Direct noninvasive monitoring of spinal cord motor function during thoracic aortic occlusion: Use of motor evoked potentials

Transcription

1 Direct noninvasive monitoring of spinal cord motor function during thoracic aortic occlusion: Use of motor evoked potentials John C. Laschinger, M.D., Jeffrey Owen, Ph.D., Michael Rosenbloom, M.D., James L. Cox, M.D., and Nicholas T. Kouchoukos, M.D., St. Louis, Mo. Spinal cord monitoring during thoracic aneurysmectomy by somatosensory evoked potentials has been criticized for its failure to measure anterior (motor) spinal cord function. We have developed a clinically applicable, noninvasive technique for intraoperative monitoring of motor evoked potentials (MEP), which allows direct functional assessment of spinal cord motor tracts during thoracic aortic occlusion. Twelve dogs tmderwent continuous intraoperative monitoring of MEP before, during, and after thoracic aortic crossclamping. Motor tract response to noninvasive cord stimulation (5 to 10 ma, 0.02 msec, 4.3 H2) was recorded by subcutaneous electrodes placed along the length of the spine (T-10, L-l, and L-4). Six animals (group I) subjected to aortic cross-damping alone demonstrated a characteristic time- and level-dependent deterioration and loss of MEP. Ischemic cord dysfunction (as determined by time from damping to loss of MEP) progressed from the distal to the proximal cord (L-4 = : 1.5 minutes; L-1 = minutes; T-10 = minutes; p < 0.05 between all levels). Reperfusion of the distal aorta 20 minutes after clamping resulted in MEP return that progressed from the proximal (T-10) to distal (L-1 and L-4) cord. In another six animals (group II), distal perfusion (mean blood pressure = 95 mm Hg) was maintained for 1 hour after cross-clamping by left atrial-femoral artery bypass. Normal configuration and amplitude of MEP was maintained throughout the cross-clamping period. These data suggest that distinctive changes in MEP indicative of reversible ischemia of spinal cord motor tracts occur after aortic cross-clamping. Such ischemia begins in the most distal cord, exhibits upward progression with time, and can be prevented by maintenance of adequate distal aortic perfusion. Clinical use of MEP monitoring during thoracic aneurysmectomy may provide a method for intraoperative assessment of the adequacy of motor tract perfusion. (J VAse SvR~ 1988;7: ) Spinal cord injury remains a major cause of morbidity after operations of the descending thoracic and thoracoabdominal aorta. Although monitoring of spinal cord function by measurement of somatosensory evoked potentials (SEP) has been used in an attempt to determine the causes ofintraoperative spinal cord ischemia and paraplegia, the technique has been criticized because of failure to monitor directly the function of anterior spinal cord motor tracts. 1,2 This artide summarizes the development of a noninvasive technique for intraoperative monitoring of From the Division of Cardiothoracic Surgery, Washington University School of Medicine. Presented at the Thirty-fifth Scientific Meeting of the North American Chapter, The International Society for Cardiovascular Surgery, Toronto, Ontario, Canada, June 8-9, Reprint requests: Nicholas T. Kouchoukos, M.D., The Jewish Hospital and Washington University Medical Center, 216 S. Kingshighway Blvd., P.O. Box 14109, St. Louis, MO spinal cord motor tract function with the use of motor evoked potentials (MEP) that may be clinically applicable. The technique permits direct on-line assessment of the adequacy of spinal cord motor tract perfusion during surgical procedures on the descending thoracic and thoracoabdominal aorta. MATERIAL AND METHODS Experimental protocol. Twelve dogs weighing 25 to 30 kg were anesthetized with intravenous sodium pentothal (15 mg/kg) and intravenous atropine (0.4 ml). After intubation, anesthesia was mainrained with enflurane (Ethrane) (0.5% to 1.5%) and respiration was controlled with room air positivepressure ventilation and low flow oxygen. Intravenous succinylcholine (0.1 ml/kg) was administered as needed to achieve and maintain muscular paralysis. A left thoracotomy was performed in the third intercostal space and the heart was suspended in a peri- 161

2 162 Laschinger et al Ioumal of VASCULAR SURGERY 4.~~Anterior CoJumn Response z5 coo,,cot,, I. / Stimuli 5-10 ma [ Signal Signal Fig. 1. Schematic representation of MEP monitoring system. See text for details. ",14-,.F7---- I l---p, - I 5 IO LATENCY (msec) Fig. 2. Normal MEP recordings. Note progression of conduction down the spinal cord (T-10 to L-4). Normal response is biphasic at all spinal cord levels. cardial cradle. Plastic catheters (16-gauge) were placed in the aortic root and left femoral artery for measurement of proximal and distal aortic pressure. MEPs were measured in all animals with a clinical evoked potential system (Nicolet Pathfinder, Nicolet Instruments; Madison, Wis.). MEP traces were generated by direct spinal cord stimulation (Fig. 1): This was achieved by complete exposure of the T-3 (an- ode) and T-4 (cathode) intervertebral disks adjacent to the left innominate artery and insertion of two 27-gauge platinum needle electrodes 6 mm in length (Grass Instruments; Quincy, Mass.). This pair of electrodes was then stimulated by a bipolar squarewave constant-current input channel (4.3 Hz, 5 to 10 ma, 0.02 msec). Anterior spinal cord conduction was recorded with similar paired electrodes placed subcutaneously along the length of the spine (T-10, L-l, and L-4) adjacent to the spinous processes. A separate grounding electrode was placed in the midportion of the abdomen. The Nicolet sensor system was used to amplify, analyze, and provide a visual display of anterior spinal cord conduction. A 20-millisecond period after each stimulus was analyzed. TwentT-five responses were recorded and electronically averaged and a composite MEP was constructed. The pass band filter was 10 Hz to 1.5 khz. All traces were stored on floppy disks for later retrieval and analysis. After a 30-minute stabilization period, baseline recordings of MEP were made. Updated recordings of MEP were made at 30- to 120-second intervals throughout the experimental period for comparison to baseline. No-stimulus control traces were recorded in each experiment for establishment of background noise levels. After baseline recordings, the descending thoracic aorta was occluded in all animals by placement of a single aortic cross-clamp just distal to the left innominate artery. Group I. In six animals, no further interventions were performed after aortic cross-clamping except for administration of sodium nitroprusside in doses ap-

3 Volume 7 Number 1 January, 1988 Spinal cord monitoring by MEP 163 Table I. Summary of MEP configuration changes after simple aortic cross-clamping without distal perfusion Aortic cross-damping Reperfusion Time to P2 loss Time to Pz loss Time to PI return Time to P2 return (min) (min) (min) (min) L ± ± ± 0.3 No return L ± ± 0.9 ~ 1.5 ± 0.3 No return T ± 0.5* 16.9 ± 0.9t 0.6 ± 0.3J- No return NOTE: All values expressed as mean ± SEM. *p < 0.05 vs L-4.?p < 0.05 vs L-1 and L-4. Table II. Summary of MEP latency and amplitude changes after simple aortic cross-clamping without distal perfusion % Change Baseline 20 rain AXC 20 rain p RP from baseline [Latency P~ T-10 t Amplitude P ± 0.5 Flat 1.86 ± 0.05 ~ 6.i ± ± ± ± 40.0 L-1 L-4 Latency P2 [Amplitude P2 Latency P Amplitude P~ / Latency P2 [Amplitude P~ t Latency P~ Amplitude PI /Latency P2 [Amplitude P ± 0.09 Flat No return ± ± 0.09 Flat 2.91 ± 0.10" 7.0 ± ± ± 8.11 i3.8 ± ± 0.12 Flat No return ± _ Flat 4.36 ± 0.22* 9.1 ± ± ± ± ] ± 0.18 Flat No return 6.49 ± 2.57 NOTE: All values expressed as mean SEM. clamping; t/2 = reperfusion. *p <0.002 vs baseline. Latency measured in rifilliseconds; amplitude measured in microvolts; AXC = aortic cross- propriate for management of proximal aortic hypertension. Serial MEP measurements were performed in all animals throughout the 20-minute aortic crossclamping and 20-minute reperfusion interval for comparison to baseline levels. Group II. Partial bypass (left atrial [32F venous cannula]-femoral artev [ 18F arterial cannula]) was performed in six animals immediately after the aortic cross-clamp was placed. This system provided partial unloading of the left ventricle while distal aortic flow was maintained below the cross-clamp. No pharmacologic agents were used for blood pressure control in these animals and proximal and distal perfusion pressures were controlled by administration of fluids and regulation of roller-pump flow. Serial MEP measurements were performed in all animals throughout a i-hour cross-clamp interval for comparison to baseline values. Statistical significance was determined by use of analysis of variance and all values are reported as mean _ the standard error of the mean. RESULTS A normal MEP trace is depicted in Fig. 2. Three parameters, configuration, latency of onset (measured in milliseconds), and amplitude of the generated MEP response, were serially monitored. These parameters of the MEP response were highly reproducible in all dogs throughout the 30-minute stabilization period before cross-clamping. The typical configuration of the normal MEP response is biphasic at all levels with an early (P1) and late (P2) response peak reflecting the response of fast and slow fiber anterior column conduction. The latency of onset of the MEP response at each level monitored (T-10, L-l, and L-4) is a function of the distance

4 ~ournai of 164 Laschinger et al. VASCULAR SURGERY I I I I I I I I I I I 5 I0 LATENCY (msec) Fig. 3. MEP response signifying early ischemia. Note that progression of conduction down the cord is maintained but second peak (P2) is absent at all levels. I 5 I0 LATENCY (msec) Fig. 4. MEP response signifying late ischemia. Note loss of conduction at all levels with loss of both peaks (P, and P2). from that level to the point of stimulation (T-3 to T-4). Latency is slightly different at each level reflecting the propagation of a wave of stimulation down anterior spinal columns in response to proximal stimulation. Group I. Mean proximal blood pressure was maintained at physiologic levels ( mm Hg)in all animals throughout the cross-clamping interval by administration of appropriate doses of sodium nitroprusside and regulation of fluid administration. Cross-clamping of the descending thoracic aorta in the absence of distal perfusion for as little as 3 minutes resulted in MEP configuration changes indicative of early spinal cord ischemia (Table I). Early ischemia was most reliably indicated by a change in configuration of the normal MEP response, specifically loss of the second peak (P2) of the bipolar response (Fig. 3). With further prolongation of the ischemic interval, the first deflection (PI) of the MEP response was also lost, signifying complete cessation of spinal cord motor tract conduction in response to proximal stimulation (Fig. 4). The changes in MEP configuration indicative of early ischemia (P2 loss) and late ischemia (Pa loss) initially affected the most distal cord levels (L-4) after cross-clamping. Further prolongation of the crossclamping interval resulted in a stepwise progression of early P2 and late PI conduction loss to more proximal cord levels (L-1 and T-10). Reperfusion of the distal aorta for 20 minutes after cross-clamping resuited in return of the P~ MEP response in all animals with failure of the P2 MEP response to return during the observation period. Return of the P1 MEP response after reperfusion progressed from the proximal to distal cord (Table I). A summary of the actual measurements of P1 and P2 latency and amplitude for all spinal cord levels throughout the experimental protocol is summarized in Table II. Simple crossclamping resulted in complete loss of motor tract conduction at all levels within the 20-minute crossclamp interval. Reperfusion resulted in return of normal P1 amplitude with slight but significant (p < 0.002) prolongation of latency. P2 return was not observed in any animal during the 20-minute reperfusion intervals. Group II. Adequate distal perfusion (mean distal pressure = mm Hg and pump flow = ml/kg/min) was maintained in all animals throughout the 1-hour cross-clamp interval by means of partial left atrial-femoral artery bypass. Partial left ventricular unloading and regulation of fluids effectively reduced mean proximal aortic pressure (110 _+ 8.4 mm Hg) without the need for pharmacologic intervention. As a result, no changes in MEP configuration were noted throughout the entire cross-damp interval (PI and P2 preserved). The changes in MEP latency and amplitude after crossclamping with maintenance of adequate distal per-

5 Volume 7 Number 1 January 1988 Sp#aal cord monitoring by MEP 165 Table IlL Summary of MEP latency and amplitude changes after simple aortic cross-clamping with adequate distal perfusion T-]0 L-1 L-3 to L-4 [ Baseline 1 hr aortic cross-damping (bypass) Latency P~ 1.65 _ _ ~ Amplitude P~ _ _ 6.73 Latency P _ ~ Amplitude 1) _ _ Latency P _ ~ Amplitude P~ 29.6 _ _ Latency P2 3.5i _ _ ~ Amplitude P _ _ Latency P~ 3.01 _ _ 0.16 Amplitude P _ _ Latency P _ _ ~+ Amplitude P _ _ NOTE: All values expressed as mean +_ SEM. Latency measured in milliseconds; amplitude measured in microvolts. ~p < vs baseline. fusion are summarized in Table III. No amplitude changes were observed; a small but significant (p < 0.002) prolongation of latency was observed at all levels. These changes were most likely a result of the effects of mild hypothermia (33 to 35 C) because a heat exchanger was not used in the simple partial bypass circuit. DISCUSSION The recent introduction of spinal cord monitoring by somatosensory evoked potentials (SEP) during surgical procedures on the descending thoracic and thoracoabdominal aorta represents a significant advancey The ability to assess the adequacy of spinal cord perfusion in an on-line fashion has allowed both experimental and clinical assessment of the factors contributing to intraoperative spinal cord ischemia and paraplegia. 6-9 Furthermore, use of SEP monitoring has allowed the systematic evaluation of the effectiveness of surgical techniques and adjuncts designed either to prevent spinal cord ischemia or ameliorate its effects. 1 q2 However, spinal cord monitoring by SEP has several limitations),2 SEP monitoring evaluates posterior and lateral spinal column function whereas paraplegia results from anterior column ischemia. SEP monitoring requires intact peripheral nerve and cortical function. Thus intraoperative functional changes in these structures could result in ischemic changes by SEP monitoring that may not be related to spinal cord ischemia. Halogenated anesthetic gases can sig- nificantly alter SEP. Thcse factors may result in either false-negative responses (normal SEP despite anterior column ischemia) or false-positive responses (abnormal SEP with normal anterior column function), thereby limiting its effectiveness. The ability to monitor motor tract function directly by MEP avoids some of the limitations of SEP, thereby allowing direct on-line assessment of the adcquacy of intraoperative spinal cord motor tract perfusion. The possibility for false-negative or falsepositive results is eliminated as the anterior (motor) cord is directly stimulated and monitored without the requirement for either peripheral nerve or cortical input. As a result, there is no requirement to maintain distal perfusion solely for the purpose of allowing effective spinal cord monitoring. In separate studies, we have shown MEP to be unaffected by any of the commonly used anesthetic agents (Laschinger JC, Owen J. Unpublished data). The clinical ability to directly assess spinal cord motor tract function every 30 to 120 seconds would render unlikely the unrecognized occurrence of significant intraoperative spinal cord ischemia. The results of this study with MEP as the spinal cord monitoring modality clearly confirm that simple cross-clamping of the descending thoracic aorta without distal perfusion results in significant ischemia of spinal cord motor tracts. The ability to monitor motor tract impulse conduction from various cord levels (T-10, L-l, and L-4) after proximal (T-3 to T-4) stimulation has shown clearly that ischemia of an-

6 166 Laschinger et al. Journal of VASCULAR SURGERY Table IV. Nomenclature of spinal cord conduction responses after thoracic aortic occlusion Type I Type II Type III Type 1V Progressive deterioration and loss of spinal cord conduction because of spinal cord ischemia that occurs as a result of proximal aortic cross-clamping without distal perfusion. Maintenance of normal configuration, latency, and amplitude after proximal aortic crossdamping that is observed when adequate distal perfusion is maintained by partial bypass or physiologic shunting via collaterals (coarctation). This response indicates that critical intercostal or lumbar vessels are not located in the aortic segment excluded from the systemic and bypass circulations. Loss of normal configuration, latency, and amplitude after proximal aortic cross-clamping despite documented adequacy of distal perfusion (distal mean pressure >70 mm Hg). This response indicates that vessels critical to spinal cord perfusion lie in the excluded aortic segment. Gradual loss of normal configuration, latency, and amplitude after proximal aortic that occurs despite use of adjunctive distal perfusion. This response is observed when distal perfusion is inadequate (distal mean pressure < 70 mm Hg) for whatever reason. terior spinal cord motor tracts results from simple aortic cross-clamping. Such ischemia affects the most distal cord (L-4) first with progression to more proximal cord levels (L-1 and T-I0) as the ischemic interval is prolonged. This finding is consistent with the known patterns of spinal cord blood flow changes that occur after simple aortic cross-clamping. 5'6,13~5 In contrast, normal spinal cord motor tract function is preserved throughout the cross-clamp interval when adequate distal pcrfusion without critical vessel exclusion is maintained by partial bypass. These resuits confirm the findings of previous studies that used SEP. 7,9 The MEP changes observed in this study correlate with the SEP type I and II responses (Table IV) in which simple aortic cross-clamping results in loss of spinal cord conduction, whereas maintenance of adequate distal perfusion without critical vessel exclusion prevents ischemic dysftmction. 6"9 Because spinal cord ischemia is probably a transcordal phenomenon, the similarity of MEP and SEP responses (Table IV) is not surprising. The principal advantage of this new technique is the ability to directly monitor spinal cord motor tract fimction noninvasively and in an on-line fashion while eliminating the major causes of false interprc- tation of SEP monitoring. Since the cord is directly stimulated and monitored, intraoperative changes in MEP are a highly sensitive and specific indicator of anterior spinal cord ischemia. The tolerable duration of such motor tract ischemia and whether the erects of such ischemia can be ameliorated by institution of various adjuncts (hypothermia, oxygen free-radical scavengers, and cerebrospinal fluid drainage) remain to be determined. It must be stated clearly that, as with other forms of spinal cord monitoring, use of MEP does not prevent paraplegia. The major value ofintraoperative spinal cord monitoring is that it permits recognition of situations where significant anterior spinal cord ischemia has occurred. The detection of intraoperarive spinal cord ischemia is associated with an increased risk of postoperative paraplegia. 9 As with other organs, the rate of progression of ischemia to frank infarction is determined by the depth and duration of the underlying ischemic state and the conditions of reperfusion. The ability to recognize such situations intraoperatively gives the surgeon an opportunity to employ specific measures designed to reverse the underlying cause of ischemia and/or ameliorate its effects. Rapid and appropriate institution of these adjuncts, guided by intraoperative MEP monitoring, may reduce the incidence of postoperative spinal cord dysfunction encountered in these situations. Prevention of spinal cord injury and paraplegia can be achieved only by complete prevention of spinal cord ischemia. Several studies have shown that the only adjunctive measure capable of preventing both anterior column ischemia (MEP) and posterior column ischemia (SEP) after proximal aortic crossdamping is maintenance of adequate distal aortic perfusion in the absence of critical vessel exclusion. 7'9 MEP monitoring of spinal cord function in conjunction with distal perfusion pressure should also confirm the adequacy of distal perfusion and permit rapid detection of spinal cord ischemia caused by interruption of flow to critical intercostal or lumbar arteries that may occur when the diseased aortic segment is excluded from the systemic and bypass circulations by placement of aortic cross-clamps. In summary, a new technique for noninvasive online assessment of spinal cord motor tract perfusion has been described. This technique directly measures anterior spinal cord motor tract fimction and is not dependent on intact cortical or peripheral nerve function or the avoidance of specific anesthetic agents. Simple aortic cross-clamping has been shown to result in distinctive changes in MEP indicative of motor

7 Volume 7 Number 1 January 1988 Spinal cord monitoring by MEP 167 tract ischcmia. Such ischemia begins in the distal cord and exhibits upward progression with time and is reversed by distal aortic reperfusion. Ischemia resulting from proximal aortic cross-clamping is prevented by maintenance of adequate distal aortic perfusion. Clinical application of this new form of intraoperative spinal cord monitoring may provide a method for rapid and accurate detection of motor tract ischemia. REFERENCES 1. Crawford ES, Crawford JL, Sail HJ, et al. Thoracoabdominal aortic aneurysms: preoperative and intraoperative factors determining immediate and long-term results of operations in 605 patients. I Vase Sutm 1986;3: Hollier LH. Protecting the brain and spinal cord, J VASe SURG 1987;5: Coles ]G, Wilson GI, Sima AF, Klement P, Tait GA. Intraoperative detection of spinal cord ischemia using somatosensory cortical evoked potentials during thoracic aortic occlusion. Ann Thorac Surg 1982;34: Cunningham IN Jr, Laschinger JC, Merldn HA, et al. Measurement of spinal cord ischemia during operations on the thoracic aorta. Ann Surg 1982;196: Laschinger JC, Curmingham IN Jr, Catinella FP, Nathan IM, Knopp EA, Spencer FC. Detection and prevention ofintraoperative spinal cord ischemia after cross-clamping of the thoracic aorta: use of somatosensory evoked potentials. Surgery 1982;92: Laschinger IC, Cunningham JN lr, Cooper MM, Baumann FG, Spencer FC. Monitoring of somatosensory evoked potentials during surgical procedures on the thoracoabdominal aorta. I. Relationship of aortic cross-clamp duration, changes in somatosensory evoked potentials, and incidence of ncurologic dysfunction. J Thorac Cardiovasc Surg 1987;94: Laschinger JC, Cunningham IN Jr, Baumann FG, Isom OW, Spencer FC. Monitoring of somatosensory evoked potentials during surgical procedures on the thoracoabdominal aorta. II. Use ofsomatosensory evoked potentials to assess adequacy of distal aortic bypass and perfusion after thoracic aortic crossclamping. I Thorac Cardiovasc Surg 1987;94: Laschinger IC, Cunninghanl JN Jr, Baumann FG, Cooper MM, Krieger KH, Spencer FC. Monitoring ofsomatosensory evoked potentials during surgical procedures on the thoracoabdominal aorta. III. Intraoperative identification of vessels critical to spinal cord blood supply. J Thorac Cardiovasc Surg 1987;94: Cunningham JN Jr, Laschinger JC, Spencer FC. Monitoring of somatosensory evoked potentials during surgical procedures on the thoracoabdominal aorta. IV. Clinical observations and results. J Thorac Cardiovasc Surg 1987;94: Laschinger IC, Cunningham IN Jr, Cooper MM, Krieger K, Nathan IM, Spencer FC. Prevention of ischemic spinal cord injury following aortic crossclamping: use of corticosteroids. Ann Thorac Surg 1984;38: Coles JG, Wilson GJ, Sima AF, et al. Intraoperative management of thoracic aortic aneurysm. Experimental evaluation of peffusion cooling of the spinal cord. J Thorac Cardiovasc Surg 1983;85: Lira KH, Connolly M, Rose DM, lacobowitz I, Cunningham JN Jr. Time response relationship of recombinant superoxide dismutase in the ischemic spinal cord syndrome. Surg Forum 1986;37: Svensson LG, Rickards E, Coull A, Rogers G, Fimme! CI, Hinder RA. Relationship of spinal cord blood flow to vascular anatomy during thoracic aortic cross-clamping and shunting. l Thorac Cardiovasc Surg 1986;91: Wadouh F, Arndt C-F, Metzger H, Hartmann M, Wadouh R, Borst HG. Direct measurements of oxygen tension on the spinal cord surface of pigs after occlusion of the descending aorta. I Thorac Cardiovasc Surg 1985;89: Gelman S, Reves JG, Fowler K, Samuelson PN, Lell WA, Smith LR. Regional blood flow during cross-clamping of the thoracic aorta and infusion of sodium nitroprusside. J Thorac Cardiovasc Surg 1983;85: DISCUSSION ~ Dr. Joseph N. Cunningham (Brooklyn, N.Y.). The study by McCullough et al. elucidates and manipulates yet another variable in the array of factors contributing to paraplegia. The idea that increases in cerebrospinal fluid pressure may contribute to neurologic injury is not a new one but the present study clearly demonstrates that clinicians and researchers should refocus on this important area. Our laboratory has long used spinal cord monitoring techniques as part of an approach to examine factors that might produce or prevent spinal cord ischemia, and this study lends itself perfectly to the use of our methodology. *This discussion combines comments and replies to McCullough et al. (pages ) and Laschinger et al. We have monitored evoked potential changes in animals to study their relationship with spinal cord perfusion pressure and changes in neurologic outcome. In a typical SEP tracing, components of latency and amplitude are evaluated and latency changes as small as 10% may be associated with a high incidence ofneurologic impairment. We manipulated spinal fluid drainage in dogs to see whether this affected the time at which a 10% latency change occurred after aortic cross-clamping ("L-10 time"). Drainage of cerebrospinal fluid to the point of neutral or negative pressure before aortic cross-clamping resulted in higher spinal cord perfusion pressures, delayed onset of latency changes, and an improved neurologic outcome (as measured by criteria devdoped by Tarlov) after clamp removal. Animals with complete spinal fluid drainage clearly

8 168 Laschinger et al. Journal of VASCULAR SURGERY demonstrated a greater ability to tolerate prolonged aortic cross-clamping without subsequent neurologic injury. Therefore these studies support the authors' suggestion that critical differences between spinal fluid pressure and distal aortic pressure are important in determining adequacy of blood supply to the cord. In a related area, Dr. Laschinger and his associates have presented an extension of their previous work and suggested that we develop clinically applicable modalities for on-line spinal cord monitoring with motor evoked potentials. We support their contention that monitoring of ventral cord columns is superior to SEP monitoring and avoids the effects of halogenated anesthesia and peripheral nerve and cortical dysfunction. This is not to infer that all previous data based on SEP studies are erroneous but simply represents an improvement in our ability to detect motor column dysfunction with more sensitive measurements. It is worth reiterating that in our clinical experience we have never wimessed a case of postoperative paraplegia in any patient in whom normal evoked potentials were maintained throughout the entire operative procedure. Paraplegia has occurred at an alarmingly high rate in our patients when evoked potential was absent or abnormal for any reason during surgical manipulation. If evoked potentials are lost (particularly for longer than 30 minutes), the incidence of neurologic injury in our series ranges from 40% to 70%. It is clear that it is impossible to provide adequate distal aortic perfusion in some patients by virtue of extensive disease or technical problems. However, it is significant that we recognize the availability of monitoring techniques that can help guide us not only clinically but experimentally in developing strategies for diminishing neurologic injury associated with thoracoabdominal aortic surgew. We would all agree, I am sure, that the advent of electrocardiography by no means prevented myocardial infarction but exists today as the "gold standard" of monitoring cardiac viability. In a similar fashion, I propose the analog for monitoring spinal cord viability is clearly that of measuring spinal cord conduction directly. We appreciate the opportunity of discussing these two papers. Dr. McCullough, because the canine model has variable collateralization to the lower cord, would it be advisable for you to employ some sort of sensor), or motor monitoring technique to ascertain initially which animals do not become ischemic with simple proximal aortic crossclamping (Laschinger type 2 response)? If you eliminated that group, it would make your entirestudy more valid. Have you used intrathecal papaverine or vasodilators in association with your current approach? Given your current interpretation of the critical relationship between mean distal aortic pressure and spinal fluid pressure, would you comment on the use of nitroprusside as a vasodilator during proximal clamping? Dr. Laschinger, at this early stage in development of motor evoked potential studies, do you predict it will be necessary to redefine your original classification of neurologic changes on the basis of somatosensory evoked potential measurements (Laschinger types 1 through 4)? Is this noninvasive cutaneous modality of MEP measurement available for use in the recovery room or the postoperative intensive care setting? Would it be useful to unravel the dilemma of the rare case of paraplegia that occurs late in the postoperative period? Dr. McCullough. Dr. Curmingham, it is encouraging to hear of your experimental results showing yet another physiologic assessment of the effects of manipulation of spinal fluid pressure on the spinal cord function. As I stated earlier, we do not use evoked potential monitoring clinically because our operative approach includes routine complete intercostal revascularization. We believe that spinal fluid drainage merely extends the safe ischemia time for the surgeon to perform intercostal reimplantation. In our experimental model we did use SEP monitoring earlier in a series of animals, and it appeared that maintenance of evoked potentials correlated with distal aortic pressures at or above 40 mm Hg. There have been several studies that suggest that sodium nitropmsside can decrease spinal cord blood flow. Therefore we now use sodiunl nitroprusside to bring the proximal aortic pressure down to a safe level but not necessarily to bring it down to the preclamp levels. We are well aware of Dr. Svensson's study, which demonstrated that a local vasodilator placed around the spinal cord can increase spinal cord blood flow during aortic clamping. In fact, we attempted to examine this in our canine model. However, when we attempted to use intrathecal papaverine, we found that as papaverine mixed with cerebrospinal fluid, it formed a precipitate. We tried various manipulations of ph but were unable to use papaverine effectively. So we have not incorporated this into our human study. Dr. Laschinger. In response to Dr. Cunningham's questions, there are four distinct SEP responses (types 1 through 4) occurring after aortic cross-clamping, and the preliminary. MEP data suggest that the same four response types will also be observed with MEP monitoring. Furthermore, our early data suggest that they will follow the same rules that were previously outlined for SEP monitoring. Although time does not permit me to demonstrate each of these responses, I think it is important, as Dr. Cunningham would agree, that a uniform classification system be used for describing evoked potential changes during thoracic aneurysm procedures. Only by adopting such a uniform classification system can data from various clinical and experimental groups be compared effectively. As far as the postoperative observations are concerned, in this animal study the size of the electrodes used permitted us to leave these disk electrodes in postoperatively. Removal of these electrodes is then achieved in essentially the same fashion as a pacemaker wire, by simply pulling them out after operation. As a result, we have obtained postoperative monitoring of motor evoked potentials in these animals for up to 7 days postoperatively. Basically we have found that in animals that become paraplegic, there are two types of responses--those in

9 Volume 7 Number 1 JanuaD Spinal cord monitoring by MEP 169 which the MEPs do not return during the reperfusion or postoperative intervals and those in which MEPs do return after reperfusion for a period of approximately 1 to 2 hours and then suddenly are lost, presumably because of some type of reperfusion injui T since no hemodynamic instability was observed. This second response occurs in about half of the dogs that do go on to become paraplegic. These data suggest that, in addition to the intraoperative ischemic injury that occurs, there is also a significant incidence of postoperative reperfusion injury that may result in further damage to the spinal cord. Dr. E. Stanley Crawford (Houston, Tex.). Dr. McCullough and associates are to be congratulated for their studies of the cause and prevention of paraplegia. They have confirmed earlier work in dogs, showing the relationship between the development of paraplegia and the reduction of distal aortic pressure below that of cerebrospinal fluid pressures by prolonged clamping of the proximal descending thoracic aorta. To my knowledge, they have been the first to apply this knowledge clinically in patients during treatment of thoracoabdominal aortic aneurysm. Their results in 24 patients with nondissecting aneurysms are encouraging. First, there were no complications from the drainage of spinal fluid and continuous intrathecal monitoring during the period of aortic clamping, indicating the probable safety of the proceduure. Second, paraplegia did not occur in any of these cases, suggesting the possible effectiveness of the method in prevention Or reduction of the development of this complication. The latter conclusion, as Dr. McCullough has indicated, is premature because, in fact, some of the experimental animals suffered paraplegia despite spinal fluid drainage and because the risk of paraplegia in patients treated for similar aneurysms by simple cross-clamping is very low (2% to 8%). Moreover, 50% of patients who have the complication do so 12 hours to 3 weeks after operation. Spinal fluid drainage limited to the period of aortic clamping would probably not prevent this delayed phenomenon. Regardless, the authors should be congratulated again for initiating the clinical application of this principle and encouraged to pursue it to its scientific conclusion by randomizing their cases between treatment by simple crossclamping and treatment by cross-clamping plus spinal fluid drainage adding patients at high risk, that is, those with extensive lesions and those with dissection, and precisely stratifying the cases according to location, extent, and origin of aneurysm replaced. Consideration may also be given to extending the period of spinal fluid monitoring and drainage in the hope that delayed paraplegia would be prevented. Dr. Laschinger has presented a new method of monitoring spinal cord function that reflects integrity of motor tract function. This method when employed during aortic operations may more accurately detect signs of motor tract ischemia and be more predictive of changes that may result in paraplegia than that achieved by monitoring sensory pathways (SEP). Perhaps this will allow development of strategies that can be employed to prevent complications, which was not possible with SEP monitoring. We attempted the latter prospectively in 198 consecutive patients. The technique could not be uscd in half of the animals because of the inability to perform adequate perfusion distally, but adequate perfusion and SEP monitoring were successfully achieved in the other hale There was no significant difference in either early or late paraplegia in the two groups. The limitations of SEP and perfusion were (1) adequate distal perfusion was impossible in 50% of cases; (2) critical sources of cord blood supply could not be identified for timely reattachment in long aortic aneurysmal segments located between widely separated clamps; and (3) a significant number of false-positive and falsenegative SEP responses. It is predicted that the former two limitations will be seen with this method of monitoring, which depends on adequate distal pcrfusion during aortic occlusion. Finally, the method does not address the phenomenon of late paraplegia. Despite these negative remarks, I encourage Dr. Laschinger and associates to continue their work and extend it to cover the postoperative period as well. Monitoring has paid off in other fields and should in this one as well. Randomization in application of this modality should also be considered. Numerous methods of preventing paraplegia have been presented as in the previous article and accepted on the basis of early successes in a few cases with little risk of paraplegia, only to be proved inadequate with additional experience. It now seems to be time to subject all methods to the real scientific test, that is, randomization, stratified according to location, extent, clamp time, and origin of disease in the segment replaced. Dr. McCullough. Dr. Crawford, we agree that intercostal reimplantation is probably the most important way today that adequate blood flow to the spinal cord can be maintained, and we certainly are aware of cases of late paraplegia. That is another reason that we believe that selective intercostal reimplantation may provide only marginal blood flow to the spinal cord. That might be sufficient intraoperatively with the patient under optimal hcmodynamic control. But postoperatively, with an episode of hypotension, these intercostal vessels that were excluded may have left this patient with marginal blood flow; then the patient's condition becomes critical with postoperative hypotension. Dr. Laschinger. Dr. Crawford, our attempts to develop monitoring techniques for spinal cord function during thoracic aneurysmectomy are directly related to the attempt to search for various causes of spinal cord ischemia and infarction during such procedures so that a unified approach may be applied to these lesions whereby we can, if possible, decrease the incidence of paraplegia. I and my co-workers would agree with Dr. Crawford that complete prevention of paraplegia is a goal that will be very difficult to achieve. Howevcr, unless we continue to search for the various causes of spinal cord ischemia and ways to prevent it, it will be a goal we will never achieve. Dr. John E. Conno~ly (Irvine, Calif.). My remarks are directed to Dr. McCullough's article. I enjoyed the pre-

10 170 Laschinger et al. }'ournat of VASCULAR SURGERY sentation and it appears that the Mayo Clinic group, through their reapplication of an old idea, may have discovered an important modaliry in the prevention of paraplegia in surgical treatment of thoracoabdominal aneurysms. Obviously, more experimental work and clinical application of spinal cord pressure control needs to be performed before we can say it is a reason that Dr. Hollier reports so many operations without paraplegia. We should also remember that Dr. Hollier has been implanting intercostal and lumbar arteries in these patients, and it may be that he is able to do that more rapidly than he was doing it before. As one who has had a long interest in the prevention of paraplegia during aortic surgery, I stress, as have others, that there are multiple factors involved. First is the particular anatomic spinal cord blood supply in a given patient. Second is the temporat T or permanent interruption of key spinal cord arterial supply, and that is so important when we are at the diaphragmatic level. Third is the presence or absence ofhypotension during operation. In recent years we have been particularly interested in selective visualization of the artery of Adamkiewicz, which is now a safe, simple radlologic procedure and I predict it will be more widely applied. The authors' technique of controlling spinal cord pressure may be working because it affords greater safe ischemia time for key intercostal and lumbar implantation. Finally, although the major risk of paraplegia occurs during surgical procedures on the thoracoabdominal area, there is a small but real danger with infrarenal aneurysms; I am sure there are some in this room who have had that catastrophe. I believe paraplegia in that circumstance happens when the artery of Adamkiewicz comes off low from a low-lying lumbar artery. If I see a suspiciously large infrarenal lumbar artery on a preoperative aortogram of a patient with an infrarenal aneurysm, I either try to avoid it, consider its reimplantation, or selectively p'erfuse that particular artery, Dr. McCullough, do you think any of your techniques will ever be applicable to the avoidance of paraplegia in infrarenal aneurysms, a complication that is usually totally unexpected and therefore even more devastating to the surgeon and patient? Dr. F. William Blaisdell (Sacramento, Calif.). Dr. McCullough, we were interested in the same problem back in 1960 and at that time we found when we measured distal aortic pressure and simultaneously measured cerebrospinal fluid pressure, that in some instances the pressure curves actually crossed. We then carried out a series of experiments in dogs in which we ligated any number of intercostal arteries up to and including all the thoracic intercostal arteries. In these dogs it did not produce paraplegia. Subsequently we cross-clamped the aorta in dogs who had variable numbers of intercostal arteries ligated and established the fact that paraplegia incidents correlated directly with the number of intercostal arteries ligated. When we ligated all thoracic intercostal arteries the incidence of paraplegia associated with aortic cross-clamping was 100%, whereas it was 40% with ~igation of two pairs of intercostal arteries. We also made the observation that by lowering cerebrospinal fluid pressure either by withdrawal of fluid or by the administration of urea, we could modify the incidence of paraplegia. We attempted to carry this out clinically and found that needle decompression, the only method available to us at that time, frequently did not work because the needle clotted or was otherwise unreliable because of displacement. Urea was not effective in the clinical circumstance in significantly lowering the pressure. We are pleased, of course, to see this concept resurrected. Dr. Wilhelm Sandmann (Duesseldorf, West Germany). I am pleased to discuss the interesting study by Dr. Laschinger and co-authors who demonstrated by the use of monitoring the motor evoked potential that spinal cord function can be maintained during cross-clamping of the thoracic aorta by perfusion of the distal aorta. However, the clinical applicability of this experimental setting seems limited because in patients with thoraeoabdominal aneurysm (who represent the true high-risk group to have paraplegia), retrograde perfusion does not reach the artery of Adanlkiewicz and high-voltage stimulation of the spinal cord has not been sufficiently investigated for its potential damage to neural tissue in humans. In our experience during the past 4 years, we have compared 32 patients after simple cross-clamping of the aorta who had extensive graft replacement of the thoracoabdominal aorta with reimplantation of visceral and renal arteries and selective reimplantation of intercostal arteries to a subsequent group of 27 patients undergoing identical procedures, except that the spinal cord function was monitored by segmental somatosensory evoked potentials recorded directly from the spinal cord by two epidural or spinal catheters and used as a guide for selective reimplanration of the intercostal and lumbar arteries. If in the latter group after restoration of the distal circulation, the evoked potentials did not reoccur or improve within 10 minutes, additional intercostal and lumbar arteries up to all available segmental vessels were reimplanted. In the group without monitoring, paraplegia or paraparesis occurred in 13% and transient deficits in 28%, which agrees with rates reported in the literature for repair of such extensive aneurysms. But in the group with monitoring, paraplegia occurred in only 1 of 27 patients, accounting for 4%. In this particular patient, we were not able to regain the evoked potential signal and we assumed that we might have missed the specific intercostal artery lying within the debris of the aortic wall. As some of the intercostal and lumbar arteries were reimplanted more than 1 hour after aortic cross-clamping and still evoked potentials were regained after reperfusion of the responsible arteries and subsequently the patient did not have paraplegia or paraparesis, we strongly recommend this type of low-voltage electrophysiologic monitoring. Dr. Laschinger, what is your clinical practice of spinal cord preservation in patients with thoracoabdominal an-

11 Volume 7 Number 1 January 1988 Spinal cord monitoring by MEP 171 eurysms involving the whole descending thoracic and abdominal aorta? Do you think that monitoring of motor evoked potentials might damage neural tissue? Do you have clinical data that confirm that distal aortic perfusion without reaching the level from which the artery of Adamkiewicz arises can prevent paraplegia or transient deficits? Transient deficits lasting from one day to a few weeks are rarely mentioned in the literature. Have you observed such deficits in your clinical routine and how do you explain them? Dr. Laschinger. In response to Dr. Sandmann's question concerning the risk of neural tissue damage resulting from MEP monitoring, I can say we have investigated that very carefi.tlly. Because the intervertebral disk is stimulated, the actual current does not directly enter the spinal cord; as a result we have fotmd it fairly impossible to damage the cord with this technique. We have gone all the way up to 75 ma (10 times threshold) without producing any functional or pathologic cord damage in dogs, pigs, and baboons; it has also been used in humans at that level. I must emphasize that distal perfusion is not required to use this technique. This technique, because it monitors the spinal cord directly, can be used without distal perfusion; therefore even in long thoracoabdominal aneurysms you can tell whether or not spinal cord ischemia occurred. Unlike SEP, MEP monitoring does not require peripheral nerve function and distal perfusion is not required. However, on the basis of experimental and clinical data, it is my belief thae for distal perfusion techniques to be effective in preventing paraplegia, vessels critical to spinal cord blood flow must bc perfused. MEP and SEP techniques allow intraoperative detection of the failure to perfuse such vessels and therefore determine the need for vessel reimplantation in each patient. Dr. Larry H. Hollier (closing comments in discussion of ' Fhoracoabdominal Aortic Aneurysm Repair Without Paraplegia"). I thank all of the discussers for their comments. I think all of us have the greatest respect for the work that Dr. Crawford has done in thoracoabdominal aneurysm repair. But, despite a seemingly perfect repair, we still see paraplegia. I believe that increased cerebrospinal fluid pressure and resukant decreased spinal cord perfusion may be implicated as ma etiologic factor in some of these patients. Obviously, there are many factors that influence the development of paraplegia and their interrelationship is important. Intraspinal pressure is one factor that was discovered in the past but, like so many things in surgery, has been abandoned and forgotten. Although spinal fluid drainage in itself does not prevent paraplegia, I think it is one modality that should be looked at much more closely on a clinical and experimental basis. I believe that cerebrospinal fluid drainage will prove to be an additional important adjunct in helping us prevent this difficult problem of paraplegia.