Dogs with a diagnosis of ANNPE or ischemic myelopathy often have similar clinical signs consisting of acute, focal, nonprogressive neurologic signs associated with spinal cord dysfunction. Acute noncompressive nucleus pulposus extrusion is an acute myelopathy characterized by extrusion of normally hydrated nucleus pulposus material spreading inside the vertebral canal and causing spinal cord contusion.1,2 This disk material does not usually have any degenerative changes, and it spreads into the vertebral canal, commonly without causing spinal cord compression.1 Ischemic myelopathy is an infarction of the spinal cord in response to a local lack of oxygen. Although most dogs with ischemic myelopathy do not have signs of pain that persist more than 24 hours after the initial clinical signs are identified, dogs with ANNPE can have continued signs of hyperalgesia.1–3 Despite the similarity in clinical signs, the etiopathogenesis of the 2 myelopathies differs substantially. Traumatic events have been associated with ANNPE. The most common cause of ischemic myelopathy in small animals is believed to be emboli of fibrocartilage originating from the nucleus pulposus of the intervertebral disk.4,5 Other causes of ischemic myelopathy include bacterial and parasitic emboli, neoplastic or fat emboli, thrombi, and atherosclerosis.5,6 Decompressive surgery is not usually required for treatment of ANNPE, and it is also not indicated for treatment of ischemic myelopathy. Furthermore, owing to the nonsurgical treatment methods and the generally good prognosis for patients affected by these conditions, antemortem histologic diagnosis is not usually obtained.
Because of its noninvasive nature and excellent soft tissue resolution, MRI remains the imaging modality of choice for the diagnosis of ANNPE and ischemic myelopathy.1–4 The MRI features compatible with ANNPE include evidence of reduced volume of the nucleus pulposus, focal hyperintensity within the overlying spinal cord on T2W FSE images, extraneous material or signal changes within the vertebral canal, vacuum phenomenon (presence of spontaneous gas within the joint space), and minimal spinal cord compression that does not require surgical intervention.1,2 The MRI findings compatible with ischemic myelopathy include presence of a focal, hyperintense, intramedullary lesion on T2W FSE images primarily involving the gray matter and having a variable degree of contrast enhancement.3–5
The presence of fat tissue in the epidural space may limit the conspicuity of epidural, meningeal, or intramedullary lesions because both fat and gadolinium have a high signal on T1W MRI. With chemical FS sequences, a preparation pulse is used to excite lipid protons selectively, followed by a spoiling gradient pulse that diphases the fat signal. Once the signal generated by fat has been removed, use of T1W fat suppression sequences results in modification of the gray scale, increasing the contrast ratio and the conspicuity of gadolinium-enhancing lesions.7
Considering that there is some overlap in the features that characterize each condition (eg, hyperintensity on T2W images), and because the diagnosis of each condition is heavily reliant on MRI characteristics, identifying MRI criteria to discriminate between ANNPE and ischemic myelopathy and allow consensus among radiologists would be beneficial for clinicians needing to determine the best diagnostic approach for these cases. The purpose of the study reported here was to determine the interobserver variability for specific MRI features previously described to be characteristics of ANNPE or ischemic myelopathy in dogs.1–5 We also sought to compare findings on postcontrast T1W FS MRI sequences for the 2 conditions and to determine whether length and directional patterns of intramedullary spinal cord hyperintensity on T2W FSE images differ between dogs with presumed ischemic myelopathy and those with presumed ANNPE. We hypothesized that dogs with suspected ANNPE would more commonly have meningeal and epidural fat enhancement, compared with those with suspected ischemic myelopathy. Furthermore, considering the types of spinal cord injury associated with ANNPE versus ischemic myelopathy, we anticipated that the length and pattern of hyperintensity within the spinal cord on T2W MRI images would differ between dogs with these 2 diagnoses.
Materials and Methods
Case selection criteria
Records in the electronic MRI database of the Centre Hospitalier Vétérinaire de l'Université de Montreal for dogs seen from January 1, 2005, to December 30, 2012, were screened by 1 investigator (SS) who did not participate in evaluations used to assess interobserver variability for presumptive diagnosis and for specific MRI features of ischemic myelopathy or ANNPE. To be included in the retrospective study, a patient must have met the following criteria: a history of acute onset of nonprogressive myelopathy for which the evaluation included MRI of the spinal cord and a complete medical record (including ≥ 1 follow-up examination) with findings compatible with a diagnosis of ischemic myelopathy or ANNPE.
MRI protocols
All MRI evaluations were performed with a 1.5-T MRI unita with a spine coil.b Imaging was performed for dogs under general anesthesia and positioned in dorsal recumbency. In addition to T2W FSE images (TR, 3,500 milliseconds; TE, 90 milliseconds; slice thickness, 3 mm) in dorsal, sagittal, and transverse planes, T2W SSFSE images (TR, 3,000 milliseconds; TE, 1,000 milliseconds; slice thickness, 3 mm) were acquired in sagittal plane with a 512 × 512 matrix for all dogs. Contrast-enhanced images were obtained following IV injection of gadobenate dimegluminec (0.2 mL/kg [0.09 mL/lb], once). T1-weighted FS sequences (TR, 450 milliseconds; TE, 20 milliseconds; slice thickness, 3 mm) were obtained in sagittal and transverse planes.
Image analysis
Recorded MRI sequences were evaluated by 3 board-certified veterinary radiologists (PP, IM, and PJ). The observers were employed at 3 different veterinary medical institutions and had different educational backgrounds in neuroimaging.
Each observer assessed the images with the aid of a DICOM (ie, Digital Imaging and Communications in Medicine) viewerd and completed an online questionnaire (created by 1 investigator [PP] through use of a Web-based documente) for each case. Observers were blinded to the identification, signalment, history, clinical signs, and outcomes of the patients at the time of online questionnaire completion. Observers were asked to assess several MRI features previously reported to be characteristics of ischemic myelopathy or ANNPE.1–5 These included the presence of intramedullary hyperintensity, narrowing of the corresponding intervertebral disk space, and reduction in volume and signal intensity of the nucleus pulposus on T2W FSE images as well as presence of a cleft in the dorsal part of the annulus fibrosus (previously identified on MRI of patients with this condition1), extraneous material or signal change within the epidural space at the level of the lesion, presence of vacuum phenomenon, and presence of spinal cord swelling on T2W FSE images (all dichotomized as yes-or-no answers). The SSFSE sequence was used in the evaluation of spinal cord swelling visible as focal circumferential loss of subarachnoid space signal. The observers were also asked to indicate gray or white matter involvement and symmetry of the spinal cord lesion on T2W FSE images (dichotomized as yes-or-no answers). Finally, each observer was asked to provide a presumptive diagnosis of either ischemic myelopathy or ANNPE on the basis of MRI criteria. Interobserver agreement was then determined for each variable included in the questionnaire.
Dogs for which perfect agreement was reached in regard to the presumptive diagnosis were classified into 2 groups (ischemic myelopathy and ANNPE). Following creation of the groups, sagittal T2W FSE and postcontrast transverse T1W FS sequences for dogs of both groups were evaluated by the same 3 observers as well as by a third-year diagnostic imaging resident (SS) until a consensus was reached. The postcontrast T1W FS sequences were evaluated for the presence of meningeal, epidural fat, or intramedullary lesion enhancement and the sagittal T2W FSE sequences were assessed for the directional pattern (longitudinal, oblique, or vertical) and length (< 1, 1 to 2, or > 2 vertebral body lengths) of the intramedullary hyperintensity. The length of the intramedullary hyperintensity was evaluated by measuring the length of the lesion relative to the length of the adjacent vertebra.1
Statistical analysis
Observer's answers to questions on the online questionnaire were automatically entered into a commercially available spreadsheet.f Interobserver agreement (paired comparisons) was determined for each MRI variable and for the final diagnosis included in the online questionnaire by calculation of the Cohen κ, with the degree of agreement defined as described elsewhere8 (1.00, perfect; 0.93 to 0.99, excellent; 0.81 to 0.92, very good; 0.61 to 0.8, good; 0.41 to 0.6, fair; 0.21 to 0.4, slight; 0.01 to 0.2, poor; and ≤ 0, none). When κ could not be calculated (ie, if ≥ 1 observer had no variation in observations), the percentage of disagreement between the remaining 2 observers was recorded as an indication of the level of divergence in their observations.
Next, a Fisher exact test was used to evaluate the association between the pattern of postcontrast signal enhancement and the presumptive radiologic diagnosis and between the length of intramedullary hyperintensity and presumptive radiologic diagnosis. Association between the directional pattern of the intramedullary hyperintensity and the diagnosis was evaluated with a Cochran-Mantel-Haenszel test because of the ordinal nature of the variables. Statistical analyses were performed with commercially available software.g Values of P < 0.05 were accepted as significant.
Results
Twenty dogs met the study inclusion criteria. Interobserver agreement for the presumptive diagnosis of ischemic myelopathy or ANNPE for the 3 observers was perfect (κ = 1 for all comparisons). On the basis of radiologic assessment, 6 dogs were categorized as having ischemic myelopathy and 14 were categorized as having ANNPE. The MRI features with the greatest degrees of interobserver agreement included the presence of a hyperintense intramedullary lesion (perfect agreement; κ = 1), narrowing of the corresponding intervertebral disk space (fair to good agreement; range of κ values, 0.5 to 0.7), reduction in volume and signal intensity of the nucleus pulposus (good to very good agreement; range of κ values, 0.69 to 0.88) on T2W FSE sequences, and presence of a cleft in the dorsal annulus fibrosus (fair to perfect agreement; range of κ values, 0.56 to 1) and presence of extraneous material or signal change within the epidural space at the level of the lesion (fair to good agreement; range of κ values, 0.44 to 0.76) on T2W FSE images. Agreement was poor to fair for gray versus white matter involvement (range of κ values, 0.10 to 0.44) and poor to slight for symmetry of the spinal cord lesion (range of κ values, 0.09 to 0.27) on T2W FSE images. Agreement could not be calculated for 2 of 3 pairs of reviewers for the presence of vacuum phenomena and the presence of spinal cord swelling because 1 of the 3 observers had no variation in observations for the categorical variables; in these cases, a percentage of disagreement was reported to provide an idea of the level of agreement between this reviewer and each of the other 2. Where determined, values indicated slight agreement (κ of 0.39 and 0.38, respectively). For the other 2 pairs of observers, disagreement was 35% and 20% for the presence of vacuum phenomenon and 35% and 45% for the presence of spinal cord swelling. Where determined, the values indicated slight agreement (κ of 0.39 and 0.38, respectively). For the other 2 pairs of reviewers, disagreement was found for 7 of 20 (35%) and 4 of 20 (20%) cases for the presence of vacuum phenomenon and 7 of 20 (35%) and 9 of 20 (45%) cases for the presence of spinal cord swelling.
On the basis of perfect agreement among the 3 reviewers for presumptive radiologic diagnosis, all dogs were included in the 2 groups formed for evaluation of the additional MRI criteria. Postcontrast T1W FS sequences were only available for 9 of 14 dogs included in the ANNPE group. Meningeal enhancement was present in all 9 of these dogs. Epidural fat enhancement and intramedullary lesion enhancement were detected in 5 and 4 of these 9, respectively (Figure 1). Postcontrast T1W FS sequences were available for 5 of 6 dogs in the ischemic myelopathy group. Intramedullary spinal cord enhancement was detected in 1 of these 5 dogs, whereas neither meningeal nor epidural fat enhancement was identified (Figure 2). Meningeal enhancement was significantly (P < 0.001) associated with a presumptive diagnosis of ANNPE. No significant association was found between intramedullary enhancement and a presumptive diagnosis of ischemic myelopathy or ANNPE.
Representative transverse T1W FS MRI images showing meningeal, epidural, and intramedullary enhancement following contrast medium administration in dogs enrolled in a retrospective study to assess and compare MRI features associated with a presumptive diagnosis of ANNPE or ischemic myelopathy. A—Image obtained at the level of L1 in a dog with a presumptive diagnosis of ANNPE. Asymmetric meningeal enhancement is evident. B—Meningeal and epidural enhancement is present in an image obtained at the level of L1 in a dog with presumed ANNPE. C—Notice asymmetric intramedullary, meningeal, and epidural enhancement in an image obtained at the level of T13 in a dog with presumed ANNPE.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Representative transverse T1W FS MRI images showing meningeal, epidural, and intramedullary enhancement following contrast medium administration in dogs enrolled in a retrospective study to assess and compare MRI features associated with a presumptive diagnosis of ANNPE or ischemic myelopathy. A—Image obtained at the level of L1 in a dog with a presumptive diagnosis of ANNPE. Asymmetric meningeal enhancement is evident. B—Meningeal and epidural enhancement is present in an image obtained at the level of L1 in a dog with presumed ANNPE. C—Notice asymmetric intramedullary, meningeal, and epidural enhancement in an image obtained at the level of T13 in a dog with presumed ANNPE.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Representative transverse T1W FS MRI images showing meningeal, epidural, and intramedullary enhancement following contrast medium administration in dogs enrolled in a retrospective study to assess and compare MRI features associated with a presumptive diagnosis of ANNPE or ischemic myelopathy. A—Image obtained at the level of L1 in a dog with a presumptive diagnosis of ANNPE. Asymmetric meningeal enhancement is evident. B—Meningeal and epidural enhancement is present in an image obtained at the level of L1 in a dog with presumed ANNPE. C—Notice asymmetric intramedullary, meningeal, and epidural enhancement in an image obtained at the level of T13 in a dog with presumed ANNPE.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal T2W FSE (A) and postcontrast sagittal (B) and transverse (C) T1W MRI images of a dog classified as having ischemic myelopathy. A—Notice the presence of a hyperintense intramedullary lesion, approximately 2.5 vertebral bodies in length, with a longitudinal directional pattern at the level of C4 through C6. B and C—There is no evidence of intramedullary enhancement at the level of the lesion (C4 through C6) on the postcontrast T1W FS images. The meninges have a normal MRI appearance, and epidural fat enhancement is not visible owing to swelling of the spinal cord.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal T2W FSE (A) and postcontrast sagittal (B) and transverse (C) T1W MRI images of a dog classified as having ischemic myelopathy. A—Notice the presence of a hyperintense intramedullary lesion, approximately 2.5 vertebral bodies in length, with a longitudinal directional pattern at the level of C4 through C6. B and C—There is no evidence of intramedullary enhancement at the level of the lesion (C4 through C6) on the postcontrast T1W FS images. The meninges have a normal MRI appearance, and epidural fat enhancement is not visible owing to swelling of the spinal cord.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal T2W FSE (A) and postcontrast sagittal (B) and transverse (C) T1W MRI images of a dog classified as having ischemic myelopathy. A—Notice the presence of a hyperintense intramedullary lesion, approximately 2.5 vertebral bodies in length, with a longitudinal directional pattern at the level of C4 through C6. B and C—There is no evidence of intramedullary enhancement at the level of the lesion (C4 through C6) on the postcontrast T1W FS images. The meninges have a normal MRI appearance, and epidural fat enhancement is not visible owing to swelling of the spinal cord.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Intramedullary hyperintensity on T2W FSE images was identified for 6 of 6 dogs with presumed ischemic myelopathy and 13 of 14 dogs with presumed ANNPE. In 1 dog in the ANNPE group, the lesion was located in the cauda equina, and the images were not included in the evaluation of this feature (Figure 3). The directional pattern of intramedullary hyperintensity was longitudinal in 6 of 6 dogs in the ischemic myelopathy group and in 4 of 13 dogs in the ANNPE group. For the remaining 9 dogs of the ANNPE group included in the evaluation, the pattern was oblique (n = 7) or vertical (2). In lesions described as oblique, the T2W FSE hyperintensity appeared to originate from the intervertebral disk space and was oriented in a caudoventral to a craniodorsal direction (Figure 4). There was a significant (P = 0.01) association between a nonlongitudinal directional pattern of intramedullary hyperintensity and a presumptive diagnosis of ANNPE. The length of intramedullary hyperintensity was greater for the ischemic myelopathy group, compared with that of the ANNPE group (P < 0.001); measurements were > 2 vertebral lengths for 3 of 6 dogs in this group and 1 to 2 vertebral lengths for the remaining 3, whereas measurements were 1 to 2 vertebral lengths for 2 of 13 dogs in the ANNPE group and < 1 vertebral length for the remaining 11.
Sagittal (A) and transverse (B) T2W FSE and postcontrast T1W FS (C) MRI images of a dog classified as having ANNPE. Because the lesion was at the level of the cauda equina, images from this dog were not included in evaluations related to the presence of intramedullary hyperintensity. A—Notice the hyperintense oblique line in the dorsal part of the annulus fibrosus, consistent with a cleft (arrow). The hypointense area in the center of the hydrated nucleus pulposus corresponds to a gas accumulation described as vacuum phenomenon (arrowhead). Notice the caudoventral to craniodorsal direction of the hydrated disk material, which has spread into the epidural space (asterisk). B—The herniated disk material (arrow) does not appear to be causing compression of the cauda equina. C—Strong epidural and meningeal enhancement can be seen around the conus medullaris extending to the nerve roots, consistent with radiculitis (arrows).
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal (A) and transverse (B) T2W FSE and postcontrast T1W FS (C) MRI images of a dog classified as having ANNPE. Because the lesion was at the level of the cauda equina, images from this dog were not included in evaluations related to the presence of intramedullary hyperintensity. A—Notice the hyperintense oblique line in the dorsal part of the annulus fibrosus, consistent with a cleft (arrow). The hypointense area in the center of the hydrated nucleus pulposus corresponds to a gas accumulation described as vacuum phenomenon (arrowhead). Notice the caudoventral to craniodorsal direction of the hydrated disk material, which has spread into the epidural space (asterisk). B—The herniated disk material (arrow) does not appear to be causing compression of the cauda equina. C—Strong epidural and meningeal enhancement can be seen around the conus medullaris extending to the nerve roots, consistent with radiculitis (arrows).
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal (A) and transverse (B) T2W FSE and postcontrast T1W FS (C) MRI images of a dog classified as having ANNPE. Because the lesion was at the level of the cauda equina, images from this dog were not included in evaluations related to the presence of intramedullary hyperintensity. A—Notice the hyperintense oblique line in the dorsal part of the annulus fibrosus, consistent with a cleft (arrow). The hypointense area in the center of the hydrated nucleus pulposus corresponds to a gas accumulation described as vacuum phenomenon (arrowhead). Notice the caudoventral to craniodorsal direction of the hydrated disk material, which has spread into the epidural space (asterisk). B—The herniated disk material (arrow) does not appear to be causing compression of the cauda equina. C—Strong epidural and meningeal enhancement can be seen around the conus medullaris extending to the nerve roots, consistent with radiculitis (arrows).
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal T2W FSE MRI images of the cervical spinal cord of a dog with a presumptive diagnosis of ischemic myelopathy (A) and the thoracolumbar spinal cord of a dog with a presumptive diagnosis of annpe (B). A—The hyperintense intramedullary lesion is long and ill-defined and has a longitudinal pattern with minimal signal changes at the level of the C3–4, C4–5, and C5–6 intervertebral disks. B—The lesion is short and well-defined with a caudoventral to craniodorsal oblique directional pattern originating from the affected T13-L1 intervertebral disk; the corresponding nucleus pulposus has normal signal intensity but decreased volume. Notice the difference between the T12–13 intervertebral disk, which has chronic degenerative changes with moderate ventral spondylosis, and the disk with suspected ANNPE.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal T2W FSE MRI images of the cervical spinal cord of a dog with a presumptive diagnosis of ischemic myelopathy (A) and the thoracolumbar spinal cord of a dog with a presumptive diagnosis of annpe (B). A—The hyperintense intramedullary lesion is long and ill-defined and has a longitudinal pattern with minimal signal changes at the level of the C3–4, C4–5, and C5–6 intervertebral disks. B—The lesion is short and well-defined with a caudoventral to craniodorsal oblique directional pattern originating from the affected T13-L1 intervertebral disk; the corresponding nucleus pulposus has normal signal intensity but decreased volume. Notice the difference between the T12–13 intervertebral disk, which has chronic degenerative changes with moderate ventral spondylosis, and the disk with suspected ANNPE.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Sagittal T2W FSE MRI images of the cervical spinal cord of a dog with a presumptive diagnosis of ischemic myelopathy (A) and the thoracolumbar spinal cord of a dog with a presumptive diagnosis of annpe (B). A—The hyperintense intramedullary lesion is long and ill-defined and has a longitudinal pattern with minimal signal changes at the level of the C3–4, C4–5, and C5–6 intervertebral disks. B—The lesion is short and well-defined with a caudoventral to craniodorsal oblique directional pattern originating from the affected T13-L1 intervertebral disk; the corresponding nucleus pulposus has normal signal intensity but decreased volume. Notice the difference between the T12–13 intervertebral disk, which has chronic degenerative changes with moderate ventral spondylosis, and the disk with suspected ANNPE.
Citation: Journal of the American Veterinary Medical Association 248, 9; 10.2460/javma.248.9.1013
Discussion
In the present study, we found perfect interobserver agreement (κ = 1 for all paired comparisons) among 3 trained radiologists with different educational backgrounds for the presumptive diagnosis of ANNPE versus ischemic myelopathy in dogs. These observers assessed specific high-field MRI features that were reported to characterize the 2 diseases in other publications1–5 and were summarized in an online questionnaire created by one of the authors (SS). Perfect agreement was also found regarding the presence of hyperintense intramedullary lesions (identified in 6/6 dogs with presumed ischemic myelopathy and 13/13 dogs with presumed ANNPE in regions cranial to the cauda equina, consistent with previous reports1–5 indicating this feature as an important MRI finding in both conditions). There was fair to good agreement among observers regarding the presence or absence of narrowing of the intervertebral disk space adjacent to the intramedullary lesion and good to very good agreement for reduced volume and signal intensity of the corresponding nucleus pulposus for both conditions. These features can also be found in dogs with chronic degenerative changes of an intervertebral disk.1 In dogs with clinical signs consistent with ischemic myelopathy or ANNPE and with multifocal chronic degenerative disk changes observed on MRI, identification of a hyperintense intramedullary lesion on T2W images is important to guide the observer toward identification of the lesion responsible for the current clinical signs. Once a hyperintense lesion is found, the observer can use other previously described MRI criteria1–5 to aid in discrimination between presumptive diagnoses of ANNPE and ischemic myelopathy.
In the present study, interobserver agreement for the presence or absence of a cleft in the annulus fibrosus ranged from fair to perfect. A cleft in the annulus fibrosus as seen on MRI has been previously described as a communicating, hyperintense tract passing through the dorsal annulus between the residual nucleus pulposus and the vertebral canal.1 This finding has been reported for dogs with ANNPE but not for those with ischemic myelopathy3–5; therefore, this may be a useful criterion to discriminate between the 2 conditions with high-field MRI. Fair to good interobserver agreement was found for the presence of extraneous material or signal change in the epidural space at the level of the lesion. This material may have represented hydrated disk material, hemorrhagic fluid, or both.2 Detection of a hypointense signal on gradient recall echo sequences has been previously reported for dogs with presumed ANNPE, but no significant association was found between this variable and outcome.2 Although gradient recall echo sequences were not evaluated for evidence of hemorrhage in the present study, future investigations including this type of evaluation may lead to a better understanding of its role in discriminating between presumptive diagnoses of ischemic myelopathy and ANNPE and whether findings are associated with patient outcome. Identification of extraneous material in the epidural space is an important finding, considering that no epidural changes are usually visible in cases of ischemic myelopathy.1,3–5
The interobserver agreement for the involvement of the gray or white matter was poor to fair, and that for symmetry of the intramedullary lesion was poor to slight. In a study4 of 11 dogs with presumed ischemic myelopathy, a hyperintense intramedullary lesion was commonly found to include predominantly gray matter, whereas in another study,1 dogs with presumed ANNPE had both white and gray matter affected. In the authors' experience, high-field MRI offers good contrast resolution, but the limited spatial resolution and artifacts such as slice misregistration can sometimes limit evaluation of small structures, especially in small patients with a low signal-to-noise ratio. We believe that those factors could have negatively influenced interobserver agreement for the involvement of the gray or white matter in this study.
Interobserver agreement regarding the presence of a vacuum phenomenon in dogs of our study was slight for 1 pair of observers, and disagreement was 35% and 20% for the other 2 pairs of observers. Vacuum phenomenon is an accumulation of gas within a synovial joint of the appendicular skeleton or an intervertebral disk space; analysis of such a gas sample from 1 human patient revealed that it contained a mixture of 90% to 92% nitrogen along with oxygen, carbon dioxide, and other gas traces.9 In a recent study,10 vacuum phenomenon was a frequent finding on CT of dogs with Hansen type I disk herniation, and the authors concluded that its recognition could be helpful in identifying a degenerated disk. In our patients, vacuum phenomenon was also occasionally found in nondegenerated disks corresponding to the site of spinal cord lesions, supporting the hypothesis suggested by others1 that this may be attributable to rapid voiding of the nucleus pulposus rather than degeneration, with gas replacing the extruded material.
The agreement for the presence of spinal cord swelling was slight for 1 pair of observers, and disagreement was found for 7 of 20 (35%) and 9 of 20 (45%) dogs for the other 2 pairs of observers. Both SSFSE and standard T2W FSE sequences were available to the observers. The SSFSE sequence is helpful in screening compressive lesions of the spinal cord.11 This sequence depicts pure fluid such as CSF within the subarachnoid space and lesion.11 The combination of a focal increase in size of the spinal cord and circumferential loss of CSF signal on the sagittal SSFSE sequence is an MRI finding indicative of spinal cord swelling. In patients with mild spinal cord swelling, obvious loss of the CSF signal is not evident on SSFSE sequences, and the subjectivity of this evaluation is a potential cause of low levels of agreement between the observers.
In assessment of postcontrast T1W FS sequences for 9 of 14 dogs with a presumptive MRI diagnosis of ANNPE, all 9 had meningeal contrast enhancement with epidural fat enhancement, intramedullary spinal cord enhancement, or some combination of these. Only 1 of 5 dogs with a presumptive radiologic diagnosis of ischemic myelopathy and images available for analysis had postcontrast intramedullary enhancement of the spinal cord, and none had enhancement of the other 2 features. To our knowledge, the present study was the first in which postcontrast T1W FS sequences have been used to evaluate both conditions. In a previous investigation regarding MRI findings in dogs with a presumptive diagnosis of ANNPE, contrast enhancement of the lesion on standard T1W images was detected in a low number of patients.2 Both normal epidural fat and gadolinium-based contrast medium have high signal intensity (hyperintensity) on standard T1W images. This makes it difficult to differentiate meningeal or epidural contrast enhancement from normal epidural fat. An FS technique can be used to suppress the signal for epidural fat. The use of postcontrast T1W FS sequences allows a viewer to better differentiate meningeal or epidural contrast enhancement (which appears hyperintense) from normal epidural fat (which shows no signal with this technique).6 We speculate that the use of standard T1W sequences could have influenced the detection of meningeal and epidural fat enhancement in previous studies,1–3 potentially resulting in some false-negative results. In dogs with a presumptive diagnosis of ANNPE, focal meningeal and epidural fat enhancement on MRI may represent inflammation secondary to trauma caused by the high-velocity extrusion of hydrated disk material and revascularization of the disk fragments within the epidural space. This is in agreement with findings of previous studies12,13 in which revascularization of extradural disk material was reported in dogs. Although the blood-brain barrier can break down within 6 hours after an ischemic event, gadolinium-based contrast medium cannot reach the lesion at this early stage owing to the lack of blood supply.14 Considering that revascularization of tissues typically takes 5 to 7 days,14 the absence of contrast enhancement in 4 of 5 dogs with a presumptive diagnosis of ischemic myelopathy in our study could have been related to a short interval of time between the onset of clinical signs and MRI, although this variable was not investigated. In patients with a presumptive diagnosis of ANNPE, contrast enhancement predominantly involves the meninges and epidural fat, which are on the outside of the blood-brain barrier. Therefore, enhancement can be achieved earlier for these patients than for those with intramedullary lesions. In dogs with a presumptive diagnosis of ischemic myelopathy, the meninges and epidural fat are not primarily involved, and the time needed to achieve contrast enhancement may depend on revascularization and the degree of damage at the level of the blood-brain barrier. We consider it likely that contrast enhancement would not typically be seen by use of MRI in the acute phase of this disease.
In the present study, 6 of 6 dogs with a presumptive diagnosis of ischemic myelopathy had intramedullary hyperintensity on T2W images, and the hyperintensity had a longitudinal directional pattern in all 6. Dogs with a presumptive diagnosis of ANNPE (13/13 assessed) also had a region of intramedullary hyperintensity, with the directional pattern oblique or vertical in 9 and longitudinal in only 4. One possible explanation for the differences in orientation of the hyperintense signal may be related to the pathophysiologic differences between the 2 conditions. In ischemic myelopathy, the intramedullary hyperintensity observed on MRI is thought to result from myelomalacia due to a regional decrease in blood supply, leading to cytotoxic or vasogenic edema, myelomalacia, or gliosis.3–5 Lesions associated with these processes may vary in size depending on the degree of extension of the vascular occlusion. In patients with ANNPE, however, the hyperintensity observed on MRI may represent spinal cord edema following contusion resulting from the impact from the herniated disk material. This hyperintensity may also represent the presence of hydrated intramedullary disk material or fluid from hemmorhage.1,15–17 For most dogs with a presumptive diagnosis of ANNPE, the intramedullary hyperintensity, when present, had a caudoventral to craniodorsal directional pattern observed on sagittal T2W FSE images (7/13 dogs). We speculated that this observation might be related to the biomechanics of the vertebral column in dogs and to the anatomic distribution of the connective fibers forming the dorsal part of the annulus fibrosus of the intervertebral disk. It is known that the dorsal part of the annulus fibrosus is thinner than the ventral part,18 predisposing to dorsal rather than ventral herniation of the intervertebral disk. It has also been reported that the direction of progressive herniation of the nucleus pulposus is influenced by the orientation of the bending axis in another quadruped species (swine).19 It is possible that a combination of the bending axis and the alignment of connective fibers of the dorsal annulus fibrosus in dogs predisposed these patients to cranial rather than caudal extrusion of hydrated disk material.
The proportions of dogs with a presumptive diagnosis of ischemic myelopathy and a region of intramedullary hyperintensity ≥ 1 or 2 vertebral bodies in length on T2W FSE images were significantly greater than those of dogs with a presumptive diagnosis of ANNPE, supporting our hypothesis that the length of T2W hyperintensity in the spinal cord on MRI would differ between dogs with these 2 conditions. In dogs with ischemic myelopathy, the length of the hyperintense region was shown to be significantly correlated with the severity of clinical signs in another study.3 The discrepancy in length of the hyperintense signal between the 2 conditions is likely attributable to differences in etiopathogenesis, considering that ANNPE lesions are thought to be caused by focal traumatic insults, whereas ischemic myelopathy lesions are attributed to regional vascular compromise.
The absence of histologic confirmation of spinal cord lesions was a substantial limitation of the present study. This was attributable to the fact that treatment for ANNPE or ischemic myelopathy generally consists of physical therapy and nursing care.2–5 Also, the prognosis for dogs with either condition depends on the extent of the injury,5 and the reported recovery rates are variable, ranging from 28 of 42 (67%) for patients with ANNPE2 to 42 of 50 (84%) for those with ischemic myelopathy.3 Thus, tissue samples are rarely available for investigation. In general, the diagnosis of these conditions is presumptive and made on the basis of clinical signs, MRI findings, and outcome.1–5 In the present study, MRI findings were used to categorize dogs (with medical histories consistent with these conditions) as having ischemic myelopathy or ANNPE, and the categories only included dogs for which all observers were in agreement in regard to the presumptive diagnosis in an effort to assure correct designations. Results of this study must be carefully interpreted given the low number of dogs included. The retrospective nature of the study limited the MRI sequences available, and postcontrast T1W FS sequences were acquired for only a limited number of patients. Furthermore, investigators were asked to choose between 2 final diagnoses, which may have added a bias; use of an open-ended answer field instead of a dichotomized choice would have potentially decreased the agreement between the observers.
Overall, the present study found fair to perfect interobserver agreement (κ, 0.44 to 1.0) among 3 radiologists for several of the high-field MRI variables assessed in our online questionnaire to discriminate ischemic myelopathy from ANNPE in dogs, with perfect agreement (κ of 1 for all paired comparisons) on the final diagnosis. To our knowledge, this was the first study in which T1W FS sequences were investigated as an additional means to distinguish between the presumptive diagnoses of ANNPE and ischemic myelopathy, and results suggested that, in addition to the length and directional pattern of intramedullary hyperintensity of the spinal cord on T2W FSE images, the finding of meningeal and epidural fat enhancement on T1W FS sequences could potentially be a useful tool to help identify dogs with ANNPE.
Acknowledgments
No third-party funding or support was received in connection with this study or the writing or publication of the manuscript. The authors declare that there were no conflicts of interest.
ABBREVIATIONS
| ANNPE | Acute noncompressive nucleus pulposus extrusion |
| FS | Fat saturation |
| FSE | Fast spin echo |
| SSFSE | Single-shot fast spin echo |
| TIW | T1-weighted |
| T2W | T2-weighted |
| TE | Echo time |
| TR | Repetition time |
Footnotes
GE Signa Echospeed HDx, General Electric Healthcare, Mississauga, ON, Canada.
8 CH CTL Spine Coil, USA Instruments Inc, Aurora, Ohio.
Multihance, Bracco Diagnostics Inc, Vaughan, ON, Canada.
OsiriX, Pixmeo, Geneva, Switzerland.
Google Documents, Google Inc, Mountain View, Calif.
Microsoft Excel 2010, Microsoft, Redmond, Wash.
SAS, version 9.4, SAS Institute Inc, Cary, NC.
References
1. Chang Y, Dennis R, Platt SR, et al. Magnetic resonance imaging of traumatic intervertebral disc extrusion in dogs. Vet Rec 2007; 160:795–799.
2. De Risio L, Adams V, Dennis R, et al. Association of clinical and magnetic resonance imaging findings with outcome in dogs with presumptive acute noncompressive nucleus pulposus extrusion: 42 cases (2000–2007). /Am Vet Med Assoc 2009; 234:495–504.
3. De Risio L, Adams V, Dennis R, et al. Magnetic resonance imaging findings and clinical associations in 52 dogs with suspected ischemic myelopathy. J Vet Intern Med 2007; 21:1290–1298.
4. Abramson CJ, Garosi L, Platt SR, et al. Magnetic resonance imaging appearance of suspected ischemic myelopathy in dogs. Vet Radiol Ultrasound 2005; 46:225–229.
5. De Risio L, Platt S. Fibrocartilaginous embolic myelopathy in small animals. Vet Clin North Am Small Anim Pract 2010; 40:859–869.
6. Cauzinille L. Fibrocartilaginous embolism in dogs. Vet Clin North Am Small Anim Pract 2000; 30:155–167.
7. D'Anjou MA, Carmel EN, Tidwell AS. Value of fat suppression in gadolinium-enhanced magnetic resonance neuroimaging. Vet Radiol Ultrasound 2011; 52:S85–S90.
8. Byrt T. How good is that agreement? Epidemiology 1996; 7:561.
9. Ford LT, Gilula LA, Murphy WA, et al. Analysis of gas in vacuum lumbar disc. AJR Am J Roentgenol 1977; 128:1056–1057.
10. Müller MK, Ludewig E, Oechtering G, et al. The vacuum phenomenon in intervertebral disc disease of dogs on computed tomography images. J Small Anim Pract 2013; 54:253–257.
11. Seiler GS, Robertson ID, Mai W, et al. Usefulness of a half-fourier acquisition single-shot turbo spin-echo pulse sequence in identifying arachnoid diverticula in dogs. Vet Radiol Ultrasound 2012; 53:157–161.
12. An HS, Nguyen C, Haughton VM, et al. Gadolinium-enhancement characteristics of magnetic resonance imaging in distinguishing herniated intervertebral disc versus scar in dogs. Spine 1994; 19:2089–2095.
13. Suran JN, Durham A, Mai W, et al. Contrast enhancement of extradural compressive material on magnetic resonance imaging. Vet Radiol Ultrasound 2011; 52:10–16.
14. Simonson TM, Yuh WT. Stroke and cerebral ischemia. In: Edelman RR, Zlatkin MB, Hesselink JR, eds. MRI clinical magnetic resonance imaging. Vol. 1. 2nd ed. Philadelphia: WB Saunders Co, 1996;767–786.
15. McConnell JF, Garosi LS. Intramedullary intervertebral disk extrusion in a cat. Vet Radiol Ultrasound 2004; 45:327–330.
16. Kent M, Holmes S, Cohen E, et al. Imaging diagnosis—CT myelography in a dog with intramedullary intervertebral disc herniation. Vet Radiol Ultrasound 2011; 52:185–187.
17. Sanders SG, Bagley RS, Patrick RG. Intramedullary spinal cord damage associated with intervertebral disk material in a dog. J Am Vet Med Assoc 2002; 221:1594–1596.
18. Evans H, de Lahunta A. The Skeleton. In: Evans H, de Lahunta A, eds. Miller's anatomy of the dog. 4th ed. St Louis: Elsevier, 2013;80–184.
19. Aultman CD, Scannel J, McGill SM. The direction of progressive herniation in porcine spine motion segments is influenced by the orientation of the bending axis. Clin Biomech (Bristol, Avon) 2005; 20:126–129.
