- Email: [email protected]
Acute changes in somatosensory evoked potentials following graded experimental spinal cord compression
62
Surg Neurol 1986;25:62-66
Acute Changes in Somatosensory Evoked Potentials following Graded Experimental Spinal Cord Compression Amadeo C. Nacimiento, M.D., Matthias Bartels, M.D., and Friedrich Loew, M.D. Research Laboratory, Department of Neurosurgery, Saarland University Medical School, Homburg/Saar, Federal Republic of Germany
Nacimiento AC, Bartels M, Loew F. Acute changes in somatosensory evoked potentials followinggraded experimental spinal cord compression. Surg Neurol 1986;25:62-66. Amplitude and latency of cortical somatosensory potentials evoked in cats by peripheral nerve stimulation were measured before, during, and for 5 hours after injury of spinal cord segment L-7 by a predetermined degree and duration of compression. An amplitude decrease, slight and transitory, was first observed after compression reduced the segmental cross section by 60%. After an 80% compression, amplitude reduction was initially larger and lasted longer, but recovered 2.5 hours after injury to a level that did not differ statistically from control values. After total (100%) compression, evoked responses disappeared abruptly and did not recover significantly. Latency was unaltered at all degrees of compression. Structural damage increased with the degree of compression. In this model, evoked potential changes neither reflect nor predict the magnitude of acute incomplete spinal cord injury. KEY WORDS: Spinal cord trauma; Spinal cord compression; Cortical evoked potentials; Spinal cord compression model; Evoked potentials; Spinal cord injury
The clinical use o f somatosensory evoked potentials to predict the outcome of spinal cord injury, evaluate degree o f trauma, or assess methods of treatment presents several unresolved problems. Neurophysiologic properties of clinical relevance include their latency, amplitude, sequential configuration, and polarity. The degree of reliability of these criteria in the evaluation of the severity of spinal cord injury remains controversial [2,14-16]. In the present work, the relation between intensity of spinal cord trauma and changes in evoked potentials was examined experimentally in a model of spinal cord compression of improved reproducibility achieved through the presetting of both degree and duAddress reprint requests to: A. C. Nacimiento, M.D., Research Laboratory, Department of Neurosurgery, Saarland University Medical School, 665 Homburg/Saar, Federal Republic of Germany.
© 1986by ElsevierSciencePublishingCo., Inc.
ration of compression [ 11]. Also, peripheral nerve input and evoked cortical output were standardized in terms of well-defined neurophysiologic criteria. Observations focused on incomplete cord lesions, because of their overwhelming clinical importance, as well as their efficacy in evaluating the functional plasticity of axonal and synaptic conduction in an injured spinal cord still possessing potential ability for recovery. Materials and Methods The experiments were done on 17 cats of either sex, with body weights ranging from 2.5 to 4.0 kg, under pentobarbital anesthesia (3 5 mg/kg intraperitoneally for induction, 3 mg/kg/hr intravenously for maintenance). After insertion of tracheal, arterial, and venous cannulae, the animals were immobilized with intravenous gallamine triethiodide and placed on artificial respiration. End-tidal CO2 volume (4%), systemic arterial blood pressure (mean, 115 mmHg), rectal temperature (38°C), and electrocorticograms were monitored throughout the experiments. Cortical potentials were evoked by stimulation of the left tibial nerve without the rami to the gastrocnemii, popliteus, and plantaris muscles, with platinum electrodes in a well of warm mineral oil. Stimulus parameters were a train-of-four pulses of 0.5 msec in duration at a frequency of 400 Hz, a strength activating both group II and III afferents, and delivered at a rate of one every 3 seconds. The right posterior sigmoid and anterior ectosylvian gyri were exposed and covered with mineral oil at 37°C. Evoked activity was mapped against ground with a ball-tipped Ag-AgC1 electrode in the hindlimb projection fields of somatosensory area I, and the restricted cortical area yielding the largest primary cortical response was determined. The amplitude and latency of the positive wave were measured. T o monitor conduction changes in dorsal column axons, the dorsal column ipsilateral to the explored hemisphere was stimulated at L-1 with single pulses, and the resulting antidromic action potentials recorded at the S-3 dorsal root. After laminectomy from T-11 to S-3, compression was applied through a metal rod lowered under control 0090-3019/86/$3.50
Evoked Potentials in Spinal Cord Trauma
Surg Neurol 1986;25:62-66
surements were made in three uninjured cats on 16 successive responses, computer-averaged at 10 minute intervals for 6 to 7 hours. After initial experiments showed no changes in evoked potentials after compression of 1 mm and 2 mm, measurements were concentrated on compressions of 3 mm (7 cats) and 4 mm (7 cats). Observations at the 5 mm compression level were not further pursued on the grounds mentioned below (see Results). Evoked potentials and dorsal column axon potentials were computer-averaged for 2 hours prior to compression according to the same schedule used in the intact animals. Posttraumatic changes were measured by averaging 16 responses at the times plotted in Figure 1. Amplitude changes were normalized as percentages with the preinjury value taken as 100%. At the end of the experiments the spinal cord was removed after fixation by perfusion through the heart, and processed for histopathological evaluation of hemorrhage and edema on serially cut frozen cross-sections of L-6, L-7, and S-1 stained with hematoxylin-benzopurpurine.
with an operating microscope upon the dorsal surface of L-7, as described in detail previously [11]. The compressing rod covered entirely the dorsal aspect of the segment, and was electromagnetically activated by a rectangular current pulse, which in turn determined duration of compression. Degree of compression was preset by calibrating movement of the rod with a displacement transducer. Typically, a compression of 3 mm was reached in 7 msec, at a speed of about 0.75 m/sec. Decompression took about 15 msec. Duration of compression was kept constant at 50 msec. The degree of compression was varied between 1 mm and 5 mm, the latter limit set by the mean average dorsoventral diameter of the spinal cord at L-7, which was 5 mm in 10 animals. Basal meaFigure 1. Time course of posttraumatic changes in amplitude of the positive wave component of the somatosensory potentials (SEP) evoked by tibial nerve stimulation. Data points were normalized to the mean value ± standard error of the mean (SEM) plotted at t = O, as obtained in the 2 hours prior to compression by computer-averaging 20 SEP every I 0 minutes. Each data point after t = 0 plots the mean ± S~M of 112 single SEP in each of two groups of 7 cats at trauma doses of 3 mm a n d 4 mm. Statistical evaluation (Student's t-test) of differences between each lesion level and control curve (not shown) ( + ) and between trauma doses (*). All points in the 3 mm curve did not differ significantly from the control curve; +, *: P < 5%; + +, **." P < 1%; + + +: P < 0.1%. Compression at t = 0 (arrow). Inset: 20 superimposed oscillograph records of the primary response in the intact animal Parameters were measured on computer averaged records.
3mm
200-
Results
The Intact Preparation The amplitude of evoked primary responses showed the random fluctuations that reflect the functional state of
50ms
• 4mm
2
0.1mY
50ms lOms
~<
, , , ~, , , , , , ,
150-
mO IXIl~
50 U.
0:0
o
[''' 0
-1
t
I 5
F
I''' 10
I 30
63
'
'
I ''' 60
I ' ' ' 1 ' ' ' 1 ' ' 120
180
240
'1 300min
64
Surg Neurol 1986;25:62-66
thalamocortical and cortical neuronal activities under moderate pentobarbital anesthesia [ 10]. Preparations with stereotyped amplitudes, indicating functional deterioration, were discarded. Average latency of the positive wave was 10 msec, with a range of 9.5-11.5 msec.
The Injured Preparation
Compression of I mm and 2 mm did not influence the amplitude of the primary response. Clear amplitude changes were first detected after a compression of 3 mm, which reduced the segmental cross section by 6 0 % and brought about at 1 minute postcompression a decrease to 74% -+ 15% (SEM) of preinjury values. The corresponding decrease after a 4 mm (80%) compression was to 36% -+ 12.4% (Figure 1). Only the latter reduction was significant, both with respect to the basal state (P < 0.001) and to the corresponding value at 3 mm (P < 0.05). Further significant differences at various times thereafter are indicated in Figure 1. The time course of posttraumatic changes show that initially, average amplitudes tended to change step-wise, apparently reflecting the corresponding degrees of compression. However, no statistically significant differences between both levels could be measured 2.5 hours after trauma. After a 5 mm compression, evoked potentials were immediately abolished, and did not recover significantly for the duration of the measurements. Histopathologically, hemorrhage and edema in-
Nacimiento et al
creased consistently with each degree of compression, from 1 to 5 mm as shown in Figure 2. At 5 mm, the spinal cord segment appeared destroyed by hemorrhagic necrosis and was grossly edematous, preventing histologic processing. The virtual transection brought about by this compression level was not further studied because: correlation of function and structure was precluded; reproducibility of measurements was reduced; any potential recovery would depend upon regeneration processes evolving in the chronic state; and incomplete spinal cord injuries constitute the bulk of clinical experience [14,16].
Changes in Antidromic Action Potentials at S-3 Dorsal Root
After a compression of 3 mm, axonal conduction across the lesioned area was immediately blocked. Partial recovery, up to about 3 0 % - 4 0 % of preinjury values, was observed within 1 to 2 hours, with later decline and block for the duration of the observations. A compression of 4 mm was followed by conduction block persisting until the end of the experiments. Figure 2. Histopathological findings (edema in the white matter, hemorrhage in the gray matter). Outlines in 10 superimposed serial sections 40 tz thick taken from the center of L-7, after compression at the levels indicated, and projected upon a typical cross section of the same segment. Samples after 5 mm compression were taken about 4 mm away, on account of hemorrhagic necrosis at the lesion center. Black .fields: edema; dashed .fields: hemorrhage.
0
lmm
3mm
4mm
2mm
5mm
Evoked Potentials in Spinal Cord Trauma
Latency Whenever detectable responses were present, latency remained unchanged after all degrees of compression.
Discussion Although hemorrhage and edema were already present at a compression level of 1 mm (20%) (Figure 2), evoked potential changes were first detected after a 3 mm (60%) compression. However, once this threshold was reached, the range between transitory decrease (at 3 ram) and abolition or drastic attenuation (at 5 mm) was very narrow. These findings suggest both a high safety factor for afferent transmission, and a steep trauma dose-response relationship. Compression may block axonal transmission by a combination of structural disruption by the mechanical force, and depolarization by posttraumatic K ~ release [4,7,9]. Experiments utilizing discrete lesions identified, in addition to the dorsal column-medial lemniscus pathway, the spinocervicothalamic tract as a parallel afferent route for eliciting the primary response [ 12]. Our results suggest that both afferent pathways mentioned above were partially, but substantially, blocked by compression. However, attenuation of evoked potentials at any compression level except 5 mm was absent or minimal. The availability of these parallel pathways may contribute to the observed high safety factor for transmission. A straightforward extrapolation of this interpretation to observations in primates and humans is not easy. Eidelberg and Woodbury [5] reported that in monkeys neither an isolated section of a dorsal column nor its combination with a dorsolateral lesion could alter amplitude, distribution, or latency of cortical responses evoked by skin stimulation at the wrist. Abolition or gross attenuation was only seen when these lesions were combined with a section of the spinothalamic tract. In contrast, some observations in primates [ 1] and humans [8] attributed primary transmission of these responses to the dorsal column-medial lemniscus pathway. Inasmuch as clinical lesions are rarely discrete, these differences may be more apparent than real. Much more work with precise posttraumatic histology is certainly needed. A property of the somatosensory system that may also contribute to enhance posttraumatic transmission is its i n p u t - o u t p u t pattern. Studies were reported relating changes in amplitude of primary cortical evoked potentials (output) in response to increased afferent activity in skin and muscle afferents, brought about by step increases in stimulation strength (input). They showed that, at inputs amounting to about 30% of maximum, cortical output reached levels of about 7 0 % - 8 0 % of maximum [13]. This steep relationship indicates there
Surg Neurol 1986;25:62-66
65
is little need for spatial summation in synaptic transmission of afferent activity to cortical projection areas [ 13]. Thus, recovery from postinjury conduction block [4,7,9], or the functional survival of only a small contingent of afferent axons, or both, would be sufficient to evoke short-latency cortical potentials after posttraumatic afferent reduction. The absence of latency changes may be due to the composite nature of the latency of posttraumatic somatosensory evoked potentials, which probably represent axonal and synaptic conduction times along parallel afferent pathways variously affected by the relative topographical distribution of lesions. After the primary response, a further potential change--the generalized secondary discharge [6]--can be recorded. It showed considerable variation in amplitude, duration, configuration, and probability of appearance. This is not surprising, because this activity is not restricted to the sensory cortical fields and is strongly dependent on the level of spontaneous cortical activity. We found that these "late waves" could not be reliably quantified, and that their qualitative assessment did not add any useful information to that already provided by the primary response. Our results agree quite well with those obtained by Ducati et al in their experiments using balloon-induced spinal cord compression in cats [3]. We have examined statistically by analysis of variance and Student's t-test the data presented in their Table 1, and found no significant correlation between the percentage of change in amplitude of evoked potentials and the corresponding percentage amount of both total and hemorrhagic lesions. Furthermore, the data show the same narrow range between posttraumatic amplitude changes and transmission block. There is, however, one aspect in the results of Ducati et al [3] that should be commented upon. We have compared statistically the data in their Tables 1 and 2, and found no significant differences in the percentage of amplitude changes in evoked potentials between untreated and treated animals. In fact, this lack of significance may already be predicted by the just mentioned absence of correlation between magnitude of lesions and changes in amplitude of evoked potentials in the same experiments. Thus, particular care is needed in testing a treatment schedule using amplitude of evoked potentials in relation to degree of injury as parameters. In patients examined shortly after injury there is no general agreement on whether amplitude [14,16], latency [2], or configuration [ 1 4 - 1 6 ] should be taken as the main criteria in the evaluation and correlation of the short-latency components of somatosensory evoked responses with neurological findings in acute incomplete spinal cord injury in humans. However, our results agree with reports of unchanged posttraumatic latency [ 14,16].
66
Surg N e u r o l 1986;25:62-66
In summary, our experiments suggest not only the coexistence of a normal evoked response with a clear compression damage, but also that moderately attenuated responses may be evoked in the presence of widespread damage through edema and hemorrhage. The mere presence of evoked somatosensory activity does not indicate the actual magnitude of acute spinal cord trauma, and any evaluation of the potential for functional recovery after spinal cord injury exclusively based on evoked potential data would thus be seriously handicapped.
Dr. E. Eidelberg made most useful comments on this work. We thank Mrs. R. Hilpert, Mrs. P. Roeder, H-J. Dejon, and D. Leonhardt for technical assistance; H. Ludt for computer-based data processing and for designing the compression device; W. Czech for the illustration; and Mrs. D. Wright for the typescript. This work was supported by Grants Na 115/1, Na 115/3, and Na 115/4-3 from the Deutsche Forschungsgemeinschaft to A.C.N.
References 1. Cusick JF, Myklebust JB, Larson SJ, Sances A. Spinal cord evaluation by evoked responses. Arch Neurol 1979;36:140-3. 2. Dorfman LJ, Perkash I, Bosley ThM, Cummins KL. Use of cerebral evoked potentials to evaluate spinal somatosensory function in patients with traumatic and surgical myelopathies. J Neurosurg 1980;52:654-60. 3. Ducati A, Schieppati M, Giovanelli A. Effects of deep barbiturate coma on acute spinal cord injury in the cat. Surg Neurol 1984;21:405-13. 4. Eidelberg E, Sullivan J, Brigham A. Immediate consequences of
N a c i m i e n t o et al
spinal cord injury: possible role of potassium in axonal conduction block. Surg Neurol 1975;3:317-21. 5. Eidelberg E, Woodbury CM. Apparent redundancy in the somatosensory system in monkeys. Exp Neurol 1972;37:573-81. 6. Forbes A, Morison RB. Cortical response to sensory stimulation under deep barbiturate narcosis. J Neurophysiol 1939;25:112-28. 7. Gelfan S, Tarlov 1M. Physiology of spinal cord, nerve root and peripheral nerve compression. Am J Physiol 1956;185:217-29. 8. Halliday AM. Changes in the form of cerebral evoked responses in man associated with various lesions of the nervous system. Electroenceph Clin Neurophysiol 1967;Suppl 25:178-92. 9. Kobrine AI, Evans DE, Rizzoli HV. Experimental acute balloon compression of the spinal cord. J Neurosurg 1979;51:841-5. 10. Nacimiento AC, Lux HD, Creutzfeldt OD. Postsynaptische Potentiale von Nervenzellen des motorischen Cortex nach elektrischer Reizung spezifischer und unspezifischer Thalamuskernen. Pflfigers Archiv 1964;281:152-69. 11. Nacimiento AC, Bartels M, Herrmann HD, Loew F. Reflex activity and axonal conduction in the L-7 spinal cord segment following experimental compression trauma. J Neurosurg 1985;62: 898-905. 12. Norrsell U, Wolpow ER. An evoked potential study of different pathways from the hindlimb to the somatosensory areas in the cat. Acta Physiol Scand 1966;66:19-33. 13. Oscarsson O, Rosen L. Short latency projection to the cat's cerebral cortex from skin and muscle afferents in the contralateral forelimb. J Physiol 1966;182:164-84. 14. Rowed DW. Value of somatosensory evoked potentials for prognosis in partial cord injuries. In: Tator Ch H, ed. Early management of acute spinal cord injury. New York: Raven Press, 1982. 15. Spielholz NI, Benjamin MV, Engler G, RansohoffJ. Somatosensory evoked potentials and clinical outcome in spinal cord injury. In: Popp AJ, Bourke RS, eds. Neural Trauma. New York: Raven Press, 1979. 16. Young W. Correlation of somatosensory evoked potentials and neurological findings in spinal cord injury. In: Tator Ch H, ed. Early management of acute spinal cord injury. New York: Raven Press, 1982.
Surg Neurol 1986;25:62-66
Acute Changes in Somatosensory Evoked Potentials following Graded Experimental Spinal Cord Compression Amadeo C. Nacimiento, M.D., Matthias Bartels, M.D., and Friedrich Loew, M.D. Research Laboratory, Department of Neurosurgery, Saarland University Medical School, Homburg/Saar, Federal Republic of Germany
Nacimiento AC, Bartels M, Loew F. Acute changes in somatosensory evoked potentials followinggraded experimental spinal cord compression. Surg Neurol 1986;25:62-66. Amplitude and latency of cortical somatosensory potentials evoked in cats by peripheral nerve stimulation were measured before, during, and for 5 hours after injury of spinal cord segment L-7 by a predetermined degree and duration of compression. An amplitude decrease, slight and transitory, was first observed after compression reduced the segmental cross section by 60%. After an 80% compression, amplitude reduction was initially larger and lasted longer, but recovered 2.5 hours after injury to a level that did not differ statistically from control values. After total (100%) compression, evoked responses disappeared abruptly and did not recover significantly. Latency was unaltered at all degrees of compression. Structural damage increased with the degree of compression. In this model, evoked potential changes neither reflect nor predict the magnitude of acute incomplete spinal cord injury. KEY WORDS: Spinal cord trauma; Spinal cord compression; Cortical evoked potentials; Spinal cord compression model; Evoked potentials; Spinal cord injury
The clinical use o f somatosensory evoked potentials to predict the outcome of spinal cord injury, evaluate degree o f trauma, or assess methods of treatment presents several unresolved problems. Neurophysiologic properties of clinical relevance include their latency, amplitude, sequential configuration, and polarity. The degree of reliability of these criteria in the evaluation of the severity of spinal cord injury remains controversial [2,14-16]. In the present work, the relation between intensity of spinal cord trauma and changes in evoked potentials was examined experimentally in a model of spinal cord compression of improved reproducibility achieved through the presetting of both degree and duAddress reprint requests to: A. C. Nacimiento, M.D., Research Laboratory, Department of Neurosurgery, Saarland University Medical School, 665 Homburg/Saar, Federal Republic of Germany.
© 1986by ElsevierSciencePublishingCo., Inc.
ration of compression [ 11]. Also, peripheral nerve input and evoked cortical output were standardized in terms of well-defined neurophysiologic criteria. Observations focused on incomplete cord lesions, because of their overwhelming clinical importance, as well as their efficacy in evaluating the functional plasticity of axonal and synaptic conduction in an injured spinal cord still possessing potential ability for recovery. Materials and Methods The experiments were done on 17 cats of either sex, with body weights ranging from 2.5 to 4.0 kg, under pentobarbital anesthesia (3 5 mg/kg intraperitoneally for induction, 3 mg/kg/hr intravenously for maintenance). After insertion of tracheal, arterial, and venous cannulae, the animals were immobilized with intravenous gallamine triethiodide and placed on artificial respiration. End-tidal CO2 volume (4%), systemic arterial blood pressure (mean, 115 mmHg), rectal temperature (38°C), and electrocorticograms were monitored throughout the experiments. Cortical potentials were evoked by stimulation of the left tibial nerve without the rami to the gastrocnemii, popliteus, and plantaris muscles, with platinum electrodes in a well of warm mineral oil. Stimulus parameters were a train-of-four pulses of 0.5 msec in duration at a frequency of 400 Hz, a strength activating both group II and III afferents, and delivered at a rate of one every 3 seconds. The right posterior sigmoid and anterior ectosylvian gyri were exposed and covered with mineral oil at 37°C. Evoked activity was mapped against ground with a ball-tipped Ag-AgC1 electrode in the hindlimb projection fields of somatosensory area I, and the restricted cortical area yielding the largest primary cortical response was determined. The amplitude and latency of the positive wave were measured. T o monitor conduction changes in dorsal column axons, the dorsal column ipsilateral to the explored hemisphere was stimulated at L-1 with single pulses, and the resulting antidromic action potentials recorded at the S-3 dorsal root. After laminectomy from T-11 to S-3, compression was applied through a metal rod lowered under control 0090-3019/86/$3.50
Evoked Potentials in Spinal Cord Trauma
Surg Neurol 1986;25:62-66
surements were made in three uninjured cats on 16 successive responses, computer-averaged at 10 minute intervals for 6 to 7 hours. After initial experiments showed no changes in evoked potentials after compression of 1 mm and 2 mm, measurements were concentrated on compressions of 3 mm (7 cats) and 4 mm (7 cats). Observations at the 5 mm compression level were not further pursued on the grounds mentioned below (see Results). Evoked potentials and dorsal column axon potentials were computer-averaged for 2 hours prior to compression according to the same schedule used in the intact animals. Posttraumatic changes were measured by averaging 16 responses at the times plotted in Figure 1. Amplitude changes were normalized as percentages with the preinjury value taken as 100%. At the end of the experiments the spinal cord was removed after fixation by perfusion through the heart, and processed for histopathological evaluation of hemorrhage and edema on serially cut frozen cross-sections of L-6, L-7, and S-1 stained with hematoxylin-benzopurpurine.
with an operating microscope upon the dorsal surface of L-7, as described in detail previously [11]. The compressing rod covered entirely the dorsal aspect of the segment, and was electromagnetically activated by a rectangular current pulse, which in turn determined duration of compression. Degree of compression was preset by calibrating movement of the rod with a displacement transducer. Typically, a compression of 3 mm was reached in 7 msec, at a speed of about 0.75 m/sec. Decompression took about 15 msec. Duration of compression was kept constant at 50 msec. The degree of compression was varied between 1 mm and 5 mm, the latter limit set by the mean average dorsoventral diameter of the spinal cord at L-7, which was 5 mm in 10 animals. Basal meaFigure 1. Time course of posttraumatic changes in amplitude of the positive wave component of the somatosensory potentials (SEP) evoked by tibial nerve stimulation. Data points were normalized to the mean value ± standard error of the mean (SEM) plotted at t = O, as obtained in the 2 hours prior to compression by computer-averaging 20 SEP every I 0 minutes. Each data point after t = 0 plots the mean ± S~M of 112 single SEP in each of two groups of 7 cats at trauma doses of 3 mm a n d 4 mm. Statistical evaluation (Student's t-test) of differences between each lesion level and control curve (not shown) ( + ) and between trauma doses (*). All points in the 3 mm curve did not differ significantly from the control curve; +, *: P < 5%; + +, **." P < 1%; + + +: P < 0.1%. Compression at t = 0 (arrow). Inset: 20 superimposed oscillograph records of the primary response in the intact animal Parameters were measured on computer averaged records.
3mm
200-
Results
The Intact Preparation The amplitude of evoked primary responses showed the random fluctuations that reflect the functional state of
50ms
• 4mm
2
0.1mY
50ms lOms
~<
, , , ~, , , , , , ,
150-
mO IXIl~
50 U.
0:0
o
[''' 0
-1
t
I 5
F
I''' 10
I 30
63
'
'
I ''' 60
I ' ' ' 1 ' ' ' 1 ' ' 120
180
240
'1 300min
64
Surg Neurol 1986;25:62-66
thalamocortical and cortical neuronal activities under moderate pentobarbital anesthesia [ 10]. Preparations with stereotyped amplitudes, indicating functional deterioration, were discarded. Average latency of the positive wave was 10 msec, with a range of 9.5-11.5 msec.
The Injured Preparation
Compression of I mm and 2 mm did not influence the amplitude of the primary response. Clear amplitude changes were first detected after a compression of 3 mm, which reduced the segmental cross section by 6 0 % and brought about at 1 minute postcompression a decrease to 74% -+ 15% (SEM) of preinjury values. The corresponding decrease after a 4 mm (80%) compression was to 36% -+ 12.4% (Figure 1). Only the latter reduction was significant, both with respect to the basal state (P < 0.001) and to the corresponding value at 3 mm (P < 0.05). Further significant differences at various times thereafter are indicated in Figure 1. The time course of posttraumatic changes show that initially, average amplitudes tended to change step-wise, apparently reflecting the corresponding degrees of compression. However, no statistically significant differences between both levels could be measured 2.5 hours after trauma. After a 5 mm compression, evoked potentials were immediately abolished, and did not recover significantly for the duration of the measurements. Histopathologically, hemorrhage and edema in-
Nacimiento et al
creased consistently with each degree of compression, from 1 to 5 mm as shown in Figure 2. At 5 mm, the spinal cord segment appeared destroyed by hemorrhagic necrosis and was grossly edematous, preventing histologic processing. The virtual transection brought about by this compression level was not further studied because: correlation of function and structure was precluded; reproducibility of measurements was reduced; any potential recovery would depend upon regeneration processes evolving in the chronic state; and incomplete spinal cord injuries constitute the bulk of clinical experience [14,16].
Changes in Antidromic Action Potentials at S-3 Dorsal Root
After a compression of 3 mm, axonal conduction across the lesioned area was immediately blocked. Partial recovery, up to about 3 0 % - 4 0 % of preinjury values, was observed within 1 to 2 hours, with later decline and block for the duration of the observations. A compression of 4 mm was followed by conduction block persisting until the end of the experiments. Figure 2. Histopathological findings (edema in the white matter, hemorrhage in the gray matter). Outlines in 10 superimposed serial sections 40 tz thick taken from the center of L-7, after compression at the levels indicated, and projected upon a typical cross section of the same segment. Samples after 5 mm compression were taken about 4 mm away, on account of hemorrhagic necrosis at the lesion center. Black .fields: edema; dashed .fields: hemorrhage.
0
lmm
3mm
4mm
2mm
5mm
Evoked Potentials in Spinal Cord Trauma
Latency Whenever detectable responses were present, latency remained unchanged after all degrees of compression.
Discussion Although hemorrhage and edema were already present at a compression level of 1 mm (20%) (Figure 2), evoked potential changes were first detected after a 3 mm (60%) compression. However, once this threshold was reached, the range between transitory decrease (at 3 ram) and abolition or drastic attenuation (at 5 mm) was very narrow. These findings suggest both a high safety factor for afferent transmission, and a steep trauma dose-response relationship. Compression may block axonal transmission by a combination of structural disruption by the mechanical force, and depolarization by posttraumatic K ~ release [4,7,9]. Experiments utilizing discrete lesions identified, in addition to the dorsal column-medial lemniscus pathway, the spinocervicothalamic tract as a parallel afferent route for eliciting the primary response [ 12]. Our results suggest that both afferent pathways mentioned above were partially, but substantially, blocked by compression. However, attenuation of evoked potentials at any compression level except 5 mm was absent or minimal. The availability of these parallel pathways may contribute to the observed high safety factor for transmission. A straightforward extrapolation of this interpretation to observations in primates and humans is not easy. Eidelberg and Woodbury [5] reported that in monkeys neither an isolated section of a dorsal column nor its combination with a dorsolateral lesion could alter amplitude, distribution, or latency of cortical responses evoked by skin stimulation at the wrist. Abolition or gross attenuation was only seen when these lesions were combined with a section of the spinothalamic tract. In contrast, some observations in primates [ 1] and humans [8] attributed primary transmission of these responses to the dorsal column-medial lemniscus pathway. Inasmuch as clinical lesions are rarely discrete, these differences may be more apparent than real. Much more work with precise posttraumatic histology is certainly needed. A property of the somatosensory system that may also contribute to enhance posttraumatic transmission is its i n p u t - o u t p u t pattern. Studies were reported relating changes in amplitude of primary cortical evoked potentials (output) in response to increased afferent activity in skin and muscle afferents, brought about by step increases in stimulation strength (input). They showed that, at inputs amounting to about 30% of maximum, cortical output reached levels of about 7 0 % - 8 0 % of maximum [13]. This steep relationship indicates there
Surg Neurol 1986;25:62-66
65
is little need for spatial summation in synaptic transmission of afferent activity to cortical projection areas [ 13]. Thus, recovery from postinjury conduction block [4,7,9], or the functional survival of only a small contingent of afferent axons, or both, would be sufficient to evoke short-latency cortical potentials after posttraumatic afferent reduction. The absence of latency changes may be due to the composite nature of the latency of posttraumatic somatosensory evoked potentials, which probably represent axonal and synaptic conduction times along parallel afferent pathways variously affected by the relative topographical distribution of lesions. After the primary response, a further potential change--the generalized secondary discharge [6]--can be recorded. It showed considerable variation in amplitude, duration, configuration, and probability of appearance. This is not surprising, because this activity is not restricted to the sensory cortical fields and is strongly dependent on the level of spontaneous cortical activity. We found that these "late waves" could not be reliably quantified, and that their qualitative assessment did not add any useful information to that already provided by the primary response. Our results agree quite well with those obtained by Ducati et al in their experiments using balloon-induced spinal cord compression in cats [3]. We have examined statistically by analysis of variance and Student's t-test the data presented in their Table 1, and found no significant correlation between the percentage of change in amplitude of evoked potentials and the corresponding percentage amount of both total and hemorrhagic lesions. Furthermore, the data show the same narrow range between posttraumatic amplitude changes and transmission block. There is, however, one aspect in the results of Ducati et al [3] that should be commented upon. We have compared statistically the data in their Tables 1 and 2, and found no significant differences in the percentage of amplitude changes in evoked potentials between untreated and treated animals. In fact, this lack of significance may already be predicted by the just mentioned absence of correlation between magnitude of lesions and changes in amplitude of evoked potentials in the same experiments. Thus, particular care is needed in testing a treatment schedule using amplitude of evoked potentials in relation to degree of injury as parameters. In patients examined shortly after injury there is no general agreement on whether amplitude [14,16], latency [2], or configuration [ 1 4 - 1 6 ] should be taken as the main criteria in the evaluation and correlation of the short-latency components of somatosensory evoked responses with neurological findings in acute incomplete spinal cord injury in humans. However, our results agree with reports of unchanged posttraumatic latency [ 14,16].
66
Surg N e u r o l 1986;25:62-66
In summary, our experiments suggest not only the coexistence of a normal evoked response with a clear compression damage, but also that moderately attenuated responses may be evoked in the presence of widespread damage through edema and hemorrhage. The mere presence of evoked somatosensory activity does not indicate the actual magnitude of acute spinal cord trauma, and any evaluation of the potential for functional recovery after spinal cord injury exclusively based on evoked potential data would thus be seriously handicapped.
Dr. E. Eidelberg made most useful comments on this work. We thank Mrs. R. Hilpert, Mrs. P. Roeder, H-J. Dejon, and D. Leonhardt for technical assistance; H. Ludt for computer-based data processing and for designing the compression device; W. Czech for the illustration; and Mrs. D. Wright for the typescript. This work was supported by Grants Na 115/1, Na 115/3, and Na 115/4-3 from the Deutsche Forschungsgemeinschaft to A.C.N.
References 1. Cusick JF, Myklebust JB, Larson SJ, Sances A. Spinal cord evaluation by evoked responses. Arch Neurol 1979;36:140-3. 2. Dorfman LJ, Perkash I, Bosley ThM, Cummins KL. Use of cerebral evoked potentials to evaluate spinal somatosensory function in patients with traumatic and surgical myelopathies. J Neurosurg 1980;52:654-60. 3. Ducati A, Schieppati M, Giovanelli A. Effects of deep barbiturate coma on acute spinal cord injury in the cat. Surg Neurol 1984;21:405-13. 4. Eidelberg E, Sullivan J, Brigham A. Immediate consequences of
N a c i m i e n t o et al
spinal cord injury: possible role of potassium in axonal conduction block. Surg Neurol 1975;3:317-21. 5. Eidelberg E, Woodbury CM. Apparent redundancy in the somatosensory system in monkeys. Exp Neurol 1972;37:573-81. 6. Forbes A, Morison RB. Cortical response to sensory stimulation under deep barbiturate narcosis. J Neurophysiol 1939;25:112-28. 7. Gelfan S, Tarlov 1M. Physiology of spinal cord, nerve root and peripheral nerve compression. Am J Physiol 1956;185:217-29. 8. Halliday AM. Changes in the form of cerebral evoked responses in man associated with various lesions of the nervous system. Electroenceph Clin Neurophysiol 1967;Suppl 25:178-92. 9. Kobrine AI, Evans DE, Rizzoli HV. Experimental acute balloon compression of the spinal cord. J Neurosurg 1979;51:841-5. 10. Nacimiento AC, Lux HD, Creutzfeldt OD. Postsynaptische Potentiale von Nervenzellen des motorischen Cortex nach elektrischer Reizung spezifischer und unspezifischer Thalamuskernen. Pflfigers Archiv 1964;281:152-69. 11. Nacimiento AC, Bartels M, Herrmann HD, Loew F. Reflex activity and axonal conduction in the L-7 spinal cord segment following experimental compression trauma. J Neurosurg 1985;62: 898-905. 12. Norrsell U, Wolpow ER. An evoked potential study of different pathways from the hindlimb to the somatosensory areas in the cat. Acta Physiol Scand 1966;66:19-33. 13. Oscarsson O, Rosen L. Short latency projection to the cat's cerebral cortex from skin and muscle afferents in the contralateral forelimb. J Physiol 1966;182:164-84. 14. Rowed DW. Value of somatosensory evoked potentials for prognosis in partial cord injuries. In: Tator Ch H, ed. Early management of acute spinal cord injury. New York: Raven Press, 1982. 15. Spielholz NI, Benjamin MV, Engler G, RansohoffJ. Somatosensory evoked potentials and clinical outcome in spinal cord injury. In: Popp AJ, Bourke RS, eds. Neural Trauma. New York: Raven Press, 1979. 16. Young W. Correlation of somatosensory evoked potentials and neurological findings in spinal cord injury. In: Tator Ch H, ed. Early management of acute spinal cord injury. New York: Raven Press, 1982.