- Email: [email protected]
Blunt injuries of the thoracic aorta
Blunt Injuries of the Thoracic Aorta Alan D. Hilgenberg, MD, Diana L. Logan, RN, Cary W. Akins, MD, Mortimer J. Buckley, MD, Willard M. Daggett, MD, Gus J. Vlahakes, MD, and David F. Torchiana, MD Surgical Cardiovascular Unit, Massachusetts General Hospital, and Department of Surgery, Harvard Medical School, Boston, Massachusetts
We managed 51 patients with thoracic aortic injuries caused by blunt trauma between 1977 and 1990. Fortynine injuries were located in the upper descending aorta and one each in the ascending aorta and aortic arch. Three patients arrived moribund and underwent thoracotomy for resuscitation, and all died. The diagnosis was confirmed by aortography in 48. One patient died of aortic rupture, 1 died of hypoxemia, and 1 refused operation and died. Forty-four patients had aortic repair, 42 with graft insertion. Gott shunts were placed in 23 with 3 cases of paraplegia (13%). Simple cross-clamping was used in 19 with 1 case of paraplegia (5.2%). We found statistically significant differences between the cross-clamp times of patients without paraplegia compared with those in whom paraplegia developed in both the shunt and no-shunt groups. Logistic regression anal-
ysis showed that the only factor significantly associated with paraplegia was cross-clamp time. There were two postoperative deaths (4.4%). Seven patients had medical therapy initially and aortic repair was delayed to allow other injuries to stabilize. Before aortic repair, 18 patients had intraarterial pressure monitoring and 34 received &blockers or antihypertensive drugs. We conclude that aortic repair with graft insertion is usually successful in nonmoribund patients, simple crossclamping is associated with a relatively low risk of paraplegia, the incidence of paraplegia is directly associated with the duration of cross-clamp time, and selected patients can be managed medically while awaiting aortic repair.
B
tially higher in patients treated with heparinization and bypass than in those who were shunted and received no heparin. Hemorrhage was the major cause of death in the bypass group. We also noted that aortic repair could be safely delayed in selected patients who were managed medically while other severe injuries were stabilized.
lunt traumatic disruption of the thoracic aorta remains a highly lethal injury. It appears that only preventive measures that may reduce the frequency of this injury could have any effect on the mortality in the prehospital phase [l]. Only the patients who reach the hospital with the false aneurysm intact and a blood pressure and cardiac rhythm present have a chance of survival. In these patients mortality and morbidity may arise from the multiple severe associated injuries, from delay in recognition of the aortic injury, from sudden rupture of the false aneurysm during angiography and transportation to the operating room, and from complications of aortic repair such as hemorrhage, paraplegia, and renal failure. Controversial issues involve diagnostic studies, timing of aortic operation, method of aortic repair, and spinal cord protection during aortic clamping [2-71. We reviewed our experience with all patients treated at the Massachusetts General Hospital with acute blunt traumatic disruption of the thoracic aorta from January 1977 to June 1990. This time period was selected because 1977 marked the end of the use of systemic heparinization and partial cardiopulmonary bypass for descending aorta repairs. Experience from our institution between 1970 and 1980 has been previously reported [7]. In that study we found that the mortality rate for aortic repair was substanPresented at the Twenty-seventh Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Feb 18-20, 1991. Address reprint requests to Dr Hilgenberg, Department of Thoracic Surgery, Massachusetts General Hospital, 32 Fruit St, Boston, MA 02114.
0 1992 by The Society of Thoracic Surgeons
(Ann Thoruc Surg 1992;53:233-9)
Material and Methods The medical records of all patients with acute blunt traumatic disruption of the thoracic aorta were reviewed. Particular attention was directed to the emergency ward record, the operative and anesthetic reports, the postoperative course, and autopsy reports. Statistical analysis included comparison of paraplegia rates by the Fisher exact test, cross-clamp times by unpaired t tests and one way analysis of variance, and factors associated with paraplegia by multivariate logistic regression analysis.
Clinical Material Between 1977 and 1990 we treated 51 patients with acute blunt injuries to the thoracic aorta. There were 43 male and 8 female patients whose average age was 35 years, with an age range of 16 to 70 years. Circumstances of the accidents were as follows: drivers of motor vehicles, 33; passengers in motor vehicles, 8; pedestrians struck by autos, 4; motorcycle riders, 4; and victims of falls, 2. Information regarding seat belt use and ejection of the victim from the vehicle was not available in sufficient numbers to draw any conclusions about these factors. 0003-4975/92/$5.00
234
Ann Thorac Surg 1992;53:23>9
HILGENBERG ET AL BLUNT AORTIC TRAUMA
Four patients came directly to our hospital, and 47 patients were referred from other institutions. Initial arterial blood pressures were unobtainable in 3 patients and less than 100 mm Hg systolic in an additional 7 patients. Initial systolic arterial pressures were greater than 140 mm Hg in 22 patients. Decreased femoral pulses were noted in 8 patients. A visible or palpable chest wall injury was apparent in only 20 patients (39%). Four patients had a flail chest. Seven patients had an audible murmur, and in 2 of these it was probably due to preexisting heart disease. Reports of chest radiographs were available for review in 48 patients, and all of them showed superior mediastinal widening, although in 3 patients the widening was not present on the initial films. Other common findings were obscuring of the aortic knob contour in 10, an extrapleural apical cap in 7, lateral shift of mediastinal structures in 13, lung contusion in 20, rib fractures in 20, pleural effusions in 20, and pneumothorax in 10. Nine patients had computed tomographic scans of the chest, and in only 4 were findings suggestive of aortic injury. Three were read as probable aortic injury, one as descending aortic dissection, and the other five showed only mediastinal hematomas and pleural effusions. Associated injuries were common and severe, and included head injuries, intrathoracic organ damage, abdominal injuries, fractures, and peripheral vascular and nerve injuries. Details of the associated injuries are summarized in Table 1. In 48 patients the diagnosis of aortic injury was confirmed by aortography, and in 3 moribund patients the aortic injury was found during emergency resuscitative thoracotomy. Forty-nine injuries were located in the upper descending aorta, one in the ascending aorta, and one in the aortic arch with involvement of the innominate and left common carotid arteries. Our recent policy is to insert radial artery catheters for continuous pressure monitoring in patients strongly suspected of aortic injury, and to treat hypertension if present. Eighteen patients had arterial lines inserted in the emergency ward. Throughout the period of this study, we have commonly administered intravenous propranolol to patients who are not hypotensive and nitroprusside infusions to those who are hypertensive during their evaluation. A total of 34 patients received pblocking drugs or vasodilators or both before their aortic repairs.
Surgical Procedures Three patients arrived in our emergency ward without pulse or blood pressure and underwent resuscitative thoracotomy, and all 3 died. One patient arrested during transport from the angiography suite to the operating room and had a thoracotomy and aortic clamping, but he could not be resuscitated. Death was due to rupture of the false aneurysm. One patient died of hypoxemia and did not have an aortic repair. One patient refused operation and died (his auto accident was probably a suicide). One patient received long-term medical therapy because of paraplegia resulting from a thoracic spine fracture with direct spinal cord damage. The remaining 44 patients had
Table 1. Associated Znjuries Injury Head injuries Concussion Cerebral hematomas requiring operation Thoracic injuries Cardiac contusion Lung contusion Right atrial rupture Tracheal laceration Left diaphragm rupture Abdominal injuries Liver laceration Ruptured spleen Retroperitoneal hematoma Fractures Ribs Sternum Clavicle Facial bones Spine Upper extremity Lower extremity Pelvis Vascular injuries Femoral artery Peripheral nerve palsies
No. of Patients
15 5 4 20 1 1 1 6
5 6 20 1 4 8 5 11 25 13 2 8
aortic repairs: 42 with descending aortic injuries and 1 each with injury to the ascending aorta and the aortic arch. Thirty-three patients underwent immediate aortic repair. In 6 patients aortic repair was delayed 4 days to 4 months to allow stabilization of other severe injuries while medical treatment was administered to reduce the risk of aortic rupture. Five of the 6 patients had severe neurological injuries that were responsible for the delays, and 1 patient had a severe pulmonary contusion. In 4 patients the diagnosis of aortic injury was not made until 2 to 5 days after the accident, and aortic repair was accomplished immediately after their arrival in our hospital. One patient had a delayed diagnosis of aortic injury and his repair was delayed while electrocardiographic changes were evaluated. Therefore, aortic repair was performed immediately after confirmation of the diagnosis of aortic injury in 84% of the patients. The ascending aortic injury was repaired through a median sternotomy on cardiopulmonary bypass with graft insertion. The aortic arch injury was repaired through a median sternotomy without bypass. A graft was placed from the ascending aorta to the innominate artery; a side arm graft was then placed to the left common carotid artery using electroencephalographic monitoring, and the aortic arch disruption was closed directly (Fig 1). The descending aortic injuries were all exposed through
HILGENBERG ET AL BLUNT AORTIC TRAUMA
Ann Thorac surg 1992;53:233-9
235
Fig 1. (A)Initial step in the repair of a false aneurysm of the aortic arch involving the innominate artery origin. An 8-mm Dacron graft is anastomosed to the ascending aorta. (B) Using electroencephalographic monitoring, the proximal and distal ends of the innominate artery are clamped and the distal anastomosis of the graft to the innominate artery is constructed. Inspection of the aortic arch showed extension of the intimal injury beyond the clamp into the origin of the left common carotid artery. Therefore a second Dacron graft, 6 mm in diameter, is anastomosed to the innominate nrtery graft as a side-arm before release of the clamps. (C) With the innominate artery perfused, the left common carotid artery is clamped and a larger clamp applied to the aortic arch. The distal anastomosis of the side-arm graft to the left common artery is completed. ( D ) To complete the repair, the aortic arch is closed directly, obliterating the intimal injury at the origins of the innominate and left common carotid arteries.
a left posterolateral thoracotomy with one-lung ventilation through a double-lumen endotracheal tube. The proximal clamp was applied between the left common carotid and left subclavian arteries in all cases. Simple cross-clamping was used in 19 patients and Gott shunts in 23. Shunt insertion sites are shown in Table 2. Dacron grafts were placed in all except 2 patients whose injuries were small and were sutured directly. Operation for associated injuries was frequently re-
Table 2. Gott Shunt lnsertion Sites in 23 Patients Site Proximal Ascending aorta Aortic arch Left ventricular apex Distal Descending aorta Left femoral artery
No. of Patients 15 4
quired. Five patients underwent intracranial procedures before aortic repair. Concomitant intrathoracic procedures included suture of tracheal laceration, closure of right atrial rupture, and repair of ruptured diaphragm. Fifteen patients had laparotomies, five of which were performed before aortic repair. Open fracture reduction was performed in 28 patients, three of these before aortic repair. One patient required reexploration for bleeding. Vascular repair was necessary in 2 patients after shunt removal; 1 had a vein patch graft repair of a femoral artery, and 1 needed multiple repair sutures in the left ventricular apex. Five patients had tracheostomies. Persistent postoperative hypertension was extremely common. A total of 39 patients received antihypertensive agents after hospital discharge. Only 3 patients discharged after aortic repair did not require drug therapy for hypertension.
4
17 6
Results The overall mortality was 8 of 51 patients (15.7%),which includes those who died before an attempt at aortic repair.
236
HILGENBERG ET AL BLUNT AORTIC TRAUMA
There were two postoperative deaths in the aortic repair group of 44 patients (4.5% mortality). The deaths, one each in the shunt and no-shunt groups, were due to respiratory failure, sepsis, and renal failure. The average age of these 2 patients was 68 years, and both were female. Paraplegia developed in 4 patients out of a total of 42 at risk (9.5%). The paraplegia rate was 1 of 19 in the no-shunt group (5.2%)and 3 of 23 in the Gott shunt group (13.0%).The difference in paraplegia rates was not statistically significant (p = 0.6 using the Fisher exact test). Aortic cross-clamp times were: no shunt, no paraplegia, 32 minutes; no shunt, paraplegia, 50 minutes; Gott shunt, no paraplegia, 45 minutes; Gott shunt, paraplegia, 80 minutes. One-way analysis of variance established differences among these groups. Unpaired t tests found significant differences between the cross-clamp times in the patients without paraplegia compared with those in whom paraplegia developed in both the shunt and noshunt groups (p = 0.02 in the no-shunt group and p = 0.0005 in the Gott shunt group). Using multivariate logistic regression analysis, we evaluated 19 factors for their association with the occurrence of paraplegia. Variables included presence of a shunt, cross-clamp time, hypotension, nitroprusside administration during aortic clamping, and administration of drugs such as mannitol, steroids, and magnesium sulfate before aortic clamping, as well as additional factors. The only factor that had a significant association with paraplegia was cross-clamp time. The presence or absence of a shunt and the administration of vasodilators or other drugs had no significant effect on paraplegia development. However, because of the relatively small number of paraplegic patients, there is a possibility of failure of association of other factors. A logistic model of probability relating the risk of paraplegia to cross-clamp time is shown in Figure 2. The calculated probability of paraplegia is less than 1%with a clamp time of 30 minutes and less than 5% with a clamp time of 45 minutes, but the curve rises steeply at 50 minutes. Although the average cross-clamp time in the Gott shunt patients was significantly longer than in the no-shunt patients, multivariate analysis indicated that a shunt did not offer protection from paraplegia or extend the “safe” cross-clamp time.
Comment If survival of patients with blunt traumatic aortic disruption is to be improved, the diagnosis must be made promptly, methods to reduce the risk of rupture instituted, and urgent surgical repair carried out in most patients. If the aortic pseudoaneurysm ruptures, even if the patient is in the hospital, it is too late to salvage the patient. In this series, none of the patients who had aortic rupture with unobtainable blood pressure survived despite attempts at aortic clamping and resuscitation. Other authors have also noted the uniformly fatal outcome of patients with massive bleeding from aortic rupture [8, 91. Perhaps a method of very rapid infusion of large volumes
Ann Thorac Surg 1992;53:23>9
1
/--
0.0 0.8 0.7 P r
0.6 b a b 0.5 i I i 0.4 t 0
Y
0.3 0.2 0.1
10
20
30
so 60 TO Clamp Time ( m i d
40
a0
90
100
Fig 2 . Logistic model of the probability of the occurrence of paraplegia compared with aortic cross-clamp time. The dotted lines are 70% confidence limits.
of blood could be of benefit in this situation [9]. However, with our current technology, the patient’s best chance for survival is to arrive in the operating room with the pseudoaneurysm still intact. We have used early continuous arterial pressure monitoring and administration of drugs to decrease the forces applied to the aortic wall in patients suspected of having aortic disruption based on the history of the accident and mediastinal widening on chest radiographs. Thoracic computed tomographic scanning is omitted because it does not confirm or exclude aortic injury reliably. The patient is taken promptly to the angiography suite for an aortogram and is attended by surgical residents and special nurses who continue aggressive monitoring, volume administration, and blood pressure control. This early treatment method appears to be generally successful as only 1patient in this series died of aortic rupture after aortography during transportation to the operating room, and he did not have a radial artery catheter and the vigorous treatment we now advocate. Timing of aortic repair requires careful consideration and individualization. In most patients prompt aortic repair is indicated, as was done in 84% of the patients in this series. However, in 7 patients aortic repair was delayed 4 days to 4 months after aortography to allow stabilization of other injuries. The most frequent reason for delaying aortic operation was a severe head injury with either the need for intracranial operation or an uncertain prognosis on initial evaluation. Head injuries were responsible for the delay in 5 of the 7 patients. One
Ann Thorac Surg 1992;5323>9
patient had severe hypoxemia from pulmonary contusion and other injuries and could not have tolerated one-lung ventilation until his lung injury improved. One other delayed patient had electrocardiographic changes that were evaluated before repair. Although we did not encounter it in this series, another reason for delay could be presence of sepsis in a patient in whom the diagnosis of aortic injury is made late. The results from this series and our previously reported experience [7] support the safety of medical therapy in these highly selected patients, as none of them experienced aortic rupture while waiting for repair. Prioritizing surgical procedures also requires careful consideration. Intracranial operation may have priority in the hemodynamically stable patient. In the patient in unstable condition with evidence of exsanguinating abdominal hemorrhage, laparotomy is indicated before aortic repair. If the airway is injured, this requires immediate attention. In this series, procedures prior to aortic repair were craniotomy or insertion of intracranial pressure monitors in 5 patients, laparotomy in 5 patients, and repair of tracheal laceration in 1 patient. The other 10 laparotomies and nearly all of the orthopedic procedures were performed after aortic repair. The only open fracture procedures done before aortic repair were in the delayed patients with head injuries. Surgical repair of the aortic injury consisted of Dacron graft insertion in 95% of the cases because the aortic intima was completely or nearly completely transected and the ends separated. The intimal damage was substantially less than circumferential in only 2 cases, and these were both sutured directly. In most instances, sutureless intraluminal grafts were considered unsatisfactory because of the small diameter of the aorta in most of these young patients and lack of space distal to the left subclavian artery for seating of the proximal cuff. The most controversial and unsolved aspects of descending aortic repair are the role of distal perfusion and the causes and incidence of paraplegia. During the period of this study, we used two techniques and the choice depended on the preference of the surgeon and the degree of stability of the patient. Forty-five percent of the patients had simple cross-clamping without distal perfusion, and in 55% a Gott shunt was placed with most patients receiving ascending to descending aortic shunting. No consistent effort was made to monitor distal aortic pressure or flow in shunted patients. Some patients received a variety of pharmacologic agents to possibly aid in spinal cord protection, and these included steroids, pentothal, magnesium sulfate, and mannitol. Nitroprusside was used for control of proximal hypertension if needed. Paraplegia occurred in 9.5% of the patients at risk, and there was no statistical difference between the incidence in the shunt and no-shunt groups (13% and 5.2%, respectively). Of all the variables evaluated by logistic regression analysis, the only factor significantly associated with paraplegia was cross-clamp time, and the association was a very strong one. Factors that had no association with paraplegia included admission systolic blood pressure less than 100 mm Hg, intraoperative
HILGENBERG ET AL BLUNT AORTIC TRAUMA
237
hypotension, administration of nitroprusside or other vasodilators during aortic clamping, administration of drugs before aortic clamping that may protect the spinal cord, and presence or absence of a shunt. A most important variable that cannot be assessed preoperatively or by statisticalanalysis is the individual variation of spinal cord blood supply. The length of the cross-clamp time is mainly a reflection of the complexity of the injury and the degree of difficulty of the repair; to a lesser extent it reflects the speed of the surgeon [101. Problems leading to longer cross-clamp times in this series were difficulty with the proximal aorta with extension of the injury into the aortic arch close to the cross-clamp, and marked obesity of a patient with a very small aorta. Cross-clamp times in our patients in whom paraplegia developed were 70, 80, and 90 minutes in the shunted patients and 50 minutes in a no-shunt patient. Although the average cross-clamp time in the Gott shunt group was significantly longer than that of the no-shunt patients, logistic regression analysis indicated that the presence of a shunt had no additional protective effect for any given cross-clamp time. This finding is similar to that reported from the Maryland Shock Trauma Center [9] and different from that reported by the Alabama group [ll]. We strongly endorse the efforts of investigators trying to improve the results of descending thoracic aortic operations, particularly with respect to reducing the incidence of paraplegia. It seems clear from our data and those of others that in the acute trauma situation, a Gott shunt without distal pressure or flow monitoring is not consistently protective of the spinal cord. We acknowledge that excellent results have been reported in predominately elective descending thoracic aortic aneurysm repairs using a Gott shunt with flow monitoring [12]. Partial cardiopulmonary bypass has its proponents, with evidence that spinal cord perfusion may be well maintained [13], but our past experience with systemic heparinization in acute trauma patients has been unsatisfactory because of hemorrhagic complications, making this an unacceptable method for use in this setting. Good results with left atrial-femoral bypass using a centrifugal pump without heparin have been reported in relatively small groups of patients [14,15]. This approach appears promising in that flows similar to those of partial cardiopulmonary bypass can be obtained and heparin is avoided. Other adjuncts such as sensory evoked potential monitoring [16], spinal fluid drainage, and pharmacologic agents are probably of minimal help, especially in the acute trauma situation. The most desirable technique for these emergency operations in unstable patients with multiple injuries is a quick and uncomplicated method of aortic repair with a low mortality rate and as low a paraplegia rate as possible. In our hands, simple cross-clamping with graft insertion meets these criteria at the present time. We hope that with additional research, improvements can be made. However, because of variability in spinal cord blood supply and variability in extent and complexity of the aortic injury, the paraplegia rate will probably never be zero.
238
HILGENBERG ET AL BLUNT AORTIC TRAUMA
We gratefully acknowledge the assistance of John Newel1 for the statistical analysis and Ruthellen McKinnon for preparation of the manuscript.
References 1. Eddy AC, Rusch VW, Fligner CL, Reay DT, Rice CL. The epidemiology of traumatic rupture of the thoracic aorta in children: a 13-year review. J Trauma 1990;30989-92. 2. Mattox KL, Holzman M, Pickard LR, Beall AC, DeBakey ME. Clamphepair: a safe technique for treatment of blunt injury to the descending thoracic aorta. Ann Thorac Surg 1985;40: 456-63. 3. Kirsh MM, Behrendt DM, Orringer MB, et al. The treatment of acute traumatic rupture of the aorta: a ten year experience. Ann Surg 1976;184:308-16. 4. Appelbaum A, Karp RB, Kirklin JW. Surgical treatment for closed thoracic aortic injuries. J Thorac Cardiovasc Surg 1976;71:458-60. 5. Orringer MB, Kirsh MM. Primary repair of acute aortic disruption. Ann Thorac Surg 1983;35:672-5. 6. Borman KR, Aurbakken CM, Weigelt JA. Treatment priorities in combined abdominal and aortic trauma. Am J Surg 1982;M728-32. 7. Akins CW, Buckley MJ, Daggett WM, McIlduff JB, Austen WG. Acute traumatic disruption of the thoracic aorta: a ten-year experience. Ann Thorac Surg 1981;31:305-9. 8. Clark DE, Zeiger MA, Wallace KL, Packard AB, Nowicki ER. Blunt aortic trauma: signs of high risk. J Trauma 1990;30: 701-5.
Ann Thorac Surg 1992;53:233-9
9. Cowley RA, Tumey SZ, Hankins JR, Rodriquez A, Attar S, Shankar BS. Rupture of thoracic aorta caused by blunt trauma. A fifteen-year experience. J Thorac Cardiovasc Surg 1990;100652-61. 10. Mattox KL. Fact and fiction about management of aortic transection. Ann Thorac Surg 1989;481-2. 11. Katz NM, Blackstone EH, Kirklin JW, Karp RB. Incremental risk factors for spinal cord injury following operation for acute traumatic aortic transection. J Thorac Cardiovasc Surg 1981;81:669-74. 12. Verdant A. 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: 147-54. 13. Laschinger JC, Isumoto H, Kouchoukos NT. Evolving concepts in prevention of spinal cord injury during operations on the descending thoracic and thoracoabdominal aorta. Ann Thorac Surg 1987;44:667-74. 14. Olivier HF Jr, Maher TD, Liebler GA, Park SB, Burkholder JA, Magovem GJ. Use of the Biomedicus centrifugal pump in traumatic tears of the thoracic aorta. Ann Thorac Surg 1984;38:58&91. 15. Hess PJ, Howe HR Jr, Robicsek F, et al. Traumatic tears of the thoracic aorta: improved results using the Bio-Medicus pump. Ann Thorac Surg 1989;486-9. 16. Crawford ES, Mizrahi EM, Hess KR, CoseIli JS, Safi HJ, Pate1 VM. The impact of distal aortic perfusion and somatosensory evoked potential monitoring on prevention of paraplegia after aortic aneurysm operation. J Thorac Cardiovasc Surg 1988;95:357-67.
DISCUSSION DR FRANCIS ROBICSEK (Charlotte, NC): I thank The Society for the opportunity to discuss this thought-provoking report. As far as the methodology is concerned, I have a few questions, all related to the lack of randomization. This study encompasses a period of 14 years. Is there a bias in comparing earlier Gott-shunted cases with more recent nonshunted cases? Also, the shunted group had a much higher paraplegia rate but also a much longer cross-clamp time. Is it possible that the more severely injured patients were shunted and the less severely injured patients were not shunted? I would also like to call attention to the fact that we are more or less fighting a strawman opponent, because a question in the mind of most of us is not to shunt with a Gott shunt or just simply cross-clamp and sew, but to use other bypass methods such as Bio-Medicus or Bio-Medicus-type bypass systems. In the material we have collected with my partner, Phillip Hess, we have 36 patients who survived reconstructive operation for traumatic aortic rupture. Femoral vein to femoral artery oxygenated bypass was used in 5. None of these patients suffered paraplegia, but we had two late deaths. Sixteen patients were operated on using a simple cross-clamp without shunt. Of the 16, 4 had paraplegia, 4 had renal shutdown, and 5 died late. Recently we have operated on a series of 15 patients using nonanticoagulated left atrium to femoral artery bypass, with cross-clamp times ranging from 21 minutes to 2 hours. There was neither death nor paraplegia in this system. We were also disturbed that in the simple clamp, no-shunt groups, we had 2 patients with
less than 32-minute cross-clamp times who ended up with paraplegia. We certainly are not here to dispute the virtues of fast and expeditious operation. Reconstruction of the ruptured aorta in a very short time is possible in some cases by some surgeons but not in all situations for all surgeons. And even if it is possible to do it in such a short time, in our experience, contrary to Hilgenberg and associates, we believe that whenever the clinical situation allows, nonanticoagulated left atrium to femoral artery bypass may provide some additional protection against the most feared complication of paraplegia. I learned a lot about this complication by serving on your Committee on Medico-Legal Affairs, and I found that about one third of major malpractice suits involving cardiovascular thoracic operations were connected to nonshunted paraplegias stemming from descending aortic reconstruction. I think the Bio-Medicus shunt is a very valuable protection not only for the patient's lower spine but also for that of the surgeons. DR STEPHEN Z. TURNEY (Baltimore, MD): I compliment Dr Hilgenberg and associates on their comprehensive report and congratulate them on their results. We recently reported our own updated experience of 114 patients from Baltimore at the meeting of the Association for Thoracic Surgery and came to essentially the same conclusions, that morbidity and mortality are high and are complex problems. Aortoaortic shunt does not necessarily protect patients from paraplegia. It is our practice not to routinely use shunt.
Ann Thorac Surg 1992;53:233-9
A natural history survival curve for blunt thoracic aortic rupture shows that most deaths occur within 12 to 24 hours of injury. Very early transport of the accident victim, as in our trauma systems, pushes the patient population one is faced with to the steep, high mortality rate portion of the curve where patients are exsanguinating from their aortic and other injuries. The other risk factors we identified included hypotension on admission, a large presenting hemothorax, the surgeon’s credentials, and the presence of other serious injury. My question is, does early pharmacologic intervention present problems of its own in management, such as the evaluation of hidden bleeding, large blood pressure swings as anesthesiologists struggle to regulate blood volume and pressure, or the steal of blood from the spinal cord during cross-clamping?We scrupulously snare intercostal arteries to avoid back-bleeding and cord steal during repair, but we remain concerned that pharmacologic steal may be taking place into a dilated, low-pressure, systemic vascular bed if the patient is receiving vasodilators. DR MANUEL DE JESUS ANTUNES (Coimbra, Portugal): In their well-presented large series, Dr Hilgenberg and associates have clearly demonstrated that the length of the aortic crossclamping is the most important risk factor for paraplegia, irrespective of the technique used. That much had also surfaced from a review of the literature done from our previous institution by Dr Svenson some years ago. Hence, all technical details that may help reducing the cross-clamp time will be of major importance. In an article published in The Annals 3 years ago, I suggested that using straight aortic cross-clamping without shunts or cardiopulmonary bypass and incision of the descending aorta, including the covering hematoma in the area of rupture, would expose the tear, usually located on the opposite wall when the tear was partial. The tear could then be repaired, from within the lumen of the aorta, more clearly and easily with a continuous suture without further dissection. When there was complete separation of the aortic edges, the repair could also be facilitated by working from inside the lumen, either with direct approximation or by interposition of a graft. Finally, the latter could also be made easier by incising it longitudinally and transforming it into a circular patch. Using these techniques, we have now operated on 17 consecutive patients without paraplegia and with only 1 case of transient paraparesis. I believe that this is a consequence of a short mean aortic cross-clamp time of 22 minutes with a range not exceeding 27 minutes. I congratulate Dr Hilgenberg and associates on their fine work.
HILGENBERG ET AL BLUNT AORTIC TRAUMA
239
Dr Hilgenberg, what is your current preference of technique with regard to the protection of the spinal cord? DR HILGENBERG Controversies and unsolved issues continue to involve the management of thoracic aortic injuries, and I am pleased to see this amount of discussion. I will attempt to answer some of the questions. Dr Robicsek, there was a general trend over this 14year period toward the use of shunts more frequently in the earlier years and less so in the later years. Your analysis of that point is correct. However, there were some shunted patients even in the most recent years. The shunts were placed based on the preference of the surgeon and the stability of the patient’s condition. I cannot explain why the cross-clamp times were longer in the shunted patients. Because it is unusual for the surgeon to determine exactly how complex the proximal aortic injury will be until after the clamps are on and the aorta is open, I do not believe that there was a bias toward insertion of shunts in the more complex patients. We are impressed with some of the data from Stanley Crawford’s recently published experience. The cases were elective descending aneurysms and not aortic trauma cases, but he looked carefully at distal perfusion with the Bio-Medicus pump and somatosensory evoked potential monitoring, and he found that those methods did not appear to confer increased protection against paraplegia. The fact remains that there are variables that we cannot necessarily predict with respect to the spinal cord and its function and blood supply. Dr Tumey, we are aware of your large experience with aortic injuries, and I was pleased to see that our data and results correspond very closely with yours. Regarding early pharmacologic intervention, we have attempted to do this judiciously, and we have not experienced major swings in blood pressure. We were also concerned about spinal cord perfusion and nitroprusside administration during cross-clamping, and we looked at this carefully with multivariate analysis. It was not a significant factor associated with paraplegia. Dr Antunes, we rarely see injuries that are less than circumferential. Perhaps there is some difference in our patient populations. There were only 2 patients out of 44 in this series in whom we thought the aortic injury was so localized that it could be easily repaired by direct suture. We usually insert a graft because the injuries are so frequently circumferential with separation of the aortic ends.
We managed 51 patients with thoracic aortic injuries caused by blunt trauma between 1977 and 1990. Fortynine injuries were located in the upper descending aorta and one each in the ascending aorta and aortic arch. Three patients arrived moribund and underwent thoracotomy for resuscitation, and all died. The diagnosis was confirmed by aortography in 48. One patient died of aortic rupture, 1 died of hypoxemia, and 1 refused operation and died. Forty-four patients had aortic repair, 42 with graft insertion. Gott shunts were placed in 23 with 3 cases of paraplegia (13%). Simple cross-clamping was used in 19 with 1 case of paraplegia (5.2%). We found statistically significant differences between the cross-clamp times of patients without paraplegia compared with those in whom paraplegia developed in both the shunt and no-shunt groups. Logistic regression anal-
ysis showed that the only factor significantly associated with paraplegia was cross-clamp time. There were two postoperative deaths (4.4%). Seven patients had medical therapy initially and aortic repair was delayed to allow other injuries to stabilize. Before aortic repair, 18 patients had intraarterial pressure monitoring and 34 received &blockers or antihypertensive drugs. We conclude that aortic repair with graft insertion is usually successful in nonmoribund patients, simple crossclamping is associated with a relatively low risk of paraplegia, the incidence of paraplegia is directly associated with the duration of cross-clamp time, and selected patients can be managed medically while awaiting aortic repair.
B
tially higher in patients treated with heparinization and bypass than in those who were shunted and received no heparin. Hemorrhage was the major cause of death in the bypass group. We also noted that aortic repair could be safely delayed in selected patients who were managed medically while other severe injuries were stabilized.
lunt traumatic disruption of the thoracic aorta remains a highly lethal injury. It appears that only preventive measures that may reduce the frequency of this injury could have any effect on the mortality in the prehospital phase [l]. Only the patients who reach the hospital with the false aneurysm intact and a blood pressure and cardiac rhythm present have a chance of survival. In these patients mortality and morbidity may arise from the multiple severe associated injuries, from delay in recognition of the aortic injury, from sudden rupture of the false aneurysm during angiography and transportation to the operating room, and from complications of aortic repair such as hemorrhage, paraplegia, and renal failure. Controversial issues involve diagnostic studies, timing of aortic operation, method of aortic repair, and spinal cord protection during aortic clamping [2-71. We reviewed our experience with all patients treated at the Massachusetts General Hospital with acute blunt traumatic disruption of the thoracic aorta from January 1977 to June 1990. This time period was selected because 1977 marked the end of the use of systemic heparinization and partial cardiopulmonary bypass for descending aorta repairs. Experience from our institution between 1970 and 1980 has been previously reported [7]. In that study we found that the mortality rate for aortic repair was substanPresented at the Twenty-seventh Annual Meeting of The Society of Thoracic Surgeons, San Francisco, CA, Feb 18-20, 1991. Address reprint requests to Dr Hilgenberg, Department of Thoracic Surgery, Massachusetts General Hospital, 32 Fruit St, Boston, MA 02114.
0 1992 by The Society of Thoracic Surgeons
(Ann Thoruc Surg 1992;53:233-9)
Material and Methods The medical records of all patients with acute blunt traumatic disruption of the thoracic aorta were reviewed. Particular attention was directed to the emergency ward record, the operative and anesthetic reports, the postoperative course, and autopsy reports. Statistical analysis included comparison of paraplegia rates by the Fisher exact test, cross-clamp times by unpaired t tests and one way analysis of variance, and factors associated with paraplegia by multivariate logistic regression analysis.
Clinical Material Between 1977 and 1990 we treated 51 patients with acute blunt injuries to the thoracic aorta. There were 43 male and 8 female patients whose average age was 35 years, with an age range of 16 to 70 years. Circumstances of the accidents were as follows: drivers of motor vehicles, 33; passengers in motor vehicles, 8; pedestrians struck by autos, 4; motorcycle riders, 4; and victims of falls, 2. Information regarding seat belt use and ejection of the victim from the vehicle was not available in sufficient numbers to draw any conclusions about these factors. 0003-4975/92/$5.00
234
Ann Thorac Surg 1992;53:23>9
HILGENBERG ET AL BLUNT AORTIC TRAUMA
Four patients came directly to our hospital, and 47 patients were referred from other institutions. Initial arterial blood pressures were unobtainable in 3 patients and less than 100 mm Hg systolic in an additional 7 patients. Initial systolic arterial pressures were greater than 140 mm Hg in 22 patients. Decreased femoral pulses were noted in 8 patients. A visible or palpable chest wall injury was apparent in only 20 patients (39%). Four patients had a flail chest. Seven patients had an audible murmur, and in 2 of these it was probably due to preexisting heart disease. Reports of chest radiographs were available for review in 48 patients, and all of them showed superior mediastinal widening, although in 3 patients the widening was not present on the initial films. Other common findings were obscuring of the aortic knob contour in 10, an extrapleural apical cap in 7, lateral shift of mediastinal structures in 13, lung contusion in 20, rib fractures in 20, pleural effusions in 20, and pneumothorax in 10. Nine patients had computed tomographic scans of the chest, and in only 4 were findings suggestive of aortic injury. Three were read as probable aortic injury, one as descending aortic dissection, and the other five showed only mediastinal hematomas and pleural effusions. Associated injuries were common and severe, and included head injuries, intrathoracic organ damage, abdominal injuries, fractures, and peripheral vascular and nerve injuries. Details of the associated injuries are summarized in Table 1. In 48 patients the diagnosis of aortic injury was confirmed by aortography, and in 3 moribund patients the aortic injury was found during emergency resuscitative thoracotomy. Forty-nine injuries were located in the upper descending aorta, one in the ascending aorta, and one in the aortic arch with involvement of the innominate and left common carotid arteries. Our recent policy is to insert radial artery catheters for continuous pressure monitoring in patients strongly suspected of aortic injury, and to treat hypertension if present. Eighteen patients had arterial lines inserted in the emergency ward. Throughout the period of this study, we have commonly administered intravenous propranolol to patients who are not hypotensive and nitroprusside infusions to those who are hypertensive during their evaluation. A total of 34 patients received pblocking drugs or vasodilators or both before their aortic repairs.
Surgical Procedures Three patients arrived in our emergency ward without pulse or blood pressure and underwent resuscitative thoracotomy, and all 3 died. One patient arrested during transport from the angiography suite to the operating room and had a thoracotomy and aortic clamping, but he could not be resuscitated. Death was due to rupture of the false aneurysm. One patient died of hypoxemia and did not have an aortic repair. One patient refused operation and died (his auto accident was probably a suicide). One patient received long-term medical therapy because of paraplegia resulting from a thoracic spine fracture with direct spinal cord damage. The remaining 44 patients had
Table 1. Associated Znjuries Injury Head injuries Concussion Cerebral hematomas requiring operation Thoracic injuries Cardiac contusion Lung contusion Right atrial rupture Tracheal laceration Left diaphragm rupture Abdominal injuries Liver laceration Ruptured spleen Retroperitoneal hematoma Fractures Ribs Sternum Clavicle Facial bones Spine Upper extremity Lower extremity Pelvis Vascular injuries Femoral artery Peripheral nerve palsies
No. of Patients
15 5 4 20 1 1 1 6
5 6 20 1 4 8 5 11 25 13 2 8
aortic repairs: 42 with descending aortic injuries and 1 each with injury to the ascending aorta and the aortic arch. Thirty-three patients underwent immediate aortic repair. In 6 patients aortic repair was delayed 4 days to 4 months to allow stabilization of other severe injuries while medical treatment was administered to reduce the risk of aortic rupture. Five of the 6 patients had severe neurological injuries that were responsible for the delays, and 1 patient had a severe pulmonary contusion. In 4 patients the diagnosis of aortic injury was not made until 2 to 5 days after the accident, and aortic repair was accomplished immediately after their arrival in our hospital. One patient had a delayed diagnosis of aortic injury and his repair was delayed while electrocardiographic changes were evaluated. Therefore, aortic repair was performed immediately after confirmation of the diagnosis of aortic injury in 84% of the patients. The ascending aortic injury was repaired through a median sternotomy on cardiopulmonary bypass with graft insertion. The aortic arch injury was repaired through a median sternotomy without bypass. A graft was placed from the ascending aorta to the innominate artery; a side arm graft was then placed to the left common carotid artery using electroencephalographic monitoring, and the aortic arch disruption was closed directly (Fig 1). The descending aortic injuries were all exposed through
HILGENBERG ET AL BLUNT AORTIC TRAUMA
Ann Thorac surg 1992;53:233-9
235
Fig 1. (A)Initial step in the repair of a false aneurysm of the aortic arch involving the innominate artery origin. An 8-mm Dacron graft is anastomosed to the ascending aorta. (B) Using electroencephalographic monitoring, the proximal and distal ends of the innominate artery are clamped and the distal anastomosis of the graft to the innominate artery is constructed. Inspection of the aortic arch showed extension of the intimal injury beyond the clamp into the origin of the left common carotid artery. Therefore a second Dacron graft, 6 mm in diameter, is anastomosed to the innominate nrtery graft as a side-arm before release of the clamps. (C) With the innominate artery perfused, the left common carotid artery is clamped and a larger clamp applied to the aortic arch. The distal anastomosis of the side-arm graft to the left common artery is completed. ( D ) To complete the repair, the aortic arch is closed directly, obliterating the intimal injury at the origins of the innominate and left common carotid arteries.
a left posterolateral thoracotomy with one-lung ventilation through a double-lumen endotracheal tube. The proximal clamp was applied between the left common carotid and left subclavian arteries in all cases. Simple cross-clamping was used in 19 patients and Gott shunts in 23. Shunt insertion sites are shown in Table 2. Dacron grafts were placed in all except 2 patients whose injuries were small and were sutured directly. Operation for associated injuries was frequently re-
Table 2. Gott Shunt lnsertion Sites in 23 Patients Site Proximal Ascending aorta Aortic arch Left ventricular apex Distal Descending aorta Left femoral artery
No. of Patients 15 4
quired. Five patients underwent intracranial procedures before aortic repair. Concomitant intrathoracic procedures included suture of tracheal laceration, closure of right atrial rupture, and repair of ruptured diaphragm. Fifteen patients had laparotomies, five of which were performed before aortic repair. Open fracture reduction was performed in 28 patients, three of these before aortic repair. One patient required reexploration for bleeding. Vascular repair was necessary in 2 patients after shunt removal; 1 had a vein patch graft repair of a femoral artery, and 1 needed multiple repair sutures in the left ventricular apex. Five patients had tracheostomies. Persistent postoperative hypertension was extremely common. A total of 39 patients received antihypertensive agents after hospital discharge. Only 3 patients discharged after aortic repair did not require drug therapy for hypertension.
4
17 6
Results The overall mortality was 8 of 51 patients (15.7%),which includes those who died before an attempt at aortic repair.
236
HILGENBERG ET AL BLUNT AORTIC TRAUMA
There were two postoperative deaths in the aortic repair group of 44 patients (4.5% mortality). The deaths, one each in the shunt and no-shunt groups, were due to respiratory failure, sepsis, and renal failure. The average age of these 2 patients was 68 years, and both were female. Paraplegia developed in 4 patients out of a total of 42 at risk (9.5%). The paraplegia rate was 1 of 19 in the no-shunt group (5.2%)and 3 of 23 in the Gott shunt group (13.0%).The difference in paraplegia rates was not statistically significant (p = 0.6 using the Fisher exact test). Aortic cross-clamp times were: no shunt, no paraplegia, 32 minutes; no shunt, paraplegia, 50 minutes; Gott shunt, no paraplegia, 45 minutes; Gott shunt, paraplegia, 80 minutes. One-way analysis of variance established differences among these groups. Unpaired t tests found significant differences between the cross-clamp times in the patients without paraplegia compared with those in whom paraplegia developed in both the shunt and noshunt groups (p = 0.02 in the no-shunt group and p = 0.0005 in the Gott shunt group). Using multivariate logistic regression analysis, we evaluated 19 factors for their association with the occurrence of paraplegia. Variables included presence of a shunt, cross-clamp time, hypotension, nitroprusside administration during aortic clamping, and administration of drugs such as mannitol, steroids, and magnesium sulfate before aortic clamping, as well as additional factors. The only factor that had a significant association with paraplegia was cross-clamp time. The presence or absence of a shunt and the administration of vasodilators or other drugs had no significant effect on paraplegia development. However, because of the relatively small number of paraplegic patients, there is a possibility of failure of association of other factors. A logistic model of probability relating the risk of paraplegia to cross-clamp time is shown in Figure 2. The calculated probability of paraplegia is less than 1%with a clamp time of 30 minutes and less than 5% with a clamp time of 45 minutes, but the curve rises steeply at 50 minutes. Although the average cross-clamp time in the Gott shunt patients was significantly longer than in the no-shunt patients, multivariate analysis indicated that a shunt did not offer protection from paraplegia or extend the “safe” cross-clamp time.
Comment If survival of patients with blunt traumatic aortic disruption is to be improved, the diagnosis must be made promptly, methods to reduce the risk of rupture instituted, and urgent surgical repair carried out in most patients. If the aortic pseudoaneurysm ruptures, even if the patient is in the hospital, it is too late to salvage the patient. In this series, none of the patients who had aortic rupture with unobtainable blood pressure survived despite attempts at aortic clamping and resuscitation. Other authors have also noted the uniformly fatal outcome of patients with massive bleeding from aortic rupture [8, 91. Perhaps a method of very rapid infusion of large volumes
Ann Thorac Surg 1992;53:23>9
1
/--
0.0 0.8 0.7 P r
0.6 b a b 0.5 i I i 0.4 t 0
Y
0.3 0.2 0.1
10
20
30
so 60 TO Clamp Time ( m i d
40
a0
90
100
Fig 2 . Logistic model of the probability of the occurrence of paraplegia compared with aortic cross-clamp time. The dotted lines are 70% confidence limits.
of blood could be of benefit in this situation [9]. However, with our current technology, the patient’s best chance for survival is to arrive in the operating room with the pseudoaneurysm still intact. We have used early continuous arterial pressure monitoring and administration of drugs to decrease the forces applied to the aortic wall in patients suspected of having aortic disruption based on the history of the accident and mediastinal widening on chest radiographs. Thoracic computed tomographic scanning is omitted because it does not confirm or exclude aortic injury reliably. The patient is taken promptly to the angiography suite for an aortogram and is attended by surgical residents and special nurses who continue aggressive monitoring, volume administration, and blood pressure control. This early treatment method appears to be generally successful as only 1patient in this series died of aortic rupture after aortography during transportation to the operating room, and he did not have a radial artery catheter and the vigorous treatment we now advocate. Timing of aortic repair requires careful consideration and individualization. In most patients prompt aortic repair is indicated, as was done in 84% of the patients in this series. However, in 7 patients aortic repair was delayed 4 days to 4 months after aortography to allow stabilization of other injuries. The most frequent reason for delaying aortic operation was a severe head injury with either the need for intracranial operation or an uncertain prognosis on initial evaluation. Head injuries were responsible for the delay in 5 of the 7 patients. One
Ann Thorac Surg 1992;5323>9
patient had severe hypoxemia from pulmonary contusion and other injuries and could not have tolerated one-lung ventilation until his lung injury improved. One other delayed patient had electrocardiographic changes that were evaluated before repair. Although we did not encounter it in this series, another reason for delay could be presence of sepsis in a patient in whom the diagnosis of aortic injury is made late. The results from this series and our previously reported experience [7] support the safety of medical therapy in these highly selected patients, as none of them experienced aortic rupture while waiting for repair. Prioritizing surgical procedures also requires careful consideration. Intracranial operation may have priority in the hemodynamically stable patient. In the patient in unstable condition with evidence of exsanguinating abdominal hemorrhage, laparotomy is indicated before aortic repair. If the airway is injured, this requires immediate attention. In this series, procedures prior to aortic repair were craniotomy or insertion of intracranial pressure monitors in 5 patients, laparotomy in 5 patients, and repair of tracheal laceration in 1 patient. The other 10 laparotomies and nearly all of the orthopedic procedures were performed after aortic repair. The only open fracture procedures done before aortic repair were in the delayed patients with head injuries. Surgical repair of the aortic injury consisted of Dacron graft insertion in 95% of the cases because the aortic intima was completely or nearly completely transected and the ends separated. The intimal damage was substantially less than circumferential in only 2 cases, and these were both sutured directly. In most instances, sutureless intraluminal grafts were considered unsatisfactory because of the small diameter of the aorta in most of these young patients and lack of space distal to the left subclavian artery for seating of the proximal cuff. The most controversial and unsolved aspects of descending aortic repair are the role of distal perfusion and the causes and incidence of paraplegia. During the period of this study, we used two techniques and the choice depended on the preference of the surgeon and the degree of stability of the patient. Forty-five percent of the patients had simple cross-clamping without distal perfusion, and in 55% a Gott shunt was placed with most patients receiving ascending to descending aortic shunting. No consistent effort was made to monitor distal aortic pressure or flow in shunted patients. Some patients received a variety of pharmacologic agents to possibly aid in spinal cord protection, and these included steroids, pentothal, magnesium sulfate, and mannitol. Nitroprusside was used for control of proximal hypertension if needed. Paraplegia occurred in 9.5% of the patients at risk, and there was no statistical difference between the incidence in the shunt and no-shunt groups (13% and 5.2%, respectively). Of all the variables evaluated by logistic regression analysis, the only factor significantly associated with paraplegia was cross-clamp time, and the association was a very strong one. Factors that had no association with paraplegia included admission systolic blood pressure less than 100 mm Hg, intraoperative
HILGENBERG ET AL BLUNT AORTIC TRAUMA
237
hypotension, administration of nitroprusside or other vasodilators during aortic clamping, administration of drugs before aortic clamping that may protect the spinal cord, and presence or absence of a shunt. A most important variable that cannot be assessed preoperatively or by statisticalanalysis is the individual variation of spinal cord blood supply. The length of the cross-clamp time is mainly a reflection of the complexity of the injury and the degree of difficulty of the repair; to a lesser extent it reflects the speed of the surgeon [101. Problems leading to longer cross-clamp times in this series were difficulty with the proximal aorta with extension of the injury into the aortic arch close to the cross-clamp, and marked obesity of a patient with a very small aorta. Cross-clamp times in our patients in whom paraplegia developed were 70, 80, and 90 minutes in the shunted patients and 50 minutes in a no-shunt patient. Although the average cross-clamp time in the Gott shunt group was significantly longer than that of the no-shunt patients, logistic regression analysis indicated that the presence of a shunt had no additional protective effect for any given cross-clamp time. This finding is similar to that reported from the Maryland Shock Trauma Center [9] and different from that reported by the Alabama group [ll]. We strongly endorse the efforts of investigators trying to improve the results of descending thoracic aortic operations, particularly with respect to reducing the incidence of paraplegia. It seems clear from our data and those of others that in the acute trauma situation, a Gott shunt without distal pressure or flow monitoring is not consistently protective of the spinal cord. We acknowledge that excellent results have been reported in predominately elective descending thoracic aortic aneurysm repairs using a Gott shunt with flow monitoring [12]. Partial cardiopulmonary bypass has its proponents, with evidence that spinal cord perfusion may be well maintained [13], but our past experience with systemic heparinization in acute trauma patients has been unsatisfactory because of hemorrhagic complications, making this an unacceptable method for use in this setting. Good results with left atrial-femoral bypass using a centrifugal pump without heparin have been reported in relatively small groups of patients [14,15]. This approach appears promising in that flows similar to those of partial cardiopulmonary bypass can be obtained and heparin is avoided. Other adjuncts such as sensory evoked potential monitoring [16], spinal fluid drainage, and pharmacologic agents are probably of minimal help, especially in the acute trauma situation. The most desirable technique for these emergency operations in unstable patients with multiple injuries is a quick and uncomplicated method of aortic repair with a low mortality rate and as low a paraplegia rate as possible. In our hands, simple cross-clamping with graft insertion meets these criteria at the present time. We hope that with additional research, improvements can be made. However, because of variability in spinal cord blood supply and variability in extent and complexity of the aortic injury, the paraplegia rate will probably never be zero.
238
HILGENBERG ET AL BLUNT AORTIC TRAUMA
We gratefully acknowledge the assistance of John Newel1 for the statistical analysis and Ruthellen McKinnon for preparation of the manuscript.
References 1. Eddy AC, Rusch VW, Fligner CL, Reay DT, Rice CL. The epidemiology of traumatic rupture of the thoracic aorta in children: a 13-year review. J Trauma 1990;30989-92. 2. Mattox KL, Holzman M, Pickard LR, Beall AC, DeBakey ME. Clamphepair: a safe technique for treatment of blunt injury to the descending thoracic aorta. Ann Thorac Surg 1985;40: 456-63. 3. Kirsh MM, Behrendt DM, Orringer MB, et al. The treatment of acute traumatic rupture of the aorta: a ten year experience. Ann Surg 1976;184:308-16. 4. Appelbaum A, Karp RB, Kirklin JW. Surgical treatment for closed thoracic aortic injuries. J Thorac Cardiovasc Surg 1976;71:458-60. 5. Orringer MB, Kirsh MM. Primary repair of acute aortic disruption. Ann Thorac Surg 1983;35:672-5. 6. Borman KR, Aurbakken CM, Weigelt JA. Treatment priorities in combined abdominal and aortic trauma. Am J Surg 1982;M728-32. 7. Akins CW, Buckley MJ, Daggett WM, McIlduff JB, Austen WG. Acute traumatic disruption of the thoracic aorta: a ten-year experience. Ann Thorac Surg 1981;31:305-9. 8. Clark DE, Zeiger MA, Wallace KL, Packard AB, Nowicki ER. Blunt aortic trauma: signs of high risk. J Trauma 1990;30: 701-5.
Ann Thorac Surg 1992;53:233-9
9. Cowley RA, Tumey SZ, Hankins JR, Rodriquez A, Attar S, Shankar BS. Rupture of thoracic aorta caused by blunt trauma. A fifteen-year experience. J Thorac Cardiovasc Surg 1990;100652-61. 10. Mattox KL. Fact and fiction about management of aortic transection. Ann Thorac Surg 1989;481-2. 11. Katz NM, Blackstone EH, Kirklin JW, Karp RB. Incremental risk factors for spinal cord injury following operation for acute traumatic aortic transection. J Thorac Cardiovasc Surg 1981;81:669-74. 12. Verdant A. 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: 147-54. 13. Laschinger JC, Isumoto H, Kouchoukos NT. Evolving concepts in prevention of spinal cord injury during operations on the descending thoracic and thoracoabdominal aorta. Ann Thorac Surg 1987;44:667-74. 14. Olivier HF Jr, Maher TD, Liebler GA, Park SB, Burkholder JA, Magovem GJ. Use of the Biomedicus centrifugal pump in traumatic tears of the thoracic aorta. Ann Thorac Surg 1984;38:58&91. 15. Hess PJ, Howe HR Jr, Robicsek F, et al. Traumatic tears of the thoracic aorta: improved results using the Bio-Medicus pump. Ann Thorac Surg 1989;486-9. 16. Crawford ES, Mizrahi EM, Hess KR, CoseIli JS, Safi HJ, Pate1 VM. The impact of distal aortic perfusion and somatosensory evoked potential monitoring on prevention of paraplegia after aortic aneurysm operation. J Thorac Cardiovasc Surg 1988;95:357-67.
DISCUSSION DR FRANCIS ROBICSEK (Charlotte, NC): I thank The Society for the opportunity to discuss this thought-provoking report. As far as the methodology is concerned, I have a few questions, all related to the lack of randomization. This study encompasses a period of 14 years. Is there a bias in comparing earlier Gott-shunted cases with more recent nonshunted cases? Also, the shunted group had a much higher paraplegia rate but also a much longer cross-clamp time. Is it possible that the more severely injured patients were shunted and the less severely injured patients were not shunted? I would also like to call attention to the fact that we are more or less fighting a strawman opponent, because a question in the mind of most of us is not to shunt with a Gott shunt or just simply cross-clamp and sew, but to use other bypass methods such as Bio-Medicus or Bio-Medicus-type bypass systems. In the material we have collected with my partner, Phillip Hess, we have 36 patients who survived reconstructive operation for traumatic aortic rupture. Femoral vein to femoral artery oxygenated bypass was used in 5. None of these patients suffered paraplegia, but we had two late deaths. Sixteen patients were operated on using a simple cross-clamp without shunt. Of the 16, 4 had paraplegia, 4 had renal shutdown, and 5 died late. Recently we have operated on a series of 15 patients using nonanticoagulated left atrium to femoral artery bypass, with cross-clamp times ranging from 21 minutes to 2 hours. There was neither death nor paraplegia in this system. We were also disturbed that in the simple clamp, no-shunt groups, we had 2 patients with
less than 32-minute cross-clamp times who ended up with paraplegia. We certainly are not here to dispute the virtues of fast and expeditious operation. Reconstruction of the ruptured aorta in a very short time is possible in some cases by some surgeons but not in all situations for all surgeons. And even if it is possible to do it in such a short time, in our experience, contrary to Hilgenberg and associates, we believe that whenever the clinical situation allows, nonanticoagulated left atrium to femoral artery bypass may provide some additional protection against the most feared complication of paraplegia. I learned a lot about this complication by serving on your Committee on Medico-Legal Affairs, and I found that about one third of major malpractice suits involving cardiovascular thoracic operations were connected to nonshunted paraplegias stemming from descending aortic reconstruction. I think the Bio-Medicus shunt is a very valuable protection not only for the patient's lower spine but also for that of the surgeons. DR STEPHEN Z. TURNEY (Baltimore, MD): I compliment Dr Hilgenberg and associates on their comprehensive report and congratulate them on their results. We recently reported our own updated experience of 114 patients from Baltimore at the meeting of the Association for Thoracic Surgery and came to essentially the same conclusions, that morbidity and mortality are high and are complex problems. Aortoaortic shunt does not necessarily protect patients from paraplegia. It is our practice not to routinely use shunt.
Ann Thorac Surg 1992;53:233-9
A natural history survival curve for blunt thoracic aortic rupture shows that most deaths occur within 12 to 24 hours of injury. Very early transport of the accident victim, as in our trauma systems, pushes the patient population one is faced with to the steep, high mortality rate portion of the curve where patients are exsanguinating from their aortic and other injuries. The other risk factors we identified included hypotension on admission, a large presenting hemothorax, the surgeon’s credentials, and the presence of other serious injury. My question is, does early pharmacologic intervention present problems of its own in management, such as the evaluation of hidden bleeding, large blood pressure swings as anesthesiologists struggle to regulate blood volume and pressure, or the steal of blood from the spinal cord during cross-clamping?We scrupulously snare intercostal arteries to avoid back-bleeding and cord steal during repair, but we remain concerned that pharmacologic steal may be taking place into a dilated, low-pressure, systemic vascular bed if the patient is receiving vasodilators. DR MANUEL DE JESUS ANTUNES (Coimbra, Portugal): In their well-presented large series, Dr Hilgenberg and associates have clearly demonstrated that the length of the aortic crossclamping is the most important risk factor for paraplegia, irrespective of the technique used. That much had also surfaced from a review of the literature done from our previous institution by Dr Svenson some years ago. Hence, all technical details that may help reducing the cross-clamp time will be of major importance. In an article published in The Annals 3 years ago, I suggested that using straight aortic cross-clamping without shunts or cardiopulmonary bypass and incision of the descending aorta, including the covering hematoma in the area of rupture, would expose the tear, usually located on the opposite wall when the tear was partial. The tear could then be repaired, from within the lumen of the aorta, more clearly and easily with a continuous suture without further dissection. When there was complete separation of the aortic edges, the repair could also be facilitated by working from inside the lumen, either with direct approximation or by interposition of a graft. Finally, the latter could also be made easier by incising it longitudinally and transforming it into a circular patch. Using these techniques, we have now operated on 17 consecutive patients without paraplegia and with only 1 case of transient paraparesis. I believe that this is a consequence of a short mean aortic cross-clamp time of 22 minutes with a range not exceeding 27 minutes. I congratulate Dr Hilgenberg and associates on their fine work.
HILGENBERG ET AL BLUNT AORTIC TRAUMA
239
Dr Hilgenberg, what is your current preference of technique with regard to the protection of the spinal cord? DR HILGENBERG Controversies and unsolved issues continue to involve the management of thoracic aortic injuries, and I am pleased to see this amount of discussion. I will attempt to answer some of the questions. Dr Robicsek, there was a general trend over this 14year period toward the use of shunts more frequently in the earlier years and less so in the later years. Your analysis of that point is correct. However, there were some shunted patients even in the most recent years. The shunts were placed based on the preference of the surgeon and the stability of the patient’s condition. I cannot explain why the cross-clamp times were longer in the shunted patients. Because it is unusual for the surgeon to determine exactly how complex the proximal aortic injury will be until after the clamps are on and the aorta is open, I do not believe that there was a bias toward insertion of shunts in the more complex patients. We are impressed with some of the data from Stanley Crawford’s recently published experience. The cases were elective descending aneurysms and not aortic trauma cases, but he looked carefully at distal perfusion with the Bio-Medicus pump and somatosensory evoked potential monitoring, and he found that those methods did not appear to confer increased protection against paraplegia. The fact remains that there are variables that we cannot necessarily predict with respect to the spinal cord and its function and blood supply. Dr Tumey, we are aware of your large experience with aortic injuries, and I was pleased to see that our data and results correspond very closely with yours. Regarding early pharmacologic intervention, we have attempted to do this judiciously, and we have not experienced major swings in blood pressure. We were also concerned about spinal cord perfusion and nitroprusside administration during cross-clamping, and we looked at this carefully with multivariate analysis. It was not a significant factor associated with paraplegia. Dr Antunes, we rarely see injuries that are less than circumferential. Perhaps there is some difference in our patient populations. There were only 2 patients out of 44 in this series in whom we thought the aortic injury was so localized that it could be easily repaired by direct suture. We usually insert a graft because the injuries are so frequently circumferential with separation of the aortic ends.