Endovascular approach to acute aortic trauma

Abstract

Traumatic thoracic aortic injury remains a major cause of death following motor vehicle accidents. Endovascular approaches have begun to supersede open repair, offering the hope of reduced morbidity and mortality. The available endovascular technology is associated with specific anatomic considerations and complications. This paper will review the current status of endovascular management of traumatic thoracic aortic injuries.

Key Words: Aorta; Complications; Outcomes; Traumatic; Endovascular

INTRODUCTION

The treatment of aortic rupture has significantly evolved since Parmley’s landmark 1958 paper[1]. Rather than angiography followed by immediate open repair, the use of computed tomography angiogram (CTA), early institution of β-blockade and pressure control to reduce the risk of early rupture, and advances in techniques including left heart bypass has allowed more selective management of patients[2-8]. In addition, increasing use of seat restraints appears to be associated with less severe aortic and associated injury, further enhancing survival[9]. Nevertheless, aortic rupture remains a leading cause of fatalities, particularly after motor vehicle accidents[10].

The incidence of traumatic aortic disruption varies from center to center. Looking at all trauma-related mortality, blunt aortic rupture may be second only to head injury as the primary cause of death. Despite this, accepting that there are approximately 8000 cases/year in the United States, and given that as many as 85% of victims die at the scene, then only 1000-1500 cases/year survive to be treated. In one of the largest contemporary series, 274 patients were admitted to 50 institutions over 2.5 years. If these cases were distributed evenly, the average institution would have seen only 2.2 cases/year[11]. In practice some centers may manage 8-15 cases/year, while the majority may encounter 1-2 at most.

Amidst this era of improved operative outcomes and medical management, endovascular stent grafts have become a possible third option. In North America, the concept of thoracic endografting, as an extension of abdominal endograft technology, was greatly stimulated by the Stanford group[12,13]. Their initial primary interest, and indeed the bulk of the literature since, was with atherosclerotic aneurysms. We now know that endografting is an attractive option that can avoid the morbidity of a thoracotomy in patients with multiple injuries, and that it appears to reduce the risk of paralysis[14-18]. Some centers have seen dramatic improvements in outcome using the endovascular approach[19-21]. In North America, currently 2/3 of blunt aortic injury are managed endovascularly[22]. As with all invasive procedures, there are specific complications and anatomic considerations that need to be incorporated into the planning of endovascular treatments of traumatic thoracic injuries. The purpose of this paper is to compare outcomes of endovascular approaches to open repair, review pertinent anatomy, imaging techniques and approaches, and to discuss complications and their management.

OUTCOMES OF ENDOVASCULAR STENTS UTILIZED IN THE TRAUMA SETTING

A number of series have been published which support the notion that endovascular stents, in the setting of traumatic aortic disruption, have low mortality (predominantly related to associated injuries) and essentially no risk of post-procedure paralysis. When reviewing these data, it is important to consider the span of time in which the experience was accrued (as stent technology has changed significantly over the past few years), recognize the difference between the acute (whether defined as within 24 h of injury or a longer period) vs chronic, and to consider what the indications for stent grafting and contra-indications to open repair were. We have selected those series published between 2002-2006 (11 reports), comprising 167 patients, the youngest being 16 years of age[5,15,16,23-31]. These series ranged from 5 to 30 cases, over time periods ranging from 1 to 7 years. Average follow up among the 10 series with at least one year follow up was 24 mo. Virtually all stents were industry made, although they varied between ‘dedicated” thoracic stents to a variety of cuff extenders. There were 7 (4%) deaths, 2 of which were procedure-related (one collapse and rupture, one stroke). Type I endoleak occurred in 8 instances (4.7%), 2 healing spontaneously, 6 requiring further stenting and/or balloon dilation. There were 2 iliac ruptures reported, and 3 (1.7%) cases of acute stent collapse requiring operative intervention. There were 2 cases of non-fatal stroke and 1 of brachial occlusion requiring thrombectomy. These experiences, in combination with an overall major non-fatal complication rate (excluding endoleak) of 4.3% and mortality of 3.6% justifies the excitement that endovascular approaches have provoked in the management of traumatic aortic rupture. Hershberger and colleagues summarized the outcome of 811 patients among 109 publications. Overall technical success rate was 93.6% and mortality 9.5% (72 patients). Eight (1%) of the deaths were directly related to complications of the endovascular procedure, of which two were due to stent collapse, one of aortic enteric fistula and others to continued hemorrhage despite stent placement[17].

In addition, endografts may also be used not as a definitive repair, but in complicated cases as a “bridge” to definitive treatment in selected patients who are not suitable candidates for either operative repair or medical management[23,32].

COMPARISON BETWEEN ENDOVASCULAR AND OPEN REPAIR

It is inherently difficult to retrospectively compare two techniques that are not necessarily applied to the same patient population with respect to risk assessment, operative experience and institutional biases. Each center has sufficiently different patient populations and management strategies to make it difficult to make broad generalizations based on an individual study. Again, recognizing that this is not a complete review of all available works, we reviewed five papers, published between 2004 and 2006, that specifically compared outcomes within their respective institutions between the 2 approaches[5,9,16,24,33]. A total of 108 patients underwent open repair. There were 15 deaths (14%) and 4 (4%) cases of new post-operative paralysis. Ninety-three patients underwent endovascular repair, 9 (9%) died and no paralysis/paraplegia was observed. Only one death was procedure-related among the stent graft group (acute stent collapse).

An American Association for the Surgery of Trauma multicenter study compared the outcomes of endovascular vs open repair. One hundred and twenty-five patients underwent endovascular repair with 20% developing stent-related complications. Of 18 patients (14%) with endoleak, 6 required open repair. Paralysis/paraplegia occurred in 0.8% compared to 2.9% among the 68 patients who underwent open repair. When adjusted for extrathoracic injuries, hypotension and age, endovascular repair had a significantly lower mortality compared to open repair[22].

Tang and associates presented the results of a meta-analysis comparing the 30-d outcomes between 278 aortic ruptures managed surgically vs 355 managed by endovascular means. There were no significant differences in injury severity or age between the groups. The endovascular group had significantly lower mortality (7.6% vs 15.2%, P = 0.008), paraplegia (0% vs 5.5%, P < 0.0001) and stroke (0.81% vs 5.1%, P = 0.003) compared to the open surgical repair cohort[34].

These small comparisons demonstrate that when feasible, at least in the population over 18 to 20 years, endovascular repair appears to be associated with a markedly lower paralysis rate than open repair, and length of stay may be reduced compared to open repair, however, acute outcome is probably more related to overall injury severity than approach. It further supports the notion that across many institutions the acute outcomes are as good as open repair, and probably better, but that within institutions with good experience of both approaches outcomes are comparable.

ENDOVASCULAR REPAIR VS MEDICAL MANAGEMENT

Institution of strict “anti-impulse” therapy should occur once the diagnosis of aortic rupture is suspected[5,35]. The “ideal” blood pressure depends upon the patient’s age and presenting blood pressure. Until recently, the goal was a systolic blood pressure of < 120 mmHg and/or mean arterial pressure < 60-70 mmHg. More recently it has been argued that a blood pressure of “less than what the patient was admitted with” may be more appropriate[10,36]. When strict blood pressure control is implemented, in stable patients, the risk of rupture in the first week may be as low as 5% or less[10]. Some series have noted improved outcomes with both delayed open and endovascular repair, but this may reflect some selection bias[5]. Nevertheless, delayed repair has been shown to be associated with improved outcomes, both for open and endovascular approaches[37].

Reasons for delaying operative intervention include severe head injury, blunt cardiac injury, solid organ injury and/or acute lung injury[7,38]. In these instances, we have favored serial surveillance imaging (usually with CTA) every 48 h for 7-10 d, to detect any change in the size or character of the lesion[2]. While the natural history of residual pseudoaneurysms appears to follow those of non-traumatic atherosclerotic aneurysms, these lesions should not, especially in young patients, be considered completely benign, and we favor early intervention as soon as medically stable.

Tight medical control of blood pressure may not be possible in every case. Many patients require other interventions, and monitoring and controlling blood pressure during these can be difficult. There are some hazards including renal and splanchnic insufficiency, and secondary brain injury especially in the setting of increased cranial pressure[32,39]. Although there is some controversy as to the value of driving up cerebral perfusion pressure, or assuming that an increased pressure translates to improved cerebral perfusion, there is general consensus that “high” pressure is associated with a lower risk of secondary brain injury[40-43]. Thus, closed head injury associated with evidence of increased intracranial pressure [by computed tomography (CT) and/or intracranial pressure monitoring] may actually mandate operative or endovascular repair. One significant advantage of endovascular repair over both operative and non-operative management is that after the stent is placed, in most cases it is possible to allow blood pressure to normalize, or even increase without the risk of bleeding or rupture. We caution that the risk of rupture, even with serial CT angiography and tight hemodynamic control, is not zero. Endovascular stents may be ideally utilized exactly in these patients who cannot undergo open operative repair because of significant co-morbidities.

The extent of injury may also impact the choice between medical and endovascular management. Minor aortic injuries, involving only small intimal defects, often heal without residual defects[4,44]. However, even small lesions can go on to rupture if blood pressure is not controlled[2]. Thus, if blood pressure can be reasonably controlled, and there are no contraindications to medical management, small intimal defects should be managed medically with close follow up. Even small pseudoaneurysms, in some cases, have healed[2]. Thus, while endovascular management appears to be an ideal solution in patients with significant co-morbidities, and who are judged to be at too high risk for prolonged medical management, it is not clear that this approach is better than medical management in patients with minimal injuries. One simple guideline is that if the lesion is minimal enough such that one would not consider open operative repair, than one should not rush to endovascular repair either.

ENDO-GRAFTS CURRENTLY AVAILABLE

The characteristics of an endograft designed for the thoracic aorta, as opposed to the abdominal aorta, include a long enough delivery system to reach the distal arch from the femoral artery, and flexibility to accommodate the curvature of the arch. There are variations between different types of graft in how they deploy, whether or not proximal and/or distal components are bare, whether or not they contain hooks, and how they are actually released from the constraining devices. In general, there has been a shift away from deploying devices in aortic trauma (and type B dissection) which rely on uncovered proximal landing zones because of concerns of aortic perforation[45]. An important consideration is that the average young trauma patient has an aortic diameter in the 20 mm range, which is too small for these devices which were designed for older, atherosclerotic, aortas[46]. Secondly, the use of endografts in the acute trauma setting is considered to be “off label” by the Federal Drug Administration (FDA) outside of trials. Currently, there is only one device approved by the FDA for use in the thoracic aorta, and then only for atherosclerotic aneurysms.

The Gore-TAG^® (W.L. Gore & Associates, Flagstaff, AZ, USA) is constructed from ePTFE, wrapped in self-expanding nitinol stents which have a sutureless attachment. These flares can be placed partially across an orifice without occluding the orifice. The device is introduced through a 30-cm sheath, the outer diameter varying depending on the diameter of the graft. The delivery catheter is fairly flexible and 100 cm in length. The device is released by pulling a rip-cord that unlaces the constraining system with rapid (less than a second) release starting in the center. This central deployment prevents a “wind sock” effect and there is rarely any need to critically lower blood pressure during deployment to beyond normal ranges to prevent distal migration. When advancing the catheter, there can be built-up tension, and so the device should be advanced just beyond the proposed landing zone and then brought back into the proper alignment just before deployment. One characteristic of the TAG device is that it rarely if ever jumps forward. Rather, it may drop a millimeter or two back (distally). In appropriately sized patients, the TAG offers an exceptionally good approach due to its flexibility and rapid deployment characteristics[9,47].

Two other companies, Cook and Medtronic, have also been developing and refining endografts for use in the thoracic aorta, and the early iterations, like the Gore, have proven very useful in trauma patients in Europe, Canada and Australia[26,27,30,33,48,49].

The Cook-Zenith TX2^® (Cook, Bloomington, IN, USA) is made of woven polyester fabric over stainless steel Z-stents. The TX2 has a proximal form, which has no bare stents but does have 5 mm long barbs. The distal component does have a bare metal distal segment. The proximal device can be uniform or tapered. The diameters range from 28 to 42 mm (in 2 mm increments) and the lengths vary from 120 to 216 mm. The tapered form narrows at the third Z-stent and the distal diameter is 4 mm less than the proximal. The sheaths are pre-curved and 75 cm long. A 20 Fr sheath (OD 23 Fr) is used for 28-34 mm stents, and a 22 Fr sheath (OD 25 Fr) for 38-42 mm endografts. Although it would be rare to need it in the trauma setting, the TX2 is designed with a distal component, to allow modification of the distal portion of the graft to fit the anatomic requirements of the distal landing zone. This portion has an uncovered bare area. The device is unsheathed, but three trigger wires continue to constrain the device, reducing any “wind sock” effect, and allowing final careful positioning before complete release.

The Medtronic Talent^® Thoracic graft has recently been released in Europe in a modified form, the Valiant^® (Medtronic, Santa Rosa, CA, USA). The Valiant differs in that while it is still made of a low-profile polyester monofilament material, the nitinol stents are on the outside, and with two proximal configurations, one covered and the other with 12 mm flexible bare stents to allow adaptation to the aortic curvature and greater fixation. In addition, there has been improved conformability of the distal end which may be covered or have a bare extension as well (FreeFlo). The Medtronic company tends to recommend the FreeFlo model over the covered proximal stent as the experience, at least with non-traumatic aortic pathology, is that the increased flexibility allows the uncovered portion to conform well with the aorta, leading to better fixation. Like the Cook-TX2, the Valiant comes in both a proximal and distal (if needed) component, although with short lesions, the proximal component alone is often sufficient. Both come in straight or tapered versions. The delivery system involves an unsheathing maneuver. The device allows retraction and repositioning to permit exact placement. The Talent does come in a slightly smaller model than the Valiant, 22 mm, which may be useful in aortas with diameters of 18-19 mm diameter.

Because of the size constraints in the “typical” trauma patient, and because the arch is often acutely angled, some groups have used abdominal aortic cuff extenders rather than dedicated thoracic aortic stent grafts[28,31,50]. These are not only smaller, but may actually fit the aortic configuration of transected aortas better, albeit at the expense of needing multiple grafts, of an increased risk of Type III endoleak, and of having to use the shorter delivery system that is designed for the infra-renal aorta[51]. The AneuRx^® (Medtronic, Santa Rosa, CA, USA) cuff extenders range from 20-28 mm diameter, are 4 cm long, and are delivered through a 21 Fr catheter which is 59 cm long. The Gore-Excluder aortic cuffs^® (W.L. Gore & Associates, Flagstaff, AZ, USA) are 3.3 cm long, have diameters of 23, 26 and 28.5 mm, and can be delivered via an 18 (for the 23 mm device) or 20 Fr sheath which is 61 cm long. On occasion a contra-lateral limb or iliac extender from the abdominal aortic set may fit the specific anatomic requirements.

Borsa and colleagues reviewed the anatomical features of 50 trauma patients with documented traumatic aortic rupture who underwent angiography[46]. The average age was 37 years. The mean distance between the superior aspect of the injury and the origin of the left subclavian artery was 5.8 mm along the inner curve and 14.9 mm along the outer curve. The mean diameter adjacent to the injury was 19.2 mm. The mean degree of curvature from the left subclavian artery and the inferior aspect of the injury was 54.0°. The mean length of the injury was 17 mm along the inner wall and 26 mm along the outer wall. In addition, the injury involved 1/4 of the circumference in 44%, 1/2 in 16%, 3/4 in 18% and the complete circumference in 22%. Finally, 28% had a bovine arch and 10% had aberrant origin of the left vertebral artery from the arch proximal to the left subclavian artery. The implications of this study are that given currently available technology, at least half of patients will require coverage of the left subclavian artery to obtain an even minimally acceptable landing zone, and at least half or more of patients will not be candidates for the TAG but will require smaller diameter devices.

ANATOMICAL CONSIDERATIONS

Anatomic considerations are listed in Table 1. The initial factor to be determined is the diameter of the proximal and distal landing zones. Measurements are taken from inner wall to inner wall. The diameter can be difficult to assess in the distal arch, but one method is to measure the transverse diameter at its widest point. In younger adults the aorta is relatively uniform through the arch. Three-dimensional reconstruction and measurements of the descending aorta just at the level of the main pulmonary artery can help confirm estimations of the proximal landing zone diameter. The stent graft diameter should be approximately 10%-14% larger than the landing zone diameters.

Table 1 Anatomic considerations.

Anatomical features to consider	Implications
Diameter of proximal and distal landing zones	Determines size of endograft that can/should be utilized
Distance from lesion to origin of left subclavian artery	Will obtaining an adequate landing zone require coverage of the left subclavian artery?
Distance from lesion to origin of left common carotid artery	If required, is there room to land distal to the origin of the left common carotid artery? Will there be room, if needed, to clamp distal to the origin or will circulatory arrest be needed if subsequent operative repair is needed?
Degree of curvature across the proximal landing zone	Is there a high likelihood that to avoid malposition along the inner curvature that the graft will have to placed more proximally?
Quality of the aorta	Is there significant thrombus and/or calcification that would pose a risk of stroke or Type I endoleak?
Quality of access vessels	Is the diameter sufficient to permit the required sheath? Are there more proximal calcifications and/or tortuosity that might prevent safe passage of the sheath?
Distance from proposed access vessel to the lesion	Does the system being used have sufficient length to reach the proposed site?
Length of the injury	If using cuffs, how many may be required to ensure fixation
Vascular anomalies	Anomalous origin of left vertebral artery? Patent LIMA graft? Aberrant origin of right subclavian artery?

Because most transections occur in proximity to the left subclavian, the next decision is whether to cover the subclavian origin or not. Deploying the graft within the distal curve (“grey zone”) of the arch may result in partial occlusion of the aorta, increase the risk of stent migration and/or collapse, and result in an endoleak. The origin of the left subclavian artery can be marked and both the “drop down” distance measured based on the number of cuts of the CT, and the transverse measurement should be estimated. Ideally at least 2 cm of proximal and distal landing zone is recommended, but in some younger patients, with otherwise normal aortas, 1-cm has proven acceptable.

The proximal aortic landing zone and arch needs to be reviewed to assess for the presence of significant thrombus and/or calcification. Focal areas of calcification can result in elevating a “lip” of endograft, resulting in increased risk of proximal endoleak. Significant thrombus increases the risk of stroke and distal embolization.

The length of the aorta that needs to be covered is based on a minimum of a 2-cm landing zone. If using cuff extenders, usually three will be required to provide stability[31,51].

Having chosen the optimal size and type of endograft, the next consideration is the length of the delivery device. Commercial thoracic endograft delivery systems have sufficient length to reach the entire thoracic aorta from the femorals, but cuff extenders have delivery systems of only 61-cm that may not reach from the groin to the arch. Additionally, the quality and diameter of the proposed access arteries need to be evaluated. The diameter, angulation and degree of calcification should be determined. Calcifications are better seen with non-contrast images. A non-calcified vessel may tolerate a slightly oversized sheath, but a severely calcified vessel may not accept a sheath that would be predicted to fit based on size criteria alone.

PROCEDURE

Imaging

Because the majority of traumatic injuries are located within 2 cm of the origin of the left subclavian artery, they are in proximity to the arch. To visualize the anatomy properly, the most “open” view of the arch should be obtained. In younger patients, usually a left anterior oblique view at the 60-90° range is required. The optimal angle can be estimated from the CTA. During angiography, using a multi-marker pig tail catheter, one can be confident that the optimal view has been obtained when the marks are equidistant throughout the image. This will facilitate the most accurate positioning of the proximal graft and determine the length of coverage needed.

Accessing either the brachial artery for angiograms and or marking with a wire has been advocated by some groups[23]. This can augment precise localization of the graft relative to the origin of the left subclavian. Alternatively, in patients with severe acute arch angulation, placing a stiff wire from the right side may permit “straightening” during deployment[23]. We have not used the right brachial approach in the trauma setting, because of concerns of dissection and/or stroke.

Patient positioning is critical. The arms can obscure visualization of the relevant anatomy. Options include abducting them above the head or placing a roll under the left side to gain some 10°. If it is anticipated that brachial catheterization will be needed, the arm should be prepped out. Prior to prepping and draping the patient, fluoroscopy in key anticipated views should be quickly carried out to make sure that there are no leads in the way, and that the arch can be easily seen without obstruction.

General conduct of the procedure

Endovascular stent-grafting is an operative procedure, and should be performed in a room designed with this in mind, including sterile set up and laminar flow. It can be performed in a standard operating suite, using portable C-arm, but the image quality offered by multi-purpose suites with fixed fluoroscopy units provides significantly better resolution. General anesthesia with a single lumen endotracheal tube is sufficient, but has been done under sedation with local anesthesia in the rare stable patient. A right radial arterial line for blood pressure monitoring is optimal because the left subclavian may be covered or the left brachial artery needed for access. Depending on the angle needed to best visualize the arch, the arms may need to be elevated above the head or a bump placed under the left side. Prior to prepping, fluoroscopy should be performed to ensure that no tubes, lines or boney structures obscure the field. We have found it helpful to place a 5- or 6-Fr sheath and multi-marker pigtail catheter through the femoral artery opposite to the side through which the device will be deployed. This permits final angiographic marking with the non-deployed stent-graft in place. It also allows access for completion pelvic angiogram (although this can be done through the sheath) and access for a balloon occluder in the event of iliac rupture. Great care should be taken to avoid air or atheroembolic. In obese patients, or those with a weak pulse, ultrasound is very helpful in accessing the common femoral artery, and avoiding areas of significant plaque. After arterial access is established, 5000 U of Heparin is administered, although it is possible to avoid heparinization if there is concern for bleeding complications[31]. A 5- or 6-Fr pigtail is advanced over a floppy 3-J wire, and then the wire is exchanged for a stiffer wire (such as an Amplatz or Lunderquist). The location of the rear of the wire should be marked on the table, so that any advancement or withdrawal can be detected. The delivery sheath is then advanced under constant fluoroscopic monitoring, usually in the AP view. Ideally it should lie in the distal aorta. Once the device is in the approximate desired location, the image intensifier is rotated to the ideal angle, and angiography performed with respirations suspended. Usually 20 mL of contrast/s for 2 s is adequate. A “road map” can be obtained and/or the landmarks marked with a felt pen. At this point the stent graft is deployed. We prefer to withdraw the pigtail used for angiogram below the stent prior to deployment. With most devices, deployment is rapid, and adenosine arrest is not needed. Hypertension should be avoided. Using cuff extenders, we do prefer adenosine, at least for the first graft, to avoid distal displacement. When deploying cuffs, in general we have started distally, which we believe helps stabilize subsequent more proximal devices. Once the device(s) is/are deployed, gentle ballooning, starting distally, is performed. When using multiple devices, the ballooning should proceed proximally, covering all areas of overlap. When deploying cuffs, the balloons do occlude the aorta, and even the tri-fasicular balloon when deployed across the curvature of the aorta, can result in acute proximal hypertension and grab the stent-graft. Ideally, in trauma cases, ballooning can be avoided altogether. If needed, rapid inflation and deflation with 1/3 contrats-2/3 saline solution prevents significant hypertension. Vigorous over ballooning should be avoided. The balloon should be withdrawn below the stent, maintaining wire access, and a completion angiogram performed. If there is a small endoleak or if there is a lip of the graft not being apposed to the inner curve of the aorta, repeat ballooning should be performed. If this fails to eradicate the issue, extending the graft with an additional module should be considered. Larger balloons can catch along the edge of the delivery sheath, but gently pulling the balloon and sheath beyond the aorto-iliac junction usually allows the balloon catheter to straighten out and be removed. The sheath is then gently withdrawn to as close to the inguinal ligament as possible (in the case of femoral access) and completion pelvic angiogram via the contra-lateral pig tail (or sheath if one was not placed) performed.

Access

Access vessel choice depends upon the size of sheath required for the chosen endograft, length of the delivery system, quality and diameter of the arteries, and clinical setting. Most trauma patients have healthy vasculature, and thus slight mismatch can be tolerated between sheath size and femoral diameter as long as there is no “tugging” and the sheath advances easily under fluoroscopy. If there is any concern, a contra-lateral sheath should be placed so that balloon occlusion can be used in the event of an iliac rupture during sheath removal. A variety of devices have been utilized for percutaneous closures with good results[31]. These devices are designed to place intra-arterial sutures that are then tied down at the end of the case[52]. The most common device used, the Perclose^® (Perclose/Abbott Labs, Redwood, CA, USA), places two sutures at right angles to each other. They are placed over a wire at the start of the case, and in general, most clinicians prefer to tie them down over a glide wire at the end so that if there is a failure of closure a sheath can be re-introduced and a cut down performed. There is an approximately 10%-20% failure rate with these devices, which are designed generally for sheaths smaller than those used during endografting procedures. We have found that accessing the femoral artery under ultrasound guidance permits us to be sure we are indeed in the common femoral artery, and are anterior to avoid areas of calcification or thrombus. When using sheaths required for dedicated thoracic stent grafts, we use two placed at 90° to each other. Complications include infection, arterial thrombosis, and pseudoaneurysm, and collectively occur at a rate of about 5%. There are no significant differences between the devices[53-57]. Percutaneous approaches can be performed as quickly as a cut down if the operators have experience but in emergency situations or if there are any concerns regarding the quality of the femoral artery (calcification, size etc.) then a cut down is safer. The groin incision can be oblique, at or above the groin skin crease, or a longitudinal incision. If there are extensive calcifications, endografts such as the TAG have been advanced without using the sheath, but this is not recommended because the graft can catch on an edge and deploy prematurely or be damaged. When withdrawing the sheath at the end of the case, particularly if a percutaneous approach has been used, it is critical that the blood pressure be monitored for 2-3 min as any acute drop is pathognomonic of an iliac rupture.

Retroperitoneal iliac exposure may be required if using cuff extenders and the device is not long enough to reach the location of the tear and/or if the femoral arteries are too small and/or calcified to use. If there has been pelvic trauma, using the side with the least hematoma is desirable. The common iliac can be accessed directly or a 10 mm silo graft is anastomosed end-to-side. If the pelvis is deep, to avoid a problem with angulation, the silo graft can be tunneled through the lower abdominal soft tissue or indeed through the femoral canal to the groin. Patients who have had prior aorto-iliac grafts can be challenging because the iliacs are often imbedded in scar tissue. The ureter should always be mobilized anteriorly, avoiding dissection on both sides to prevent devascularization. In the vast majority of cases, the best that can be achieved is that enough dissection of the iliac limb of the graft allows application of a partial occlusion clamp or direct graft puncture. Having completed the procedure, whether anastomosing to graft or native vessel, the conduit is simply truncated and over sewn as a patch. In some circumstances, it may be advisable to convert the conduit to an ilio-femoral artery bypass. This allows a relatively easier access route for later percutaneous interventions should the need arise.

Some patients may already have an open abdomen, and in these cases direct infra-renal aortic access can be used[51]. This would not be a good choice if there has been visceral spillage.

ILLUSTRATIVE CASE STUDY

An 84-year-old male presented after a motor vehicle accident with multiple deep facial lacerations, right hip dislocation, and traumatic rupture of the thoracic aorta. His history was significant for extensive coronary artery disease and he had electrocardiographic and laboratory evidence for ischemia vs blunt cardiac injury (new S-T depressions and elevated cardiac troponin I levels). Because of his co-morbidities we elected to repair the injury in the hybrid operating suite. The injury was very close to the origin of the subclavian artery (Figure 1A). To mark the origin of the subclavian, and because of the short distance across an acutely angled neck, brachial access was obtained to permit exact measurement and allow access for stenting if required (Figure 1B). After deployment, a Type I endoleak was noted (Figure 1C). Gentle ballooning to push the inferior lip of the graft into better apposition with the inner curve was successful (Figure 1D). Completion angiogram demonstrated retrograde filling of the subclavian without type II endoleak, thus obviating the need for a kissing stent via the subclavian artery or carotid-subclavian bypass. At the end of the procedure his facial injuries were repaired and the hip re-located. The patient was extubated the next morning but spent 2 wk in hospital because of his hip dislocation. At 9 mo follow up he is doing well with no evidence of endoleak.

Figure 1 An 84-year-old male’s injury was repaired in the hybrid operating suite. A: Traumatic injury. There is one centimeter along the lesser curve in a straight portion of the aorta that could serve as a landing zone, but much less along the outer curve (straight line). The “no-man’s land” is indicated by the stippled line, and would also act to destabilize the endograft if such a short landing zone was accepted. Thus, we felt subclavian coverage was required; B: Using the pig tail as a marker the exact location of the subclavian artery is marked during all maneuvers; C: After deployment of a Gore-TAG^® 31 mm × 15 cm (W.L. Gore & Associates, Flagstaff, AZ, USA). This length was chosen to gain better distal stability because of the acute aortic curvature. A Type I endoleak is noted. The proximal portion of the graft is in the “proximal no-man’s land” of the ascending aorta and does not appose well against the inner curve here; D: After gentle ballooning, the endoleak is obliterated and although a lip still remains (arrow) the graft has improved apposition.

FOLLOW UP

The protocols for follow up are based on the various clinical trials designed predominantly to evaluate thoracic endografting for atherosclerotic aneurysms. Typical guidelines include CT angiography at 48 h, discharge, 1-, 6- and 12-mo and then annually. These protocols are designed to detect graft collapse, migration or persistent endoleak with aneurysmal growth. To a large extent, these guidelines were laid out because the cases involved patients with diseased landing zones with a potential for ongoing dilation of the aorta. Obvious concerns include following patients with renal insufficiency, as well as the burden of a large number of radiation exposures. Patients with renal insufficiency can be surveyed with intravascular ultrasound, trans-esophageal echo, magnetic resonance or even CT without contrast. The primary concern is whether or not there is pseudoaneurysm regression or growth. Simple chest radiography can detect stent deformation or migration. For aortic transaction cases we tend to obtain a CT angiography at 48 h, at 1 mo, at one year and then follow with chest radiographs. When obtaining a CT angiogram, it is important to make sure that the study is performed in a uniform manner: triphasic with unenhanced, enhanced and delayed images. Indwelling pressure transducers, placed at the time of abdominal aortic endografting, continue to be studied and may, at least in select patients, provide a useful alternative to serial imaging[58].

We have not used antiplatelet agents for thoracic aortic stent grafting. However, we treat these like any other implant, and recommend antibiotic prophylaxis for any invasive procedure (such as dental work).

COMPLICATIONS

Endoleak and endotension

In brief, endoleak can be categorized as Type I (leak around the proximal, A, or distal, B, ends of the graft), II (leak form an artery feeding into the aneurysm sac), III (leak between components) or IV (failure of graft integrity). Typically, distal Type I endoleaks are rare in the trauma setting. These generally are found in patients with extensive atheromatous disease and dilated, short distal landing zones. Proximal Type I endoleaks occur in approximately 5% of cases[17,22]. Persistent Type I endoleak is associated with a risk of late rupture[48]. The predominant mechanism in trauma patients is the combination of a short landing zone and lack of apposition along the inner curvature of the arch. Gentle ballooning should be tried first. If this is not sufficient, then extending proximally with another graft should follow[27,59]. Type I leaks that are visualized only on delayed images immediately following deployment may resolve following heparin reversal. We assess these at 48 h with a repeat CT angiogram. Blood pressure should be controlled with β-blockade during this period. Proximal Type I endoleaks found on follow up imaging can usually be managed by repeat interventions. Significant leaks seen at the time of implant or at follow up, that do not respond to further ballooning or extension should undergo operative repair. There have been attempts to coil embolize the pseudoaneurysm in the hope that thrombosis will ensue, but clinically and experimentally this does not appear to reduce the endotension and is associated with a risk of late rupture. Most Type I endoleaks occur within 30 d, but occasionally can be found up to 2 years later, reinforcing the need for strict surveillance.

Type II endoleaks should be managed based on whether or not the left subclavian is the source. If it is, coil occlusion of the subclavian, carotid to subclavian bypass with proximal ligation or carotid subclavian to carotid transposition should be performed. Left subclavian arterial causes of Type II endoleak are less common in the trauma setting than the more typical atherosclerotic aneurysm case where there is circumferential dilation and the subclavian is more likely to feed into the aneurysm. Type II endoleaks believed to be secondary to patent bronchial or intercostal arteries are more problematic, but again are less problematic in the trauma setting as there are fewer branches in the proximal descending thoracic aorta. Some investigators believe that these are more benign than in the setting of abdominal endografting, and that in the majority of cases they will seal spontaneously[60]. Rarely a branch vessel can be accessed and coiled using micro-catheter techniques.

Type III endoleaks represent a leak between endograft components, and thus would be expected to be seen predominantly when cuff extenders are used, particularly if deployed along the curvature of the distal arch/proximal descending aorta. These are usually seen at the time of implant, and are managed by repeat ballooning and/or deploying another cuff across the site of the leak.

Type IV endoleak has not been described to our knowledge in the trauma population but implies a general failure of the graft material. Endotension refers to persistent pressure within an aneurysm or pseudoaneurysm without a documented endoleak. Clearly, if there is a source that can be defined it should be addressed. If no source is identified, but there is continued aneurysm growth, most surgeons would opt for operative repair rather than re-endografting. There is debate about whether absence of regression without growth represents evidence of endotension. Regardless, this does not appear to be a concern in the majority of trauma patients, and our own bias is that if the pseudoaneurysm is completely thrombosed following endografting, then routine follow up is all that is required.

Stent graft collapse

This is a catastrophic complication that can occur immediately, or within the first 48 h, but it has been seen up to 3 mo post procedure[61]. It is felt that this represents a combination of graft oversizing and a lack of apposition along the inner curve of the aorta[17]. In young hyperdynamic aortas, with their degree of pliability, the force of the cardiac ejection that hits the underside of the graft causes collapse of the graft. This usually leads to immediate aortic occlusion and possibly rupture. If this occurs post-implant, the patient will develop signs of acute coarctation, and the rapid onset of paralysis and renal failure can occur. This may not be immediately apparent if the patient is still on the ventilator and sedated. Alternatively, the collapse may only occur in systole, and require fluoroscopy for detection[62]. Prevention includes very accurate sizing, choosing a graft that approximates a 10% oversizing rather than 20%. It is also important to plan pre- and intra-operatively to avoid landing in the “no man’s land” of the aorta. If the proximal portion of the graft is not apposed or at least close to the inner curve, particularly if there is only a short zone of apposition, perhaps less than 50%, then options include extending the graft proximally and/or repeat ballooning[61]. Uncovered bare metal stents deployed within the stent graft have also been used both acutely and when collapse occurs in a delayed fashion. There has been some concern that these bare stents may either erode over time through the graft fabric or create proximal aortic perforations[45]. There is not enough data to determine the real risk of this occurring, but theoretically a short bare stent will conform more closely to the aortic curvature than a bare proximal portion that is secured to an endograft and has reduced flexibility. Across the country there have been numerous anecdotal reports of bare stent extenders being used for proximal partial or complete collapse with good short term results. There is growing consensuses that perhaps cuff extenders, which may be deployed sequentially and thus fit the curvature of the aorta better, may prove to perform better than longer thoracic stent grafts in patients with aortic diameters smaller than 24 mm.

If stent graft collapse occurs post-operatively, it can often be detected by plain chest radiography, or by non-contrast CT. Immediate intervention is required. If complete collapse has occurred, explanting and operative repair is prudent, but ballooning and extending the device with a bare stent has been used with success[31]. Anecdotally, axillary-femoral-femoral bypass has been used as a temporizing measure, but ultimately the stent must be removed.

Subclavian occlusion

Coverage of the left subclavian artery origin with subclavian to carotid bypass or transposition raises the question of arm ischemia, vertebral-basilar insufficiency and/or type II endoleak. Critical arm ischemia is rare, affecting less than 2% of patients, and if it occurs can be managed electively in most cases[15,47,49,63,64]. Type II endoleak arising by back flow into the pseudoaneurysm is also uncommon as most tears arise from the inner curve. Should a Type II endoleak occur, or if there is concern regarding this prior to the procedure, the left brachial artery can be accessed and once the graft is deployed, the subclavian can be coiled or closed with a peripheral closure device[65]. Vertebral steal phenomenon can also be addressed electively[66]. Patients with patent left internal mammary grafts should undergo carotid-subclavian bypass prior to left subclavian coverage[47,66]. If the proximal landing zone is felt to encroach upon the origin of the left subclavian, but complete coverage is not required, it is possible to access the left brachial artery and leave a wire in the arch, which allows precise placement of the device and can permit stenting of the subclavian origin if narrowing occurs[27].

Stroke

Stroke remains the “Achilles heel” of thoracic endografting, occurring at a rate of 3%-5% of cases[67]. In the trauma population the incidence is lower, at 0%-2% depending on the series[17,22,30]. Occlusion of a critical left vertebral artery is one cause. There is some data that an aberrant vertebral artery arising from the arch directly may be more likely to be a major source of posterior circulation, and in these cases we are more apt to perform vertebral transfer first. In older patients, arch thrombus is a concern for its embolic risk. Manipulation with the stiff wire across the arch can cause embolization. This has not been demonstrated in the trauma population to our knowledge, but should it happen, theoretical options include immediate cerebral angiography and if there is a definitive lesion, intervention with catheter-based techniques. Finally, inadvertent guide wire advancement into the carotid, or air embolism because of a loose connection, can result in cerebrovascular events. Whenever a catheter is proximal to the left subclavian artery, meticulous de-airing of the catheter and equipment is mandatory. Wire position should be monitored and stabilized at all times. Never advance a catheter or wire blindly.

Impeding the flow to the left vertebral may pose a risk of posterior cerebellar circulatory insufficiency or stroke. In our experience, this has never happened in the younger population, but we are concerned in older patients with diffuse vascular disease. Assessing cerebral circulation is clearly difficult under emergent conditions. Anatomic assessments can be made by CT-angiography or MRA of the head and neck, or cerebral angiography either prior to placement or at the time. We have found trans-cranial doppler to be a useful adjunct. If the basilar artery and the posterior communicating artery can be seen, then flow from both vertebral arteries, the basilar arteries and posterior communicating arteries can be measured while temporally occluding the origin of the left subclavian artery with an occlusion balloon. Carotid-subclavian bypass is generally well tolerated, but some investigators have noted an increased stroke risk when this is performed in patients with atherosclerotic aneurysmal disease[66,67]. This may be related to the increased degree of calcification in this older population, and may not apply to the younger trauma patient. Demonstrating intact vertebral-basilar flow upon left subclavian occlusion precludes the need for prophylactic subclavian bypass or transposition[68].

Embolization

Distal embolization is more of a concern in older patients with diffuse atheromata. Rarely, a traumatic injury can present with thrombus arising from the injury site. In any patient where there is a recognized risk that wire or device manipulation might cause distal embolization, one should be prepared to perform distal angiography and to record distal pulses at the start and end of the procedure.

Bronchial obstruction

Grafts placed in the mid descending aorta have rarely been associated with compression of the left main bronchus[17,69]. This may be detected on follow up CT or chest radiograph, bronchoscopy for other reasons, or clinically with new onset asthma. The optimal management is open repair of the aorta and explanting the device when medically feasible. Bronchial stenting has been used but in our opinion will predispose to the development of aorto-bronchial fistula[17].

Implant syndrome

Occasionally patients will experience persistent or new back pain, fever, malaise and/or a flu-like syndrome following implantation. This has been reported more commonly with abdominal aortic endografting, but Rousseau and associates found it in nearly 1/5 of their series[5,70,71]. It is not clear whether the back pain is due to residual endotension, inflammation secondary to thrombosis in the excluded aorta and/or a systemic inflammatory response to the graft. Repeat CT angiography is required to rule out a complication, but the presence of elevated erythrocyte sedimentation rate, with or without leukocytosis, and a normal CT suggest the diagnosis. Many patients will also complain of various degrees of pleurisy. This condition is self-limited, usually resolving within 7 d, and can be treated with anti-inflammatory medications.

Dissection/rupture

Free rupture can occur at any time. Prevention is strict blood pressure control, particularly during periods of transfer or other procedures that might acutely elevate the heart rate and/or blood pressure. At the time of initial wire passage, great care should be given to watching the wire advance. If there is difficulty negotiating the aortic curvature or there is narrowing at the injury site, a directional catheter such as a vertebral and/or hydrophilic catheter can be invaluable.

There have been cases of delayed or immediate rupture after the graft has been deployed. Endografts which feature a bare metal proximal extension have been implicated in perforating the aortic wall[72]. Even covered grafts which do not have this feature have been implicated if there is poor apposition of the aortic wall with resultant motion against the wall. All three dedicated thoracic endografts discussed here have been implicated in acute or delayed perforation with/or without dissection, at least anecdotally, in the non-trauma experience. Proper graft sizing is essential to facilitate good graft-aortic apposition.

Proximal dissection has also been documented. One mechanism is that during ballooning of the proximal cuff, the ends of the graft can create a dissection flap that rapidly progresses retrograde. To avoid this, initial ballooning should be as gentle as possible, just enough to document profiling of the balloon along the side of the graft. Ballooning should only be done within the graft.

Migration

If the proximal landing zone is not long enough, and the aneurysm itself is large, stents can migrate distally. This may be detected on routine chest radiography, or may present with a new endoleak. In younger patients, as aortic growth occurs, an endograft may loose its fixation. If this should occur, options include both operative explanting and grafting, or proximal extension with another endograft. This is one of the reasons that life-long surveillance is necessary.

Paralysis

The risk of paralysis with endografting in the trauma setting has not been demonstrated in recent series. This is presumably because in the vast majority of cases only a short segment of the proximal descending aorta is covered. However, based on experience acquired from treating atherosclerotic aneurysms, there may be uncommon circumstances (such as treating a chronic post-traumatic aneurysm in an older patient) where there is increased risk. The major risk factors include covering more than 20 cm of aorta and prior abdominal aortic procedures. These risks are enhanced if the left subclavian must be covered, or if there is extensive iliac disease with hypogastric occlusion. In the elective setting, placement of a lumbar drain and allowing increased pressure after the endograft procedure can both reduce the risk of paralysis or, in as many as 1/3-1/2 of patients who do develop neurological changes, reverse the spinal cord ischemia[73,74]. Clearly this is not a viable option in most acute trauma settings, but may be considered in delayed or elective settings.

Other long term complications

Aortobronchial and/or aortoesophageal fistula have been reported years after endografting[17,75]. Sayed and colleagues reviewed 1518 cases reported in the literature of thoracic endografting (primarily for non-traumatic thoracic aneurysmal disease) and found 6 cases (0.4%) of thoracic graft infection, paralleling the experience with abdominal grafts[21,76]. The primary risk factors were other intravascular sources of infection. Surgical management was associated with a 16% mortality compared with 36% with medical management alone. Any evidence of aneurysmal changes or persistent bacteremia should prompt consideration for surgical intervention.

THE PEDIATRIC AND ADOLESCENT PATIENT

Endovascular aortic stent grafts are not commonly used in pediatric and adolescent patients[77]. These patients usually have aortas that are too small for currently available devices, will likely grow and the lack of long term durability has greater significance in this population. Rarely, an endograft has been used as a “bridge to definitive treatment” in patients who cannot undergo operative repair initially[32].

FUTURE TRENDS

Endograft technology is continuing to evolve, but perhaps even more significant, experience and longer follow up data is also beginning to accrue. Branched grafts are beginning to be designed for both arch and abdominal visceral vessels. Specific for the trauma population, a variety of grafts which are shorter, pre-curved and smaller are being developed which will allow more precise deployment and potentially reduce complication rates.

CONCLUSION

Endovascular repair of the traumatically injured thoracic aorta has emerged as an exceptionally promising modality that is typically quicker than open repair, with a reduced risk of paralysis. There are a specific set of anatomic criteria that need to be applied, which can be rapidly assessed by the CT angiogram. The enthusiasm for endovascular repair must be tempered by recognition of the complications and lack of long term follow up, particularly in younger patients. Surgeons who are skilled in open aortic repair must not only be involved, but should take on a leadership role during the planning, deployment and follow up of these patients[78]. Familiarity with all of the available devices expands treatment options. As more specific devices become available, and more follow up is accrued, the role of endovascular stents will continue to grow.