Thoracolumbar spinal cord injury in dogs is often caused by IVDH1,2 and is particularly common in chondrodystrophic dogs because of accelerated intervertebral disk degeneration relative to that in other breeds.3–5 The resultant injury to the spinal cord almost always involves both an impact injury, conventionally referred to as contusion, and a persistent deformation, referred to as compression.2 Contusive injury to the spinal cord has been extensively evaluated in laboratory experiments and leads to a chain of events causing tissue destruction, the extent of which depends on the severity of the primary event.6,7
Treatment for dogs with IVDH varies with the severity of clinical signs.1 Intensive efforts over the past several decades have failed to identify an effective treatment for contusive spinal cord injury beyond maintenance of physiologic stability (eg, blood flow and oxygenation).8 In contrast, persistent compression of the spinal cord can clearly be alleviated by surgery. Routinely, dogs that become unable to walk independently are considered as potential surgical candidates and, when spinal cord compression can be detected through imaging, will undergo decompressive surgery.2
The prognosis for surgically treated dogs with thoracolumbar IVDH varies and appears to be particularly predicted by the presence or absence of deep pain perception in the pelvic limbs.9,10 Although the subject is controversial,11 testing for deep pain perception is a routine component of neurologic examinations involving small animals with spinal cord injury. Deep pain perception is deemed to be intact when a behavioral response consistent with that of a conscious perception of pain is observed during application of a severely noxious stimulus.
The prognosis for dogs that retain deep pain perception is generally excellent; almost 100% of dogs with thoracolumbar IVDH will recover the ability to walk independently.10,12,13 However, the recovery rate for dogs that lack deep pain perception is far worse. Retrospective studies10,14–17 have revealed an overall recovery rate of approximately 50%, although rates differ greatly among reports, which may reflect differences in methods of testing.1 Investigations into alternative methods to estimate prognosis after severe spinal cord injury in dogs have included evoked potential testing18,19 and MRI measurements.20–22 However, neither of those methods alone provides prognostic accuracy superior to that of testing for deep pain perception. Measurement of various biomarker concentrations in serum or CSF23–26 holds considerable promise for future prognostication but is insufficiently sensitive and specific at present.
Information is lacking to help clinicians predict whether a dog without deep pain perception will recover its ability to walk again, making it challenging to provide owners with a prognosis. Results of experimental research27 involving combined contusive and compressive lesions in rodents suggest that greater magnitude of contusive injury, more severe compression, and longer delay between onset and removal of compression reduce the extent of recovery. Similar relationships could be expected in dogs with a spinal cord injury involving contusive-compressive lesions as a result of IVDH.
Magnitude of mechanical injury cannot be measured in clinically affected dogs and will vary among individuals, depending on the mass, velocity, and physical character of the herniated material. However, the other factors can be estimated from owners' reports and from measurements made on preoperative diagnostic images. Indeed, some evidence from retrospective veterinary clinical research suggests that a longer (vs shorter) duration of absence of deep pain perception9,14,15,28 or more rapid (vs less rapid) onset of paraplegia15 is associated with poorer outcomes. Although it has been suggested that decompressive surgery is unlikely to improve the chance of recovery when the delay between onset of loss of deep pain perception and surgical intervention is > 48 hours,29 evidence also exists to contradict this opinion.15,16 Overall, considerable discrepancies exist among reported recovery rates, probably because those rates largely originated from retrospective case series. Such data often contain many sources of bias, and many analyses have involved small sample sizes, which yield a higher risk of false discovery than larger sample sizes.30
The purpose of the study reported here was to investigate relationships between recovery of independent (unassisted) locomotion and selected, readily assessed putative prognostic indicators in dogs with loss of deep pain perception in the pelvic limbs after sudden-onset IVDH and to identify factors associated with recovery of locomotion within a 3-month follow-up period. Our hypothesis was that recovery would be more likely for dogs with less severe and less prolonged spinal cord compression.
Materials and Methods
Dogs
Three university veterinary clinical centers (Iowa State University, Texas A&M University, and the University of Bristol) participated in collecting information on dogs for this prospective cohort study. The study was initiated in November 2011 and was planned to continue until a sufficient number of dogs had been enrolled to allow for multivariate analysis, which occurred in February 2014. One author (NG) collected data on dogs evaluated at a private veterinary referral center (Oncovet, Lille, France) from November 2011 to August 2012 and at the University of Bristol from September 2012 until the end of the collection period.
Dogs evaluated at any of the participating clinics were considered for inclusion in the study when they had paraplegia and absence of deep pain perception, as defined by a lack of behavior consistent with conscious perception of pain (eg, crying, whining, or attempting to bite) when noxious pressure was applied with pliers to the digits, metatarsals, and distal aspect of the tibia of both pelvic limbs and the tail. Dogs were excluded from the study when the cause of paraplegia was a lesion other than extrusive IVDH, the disk extrusion had occurred outside the T10-L3 region of the spinal cord, or the dogs had concurrent neurologic or other medical conditions that might have interfered with collection or interpretation of data (eg, incomplete recovery after a previous spinal lesion).
Putative prognostic factors
History taking and physical and neurologic examination were conducted routinely. If absence of deep pain perception was identified, a specific study record was completed for each dog. In clinical practice, it is not possible to determine the precise timing of loss of deep pain sensation (because dog owners do not make this observation); therefore, 2 intervals representing important periods after spinal cord injury were measured: the interval between the point when clinical signs related to spinal cord disease were first identified by the owner and the point when the dog became unable to walk (duration) and the interval between the point when the dog was first observed to be unable to walk and the point when it was initially evaluated at the participating clinic (delay).
Specific items recorded at the initial referral evaluation included dog signalment (age, breed, and sex), interval between first identification of clinical signs and loss of an ability to walk (loss of independent locomotion), interval between onset of loss of independent locomotion and initial referral evaluation, and whether corticosteroid drugs had been administered during the episode of spinal disease for which the dog was being evaluated. Corticosteroid drug administration was included as a putative prognostic factor because that variable was associated with a poorer prognosis than that achieved with no such administration in a previous study31 involving conservative management (but not one involving surgical treatment32) of affected dogs. Dosage, route, and timing of corticosteroid drug administration by referring veterinarians were highly variable, and no attempt was made to subclassify this intervention. No corticosteroid drugs were administered after the initial referral consultation.
Clinical findings were discussed with owners, including the implication that absence (vs presence) of deep pain perception is associated with a poorer prognosis, and each owner made a decision regarding whether to continue with imaging and possible surgical intervention. After CT or MRI imaging was performed (choice made on the basis of clinician preference or equipment availability), additional entries were made in the study record regarding the craniocaudal and circumferential location of the herniated disk material. Dogs were subsequently treated in accordance with the recommendations of the attending clinician and wishes of the owners. When the target number of dogs had been enrolled, MRI or CT scans of each dog were examined independently by 2 observers and measurements made to define the maximum severity of spinal cord compression.
Outcome
The primary outcome assessed was the ability of each dog to walk independently (ie, recovery of independent locomotion) within 3 months after spinal decompression surgery. Independent locomotion was deemed to exist when a dog could walk unaided for 10 consecutive steps on the pelvic limbs, without falling to 1 side sufficiently for the lateral aspect of any part of the limb other than the digits to touch the ground surface. Follow-up information for dogs that had failed to recover by the point of clinic discharge was collected through regular (approx twice-weekly) reexaminations or telephone calls to the owners and referring veterinarians for at least 3 months after spinal decompression surgery or until dogs had been euthanized.
Diagnostic imaging and severity of spinal cord compression
Computed tomographic images were acquired in the transverse plane, and MRI scans were acquired in dorsal, sagittal, and transverse planes. Observers blinded to outcome status used dorsal and sagittal (or reconstructed images for CT) views to identify the site of disk herniation and transverse T2-weighted MRI or bone-window CT images to determine the severity of spinal cord compression. Transverse images were captured at the point where the spinal cord appeared maximally compressed and exactly 1 spinal cord segment cranial to this site where the spinal cord appeared uncompressed, guided by the orientation lines on the sagittal (or reconstructed) images. In situations in which the herniated disk material extended to affect the contour of the adjacent cranial spinal cord segment, the next most cranial segment was used to measure the expected uncompressed cord area. Slices representing sites of maximal compression and no compression were exported to image analysis softwarea,b to determine the ratio of compressed to uncompressed areas for each dog (therefore smaller values indicated more severe compression). The entire process, from identifying sites of maximal compression and no compression to deriving the compression ratio, was repeated in the same manner by a second observer. Area ratios were then averaged between the 2 observers for subsequent statistical analysis. In situations in which interobserver disagreement was > 20%, a third observer performed the same analysis and the mean of all 3 values was used for statistical analysis.
Treatment
On arrival at the participating referral institution, each dog underwent neurologic examination followed in most situations by imaging and decompressive surgery. Time of arrival was used to calculate the delay between onset of loss of independent locomotion and initial referral evaluation, and, when performed, surgery occurred as rapidly as possible after arrival. Choice of decompressive surgical technique was made on the basis of the location of the compressive material and preference of the attending surgeon, but included hemilaminectomy, pediculectomy, and partial corpectomy. Postoperative care consisted of opioid analgesics, manual bladder expression, and physical therapy tailored to each dog's requirements.
Statistical analysis
Overall, the study was designed to investigate associations between recovery of locomotion (recorded as a dichotomous outcome [yes vs no]) at 3 months and 5 putative prognostic factors: interval between initial clinical signs and onset of loss of independent locomotion (duration), interval between loss of independent locomotion and initial referral evaluation (delay), severity of spinal cord compression (recorded as a ratio), corticosteroid drug administration (recorded as a dichotomous variable [yes vs no]), and dog age.
The target study sample size was 80; however, determination of appropriate sample sizes for observational studies such as the present study is not straight-forward.33 Our pragmatic approach to sample size was to use the rule of 10 (ie, 10 is often considered a minimum sample size for analysis by logistic regression).34 We considered it unlikely that all 5 factors would be included in the final statistical model; therefore, the plan was to enroll a sufficient number of dogs to yield 40 dogs with recovery of locomotion. On the basis of findings of previous studies,12–17 we estimated that approximately 80 dogs would be necessary to achieve this goal. Dogs were enrolled until this total was attained.
All analyses were performed with statistical software.c Initially, each of the putative prognostic factors, which were included as continuous data when recorded as such, was examined for an association with recovery of independent locomotion by means of univariate logistic regression. Then, all 5 factors were included in a multivariable model to examine their relationships with recovery of locomotion within the 3-month period, thus allowing for possible interactions among factors. Subsequently, this model was reexamined by both stepwise inclusion and backward elimination of factors in accordance with their probability of a significant association with the outcome; each sequential model was tested for a significant difference from the prior model by means of the likelihood ratio test, with the aim of identifying the most parsimonious model that included all significant associations with outcome.
Subsequent analysis with the prognostic factors restructured into categories was used to check for the possibility of nonlinear relationships, reduce the possible effects of inaccuracy in data recording (through owner misjudgment or inability to determine exact timings), and allow comparison of results with those obtained in previous studies.9,14,15,27–29 For compression ratio, the categories were 0 to 0.50, 0.50 to 0.65, 0.65 to 0.80, and 0.80 to 1.00 (to correspond to categories of compression experimentally applied in another study27). For duration and delay, the categories were 0 to 12, 12 to 24, 24 to 48, and > 48 hours. For all analyses, ORs and 95% CIs were calculated, and tests of trend for odds were performed to determine whether linear trends existed between series of recorded ORs. Values of P < 0.05 were considered significant.
Results
Dogs
Data from 93 dogs were collected and examined for eligibility (Figure 1).35 Seven dogs were excluded from the study because they did not have imaging performed; of these, 4 were immediately euthanized at the owners' request and 3 were treated conservatively. All 3 conservatively treated dogs were later euthanized because of signs of myelomalacia (ie, cutoff for cutaneous trunci reflex became progressively more cranial and trunk muscle flaccidity was evident) between 3 and 5 days after imaging. An additional 5 dogs with no evidence of myelomalacia were euthanized at the owners' request at various stages between 0 and 28 days after imaging, and data pertaining to those dogs were excluded from logistic regression analyses (because they did not fairly represent the potential of the study population to recover). One dog died within 24 hours after surgery from suspected pulmonary thromboembolism (acute respiratory distress and rapid death), and data pertaining to it were also excluded from statistical analysis because although this complication might have been a risk of surgery, it was unlikely to have been related to the underlying spinal cord injury. Data pertaining to an additional 2 dogs were excluded from analysis because those dogs were lost to follow-up after clinic discharge (at 5 days after surgery).
Data pertaining to all remaining dogs (n = 78) were included in the analysis. Forty-four dogs were Dachshunds, and the remaining 34 primarily included various small- to medium-sized purebred dogs or dogs of mixed breeding. Estimates of the severity of spinal cord compression (compression ratio) for 23 (29%) of the included dogs differed between the initial 2 observers by > 20%, and discrepancies were resolved by a third blinded observer.
Outcome
Forty-five (58%) dogs had recovery of independent locomotion within the 3-month follow-up period (Table 1). Median interval to recovery for all dogs was 49 days; dogs that recovered within 3 months did so within a median of 27 days. Of the 33 dogs that did not recover, 10 (13% of all dogs) developed signs consistent with progressive (ie, ascending or descending) myelomalacia at a postoperative interval of between 2 and 8 days and were euthanized.
Characteristics of dogs with loss of deep pain perception in the pelvic limbs following IVDH that recovered (n = 45) or did not recover (33) independent locomotion within 3 months after spinal cord decompression surgery.
Factor | Recovered | Did not recover |
---|---|---|
Female | 23 (51) | 18 (55) |
Age (y) | 5.5 ± 2.2 | 5.2 ± 1.9 |
Received corticosteroid drugs | 15 (33) | 8 (24) |
Duration of clinical signs (h) | 47.5 ± 69.2 | 32.4 ± 49.8 |
Delay between onset of loss of independent locomotion and initial referral evaluation (h) | 25.1 ± 23.0 | 23.6 ± 22.9 |
Spinal cord compression ratio* | 0.443 ± 0.188 | 0.529 ± 0.220 |
Values are number (%) for sex and receipt of corticosteroid drugs and mean ± SD for all other factors.
Compression ratio (ratio of compressed to uncompressed area) was calculated from transverse MRI or CT images captured at the point where the spinal cord appeared maximally compressed and exactly 1 spinal cord segment cranial to this site where the spinal cord appeared uncompressed. The compression ratio was unobtainable (ie, images were not retrievable) for 2 dogs, both of which recovered.
Association of putative prognostic factors with outcome
Univariate logistic regression revealed that none of the factors hypothesized to be associated with recovery of locomotion were significantly associated with prognosis (Table 2). This lack of effect was confirmed through inclusion of all factors in a multivariate analysis, which again revealed that none had a significant association. In additional exploratory analysis, the interaction between spinal cord compression ratio and delay between onset of loss of independent locomotion and initial referral evaluation was also examined and, again, no significant effect was discerned (OR, 1.007; 95% CI, 0.971 to 1.045). When stepwise elimination of the various prognostic factors was used, no significant change was identified in model fit. The last remaining factor (ie, the factor with the lowest P value) was spinal cord compression ratio, and the association of this factor with outcome was subsequently examined in more detail.
Results of univariate and multivariate analyses to examine relationships between putative prognostic factors and recovery of locomotion in 78 dogs with loss of deep pain perception in the pelvic limbs following IVDH.
Factor | OR | SE | ||
---|---|---|---|---|
Univariate analysis | ||||
Age (y) | 1.130 | 0.137 | ||
Compression ratio* | 0.149 | 0.180 | ||
Duration of clinical signs (h) | 1.005 | 0.004 | ||
Delay between onset of loss of independent locomotion and initial referral evaluation (h) | 1.007 | 0.011 | ||
Corticosteroid drugs received | 1.334 | 0.670 | ||
Multivariate analysis | ||||
Age (y) | 1.051 | 0.141 | ||
Compression ratio* | 0.188 | 0.235 | ||
Duration of clinical signs (h) | 1.004 | 0.005 | ||
Delay between onset of loss of independent locomotion and initial referral evaluation (h) | 1.003 | 0.012 | ||
Corticosteroid drugs received | 1.104 | 0.625 |
See Table 1 for key.
Additional univariate analysis involving categorization of the continuous variables duration, compression ratio, and delay resulted in odds of recovery that differed among categories for each variable (Table 3). However, the associated 95% CIs were wide, meaning that there was insufficient evidence to conclude that any of these factors was prognostically important. Specifically, no evidence was identified to support the notion that prognosis for recovery of locomotion was poorer when the delay was greater between onset of loss of independent locomotion and initial evaluation at the participating referral center. However, the power of this analysis was low because few dogs were included in the categories representing the greatest delay.
Results of univariate analysis to examine relationships between putative prognostic factors and recovery of independent locomotion in dogs with loss of deep pain perception in the pelvic limbs following IVDH (n = 78), after reclassification of selected continuous factors as categorical variables.
Factor | No. (%) that recovered | No. (%) that did not recover | ||
---|---|---|---|---|
Compression ratio* | ||||
0–0.50 | 20 (47) | 14 (42) | ||
0.50–0.65 | 18 (42) | 10 (30) | ||
0.65–0.80 | 4 (9) | 6 (18) | ||
0.80–1.00 | 1 (2) | 3 (9) | ||
Duration of clinical signs (h) | ||||
0–12 | 20 (44) | 15 (45) | ||
12–24 | 8 (18) | 12 (36) | ||
24–48 | 6 (13) | 2 (6) | ||
> 48 | 11 (24) | 4 (12) | ||
Delay between onset of loss of independent locomotion and initial referral evaluation (h) | ||||
0–12 | 18 (40) | 16 (48) | ||
12–24 | 14 (31) | 11 (33) | ||
24–48 | 9 (20) | 3 (9) | ||
> 48 | 4 (9) | 3 (9) |
For percentage calculations, the denominator is the total number of dogs that did (43 dogs for compression ratio and 45 dogs for the other 2 factors) or did not (33 dogs) recover for which data were available.
— = Not applicable.
See Table 1 for remainder of key.
Discussion
The exploratory study reported here was designed to investigate whether recovery of locomotion by dogs with absence of deep pain perception in the pelvic limbs following IVDH would be associated with prognostic factors that could readily be determined in the clinic. Most importantly, we found no evidence to support the notion that the 2 factors over which veterinarians presumably have the most control (delays between onset of paraplegia and spinal decompression surgery or corticosteroid drug administration) had an effect on prognosis within the 3-month follow-up period. Most surprisingly, no evidence was obtained to suggest a poorer prognosis for dogs that had absence of deep pain perception for > 48 hours after onset of paraplegia than the prognosis for dogs in which it had been absent for < 48 hours.
The lack of associations identified in the present study conflicted with other reported findings from retrospective studies.9,14,15,28 In studies in which little or no evidence of an association is detected, the possibility typically exists that inadequate power was available to detect a true association. However, results of the present study could be considered more reliable than those of the other studies9,14,15,28 because data were collected prospectively on sequentially evaluated dogs, more dogs were included than in the other studies, and almost complete follow-up information was available for each. Indeed, the present study involved the largest sample of dogs with loss of deep pain sensation following IVDH to be analyzed for prognostic factors. Furthermore, the point estimates of effect obtained with both univariate and multivariate analyses in the present study were centered on the null value (ie, OR = 1) and many had reasonably narrow 95% CIs, suggesting that increasing the number of dogs would have been unlikely to have yielded considerably different results.
Results of corollary analyses yielded little reason to believe that interactions between severity of spinal cord compression and either of the temporal factors (duration of clinical signs before onset of paraplegia and delay between onset of clinical signs and initial referral evaluation) would affect prognosis, although those analyses involved a smaller number of dogs and therefore had less power to detect associations. Provision of sufficiently precise 95% CIs to completely rule out significant interactions between factors would have required data from many hundreds of dogs, necessitating collaboration between many neurosurgical centers over many years.
The present findings were surprising, considering that evidence from experimentally induced disease in small groups of rodents (42 rodents separated into 5 groups) suggests that severity and duration (equivalent to delay in the present study) of spinal cord compression are associated with extent of recovery in that species.27 Some evidence also exists to suggest that immediacy of surgery in humans with spinal cord injury is associated with a superior recovery rate,36,37 although this finding is highly controversial.37 Indeed, results of a study38 involving humans with a severity of spinal injury similar to that of dogs without deep pain sensation (ie, American Spinal Injury Association grade A patients) suggest that an increase in delay to surgery does not affect outcome.39 Prior experimental laboratory studies27,40–42 and the present study differed in 2 important ways. First, in the most relevant experiment involving rats,27 spinal cord compression persisted for a matter of hours, whereas in the present study, compression may have been left untreated for several days (the duration of clinical signs). Second, injury causing loss of deep pain perception in dogs is more severe than that generally evaluated in rats.40–42
Instead, findings of the study reported here suggested that the prognosis for many dogs lacking deep pain perception following IVDH may be more or less fixed at the point of injury (ie, the damage may be irreparable in some dogs). The unknown factor responsible for induction of irreversible injury could perhaps be the primary mechanical injury, the magnitude of which cannot be measured in clinically affected dogs but which can be controlled in laboratory investigations. Contusion initiates a cascade of destructive secondary events, and laboratory experiments involving rats have revealed that subtle differences in various aspects of the mechanical injury can yield quite profound differences in outcome.43 A so-called switch effect has been suggested on the basis of findings involving humans with brain injury after stroke,44 in whom pericyte death causes shutdown of regional blood flow. With the techniques available at the time of the present study, we were unable to identify dogs that would be unable to recover.
Ascending myelomalacia was identified in approximately 10% of the surgically treated dogs in the present study, which was similar to proportions reported previously.10,14–17 This small proportion made it difficult to identify what may be important causal factors. Instead, a cohort study with a sufficient number of dogs to examine this outcome (approx 500 dogs with absence of deep pain perception) or a case-control study would be required to identify specific risk factors.
Although the present study yielded no evidence that the physical and temporal factors examined had prognostic importance, results should be interpreted with caution. Some of the measurements may have been unreliable. For example, owner-reported timing of specific events may have been inaccurate because of difficulty in owner assessment or recall, thereby obscuring the results. However, we considered this potential problem before performing the analyses and therefore used logistic regression (for dichotomous differentiation of recovery of locomotion rather than time-to-event analysis) plus reanalysis of continuous data as categorical data (to reduce the effect of imprecise measurements). We consequently believe that the results were generally reliable. Furthermore, many of the factors examined cannot be more accurately measured in clinical practice than they were in the present study, so measurements obtained in the study should generally represent those available to veterinary neurosurgeons in other clinics.
Other potential sources of inaccuracy in the study reported here included measurements of the severity of spinal cord compression. Moderate disagreement was common between measurements made by 2 observers, and the consequent likely loss of accuracy might have obscured important relationships between compression severity and outcome. Such inaccuracy might be resolved in the future as the quality of clinical imaging increases. No constraints were placed on mode of imaging (CT or MRI) used for measurements of spinal cord compression because the study involved many clinics and many clinicians with different preferences or available equipment. To control for these differences, we chose to assess severity of spinal cord compression by calculating ratios of compressed to uncompressed areas (rather than measuring only the compressed area) by use of the same sequence and imaging modality for each dog. However, failure to use a single modality may have obscured relationships between compression severity and outcome.
Lastly, assessment of absence of deep pain sensation and detection of independent locomotion could have involved some imprecision. Assessment of deep pain sensation was subjective, although no different in nature from other neurologic assessments used for veterinary species or those used in human clinical trials.45 Of course, other observers may have graded the degree of deep pain perception differently than the clinicians involved in the present study, through differences in technique or interpretation. In addition, recovery of independent locomotion as defined in the present study may have been difficult for owners to accurately identify. The study design also did not account for the possibility that some dogs might have been able to walk without recovery of deep pain perception, although this ability usually takes many months to develop.46 To rule out that possibility, we collected information on other variables such as recovery of voluntary tail wag and development of urinary continence in situations requiring follow-up telephone conversations, and many dogs were reexamined at the participating clinics at or after the 3-month follow-up period.
Results of the study reported here indicated that the putative temporal and physical prognostic factors evaluated had little value in predicting recovery of independent locomotion (within 3 months) in dogs with paraplegia and loss of pain sensation in the pelvic limbs. Most notably, the data suggested that, as a whole, dogs with paraplegia and loss of deep pain sensation did not necessarily require immediate surgical intervention. Findings also suggested the possibility that the ability to recover locomotion may have already been determined prior to initial referral evaluation. The mechanism for such a determination likely involved control of blood flow to the damaged region; therefore, we recommend that future investigations of prognostic factors for recovery of independent locomotion in dogs with loss of deep pain perception following IVDH focus on detection of biomarkers for this type of injury.
Acknowledgments
Dr. Hu is supported by a grant from the International Spinal Research Trust.
ABBREVIATIONS
CI | Confidence interval |
IVDH | Intervertebral disk herniation |
Footnotes
Image J, version 1.50a, National Institutes of Health, Bethesda, Md.
Osirix, version 58.2. Pixmeo, Bernex, Switzerland.
Stata, version 11.2, StataCorp, College Station, Tex.
References
1. Griffin JF, Levine JM, Kerwin SC, et al. Canine thoracolumbar intervertebral disk disease: diagnosis, prognosis and treatment. Compend Contin Educ Pract Vet 2009; 31: E1–E14.
2. Jeffery ND, Levine JM, Olby NJ, et al. Intervertebral disk degeneration in dogs: consequences, diagnosis, treatment, and future directions. J Vet Intern Med 2013; 27: 1318–1333.
3. Goggin JE, Li AS, Franti CE. Canine intervertebral disk disease: characterization by age, sex, breed, and anatomic site of involvement. Am J Vet Res 1970; 31: 1687–1692.
4. Bergknut N, Egenvall A, Hagman R, et al. Incidence of intervertebral disk degeneration-related diseases and associated mortality rates in dogs. J Am Vet Med Assoc 2012; 240: 1300–1309.
5. Smolders LA, Bergknut N, Grinwis GC, et al. Intervertebral disc degeneration in the dog. Part 2: chondrodystrophic and non-chondrodystrophic breeds. Vet J 2013; 195: 292–299.
6. Rowland JW, Hawryluk GW, Kwon B, et al. Current status of acute spinal cord injury pathophysiology and emerging therapies: promise on the horizon. Neurosurg Focus 2008; 25: E2.
7. Olby N. Current concepts in the management of acute spinal cord injury. J Vet Intern Med 1999; 13: 399–407.
8. Vale FL, Burns J, Jackson AB, et al. Combined medical and surgical treatment after spinal cord injury: result of a prospective pilot study to assess the merits of aggressive medical resuscitation and blood pressure management. J Neurosurg 1997; 87: 129–146.
9. Gambardella PC. Dorsal decompressive laminectomy for treatment of thoracolumbar disc disease in dogs: a retrospective study of 98 cases. Vet Surg 1980; 9: 24–26.
10. Aikawa T, Fujita H, Kanazono S, et al. Long-term neurologic outcome of hemilaminectomy and disk fenestration for treatment of dogs with thoracolumbar intervertebral disk herniation: 831 cases (2000–2007). J Am Vet Med Assoc 2012; 241: 1617–1626.
11. Speciale J. Common method for pain perception may be inappropriate. J Am Vet Med Assoc 2003; 222: 1502–1503.
12. Davis GJ, Brown DC. Prognostic indicators for time to ambulation after surgical decompression in nonambulatory dogs with acute thoracolumbar disk extrusions: 112 cases. Vet Surg 2002; 31: 513–518.
13. Ruddle TL, Allen DA, Schertel ER, et al. Outcome and prognostic factors in non-ambulatory Hansen type I intervertebral disc extrusions: 308 cases. Vet Comp Orthop Traumatol 2006; 19: 29–34.
14. Duval J, Dewey C, Roberts R, et al. Spinal cord swelling as a my-elographic indictor of prognosis: a retrospective study in dogs with intervertebral disc disease and loss of deep pain perception. Vet Surg 1996; 25: 6–12.
15. Scott HW, McKee WM. Laminectomy for 34 dogs with thoracolumbar intervertebral disc disease and loss of deep pain perception. J Small Anim Pract 1999; 40: 417–422.
16. Olby NJ, Levine J, Harris T, et al. Long-term functional outcome of dogs with severe injuries of thoracolumbar spinal cord: 87 cases (1996–2001). J Am Vet Med Assoc 2003; 222: 762–769.
17. Loughin CA, Dewey CW, Ringwood PB, et al. Effect of durotomy on functional outcome of dogs with type I thoracolumbar disc extrusion and absent deep pain perception. Vet Comp Orthop Traumatol 2005; 18: 141–146.
18. Sylvestre AM, Cockshutt JR, Parent JM, et al. Magnetic motor evoked potentials for assessing spinal cord integrity in dogs with intervertebral disc disease. Vet Surg 1993; 22: 5–10.
19. Poncelet L, Michaux C, Balligand M. Somatosensory potentials in dogs with naturally acquired thoracolumbar spinal cord disease. Am J Vet Res 1993; 54: 1935–1941.
20. Ito D, Matsunaga S, Jeffery ND, et al. Prognostic value of magnetic resonance imaging in dogs with paraplegia caused by thoracolumbar intervertebral disk extrusion: 77 cases (20002003). J Am Vet Med Assoc 2005; 227: 1454–1460.
21. Penning V, Platt SR, Dennis R, et al. Association of spinal cord compression seen on magnetic resonance imaging with clinical outcome in 67 dogs with thoracolumbar intervertebral disc extrusion. J Small Anim Pract 2006; 47: 644–650.
22. Boekhoff TM, Flieshardt C, Ensinger EM, et al. Quantitative magnetic resonance imaging characteristics: evaluation of prognostic value in the dog as a translational model for spinal cord injury. J Spinal Disord Tech 2012; 25: E81–E87.
23. Witsberger TH, Levine JM, Fosgate GT, et al. Associations between cerebrospinal fluid biomarkers and long-term neurologic outcome in dogs with acute intervertebral disk herniation. J Am Vet Med Assoc 2012; 240: 555–562.
24. Roerig A, Carlson R, Tipold A, et al. Cerebrospinal fluid tau protein as a biomarker for severity of spinal cord injury in dogs with intervertebral disc herniation. Vet J 2013; 197: 253–258.
25. Nishida H, Nakayama M, Tanaka H, et al. Evaluation of serum phosphorylated neurofilament subunit NF-H as a prognostic biomarker in dogs with thoracolumbar intervertebral disc herniation. Vet Surg 2014; 43: 289–293.
26. Taylor AR, Welsh CJ, Young C, et al. Cerebrospinal fluid inflammatory cytokines and chemokines in naturally-occurring canine spinal cord injury. J Neurotrauma 2014; 31: 1561–1569.
27. Dimar II JR, Glassman SD, Raque GH, et al. The influence of spinal canal narrowing and timing of decompression on neurologic recovery after spinal cord contusion in a rat model. Spine 1999; 24: 1623–1633.
28. Laitinen OM, Puerto DA. Surgical decompression in dogs with thoracolumbar intervertebral disc disease and loss of deep pain perception: a retrospective study of 46 cases. Acta Vet Scand 2005; 46: 79–85.
29. Sharp NJH, Wheeler SJ. Thoracolumbar disc disease. In: Small animal spinal disorders: diagnosis and surgery. 2nd ed. Beijing: Elsevier-Mosby, 2005;121–160.
30. Button KS, Ioannidis JP, Mokrysz C, et al. Power failure: why small sample size undermines the reliability of neuroscience. Nat Rev Neurosci 2013; 14: 365–376.
31. Levine JM, Levine GJ, Johnson SI, et al. Evaluation of success of medical management for presumptive thoracolumbar intervertebral disk herniation in dogs. Vet Surg 2007; 36: 482–491.
32. Levine JM, Levine GJ, Boozer L, et al. Adverse effects and outcome associated with dexamethasone administration in dogs with acute thoracolumbar intervertebral disk herniation: 161 cases (2000–2006). J Am Vet Med Assoc 2008; 232: 411–417.
33. Vandenbroucke JP, von Elm E, Altman DG, et al. STROBE initiative. Strengthening the reporting of observational studies in epidemiology (STROBE): explanation and elaboration. PLoS Med [serial online] 2007; 4: e297. Available at: journals.plos.org/plosmedicine/article?id=10.1371/journal.pmed.0040297. Accessed Oct 23, 2015.
34. Vittinghoff E, McCulloch CE. Relaxing the rule of ten events per variable in logistic and Cox regression. Am J Epidemiol 2007; 165: 710–718.
35. von Elm E, Altman DG, Egger M, et al. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. BMJ 2007; 335: 806–808.
36. Furlan JC, Noonan V, Cadotte DW, et al. Timing of decompressive surgery of spinal cord after traumatic spinal cord injury: an evidence-based examination of pre-clinical and clinical studies. J Neurotrauma 2011; 28: 1371–1399.
37. Fehlings MG, Vaccaro A, Wilson JR, et al. Early versus delayed decompression for traumatic cervical spinal cord injury: results of the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS). PLoS ONE [serial online] 2012; 7: e32037. Available at: www.ncbi.nlm.nih.gov/pmc/articles/PMC3285644/. Accessed Oct 23, 2015.
38. van Middendorp JJ. Letter to the editor regarding: “Early versus delayed decompression for traumatic cervical spinal cord injury: results of the Surgical Timing in Acute Spinal Cord Injury Study (STASCIS) (lett). Spine J 2012; 12: 540.
39. Dvorak MF, Noonan VK, Fallah N, et al. The influence of time from injury to surgery on motor recovery and length of hospital stay in acute traumatic spinal cord injury: an observational Canadian cohort study. J Neurotrauma 2015; 32: 645–654.
40. Cheriyan T, Ryan DJ, Weinreb JH, et al. Spinal cord injury models: a review. Spinal Cord 2014; 52: 588–595.
41. Stokes BT, Jakeman LB. Experimental modelling of human spinal cord injury: a model that crosses the species barrier and mimics the spectrum of human cytopathology. Spinal Cord 2002; 40: 101–109.
42. Smith PM, Jeffery ND. Histological and ultrastructural analysis of white matter damage after naturally-occurring spinal cord injury. Brain Pathol 2006; 16: 99–109.
43. Lam CJ, Assinck P, Liu J, et al. Impact depth and the interaction with impact speed affect the severity of contusion spinal cord injury in rats. J Neurotrauma 2014; 31: 1985–1997.
44. Hall CN, Reynell C, Gesslein B, et al. Capillary pericytes regulate cerebral blood flow in health and disease. Nature 2014; 508: 55–60.
45. Bracken MB, Shepard MJ, Collins WF, et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury. Results of the Second National Acute Spinal Cord Injury Study. N Engl J Med 1990; 322: 1405–1411.
46. Handa Y, Naito A, Watanabe S, et al. Functional recovery of locomotive behavior in the adult spinal dog. Tohoku J Exp Med 1986; 148: 373–384.