications, insurance denial, knowledge of acquaintances who worsened after surgery, or desire to change lifestyle and live with the pain. The patients had a mean follow-up of 4.9 years and completed clinical evaluation including Roland-Morris disability scale and pain scale of Million et al. These scales were filled out twice, one representing their current status and the second to represent the recall of their status at the time of discography. Based on this assessment, 68% of patients improved, 8% had no perceived change in status, and 24% worsened. No patient was pain-free or free of disability. Of those who improved, pain improved an average of 39% and disability decreased an average of 42%. Patients who reported improvement had a shorter history of low back pain (3.5 years versus 11 years) and were older (45 years old versus 33 years old). Psychiatric disease was present in 66.7% of those who worsened. This study is certainly flawed in that not only is it a retrospective study, but the subjects were asked to retrospectively enter data to reflect their status an average of 5 years prior to the time of the study. However, it does demonstrate that 68% perceived an improvement.

In summary, these studies suggest that provocation discography is useful in the preoperative planning and prognostication for certain procedures. Discs that have been found to cause concordant pain on disc stimulation may also improve spontaneously. However, due to methodological insufficiencies and differences in these studies, more research is needed in the form of prospective, randomized trials with adequate sample size and validated outcome measures. The use of manometry may decrease the false-positive rate of provocation discography.

VALIDITY AND FALSE-POSITIVE RATES OF PROVOCATION DISCOGRAPHY AS A DIAGNOSTIC TOOL

Holt’s classic study questioned discography’s value, concluding that the false-positive rate of discography in asymptomatic volunteers was 36% based on the ratio of abnormal discographic images to the number of satisfactory disc injections.³¹ Holt’s work has been widely criticized. Simmons et al. ‘acknowledge[d] that Holt’s study was appropriate for its time,’ but noted subsequent advances in nonirritating contrast dye and radiologic equipment.^32–34 In their reassessment they noted Holt’s high failure rate for proper needle placement, which was 30% for the lower two disc levels (site of most disc pathology), 37% for L5–S1, and 23% for L4–5. The selection process for asymptomatic volunteers (who were prison inmates) was not clearly defined and there were discrepancies within the text. Additionally, only disc morphology, and not pain provocation, was considered in calculation of false positives. If pain response and disc morphology were considered, then reanalysis demonstrates that the true negative rate would be 74%. If only patients with successful injections were considered, then the accuracy rate of discography would be 81.9% despite the use of highly irritating contrast material (Hypaque).

‘Major advances in the techniques of discography’ occurring after Holt’s 1968 study prompted Walsh et al.³³ to perform a similar study in 1990, reevaluating discography’s value and specificity using ‘current techniques’ and more precise methodology. They used the much less irritating contrast, iopamidol, instead of diatrizole. More importantly, they included pain provocation evaluation. They rated the pain reproduction according to intensity, presence of pain-related behavior, and pain similarity (concordancy) for symptomatic patients. A ‘positive pain-related response’ to disc injection was recorded if the subject exhibited two or more types of pain behavior and rated the pain intensity a 3 or more (‘bad’ on a scale 0 to 5). Pain behaviors considered were guarding/bracing/withdrawing, rubbing, grimacing, sighing, or verbalizing. (Inter-rater reliability for these was 92.6%.) Because asymptomatic subjects could not report ‘concordant’ pain, they were termed ‘positive’ if there was an abnormal disc morphology with positive ratings for pain intensity and pain behavior. Symptomatic control patients required a typical rating for pain similarity to be called ‘positive.’

They included two groups: an asymptomatic group of young men and a symptomatic control group determining interobserver reliability for pain response evaluation and semiblinding the raters as to which participants were asymptomatic. The asymptomatic test subjects demonstrated an abnormal discographic disc morphology in 17% but a false-positive rate of 0% based on pain response. Even if more liberal criteria were used such as only minimal pain (at least 1 on a scale of 5) or presence of only one pain behavior, the false-positive rate would have been 3% and 7%, respectively.

Because the false-positive rate was lower than that of the Holt study (0% as compared to 36%), Walsh et al. concluded that the ‘single most important way to improve the specificity of discography is to incorporate assessment of pain into the definition of a positive discogram.’

Carragee and colleagues have extensively studied the possibility of false-positive results obtained from provocation discography in clinical practice (Table 94.1). Tanner and Carragee³⁵ suggest that ‘the reproduction of concordant pain has less diagnostic utility that [sic] often assumed, particularly if there is pathology in a similar sclerotomal region.’

Table 94.1 Summary of studies from Eugene J. Carragee, M.D., et al.

Initially, Carragee et al. presented case studies in which patients with positive lumbar discography were later found to have other painful processes such as sacroiliac joint abnormalities and posterior element neoplasm.³⁶ With these in mind, they appropriately sought to test the validity of provocation discography and determine its false-positive rates among various subjects by replicating and extending Walsh et al.’s study. They reasoned that in order to check the validity of any clinical test, it is essential to know how many subjects without a given disease will test positive with a particular test for that disease. Also, the relative risk of certain subsets of subjects will affect the meaning of the test. Therefore, they took as their starting point the use of asymptomatic control subjects.³⁵

Their first study applied ‘experimental disc injections’ to subjects with no previous history of low back pain to evaluate the pain responses and pain-related behaviors in the experimental setting. Its aim was to test the ‘first assumption in discography’ that ‘stimulation of a disc in an asymptomatic individual will not cause a significant sensation of pain.’³⁷ Twenty-six volunteers with an mean age of 43 years were selected from three sources: patients being followed after cervical surgery with best results and pain free (n=10), patients from the same cervical surgery cohort but with the worst results and with cervical-related chronic pain(n=10), and finally patients meeting the DSM-IV criteria for somatization disorder (initial n=10). All were asymptomatic for low back pain. Provocation discography was performed according to the protocol of Walsh et al.³³ with a ‘positive’ result scored if the pain response was greater than or equal to 3 out of 5, but only if accompanied by two or more pain behaviors (inter-rater agreement was 97.4%).³⁷

They found that disc stimulation was ‘false positive’ in at least one disc in 10% of subjects in the pain-free group and 40% of those with chronic pain (cervical-related chronic pain). Of the somatization group, 40% (n=4) dropped out before discographic injection and two others stopped the procedure after only one or two disc levels were injected. Of those remaining in the somatization group who completed all disc injections, 75% (3 of 4 subjects) had a least one-level positive discogram. If the somatization subjects who had at least one disc injected were considered, then there was an 83% rate of subjects with false-positive pain provocation.

When correlating radiologic (MRI and discogram) findings with pain provocation in subjects who completed at least 3 disc injections, there were no positive injections in the 31 radiologically normal discs, 11% (2 of 18) positive disc injections in the intermediate disruption discs (abnormal MRI and discographic nucleogram but no extension of dye to the outer anulus), and 37% (10 of 26) positive disc injections in the annular disruption discs (discogram with dye extension to or through the anulus).

Patients on disability who completed three disc injections (4 of 5) had an 80% false-positive rate. Subjects with active worker’s compensation or personal injury claims had an 89% rate (8 of 9) of false-positive injections. All subjects in the somatization group had some type of ongoing compensation claim which confounded separation of those two variables. A positive discogram correlated with an elevated modified Zung Depression Test and Modified Somatic Pain Questionnaire.

This study has been criticized on the basis that ‘concordant pain,’ paramount in provocation discography, cannot, by definition, be tested in asymptomatic individuals.³⁴ In response, Tanner and Carragee point out that all clinical tests must be evaluated in part by the results obtained in asymptomatic or disease-free persons.³⁵

Bogduk re-analyzed Carragee’s data in this study using IASP criteria (Table 94.2): that is, concordant pain with an intensity of 6/10, abnormal morphology, and at least one painless control level.³⁸ Specifically, he imposed the criteria that the suspect disc(s) be surrounded by at least one painless adjacent disc. He also applied the manometric criteria for pressure of injection. When at least one adjacent disc was required to be painless in order to call a given level positive, the false positives in the chronic pain group fell from 40% to 20%. But the rate remained stable for the no-pain group (10%) and somatization group (75%). Bogduk then used two different manometric parameters to analyze Carragee’s data. Under the criterion that the pressure of disc injection be less than 50 psi, the false-positive rates of both the chronic pain and somatization groups fell. The false-positive rates for the chronic pain group decreased to 10% (original 40%) while the somatization group had a resultant 50% false-positive rate (original 75%). If the stricter criterion of disc injection pressure less than 15 psi (called the ‘chemically sensitive disc’ by some) is used, then the false-positive rates fall to 0% in the pain-free group, 0% in the chronic pain group, and 25% in the somatization group. Additionally, Bogduk stated that because of the small sample size, confidence intervals should be used, and with these adjustments found that the false-positive rates either became zero or the confidence intervals overlapped zero even in the somatization group. Based on this reanalysis of Carragee et al.’s data, Bogduk suggested using the criterion of 50 psi in general clinical practice to restore sensitivity but 15 psi in patients with suspected or known somatization.

Table 94.2 IASP criteria for positive discography

In order to evaluate false-positive rates of ‘concordant’ back and buttock pain, Carragee et al. devised a study in which patients who had never had low back pain prior to posterior iliac crest bone grafting for nonlumbar spinal reasons were evaluated with provocation discography for concordant bone graft donor site pain.³⁹ Eighty-five percent of subjects (7 of 8) experienced similar or exact pain. Sixty-four percent of discs caused similar or exact pain. Of the discs with annular tears, 70% (7 of 10) caused similar or exact pain reproduction of the patient’s usual iliac crest bone graft harvest site. In this small sample of 8 patients and 24 discs, 50% of subjects had false-positive concordant donor site pain on disc stimulation if pain magnitude, concordancy, and demonstration of pain behavior are all considered. Carragee et al., again, did not use the IASP criteria. If these data are reanalyzed with IASP criteria, 25% of subjects would have false-positive findings if disc injection pressure was less than 50 psi, or a 12.5% false-positive rate for less than 15 psi. Still, these results call into question whether a patient can reliably distinguish pain from discogenic versus nondiscogenic sources.

After the rates of false positives in subjects without previous history of low back and/or buttock pain due to spinal pathology were evaluated, Carragee et al. next sought to determine the rate of positive disc injection in patients who had a previous history of lumbar-related complaints.⁴⁰ They no longer called these ‘false positive,’ presumably to eliminate the association with concordancy of pain response. They performed provocative discography on 20 patients from a cohort who had a previous history of symptomatic single-level lumbar pathology, underwent subsequent limited posterior discectomy 2–10 years prior to the study, but were now asymptomatic. They compared the rates of positive injection in this group to a control group of 27 subjects who were symptomatic with persistent or recurrent lumbar or lower limb symptoms after similar surgery 14 months to 6 years prior to the study. Twenty-six percent of these symptomatic patients had normal psychometric scores.

In the asymptomatic group, 40% of subjects had positive provocation disc injection at the level of previous surgery while only 10% (2 of 20) of asymptomatic subjects had a positive provocation if only the nonsurgically treated discs were considered. In total, 45% of the asymptomatic group had at least one positive provocation injection. All previously surgically treated discs that had significant pain on injection had discographic grade 2 to 3 tears with dye penetrating to or through the outer anulus. Sixty-three percent of symptomatic subjects had a positive pain response at the level of previous surgery. If the symptomatic group was subdivided according to psychometric scores, those with normal psychometrics had a 43% rate of positive provocation (all concordant) while those with abnormal psychometrics had a 70% rate of positive provocation (86% of which were concordant). The intensity of pain was significantly rated as higher in the symptomatic group with abnormal psychometrics (3.4 of 5) versus those in both the asymptomatic group and the symptomatic group with normal psychometrics (both 2.1 of 5).

In reanalyzing this asymptomatic group data with IASP criteria for manometric injection, the positive provocation rate in previously surgically treated discs would decrease to 35% if pressure on injection was less than 50 psi or to 25% if pressure on injection was less than 15 psi. Of unoperated discs, the positive rate would fall to 5% if either the less than 50 psi or less than 15 psi criteria was used. If the requirement for a pain-free adjacent control disc is also required, then the positive rate falls to 25% if less than 50 psi is used, or 15% if less than 15 psi is used. Still, previously operated discs may be more likely to be painful on injection than discs which have never been operated on.

In summary, the Carragee et al.⁴⁰ studies point out the fact that false-positive results can occur in discography as with any clinical test. These false positives are more likely in those with abnormal psychometric scores, specifically somatization disorder, and in previously operated discs. Bogduk has pointed out that adherence to the IASP criteria can decrease the false-positive results to an acceptable level.

O’Neill and Kurgansky published a retrospective study of 253 patients in an attempt to further delineate false-positive rates and offer suggestions to increase specificity.⁴¹ Particularly, they wanted to determine if the distribution of disc pain thresholds would organize into subgroups to help identify false positives. To accomplish this, they incorporated pressure-controlled discography through manometry. By correlating the manometry information with false-positive findings from Carragee’s asymptomatic populations,^{36,37,39,40,42} they were able to select out for false positives. They concluded that there was a 100% chance of false positives above 50 psi. Pressures ranging 10–25 psi had a 50% chance of being a false positive. However, discs with pain responses at 0–10 psi were most likely true positives. Based on these data, there appears to be a bimodal distribution of true positives. One population of true positive were those who had positive pain responses with cut-off values above 25 psi and below 50 psi; the other population of patients were those who had pain responses at less than 10 psi. However, it was difficult to confirm that the latter are true positive due to testing limitations.

DISCOGRAPHY COMPLICATIONS

Passing a needle into the disc has risks that should be recognized and anticipated before they occur. Perhaps the most common reaction or complication to any procedure is a vasovagal episode. This involuntary reaction of the parasympathetic nervous system causes reflex bradycardia and vasodilatation. Patients should be monitored with pulse oximetry, pulse rate, and blood pressure. A vagal episode is recognized by an audible slowing of the pulse rate and is usually accompanied by perfuse sweating and nausea. Increasing the intravenous fluids and raising the legs will usually restore cerebral perfusion, but 0.5–1 mg of intravenous atropine can be administered and incremental intravenous doses of 5–10 mg of ephedrine may be required in more severe reactions Other common reactions include paravertebral muscle pain and potential contusions from local punctures.

Less common complications include nerve injury, dural puncture, bowel perforation, epidural abscess, local cellulitis, and allergic reactions to contrast agents, topical iodine and prophylactic antibiotics. These risks can usually be avoided by using proper techniques and due diligence. The ventral ramus crosses the posterolateral quadrant of the target disc and is vulnerable to direct trauma, but the risk is minimized by slowly advancing the needle medial and under the root in the safe triangle and redirecting the needle when the patient complains of buttock or leg pain. Dural puncture and bowel perforation is possible but should not occur if one uses proper technique and carefully monitors the advancing needle with both AP and lateral imaging. An individual with a pendulous abdomen may have an increased risk of bowl perforation and subsequent disc space infection because the retroperoteneum is pushed posteriorly. Placing these patients in a lateral decubitus position may help avoid this potential complication.

Local cellulitis can be avoided with strict adherence to sterile techniques, but if it occurs it is treated with cold compresses and empiric coverage for group A streptococci and Staphylococcus aureus. Allergic reactions to drugs utilized can potentially be avoided with through histories. Patients with a serous allergic reaction to iodine and shellfish can be pretreated with prednisone 50 mg p.o. 24 hours before the procedure, prednisone 50 mg p.o. 12 hours before the procedure, and benadryl 50 mg p.o. 1 hour before the procedure or i.v. at the time of the procedure.

The most serious potential complication is disc space infection. Before antibiotic prophylaxis became routine, the reported rate of discitis ranged from 2.3% per patient and 1.3% per disc to 0.1% per patient and 0.05% per disc.⁴³ The most common causative organisms are S. aureus and S. epidermidis. The reported incidence of discitis may be reduced by using a double-needle techniques and preprocedural antibiotics.⁴⁴ In 1990, Osti et al.⁴⁵ designed an experimental discography model using sheep. They added cefalozin to the intradiscal suspension or administered it intravenously 30 minutes prior to intradiscal inoculation of bacteria. The prophylactic treatment prevented any radiographic, macroscopic, or histological signs of discitis. In a follow-up study of 127 patients this protocol also prevented disc space infection following discography and became the basis of the routine use of intravenous antibiotics (e.g. 1 g of cefalozin) prior to discography, adding antibiotics to the injected contrast, or the use of both intravenous and intradiscal antibiotics.

CONTROVERSIES

Although the issue of discography itself remains controversial, there are subjects within the procedure itself which should be addressed in order to further increase its diagnostic utility.

CONCLUSIONS

Lumbar discography is a diagnostic test that helps one determine whether a particular intervertebral disc is a source of pain. Properly performed and interpreted, discography is an invaluable tool. Together with the history, physical examination, and radiological studies, discography will identify asymptomatic discs and provide confirmatory evidence that a particular disc is the source of the patient’s pain. Because this information is often used to determine at which levels to perform a percutaneous or open surgical procedure, accurate and precise interpretation of the results is vital.