2009 ISSLS Prize Winner: Does Discography Cause Accelerated Progression of Degeneration Changes in the Lumbar Disc: A Ten-year Matched Cohort Study

Eugene J. Carragee, MD; Angus S. Don, FRACS; Eric L. Hurwitz, DC, PhD; Jason M. Cuellar, MD, PhD; John Carrino, MD; Richard Herzog, MD


Spine. 2009;34(21):2338-2345. 

In This Article

Materials and Methods

Study Design

This is a prospective, matched cohort study to investigate the morphologic effects of provocative lumbar discography on the intervertebral disc. At 7 to 10 years after baseline MR imaging of the lumbar spine, the progression of disc degeneration was assessed in subjects undergoing provocative discography (L3-S1) and compared to matched subjects having the same baseline MRI evaluation.

Primary Hypothesis

Lumbar discography using modern techniques (small needle gauge, pressure controls, antibiotics, and nonirritating contrast) will not be associated with an observable increase in disc degeneration findings on MRI 7 to 10 years after disc puncture and injection.

Subject Recruitment

Between 1997 and 1998, a cohort of 75 subjects were recruited and enrolled in a study of discography (L3-S1) in persons asymptomatic or minimally symptomatic for low back pain.[14,16,20] Subjects were recruited from 1 of 3 patient pools: subjects having documented cervical disc disease (n = 40); subjects having previous lumbar disc herniation with complete symptom resolution (n = 25); and subjects with no history of either cervical nor lumbar disc illness but who did have a history of serious psychological distress consistent with a somatization disorder (n = 10). From the same 3 subject pools, 75 matched subjects who would not have discography performed, were simultaneously recruited to act as control subjects: again these were recruited in the same proportions from the previous cervical disease (n = 40), previous symptomatic lumbar disc herniation (n = 25), and somatization groups (n = 10). These subjects were matched to age, sex, and previous cervical or lumbar disc troubles, and psychometric profiles, and had the same baseline evaluations (LBP questionnaires, psychometric questionnaires, radiographs, and MRI scan) with the sole exception of the provocative discography (L3-S1).

After the baseline evaluations, 75 discography subjects were scheduled to undergo lumbar discography (L3-S1). The results of these studies have been previously reported.[14,16,20] Discography and control subjects were then followed at intervals for clinical low back pain problems. Clinical low back pain problems up to 5 years after enrollment in these discography and control subjects has previously been reported.[19,21]

At 10 years after the start of the study, discography subjects and controls were contacted. If subjects had received a high-quality lumbar MRI more than 7 years after the discogram, these were obtained and compared to the baseline MRI. If no MRI had been performed for clinical reasons, or if the MRI obtained was not consistent with the protocol for MRI reading, subjects were asked to undergo a new MRI at Stanford University School of Medicine. (See Follow-up MRI Collection).

Baseline Entry Criteria

The detailed entry criteria were described in each previous study.[14,16,20] All subjects were screened for current low back problems using a screening questionnaire and the Oswestry Disability Index. Where subjects were involved as patients or subjects of concurrent studies, the original medical charts, diagnosis criteria, or study questionnaires were examined to confirm the absence of significant LBP symptoms. Inclusion criteria required that subjects were not receiving or seeking medical treatment for LBP, be taking no medications for backache, and have no activity restrictions because of LBP. Of the subjects with minimal back pain and those recruited after lumbar discectomy, there was a further criterion that subjects have normal psychometric scores. Approval was obtained from the institutional review board and the Administrative Panel of Human Subjects in Medical Research according to US Department of Health and Human Services regulations at Stanford University School of Medicine. Informed consent according to University and US Department of Health and Human Services guidelines was obtained from all prospective participants at the original studies.

Psychometric Studies

A Modified Zung Depression Test, Modified Somatic Pain Questionnaire (full score) (MSPQ), and Medication Scales were administered to each patient as a routine intake step. Both the Zung Depression Test and the MSPQ are validated tests.[22]


The discography protocol that was used in these studies was uniform across all study groups. The discographer with more than 10-years experience and an average of 50 to 100 discography examinations per year performed all the injections. The discographer and his assistant/observer were blinded to the group assignments and performed all discography on a mix of asymptomatic volunteers and active clinical patients with serious low back pain. Similarly, the assistants/observers who graded the response to injections were also blinded to whether the injection was performed on an active clinical patient or an asymptomatic volunteer. The side of needle entry (right or left) was determined by the predominant site of pain in the active clinical LBP patients; whereas in the asymptomatic volunteers, research assistants randomly assigned the side by random number sequence, immediately before the discography. Pressure measurements were made during the injection indicating continuous recordings in pounds per square inch on a manometer, with each 0.5 mL of the injection (Hewlett-Packard, Palo Alto, CA). A pressure relief valve set at 100 psi limited the maximum possible injection pressure to this level. The lowest static pressure (relative to the opening pressure) associated with a pain response was noted. In those subjects without pain on injection, the highest static pressure was recorded. The criteria for pain response, pain behavior, and determining a positive injection was the same as used in the original work of Walsh et al[12] A low pressure positive injection met these criteria with a static pressure of less than 22 psi when pain was provoked.[23] In all subjects, a negative control disc was required to determine a positive finding.

Follow-up MRI Collection

Discography and control group subjects were contacted at 10-years after the baseline testing. An independent research assistant (EvdH) who was blinded to patient baseline data conducted scripted telephone interviews, This examiner had no knowledge of control or discogram status and was not involved in the study design. The interview was conducted by telephone, including an interval medical history, interval lumbar imaging studies history, occupational history, medication usage, serious injury history and spinal procedure history since the last (5-year) follow-up record. Subjects were enrolled for 10-year follow-up MRI evaluation unless they were noted to have one or more of the following exclusion criteria: subjects were excluded if since the baseline evaluation they had a history of new lumbar spine fracture or dislocation; interval lumbar spine surgery or intradiscal therapy of any type; another lumbar discogram; lumbar infection or tumor; new rheumatic disease diagnosis which required treatment; or was unable to have a new MRI due to imaging contraindications (e.g., cardiac pacemaker, ocular injury, etc.). If subjects had received an MRI of the lumbar spine during the last 3 years, and that MRI met the analysis protocol, that scan was used in lieu of another MRI at the 10-year point.

MRI Protocols

All baseline images were obtained from one center. All 10-year follow-up studies by protocol were obtained from the same center and used the same protocol as the baseline examinations.[24,25] MR images performed for clinical indications (back or leg pain) were acquired from several clinical practices when done before the 10-year protocol examination; as a result, variable MR imaging techniques were performed in a minority of studies. However, all MR imaging protocols had the following pulse sequences in common: T1-weighted sagittal conventional spin echo, T2-weighted sagittal fast spin echo (FSE) or turbo spin echo without or with fat suppression, and T2-weighted axial fast spin echo/turbo spin echo parallel to the intervertebral disc. All baseline and follow-up images were collected electronically and stored directly as DICOM files. All images were deidentified for patient confidentiality and readers were blind to discography versus control status of the subjects.

All lumbar levels (i.e., 5 discs, L1-S1) were scored for qualitative and quantitative findings of disc degeneration, disc contour changes, anular disruption, and endplate changes by the protocol below.


The primary comparison was between the progression of degenerative findings in the injected discs (L3-S1) of the discography group compared to the progression in same levels in the matched control group.

Secondary comparison between groups was made for the noninjected levels (L1-L3) to determine whether the rate of degeneration in the noninjected discs was otherwise similar between groups independent of the discography intervention at lower levels.

Within the discography group, progression of disc findings was analyzed for correlation with (1): with the size of the needle puncture (22 vs. 25 gauge) (2); whether the disc injection had been painful; and (3) whether degeneration changes (e.g., disc protrusion) appeared to be greater on the side ipsilateral to the needle puncture.

Quantitative Image Evaluation. Quantitative measurements of disc height and disc signal intensity was performed using the method described by Videman et al.[26] Disc signal intensity was determined relative to the adjacent CSF signal. Interval signal change is expressed as the percentage signal change compared to the baseline images (i.e., signal loss is recorded as a negative percentage, brighter signal at follow-up is recorded as a positive percentage).

Qualitative Image Evaluation. The observers for the qualitative and semiquantitative evaluation were four physicians experienced in spine MRI interpretation (2 radiologists and 2 orthopedic spine surgeons). Similar training on this grading system was provided to all examiners. A handbook containing standardized definitions of imaging characteristics with pictorial and diagrammatic examples was provided to each reader. Definitions were derived from the literature or by consensus when no relevant publication was available. This MR reading protocol has been previously described and has acceptable inter and intrarater reliability.[24,25] Before study initiation, readers evaluated a sample set of images and met to review images and refine the standardized definitions.

All MR images were graded by 2 examiners for findings at the lowest 5 disc segments. If there was disagreement in grading, the images were independently regraded by 2 additional tie-breaker examiners. If these were both consistent with one of the original grades, this grade was accepted. If disagreement still existed between the 2 tie-breaker examiners, a consensus reading would be obtained with at least 3 examiners reviewing the films together.

The spine MR imaging findings assessed were the following: disc degeneration, disc herniation, marrow endplate abnormality (Modic changes), intervertebral disc posterior, posterior-lateral or foramenal anular hyperintense zone, disc height, and relative signal intensity compared to the adjacent CSF.

Image interpretation was recorded using a standardized data collection form prompting the reader to select from multiple choice lists of findings for imaging characteristics at each intervertebral disc level. This method of qualitative grading of common degenerative findings has previously been evaluated for inter-rater and intrarater reliability.[24,25] Classification of disc morphology by this method showed substantial intra- and inter-reader agreement (kappa 0.81). 3 of the readers (EC, JC, RH) in the current study participated as readers in the Lurie et al study, and have had extensive experience with this method. The other reader was trained in the same fashion as the readers in the Lurie et al study.

Disc degeneration grade was ranked on a 5-level ordinal scale as described by Pfirrman et al in 2001.[27] The type of endplate marrow abnormality was designated using the Modic classification for simplicity of reference.[28] Each vertebral level was scored by the dominant Modic change. Semiquantitative assessment of extent of endplate marrow signal alteration was performed for each endplate involved above (inferior) and below (superior) a specified intervertebral disc level. The observer assessed the amount of signal alteration from anterior-posterior (AP): less than or equal to 50%, or greater than 50%; and cranial-caudal (CC): less than 25%, 25%-50%, or greater than 50%). A hyperintense zone (HIZ) was defined based on the original description as an area of bright signal intensity in the posterior anulus that is brighter than nucleus on T2-weighted images.[29] The location of the HIZ was designated as central, posterolateral, or foraminal (as well as right and left if not central). Presence of an anterior HIZ was not assessed. Disc herniation was graded as focal protrusion, broad protrusion, extrusion and sequestration. Herniations were also graded for location (central, paracentral, foraminal, and far-lateral), right and left sided, and root displacement or compression.[24]

Statistical Methods

Descriptive statistics were used to summarize patient socio-demographic and MRI characteristics measured at baseline, and at follow up. Means, standard deviations, and medians were calculated for continuous variables; frequency distributions were generated for categorical variables. Comparative between group analysis of categorical MRI findings (disc grade, HIZ, herniation, and Modic changes) used a χ2 test. Comparative between group analysis of continuous, normal distribution data (disc signal intensity and disc height) was performed using a student t test. StatView statistical program (SAS Institute, Cary, NC) was used for all analyses.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.