IV Versus Myelography Computed Tomography for Thoracolumbar Intervertebral Disc Extrusion Surgical Planning in French Bulldogs
ABSTRACT
Accurate diagnostic imaging is required for surgical planning of acute thoracolumbar intervertebral disc extrusion. However, data comparing the accuracy of conventional IV contrast-injected computed tomography (CT) and myelography CT for hemilaminectomy localization and size assessment are sparse. In this study, IV contrast-injected CT and myelography CT were performed in 48 French bulldogs presenting with acute disc herniation and compared with postsurgical CT. CT images were evaluated by three raters. IV contrast-injected CT erroneously identified the compressive lesion site significantly more often than myelography CT. The length of the compressive lesion was significantly higher using myelography CT compared with conventional CT, but this did not lead to relevant consequences on the surgical opening site length. Myelography CT should therefore be recommended when thoracolumbar disc disease is suspected and multiple compressive lesions are visualized on IV-injected CT in French bulldogs.
Introduction
Intervertebral disc disease is the most common cause of myelopathy in French bulldogs, considering the chondrodystrophic nature of the breed.1,2 In acute intervertebral disc extrusion (IVDE), rupture of the annulus fibrosus and nuclear herniation are preceded by progressive chondroid degeneration of the gelatinous nucleus pulposus. Localization of the IVDE is critical for accurate surgical planning; it allows decreased anesthetic time and effective decompression, thereby shortening recovery duration.3,4
Myelography is a radiographic technique in which spinal radiographs are obtained after injection of radiopaque contrast media into the subarachnoid space. Historically, it has been the recommended imaging technique for IVDE diagnosis. The development of computed tomography (CT) has improved extradural disc herniation conspicuity.5 CT shows better sensitivity for IVDE identification in dogs than myelography (82% vs. 84% or 100% vs. 90% depending on the study).6,7 Conventional CT with IV injection of iodinated contrast medium (ivCT) was described as more precise than noninjected CT in detecting extradural compression in dogs.8,9 Myelography CT (mCT) combines the advantages of conventional CT and myelography. It allows the assessment of the entire spine within a short acquisition time, and the use of multiplanar reconstructions in mCT is highly beneficial when planning surgical openings. MRI requires longer anesthesia time than conventional CT.10 No difference in outcome was found between dogs undergoing myelography and those undergoing MRI in a recent large-scale retrospective study.11
The aims of this study were to compare in a population of French bulldogs (1) the assessment of the extent of spinal cord compression between ivCT and mCT, (2) the recommended surgical openings (localization and site) and eventual discrepancies between the two techniques, and (3) the length of the estimated compression to the size of the final surgical opening that the surgeon considered sufficient for good spinal cord decompression on postoperative CT.
Materials and Methods
Recruitment of Dogs
Sample size was calculated before the beginning of the study to determine the minimum number of cases required for statistical significance. A sample size of 42 dogs was estimated to show an expected mean difference in the length ratio of 0.75 between ivCT and mCT, with an expected standard deviation of 1.3, estimated dropout rate of 50%, alpha of 0.025, and power of 80%.
French bulldogs presenting in Fregis Hospital (Arcueil, France) with a first episode of acute-onset T3–L6 disc herniation between January 2017 and August 2019 were prospectively recruited. Dogs with known systemic diseases that would increase the risk of anesthesia were excluded. Dogs were surgically treated by hemilaminectomy. Dogs with no improvement at the 2 mo follow-up were excluded. Dogs with persistent compressive material on postoperative CT were excluded. The aim of these two last exclusion criteria was to exclude dogs with erroneous surgical site.
Study Population
Fifty French bulldogs were initially included in this study. However, two dogs were excluded due to the lack of clinical improvement after surgery, leading to the inclusion of 48 dogs. The median age of the patients was 4 yr (range, 2–8 yr). The group consisted of 9 (19%) castrated males, 18 (38%) males, 17 (35%) spayed females, and 4 (8%) females. The median Thompson grade was 3 (range, 2–5), with 21 (44%) grade 2 dogs, 18 (38%) grade 3, 8 (15%) grade 4, and 1 (2%) grade 5 dog.
Procedures
All dogs underwent complete physical and neurological examinations, complete blood cell counts, and serum biochemical profiling at admission. Written informed owner consent was obtained for every dog. Neurological signs were evaluated and localized by a board-certified neurologist or neurology resident and graded according to the Thompson grading score. Dogs were premedicated by IV injection of 0.5 mg/kg diazepama and 0.2 mg/kg morphine.b Anesthesia was induced by an IV propofolc dose response and maintained with isoflurane and 100% oxygen.
Spinal CT was performed using a 16-slice CT scannerd, with the patient positioned in dorsal recumbency. Transverse images of the spine were obtained from the third thoracic vertebra to the first sacral vertebra using the following parameters: helicoidal mode, 165 mA, 120 KVp, 1.25 mm slice thickness, 0.625 mm interval, and 0.938 pitch. Sagittal and dorsal plane images were reformatted using 0.9 mm thickness for all dogs.
After the initial acquisition, an IV bolus of 320 mgI/mL iodixanol solutione was administered at a dosage of 2 mL/kg, and image acquisition was repeated 2 min after injection. A myelogram was then performed by a board-certified neurologist or a neurology resident. Lumbar puncture of the subarachnoid space was performed under radiographic guidance in all dogs. The intrathecal contrast agent used was 300 mg/mL iopamidol.f Subarachnoid contrast injection and appropriate thoracic extension of the contrast agent were ensured with the guidance of digital radiographs. An additional CT scan was then performed using the same parameters as those previously mentioned. The dogs underwent decompressive surgery immediately after CT completion. A postoperative CT scan was performed immediately after surgery.
The median duration for myelography procedures (including puncture, contrast agent injection, and radiographic assessment) was 18 min (range, 10–47 min). The median volume of the contrast agent injected was 0.2 mL/kg (range, 0.1–0.4 mL/kg). None of the dogs developed postmyelography seizures or irreversible nervous system degradation. All lesions were confirmed as type-1 (acute) disc extrusions during surgery.
Collection of Clinical Data
Age, sex, body weight, duration of clinical signs before imaging, and duration of imaging studies were recorded.
Surgical Approach
Surgical decompression by hemilaminectomy was performed by a board-certified surgeon or surgery resident, after interpretation of the ivCT and mCT by the neurologist and surgeon at the time of diagnosis with direct visualization of spinal cord compression during surgery. The nature of the compressive material was characterized by the surgeons as mineralized, gelatinous, hemorrhagic, or mixed (gelatinous and hemorrhagic). Postoperative care was individualized according to the needs of each dog.
Image Evaluation
Transverse thoracolumbar ivCT and thoracolumbar mCT were reviewed separately for each patient. The images were anonymously evaluated by three raters: one third-year neurology resident (rater 1), one board-certified neurologist (rater 2), and one third-year diagnostic imaging resident under the supervision of a board-certified radiologist (rater 3). Window width and level adjustments were allowed according to the rater’s preferences, using a multiplanar reformatting software (Osirix). The raters were unaware of the patient’s signalment, history, and the neurolocalization of the suspected lesion. ivCT and mCT images were randomly presented to the raters. Compression was defined as a distortion of spinal cord edges. They were asked to identify the intervertebral disc herniation site with the maximal spinal cord compression in each sequence and to determine the following:
Transverse ratio (TR): Ratio of the cross-sectional area of the cord and subarachnoid space to the area of the vertebral canal (placed perpendicular to the floor of the vertebral canal) (Figure 1).
Length ratio (LR): Ratio between the length of the compression (identified on transverse images) divided by the length of the second lumbar vertebral body (Figure 1).
A recommended site accuracy index (R), with a precision of one-third of the vertebral body. Each rater was asked for the recommended site of hemilaminectomy from the ivCT and mCT images. The minimum length used was one-third of a vertebral body, and there was no maximum length. If the opening site on postoperative CT was strictly outside the recommended zone, R was rated as 0. Otherwise, R was rated 1 if the opening site had at least one-third of the vertebral body in common with the recommended site, which means that at least one segment of the postsurgical site is comprised in the recommended area.
Estimated deviation between the recommended site of hemilaminectomy and postsurgical findings (E): Only dogs with R = 1 were included in the E measurement. E is the sum of the cranial and caudal differences between the recommended site and the opening site on postoperative CT (with one unit representing one-third of the vertebral body). For example, an estimated site of maximal compression going from the cranial third of T13 to the second-third of L2 was considered. Additionally, the postoperative opening was from the second-third of T13 to the caudal third of L1. In this case a difference of one-third of a vertebra cranially and two-thirds of a vertebra caudally would be found: E = 3 (i.e., 1 e + 2 e).
The lateralization for the surgical procedure (right or left): Side of the spinal cord most affected by the compressive lesion.
The average attenuation of the compressive material: A region of interest including all the compressive material was manually drawn, and the mean attenuation value was calculated in this area.
The presence of thinning of contrast columns in continuity with the compressive lesion on mCT.
The presence of any other extradural or intradural extramedullary lesion along the thoracolumbar spine (disc protrusion, vertebral malformations, subarachnoid diverticulum, etc.)
Citation: Journal of the American Animal Hospital Association 59, 5; 10.5326/JAAHA-MS-7318
Statistical Analysis
Statistical analyses were performed using a commercially available software.g For descriptive statistics, continuous data were assessed for Gaussian distribution by histogram evaluation and the Shapiro-Wilk test (Gaussian if P > 0.05). Gaussian data were presented using mean (standard deviation) and non-Gaussian data using median (minimum–maximum). Categorical data were presented as number of dogs (percentage).
Differences in TR, LR, E, and P between ivCT and mCT were assessed using mixed-effects analysis of variance models with identity covariance matrices to consider the repeated-measures design of the study. For each model, the dependent variable was either TR, LR, E, or P; the fixed-effects independent variables were the CT technique (i.e., ivCT or mCT), rater (i.e., rater 1, 2, or 3), and interaction between the two; the random-effects variable was the individual dog. The Gaussian distribution and homoscedasticity of the residuals were assessed using a graphical assessment of frequency distribution histograms and residual plots, respectively. Differences were further investigated using contrast and predictive margins if necessary. The difference in R between ivCT and mCT was assessed using mixed-effects logistic regression models with identity covariance matrices to consider the repeated-measures design of the study. In this model, the dependent variable was R, fixed-effects independent variables were the CT technique, rater, and interaction between the two, and random-effects variable was the individual dog. For all statistical analyses, significance was set at P < 0.05.
Results
Comparison Between ivCT and mCT
The median values of the TR, LR, and E are listed in Table 1. TR values were significantly higher on ivCT than on mCT by rater 1 (neurology resident) and rater 2 (board-certified neurologist) but not by rater 3 (diagnostic imaging resident) (Figure 2). For TR, the interaction between the CT technique and raters was statistically significant (P < 0.001), indicating that the differences in TR values measured by ivCT and mCT might vary depending on the rater.
Citation: Journal of the American Animal Hospital Association 59, 5; 10.5326/JAAHA-MS-7318
LR values were significantly higher on mCT than on ivCT (P < 0.001), independent of the raters (Figure 3). Thinning of contrast columns in continuity with the compressive lesion was reported in 32 (67%) dogs by rater 1 and 2 and in 39 (81%) dogs by rater 3.
Citation: Journal of the American Animal Hospital Association 59, 5; 10.5326/JAAHA-MS-7318
The surgical site on postoperative CT was strictly outside the recommended zone in three (6%) and five (10%) dogs for rater 1 and 2 on ivCT, respectively, and in one (2%) dog for rater 2 on mCT. Thus, ivCT reading was statistically more frequently associated with erroneous predictions of the operative site than mCT, independent of the raters (P = 0.04). The average densities of the compressive material of the eight dogs (17%) with erroneous ivCT operative site predictions were 151, 126, 88, 255, 255 162, 170, and 86 Hounsfield units. This is higher than reported spinal cord attenuation12 and density of compressive material in the correct site of extrusion (respectively, 162, 96, 86, 113, 113, 96 and 63 Hounsfield units). Compressive material in these sites of extrusion was characterized as nonmineralized by the surgeons in all cases incorrectly localized by ivCT.
In 41 (85%) dogs with correct operative site predictions (i.e., R = 1), the E values were not significantly different between ivCT and mCT (P = 0.34), independent of the rater. Similarly, no significant difference was found in extruded material lateralization and opening side between the two CT techniques. The confounding effect of the raters was identified regarding P. This means that the effect of the rater was mixed with the effect of imaging modality. We can therefore not establish a clear link between imaging modality and number of protrusions identified.
A subarachnoid diverticulum was visualized only on mCT in two dogs (not observed on ivCT). Kyphosis associated with vertebral canal stenosis was observed in two dogs. A reduction in spinal cord diameter secondary to vertebral malformation was identified in two dogs by mCT that was not seen on ivCT.
Discussion
In this study, we demonstrated differences in the conspicuity of thoracolumbar disc extrusion assessment between ivCT and mCT in French bulldogs. The length of the compression was higher in mCT than in ivCT. However, no relevant difference in spinal cord compression measured in the transverse plane was shown between the two methods. ivCT was more often associated with erroneous identification of the appropriate surgical site than mCT. However, no meaningful differences in deviation between the recommended site of hemilaminectomy and postsurgical findings were observed in-between both techniques when the correct extradural compressive site was identified. Additional findings such as subarachnoid diverticulum and kyphosis-associated reduction of spinal cord diameter were visualised in 8% of dogs.
Errors in identifying the correct location of acute disc herniation reached 17% with ivCT and 2% with mCT. Discrimination between acute and chronic compression has already been reported to be more difficult without mCT.12 In cases incorrectly localized with ivCT, the CT attenuation measurement was high at the level of the erroneously identified compression. Furthermore, compressive material in the site of extrusion was characterized as nonmineralized material by the surgeon in all cases of incorrect localization with ivCT. Mineralization or acute hemorrhage may have caused hyperattenuating extradural lesions and contributed to the errors in disc extrusion localization. This could erroneously guide the rater to falsely localize the compression site because the true acute compression could be done by material that has, at least partially, the same attenuation as the spinal cord.
Our results were similar to a previous study comparing noninjected CT and mCT and showing that some compressive lesions are identified only with mCT.13 Only one previous study had already compared ivCT and mCT for evaluation of acute disc herniation and did not find any relevant differences.7 However, the comparison was conducted between ivCT and mCT from different dogs. We recommend mCT rather than ivCT to avoid incorrect localization of disc extrusion.
The medullary compression length was significantly higher with mCT than with ivCT. Extradural material causing subtle subarachnoid contrast deviation or thinning was missed on ivCT. This could explain the high LR values with this modality compared with ivCT. Although the surgeon was provided with more accurate information about the disc extrusion, the length of the hemilaminectomy did not seem to be affected (E was not significantly different between the two techniques). This may be related to the absence of a surgical opening at minimally compressive sites.mCT can indicate spinal cord edema14 or contusion in areas of circumferential thinning to loss of subarachnoid contrast associated with an increase in the spinal cord diameter. Spinal cord edema has been negatively associated with prognosis in dogs without pain perception.15 Its visualization could influence the incision size during laminectomy because a larger surgical opening could be advised if edema is suspected. Myelomalacia, particularly frequent in French bulldogs,16 can also be identified with mCT because the subarachnoid contrast usually diffuses with the spinal cord parenchyma.16,17 No such case was present in this study, owing to the low number of grade 5 cases.
Moreover, mCT may be useful in assessing concurrent lesions. For instance, subarachnoid diverticulum and malformation-associated spinal cord compression/atrophy were identified in 8% of the dogs using mCT but not using ivCT.
Finally, cerebrospinal fluid tap is routinely performed during mCTand could provide additional preoperative prognosis criterion and identify meningitis and meningomyelitis.18 The median time of the myelography process (18 min) in our study was no longer than MRI acquisition time.
Although mCT is associated with a more accurate identification of IVDE, adverse reactions must be considered. Seizures and neurological deterioration are the most commonly reported adverse reactions.6,9,19,20 Sporadic cases of apnea, meningitis, cardiac arrhythmias, and intracranial subarachnoid hemorrhage have also been reported.19,21,22 No adverse effects were observed in this study. We emphasize that the amount of contrast media needed to achieve good visualization of the arachnoid space should be lower using CT than radiography. We also exclusively performed lumbar punctures and used nonionic contrast agents.
The limitations of our study primarily lie in the use of a surgical confirmation for diagnosis of the intervertebral disc herniation site. Moreover, the surgeon decided the length of the opening case by case, which biased the E calculation. Comparison with MRI or the use of a manual caliper, as previously described, could improve accuracy.23 Moreover, no confirmation of disc disease was conducted by histological evaluation of the removed material.
It was difficult to definitively prove that the compressive site determined by the raters from CT data was truly the site of highest clinical relevance. To limit this, we excluded from the study dogs with no clinical recovery, in which localization errors on CT and surgery would have been possible. However, the authors are aware that postoperative patient care (rest, anti-inflammatory drugs, and physiotherapy) could induce recovery despite medullary compression localization errors.
Conclusion
In conclusion, a significant increase in length of compression evaluation was found on mCT compared with ivCT in French bulldogs, but no significant differences in surgical advice were observed when the correct disc extrusion site was identified between the two imaging modalities. However, the use of ivCT was more often associated with erroneous identification of the surgical site than mCT. Incorrect localization appeared often related to the presence of hyperattenuating material (mineralized disc or acute hemorrhage) at the erroneously identified site, whereas the clinically relevant compressive lesion is caused by less easily seen material that is similar in attenuation to the spinal cord. mCT should therefore be recommended to limit erroneous localization of disc extrusion. Moreover, some other spinal thoracolumbar affliction cannot be identified without mCT, and their detection can modify the treatment or prognosis. Investigation of other breeds, especially nonchondrodystrophic breeds is required to be investigated in future prospective studies.
Schematic representation of LR—the length of compression (L) divided by the length of the L2 vertebral body—and TR—compression (D) area divided by canal (C) area in the zone of maximal compression. The compressive disc is pictured in blue and the spinal cord in pink. LR, length ratio; TR, transverse ratio.
Graph plots showing predictive margins of TR values measured on mCT and ivCT depending on the rater of the 48 French bulldogs included in the study. Error bars represent 95% CIs. TR values were found to be significantly higher on ivCT than on mCT by rater 1 and rater 2 but not by rater 3. CI, confidence interval; ivCT, IV computed tomography; mCT, myelography computed tomography; TR, transverse ratio.
Graph plots showing predictive margins of LR values measured on mCT and ivCT depending on the rater in the 48 French bulldogs included in the study. Error bars represent 95% Cis. LR values were significantly higher on mCT than on ivCT (P < 0.001), independent of the rater. CI, confidence interval; ivCT, IV computed tomography; LR, length ratio; mCT, myelography computed tomography.
Contributor Notes
