The Influence of Spinal Canal Narrowing and Timing of Decompression on Neurologic Recovery After Spinal Cord Contusion in a Rat Model

Transcription

1 The Influence of Spinal Canal Narrowing and Timing of Decompression on Neurologic Recovery After Spinal Cord Contusion in a Rat Model SPINE Volume 24, Number 16, pp , Lippincott Williams & Wilkins, Inc. John R. Dimar, II, MD,* Steven D. Glassman, MD,* George H. Raque, MD, Yi Ping Zhang, MD, and Christopher B. Shields, MD, FRCS, (C) Study Design. The effect of spinal canal narrowing and the timing of decompression after a spinal cord injury were evaluated using a rat model. Objective. To evaluate whether progressive spinal canal narrowing after a spinal cord injury results in a less favorable neurologic recovery. Additionally, to evaluate the effect of the timing of decompression after spinal cord injury on neurologic recovery. Summary of Background Data. Results in previous studies are contradictory about whether the amount of canal narrowing or the timing of decompression after a spinal cord injury affects the degree of neurologic recovery. Methods. Forty adult male Sprague Dawley rats were equally divided into a control group, in which spacers of 20%, 35%, and 50% were placed into the spinal canal after laminectomy, and an injury group in which the spacers were placed after a standardized incomplete spinal cord injury. After spacer removal, neurologic recovery in both was monitored by Basso, Beattie, Bresnahan (BBB) Locomotor Rating Scale (Ohio State University, Columbus, OH) motor scores and transcranial magnetic motor evoked potentials for 6 weeks followed by histologic examination of the spinal cords. Subsequently, 42 rats were divided into five groups in which, after spacer placement, the time until decompression was lengthened 0, 2, 6, 24, and 72 hours. Again, serial BBB motor scores and transcranial magnetic motor evoked potentials were used to assess neurologic recovery for 6 weeks until the animals were killed for histologic evaluation. Results. Spacer placement alone in the control animals resulted in no neurologic injury until canal narrowing reached 50%. All of the control groups (spacer only) exhibited significantly better (P 0.05) motor scores compared with the injury groups (injury followed by spacer insertion). Within the injury groups the motor scores were progressively lower as spacer sizes increased from the no-spacer group to the 35% group. The results in the 35% and 50% groups were not statistically different. The results of the time until decompression demonstrated that the motor scores were consistently better the shorter the duration of spacer placement (P 0.05) for each of the time groups (0, 2, 6, 24, and 72 hours) over the 6-week recovery period. Histologic analysis showed more severe spinal cord damage as both spinal canal narrowing and the time until decompression increased. From the Departments of *Orthopaedic Surgery and Neurological Surgery, University of Louisville, and the Kenton D. Leatherman Spine Center, Louisville, Kentucky. Acknowledgment date: May 26, First revision date: October 7, Acceptance date: February 24, Device status category: 1. Conclusion. The results in this study present strong evidence that the prognosis for neurologic recovery is adversely affected by both a higher percentage of canal narrowing and a longer duration of canal narrowing after a spinal cord injury. The tolerance for spinal canal narrowing with a contused cord appears diminished, indicating that an injured spinal cord may benefit from early decompression. Additionally, it appears that the longer the spinal cord compression exists after an incomplete spinal cord injury, the worse the prognosis for neurologic recovery. [Key words: canal narrowing, motor behavior scores, neurologic reconstruction, spinal cord injury, transcranial magnetic motor evoked potentials] Spine 1999;24: The importance of restoring spinal stability after spinal trauma, either by operative or nonoperative means is widely accepted. 10,12,13,22,27,28 However, the benefit and timing of spinal canal decompression are less clear. The debate centers on the question of whether the neurologic deficit is caused by the initial contusion injury, or whether the concurrent spinal canal narrowing results in ongoing injury that impedes neurologic recovery. 5,11,12,14,16,23,29 This clinical dilemma has been difficult to resolve because of the difficulty in assessing the degree of initial spinal cord injury and the unpredictable effect of several other factors such as the secondary metabolic changes of spinal cord injury, the extent of spinal canal narrowing, the degree of spinal instability, and the associated nonspinal injuries. Efforts to develop an analogous animal model have been hampered by the inability to replicate the multiple parameters that simulate the clinical situation. Despite these difficulties, the value of determining whether decompression after spinal cord injury is beneficial to neurologic recovery and the optimal timing of surgery justifies ongoing studies. The purpose of this study was to evaluate the influence of spinal canal narrowing on locomotor recovery after a spinal cord injury. Additionally, the effect of early versus delayed decompression after spinal cord injury was assessed. These experiments were performed using a well-established rat spinal cord injury model, a precise method to simulate mechanical spinal canal narrowing, and electrophysiologic and locomotor testing of the hindlimbs of rats. 1623

2 1624 Spine Volume 24 Number Figure 1. Diagram of spacer showing its dimensions. Materials and Methods In a preparatory study, the average anteroposterior spinal canal diameter was determined from the spines of rats of similar weight and age. This allowed for determination of the spacer size needed to produce a precise degree of narrowing of the spinal canal diameter. These spacers (Figure 1) were manufactured to decrease the rat s average spinal canal diameter by 20% (0.69-mm thick spacer), 35% (1.02-mm thick spacer), and 50% (1.35-mm thick spacer). Neurologic Recovery Versus Percentage of Canal Narrowing. Forty Sprague Dawley rats with an average age of 12 weeks 3 days and weight of g were divided equally into two groups. Baseline neurologic and electrophysiologic testing was performed on all rats. The control group consisted of the rats that had a spacer placed without a prior spinal cord contusion injury. The control group was divided into four subgroups: laminectomy only, laminectomy with insertion of a 20% spacer, laminectomy with insertion of a 35% spacer, and laminectomy with insertion of a 50% spacer. The injury group consisted of rats that had a spinal cord contusion injury followed by placement of a spacer in the dorsal epidural space immediately rostral to the injury. This spacer placement was consistently adjacent and partially over the epicenter of the contusion, thus providing precise mechanical narrowing of the spinal canal. The injury group was also divided into four subgroups consisting of: laminectomy and contusion, laminectomy with contusion and insertion of a 20% spacer, laminectomy with contusion and insertion of a 35% spacer, and laminectomy with contusion and insertion of a 50% spacer. The rats were anesthetized intraperitoneally with ketamine and xylazine (mixture 1.5 ml/kg combining 37.5 mg/ml ketamine and 5 mg/ml xylazine) followed by the administration of 5 mg gentamicin subcutaneously. A midline lower thoracic incision was made with exposure of the T10 lamina. Under 10 magnification, a T10 laminectomy was performed by one surgeon. In rats in the control group (n 20) a Teflon spacer was placed beneath the T9 lamina, dorsal to the dural tube. The spacers were manufactured using a Teflon rod that was micromachined into the appropriate spacer size under 10 binocular magnification. The spacers were then precisely measured to ensure correct dimensions using microcalipers. Each spacer had a tungsten wire attached to facilitate its removal 6 hours after surgery. The pullout wire was then tied subcutaneously with a removable suture. The paraspinal fascia and muscle were closed and the skin stapled. After laminectomy, rats in the injury group (n 20) were placed on a test frame where a 12.5 g/cm spinal cord contusion was produced using an impactor (developed at New York University [NYU]; New York, New York) impactor (Figure 2). The NYU impactor can precisely produce an injury in the test animals through an electronically released plunger. The weight of the plunger is 10 g, whereas the height of the drop may be 6.25 mm, 12.5 mm, 25 mm, or 50 mm, thus creating a spectrum of injury from mild to severe. The 12.5-g/cm lesion was chosen as the optimal severity of injury, because at 6 weeks there is sufficient locomotor recovery to enable the rats to stand and walk. The dural tube was noted to swell immediately after the contusion. The first subgroup of rats with a spinal cord injury had no spacer placed after contusion. The next three injury subgroups (n 5 rats in each group) had 20%, 35%, and 50% spacers placed between the T9 lamina and the dural tube after spinal cord injury. The pullout wires of the spacers were secured to the fascia of the paraspinal muscles followed by closure of the skin with staples. After surgery, the rats were placed on a heating pad ( C) until they regained consciousness. Rats in both the control and spinal cord injury groups had the spacers removed 6 hours after surgery. The rats were then returned to the animal-holding facilities where they were maintained and monitored for 6 weeks. The bladder was manually expressed twice a day until normal voiding was obtained or the bladder became spastic. Neurologic status was monitored weekly by two methods: transcranial magnetic motor evoked potentials (tcmmeps) and the Basso, Beattie, Bresnahan (BBB) 2 Locomotor Rating Scale (developed at Ohio State University, Columbus, OH). Transcranial MMEPs are an extremely sensitive electrophysiologic method to measure the integrity of spinal cord function and were used as one method to evaluate neurologic outcomes in the test rats. They provide an objective measurement that can be assessed in one of two ways: quality of the tcmmeps (i.e., measurement of the changes in amplitude) and latency or the quantity of tcmmeps (i.e., presence or absence of response). During preparatory studies, it became clear that even with a very mild spinal cord injury, there was too great a reduction in the amplitude of the tcmmep responses, making it impractical to record amplitudes. Therefore, for this study, quantity of the tcmmep response rate was used, (i.e., presence or absence of response). The amplitude of compound muscle action potentials (in microvolts) were set at 200 V or above, Figure 2. Spinal cord test apparatus used to create a 12.5-g/cm spinal cord injury.

3 Neurologic Recovery After Spinal Cord Contusion Dimar et al 1625 Table 1. BBB Behavior Rating Scale I. Early stage of recovery (hindlimb joint movements) 0 No observable hindlimb (HL) movement 1 Slight movement of 1 or 2 joints, usually the hip and/or knee 2 Extensive movement of 1 joint or extensive movement of 1 joint and slight movement of 1 other joint 3 Extensive movement of 2 joints 4 Slight movement of all 3 joints of the HL (hip, knee, and ankle) 5 Slight movement of 2 joints and extensive movement of the third 6 Extensive movement of 2 joints and slight movement of the third 7 Extensive movement of all 3 joints of the HL II. Intermediate stage of recovery (coordination in stepping ability) 8 Sweeping with no weight support or plantar placement of the paw with no weight support 9 Plantar placement of the paw with weight support in stance only (i.e., when stationary) or occasional, frequent or consistent weight supported dorsal stepping and no plantar stepping 10 Occasional weight supported plantar steps, no FL-HL (forelimbhindlimb) coordination 11 Frequent to consistent weight supported plantar steps and no FL-HL coordination 12 Frequent to consistent weight supported plantar steps and occasional FL-HL coordination 13 Consistent weight supported plantar steps and frequent FL-HL coordination III. Late stage of recovery (details, refinement of locomotion) 14 Consistent weight supported steps, consistent FL-HL coordination, and predominant paw position during locomotion is rotated or frequent plantar stepping, consistent FL-HL coordination and occasional dorsal stepping 15 Consistent FL-HL coordination and no toe clearance or occasional toe clearance during forward limb advancement; predominant paw position is parallel to the body at initial contact 16 Consistent FL-HL coordination during gait, and toe clearance occurs frequently during forward limb advancement; predominant paw position is parallel at initial contact and rotated at lift off 17 Consistent FL-HL coordination during gait, and toe clearance occurs frequently during forward limb advancement; predominant paw position is parallel at initial contact and lift off 18 Consistent FL-HL coordination during gait, and toe clearance occurs consistently during forward limb advancement; predominant paw position is parallel at initial contact and rotated at lift off 19 Consistent FL-HL coordination during gait, and toe clearance occurs consistently during forward limb advancement; predominant paw position is parallel at initial contact and lift off 20 Consistent coordinated gait, consistent toe clearance; predominant paw position is parallel at initial contact and lift off, but there is trunk instability; the tail is consistently up 21 Coordinated gait, consistent toe clearance, predominant paw position is parallel throughout stance, consistent trunk stability, and tail is consistently up. cord injury groups by averaging the left and right hindlimb motor scores. The BBB locomotor score consists of a precise, scaled scoring system that evaluates all of the rats hindlimbs major motor groups, bladder function, and the ability to ambulate. (Table 1 modified BBB motor score table). All rats were killed after a lethal dose of phenobarbital (100 mg/kg), 6 weeks after surgery. Histologic studies were performed on each of the control and injury test groups with two spinal cords sectioned in the sagittal and axial planes from each subgroup. Each cord was stained using hematoxylin and eosin stain and examined microscopically. Neurologic Recovery Versus Duration of Canal Narrowing. The second phase of the study evaluated the effect on neurologic recovery of time elapsed until decompression after a 12.5-g/cm spinal cord contusion with 35% spinal canal narrowing. The decision to use the 35% spacer was based on the finding that 35% canal narrowing is the critical point at which a significant inhibition of neurologic recovery occurred in the first section of this study. Forty-two Sprague Dawley rats with an average age of 12 weeks 3 days and weighing g were segregated into five groups based on the duration of spinal cord compression: 0 hours (n 8), 2 hours (n 8), 6 hours (n 7), 24 hours (n 10), and 72 hours (n 9). The 6-hour group included five animals previously tested in the neurologic recovery versus canal narrowing phase of the study. A T10 laminectomy was performed followed by the creation of a spinal cord contusion at T10 using the NYU impactor. After contusion, a 35% spacer was placed dorsally between the T9 lamina and the dural tube. In the 0-hour test rats, the spacer was placed into the canal and immediately removed. In the remaining groups (2, 6, 24, and 72 hours) the spacer was placed into the canal with the removal wire secured subcutaneously. At the specified times, the spacers were removed. The animals were maintained in the animal care facility for 6 weeks. Weekly tcmmeps and BBB scores were obtained in all an- because responses smaller than 200 V could not be differentiated from the background noise. This study s tcmmep data were compared with normative tcmmep data, determined by Linden et al. 24 The tcmmeps were performed using a standard technique with recording electrodes placed in the gastrocnemius muscles of both hindlimbs. 17,18 A ground electrode was placed into the rat s lumbar paraspinal muscles. A magnetic stimulator (Excel MES-10; Cadwell Laboratories, Kenewick, WA) delivered a magnetic pulse of 70 sec (2-T maximum output) through a 5.5-cm diameter coil. The tip of the coil was positioned at the rat s inion. Each tcmmep recording was duplicated for reproducibility. Latency of onset (time between stimulation and initiation of the muscle evoked potential) and amplitude (peakto-peak) were measured for each tcmmep pair. A sample tracing is shown in Figure 3. The BBB locomotor score was performed after surgery, at 24 hours, and weekly on each rat in both the control and spinal Figure 3. Both rats demonstrated good baseline tcmmep responses (onset latency, 6 msec; amplitude, 5 mv). C-35: The tcmmep tracing on the left represents the tracings obtained immediately before and 3 weeks after the insertion of a 35% spacer into the spinal canal for 6 hours. Note at 3 weeks a slight decrease in amplitudes but no increase in latency on both the left and right gastrocnemius muscles. I-35: The tcmmep tracing on the right represents the tracings obtained immediately before and 3 weeks after the creation of a moderate (12.5 g/cm) spinal cord injury plus the insertion of a 35% spacer for 6 hours. Note at 3 weeks the complete loss of responses in the left gastrocnemius muscle, whereas on the right a response was obtained, but the latency was more than double (17 msec), with significantly decreased amplitude. tcmmeps, transcranial magnetic motor evoked potentials.

4 1626 Spine Volume 24 Number Figure 4. Control group with spacer only. The BBB motor score graph demonstrating motor score ratings from day 1 through 6 weeks. Note that the 0%, 20%, and 35% groups show no decrease in normal scores, which is 21. The 50% spacer demonstrates transient scores that return to 19/21 (good gait). BBB, Basso, Beattie, Bresnahan 2 Locomotor Rating Scale. imals. At 6 weeks after surgery the rats were killed using a lethal dose of intraperitoneal phenobarbital (100 mg/kg). Histology of the spinal cord was performed in each subgroup with two spinal cords sectioned in the sagittal and axial planes. Each spinal cord was examined microscopically using hematoxylin and eosin. Statistical Methods. Analysis of the neurologic recovery versus canal narrowing BBB scores and tcmmeps was performed using the nonparametric Wilcoxon signed rank test for matched pairs. The neurologic recovery versus duration of canal narrowing data using the BBB and tcmmeps values were analyzed with the Wilcoxon signed rank test and the analysis of covariance (ANCOVA). Additionally, the data were plotted using the BBB means standard errors of the mean and the line of best fit (linear). The slopes of the lines of best fit for each time group were evaluated for trend significance. Results Neurologic Recovery Versus Canal Narrowing All rats survived until the end of the study, and there were no infections. The control groups (spacer only) exhibited normal BBB scores except in the group with 50% spinal canal narrowing, which demonstrated an initial deterioration of hindlimb locomotor function that recovered to 90% of baseline motor function (19/21 BBB score) by the sixth week after injury (Figure 4). The injury groups (contusion plus spacer) exhibited abnormal BBB motor scores in all groups and demonstrated neurologic recovery at 6 weeks that was directly related to the spacer size (Figure 5). Assessment of the rats in the control group demonstrated that those with the 0%, 20%, and 35% spacers inserted had normal motor scores after day 1. The 35% control group locomotor scores were significantly higher (P 0.05) when compared with those in the 50% spacer group. According to the results of a repeated-measures analysis of variance, even the 50% spacer group demonstrated significant motor recovery of the hindlimbs at 6 weeks (P 0.05). The motor scores of the control groups (normal, 21) when compared with scores of the injured groups were higher for each wedge size studied (0%, 20%, 35%, 50%). Results of the statistical analysis of BBB motor scores for control rats (spacer only) versus injured rats (contusion plus spacer) showed that all rats in the injured group demonstrated a significantly increased neurologic (P 0.05) deficit when compared with the control group. The effect of the spacer size on rats in the injured group was assessed during the 6-week postoperative period to determine whether neurologic recovery of the hindlimbs correlated with spacer size. The results of the Wilcoxon signed rank test demonstrated that the BBB motor scores were consistently higher in the 0% spacer group than in the 20% spacer group (P 0.05) and in the 20% spacer group than in the 35% spacer group (P 0.05). There was no significant difference in the 35% spacer group when compared with the 50% spacer group after statistical testing with Student s t test. Over the 6-week postoperative period, each increase in spacer size resulted in a significant (P 0.05) decrease of neurologic recovery within the first three groups (0%, 20%, and 35% spacers). The tcmmep data (Figure 6, MEP Controls) for the rats in the control groups demonstrated no significant differences between the 0%, 20%, and 35% spacer groups. There was a minimal decrease in tcmmeps in the 35% spacer control group that returned to normal levels at 6 weeks. The 50% spacer caused a loss of tcm- MEPS that was significant when compared with transient deterioration seen in the 35% spacer group (P 0.05). The tcmmeps in the injured group (Figure 7, MEP Injury) demonstrated significant differences between the 0% and 20% spacer groups (P 0.001), and the 20% and 35% spacer groups (P 0.05), whereas no Figure 5. Injury group: spacer with spinal cord injury. The BBB motor score graph demonstrating significantly abnormal motor score ratings from day 1 through 6 weeks. Significant differences existed among the 0%, 20% and 35% spacer groups but not between the 35% and 50% groups. Note the abrupt worsening of the scores with the 35% and 50% spacers (motor scores decreased from 15/21 to 7/21). BBB, Basso, Beattie, Bresnahan 2 Locomotor Rating Scale.

5 Neurologic Recovery After Spinal Cord Contusion Dimar et al 1627 Figure 6. Control group: spacer only. A transcranial magnetic motor evoked potential graph demonstrating responses by 6 weeks. The 50% group responses were lost on day 1 and minimal improvement was shown at 6 weeks. MEP, motor evoked potential. significant differences were noted in the 35% and 50% spacer groups because of inconsistent responses. Comparisons of the tcmmeps in the injured and control groups demonstrated significant deterioration of tcmmeps in each injury group, with the exception of the 50% control spacer group, when compared with the same size spacer in the injury group. The 0% injury group demonstrated significant worsening of the tcmmeps when compared with the 0% control group at 6 weeks (P 0.05). The 20% and 35% spacer control groups also demonstrated significant worsening of the tcmmeps when compared with the same spacer size injury group (P 0.05). The tcmmeps were so sensitive to spinal cord electrophysiologic disruption that no significant potentials or differences between the potentials Figure 7. Injury group: spacer with spinal cord injury. A tcmmep graph showing responses from day 1 through 6 weeks. The tcmmep responses were significantly different among the 0%, 20%, and 35% groups, but not between the 35% and 50% groups. The tcmmep was extremely sensitive to cord injury, with complete loss of responses in the 35% and 50% groups. tcmmep, transcranial magnetic motor evoked potentials. Figure 8. Neurologic recovery versus duration of compression: motor score means. The BBB motor scores after injury with a 35% spacer over 0-, 2-, 6-, 24-, and 72-hour intervals before removal. Each line represents one of these interval group means standard error of means plotted from day 1 through 6 weeks. Note the progressive worsening of the motor scores as spacer placement time increases. BBB, Basso, Beattie, Bresnahan 2 Locomotor Rating Scale. were identified in the 35% versus 50% spacer control or injury groups. Neurologic Recovery Versus Duration of Canal Narrowing During this phase of the experiment, a 35% spacer after contusion was used as a constant, and the time until decompression of the spinal cord was varied (0, 2, 6, 24 and 72 hours). There were no infections; however, two animals died before final evaluation at 6 weeks. One rat in the 24-hour group died of dehydration 3 weeks after surgery, and one rat in the 0-hour group was euthanized because of technical problems causing spinal cord injury. The BBB locomotor scores of each time group was plotted each week using BBB mean scores standard error of the mean (Figure 8, BBB means SEM). Each of the compression intervals (0, 2, 6, 24, and 72 hours) was compared using the Wilcoxon signed rank test and ANCOVA. These values were plotted using a linear line of best fit. (Figure 9, linear BBB scores). The Wilcoxon signed rank test results showed that in each group the BBB motor score means were consistently higher (P 0.05) over the 6-week recovery period as the duration of spacer insertion time decreased. Rats with the spacers in place for 0, 2, and 6 hours had sufficient motor recovery to ambulate, measured by the BBB motor scores, whereas the 24-hour group had hindlimb extensile motion and the 72-hour group remained essentially paraplegic. The ANCOVA test showed that the 0-hour and 2-hour groups overlapped and were not statistically different (P 0.05). All other groups demonstrated statistically improved BBB scores the shorter the duration of spinal cord compression (P 0.001). Neurologic recovery in the 2-hour group was better than in the 6-hour group, in the 6-hour group than in the 24-hour group, and in the 24-hour group than in the 72-hour group. The

6 1628 Spine Volume 24 Number Figure 9. Neurologic recovery versus duration of compression: line of best fit. Lines of best fit and BBB motor scores with a 35% spacer over 0-, 2-, 6-, 24- and 72-hour intervals before removal. All slopes of each group showed significant improvement during the 6-week postoperative period (P 0.05). Results of regression analysis of the slopes demonstrated that all slopes increased significantly over time. 0 hours, P 0.05; 2 hours, P 0.001; 6 hours, P 0.001; 24 hours, P 0.001; and 72 hours, P Each of the time groups means are statistically significant and consistently higher the shorter the duration of cord compression. Clinically, rats in the 0- and 2-hour groups could walk, but the 72-hour animals were essentially paraplegic. BBB, Basso, Beattie, Bresnahan 2 Locomotor Rating Scale. BBB motor score results, as analyzed by both statistical methods, demonstrated that neurologic recovery was significantly higher the shorter the duration of spinal cord compression, with the exception of the 0-hour and 2-hour groups, which demonstrated the same return of motor function according to the results of ANCOVA. All test groups demonstrated parallel improvement of neurologic recovery over 6 weeks, with each group demonstrating significant (P 0.001) improvement of BBB motor scores when comparing the immediate postoperative motor scores with those at 6 weeks. The BBB motor scores improved 1 grade per week in the 0-, 2-, 6-, and 24-hour groups and 0.5 of a grade per week in the 72- hour group. Analysis of the tcmmep responses also showed significant differences between the duration of compression in the various groups. (Figure 10, tcmmeps responses). The tcmmeps response rates of the group with 0 hours of compression were significantly better than the those in the 2-, 6-, 24-, and 72-hour groups (P 0.05). The tcmmep response rates of the 2-hour group were significantly better than those of the 6-, 24-, and 72-hour groups (P 0.05). However, because of the sensitivity of the tcmmeps, at 6 hours the responses were obtained in only 50% of the rats, whereas no tcmmep responses were obtained from the 24- and 72-hour groups. Therefore, no statistically significant difference was found between the 6-, 24-, and 72-hour groups. Histology Histologic analysis by light microscopic examination of the spinal cord in the region of spacer insertion in the neurologic recovery versus canal narrowing spinal cord specimens showed significant changes in the architecture in both the control and injury groups. Cross sectional views (Figure 11A) demonstrated no significant damage within the control groups until the 50% spacer was inserted, which caused injury to the dorsal columns of the spinal cord, characterized by cavitation, marginal gliosis, and atrophy. Midsagittal sections (Figure 12) also showed no significant microscopic changes until the 50% spacer was inserted, in which case similar changes occurred. Rats in the spinal cord injury groups demonstrated more severe histologic changes (Figure 11B). Cross sectional views of the spinal cord showed a progressive increase in dorsal column injury with extension of necrosis and gliosis with an increase in the spacer size. The zone of necrosis and cavitation extended into the central and ventral portion of the spinal cord, resulting in progressive myelomalacia with incremental increase in spacer size. The damage was most severe in the rats with spinal cord injury plus 35% and 50% spacer insertion. Histologic changes were also noted in the midsagittal sections of the spinal cord in which dorsal cavitation also occurred with progressive fibrosis, gliosis, and atrophy of the cord. There was marked cephalad and caudad extension of the spinal cord cavitation when larger spacers were used. Necrotic changes extended far beyond the area of injury and spacer placement, reaching more than 10 mm in both directions in rats with the 50% spacer inserted (Figure 13). Figure 10. Neurologic recovery versus duration of compression: tcmmep responses. The tcmmep responses with a 35% spacer over 0-, 2-, 5-, 24- and 72-hour intervals before removal. the TcMMEPs worsened the longer the duration of compression with no significant difference at 6, 24, and 72 hours. The responses disappeared in these three groups. tcmmep, transcranial magnetic motor evoked potentials.

7 Neurologic Recovery After Spinal Cord Contusion Dimar et al 1629 Histologic examination of the spinal cords used in the neurologic recovery versus duration of canal narrowing phase of the study demonstrated progressively more severe central and dorsal cavitation as the time until decompression increased. Cross-sectional views demonstrated progressive ventral cord damage that became progressively more severe, particularly in the rats with 6 hours or more of spinal cord compression. By 1 day, minimal ventral spinal cord substance remained, and by 3 days, virtually all of the spinal cord was destroyed (Figure 14). Midsagittal sections (Figure 15) showed similar progressive changes as the duration of compression increased. The ventral substance of the spinal cord was preserved in the rats subjected to 6 hours of compression, but by 1 day the extent of spinal cord myelomalacia increased with almost complete atrophy occurring in those rats in which the spinal cord was subjected to 72 hours of compression. Midsagittal sections also demonstrated progressive cephalad and caudad cord necrosis and cavitation, which worsened the longer the duration of compression. These changes were most severe in the 24- and 72-hour specimens. Figure 11. Histology: neurologic recovery versus canal narrowing. A, Axial view of gross sections of cord for each control group (0%, 20%, 35%, 50%) demonstrating no cord damage until the 50% spacer was placed, which caused dorsal cavitation. B, Axial view of gross sections of cord for each injury group demonstrating progressively more severe dorsal cord cavitation and myelomalacia the larger the spacer. Figure 12. Histology: neurologic recovery versus canal narrowing. Midsagittal gross sections of spinal cord in each control group demonstrating no significant changes until dorsal cavitation in the 50% group.

8 1630 Spine Volume 24 Number Figure 13. Histology: neurologic recovery versus canal narrowing. Midsagittal gross sections of spinal cord in each injury group demonstrating progressive dramatic cephalad and caudal extension of destructive cavitation the longer the larger the spacer. Discussion The role played by canal narrowing after spinal cord contusion on the ultimate return of neurologic function remains controversial. Some investigators have suggested that there is no correlation between the amount of neurologic recovery and the degree of spinal canal encroachment. 21 Others have reported that early surgical decompression results in no improvement or even worsening of neurologic function. 1,24 However, contrary to these opinions there is an extensive body of experimental data implying that early decompression of canal narrowing after spinal cord contusion may result in improved neurologic outcomes. 7 9,19,25,26 Figure 15. Histology: neurologic recovery versus duration of canal narrowing. Midsagittal spinal cord sections from groups with 0, 2, 6, 24, and 72 hours until decompression injury demonstrating more severe dorsal cavitation the longer the duration of the compression and marked cephalad and caudal extension of cord myelomalacia most dramatic in the 72-hour specimens. The models used to simulate spinal cord injury and the effect of early decompression included aneurysm clips, nylon electrical cables, compressive plates, translaminar screws, balloon catheters, and weight Figure 14. Histology: neurologic recovery versus duration of canal. Narrowing axial gross sections of time until decompression injury groups (0, 2, 6, 24, and 72 hours) demonstrating progressively more severe dorsal cord cavitation. Note the almost total myelomalacia of the cord at 72 hours.

9 Neurologic Recovery After Spinal Cord Contusion Dimar et al 1631 drops. 1,4,7 9,19,20,25,26 In each of these experimental models the efficacy of early decompression has been demonstrated. Several mechanisms of injury were proposed in these models as the cause of spinal cord injury and included hemorrhage, arterial or venous stasis, decreased tissue oxygen, or secondary metabolic effects. 5,11,12,15,16,23,29 In the current study a standardized, reproducible spinal cord injury model was used to control precisely both the degree and duration of postinjury spinal canal narrowing and to measure accurately the effect of these parameters on neurologic recovery. There were three major challenges in designing the study: the creation of a reproducible spinal cord injury in each rat; the creation of reproducible, measurable spinal canal narrowing to simulate encroachment of bone fracture fragments; and the precise measurement of locomotor and electrophysiologic hindlimb functional recovery. The rat spinal cord injury model induced by the NYU impactor was selected because of its widespread acceptance in previous studies of spinal cord trauma and regeneration. 6 To quantify spinal canal narrowing the anteroposterior spinal canal diameter was measured in genetically homologous Sprague Dawley rats of similar size and age to the study animals. Spacers were manufactured to narrow the spinal canal diameter by precisely 20%, 35%, or 50%. After induction of a spinal cord contusion, immediate swelling of the dural tube was observed at the site of impaction with extension of the swelling under the adjacent lamina. Although the spacers could not be placed directly over the laminectomy site, they were consistently adjacent and partially over the expanding epicenter of the spinal cord injury, thus providing exact and reproducible segmental spinal canal narrowing. Also critical to this study was the establishment of accurate testing methods to document neurologic recovery of the hindlimbs after thoracic spinal cord injury. Two techniques were selected to assess neurologic recovery: BBB motor scores and tcmmeps. 2,18,17 The significance of the magnitude and duration of spinal cord compression on neurologic outcome has wide-ranging potential clinical ramifications. The relevance of this issue is underscored by the trend toward pharmacologic treatment of spinal cord injury, which does not address the ongoing cord damage associated with bony compression as edema develops after contusion. 3,30 Indeed, timely decompression, as well as pharmacologic therapy, may provide the best opportunity for neurologic recovery after a spinal cord injury with concurrent bony encroachment. Neurologic Recovery Versus Canal Narrowing During this phase of the study, the degree of spinal canal narrowing at which significant neurologic deterioration occurred was established in this model. Each increase in spacer size (0 35%; the 35% and 50% spacers produced similar injury) after contusion was less favorable to neurologic recovery. The threshold of canal narrowing at which precipitous neurologic deterioration occurred was 35%. Evaluation of the control animals (i.e., laminectomy only with spacer insertion in the epidural space) established that spacer placement alone resulted in no neurologic injury (P 0.05) until the canal was narrowed to 50% of its normal diameter. Further, the 50% control group regained good locomotor function (BBB score, 19/ 21) by week 6. The tcmmeps were normal in the rats subjected to 0% and 20% spinal canal narrowing but were depressed transiently in the 35% group and were obtained infrequently in the 50% spacer group. It was suspected in the rats in which the canal was narrowed 50% that the spacer induced either spinal cord ischemia or a contusion. The spinal cord histology was normal in the 0%, 20%, and 35% control groups, whereas the 50% control group spinal cords demonstrated significant cavitation, gliosis, and atrophy of the dorsal columns. Therefore, spacer insertion alone (i.e., narrowing of the canal as an isolated event) did little to affect neurologic function and caused no significant histologic changes until extreme narrowing (50%) of the canal occurred. An injury group was created by the induction of a spinal cord contusion before spacer insertion (no spacer or 20%, 35%, or 50% spacer). Spacer insertion in the injured rats produced progressively more severe neurologic deficit in the hindlimbs as the spacer size increased. The injured rats with no spacer insertion demonstrated enough neurologic recovery at 6 weeks (BBB score, 18/ 21) to have a normal-appearing gait, whereas the 20% spacer insertion group exhibited clumsy walking ability (BBB score, 15/21). The 35% and 50% spacer insertion groups exhibited similar and not statistically different antigravity extensile hindlimb movement (BBB score, 7/21). The insertion of both the 35% and 50% spacers caused a precipitous increase in the severity of the neurologic injury. The tcmmep data demonstrated significant worsening of the potentials between the 0% and 20% injury groups (P 0.001) and between the 20% and 35% groups (P 0.05). No significant difference was noted in the tcmmeps between the 35% and 50% compression groups. The tcmmeps in rats with no spacer or the insertion of a 20% spacer returned to 50% of the baseline level. The tcmmeps were so sensitive in the injury groups that no significant differences were detected when comparing the 35% and 50% spacer groups. Animals in the spinal cord injury groups all had incomplete lesions that demonstrated parallel neurologic improvement over 6 weeks, when measured by their BBB locomotor scores, whereas the tcmmeps were less reliable in predicting neurologic recovery because of their sensitivity. The degree of return of neurologic function was significantly worse the larger the size of the spacer inserted, particularly in the 35% and 50% injury groups. This demonstrated that the vulnerability of the spinal cord to damage after spinal canal narrowing differed be-

10 1632 Spine Volume 24 Number tween the control group (intact cord) and injury group (swollen cord) and that after an injury, increasing the spacer size produced a more severe neurologic injury. This result indicates that the injured spinal cord was far more vulnerable and sensitive to canal narrowing in the vicinity of the injury. Neurologic Recovery Versus Duration of Canal Narrowing One of the primary goals of this model was the establishment of the degree of spinal canal narrowing (i.e., 20%, 35%, or 50%) where neurologic damage accelerated. A second goal was to show that the recovery of neurologic function was affected by the duration of the spinal canal narrowing. By establishing a critical threshold of spinal canal narrowing in this model of neurologic injury, the critical effect of time, (i.e., when a decompression should be performed) could be studied. It was established that the critical threshold of spinal canal narrowing was 35% after spinal cord injury, based on the findings in the first part of the study that the rats in the control group after insertion of a 35% spacer demonstrated no changes in the BBB motor scores, transient tcmmeps changes, and no histologic changes. The injury groups demonstrated a dramatic decrease in recovery at 6 weeks between the rats sustaining spinal cord injury and insertion of the 20% versus 35% spacers. Therefore, we concluded, because there were no changes in the 35% control animals and precipitous neurologic changes in the 35% injury groups, that 35% canal narrowing would be used for the time studies. The 35% spacer insertion in all time groups (0, 2, 6, 24, and 72 hours) showed that the BBB scores were consistently higher (P 0.001) in rats that had shorter durations of compression over the 6-week postoperative period when analyzed by the Wilcoxon signed rank test. Results of repeat analysis using ANCOVA were similar, with the exception of the 0- and 2-hour groups, which showed no statistical difference in BBB scores. Thus, the longer spinal canal narrowing existed at the same level as a partial spinal cord injury, the worse the prognosis for neurologic recovery. The 0- and 2-hour groups regained a good gait pattern, the 6-hour group could support their body weight and step, the 24-hour group had extensile antigravity leg movement, but the 72-hour group remained paraplegic. Postoperative neurologic recovery in the groups with different times until decompression was parallel, indicating that all groups had improved BBB scores at the same rate (with the exception of the 72-hour group in which slower recovery occurred). Additionally, in all the groups, the level of neurologic improvement increased over time regardless of the injury severity. Therefore, the progressive differences in final tcmmeps for each time group were owing to the duration of compression, because the initial contusions were identical in each test rat, and the size of the spacer remained constant. It seems that after 6 hours, the prognosis for the recovery of ambulatory potential was poor, and by 72 hours paraplegia was a certainty. Clearly, these motor scores demonstrate that after a reproducible spinal cord injury, the longer the duration of compression the worse the prognosis for neurologic recovery. Histologic examination of the spinal cords demonstrated similar specific changes within the spinal cord architecture that other investigators have noted. 7 Histologic findings in the neurologic recovery versus canal narrowing portion of the current study confirmed that the extent of spinal cord damage worsened with increasing spacer size. These changes were particularly evident in midsagittal sections, in which cephalad and caudad extension of cord necrosis and cavitation worsened as spacer size increased. The histologic changes correlated with the BBB locomotor scores and tcmmeps (i.e., the larger the spacer after contusion the worse the specimens dorsal necrosis and cavitary changes). These findings were in contrast to those in the uninjured animals in which no histologic changes were found until 50% canal narrowing, when a probable acute injury or vascular ischemia occurred. In the injury group, animals with no spacer exhibited spinal cord dorsal cavitation; nevertheless, more severe necrosis and progressive cavitation occurred with each incremental increase in spacer size. Examination of the spinal cord specimens from the second phase of the study, neurologic recovery versus duration of compression, demonstrated clear evidence that spinal cord damage worsened the longer the duration of compression. Cavitary changes progressed to almost complete cord myelomalacia by 72 hours. Again, marked cephalad and caudad extensions of necrosis and cavitation were observed the longer the duration of compression. Therefore, more severe spinal cord damage was observed with both increasing spinal canal narrowing and as the time until decompression increased. These results are strong evidence that the prognosis for neurologic recovery after spinal cord injury was adversely affected by both a greater degree and a longer duration of canal narrowing. The tolerance for spinal canal narrowing within an uninjured spinal cord versus a contused, swollen cord was demonstrated to be significantly different. This indicates that an injured spinal cord was far more vulnerable and sensitive to canal narrowing in the vicinity of the injury. Additionally, the longer this narrowing was present, the worse the spinal cord injury. Histologic analysis also confirmed increased severity of injury with increasing percentage of canal narrowing and duration of compression. Although this animal model may not be directly comparable to the human clinical situation, it presents compelling evidence supporting the need for early decompression after spinal cord contusion. Acknowledgment The authors would like to thank Safamor Danek Corp. for providing funding for this basic research.

11 Neurologic Recovery After Spinal Cord Contusion Dimar et al 1633 References 1. Allen AR. Remarks on the histopathological changes in the spinal cord due to impact. An experimental study, J Nerv Ment Dis 1914;41: Basso DM, Beattie MS, Bresnahan JC. A sensitive and reliable locomotor rating scale for open field testing in rats. J Neurotrauma 1995;12: Bracken MB. Shepard MJ, Collins WF, et al. A randomized, controlled trial of methylprednisolone or naloxone in the treatment of acute spinal-cord injury. N Engl J Med 1990;322: Bresnahan JC, Beattie MS, Todd III FD, Noyes DH A behavioral and anatomical analysis of spinal cord injury produced by a feedback-controlled impaction device. Exp Neurol 1987;95: Carlson GD, Warden KE, Barbeau JM, et al. Viscoelastic relaxation and regional blood flow response to spinal cord compression and decompression. Spine 1997;22: Constantini S, Young W. The effects of methylprednisone and ganglioside GMI on acute spinal cord injury in rats. J Neurosurg 1994;80: Delamarter RB, Sherman J, Carr JB. Pathophysiology of spinal cord injury. J Bone Joint Surg [Am] 1995;77: Delamarter RB, Sherman JE, Carr JB Volvo Award in Experimental Studies. cauda equina syndrome: neurologic recovery following immediate, early, or late decompression. Spine 1991;16: Dolan EJ, Tator CH, Endrenyi L. The value of decompression for acute experimental spinal cord compression injury. J Neurosurg 1980;53: Ducker TB, Salcman M, Perot PL, Ballantine D. Experimental spinal cord trauma. I: Correlation of blood flow, tissue oxygen and neurologic status in the dog, surg. Neurology 1978;10: Ducker TB, Salcman M, Lucas JT, Garrison WB, Perot PL. Experimental spinal cord trauma, II: Blood flow, tissue oxygen, evoked potentials in both paretic and plegic monkeys. Surg Neurol 1978;10: Ducker TB, Salcman M, Daniell HB. Experimental spinal cord trauma. III: Therapeutic effect of immobilization and pharmacologic agents. Surg Neurol 1978;10: Ducker TB, Bellegarrigue R, Salcman M, Walleck C. Timing of operative care in cervical spinal cord injury. Spine 1984;9: Duh MS, Shepard MJ, Wilberger JE, Bracken MB. The effectiveness of surgery on the treatment of acute spinal cord injury and its relation to pharmacological treatment. Neurosurgery 1994;35: Eidelberg E. The pathology of spinal cord injury. Radiol Clin North Am 1997; Fehlings MG, Tator CH, nd Linden RD. The relationship among the severity of spinal cord injury, motor and somatosensory evoked potentials and spinal cord blood flow. Electroencephalography and Clinical Neurophysiology 1989; 74: Ireland: Elsevier Scientific Publishers, Ltd. 17. Fishback AS, Shields CB, Linden RD, Johnson JR, Zhang YP. The effect of propofol on transcranial motor evoked potentials in rats. J Neurosurg, 1995;37: Glassman SD, ZhangYP, Shields CB, Linden RD, Johnson JR. An evaluation of motor-evoked potentials for detection of neurologic injury with correction of an experimental scoliosis. Spine, 1995;20: Guha A, Tator CH, Endrenyi L, Piper I. Decompression of the spinal cord improves recovery after acute experimental spinal cord compression injury. Paraplegia 1987;25: Hashimoto T, Fukuda N. New spinal cord injury model produced by spinal cord compression in the rat. J Pharmacol Methods 1990;23: Herndon WA, Galloway D. Neurologic return versus cross-sectional canal area in incomplete thoracolumbar spinal cord injuries. J Trauma 1988;28: Holdsworth FW, Hardy A, Sheffield E. Early treatment of paraplegia from fractures of the thoraco-lumbar spine. J Bone Joint Surg [Br] 1953;35: Kobrine AI, Evans DE, Rizzoli HV. The effects of ischemia on long-tract neural conduction in the spinal cord. J Neurosurg 1979;50: Linden RD, Zhang YP, Burke DA, Hunt MA, Harpring JE, Shields CB. Magnetic motor evoked potentials recorded from the rat: 1: Normative data. J Neurosurg (In press). 25. Marshall LF, Knowlton S, Garfin SR, et al. Deterioration following spinal cord injury. J Neurosurg 1987;66: Nystrom B, Berglund JE. Spinal cord restitution following compression injuries in rats. Acta Neurol Scand 1988:78: Rivlin AS, Tator CH. Effect of duration of acute spinal cord compression in a new acute cord injury model in the rat. Surg Neurol 1978;10: Schlegel J, Bayley J, Yuan H, Fredricksen B. Timing of surgical decompression and fixation of acute spinal fractures. J Orthop Trauma 1996;10: Tator CH. Early management of acute spinal cord injury. Semin Neurol Surg Chapter 5, 1982;53 8. Address reprint requests to John R. Dimar, II, MD Spine Institute for Special Surgery, PSC 210 East Gray Street, Suite 900 Louisville, KY 40202