Traumatic Rupture of the Aorta

Abstract

Objective

To determine whether delay of the repair of the ruptured thoracic aorta in patients with other major injuries is safe and has a potential positive impact on survival.

Summary Background Data

The accepted treatment for acute traumatic rupture of the thoracic aorta has been repair of the injury as soon as possible. This form of management, however, has been accompanied by a death rate of 0% to 54% mortality, often related to the presence of other injuries.

Methods

The records of 30 consecutive patients with rupture of the thoracic aorta from blunt trauma treated from 1995 to 2001 were retrospectively reviewed. Two of them died shortly after admission and were excluded from further consideration. The remaining 28 patients were divided according to the time of the repair of the rupture into two groups. Group 1 patients underwent repair of the rupture immediately after the diagnosis was made. Group 2 patients, who had associated injuries that were likely to increase the risk of surgical death, had either repair more than 48 hours after injury (subgroup 2A) or had no repair (subgroup 2B). The patients in group 2 had their mean arterial pressure maintained at less than 70 mm Hg with medication to eliminate shear stress on the aortic tear while being observed.

Results

Twenty-eight patients (22 men, 6 women) with an average age of 36 years (range 19–76) were treated. Twenty-five had rupture of the descending thoracic aorta and three had rupture in the ascending thoracic aorta. Group 1 comprised 14 patients, 5 of whom died during surgery or in the early postoperative period. Group 2 comprised 14 patients, 9 in subgroup 2A and 5 in subgroup 2B. Two patients in subgroup 2A and three patients in subgroup 2B died of associated injuries or illnesses. Rupture of the traumatic pseudoaneurysm of the thoracic aorta did not occur in any of the patients in group 2.

Conclusions

Delayed repair of acute traumatic aortic rupture is safe under appropriate treatment and should be considered in selected patients.

The traditional management of acute traumatic rupture of the thoracic aorta has been repair of the injury as soon as possible. The recommendation for this treatment is primarily based on the premise that as many as 90% of patients will die within the first 24 hours after their injury. ^1,2 Immediate repair of the transected aorta has been associated with a surgical death rate of 0% to 54.2%. ³ This high death rate is often due to the other associated injuries that may be present in the major trauma patient. Delayed repair of the acute aortic tear, allowing the patient to recover from other major injuries, has been reported in the literature with increasing frequency. ^4–9 This review was conducted to determine the safety of delayed repair of acute rupture of the thoracic aorta, and this strategy’s impact on the patient’s outcome.

METHODS

The records of all patients with acute rupture of the thoracic aorta who were treated from 1995 to 2001 were reviewed retrospectively. All patients were brought to the State of Georgia Level I Trauma Center at Grady Memorial Hospital (Atlanta, GA) directly from the scene of the accident. Data on the site of aortic rupture, the associated injuries, the time between the diagnosis and repair of the aortic tear, the type of operation performed, and the patient’s outcome were collected for this review. During this 6-year period, we treated 30 patients, including 2 who died shortly after admission during exploratory thoracotomy for massive intrathoracic bleeding. The diagnosis of blunt rupture of the thoracic aorta was made at thoracotomy. Because these two patients had complete tear of the aortic wall when they arrived in the emergency department and exsanguinated quickly, they are excluded from further consideration.

The remaining 28 patients with aortographic confirmation of a blunt rupture of the thoracic aorta were separated into two groups as follows. Patients in group 1 underwent repair of the ruptured thoracic aorta immediately after the diagnosis was made. Patients in group 2 were at risk to have an increased risk of surgical death due to associated injuries and underwent either repair of the aortic tear more than 48 hours after admission when their condition was optimized (subgroup 2A) or had no repair at all (subgroup 2B). While they received treatment for other injuries, their mean arterial pressure was maintained at less than 70 mmHg with beta blockers and vasodilators (sodium nitroprusside, calcium channel blockers, and nitrates), and the mediastinal silhouette was followed with serial chest roentgenograms.

The diagnosis of aortic transection was confirmed by aortography in all 28 patients, which demonstrated a false aneurysm.

Surgical repairs in patients with rupture of the descending thoracic aorta were performed under left atrial to left femoral artery bypass using a centrifugal pump and the administration of heparin (3.0 mg/kg body weight) to achieve active clotting time (ACT) levels of 350 to 400. The repairs in patients with rupture of the ascending thoracic aorta were performed under conventional total cardiopulmonary bypass. Graft interposition was used in 13 patients; primary repair was used in 8 patients.

RESULTS

Patient Demographics

Twenty-eight patients at an average age of 36 years (range 19–76) were treated during the 6-year period (22 men, 6 women). Twenty-five had rupture of the descending thoracic aorta just distal to the left subclavian artery; three had rupture of the ascending thoracic aorta. Two of the three patients with rupture of the ascending thoracic aorta have been previously described. ¹⁰ Of interest, group 1 patients had an average of 1.3 other organ injuries (Table 1), whereas those in subgroups 2A and 2B had an average of 3.7 other injuries (Table 2) and 4.4 other injuries (Table 3), respectively.

Table 1. PATIENTS WHO UNDERWENT IMMEDIATE REPAIR OF AORTIC RUPTURE

MVC, motor vehicle crash.

Table 2. PATIENTS WHO UNDERWENT DELAYED REPAIR OF AORTIC RUPTURE

MVC, motor vehicle crash.

Table 3. PATIENTS WHO DID NOT UNDERGO REPAIR OF AORTIC RUPTURE

MVC, motor vehicle crash.

Patient Outcomes

Group 1 comprised 14 patients who underwent repair of the aortic rupture immediately after the diagnosis was made. Nine patients survived and were eventually discharged. Six of these nine patients, however, suffered life-threatening complications after surgery (see Table 1). Four of the 14 patients in group 1 died in the operating room. Two elderly patients (ages 68 and 76 years) died of cardiac arrest that occurred shortly after the placement of the aortic cross-clamp. One patient died of hemorrhage and another died of hypoperfusion and cardiac arrest, both due to technical problems. One other patient died within 10 days of surgery as a result of a pulmonary embolus. The surgical death rate for group 1 was 28.5%, the late death rate was 7%, and the overall death rate was 35.5%.

Group 2 comprised 14 patients, 9 in subgroup 2A (delayed aortic repairs) and 5 in subgroup 2B (no surgical repair). The nine patients in subgroup 2A had their repairs delayed 2 to 129 days (average 27) after admission (see Table 2). Five of these nine patients suffered no complications and were discharged in good condition. One patient developed paralysis of the left vocal cord secondary to an injury to the recurrent laryngeal nerve, and another developed a chylothorax. Two patients died in the late postoperative period, one of acute respiratory failure and complications from an abdominal operation and the second of sepsis. The surgical death rate for this subgroup was 0% and the late and overall death rates were 22.2%. Three of the subgroup 2B patients never underwent aortic repair and subsequently died. Two of them died of associated injuries 12 and 13 days after admission. The third patient died 232 days after injury from complications of Guillain-Barré syndrome, which manifested in the early postinjury period. The other two patients in subgroup 2B were transferred elsewhere 36 and 72 days after admission. In no group 2 patient did alteration of the mediastinal silhouette occur during the medical treatment regimen, and no patient died of sudden exsanguination as a result of rupture of the traumatic pseudoaneurysm of the thoracic aorta. Also, postoperative paraplegia did not occur in the series.

DISCUSSION

Blunt rupture of the thoracic aorta is a devastating injury, with the vast majority of patients dying at the time of injury. ^1,2 In order of frequency, rupture occurs at the aortic isthmus, the ascending aorta, the aortic arch, the distal descending aorta, and the abdominal aorta. Feczko et al. ¹¹ and Williams et al. ¹² reviewed autopsies of victims who suffered blunt trauma and found that 55% to 65% of injuries occurred at the aortic isthmus and 10% to 14% occurred in the ascending aorta or aortic arch. Twenty-five (89.3%) of our patients had rupture at the aortic isthmus, and three (10.7%) had tears in the ascending aorta.

The force from rapid deceleration necessary to tear the aorta often leads to other organ injuries. Pate et al. ⁶ found that associated injuries were present in more than 90% of patients with aortic transection, and 24% of them required a major operation before aortic repair. In our series, 93% of the patients had one or more other organ injuries.

The aortic tear may be limited to the intima or may extend to both the intima and media or to all aortic layers. ^1,13 In the vast majority of patients who survive the initial traumatic impact, the tear involves both the intima and the media. ¹ In three subgroup 2A patients in whom histologic studies of the aorta were available, the wall of the traumatic false aneurysm consisted mainly of fibrous tissue.

Since Parmley et al.’s classic study in 1958, ¹ the extremely high death rate of acute blunt rupture of the thoracic aorta has led surgeons to repair the aortic tear as quickly as possible. This form of management, however, has resulted in high rates of death and complications. The overall death rate found by Von Oppell et al. ³ in a meta-analysis of articles reporting outcomes of 1,492 patients who reached the operating room hemodynamically stable was 21.3%. It ranged from 0% to 54.2%, and the majority of the deaths occurred in the postoperative period. The high death and complication rates are often due to associated injuries in other organs. ¹⁴ Two patients in group 1 died in the operating room after cross-clamp application. These two patients were the oldest in the study (68 and 76 years old). One of them had diffuse subepicardial ecchymosis over the right ventricle, indicating the presence of a myocardial contusion. Although no preoperative evaluation for coronary artery disease was performed in either of these two patients, they may have suffered acute ischemia from the rapid increase in afterload from the aortic cross-clamping, even though they were on left atrial to femoral artery bypass. A delay of the repair to investigate the status of their coronary arteries may have resulted in a better outcome.

During the early 1970s, Akins et al. ⁴ began to delay the repair of the blunt aortic tear in selected patients with other major injuries. Before the repair they were treated with antihypertensives, and rupture of the traumatic false aneurysm did not occur. Similarly, in 1995 Pate et al. ⁶ reported that rupture of the pseudoaneurysm in the thoracic aorta did not occur in 41 patients whose arterial pressure was maintained at less than 140 mm Hg, and who underwent repair of the aortic tear between 12 hours and 24 weeks. Based on these reports and on the recent experience reported here, our current indications for delaying the aortic repair include the following: trauma to the central nervous system, contaminated wounds, respiratory insufficiency from lung contusion or other causes, body surface burns, blunt cardiac injury (myocardial contusion), tears of solid organs that will undergo nonoperative management, retroperitoneal hematoma, age 50 years or older, and medical comorbidities.

Each patient in whom we delayed the repair of the thoracic aorta was admitted to the surgical intensive care unit, where the arterial blood pressure was closely monitored. The mean arterial pressure was maintained at less than 70 mm Hg with the administration of beta blockers and vasodilators. In patients in whom surgical repair of the blunt rupture of the thoracic aorta was delayed or never performed, rupture with death from exsanguination did not occur. Two of the subgroup 2A patients died 14 and 22 days after aortic repair. One died of complications after an abdominal operation, and the other died of sepsis and multiorgan failure. Three of the subgroup 2B patients with major associated injuries never underwent aortic repair and died. Two of them died on hospital days 12 and 23 as a result of the other injuries. The third patient also sustained life-threatening injuries, but survived and was eventually discharged; however, this third patient developed Guillain-Barré syndrome with complete quadriplegia in the early postinjury period. As a result, he was ventilator-dependent and died of pulmonary and other complications 232 days later. Despite the low number of other organ injuries (1.3) in group 1 patients undergoing immediate repair, their overall death rate remained high (35.5%). The death rate for patients who had delayed repair in subgroup 2A was 22.2%, even though these patients had a much higher average number (3.7) of other organ injuries.

In conclusion, our study supports previous reports that free rupture of a contained acute traumatic rupture of the thoracic aorta from blunt trauma is unlikely to occur under proper monitoring and control of blood pressure. Therefore, it appears safe to allow patients who have suffered major associated trauma to be stabilized, undergo other emergent operations if needed, and then have elective surgical repair of the aortic tear. Also, our study suggests that patients who are treated in this delayed manner have better outcomes, with fewer intraoperative and postoperative complications.

Discussion

Dr. James W. Pate (Memphis, TN): Elective delay in the repair of a ruptured aorta is not yet widely accepted, but I would like to strongly support and endorse the recommendations of Dr. Symbas.

Building on the observation of animals, Bill Wheat, of Florida, in 1965 introduced the use of drugs in controlling aortic wall stress in aortic dissections and showed that this was clinically useful. Adkins in Boston and we in Memphis took this concept and applied it to traumatic rupture of the aorta, treating them in a similar way with a regimen for decreasing aortic wall stress. In 1984 we had accumulated 12 patients with this form of delayed therapy. This convinced us that it was really a useful method. We then established a formal protocol of pharmacologic reduction in aortic wall stress and hemodynamic monitoring instituted before an aortogram or invasive test, and continued until the aorta was cross-clamped at surgery some optimal time in the future.

The critical component of the therapy is the beta blockers, not just the hypotensives. The beta blocker decreases the systolic ejection slope of the left ventricle, which decreases the tearing force on the aortic wall. If one gives vasodilators alone, the pulse pressure increases as the diastolic pressure disproportionately falls, and one may actually increase the probability of rupture because of these increased stresses in the aortic wall. Patients who have major intraabdominal or intracranial injuries, significant pulmonary lacerations, serious cardiac or pulmonary chronic diseases, or those who were just hemodynamically unstable, hypothermic, hypoxic, or fluid overloaded, were selected for elective repair anywhere from 2 to 29 months later (with a median of just over 8 days). This slide summarizes our results in the ensuing 17 years with this protocol. In type 1B patients, “1” refers to the fact that they are acute (in the first few days) and “B,” where the hematoma is contained within the mediastinum at presentation to the trauma center. Patients with ascending aortic tears or patients who have freely exsanguinated into the chest before reaching the emergency department obviously are not candidates for this form of therapy and are not included. As you see, 147 patients have been treated during that time. Two of them died preoperatively. Those are interesting. In one, the radiologist discontinued the IV drips before doing the aortogram and the hematoma ruptured during injection of the dye. The other, the anesthesiologist discontinued the beta blocker before carrying the patient to the operating room; in that patient the hematoma ruptured on induction of anesthesia before the chest was opened. So, only 2 patients out of 147 actually ruptured the mediastinal hematoma and exsanguinated and, in both, the protocol was violated. We repaired the aorta in 132 patients, 92 of them under emergency conditions—that is, shortly after diagnosis—and 40 patients were electively delayed for various periods of time. We followed one patient who was lost to follow-up at 49 months, without any change in his chest x-ray or rupture. Thirteen patients were not repaired. Six of these died, usually of intracranial injuries or multiple organ failure. So I would suggest that Dr. Symbas turn his conclusions around: my personal opinion is that an emergency repair for traumatic ruptured aorta is done only on good-risk patients without other major physiologic derangements.

I would like to ask a few questions. One, how do you manage the ascending aorta? Second, you had two or three patients who died of cardiac arrest on aortic cross-clamping. We had a similar experience back in the 1970s, and so we changed our method to partial cardiopulmonary bypass with an oxygenator in the circuit so that we could quickly shift to total cardiopulmonary perfusion if the heart dilated or fibrillated with aortic cross-clamping. The risk of cardiac arrest is particularly high in elderly patients or those with known heart disease. Do you concur? Third, the mortality is still quite high in emergent cases. What are your thoughts about decreasing that mortality? Last, other than the stabilizing the patient before repair to decrease mortality rate, as you reported today, what are the other advantages of delay of surgical repair, and what are your comparisons of the different methods of left-heart bypass or distant perfusion?

Thank you, Dr. Symbas; as usual, I enjoyed your contributions to aortic surgery.

Dr. Anthony A. Meyer (Chapel Hill, NC): I enjoyed the discussion. I would concur that there are patients with multiple injuries, pulmonary contusion, other things, in whom partial bypass or even general anesthesia can greatly increase the risk of immediate death not from rupture but from the other related problems. I do worry, however, that sometimes this report would be used as a way to not operate on people who don’t have associated things, because somebody says that if you can wait for some, it is okay to wait with all such patients. I agree that the key is to find the patients for whom there is a benefit to wait and treat those appropriately. The question I would ask is, what type of regimen did you use? Beta blockers seem to be the key. Do you have a special or specific regimen of a blood pressure target or something else that you use with your beta blocker treatment? I guess the last question I have, is this indeed the total experience in your institution of 29 blunt injuries to the aorta, or were there other ones that may not have been captured? From experience with the abdominal aortic aneurysms, oftentimes the patients are counted for the ones that go to the operating room, and the ones that die of rupture and never make it to the operating room are never swept up into the analysis.

Dr. Gordon F. Murray (Morgantown, WV): I would like to congratulate Peter on a very fine presentation and thank the authors for the opportunity to review the manuscript. This report of Dr. Symbas and his colleagues effectively has added to a growing body of literature which suggests that thoughtful delay in the repair of traumatic aortic transection in selected patients may improve clinical outcomes. Intuitively it seems reasonable to assume that coexistence of severe head injury, intraabdominal hemorrhage, cardiac and/or pulmonary contusion will all add to the overall mortality and morbidity in these patients undergoing immediate repair. The Emory experience also reminds us that there is minimal current clinical evidence to routinely justify an urgent operative response to traumatic rupture of the aorta with presently available diagnostic and therapeutic techniques. Early and accurate detection of variable aortic injury and aggressive antihypertensive therapy attain an acceptably low incidence of interval free rupture compared with historic control subjects. I am more hesitant to accept the authors’ implication that delayed repair per se has an inherent benefit; that is, a group 1 mortality of 36% and group 2A of 22%. First, the modern incidence of interval rupture remains real, at least 5% to 10% in the first 24 to 48 hours of admission. Second, relatively few of the authors’ group 1 patients had no or negligible associated injuries, a scenario in which operative survival really should approach 100%. Indeed, conditions contributing to the mortality and morbidity of patients who underwent immediate repair include splenic rupture, diaphragmatic rupture, kidney laceration prompting nephrectomy, and cardiac contusion. I would like the authors to comment on this observation and answer the question: Would these group 1 patients now be selected to group 2A in their current practice? It is also striking that the two deaths occurring in group 2A were patients undergoing early repair within 24 to 48 hours following laparotomy. Much like protocols for revascularization following myocardial infarction, have the authors developed practice guidelines to help us establish the appropriate timing for intervention in thoracic aortic transection, when delay is elected?

Overall, this report helps put a difficult clinical problem into perspective, at least momentarily. Investigational work is under way in many centers in this country for the application of endoluminal aortic stent-graft repair in the treatment of traumatic aorta injury. Endovascular repair of thoracic aortic transection has several potential benefits in managing the severely injured patient, including elimination of a thoracotomy, aortic cross-clamping, single lung ventilation, and the use of minimal anticoagulation. I would appreciate Dr. Symbas’ insight into the impact this new modality may have on his treatment paradigm for thoracic transection.

Dr. Michael L. Hawkins (Augusta, GA): I rise as a nonthoracic surgeon to perhaps point out the obvious. The antihypertensive treatment as proposed by Wheat for dissecting aneurysms is not the same situation as the moderate or mild hypotensive treatment currently proposed. I would like to ask the authors how they collaborate with their neurosurgeons to maintain cerebral perfusion pressure in these patients, many of whom have head injuries and perhaps associated increased intracranial pressure. A second question: what is the role, if any, in your institution for shunts instead of systemic anticoagulation?

Dr. Martin L. Dalton, jr. (Macon, GA): I appreciated your paper very much. We have followed your recommendations for several years, following the untimely death of a head injury patient 3 days after successful repair of his thoracic aortic injury. Now he would fall into your group 2B, the nonoperative management. Acting on the recently reported success from the Stanford group with endograft treatment of this condition, I would like to know if you plan to use that modality.

Dr. Panagiotis N. Symbas (Atlanta, GA): I would like to thank the discussants for their comments and questions. I will try to answer the questions in sequence.

Dr. Pate, as far as changing the current form of bypass, as you know early in our experience we were using femoral vein to femoral artery cardiopulmonary bypass. In order to minimize the adverse effects of the bypass during the last 10 years we switched to left atrial to femoral artery bypass. Considering the fact that the patients who are at risk for bleeding can be safely delayed, it appears prudent now to go back to the old form of circulatory support (i.e., femoral vein to femoral artery cardiopulmonary bypass), as we recently started doing and we will continue doing from now on. This leads me to the answer of his second question, what are my thoughts as to how to reduce our operative mortality in the group 1 patients. The use of femoral vein to femoral artery total cardiopulmonary, as we reinstituted recently, should increase the possibility of saving patients who suffer cardiac arrest, as the two of our group 1 patients. Such a patient, as you know, is almost impossible to salvage with the form of bypass that we were using (i.e., partial left atrial to femoral artery bypass with centrifugal pump). Similarly, the use of femoral vein to femoral artery cardiopulmonary should increase the salvage rate of patients with uncontrollable bleeding by having the capability to cool them and instituting total circulatory arrest if needed to control the hemorrhage. In addition, as I mentioned in my presentation, the two patients who suffered cardiac arrest were the oldest in the group, 68 and 76 years respectively. Preoperatively their coronary artery status was not evaluated, contrary to what is done routinely in all of such age patients who undergo other forms of aortic surgery. Having been shown that, under good treatment, it is safe to delay the repair of aortic rupture, in the future patients with aortic rupture older than 50 years should no longer undergo aortic repair before the status of their coronary arteries is assessed. As far as the question of what technique we used for the repair of ascending aortic rupture, we used conventional total cardiopulmonary bypass. Finally, the last of Dr. Pate’s questions: what are other, if any, advantages of the delayed repair of the aortic rupture? Due to the extensive fibrosis at the site and around the aortic tear and the dense scar formation, the possibility of precipitating uncontrollable hemorrhage during the dissection of the aorta or the placement of the cross-clamp between the left common carotid and left subclavian artery is greatly minimized if not totally eliminated. Also the repair, particularly in those cases in whom the rupture is next to the orifice of the left subclavian artery, is easier and certainly more secure. The downside of the delayed repair is that due to the extensive scarring at the site and around the injury, the dissection of the aorta is more difficult and tedious. As far as the comment on the preferential use of beta blockers, we agree that they are the drugs of choice. We added vasodilators only in those cases in whom we could not bring the mean arterial pressure to less than 70 mm Hg.

Dr. Murray raised the question whether the two patients who arrested should have been moved to subgroup 2A. We do agree that they should, and as I stated earlier, it is prudent to delay the repair of the aortic tear in patients 50 years or older until the status of their coronary arteries is assessed.

As far as stenting the rupture, which several discussants raised, certainly it appears to be a good therapeutical modality. However, due to the proximity of the rupture to the orifice of the left subclavian artery, an extrathoracic bypass to the subclavian will be needed. Because of this and especially because the performance of the stent over long periods has not been determined, I would defer my view until more data are accumulated.

The management of patients with central nervous system injury is problematic when, as we suggest, we need to maintain the mean arterial pressure at less than 70 mm Hg and the neurosurgeons want to have a perfusion pressure of 60 to 80 mm Hg. However, talking to them, it is possible to manage such cases with cerebral perfusion pressure of 65 mm Hg and under close observation. In addition, the perfusion of the brain can be sufficient with mean arterial pressure of less than 70 mm Hg as long as the intracranial pressure is well controlled. The bottom line is that such patients should be managed in close collaboration with the neurosurgeons in order to maintain balance between the needed mean pressure to protect the patient from rupture of the pseudoaneurysm and the needed perfusion pressure of the central nervous system.