|
By:
Timothy S. Huang, MD
James F. Zucherman, MD
Ken Y. Hsu, MD
Michael Shapiro, MD
Daniel Lentz, MD
John Gartland, MD
All authors from:
St. Mary's Spine Center
One Shrader Street, Suite 450
San Francisco, CA 94117
Corresponding author:
Timothy S. Huang, MD
St. Mary's Spine Center
One Shrader Street, Suite 450
San Francisco, CA 94117
Phone 415-750-4955
Fax 415-750-8103
huangts@yahoo.com
Introduction
Of the 692 discs injected during lumbar discograms, end-plate disruptions with leakage of contrast material into the vertebral bodies were noted in 14 discs. Although gentle pressure was applied during the injections, severe fully concordant pain was reproduced in four (28.3%) discs, moderately severe and fully concordant pain in nine (64.3%) discs, and mild discordant pain in one (7.4%) disc. This is compared to 11.2% of the remaining 678 discs without end-plate disruption that reproduced severe concordant pain, 31.1% with moderately severe concordant pain, 17.1% with mild pain, and 40.6% without any pain reproduction. The difference between pain frequency in discs with end-plate disruption and those without is statistically significant (P < .001). This suggests that endplate disruptions may be related to painful segments. [Key words: discography, end-plate disruption, pain reproduction]
Study Design
This was a prospective study of MRI-discography using gadolinium as an intradiscal contrast agent.
Objective
To evaluate the utility of MRI-discography using gadolinium as an intradiscal contrast agent compared to the current standard of CT-discography using intradiscal iodinated contrast agent.
Background
MR arthrography with gadolinium has been used to evaluate shoulder, elbow, wrist, hip, knee, and other joints. Gadolinium has not been used as an intradiscal contrast agent. We previously reported our preliminary results using gadolinium as an intradiscal contrast agent for MRI discography. We now report the results of our completed study.
Methods
Forty-two disc levels were studied in 13 patients. Water soluble iodinated contrast and gadopentetate dimeglumine were injected at each disc level. Post discogram AP and lateral radiographs, CT scans and T1-weighted MRI scans were performed. Six physicians interpreted the results of each disc level in a blinded fashion. Interscan and interobserver interpretation variability was determined for MRI-discography and CT-discography using the Pearson correlation coefficient (r) test.
Results
Interscan variability was highly correlated between CT-discography and MRI-discography when used to determine disc normality, general degeneration, annular fissure, disc herniation, and contrast leakage (average r = 0.87, 0.87, 0.89, 0.92, and 0.77, respectively). Interscan variability was assessed by the Pearson test and all values of (r) for all readers were noted to be statistically significant at p<0.01. Interobserver variability was significantly correlated among the four more experienced readers (neuroradiologists and spine surgeons) but not among the spine fellows. Interobserver variability was also significantly correlated by CT- and MRI-discography for disc normality (CT r = 0.60, MRI r = 0.56), general degeneration (CT r = 0.76, MRI r = 0.71), annular fissure (CT r = 0.79, MRI r = 0.84), and disc herniation (CT r = 0.63, MRI r = 0.64). The readings for contrast leakage did not reach statistical significance for CT- or MRI-discography.
Conclusion
The high interscan and interobserver correlation rates obtained for MRI-discography using gadolinium compared with the standard CT-discography technique indicate that MRI-discography may be an acceptable substitute for the imaging of disc pathology. MRI-discography with gadolinium can be recommended for those patients with allergy to iodinated contrast agents and for patients who wish to limit their radiation exposure.
[Key words: disc degeneration, discography, iodinated contrast media, gadolinium, magnetic resonance imaging]
Precis
MRI-discography using gadolinium as an intradiscal contrast agent was compared to CT-discography using intradiscal iodinated contrast agent. MRI- and CT-discograms were evaluated by six physicians to assess for interscan and interobserver variability. The high correlation values obtained suggest that MRI-discography may be a suitable substitute for CT-discography in those patients with allergy to iodinated contrast agents and is an accurate alternative that also decreases radiation exposure.
Introduction
MR arthrography can safely and effectively evaluate intraarticular anatomy and detect abnormalities. Joints evaluated include shoulders, elbows, wrists, thumbs, hips, knees, and ankles7. In studies performed by Hajek, gadolinium was considered the ideal contrast agent for MR arthrography6. Gadolinium is a paramagnetic substance and produces a strong relaxation effect on adjacent hydrogen nuclei26. Gadolinium, a toxic heavy metal, can form safe stable chelates such as gadopentetate dimeglumine for medical use13. Gadolinium chelates have been used safely as intravascular and intra-articular MR contrast agents7. We previously reported our preliminary results using gadopentetate dimeglumine as an intradiscal contrast agent for MRI-discography27. This study was performed to further clarify if MRI-discography using gadopentetate dimeglumine as an intradiscal agent is a suitable substitute for the current standard of CT-discography using iodinated intradiscal contrast agents.
Methods
Forty-two discograms were performed in 13 patients. Discography was performed using a double needle technique via a posterolateral approach. Both iodinated contrast (Conray 60®, meglumine iothalamate) and dilute gadopentetate dimeglumine (Magnevist, in a concentration of 3mg/cc) were injected at each level. The two contrast agents were mixed and injected together at the time of discography. Following discography, AP and lateral radiographs, CT scans (contiguous 3mm axial sections on a GE CTi Scanner with sagittal and coronal reformations) and MRI scans (GE Signa 1.5T, T1-weighted FSE with fat suppression prepulse) were performed. Six physicians-two neuroradiologists, two orthopedic spine surgeons, and two spine fellows-interpreted the images of each disc level in a blinded fashion. Interscan and interobserver interpretation variability was determined for CT discography and MRI discography using the Pearson correlation (r) test. The parameters evaluated for our discography study were the following: general degeneration, annular fissure, disc normality, disc herniation, and contrast leakage (according to the Dallas Discogram Description)19.
Results
Table 1 presents the results of the interscan variability based on calculation of the Pearson correlation (r) coefficient. The interscan variability between CT-discography and MRI-discography correlated closely for all six readers in all parameters evaluated. For disc normality, the average (r) among the six readers was 0.87 with a range of 0.67 to 1.0 (perfect correlation). For general degeneration, the average (r) was 0.87 with a range of 0.75 to 0.97. For annular fissure, the average (r) was 0.89 with a range of 0.76 to 0.98. For disc herniation, the average (r) was 0.92 with a range of 0.80 to 1.0. For contrast leakage, the average (r) was 0.77 with a range of 0.61 to 1.0. All values of (r) for all readers were noted to be statistically significant at p<0.01.
Table 2 presents the results of the interobserver variability as assessed by the Pearson test. The interobserver variability significantly correlated only among the four more experienced readers (neuroradiologists and spine surgeons) but not among the less experienced readers (spine fellows). For disc normality as assessed by CT-discography, the average (r) among the four readers was 0.60 with a range of 0.40 to 0.82; when assessed by MRI-discography, the average (r) was 0.56 with a range of 0.36 to 0.88. For general degeneration as assessed by CT-discography, the average (r) was 0.76 with a range of 0.64 to 0.89; when assessed by MRI-discography, the average (r) was 0.71 with a range of 0.60 to 0.87. For annular attenuation as assessed by CT-discography, the average (r) was 0.79 with a range of 0.73 to 0.86; when assessed by MRI-discography, the average (r) was 0.84 with a range of 0.76 to 0.89. For disc herniation as assessed by CT-discography, the average (r) was 0.63 with a range of 0.47 to 0.90; when assessed by MRI-discography, the average (r) was 0.64 with a range of 0.45 to 0.90. All values of (r) for the above four parameters were noted to be statistically significant at p<0.01 with the exception of the parameter of disc normality as assessed by MRI (statistically significant at p<0.02). The interobserver readings for contrast leakage did not reach statistical significance for either CT-discography or MRI-discography. For contrast leakage as assessed by CT-discography, the average (r) was 0.22 with a range of -0.09 to 0.29; when assessed by MRI-discography, the average (r) was 0.41 with a range of 0.26 to 0.54.
Examples of MRI-discogram compared with CT-discogram are shown in Figures 1-3. Figures 1A and 1B show axial images of CT- and MRI-discogram, respectively, of a normal disc. Figures 2A and 2B show images of a herniated disc. Figures 3A and 3B show sagittal images of CT- and MRI-discogram of multilevel degenerative disc disease.
Discussion
The clinical value of discography continues to be controversial2, 8, 17, 18, 20. Historically, discography has been reported by various authors to be a sensitive reflection of the various stages of disc degeneration1,3. The main criticism of discography lies in its level of accuracy. False positive rates of up to 37% have been reported in the literature8. The specificity of discography can be improved by using stricter criteria for what constitutes a positive exam. Familiar and significant pain, along with degenerative morphology on discogram, constitute the necessary criteria for a positive exam. The provocative aspect of discography with reproduction of exact or familiar pain is the most important part of the exam. Using such strict criteria, Walsh reported a false positive rate of 0% and a specificity of 100% for discography in a population of young asymptomatic volunteers25.
Computed tomography scanning after discography has proven helpful because of the additional information obtained. The axial views provided by CT-discography allow a classification of both disc degeneration and annular ruptures, as well as the identification of disc herniations and other pathology19. Studies from the Texas Back Institute have shown that certain pathologic features (disc degeneration, annular disruption and disc herniation) and personal or structural variables (age and lumbar level) affect the reporting of pain during discography19, 21-24. Other authors have disputed their findings and reported that disc herniation and the presence of outer annular ruptures are the best predictors of pain reproduction12,14. Disc degeneration, if not associated with outer annulus rupture, was not a predictor of familiar pain. Despite the conflicting claims, there is general agreement that CT-discography provides crucial information in the diagnosis and management of patients with discogenic back pain2,4,19.
For physicians that use discography as a diagnostic aid, our study offers an additional method of obtaining the same visual information as CT-discography. Our data demonstrate that MRI-discography using gadolinium as an intradiscal agent can be used as an effective substitute for the current standard of CT-discography using iodinated contrast material. The high interscan correlation values (Table 1) give credence to this claim. This is especially noteworthy since relatively inexperienced readers (spine fellows) were able to read the MRI and CT scans with a high degree of interscan correlation. Very high values of (r) were obtained for disc herniation and annular fissure (0.92 and 0.89, respectively), the two factors most predictive of familiar pain on discography12,14.
A comparison of the interobserver correlation values obtained (Table 2) also suggests that MRI-discography to be an acceptable substitute for CT-discography. Although the interobserver correlation values calculated were not as high as that for the interscan correlation values, they did reach statistical significance. For each parameter studied, the correlation values for the MRI-discography were very similar to the correlation values for the CT-discography. This suggests that readers were as likely to agree/disagree with each other regardless of the visual medium (CT or MRI), that one medium was not any better than the other. The fact that interscan correlations were very high, and higher than interobserver correlations, supports the point that MRI-discography produces the same graphic anatomic information as CT-discography.
In our study, iodinated contrast media (Conray) and gadopentetate dimeglumine (Magnevist) were mixed and injected together at the time of discography. Conray is not known to interact with gadolinium. The mixture of Conray with gadolinium does not adversely affect the MR images obtained as long as small amounts of Conray are used. When performing MR arthrography, confirmation of intraarticular needle placement is often made by first injecting a small amount of iodinated contrast material. Gadolinium is then injected into the joint and the joint imaged. Kopka mixed 10 ml of iodinated contrast material with gadolinium and injected the solution into shoulder joints and noted no untoward effects on the MR images obtained11. The addition of gadolinium to saline is unlikely to change the viscosity of the fluid. The concentration used (3mg/cc) is equivalent to one drop of gadolinium in 10cc of saline. The injection of a mixture of Conray and gadolinium elicited concordant pain in a similar manner as the injection of any other agent of similar viscous characteristic.
There are no published studies on the effect of gadolinium on nucleus pulposus tissue. Gadolinium has been widely used intravascularly and is safe for intraarticular use7. The composition of nucleus pulposus most closely resembles articular cartilage-high type II collagen and proteoglycan content. Given its safety in intraarticular use, one can surmise its safety for intradiscal use. However, we do not know the long-term effect of gadolinium (even in the very small concentration used) on an avascular tissue such as the disc. Since 1997, we have used gadolinium in our patients with allergy to iodinated contrast media and have noted no adverse effects. More work is needed in this area.
Allergy reactions to iodinated contrast agents are a potential complication of discography. Older, ionic iodinated contrast agents (e.g. Conray) have been reported to cause adverse reactions in about 10-15% of patients5. Newer nonionic iodinated contrast agents (e.g. Omnipaque) have lower osmolality and is though to reduce the incidence of adverse reactions. Katayama reported on 169,284 cases injected with either ionic or nonionic contrast agents.10 The incidence of total adverse drug reactions (ADRs) was 12.66% in the ionic contrast media group compared to 3.13% for the nonionic contrast media group. The incidence of severe ADRs (defined as dyspnea, sudden hypotension, cardiac arrest, or loss of consciousness) was reported to be 0.22% (ionic contrast) and 0.04% (nonionic contrast). One death was reported in each of the two group10.
Patients with a history of asthma/bronchospasm, previous adverse reaction, or allergy to iodides/other foods have a 5-10 times increased rate of an adverse reaction to contrast media5. Other known risk factors are patients with diabetes, cardiac, renal, or hematological diseases, as well as patients with inadequate hydration7. Premedication with steroids and antihistamine is advised for patients with known risk factors. There are no large published studies on the rate of ADRs in patients with known risk factors and receiving steroid/antihistamine Premedication prior to contrast media injection. Such data would be extremely helpful.
The frequency of adverse reactions to gadolinium is lower than the frequency of adverse reactions to iodinated contrast media. Murphy reported on 21,000 patients in whom gadolinium contrast media were administered15. The incidence of total ADRs was 0.138% while the incidence of severe ADRs was 0.01% when gadopentetate dimeglumine was used. In a survey encompassing 108 centers and involving 687,255 patients receiving gadopentetate dimeglumine, an overall ADRs incidence of 0.06% and a severe ADRs incidence of 0.0014% was reported16. Though safer than iodinated contrast agents, gadolinium can cause severe life-threatening reactions. As with iodinated contrast media, mortality has been reported with the use of gadolinium contrast agents9.
At our institution, the hospital charge for a lumbar MRI is $1328. The hospital charge for a lumbar CT is also $1328, with an additional $765 for sagittal and coronal reformation. The charge for a limited lumbar CT-a few slices at one or two disc levels-is $1063. Excluding other costs associated with the procedure, the cost of a lumbar MRI represents a 25% increase over the cost of a limited lumbar CT. This is a mild premium given its safety and potential savings from engendering less patient adverse reactions.
Another potential advantage of MRI-discography compared with CT-discography is the reduced radiation exposure. MR imaging does not expose the patient to any radiation. The amount of radiation with CT imaging varies depending on whether a "limited" CT was performed or whether a full lumbar CT was performed. In the middle-aged and elderly patients, radiation exposure will be of less concern. In younger patients, radiation exposure should be minimized.
One shortcoming of gadolinium is the inability to fluoroscopically visualize this agent during intradiscal needle placement. During MR arthrography, needle placement is confirmed with the injection of a small amount of iodinated contrast agent. For those with a true iodine allergy, this option will not be possible. Accurate needle placement under biplane fluoroscopy is critical before intradiscal injection with gadolinium contrast media.
In summary, the utility and value of discograms continue to be controversial. For those who find CT-discography helpful in diagnosing and treating discogenic low back pain, MRI-discography may prove to be a suitable substitute. Our data show that high interscan correlation rates were obtained for MRI-discography using gadolinium as an intradiscal agent compared with CT-discography using iodinated contrast media. Gadolinium contrast agents are safer and cause less adverse reactions than iodinated contrast agents. The cost of MRI-discography compared with CT-discography is only mildly higher. Patients are exposed to less radiation with MRI-discography when compared to CT-discography. MRI-discography may be recommended for those with true iodine allergy. The excellent visual information obtained, along with the pain provocation response, may prove helpful in the diagnosis and treatment of discogenic low back pain.
| Table 1: Interscan Variability |
| Parameter studied |
Average (r) |
Range (r) |
p value |
|
| Disc normality |
0.87 |
0.67-1.0 |
< 0.01 |
| General degeneration |
0.87 |
0.75-0.97 |
< 0.01 |
| Annular fissure |
0.89 |
0.76-0.98 |
< 0.01 |
| Disc herniation |
0.92 |
0.80-1.0 |
< 0.01 |
| Contrast leakage |
0.77 |
0.61-1.0 |
< 0.01 |
|
| Table 2: Interobserver Variability |
| Parameter studied |
Average (r) |
Range (r) |
p value |
|
Disc normality--CT
Disc normality--MRI |
0.60
0.56 |
0.40-0.82
0.36-0.88 |
<0.01
<0.02 |
General degeneration--CT
General degeneration--MRI |
0.76
0.71 |
0.64-0.89
0.60-0.87 |
<0.01
<0.01 |
Annular fissure--CT
Annular fissure--MRI |
0.79
0.84 |
0.73-0.86
0.76-0.89 |
<0.01
<0.01 |
Disc herniation--CT
Disc herniation--MRI |
0.63
0.64 |
0.47-0.90
0.45-0.90 |
<0.01
<0.01 |
Contrast leakage--CT
Contrast leakage--MRI |
0.22
0.41 |
-0.09-0.29
0.26-0.54 |
not significant
not significant |
|
|
 |
