Prevention of Spinal Cord Injury After Repair of the Thoracic or Thoracoabdominal Aorta

Transcription

1 CURRENT REVIEW Prevention of Spinal Cord Injury After Repair of the Thoracic or Thoracoabdominal Aorta Michael C. Mauney, MD, Lorne H. Blackbourne, MD, Scott E. Langenburg, MD, Scott A. Buchanan, MD, Irving L. Kron, MD, and Curtis G. Tribble, MD Department of Surgery, University of Virginia Health Sciences Center, Charlottesville, Virginia Spinal cord injury occurring as the result of surgical repair of thoracic and thoracoabdominal aortic disease remains a devastating complication. The incidence of postoperative neurologic deficits varies from 4% to 38%. Factors associated with a greater risk for injury include the presence of dissection or extensive thoracoabdominal disease, and a prolonged cross-clamp time. Spinal cord ischemia initiates a deleterious cascade of biochemical events that ultimately result in an increased intracellular calcium concentration. Calcium-activated proteases, lipases, and nucleases mediate the processes that cause cell injury. The accumulation of oxygen-derived free radicals and the occurrence of hyperemia during reper- fusion are also contributing causes of spinal cord injury. Increasing the spinal cord blood flow with shunts, oxygenated bypass circuits, cerebrospinal fluid drainage, the intrathecal administration of vasodilators, and the reattachment of intercostal arteries has been tried in an effort to increase spinal cord perfusion. Pharmacologically based measures to prevent spinal cord injury have been pursued, and these have consisted of hypothermia, anesthetic agents, calcium channel blockers, free radical scavengers, and immune system modulation. However, no single technique has proved to be consistently effective in preventing ischemia-induced spinal cord injury. (Ann Thorac Surg 1995;59:245-52) D espite a dramatic reduction in the mortality associated with surgical procedures for the treatment of descending thoracic and thoracoabdominal aortic disease over the past 30 years, spinal cord injury continues to be a devastating complication. Improved methods for detecting thoracoabdorninal aortic disease coupled with a sizable aging population have led to the performance of an increased number of surgical repairs. Investigators continue to study the frequency of spinal cord injury in an effort to develop improved methods of spinal cord protection. The goal of this article is to review the epidemiology of spinal cord injury, the pathophysiologic features associated with occlusion of the thoracoabdominal aorta, and the current techniques for preventing spinal cord injury. Frequency of Spinal Cord Injury The exact incidence of spinal cord injury occurring during aortic surgical procedures remains controversial, in part due to the relatively small number of patients involved. Furthermore, most reports are retrospective studies, which tend to reveal lower incidences of paraplegia than do prospective ones. Nevertheless, the findings from the largest clinical series suggest that the risk of spinal cord injury is related to several variables, including the complexity of the repair, the time of aortic occlusion, and the cause of the aortic disease. In the largest series reported on to date, Svensson and associates [1[ noted a 16% incidence of paraplegia or Address reprint requests to Dr Tribble, Department of Surgery, University of Virginia Health Sciences Center, Box 181, Charlottesville, VA paraparesis in 1,509 patients who had undergone repairs for the treatment of thoracoabdominal aortic disease. The complexity of the repair was found to be a significant predictor of spinal cord injury. The Crawford classification defines thoracoabdominal aortic aneurysms as follows: type L proximal descending thoracic aorta to the upper abdominal aorta; type II, proximal descending thoracic aorta to below the renal arteries; type III, distal half of the descending thoracic aorta into the abdomen; and type IV, most or all of the abdominal aorta. Repairs of type I, II, Ill, and IV aneurysms were associated with paraplegia or paraparesis rates of 15%, 31%, 7%, and 4%, respectively. Other authors have also reported an increased frequency of spinal cord injury in patients undergoing repairs of complex (ie, types I and I1) thoracoabdominal aortic disease [2, 3]. Repairs limited to the descending thoracic aorta are generally associated with a lower incidence of paraplegia. Livesay and colleagues [3] found the incidence of paraplegia associated with 360 repairs of proximal and extensive descending thoracic lesions was 3% and 9%, respectively. Lerberg and associates [4l reported a 1.5% risk of paraplegia for patients undergoing repair of coarctation of the aorta. Survivors of acute traumatic disruption of the thoracic aorta represent a small subset of patients with thoracic aortic disease. Only 1,188 cases have been reported in the past 20 years; not surprisingly, the incidence of postoperative spinal cord injury varies from 0 to 20% [5-7]. Most authors agree that the presence of dissection or the need for an emergent operation because of aneurysreal rupture increases the risk for spinal cord injury. In 1995 by The Society of Thoracic Surgeons /95/$ (94)00815-O

2 246 REVIEW MAUNEY ET AL Ann Thorac Surg SPINAL CORD ISCHEMIA 1995;59: Svensson and associates' [1] experience, when dissection occurred, the risk of paraplegia was significantly increased after repairs for all types of thoracoabdominal aortic aneurysms. Livesay and colleagues [3] demonstrated that emergent repairs carried a 17% risk of injury versus a 6% risk for elective repairs. Hollier and coworkers [8] reported a combined paraplegia and paralysis rate of only 5% for 83 elective repairs of nondissecting thoracoabdominal aortic aneurysms. A frequently debated question with regard to aortic procedures concerns the influence of aortic cross-clamp times on spinal cord injury. Svensson and colleagues [1] determined that the aortic cross-clamp time was an independent predictor of paraplegia or paraparesis. Repairs requiring more than 60 minutes of aortic occlusion carried a 20% risk, whereas the risk for those repairs completed in less than 30 minutes was less than 10%. The median cross-clamp time for all patients in their series was 43 minutes. Likewise, Livesay and co-workers [3] showed that the risk of injury increased from 3% to 11% when the cross-clamp times exceeded 30 minutes. Conversely, other studies have not shown a clear-cut relationship between spinal cord injury and longer crossclamp times. In Hollier and associates' [8] series, the cross-clamp times in all 5 patients who suffered postoperative paraplegia or paraparesis were less than 20 minutes. Najafi [9] observed no postoperative spinal cord injury in 31 patients who underwent repairs with mean aortic cross-clamp times of 58 minutes. However, distal aortic perfusion with partial femoral-femoral cardiopulmonary bypass was used in all patients. In a similar report, Svensson and associates [10] concluded that cross-clamp times greater than 40 minutes were not associated with spinal cord injury if distal aortic perfusion was used. Hence, the use of distal aortic perfusion may offset any influence cross-clamp times have on spinal cord injury. Mechanisms of Spinal Cord Injury Spinal Cord Blood Supply Ischemia is the common pathway to spinal cord injury in the perioperative period. A knowledge of the nature of the spinal cord blood supply is therefore critical in understanding the mechanism of spinal cord ischemia. The human spinal cord is perfused by a single anterior and two posterolateral spinal arteries. These arise from the vertebral arteries superiorly and continue inferiorly within the subarachnoid space, communicating with each other through poorly defined circumferential anastomotic channels that are unpredictable in number. Occasionally, the posterolateral spinal arteries arise from the posterior inferior cerebellar arteries. The anterior spinal artery supplies 75% of the blood to the spinal cord and runs in a continuous fashion, whereas the posterolateral spinal arteries are more plexiform [11]. There are 25 to 30 pairs of segmental vessels arising from the aorta and its branches that can potentially anastomose with the anterior spinal artery. Twelve to 14 of these arteries arise directly from the aorta and the remainder originate from Posterolateral Spinal Arteries Anterior Spinal Artery Anterior Radicular Artery Fig 1. Branches of segmental arteries. Posterior Radicular Artery Spinal Artery o~,~# ~'i "! ~ Muscular Branch Posterior Ramus "- ~ Anterior Ramus Segmental Artery "'Anrta the vertebral arteries, thyrocervical and costocervical trunks, and the iliac arteries. These segmental vessels course posteriorly to split into anterior and posterior rami. The anterior rami continue as the intercostal arteries in the chest. In the abdomen, they contribute to the posterolateral blood supply of the abdominal wall and flank. The posterior rami give rise to spinal arteries that further divide into anterior and posterior radicular arteries. The anterior radicular artery merges with the anterior spinal artery to supply the spinal cord and the nerves that arise directly from the spinal cord (Fig 1). Only a fraction of the segmental arteries eventually reach the spinal cord (Fig 2). The superior cord (C1 to T2/T3) is well supplied with three to five anterior radicular arteries. The midthoracic cord (T4 to T7/T8) has only one or no anterior radicular artery. The remaining lower thoracic and lumbar spinal cord is supplied with three to five anterior radicular arteries, including the most important one, the arteria radicularis magna (ARM), or artery of Adamkiewicz [11]. This artery arises between T8 and L3 in 90% of human beings [12] and has a left-sided origin in 80% [13]. Wadouh and associates [14] illustrated the important contribution this artery makes to the lumbar spinal cord blood supply by their finding of a 70% paraplegia rate in pigs after ligation of the artery of Adamkiewicz. Like the brain, the spinal cord blood supply is under autoregulatory control, such that flow is constant for spinal cord perfusion pressures of 50 mm Hg or more [15]. This pressure is the difference between the distal aortic blood pressure and the cerebrospinal fluid pressure. The relationship between the cerebrospinal fluid pressure and spinal cord blood flow has been studied extensively. Thoracic aortic occlusion is associated with distal aortic hypotension and increased cerebrospinal fluid pressure. Blaisdell and Cooley [16] believed that proximal aortic hypertension occurring after aortic occlusion resulted in increased cerebrospinal fluid pressures and decreased spinal cord blood flow. However, Svensson and colleagues [17] noted an increase in the cerebrospinal fluid pressure before aortic occlusion and correlated this with an increased central venous pressure. Maughn and co-workers [18] reinforced this theory by demonstrating a decrease in the cerebrospinal fluid pres-

3 Ann Thorac Surg REVIEW MAUNEY ET AL ;59: SPINAL CORD ISCHEMIA Vertebral Artery Ascending Cervical Fig 2. Origin and distribution of anterior radicular arteries supplying the anterior spinal artery. Anterior Spinal Arter C1~ Superior Spinal Cord (3-5) Thyrocervical Trunk Ao~a Subclavian Artery T 2,3-- Mid-Thoracic Spinal Cord (1-2) Lower-Thoracic and Lumbar Spinal Cord (3-5) TS-- L5-- \ Artery of Adamkiewicz sure and an increase in the spinal cord blood flow with exsanguination during aortic occlusion. Regardless, increased cerebrospinal fluid pressure is not always associated with a decreased spinal cord blood flow, as findings from the work done by Griffiths and colleagues [15] suggest that spinal cord blood flow can be maintained despite an elevated cerebrospinal fluid pressure. Spinal Cord Ischemia Spinal cord ischemia in the perioperative period can result from preoperative hypotension, distal aortic hypotension after aortic occlusion, the interruption of critical intercostal and lumbar arteries, thrombosis and embolism of intercostal arteries, and postoperative hypotension or hypoxia. Neural tissue is exquisitely sensitive to ischemia. At normal temperatures, the stores of adenosine triphosphate are completely depleted after 3 to 4 minutes of ischemia as mitochondrial oxidative phosphorylation ceases. When the adenosine triphosphate levels approach zero, adenosine triphosphate-dependent membrane pumps necessary for intracellular ionic homeostasis fail [19]. This disruption of calcium homeostasis is critically important in initiating the process of irreversible cellular injury. During ischemia, intracellular calcium rapidly accumulates. The increased calcium levels activate the release of cytoplasmic proteases and nucleases, resulting in damage to structural proteins and DNA. Calciumactivated phospholipases metabolize membrane lipids to arachidonic acid and several other vasoactive metabolites. Endothelial xanthine dehydrogenase is converted to xanthine oxidase, an enzyme critical in mediating free radical production during reperfusion [19]. And, finally, an elevated intracellular calcium concentration results in an increased release of the excitatory amino acids aspartate and glutamate, both of which have been shown to be neurotoxic [20]. Reperfusion Injury Braughler and Hale [21] recently reviewed the current state of knowledge about free radical injury and its importance in ischemic neural tissue after reperfusion with oxygenated blood. Most free radicals are derived from the conversion of molecular oxygen to the superoxide radical by xanthine oxidase in the presence of NADPH (the reduced form of nicotinamide-adenine dinucleotide phosphate) and xanthine or hypoxanthine. Xanthine and hypoxanthine levels rise during ischernia as adenosine triphosphate is quickly metabolized. Other mechanisms of free radical production include the respiratory burst of activated neutrophils that occurs in response to tissue injury, the metabolism of arachidonic acid to prostaglandins G 2 and H2, and the autooxidation of catecholamines in the presence of metals such as iron [19]. Free radicals damage DNA, initiate lipid peroxidation, and degrade structural elements like collagen and hyaluronic acid. Lipid peroxidation results in a loss of membrane integrity, inactivation of critical membrane-

4 248 REVIEW MAUNEY ET AL Ann Thorac Surg SPINAL CORD ISCHEMIA 1995;59: bound enzyme systems, and an increased ratio of vasoconstrictive prostaglandins like thromboxane A 2 [19]. This adverse shift in prostaglandin production can lead to the no-flow phenomenon of microvascular vasospasm and thrombosis secondary to an increased aggregation of platelets and inflammatory cells. Barone and co-workers [22] observed a significant increase in lower thoracic spinal cord blood flow in dogs compared to the baseline flow after 30 minutes of thoracic aortic occlusion. A significantly greater degree of hyperemia was noted in those dogs with neurologic deficits compared with that seen in neurologically intact dogs. Jacobs and co-workers [23] were able to correlate spinal cord injury with hyperemia and an increased spinal cord blood-brain barrier permeability to albumin, and stated that this increased permeability contributed to the development of spinal cord edema and a possible compartment syndrome. Other authors have correlated a reduction in spinal cord injury with a reduced hyperemic response during reperfusion [24]. Mechanisms of Spinal Cord Protection Maintaining Spinal Cord Blood Flow DISTAL AORTIC PERFUSION. In most patients undergoing thoracic and limited thoracoabdominal repairs, the critical radicular arteries supplying the lower spinal cord originate inferior to the distal margin of the repair. In these patients, blood flow to the lower spinal cord could be maintained with shunts or bypass circuits. The heparin-impregnated Gott shunt can be placed proximally in the ascending aorta, aortic arch, or left ventricular apex, and distally in the descending aorta or femoral artery. Although systemic anticoagulation is not required, a distinct disadvantage of the Gott shunt compared with oxygenated bypass circuits is the lack of control over flow rates and proximal and distal aortic pressures. In two recently published reports concerning series of patients with traumatic tears of the thoracic aorta repaired using the Gott shunt, the incidence of spinal cord injury was not decreased compared with the incidence in those who had cross-clamping alone [5, 7]. However, Verdant and associates [25] reported excellent results with the Gott shunt in patients undergoing elective repairs of thoracic aortic aneurysms. Femoral vein-to-femoral artery bypass grafting involving the use of a pump oxygenator has the disadvantage of requiring systemic heparinization. The Bio-Medicus pump (Eden Prairie, MN), a heparinless left atrial-tofemoral artery bypass circuit, has been shown to adequately control proximal and distal aortic hemodynamics [26]. This approach appears to be the emerging technique of choice for distal aortic perfusion, especially in patients with traumatic injuries [5, 6]. However, in the setting of nontraumatic thoracoabdominal aortic disease, Crawford and associates [1, 27] and earlier Najafi and colleagues [28] found in their reviews that the rate of paraplegia was either equal to or increased in association with distal aortic perfusion techniques, compared with the rate noted for clamp and repair techniques. Nevertheless, Najafi [9] has more recently reported a zero paraplegia rate associated with distal aortic perfusion using femoral-femoral bypass. To explain the apparent lack of protection afforded by distal aortic perfusion techniques, Laschinger [29] pointed out that many surgeons fail to properly monitor the spinal cord perfusion pressure and spinal cord function intraoperatively. He stressed the importance of maintaining mean distal aortic pressures greater than 60 mm Hg and of monitoring spinal cord integrity by observing the somatosensory evoked potentials. Somatosensory evoked potentials reflect activity in the sensory tracts of the posterior and lateral spinal cord. Loss of the signal indicates spinal cord ischemia, and this method has been used to assess both the need for and adequacy of distal aortic perfusion. Intraoperative somatosensory evoked potential monitoring has been criticized for being associated with a significant rate of false-positive and false-negative readings and for its inability to monitor function in the more vulnerable motor tracts of the spinal cord. The monitoring of motor evoked potentials has been shown to be superior to the monitoring of somatosensory evoked potentials because of its high degree of sensitivity and specificity in detecting clinically important motor tract injury. Regardless, Svensson and Crawford [30] have argued that evoked potential recording is of little use, except to confirm the patency of reattached intercostal arteries, if protective measures have been maximally implemented. Distal aortic perfusion techniques also fail to prevent spinal cord injury when the ARM, the artery critical for furnishing the blood supply to the lower spinal cord, is not perfused. In patients with extensive thoracoabdominal atherosclerotic disease, the ARM is likely to be included in that portion of the aorta being repaired, and therefore it is not perfused by any shunt in the distal aorta. Even with perfusion of the ARM, blood flow to all regions of the spinal cord may not be adequate. Svensson and associates [12] demonstrated in baboons that blood flow to the lower thoracic spinal cord was not increased by a shunt perfusing the ARM even though blood flow to the lumbar spinal cord was increased. This discrepancy was attributed to the smaller diameter and higher resistance in that portion of the anterior spinal artery above the ARM that supplies the lower thoracic spinal cord. They suggested that a similar phenomenon of inadequate perfusion of the lower thoracic spinal cord despite the presence of a functioning shunt exists in human subjects. INTRATHECAL VASODILATORS. Several investigators have used the intrathecal route for the delivery of potentially neuroprotective agents. Svensson and colleagues [31] used intrathecally administered papaverine to dilate the anterior spinal artery and found this produced increased lower thoracic spinal cord blood flow and conferred complete protection from paraplegia in a baboon model after 60 minutes of aortic occlusion. In a nonrandomized human trial, a decreased incidence of spinal cord injury

5 Ann Thorac Surg REVIEW MAUNEY ET AL ;59: SPINAL CORD 1SCHEMIA was found in those patients receiving papaverine intrathecally via a lumbar catheter [32]. In a dog model of 70 minutes of aortic occlusion, Maughan and co-workers [33] demonstrated uniform protection from spinal cord injury using intrathecal Fluosol, an oxygenated perfluorocarbon possessing known vasodilatory effects. REATTACHMENT OF INTERCOSTAL AND LUMBAR VESSELS. In 1986, Crawford and associates published a review [34] of the data in 605 patients who had undergone repair of thoracoabdominal aneurysms and reported an increased risk of neurologic injury in the patients who underwent reattachment of intercostal and lumbar arteries. They attributed this to the increased operative time associated with localizing the vessels that supplied the spinal cord. Preoperative localization using highly selective angiography is also difficult and not without complications [35]. Accordingly, most surgeons forego any attempt to reattach segmental arteries. Nevertheless, Svensson and co-workers [36] described an intraoperative technique for identifying those arteries supplying the spinal cord that consisted of placement of a platinum electrode on the spinal cord to detect hydrogen ions injected into the segmental arteries. Preservation or division of the arteries so identified significantly influenced the development of postoperative neurologic deficits in pigs. Consequently, they strongly recommend the reattachment of intraoperatively identified intercostal and lumbar arteries be performed in conjunction with distal aortic perfusion or the intrathecal administration of papaverine, or both measures. Similarly, Hollier and associates [8] reported on 24 consecutive patients who underwent thoracoabdominal aortic aneurysm repair with spinal fluid drainage and intercostal reimplantation without a single incident of postoperative paraplegia. DECREASING CEREBROSPINAL FLUID PRESSURE. As mentioned earlier, Blaisdell and Cooley [16] believed that the sudden increase in cerebrospinal fluid pressure after aortic occlusion resulted in reduced spinal cord blood flow. They recommended the use of intraoperative cerebrospinal fluid drainage to prevent spinal cord injury. However, animal studies have yielded conflicting results regarding the impact of spinal fluid drainage on spinal cord injury after aortic occlusion [14, 31, 37]. The results of clinical trials have been equally conflicting. McCullough and colleagues [38] kept cerebrospinal fluid pressure to less than 10 mm Hg in 24 patients with nondissecting thoracoabdominal aneurysms and encountered no postoperative neurologic deficits in these patients. Reattachment of the intercostal arteries was performed in some of the patients with extensive disease. Acher and associates [39] reported a 2% incidence of neurologic deficits in 56 patients with thoracic and thoracoabdominal aneurysms whose treatment included spinal fluid drainage and intravenously administered naloxone; this compared with a 23% incidence in control patients. No intercostal reimplantation was performed. Neither of these studies was randomized. In a prospective, randomized trial, Crawford and co-workers [40] found no benefit to using spinal fluid drainage in patients with extensive thoracoabdominal aneurysms regardless of the cerebrospinal fluid pressure during aortic occlusion. Increasing Spinal Cord Tolerance to Ischemia HYPOTHERMIA. Hypothermia causes a generalized reduction in all energy-consuming processes of neural tissue in direct proportion to the degree of cooling [41]. Kouchoukos and colleagues [42] found systemic hypothermic circulatory arrest to be neuroprotective and relatively safe in selected patients with extensive thoracoabdominal disease. However, hemorrhagic, pulmonary, and central nervous system complications limit the widespread application of this technique. Animal models of regional spinal cord hypothermia achieved through intrathecal [43] or arterial [44] perfusion have shown hypothermia to be neuroprotective. More recently, the use of regional perfusion cooling of the spinal cord in human subjects has been reported [45]. ANESTHETIC AGENTS. Barbiturates like thiopental block synaptic transmission and relax vascular smooth muscle, and have been reported to be neuroprotective in animals [41]. Kirshner and co-workers [46], however, found no benefit provided by thiopental alone, but did find a moderate amount of protection conferred by the combination of thiopental, hypothermia, and superoxide dismutase (SOD). Cocaine-derived local anesthetic agents stabilize cell membranes by inhibiting membrane-bound ion pumps and channels and have been shown to be neuroprotective in animals [47]. Similarly, opiate receptor antagonists like naloxone have been found to cause a decrease in central nervous system metabolic rates and to provide some neuroprotection in animal models of spinal cord ischemia [41, 48]. CALCIUM-CHANNEL BLOCKERS. Calcium-channel blockers augment cerebral blood flow and neurologic recovery after cerebral ischemia [49]. Schittek and associates [50] observed almost complete protection from spinal cord injury afforded by the intravenous administration of nimodipine after 30 minutes of spinal cord ischemia in pigs. In a rabbit model of spinal cord ischemia, however, Lyden and colleagues [51] found no protective effects conferred by three calcium-channel blockers selective for the cerebrovascular system. They proposed that calcium enters the injured neuron by way of channels other than those blocked by these drugs. EXCITATORY AMINO ACID ANTAGONISTS. As mentioned earlier, recent attention has been focused on the excitotoxic neurotransmitter theory of spinal cord injury. Three distinct receptors for aspartate and glutamate have been identified, designated K, Q, and NMDA I20]. Administration of the NMDA receptor antagonists MK801 ]48] and LY [52] has been found to provide some degree of spinal cord protection in rabbit models of spinal cord ischemia.

6 250 REVIEW MAUNEY ET AL Ann ~fhorac Surg SPINAL CORD 1SCHEM1A 1995;59: Decreasing Reperfusion Injury FREE RADICAL SCAVENGERS. A large body of research has been focused on limiting free radical-induced injury during the reperfusion of ischemic tissue. The intravenous administration of the endogenous free radical scavenger SOD just before and after termination of aortic occlusion in animals was found to protect against ischemic periods of 30 minutes or less [531. Our laboratory [54] and other authors [46, 53], however, found that SOD alone is not protective against periods of ischemia of 40 minutes or more. With the conjugation of SOD to polyethylene glycol, the half-life of SOD is increased from minutes to hours. Our laboratory [54] and Granke and associates [55] found that this form of SOD alone effects partial but not complete prevention of neurologic injury after 40 and 60 minutes of spinal cord ischemia, respectively. Despite success in limiting myocardial reperfusion injury [56], the xanthine oxidase inhibitor allopurinol has been shown to be of no benefit in animal models of spinal cord ischemia [31, 53]. Deferoxamine has also been proved beneficial in preventing reperfusion injury in myocardium by chelating free iron, a catalyst for hydroxyl radical formation [57]. Qayumi and colleagues [53] recently reported that deferoxamine is neuroprotective in the setting of spinal cord ischernia; however, a dosedependent toxicity has been observed [57]. Laschinger and co-workers [58] reported that ischemic spinal cord injury was prevented in animals receiving methylprednisolone intravenously. The protective effects of corticosteroids were thought to be related to their ability to stabilize membranes, modulate the immune system, and scavenge for free radicals. The 21-aminosteroids lack any mineralocorticoid or glucocorticoid activity and are potent scavengers of superoxide and lipid peroxyl radicals, as well as extremely potent inhibitors of iron-dependent lipid peroxidation under in vitro conditions [59]. Fowl and associates [60] reported a decreased rate of spinal cord injury in rabbits given the 21- aminosteroid U74006F intravenously. IMMUNE SYSTEM MODULATION. Immune system modulation is a relatively new technique for attenuating reperfusion injury. Giulian and Robertson [61] were able to inhibit the phagocytic and secretory functions of mononuclear phagocytes with chloroquine and colchicine, and demonstrated that the clinical and histopathologic outcomes were improved after 20 minutes of spinal cord ischemia in rabbits treated with these agents. Clark and associates [62] studied the effects of an antibody to the surface glycoprotein CD18, a substance required for leukocytes to adhere to the endothelium, and found a significant reduction in neurologic deficits after 30 minutes of spinal cord ischemia in rabbits given the antibody. ADENOSINE. An expanding body of literature has documented the neuroprotective effects of adenosine and its analogues [63-65]. Adenosine-1 receptors are primarily located in the neural tissue, whereas adenosine-2 receptors reside in the smooth muscle and endothelium of the vasculature. The activation of adenosine-1 receptors decreases neuronal and membrane excitability, thereby limiting the damaging influx of calcium through voltagegated channels. Aspartate and glutamate release is inhibited [63, 64]. Adenosine-2 receptor activation promotes vasodilation, inhibits platelet aggregation, and inhibits neutrophil activation and subsequent free radical production, thus theoretically reversing the tendency toward the no-reflow phenomenon [65]. Recently, in a gerbil model of cerebral ischemia, von Lubitz and Marangos [66] demonstrated significant neuroprotective effects from an exogenously administered adenosine analogue that was given after reperfusion. On the basis of findings from work in our laboratory, we have concluded that a regional infusion of hypothermic adenosine into the excluded infrarenal aorta of the rabbit provides complete protection from spinal cord injury after 40 minutes of spinal cord ischemia, as shown by both clinical and histologic evidence [67]. Summary After decades of research, the hemodynamic and biochemical changes that occur after cross-clamping of the thoracic aorta are beginning to be characterized, as well as the way in which these changes contribute toward producing ischemic spinal cord injury. Despite significant success in preventing ischemic spinal cord injury in animal models through the use of mechanical and pharmacologic interventions, controversy remains regarding the optimal approach for spinal cord protection in human beings. Heparinless perfusion techniques and intercostal reimplantation maximize regional spinal cord blood flow during and after repair. Pharmacologic agents lower spinal cord metabolism, reduce the levels of neurotoxic neurotransmitters, and decrease the formation of free radicals and inflammatory mediators of reperfusion injury. However, no single technique has proved consistently safe and effective in preventing spinal cord injury. It would appear that a combination of distal aortic perfusion and pharmacologic interventions that attenuate ischemic and reperfusion injury holds the most promise. References 1. Svensson LG, Crawford ES, Hess KR, Coselli JS, Sail HJ. Experience with 1509 patients undergoing thoracoabdominal aortic operations. J Vasc Surg 1993; Cox GS, O'Hara PJ, Hertzer NR, Piedmonte MR, Krajewski LP, Beven EG. Thoracoabdominal aneurysm repair: a representative experience. J Vasc Surg 1992;15: Livesay JL, Cooley DA, Ventemiglia RA, et al. Surgical experience in descending thoracic aneurysmectomy with and without adjuncts to avoid ischemia. Ann Thorac Surg 1985;39: Lerberg DB, Hardesty RL, Siewers RD, Zuberbuhler JR, Bahnson HT. Coarctation of the aorta in infants and children: 25 years of experience. Ann Thorac Surg 1982;33:

7 Ann Thorac Surg REVIEW MAUNEY ET AL ;59: SPINAL CORD ISCHEMIA 5. Duhaylongsod FG, Glower DD, Wolfe WG. Acute traumatic aortic aneurysm: the Duke experience from 1970 to J Vasc Surg 1992;15: Hess PJ, Howe HR, Robicsek F, et al. Traumatic tears of the thoracic aorta: improved results using the Bio-Medicus pump. Ann Thorac Surg 1989;48: Hilgenberg AD, Logan DL, Akins CW, et al. Blunt injuries of the thoracic aorta. Ann Thorac Surg 1992;53: Hollier LH, Symmonds JB, Pairolero PC, Cherry KJ, Hallett JW, Gloviczki P. Thoracoabdominal aortic aneurysm repair: analysis of postoperative morbidity. Arch Surg 1988;123: Najafi H. Descending aortic aneurysmectomy without adjuncts to avoid ischemia. Ann Thorac Surg 1993;55: Svensson LG, Crawford ES, Hess KR, Coselli JS, Safi JH. Variables predictive of outcome in 832 patients undergoing repairs of the descending thoracic aorta. Chest 1993;104: Sliwa JA, Maclean IC. Ischemic myelopathy: a review of spinal vasculature and related clinical syndromes. Arch Phys Med Rehabil 1992;73: Svensson LG, Klepp P, Hinder RA. Spinal cord anatomy of the baboon: comparison with man and implications of spinal cord blood flow during thoracic aortic cross-clamping. S Afr J Surg 1986;24: Dommisse GF. The blood supply of the spinal cord. A critical vascular zone in spinal surgery. J Bone Joint Surg [Br] 1974;56: Wadouh F, Lindemann EF, Arndt CM, Hetzer R, Borst HG. The arteria radicularis magna anterior as a decisive factor influencing spinal cord damage during aortic occlusion. J Thorac Cardiovasc Surg 1984;88: Griffiths IR, Pitts LH, Crawford RA, Trench JG. Spinal cord compression and blood flow I: The effect of raised cerebrospinal fluid pressure on spinal cord blood flow. Neurology 1978;28: Blaisdell FW, Cooley DA. The mechanism of paraplegia after temporary thoracic aortic occlusion and its relationship to spinal fluid pressure. Surgery 1962;51: Svensson LG, Grum DF, Bednarki M, Cosgrove DM, Loop FD. Appraisal of cerebrospinal fluid alterations during aortic surgery with intrathecal papaverine administration and cerebrospinal fluid drainage. J Vasc Surg 1990;11: Maughn RE, Mohan C, Levy R, Cunningham JN, Jacobwitz I, Marini C. Effects of exsanguination and sodium nitroprusside on compliance of the spinal canal during aortic occlusion. J Surg Res 1992;52: Krause GS, White BC, Aust SD, Nayini NR, Kumar K. Brain cell death following ischemia and reperfusion: a proposed biochemical sequence. Crit Care Med 1988;16: Rothman S. Synaptic release of excitatory amino acid neurotransmitter mediates anoxic neuronal death. J Neurosci 1984;4: Braughler JM, Hall ED. Central nervous system trauma and stroke. I. Biochemical considerations for oxygen radical formation and lipid peroxidation. Free Radic Biol Med 1989; 6: Barone GW, Joob AW, Flanagan TL, Dunn CE, Kron IL. The effect of hyperemia on spinal cord function after temporary thoracic aortic occlusion. J Vasc Surg 1988;8: Jacobs TP, Kempski O, McKinley D, Dutka AJ, Hallenbeck JM, Feuerstein G. Blood flow and vascular permeability during motor dysfunction in a rabbit model of spinal cord ischemia. Stroke 1992;23: Rokkas CK, Sundaresan S, Shuman TA, et al. Profound systemic hypothermia protects the spinal cord in a primate model of spinal cord ischemia. J Thorac Cardiovasc Surg 1993;106: Verdant AP, Page A, Cossette R, Dontigny L, Page P, Baillot R. Surgery of the descending thoracic aorta: spinal cord protection with the Gott shunt. Ann Thorac Surg 1988;46: Olivier HF, Maher TD, Liebler GA, Park SB, Burkholder JA, Magovern GJ. Use of the Bio-Medicus centrifugal pump in traumatic tears of the thoracic aorta. Ann Thorac Surg 1984;38: Crawford ES, Walker HS, Saleh SA, Normann NA. Graft replacement of aneurysm in descending thoracic aorta: results without bypass or shunting. Surgery 1981;89: Najafi H, Javid H, Hunter J, Serry C, Monson D. Descending aortic aneurysmectomy without adjuncts to avoid ischemia. Ann Thorac Surg 1980;30: Laschinger JC. Spinal cord injury following surgical correction of acute aortic disruption. Semin Thorac Cardiovasc Surg 1992;4: Svensson LG, Crawford ES. Aortic dissection and aortic aneurysm surgery: clinical observations, experimental investigations, and statistical analysis, Part III. Curr Probl Surg 1993;30: Svensson LG, Von Ritter CM, Groeneveld HT, et al. Crossclamping of the thoracic aorta: influence of aortic shunts, laminectomy, papaverine, calcium channel blocker, allopurinol, and superoxide dismutase on spinal cord blood flow and paraplegia in baboons. Ann Surg 1986;206: Svensson LG, Stewart RW, Cosgrove DM, et al. Intrathecal papaverine for the prevention of paraplegia after operation on the thoracic or thoracoabdominal aorta. J Thorac Cardiovasc Surg 1988;96: Maughan RE, Mohan C, Nathan IM, et al. Intrathecal perfusion of an oxygenated perfluorocarbon prevents paraplegia after aortic occlusion. Ann Thorac Surg 1992;54: 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. J Vasc Surg 1986;3: Fereshatian A, Kadir S, Kaufman SL, et al. Digital subtraction spinal cord angiography in patients undergoing thoracic aneurysm surgery. Cardiovasc Intervent Radiol 1989;12: Svensson LG, Patel V, Robinson MF, Ueda T, Roehm JO, Crawford ES. Influence of preservation or perfusion of intraoperatively identified spinal cord blood supply on spinal motor evoked potentials and paraplegia after aortic surgery. J Vasc Surg 1991;13: Bower TC, Murray MJ, Gloviczki P, Yaksh TL, Hollier LH, Pairolero PC. Effects of thoracic aortic occlusion and cerebrospinal fluid drainage on regional spinal cord blood flow in dogs: correlation with neurologic outcome. J Vasc Surg 1988;9: McCullough JL, Hollier LH, Nugent M. Paraplegia after thoracic aortic occlusion: influence of cerebrospinal fluid drainage. J Vasc Surg 1988;7: Acher CW, Wynn MM, Hoch JR, Popic PM, Turnipseed WD. Combined use of cerebral spinal fluid drainage and naloxone reduces the risk of paraplegia in thoracoabdominal aneurysm repair. J Vasc Surg 1994;19: Crawford ES, Svensson LG, Hess KR, et al. A prospective randomized study of cerebrospinal fluid drainage to prevent paraplegia after high-risk surgery on the thoracoabdominal aorta. J Vasc Surg 1990;13: Robertson CS, Foltz R, Grossman RG, Goodman JC. Protection against experimental ischemic spinal cord injury. J Neurosurg 1986;64: Kouchoukos NT, Wareing TH, Izumoto H, Klausing W, Abboud N. Elective hypothermic cardiopulmonary bypass and circulatory arrest for spinal cord protection during operations on the thoracoabdominal aorta. J Thorac Cardiovasc Surg 1990;99: Colon R, Frazier OH, Cooley DA, McAllister HA. Hypothermic regional perfusion for protection of the spinal cord during periods of ischemia, Ann Thorac Surg 1987;43: Berguer R, Porto J, Fedoronko B, Dragovic L. Selective deep hypothermia of the spinal cord prevents paraplegia after aortic cross-clamping in the dog model. J Vasc Surg 1992;15: Fehrenbacher JW, McCready RA, Hormuth DA, et al. One

8 252 REVIEW MAUNEY ET AL Ann Thorac Surg SPINAL CORD ISCHEMIA 1995;59: stage segmental resection of extensive thoracoabdominal aortic aneurysms with left-sided heart bypass. J Vasc Surg 1993;18: Kirshner DL, Kirshner RL, Heggeness LM, Deweese JA. Spinal cord ischemia: an evaluation of pharmacologic agents in minimizing paraplegia after aortic occlusion. J Vasc Surg 1989;9: Breckwoldt WL, Genco CM, Connolly RJ, Cleveland RJ, Diehl JT. Spinal cord protection during aortic occlusion: efficacy of intrathecal tetracaine. Ann Thorac Surg 1991;51: Yum SW, Faden AI. Comparison of the neuroprotective effects of the n-methyl-d-aspartate antagonist MK-801 and the opiate-receptor antagonist nalmefene in experimental spinal cord ischemia. Arch Neurol 1990;47: Steen PA, Newberg LA, Milde JH, Mitchenfelder JD. Nimodipine improves cerebral blood flow and neurologic recovery after complete cerebral ischemia in the dog. J Cereb Blood Flow Metab 1983;3: Schittek A, Bennink GB, Cooley DA, Langford LA. Spinal cord protection with intravenous nimodipine: a functional and morphologic evaluation. J Thorac Cardiovasc Surg 1992; 104: Lyden PD, Zivin JA, Kochler A, Mazzarella V. Effects of calcium channel blockers on neurologic outcome after focal ischemia in rabbits. Stroke 1988;19: Madden KP, Clark WM, Kochhar A, Zivin JA. Efficacy of LY233053, a competitive glutamate antagonist, in experimental central nervous system ischemia. J Neurosurg 1992; 76: Qayumi AK, Janusz MT, Jamieson WE, Lyster DM. Pharmacologic interventions for prevention of spinal cord injury caused by aortic cross-clamping. J Thorac Cardiovasc Surg 1992;104: Agee JM, Flanagan T, Blackbourne LH, Kron IL, Tribble CG. Reducing postischemic paraplegia using conjugated superoxide dismutase. Ann Thorac Surg 1991;51: Granke K, Hollier LH, Zdrahal P, Moore W. Longitudinal study of cerebral spinal fluid drainage in polyethylene glycol-conjugated superoxide dismutase in paraplegia associated with thoracic aortic cross-clamping. J Vasc Surg 1991; 13: Godin DV, Garnett ME. Altered antioxidant states in the ischemic/reperfused rabbit myocardium: effects of allopurinol. Can J Cardiol 1989;5: Reddy BR, Kloner RA, Przynlerk K. Early treatment with deferoxamine limits myocardial ischemic/reperfusion injury. Free Radic Biol Med 1989;7: Laschinger JC, Cunningham JN, Cooper MM, Krieger K, Nathan IM, Spencer FC. Prevention of ischemic spinal cord injury following aortic cross-clamping: use of corticosteroids. Ann Thorac Surg 1984;38: Braughler JM, Pregenzer JF, Chase RL, Duncan LA, Jacobsen EJ, McCall JM. Novel aminosteroids as potent inhibitors of iron dependent lipid peroxidation. J Biol Chem 1987;262: Fowl RJ, Patterson RB, Gewirtz RJ, Anderson DK. Protection against postischemic spinal cord injury using a new 21- aminosteroid. J Surg Res 1990;48: Giulian D, Robertson C. Inhibition of mononuclear phagocytes reduces ischemic injury in the spinal cord. Ann Neurol 1990;27: Clark WM, Madden KP, Rothlein R, Zivin JA. Reduction of central nervous system ischemic injury in rabbits using leukocyte adhesion antibody treatment. Stroke 1991;22: Schubert P, Kreutzberg GW. Cerebral protection by adenosine. Acta Neurochir 1993;57(Suppl): Simpson RE, O'Regan MH, Perkins LM, Phillis JW. Excitatory transmitter amino acid release from the ischemic rat cerebral cortex: effects of adenosine receptor agonists and antagonists. J Neurochem 1992;58: Rudolphi KA, Schubert P, Parkinson FE, Fredholm BB. Neuroprotective role of adenosine in cerebral ischemia. Trends Pharmacol Sci 1992;13: von Lubitz DK, Marangos PJ. Cerebral ischemia in gerbils: postischemic administration of cyclohexyl adenosine and 8-sulfophenyl-theophylline. J Mol Neurosci 1990;2: Herold JA, Kron IL, Langenburg SE, Blackbourne LH, Tribble CG. Complete prevention of post-ischemic spinal cord injury by means of regional infusion with hypothermic saline and adenosine. J Thorac Cardiovasc Surg 1994;107: