- Email: [email protected]
Management of Blunt Traumatic Thoracic Aorta Injuries With Endovascular Stent-Grafts in a Tertiary Hospital in an Urban Asian City
Management of Blunt Traumatic Thoracic Aorta Injuries With Endovascular Stent-Grafts in a Tertiary Hospital in an Urban Asian City Glenn Wei Leong Tan,1 Chong Han Pek,1 Daniel Wong,2 Sundeep Punamiya,2 Ming Terk Chiu,3 Vijayan Appasamy,3 Kok Hoong Chia,1 and Chee Wei Lee,1 Singapore
Background: Thoracic aortic injury from blunt trauma is a life-threatening condition with significant mortality and morbidity with open surgical repair. Endovascular means of treatment is emerging as an attractive and less invasive option. We report our experience with endovascular stent-graft repair for blunt traumatic thoracic aorta injury. Methods: Between January 2003 and August 2009, six patients underwent endovascular repair of blunt traumatic thoracic aorta injury. Data from the medical records of these patients were analyzed. Results: All patients had deceleration injury resulting in thoracic aorta pseudoaneurysm at the aortic isthmus just distal to the left subclavian artery. Four patients were motorcyclists who had been involved in road-traffic accidents and two had fallen from a height. The mean Injury Severity Score was 33.8 (range, 21-43). All procedures were performed within 48 hours of admission and technical success was 100%. Five patients (83.3%) had the left subclavian artery intentionally covered by the stent-graft to achieve adequate proximal landing zone. None of them had any left upper limb ischemic complications or cerebrovascular events after the procedure. One patient had preservation of the left subclavian artery because intraoperative vertebral angiogram showed a hypoplastic right vertebral artery. No other procedural complications, paraplegia, or deaths were reported. Mean follow-up was 18.8 months (range, 1.5-42 months). Conclusions: Our early experience of endovascular stent-grafting for blunt traumatic thoracic aorta injury suggests that this emerging technique is safe for treatment of such pathology in our inherently Asian population. Left subclavian artery may be sacrificed if adequate proximal landing zone is required for the stent-graft; and when vertebral angiogram was performed which did not show a dominant left vertebral artery.
INTRODUCTION 1
Department of General Surgery, Vascular Surgery Service, Tan Tock Seng Hospital, Singapore. 2
Department of Radiology, Interventional Radiology Service, Tan Tock Seng Hospital, Singapore. 3 Department of General Surgery, Trauma Surgery Service, Tan Tock Seng Hospital, Singapore.
Correspondence to: Glenn Wei Leong Tan, MB, ChB, MRCS (Glasg), M Med (Surgery), Department of General Surgery, Tan Tock Seng Hospital, 11 Jalan Tan Tock Seng, Singapore 308433, Singapore, E-mail: [email protected] Ann Vasc Surg 2011; 25: 605-611 DOI: 10.1016/j.avsg.2010.12.009 Ó Annals of Vascular Surgery Inc. Published online: March 24, 2011
Traumatic aortic injury is a life-threatening condition with high prehospital mortality rates to a maximum of 85%.1 Most of the patients die at the scene of the injury. The postulated mechanisms of blunt great vessel injury include shear forces caused by relative mobility of a portion of the vessel adjacent to a fixed portion, compression of the vessel between bony structures, and profound intraluminal hypertension during the severe traumatic event. The most common portion of the aorta to be affected is the proximal descending thoracic aorta. 605
606 Glenn Tan et al.
Descending thoracic aortic injuries often occur at the isthmus just distal to the left subclavian artery because of the fixation of the descending thoracic aorta at the ligamentum arteriosum. The traditional management of descending thoracic aortic injury is open surgery with some form of bypass shunt or cardiopulmonary bypass or, for centers without facilities for bypass, simple clamp-and-repair technique. Mortality rates for open surgery can be as high as between 15% and 30%.2,3 Recently, endovascular techniques with low mortality and morbidity rates have become popular in the treatment of traumatic thoracic aortic injuries.4,5 Most of the large series of endovascular stenting for blunt traumatic aortic injuries have emerged from Western centers, with only a small series from the Far East.6 We report our experience with this technique in a tertiary hospital in an urban Asian city.
MATERIALS AND METHODS Between January 2003 and August 2009, a search was made for traumatic thoracic aorta injuries from the prospective trauma registry in Tan Tock Seng Hospital (TTSH), Singapore. The registry is staffed by two permanent trauma coordinators and is regularly audited for accuracy. The trauma registry includes all patients admitted to TTSH for trauma. TTSH is a tertiary hospital located in the central part of Singapore housing the busiest trauma center in the island country with an estimated population of five million. It is well accessible with good motorway links to the rest of the country. During the study period of 80 months, 45 patients were identified as having traumatic thoracic aorta injury. A total of 31 patients died in the Accident and Emergency Department during resuscitation before any intervention or investigations could be performed, and the thoracic aortic injury was only discovered at the postmortem examination. Of the remaining 14 patients, only six underwent endoluminal stenting of the thoracic aorta. The other eight died from their injuries. Of these, three patients were brought to the operation theater for resuscitation with surgery planned to achieve hemostasis: only one patient was stable enough to undergo thoracotomy and the remaining two died before thoracotomy could be performed. However, the patient who underwent thoracotomy died intraoperatively before aortic repair was performed. The remaining five patients had ongoing resuscitation in the intensive care unit but were never stable enough to be transported to the operation theater for surgery and subsequently died. The medical
Annals of Vascular Surgery
records were reviewed for the six patients who underwent endovascular stenting. All trauma patients are brought to the Accident and Emergency Department’s Resuscitation room on arrival and stabilized by the Accident and Emergency staff and General Surgery and Trauma Surgery services according to Advanced Trauma Life Support protocols. Multidetector computed tomographic scan of the thorax with 1-mm fine cuts was performed for all trauma patients who demonstrated a widened mediastinum on chest X-ray, which was part of the trauma series of X-rays performed in the Accident and Emergency Department. All but one patient had a widened mediastinum on the chest X-ray. The patient who had a normal chest X-ray had to undergo a computed tomographic (CT) scan of the thorax because of the severity of the mechanism of the injury (fall from the fourth storey of a building), which incidentally showed the pseudoaneurysm of the thoracic aorta. On diagnosis of the thoracic aortic pathology, the vascular surgery service was consulted for all cases and decision and timing for surgery was discussed and agreed on between the trauma and vascular surgery teams. All the patients who were confirmed to have a traumatic thoracic aorta injury on CT scan were offered endovascular stenting. Life-threatening and more severe injuries were treated first before endovascular stenting of the thoracic aorta could be performed (Figs. 1 and 2). The general criteria used to determine anatomic suitability for endovascular repair in this series was a proximal and distal aortic neck of a minimum of 20 mm in diameter. The left subclavian artery was sacrificed and covered in cases when adequate landing zone was required. The endografts were oversized by approximately 20% for our study group. Iliac vessels also had to be >7 mm in diameter to accommodate the delivery system. All devices used were one of the following two types: the Talent or Valiant endovascular stent-graft (Medtronic Inc.). In 2006, the Talent thoracic stent-graft was replaced by the Valiant thoracic stent-graft at our center. Both endovascular devices are straight stentgrafts consisting of an external nitinol wire support structure with an outer polyester covering. The Talent device was used in the first three patients, whereas the subsequent patients had the Valiant variety. The Talent graft was available in only 100mm lengths, with diameters ranging from 24 to 46 mm, whereas the Valiant devices have lengths ranging from 100 to 150 mm and diameters ranging from 22 to 46 mm. There is a bare portion of proximal stent-graft of 12 mm in the Valiant and 15 mm in the Talent.
Vol. 25, No. 5, July 2011
Management of blunt traumatic thoracic aorta injuries 607
Fig. 2. Angiographic film of thoracic aorta pseudoaneurysm just distal to the left subclavian artery. Fig. 1. Sagittal image of computed tomographic aortogram showing thoracic aorta pseudoaneurysm just distal to the left subclavian artery.
All procedures were performed under general anesthesia in the angiography suite by a vascular surgeon and an interventional radiologist under fluoroscopic guidance. Patients were positioned supine, prepped, and draped as for thoracotomy in case of stent-graft failure. Lumbar drains were not used. In all cases, the right common femoral artery was surgically exposed for direct puncture, sheath placement, and subsequent delivery of the stentgraft. Percutaneous puncture of the left common femoral artery was also performed. For all our patients, the ileofemoral artery was >7 mm in diameter and none required a conduit for stent-graft delivery. An aortography was performed initially to confirm the aortic pathology before proceeding with stent-graft deployment. A selective vertebral angiogram was also performed intraoperatively for all patients for assessment of posterior circulation to the brain to ensure patency of the vertebral arteries in case there was a need to preserve the left subclavian artery when it was the dominant supply to the basilar artery. Sizing of the stentgraft was determined by the preoperative CT scan and intraoperative aortogram. A completion aortogram at the end of the procedure was also carried out in all cases to ensure accurate placement and no endoleaks. Surgical arterial closure was performed for the right groin arterial cutdown (Figs. 3 and 4).
Technical success was defined as deployment of the endograft in the correct position to exclude the aortic pathology and to ensure the absence of endoleak on completion aortogram. Follow-up of the patients was carried out with clinical examination and repeat CT planned at 6-month intervals for the first year, then every 12 months or earlier if necessary. Plain chest radiography was also performed to monitor for stent position and fractures in cases when CT scan was not performed at follow-up. All patients were given antiplatelet therapy after endovascular stenting.
RESULTS All the patients had undergone high impact accidents with some form of deceleration injury resulting in a traumatic pseudoaneurysm of the thoracic aorta just distal to the left subclavian artery. Mechanism of injury in the case of two patients was fall from a height: one patient fell from the seventh storey of a building in a suicide attempt and another was a domestic helper who fell from the fourth storey while cleaning windows. The other four patients were motorcyclists who were involved in road-traffic accidents and were flung to a distance from their vehicles. There were four men and two women. All patients were of Asian descent (three Chinese, two Malay, and one Indian). The mean age was 45 years (range, 24-72 years). All patients had a variety of associated injuries, including fractures of long bones, ribs, and pelvis; head injuries;
608 Glenn Tan et al.
Fig. 3. Angiographic film showing a hypoplastic right vertebral artery.
abdominal injuries; spine injuries; pulmonary injuries; and other various orthopedic injuries, as a result of the severity of the trauma. The mean Injury Severity Score was 33.8 (range, 21-43). None of the patients had any severe comorbidities (hypertension, diabetes mellitus, ischemic heart disease, cerebrovascular disease, renal impairment, pulmonary disease, or hyperlipidemia), with the exception of patient #2 who was a chronic smoker. The length of intensive care unit stay and hospital stay was prolonged mainly because of the severity of the associated injuries from the trauma. The length of hospital stay was prolonged in patients who had severe orthopedic limb injuries or amputations requiring rehabilitation and intensive physiotherapy. The traumatic aorta injury was not the main reason in any of the patients for the prolonged hospital stay. All patients had stent-graft insertion within 48 hours of admission. The mean time from admission to stent-graft insertion was 21.8 hours (range, 6-48 hours). The patients’ injuries were treated according to the risk of threat to the life of the patient, which explains the delay from admission before some of these patients received their stent-grafts. This assessment was made jointly and with close collaboration between the trauma surgeons and vascular surgeons. All patients had high surgical risk, as reflected by the American Society of Anesthesiologists (ASA) score ranging between 3 and 4. The
Annals of Vascular Surgery
Fig. 4. Angiographic film of normal caliber right vertebral artery.
mean operating time was 129.2 minutes (range, 105-180 minutes). A summary of the procedural information is displayed in Table I. All patients had a traumatic thoracic aorta pseudoaneurysm occurring just distal to the left subclavian artery. The average diameter of the thoracic aorta was 23 mm (range, 20-28 mm). All patients had a landing zone of 20 mm between the left subclavian artery and the site of aortic injury. To gain enough proximal landing zone for the stentgraft so as to obtain a good seal, the left subclavian artery had to be sacrificed and covered with the stent-graft for the first five patients (83.3%). In patient #6, the selective vertebral angiogram showed a hypoplastic-attenuated right vertebral artery, suggesting that the left vertebral artery may be supplying most of the blood to the basilar artery, and a decision was made to preserve the left subclavian artery, which was done successfully because the proximal landing zone was 20 mm in length. None of the patients needed the left common carotid artery or innominate artery to be covered by the stent-graft to gain adequate landing zone. Of the patients who had the left subclavian artery covered, all did not have a radial or brachial pulse after the procedure, but none of the patients complained of ischemic rest pain or claudication in the left upper limb. None of the patients required carotide subclavian bypass surgery to revascularize the left
Vol. 25, No. 5, July 2011
Management of blunt traumatic thoracic aorta injuries 609
Table I. Procedural information Average size Age/ of thoracic Patients gender aorta (mm) Device
Device size (mm)
1 2 3 4 5 6
30 24 24 28 32 24
40/M 38/M 24/F 42/M 72/F 54/M
25 21 20 23 28 21
Talent Talent Talent Valiant Valiant Valiant
30 24 24 28 32 24
Elapsed time Operating from injury to time Landing Left subclavian Follow-up repair (hours) (minutes) zone (mm) artery covered (months)
100 37.5 100 25 100 48 150 6 150 7 150 7.5
subclavian artery or chimney stent to preserve the left subclavian artery. None of the patients had a stroke after the procedure. Only one stent-graft was required for each patient. All the patients had primary technical success of the stent-graft deployment (100%). None of the patients had adverse characteristics to endovascular stenting such as calcium plaques, mural thrombus, small iliofemoral access arteries, or tortuosity of the aorta. Blood loss was minimal (<100 mLs) for all patients, except for patient #4 (300 mLs). Only patient #1 received 5,000 units of intravenous unfractionated heparin during deployment of stent-graft, but this did not result in any adverse hemorrhagic events. The remaining patients were not given heparin because they had suffered multiple other injuries and the surgeon did not want to induce coagulopathy in these patients. There were no early endograft-related morbidities such as graft migration, collapse, infection, or thrombosis. None of the patients had any postoperative cardiovascular, respiratory, renal, or cerebrovascular events. None of the patients suffered from paraplegia. There were no access site problems such as groin wound infections, bleeding, hematoma, or pseudoaneurysm formation. No paraplegia, bowel ischemia, or limb ischemia was observed. No intraoperative, in-hospital, or 30-day mortalities were observed. None of the patients required any further procedures for the thoracic aorta pseudoaneurysm. Mean follow-up was 18.8 months. Compliance to follow-up was reduced in two patients. Patient #2 had underlying psychiatric disorder and was noncompliant to his hospital appointments and was lost to follow-up after 12 months. Patient #3 was a domestic worker from Indonesia and chose to return home for further treatment, explaining the follow-up of only 1.5 months. In the remaining four patients, there were no documented late complications (Fig. 5).
125 180 110 105 120 135
5 16 17 18 12 20
Yes Yes Yes Yes Yes No
42 12 1.5 36 18 3
DISCUSSION Since the introduction of endovascular stenting of the thoracic aorta, there has been a paradigm shift toward using this less invasive method for treatment of thoracic aorta pathology such as aneurysmal disease, dissections, and atherosclerotic ulcers. This has been fuelled by the benefit gained by the patient in the form of lower morbidity and mortality rates5 and more economical utilization of hospital resources resulting from shorter hospital stays and usage of intensive care facilities associated with this technique. Trauma patients often have associated multiple injuries, usually requiring several operations. The traditional gold standard for traumatic thoracic aorta injury is open surgery, with its high mortality rates in the range of 15-30%.2,3 Avoiding a thoracotomy, cardiopulmonary bypass and aortic cross-clamping together with its morbid complications of prolonged ventilatory support, pulmonary complications, risk of renal failure, and paraplegia is certainly advantageous to the already critically ill patient. With multiple injuries and operations often required in trauma patients, the need for massive blood transfusion coupled with hypothermia common in this set of patients, concomitant coagulopathy remains a major concern. Systemic heparinization usually given before stent deployment can further worsen the coagulopathy. The role of heparin is to prevent thrombus formation in the aorta and iliac vessels around the endovascular deployment device and also in the important arteries arising from the aorta such as the visceral vessels. In our study, only one patient (16.7%) received intraoperative heparin, but there were no thromboembolic complications noted despite omission of heparin in most of our patients. Because most injuries occur at the aortic isthmus with limited landing zone for the stent-graft, left subclavian artery coverage may be inevitable. Several studies have shown the safety of left
610 Glenn Tan et al.
Fig. 5. Post-stenting with left subclavian artery preserved.
subclavian artery sacrifice,7-10 which we have further emphasized in our study in which 83.3% of our patients had uneventful sacrifice of the left subclavian artery. Although some investigators have reported complications from left subclavian artery coverage11 and the necessity for prophylactic carotidesubclavian bypass12 or creating a window in the stent-graft for the left subclavian artery ostium,13 we have not experienced the need for these extra procedures that will lengthen the operative time and may add further stress to an already compromised trauma patient. We do advocate selective vertebral angiogram during the procedure to assess the posterior circulation to the brain. This practice helped pick a patient in our cohort with an attenuated hypoplastic right vertebral artery and we opted to preserve the left subclavian artery circulation in this case. In cases when the left subclavian artery has to be sacrificed with potential reduction of blood flow to the posterior circulation, the adjunct maneuvers described by Rousseau et al.13 can be used or carotidesubclavian bypass operation should be performed before stent-graft deployment. CT angiogram is another option that can be used to define the anatomy of the vertebral arteries but will necessitate another journey to the CT scanner because most CT scans of the thorax do not include the vertebrobasilar arterial system. We prefer intraoperative selective vertebral angiogram before stent placement because it gives a more physiologic representation of the dominance of the posterior circulation and Circle of Willis and does not significantly increase the operative time and can potentially prevent a catastrophic posterior circulation stroke.
Annals of Vascular Surgery
There has been some controversy regarding the timing of intervention for traumatic thoracic aortic injuries with some advocating delayed treatment,13,14 whereas others have recommended immediate repair.15 In patients with traumatic aortic injury, blood flow is maintained in the disrupted aorta by the adventitia and surrounding mediastinal structures. This is a precarious situation with a potential for acute exsanguination into the thorax at any time. Some previous studies have suggested that repair can be delayed for patients who are stable with concomitant injuries, presumably owing to the stress of additional surgery.16,17 In our study, all patients had endovascular repair as soon as other more life-threatening injuries were treated and the patient was stable for transfer to the angiography suite. We have not experienced any major morbidity or mortality from early endovascular stent-grafting in our series despite the fact that all our patients had associated injuries, with many requiring multiple operations. Despite all the advantages of endovascular repair in thoracic aorta injuries, there remain some challenges to this technique. The typical trauma patient is younger than the average patient with degenerative aortic disease for whom the stent-grafts were initially designed. Furthermore, most injuries occur at the isthmus, thereby requiring for the stent-grafts to be placed across a tighter more angulated aortic arch. Younger subjects also have smaller caliber aorta and iliac access vessels, which may present difficulties in inserting the devices, stent-graft conformity to aortic wall, or obtaining an accuratesized stent-graft. These factors may lead to higher risk of endoleaks, stent collapse,18-20 or iliac artery avulsions.4,5 Asian patients are of smaller build and thus are perceived to have smaller aortas and access vessels, which may be a contraindication to endovascular approach to the management of traumatic thoracic aortic injuries. However, we have not encountered any problems with small iliofemoral arteries requiring access conduits or small aortas where a suitably sized stent-graft was not available. In our series, we did not need to use any uncovered bare stents, although we acknowledge their usefulness in certain situations. It can be deployed within the stent-grafts in the acute setting to allow better conformity to the native aortic curvature in cases when there is poor apposition of the proximal portion and also in delayed partial or complete collapse of the stent-grafts. However, there are concerns that bare stents may erode though the graft fabric over time or result in proximal aortic perforations and hence our reservations about their use.21
Vol. 25, No. 5, July 2011
Endovascular stent-grafts require long-term surveillance to assess for stent-graft durability and endoleaks, usually with regular CT scans. This results in large amounts of radiation exposure, especially undesirable in young patients. Only long-term results will demonstrate the robustness of the endovascular stents and provide a clearer direction or alternatives for follow-up.
CONCLUSION Endovascular stent-graft for traumatic thoracic aorta injury has been shown to be a feasible and safe method of repair. This is particularly advantageous in trauma patients who have multiple injuries, where avoiding a major thoracotomy can reduce morbidity for the patient. Left subclavian artery coverage by endovascular stent-graft to obtain adequate landing zone is also safe when intraoperative vertebral angiogram has shown adequate posterior circulation to the brain from the right vertebral artery. Omitting systemic heparinization during stent-graft deployment does not increase the risk of thromboembolic complications from the procedure and may be desirable in trauma patients to reduce the risk of coagulopathy. Our study demonstrates feasibility of endovascular management of thoracic aorta injury in Asian patients and did not encounter any problems with small access vessel and aorta size that is widely perceived in Asian patients because of their smaller build.
The authors thank Ms Karen Go and Ms Yen Teng Yeo for the use of information from the Trauma registry and Ms Wenzhao Sun for help in tracing the medical records. The authors would also like to thank the library services (Ms Cecilia James and Ms Norina Hamid) in TTSH for providing reference materials. REFERENCES 1. Parmley LF, Mattingly TW, Manion WC, et al. Non-penetrating traumatic injury of the aorta. Circulation 1958;17: 1086-1101. 2. Von Oppell UO, Dunne TT, De Groot MK, et al. Traumatic aortic rupture: twenty-year meta-analysis of mortality and risk of paraplegia. Ann Thorac Surg 1994;58:585. 3. Jamieson WR, Janusz MT, Gudas VM, et al. Traumatic rupture of the thoracic aorta: third decade of experience. Am J Surg 2002;183:571-575.
Management of blunt traumatic thoracic aorta injuries 611
4. Bent CL, Matson MB, Sobeh M, et al. Endovascular management of acute blunt traumatic thoracic aortic injury: a single center experience. J Vasc Surg 2007;46:920-927. 5. Stone DH, Brewster DC, Kwolek CJ, et al. Stent-graft versus open surgical repair of the thoracic aorta: mid term results. J Vasc Surg 2006;44:1188-1197. 6. Fujikawa T, Yukioka T, Ishimaru S, et al. Endovascular stent grafting for the treatment of blunt thoracic aortic injury. J Trauma 2001;50:223-229. 7. Sandroussi C, Waltham M, Hughes C, et al. Endovascular grafting of the thoracic aorta, an evolving therapy: tenyear experience in a single centre. ANZ J Surg 2007;77: 974-980. 8. Riesenman PJ, Farber MA, Mendes RR, et al. Endovascular repair of lesions involving the descending thoracic aorta. J Vasc Surg 2005;42:1063-1074. 9. Melnitchouk S, Pfammatter T, Kadner A, et al. Emergency stent-graft placement for haemorrhage control in acute thoracic aortic rupture. Eur J Cardiothorac Surg 2004;25: 1032-1038. 10. Moainie SL, Nechis DG, Gammie JS, et al. Endovascular stenting for traumatic aortic injury: an emerging new standard of care. Ann Thorac Surg 2008;85:1625-1630. 11. Riesenman PJ, Farber MA, Mendes RR, et al. Coverage of the left subclavian artery during thoracic endovascular aortic repair. J Vasc Surg 2007;45:90-94. 12. Leurs LJ, Bell R, Degrieck Y, et al. Endovascular treatment of thoracic aortic diseases: combined experience from the EUROSTAR and United Kingdom Thoracic Endograft registries. J Vasc Surg 2004;40:670-680. 13. Rousseau H, Soula P, Perreault P, et al. Delayed treatment of traumatic rupture of the thoracic aorta with endoluminal covered stent. Circulation 1999;99:458-504. 14. Pacini D, Angeli E, Fattori R, et al. Traumatic rupture of the thoracic aorta: ten-years of delayed management. J Thorac Cardiovasc Surg 2005;129:880-884. 15. Thompson CS, Rodriguez JA, Ramaiah VG, DiMugno L, Shafique S, Olsen D, Diethrich EB. Acute traumatic rupture of the thoracic aorta treated with endoluminal stent grafts. J Trauma 2002;52:1173-1177. 16. Walker WA, Pete JW. Medical management of acute traumatic rupture of thoracic aorta. Ann Thorac Surg 1990;50: 965-967. 17. Maggisano R, Nathens A, Alexandrova NA, Cina C, Boulanger B, McKenzie R, Harrison AW. Traumatic rupture of the thoracic aorta: should one always operate immediately? Ann Vasc Surg 1995;9:44-52. 18. Neschis DG, Moaine S, Gutta R, et al. Twenty consecutive cases of endografts repair of traumatic aortic disruption: lessons learned. J Vasc Surg 2007;45:487-492. 19. Idu MM, Reekers JA, Balm R, et al. Collapse of stent-graft following treatment of traumatic thoracic aortic rupture. J Endovasc Ther 2005;12:503-507. 20. Steinbauer MG, Stehr A, Pfister K, et al. Endovascular repair of proximal endograft collapse after treatment of thoracic aortic disease. J Vasc Surg 2006;43:609-612. 21. Malina M, Brunkwall J, Ivancev K, et al. Late aortic arch perforation by graft anchoring stent: complication of endovascular thoracic aneurysm exclusion. J Endovasc Surg 1998;5:274-277.
Background: Thoracic aortic injury from blunt trauma is a life-threatening condition with significant mortality and morbidity with open surgical repair. Endovascular means of treatment is emerging as an attractive and less invasive option. We report our experience with endovascular stent-graft repair for blunt traumatic thoracic aorta injury. Methods: Between January 2003 and August 2009, six patients underwent endovascular repair of blunt traumatic thoracic aorta injury. Data from the medical records of these patients were analyzed. Results: All patients had deceleration injury resulting in thoracic aorta pseudoaneurysm at the aortic isthmus just distal to the left subclavian artery. Four patients were motorcyclists who had been involved in road-traffic accidents and two had fallen from a height. The mean Injury Severity Score was 33.8 (range, 21-43). All procedures were performed within 48 hours of admission and technical success was 100%. Five patients (83.3%) had the left subclavian artery intentionally covered by the stent-graft to achieve adequate proximal landing zone. None of them had any left upper limb ischemic complications or cerebrovascular events after the procedure. One patient had preservation of the left subclavian artery because intraoperative vertebral angiogram showed a hypoplastic right vertebral artery. No other procedural complications, paraplegia, or deaths were reported. Mean follow-up was 18.8 months (range, 1.5-42 months). Conclusions: Our early experience of endovascular stent-grafting for blunt traumatic thoracic aorta injury suggests that this emerging technique is safe for treatment of such pathology in our inherently Asian population. Left subclavian artery may be sacrificed if adequate proximal landing zone is required for the stent-graft; and when vertebral angiogram was performed which did not show a dominant left vertebral artery.
INTRODUCTION 1
Department of General Surgery, Vascular Surgery Service, Tan Tock Seng Hospital, Singapore. 2
Department of Radiology, Interventional Radiology Service, Tan Tock Seng Hospital, Singapore. 3 Department of General Surgery, Trauma Surgery Service, Tan Tock Seng Hospital, Singapore.
Correspondence to: Glenn Wei Leong Tan, MB, ChB, MRCS (Glasg), M Med (Surgery), Department of General Surgery, Tan Tock Seng Hospital, 11 Jalan Tan Tock Seng, Singapore 308433, Singapore, E-mail: [email protected] Ann Vasc Surg 2011; 25: 605-611 DOI: 10.1016/j.avsg.2010.12.009 Ó Annals of Vascular Surgery Inc. Published online: March 24, 2011
Traumatic aortic injury is a life-threatening condition with high prehospital mortality rates to a maximum of 85%.1 Most of the patients die at the scene of the injury. The postulated mechanisms of blunt great vessel injury include shear forces caused by relative mobility of a portion of the vessel adjacent to a fixed portion, compression of the vessel between bony structures, and profound intraluminal hypertension during the severe traumatic event. The most common portion of the aorta to be affected is the proximal descending thoracic aorta. 605
606 Glenn Tan et al.
Descending thoracic aortic injuries often occur at the isthmus just distal to the left subclavian artery because of the fixation of the descending thoracic aorta at the ligamentum arteriosum. The traditional management of descending thoracic aortic injury is open surgery with some form of bypass shunt or cardiopulmonary bypass or, for centers without facilities for bypass, simple clamp-and-repair technique. Mortality rates for open surgery can be as high as between 15% and 30%.2,3 Recently, endovascular techniques with low mortality and morbidity rates have become popular in the treatment of traumatic thoracic aortic injuries.4,5 Most of the large series of endovascular stenting for blunt traumatic aortic injuries have emerged from Western centers, with only a small series from the Far East.6 We report our experience with this technique in a tertiary hospital in an urban Asian city.
MATERIALS AND METHODS Between January 2003 and August 2009, a search was made for traumatic thoracic aorta injuries from the prospective trauma registry in Tan Tock Seng Hospital (TTSH), Singapore. The registry is staffed by two permanent trauma coordinators and is regularly audited for accuracy. The trauma registry includes all patients admitted to TTSH for trauma. TTSH is a tertiary hospital located in the central part of Singapore housing the busiest trauma center in the island country with an estimated population of five million. It is well accessible with good motorway links to the rest of the country. During the study period of 80 months, 45 patients were identified as having traumatic thoracic aorta injury. A total of 31 patients died in the Accident and Emergency Department during resuscitation before any intervention or investigations could be performed, and the thoracic aortic injury was only discovered at the postmortem examination. Of the remaining 14 patients, only six underwent endoluminal stenting of the thoracic aorta. The other eight died from their injuries. Of these, three patients were brought to the operation theater for resuscitation with surgery planned to achieve hemostasis: only one patient was stable enough to undergo thoracotomy and the remaining two died before thoracotomy could be performed. However, the patient who underwent thoracotomy died intraoperatively before aortic repair was performed. The remaining five patients had ongoing resuscitation in the intensive care unit but were never stable enough to be transported to the operation theater for surgery and subsequently died. The medical
Annals of Vascular Surgery
records were reviewed for the six patients who underwent endovascular stenting. All trauma patients are brought to the Accident and Emergency Department’s Resuscitation room on arrival and stabilized by the Accident and Emergency staff and General Surgery and Trauma Surgery services according to Advanced Trauma Life Support protocols. Multidetector computed tomographic scan of the thorax with 1-mm fine cuts was performed for all trauma patients who demonstrated a widened mediastinum on chest X-ray, which was part of the trauma series of X-rays performed in the Accident and Emergency Department. All but one patient had a widened mediastinum on the chest X-ray. The patient who had a normal chest X-ray had to undergo a computed tomographic (CT) scan of the thorax because of the severity of the mechanism of the injury (fall from the fourth storey of a building), which incidentally showed the pseudoaneurysm of the thoracic aorta. On diagnosis of the thoracic aortic pathology, the vascular surgery service was consulted for all cases and decision and timing for surgery was discussed and agreed on between the trauma and vascular surgery teams. All the patients who were confirmed to have a traumatic thoracic aorta injury on CT scan were offered endovascular stenting. Life-threatening and more severe injuries were treated first before endovascular stenting of the thoracic aorta could be performed (Figs. 1 and 2). The general criteria used to determine anatomic suitability for endovascular repair in this series was a proximal and distal aortic neck of a minimum of 20 mm in diameter. The left subclavian artery was sacrificed and covered in cases when adequate landing zone was required. The endografts were oversized by approximately 20% for our study group. Iliac vessels also had to be >7 mm in diameter to accommodate the delivery system. All devices used were one of the following two types: the Talent or Valiant endovascular stent-graft (Medtronic Inc.). In 2006, the Talent thoracic stent-graft was replaced by the Valiant thoracic stent-graft at our center. Both endovascular devices are straight stentgrafts consisting of an external nitinol wire support structure with an outer polyester covering. The Talent device was used in the first three patients, whereas the subsequent patients had the Valiant variety. The Talent graft was available in only 100mm lengths, with diameters ranging from 24 to 46 mm, whereas the Valiant devices have lengths ranging from 100 to 150 mm and diameters ranging from 22 to 46 mm. There is a bare portion of proximal stent-graft of 12 mm in the Valiant and 15 mm in the Talent.
Vol. 25, No. 5, July 2011
Management of blunt traumatic thoracic aorta injuries 607
Fig. 2. Angiographic film of thoracic aorta pseudoaneurysm just distal to the left subclavian artery. Fig. 1. Sagittal image of computed tomographic aortogram showing thoracic aorta pseudoaneurysm just distal to the left subclavian artery.
All procedures were performed under general anesthesia in the angiography suite by a vascular surgeon and an interventional radiologist under fluoroscopic guidance. Patients were positioned supine, prepped, and draped as for thoracotomy in case of stent-graft failure. Lumbar drains were not used. In all cases, the right common femoral artery was surgically exposed for direct puncture, sheath placement, and subsequent delivery of the stentgraft. Percutaneous puncture of the left common femoral artery was also performed. For all our patients, the ileofemoral artery was >7 mm in diameter and none required a conduit for stent-graft delivery. An aortography was performed initially to confirm the aortic pathology before proceeding with stent-graft deployment. A selective vertebral angiogram was also performed intraoperatively for all patients for assessment of posterior circulation to the brain to ensure patency of the vertebral arteries in case there was a need to preserve the left subclavian artery when it was the dominant supply to the basilar artery. Sizing of the stentgraft was determined by the preoperative CT scan and intraoperative aortogram. A completion aortogram at the end of the procedure was also carried out in all cases to ensure accurate placement and no endoleaks. Surgical arterial closure was performed for the right groin arterial cutdown (Figs. 3 and 4).
Technical success was defined as deployment of the endograft in the correct position to exclude the aortic pathology and to ensure the absence of endoleak on completion aortogram. Follow-up of the patients was carried out with clinical examination and repeat CT planned at 6-month intervals for the first year, then every 12 months or earlier if necessary. Plain chest radiography was also performed to monitor for stent position and fractures in cases when CT scan was not performed at follow-up. All patients were given antiplatelet therapy after endovascular stenting.
RESULTS All the patients had undergone high impact accidents with some form of deceleration injury resulting in a traumatic pseudoaneurysm of the thoracic aorta just distal to the left subclavian artery. Mechanism of injury in the case of two patients was fall from a height: one patient fell from the seventh storey of a building in a suicide attempt and another was a domestic helper who fell from the fourth storey while cleaning windows. The other four patients were motorcyclists who were involved in road-traffic accidents and were flung to a distance from their vehicles. There were four men and two women. All patients were of Asian descent (three Chinese, two Malay, and one Indian). The mean age was 45 years (range, 24-72 years). All patients had a variety of associated injuries, including fractures of long bones, ribs, and pelvis; head injuries;
608 Glenn Tan et al.
Fig. 3. Angiographic film showing a hypoplastic right vertebral artery.
abdominal injuries; spine injuries; pulmonary injuries; and other various orthopedic injuries, as a result of the severity of the trauma. The mean Injury Severity Score was 33.8 (range, 21-43). None of the patients had any severe comorbidities (hypertension, diabetes mellitus, ischemic heart disease, cerebrovascular disease, renal impairment, pulmonary disease, or hyperlipidemia), with the exception of patient #2 who was a chronic smoker. The length of intensive care unit stay and hospital stay was prolonged mainly because of the severity of the associated injuries from the trauma. The length of hospital stay was prolonged in patients who had severe orthopedic limb injuries or amputations requiring rehabilitation and intensive physiotherapy. The traumatic aorta injury was not the main reason in any of the patients for the prolonged hospital stay. All patients had stent-graft insertion within 48 hours of admission. The mean time from admission to stent-graft insertion was 21.8 hours (range, 6-48 hours). The patients’ injuries were treated according to the risk of threat to the life of the patient, which explains the delay from admission before some of these patients received their stent-grafts. This assessment was made jointly and with close collaboration between the trauma surgeons and vascular surgeons. All patients had high surgical risk, as reflected by the American Society of Anesthesiologists (ASA) score ranging between 3 and 4. The
Annals of Vascular Surgery
Fig. 4. Angiographic film of normal caliber right vertebral artery.
mean operating time was 129.2 minutes (range, 105-180 minutes). A summary of the procedural information is displayed in Table I. All patients had a traumatic thoracic aorta pseudoaneurysm occurring just distal to the left subclavian artery. The average diameter of the thoracic aorta was 23 mm (range, 20-28 mm). All patients had a landing zone of 20 mm between the left subclavian artery and the site of aortic injury. To gain enough proximal landing zone for the stentgraft so as to obtain a good seal, the left subclavian artery had to be sacrificed and covered with the stent-graft for the first five patients (83.3%). In patient #6, the selective vertebral angiogram showed a hypoplastic-attenuated right vertebral artery, suggesting that the left vertebral artery may be supplying most of the blood to the basilar artery, and a decision was made to preserve the left subclavian artery, which was done successfully because the proximal landing zone was 20 mm in length. None of the patients needed the left common carotid artery or innominate artery to be covered by the stent-graft to gain adequate landing zone. Of the patients who had the left subclavian artery covered, all did not have a radial or brachial pulse after the procedure, but none of the patients complained of ischemic rest pain or claudication in the left upper limb. None of the patients required carotide subclavian bypass surgery to revascularize the left
Vol. 25, No. 5, July 2011
Management of blunt traumatic thoracic aorta injuries 609
Table I. Procedural information Average size Age/ of thoracic Patients gender aorta (mm) Device
Device size (mm)
1 2 3 4 5 6
30 24 24 28 32 24
40/M 38/M 24/F 42/M 72/F 54/M
25 21 20 23 28 21
Talent Talent Talent Valiant Valiant Valiant
30 24 24 28 32 24
Elapsed time Operating from injury to time Landing Left subclavian Follow-up repair (hours) (minutes) zone (mm) artery covered (months)
100 37.5 100 25 100 48 150 6 150 7 150 7.5
subclavian artery or chimney stent to preserve the left subclavian artery. None of the patients had a stroke after the procedure. Only one stent-graft was required for each patient. All the patients had primary technical success of the stent-graft deployment (100%). None of the patients had adverse characteristics to endovascular stenting such as calcium plaques, mural thrombus, small iliofemoral access arteries, or tortuosity of the aorta. Blood loss was minimal (<100 mLs) for all patients, except for patient #4 (300 mLs). Only patient #1 received 5,000 units of intravenous unfractionated heparin during deployment of stent-graft, but this did not result in any adverse hemorrhagic events. The remaining patients were not given heparin because they had suffered multiple other injuries and the surgeon did not want to induce coagulopathy in these patients. There were no early endograft-related morbidities such as graft migration, collapse, infection, or thrombosis. None of the patients had any postoperative cardiovascular, respiratory, renal, or cerebrovascular events. None of the patients suffered from paraplegia. There were no access site problems such as groin wound infections, bleeding, hematoma, or pseudoaneurysm formation. No paraplegia, bowel ischemia, or limb ischemia was observed. No intraoperative, in-hospital, or 30-day mortalities were observed. None of the patients required any further procedures for the thoracic aorta pseudoaneurysm. Mean follow-up was 18.8 months. Compliance to follow-up was reduced in two patients. Patient #2 had underlying psychiatric disorder and was noncompliant to his hospital appointments and was lost to follow-up after 12 months. Patient #3 was a domestic worker from Indonesia and chose to return home for further treatment, explaining the follow-up of only 1.5 months. In the remaining four patients, there were no documented late complications (Fig. 5).
125 180 110 105 120 135
5 16 17 18 12 20
Yes Yes Yes Yes Yes No
42 12 1.5 36 18 3
DISCUSSION Since the introduction of endovascular stenting of the thoracic aorta, there has been a paradigm shift toward using this less invasive method for treatment of thoracic aorta pathology such as aneurysmal disease, dissections, and atherosclerotic ulcers. This has been fuelled by the benefit gained by the patient in the form of lower morbidity and mortality rates5 and more economical utilization of hospital resources resulting from shorter hospital stays and usage of intensive care facilities associated with this technique. Trauma patients often have associated multiple injuries, usually requiring several operations. The traditional gold standard for traumatic thoracic aorta injury is open surgery, with its high mortality rates in the range of 15-30%.2,3 Avoiding a thoracotomy, cardiopulmonary bypass and aortic cross-clamping together with its morbid complications of prolonged ventilatory support, pulmonary complications, risk of renal failure, and paraplegia is certainly advantageous to the already critically ill patient. With multiple injuries and operations often required in trauma patients, the need for massive blood transfusion coupled with hypothermia common in this set of patients, concomitant coagulopathy remains a major concern. Systemic heparinization usually given before stent deployment can further worsen the coagulopathy. The role of heparin is to prevent thrombus formation in the aorta and iliac vessels around the endovascular deployment device and also in the important arteries arising from the aorta such as the visceral vessels. In our study, only one patient (16.7%) received intraoperative heparin, but there were no thromboembolic complications noted despite omission of heparin in most of our patients. Because most injuries occur at the aortic isthmus with limited landing zone for the stent-graft, left subclavian artery coverage may be inevitable. Several studies have shown the safety of left
610 Glenn Tan et al.
Fig. 5. Post-stenting with left subclavian artery preserved.
subclavian artery sacrifice,7-10 which we have further emphasized in our study in which 83.3% of our patients had uneventful sacrifice of the left subclavian artery. Although some investigators have reported complications from left subclavian artery coverage11 and the necessity for prophylactic carotidesubclavian bypass12 or creating a window in the stent-graft for the left subclavian artery ostium,13 we have not experienced the need for these extra procedures that will lengthen the operative time and may add further stress to an already compromised trauma patient. We do advocate selective vertebral angiogram during the procedure to assess the posterior circulation to the brain. This practice helped pick a patient in our cohort with an attenuated hypoplastic right vertebral artery and we opted to preserve the left subclavian artery circulation in this case. In cases when the left subclavian artery has to be sacrificed with potential reduction of blood flow to the posterior circulation, the adjunct maneuvers described by Rousseau et al.13 can be used or carotidesubclavian bypass operation should be performed before stent-graft deployment. CT angiogram is another option that can be used to define the anatomy of the vertebral arteries but will necessitate another journey to the CT scanner because most CT scans of the thorax do not include the vertebrobasilar arterial system. We prefer intraoperative selective vertebral angiogram before stent placement because it gives a more physiologic representation of the dominance of the posterior circulation and Circle of Willis and does not significantly increase the operative time and can potentially prevent a catastrophic posterior circulation stroke.
Annals of Vascular Surgery
There has been some controversy regarding the timing of intervention for traumatic thoracic aortic injuries with some advocating delayed treatment,13,14 whereas others have recommended immediate repair.15 In patients with traumatic aortic injury, blood flow is maintained in the disrupted aorta by the adventitia and surrounding mediastinal structures. This is a precarious situation with a potential for acute exsanguination into the thorax at any time. Some previous studies have suggested that repair can be delayed for patients who are stable with concomitant injuries, presumably owing to the stress of additional surgery.16,17 In our study, all patients had endovascular repair as soon as other more life-threatening injuries were treated and the patient was stable for transfer to the angiography suite. We have not experienced any major morbidity or mortality from early endovascular stent-grafting in our series despite the fact that all our patients had associated injuries, with many requiring multiple operations. Despite all the advantages of endovascular repair in thoracic aorta injuries, there remain some challenges to this technique. The typical trauma patient is younger than the average patient with degenerative aortic disease for whom the stent-grafts were initially designed. Furthermore, most injuries occur at the isthmus, thereby requiring for the stent-grafts to be placed across a tighter more angulated aortic arch. Younger subjects also have smaller caliber aorta and iliac access vessels, which may present difficulties in inserting the devices, stent-graft conformity to aortic wall, or obtaining an accuratesized stent-graft. These factors may lead to higher risk of endoleaks, stent collapse,18-20 or iliac artery avulsions.4,5 Asian patients are of smaller build and thus are perceived to have smaller aortas and access vessels, which may be a contraindication to endovascular approach to the management of traumatic thoracic aortic injuries. However, we have not encountered any problems with small iliofemoral arteries requiring access conduits or small aortas where a suitably sized stent-graft was not available. In our series, we did not need to use any uncovered bare stents, although we acknowledge their usefulness in certain situations. It can be deployed within the stent-grafts in the acute setting to allow better conformity to the native aortic curvature in cases when there is poor apposition of the proximal portion and also in delayed partial or complete collapse of the stent-grafts. However, there are concerns that bare stents may erode though the graft fabric over time or result in proximal aortic perforations and hence our reservations about their use.21
Vol. 25, No. 5, July 2011
Endovascular stent-grafts require long-term surveillance to assess for stent-graft durability and endoleaks, usually with regular CT scans. This results in large amounts of radiation exposure, especially undesirable in young patients. Only long-term results will demonstrate the robustness of the endovascular stents and provide a clearer direction or alternatives for follow-up.
CONCLUSION Endovascular stent-graft for traumatic thoracic aorta injury has been shown to be a feasible and safe method of repair. This is particularly advantageous in trauma patients who have multiple injuries, where avoiding a major thoracotomy can reduce morbidity for the patient. Left subclavian artery coverage by endovascular stent-graft to obtain adequate landing zone is also safe when intraoperative vertebral angiogram has shown adequate posterior circulation to the brain from the right vertebral artery. Omitting systemic heparinization during stent-graft deployment does not increase the risk of thromboembolic complications from the procedure and may be desirable in trauma patients to reduce the risk of coagulopathy. Our study demonstrates feasibility of endovascular management of thoracic aorta injury in Asian patients and did not encounter any problems with small access vessel and aorta size that is widely perceived in Asian patients because of their smaller build.
The authors thank Ms Karen Go and Ms Yen Teng Yeo for the use of information from the Trauma registry and Ms Wenzhao Sun for help in tracing the medical records. The authors would also like to thank the library services (Ms Cecilia James and Ms Norina Hamid) in TTSH for providing reference materials. REFERENCES 1. Parmley LF, Mattingly TW, Manion WC, et al. Non-penetrating traumatic injury of the aorta. Circulation 1958;17: 1086-1101. 2. Von Oppell UO, Dunne TT, De Groot MK, et al. Traumatic aortic rupture: twenty-year meta-analysis of mortality and risk of paraplegia. Ann Thorac Surg 1994;58:585. 3. Jamieson WR, Janusz MT, Gudas VM, et al. Traumatic rupture of the thoracic aorta: third decade of experience. Am J Surg 2002;183:571-575.
Management of blunt traumatic thoracic aorta injuries 611
4. Bent CL, Matson MB, Sobeh M, et al. Endovascular management of acute blunt traumatic thoracic aortic injury: a single center experience. J Vasc Surg 2007;46:920-927. 5. Stone DH, Brewster DC, Kwolek CJ, et al. Stent-graft versus open surgical repair of the thoracic aorta: mid term results. J Vasc Surg 2006;44:1188-1197. 6. Fujikawa T, Yukioka T, Ishimaru S, et al. Endovascular stent grafting for the treatment of blunt thoracic aortic injury. J Trauma 2001;50:223-229. 7. Sandroussi C, Waltham M, Hughes C, et al. Endovascular grafting of the thoracic aorta, an evolving therapy: tenyear experience in a single centre. ANZ J Surg 2007;77: 974-980. 8. Riesenman PJ, Farber MA, Mendes RR, et al. Endovascular repair of lesions involving the descending thoracic aorta. J Vasc Surg 2005;42:1063-1074. 9. Melnitchouk S, Pfammatter T, Kadner A, et al. Emergency stent-graft placement for haemorrhage control in acute thoracic aortic rupture. Eur J Cardiothorac Surg 2004;25: 1032-1038. 10. Moainie SL, Nechis DG, Gammie JS, et al. Endovascular stenting for traumatic aortic injury: an emerging new standard of care. Ann Thorac Surg 2008;85:1625-1630. 11. Riesenman PJ, Farber MA, Mendes RR, et al. Coverage of the left subclavian artery during thoracic endovascular aortic repair. J Vasc Surg 2007;45:90-94. 12. Leurs LJ, Bell R, Degrieck Y, et al. Endovascular treatment of thoracic aortic diseases: combined experience from the EUROSTAR and United Kingdom Thoracic Endograft registries. J Vasc Surg 2004;40:670-680. 13. Rousseau H, Soula P, Perreault P, et al. Delayed treatment of traumatic rupture of the thoracic aorta with endoluminal covered stent. Circulation 1999;99:458-504. 14. Pacini D, Angeli E, Fattori R, et al. Traumatic rupture of the thoracic aorta: ten-years of delayed management. J Thorac Cardiovasc Surg 2005;129:880-884. 15. Thompson CS, Rodriguez JA, Ramaiah VG, DiMugno L, Shafique S, Olsen D, Diethrich EB. Acute traumatic rupture of the thoracic aorta treated with endoluminal stent grafts. J Trauma 2002;52:1173-1177. 16. Walker WA, Pete JW. Medical management of acute traumatic rupture of thoracic aorta. Ann Thorac Surg 1990;50: 965-967. 17. Maggisano R, Nathens A, Alexandrova NA, Cina C, Boulanger B, McKenzie R, Harrison AW. Traumatic rupture of the thoracic aorta: should one always operate immediately? Ann Vasc Surg 1995;9:44-52. 18. Neschis DG, Moaine S, Gutta R, et al. Twenty consecutive cases of endografts repair of traumatic aortic disruption: lessons learned. J Vasc Surg 2007;45:487-492. 19. Idu MM, Reekers JA, Balm R, et al. Collapse of stent-graft following treatment of traumatic thoracic aortic rupture. J Endovasc Ther 2005;12:503-507. 20. Steinbauer MG, Stehr A, Pfister K, et al. Endovascular repair of proximal endograft collapse after treatment of thoracic aortic disease. J Vasc Surg 2006;43:609-612. 21. Malina M, Brunkwall J, Ivancev K, et al. Late aortic arch perforation by graft anchoring stent: complication of endovascular thoracic aneurysm exclusion. J Endovasc Surg 1998;5:274-277.