Applied Radiology

Dr. Wu graduated with a BS in Biology from the Massachusetts Institute of Technology, Boston, MA, in 1990. He received his MD in 1999 from Baylor College of Medicine, Houston, TX. He is currently Chief Resident and a third-year Radiology Resident at Yale/New Haven Hospital, New Haven, CT. Dr. Wu plans to complete a Musculoskeletal Imaging Fellowship following his residency.

Traumatic aortic injury (TAI) is a major cause of trauma-related fatalities, with the majority of the victims dying at the scene of the accident. For the few who reach the hospital, rapid diagnosis and treatment is essential in order to prevent aortic rupture. Unfortunately, the diagnosis of TAI is difficult, as clinical signs and plain radiography are poor screening methods. Helical computed tomographic angiography (CTA) has vastly improved the evaluation of TAI, and its frequent use as a screening tool has led to a reduction in the number of aortograms. This paper will provide an overview of traumatic aortic injury and the role of CTA in its diagnosis.

More than 100,000 people sustain acute chest trauma in the United States each year, often resulting in traumatic aortic injury (TAI) and death. ¹ In fact, 80% to 90% of people with TAI die at the scene of the accident as a result of aortic rupture. ^2,3 Of those who reach the hospital, 60% to 70% survive if the aortic injury is diagnosed and treated promptly. ^4,5 However, mortality increases as the time to diagnosis is prolonged. Thirty percent of patients with untreated TAI die within the first 6 hours of arrival at the hospital, and 40% die within the first 24 hours. ^2,6 If the aortic injury is never diagnosed, 90% die within 4 months. ² Thus, a rapid and effective diagnostic test for TAI is essential.

Unfortunately, the diagnosis of TAI is difficult. Physical examination and chest radiography are of limited value, especially in hemodynamically stable patients. Aortography is the gold standard; however, it is not without risks and can be time consuming. Thoracic aortography has a systemic morbidity rate of 1.7% and fatalities have been associated with catheter-induced aortic injury. ^7,8 Computed tomographic (CT) evaluation of the thoracic aorta is a fast, safe, and effective screening tool for TAI. Innovations in CT technology, especially helical multidetector scanning, enable faster scans and better z-axis resolution, which lead to the earlier diagnosis and treatment of TAI. This paper will provide an overview of traumatic aortic injury and the role of CT angiography (CTA) in its diagnosis.

Overview of traumatic aortic injury

Anatomic considerations and mechanisms of injury

The descending thoracic aorta is fixed to the spine, whereas the ascending aorta and aortic arch are fairly mobile. Most blunt aortic injuries occur in the ascending aorta, often at the transition between the mobile and fixed portions of the thoracic aorta. In fact, 90% of aortic injuries occur at the aortic isthmus, in this area of transition. ^4,5,9-11 The isthmus is the portion of the aorta between the origin of the left subclavian artery and site of attachment of the ligamentum arteriosum. Other sites of injury include the aortic root and arch. However, injuries at these two locations are usually more serious and often fatal. ^4,9,11

During a deceleration injury, such as from a motor vehicle collision or fall, the aorta is subjected to both shearing and bending forces. Shearing forces occur when the mobile arch decelerates more slowly than the fixed descending aorta. Bending forces occur when the aorta is flexed across the left pulmonary artery and left mainstem bronchus. These bending and shearing forces also lead to the "osseous pinch," in which the aorta is injured as it is pinched between the anterior bony structures and the spine during chest compression. ^9,12

A review of the composition and characteristics of the aortic wall can help in understanding aortic injuries and the signs seen on CT (Figure 1). ⁹ The aorta is composed of three layers: intima, media, and adventitia. The media is the middle and strongest layer, providing most of the aorta's support and elasticity. Tears of the intima and portions of inner media lead to intraluminal flaps and filling defects (Figure 2). Generally, however, the contour of the aorta is preserved, as most of the media remains intact. A pseudo-aneurysm occurs when there is complete disruption of the intima and media, without disruption of the adventitia. The weak adventitia bulges outward, forming a characteristic outpouching (Figure 3). These pseudoaneurysms can rupture when treatment is delayed. ⁹ Both intimal flaps and pseudoaneurysms can lead to emboli of thrombus or intimal debris. Moreover, aortic dissection can occur as the intimal flap extends longitudinally along the wall of the aorta. ⁹

Treatment of aortic injury

Aortic rupture with immediate death is the greatest concern in TAI. Treatment consists of urgent thoracotomy with surgical repair once the diagnosis is made. Unfortunately, surgery is associated with significant mortality (20% to 54%) and morbidity (5% to 10% will develop paraplegia). ^13,14 Aortic cross-clamping during surgery can lead to paraplegia secondary to under perfusion of the spinal cord. ¹⁵ Moreover, these patients are often poor surgical candidates as a result of other coexisting injuries. The use of antihypertensive medications, such as beta-blockers, improves outcome and decreases the risk of aortic rupture. ¹⁶ These medications are especially important in patients with multi-organ injuries who cannot safely undergo immediate surgery.

In the past, all patients with aortic injuries, no matter how small, were surgically treated. However, a small number of studies have described the nonoperative management of patients with minimal aortic injury (MAI). ^17-19 Characterized by intimal flaps of <1 cm without significant periaortic hema-toma, MAI occurs in 10% of patients with TAI. ¹⁸ Such patients have been treated with antihypertensive medication alone and closely followed-up with serial CT scans. Animal studies have shown that arterial injuries limited to the intima and inner media heal without intervention in a few days. ^18-20 However, nonoperative management of MAI is not a universally accepted practice.

The few patients with untreated TAI who survive can develop chronic pseudoaneurysms (Figure 4). Forty-two percent of patients with chronic pseudoaneurysms develop symptoms within 5 years, 85% within 20 years. ^9,18,21 Such patients require surgical repair, as they remain at risk for aortic rupture, embolism, and dissection of the aortic wall.

Diagnosis of TAI

Diagnosis of TAI is difficult. Physical examination in the hemodynamically stable patient is of limited value, and chest radiography (CXR) has variable sensitivity (53% to 100%) and poor specificity (1% to 60%). ^4,9 Signs of mediastinal widening, left apical cap, tracheal deviation, and depression of the left mainstem bronchus on CXR are not specific to patients with TAI. The positive predictive value (PPV) for chest radiography is only 4% to 20%. ^22,23 Despite these shortcomings, CXR was used for several decades as the primary screening tool for TAI. Patients with abnormal CXR were taken to angiography for confirmatory aortograms. Prior to the adoption of helical CT for screening, only 10% of aortograms were positive for aortic injury. ^15,23

In 1983, Heiberg et al ²⁴ used conventional CT to evaluate a small series of patients with TAI. Based on the initial studies, conventional CT was most useful in the detection of mediastinal hematoma (Figure 5). ^9,25 If none was detected, TAI could be confidently excluded. If mediastinal hematoma was observed, patients were sent to angiography. Early studies showed good sensitivity, 90% to 100%, and negative predictive value (NPV), 94% to 100% for the detection of TAI with conventional CT. However, specificity and positive predictive value (PPV) were poor, 19% to 45% and 0% to 50%, respectively. ^9,25-27 These low values were due to the fact that mediastinal and periaortic hematomas can be seen in other chest injuries, such as sternal or spinal fractures. In fact, early studies showed that only 14% to 56% of patients with mediastinal hematomas had TAI. ¹⁶ Although conventional CT was an improvement from plain radiography, it was imperfect. A large number of patients still underwent unnecessary aortograms. ^16,17,25

Long scan times precluded performing CT angiography with conventional scanners. Moreover, suboptimal opacification of the aorta made intimal flaps and small pseudoaneurysms difficult to detect with conventional scanners. With the introduction of helical CT and followed by multidetector computed tomography (MDCT), the direct signs of aortic injury could be visualized, thereby raising the specificity of CT in the diagnosis of TAI. Conventional CT yields a specificity of only 19% to 45%, whereas the specificity rises to 92% to 99% with helical CT. ^{9,16-18,25-27} Helical CT angiography is now a major screening tool for the evaluation of TAI.

Technique

The developments in CT technology from conventional to helical scanning, and then from single-detector CT (SDCT) to the multidetector CT (MDCT) have led to unprecedented speed and quality of CT imaging. Most trauma centers are currently equipped with MDCT and routinely employ this technology in the evaluation of TAI. The most common type of MDCT in practice is a four-channel scanner, however, 16-channel scanners are becoming more available. The advantages of MDCT over SDCT include shorter acquisition time, retrospective compilation or creation of thinner slices from the same raw data, and improved three-dimensional (3D) renderings with fewer artifacts. ²⁸ Because of shorter acquisition time and the ability to reliably image during peak vascular enhancement, smaller contrast volumes (up to 30% less) may be used as compared with SDCT. ^28-30

Prior to beginning the examination, all possible external sources of artifact should be removed and the patient's arms should be raised above the head. Moreover, for intubated patients, the ventilator should be paused just prior to imaging. With both SDCT and MDCT, the aorta must be optimally enhanced. Typically, 90 to 120 mL of nonionic contrast media with an iodine concentration of 240 to 340 mg/mL is administered using a power injector at 3 to 4 mL/sec, through a 20-gauge or larger peripheral venous catheter. ^{5,16,17,25,31-33} Injection through a right antecubital vein is recommended, as the use of a left antecubital vein can result in contrast artifact in the left brachiocephalic vein and limit evaluation of the mediastinum. ^5,32 Following injection, an immediate saline flush of 10 mL of isotonic solution can improve vascular enhancement profile and decrease streak artifact. ³⁴ Using this technique, one can also decrease the contrast requirement by up to 20%. ³⁵

The scanning time can be determined using automated systems (eg, Smartprep, GE Medical Systems, Milwaukee, WI; Care Bolus, Siemens Medical Systems, Erlangen, Germany). Alternatively, a test injection can be used; 20 mL of nonionic contrast material is injected at 2.5 to 3 mL/sec and multiple sequential images are then obtained and coupled with region of interest analysis of the aorta in order to determine the appropriate scanning delay. In general, the optimal delay for the thoracic aorta is 20 to 25 seconds. ^5,16,17,32

With SDCT, the slice collimation and slice width are the same. Five to 10-mm collimation with 2.5- to 5-mm reformation is commonly used with SDCT angiography, producing limited axial images and poor reconstructions. However, with MDCT, there is more flexibility and the user can select the detector collimation for each study. For the 4-channel scanners, typical collimation choices are 1 mm, 2.5 mm, or 5 mm. Five- to 2.5-mm collimation with 1-mm reformations is commonly used when evaluating for TAI. With the 16-channel scanners, 0.75 to 0.5 mm collimation is achievable. ^29,31,36

Although cardiac gating is not widely used when evaluating for TAI, electrocardiogram (EKG)-gated imaging can improve detection of aortic injury by limiting cardiac motion artifact and improving reconstructions. Since the heart chambers move the least during diastole, imaging during this phase of the cardiac cycle is ideal. Electrocardiogram-gated imaging can be divided into prospective and retrospective gating. In prospective EKG-gating, the start of each axial scan is triggered by the R-peak from the EKG signal, ensuring that imaging occurs during diastole. In retrospective EKG-gating, several axial images are obtained for every z-axis location and only those images that correspond to diastole are utilized for image reconstruction. Retrospective gating yields higher quality reconstruction images compared with prospective gating; however, the radiation dose is higher with retrospective cardiac gating. ³⁷

Postprocessing of the axial slice data is an important aspect of CT angiography. Although studies have shown that sensitivity and specificity do not improve with reconstruction, additional images are useful for surgical planning. ^5,16,33 Many surgeons, accustomed to sagittal and oblique views of the thoracic aorta, still request aorto-grams despite obvious findings on the axial images. Two-dimensional multiplanar reformations (MPR) in the coronal and sagittal planes are commonly performed on studies positive for signs of TAI. Three-dimensional images can be generated, and the quality of these images has improved with advances in MDCT (Figures 6 and 7). The newer 16-channel scanners allow for thinner collimation, which improves z-axis resolution and reduces artifacts during 3D image reconstruction. ³¹

Initially, most 3D reconstructions were performed with surface rendering or maximum intensity projection (MIP). With surface rendering, each voxel in the data set is determined to be a part of or not a part of the object of interest, thereby defining the surface of the object. Once the surface of the object is determined, the rest of the data is discarded. In MIP, each voxel along a line from the viewer's eye is evaluated and the maximum voxel value is used for the corresponding display pixel. Currently, most institutions use volume rendering, as this type of postprocessing has proven to be superior to MIP and surface rendering. ^38,39 In volume rendering, the computer sums the contribution of each voxel along a line from the viewer's eye through the data set. This is done repeatedly in order to determine the value of each pixel in the final display image. Therefore, the entire data set is incorporated in the final image. ^38,39

CT findings in traumatic aortic injury

CT findings suggestive of traumatic aortic injury can be divided into indirect and direct signs. ³⁶ Indirect signs of trauma include mediastinal and periaortic hematomas, which appear as soft-tissue-attenuating material surrounding the mediastinal structures or aorta, respectively (Figures 5 and 8). With conventional CT scanners, only indirect signs of TAI can be visualized. Although the indirect signs are highly sensitive for TAI, they are not very specific, 19% to 45%. ^{9,10,17,25,33} Mediastinal and periaortic hematomas can also be seen in patients with sternal or spine fractures but without TAI.

With helical CT and CTA, direct signs of TAI can be visualized. The direct signs of aortic trauma are most specific and refer to visualized defects in the aorta itself. These direct signs range from small intraluminal filling defects to focal pseudoaneurysms to gross aortic transection with active extravasation. ^22,36 Intraluminal filling defects can be caused by small intimal flaps or intimal thrombus. The intimal flaps are seen as small linear low-density intraluminal filling defects (Figure 2). With more serious injuries, the intimal flap can extend into the media or adventitia. If the media is completely disrupted, the weak adventitia will bulge outward producing a pseudo-aneurysm (Figures 3 and 9). If the adventitia is breached, extravasation of contrast can occur, ranging from small leaks to gross extravasation. Gross active extravasation is rarely seen on CT, as patients who survive an adventitial breach are most likely hemodynamically unstable. Intimal flaps and pseudoaneurysms can produce aortic contour abnormalities. Best seen on sagittal reconstructions, aortic contour abnormalities are also a direct sign of TAI (Figure 10). ^22,31

Pitfalls to CTA imaging

Occasionally, the diagnosis of traumatic aortic injury on CTA can be difficult to determine, as there are many conditions that mimic direct and indirect signs of TAI. Prominent thymic tissue, especially in children and young adults, can mimic mediastinal hematoma. Motion, volume-averaging, and pulsation artifacts can be mistaken for intimal flaps or hematomas. Cardiac motion can also produce artifacts; which can be reduced with EKG-gated imaging. Moreover, irregularities in the contour of the aortic wall can appear similar to a pseudo-aneurysm. The most common aortic wall contour abnormalities mistaken for TAI are atherosclerotic pseudoaneuryms, prominent bronchial arteries, and ductus diverticulums. ^17,18,40 Several false-positive cases on CT and aortography stem from a normal ductus diverticulum being mistaken for a pseudo-aneurysm. ^16,41 Finally, high-density contrast in the brachiocephalic vein and superior vena cava can cause streak artifacts, limiting evaluation of the aorta and mediastinum. These equivocal cases should go to angiography for further evaluation.

Accuracy of CTA and aortography

In screening for TAI, sensitivity and NPV are more important than specificity and PPV. A screening tool with 100% sensitivity and NPV will ensure that all patients with TAI are discovered. Missing the diagnosis of TAI is unacceptable, as the chief consequence is aortic rupture and death.

Single-detector CT and MDCT angiography enabled visualization of direct signs of TAI, specifically intimal flaps and pseudoaneurysms. Studies have shown that these direct signs identify TAI with a sensitivity of 100%, an NPV of 100%, a specificity of 92% to 99%, and a PPV of 80% to 89%. ^{16-18,25,32,40,42,43} All patients with direct signs of injury on CTA had TAI, regardless of whether they also had indirect signs. Most authors do not recommend angiography if only mediastinal or periaortic hematomas are seen. ^16,25,32,43 Mirvis et al ²⁵ evaluated 100 patients with CT evidence of mediastinal and/or periaortic hematomas and concluded that if hematomas were seen only in the anterior or posterior mediastinum, TAI was very unlikely.

Currently, aortography is the gold standard. However, in comparison to CTA it is more expensive, time consuming, and invasive. Interventional teams generally are not readily available after normal working hours, necessitating long waiting periods or transfer to a tertiary center for aortography. Hunt et al ³ showed in 144 patients with TAI that the average time from admission to angiography was 147 minutes. Some authors argue that if direct signs of TAI are seen on CTA, conventional angiography is redundant and can delay treatment. ^9,43-45 CT angiography adds only a few additional minutes to the patient work-up, as most patients undergo imaging for other injuries. Moreover, aortograms have serious risks, including iatrogenic extension of intimal flaps by the guide- wire or catheter, as well as entry of the wire into the pseudoaneurysm. ^8,9

Furthermore, aortography is not 100% sensitive. Gavant et al ¹⁷ studied 1518 patients with nontrivial chest trauma and found aortography to have a sensitivity of 92%, whereas the sensitivity of helical CTA was 100%. CT has the additional benefit of being able to diagnose other injuries. Pneumothorax, pulmonary contusion, pericardial effusion, and rib fractures may not be seen on aortography, but are easily seen on CT.

Aortography appears to be slightly more specific than CT, 99% to 100% for aortography versus 92% to 99% for helical CTA. ^16,17,25 Thus, equivocal CT studies should go to aortography. However, most authors advocate sending patients directly to surgery if direct signs of TAI are obvious, in order to save time. ^9,43-45 Most studies conclude that indirect signs of TAI are sensitive but nonspecific and, therefore, should not be the sole indication for aortography. Many centers advocate the use of aortography only when direct signs of TAI are equivocal or large periaortic hematomas are seen. ^32,43,45

Management of TAI based on CTA

The main goal of CTA in aortic trauma is to diagnose all patients with aortic injury quickly, so that surgery can be performed prior to rupture of the pseudoaneurysm. Another goal is to reduce the number of unnecessary aortograms, as they are invasive and can delay treatment. Management of patients with traumatic aortic injury depends on identifying the indirect and direct signs of TAI on CT and knowing the appropriate next step. If there are no indirect or direct signs of TAI, the patient is considered to be free of aortic injury. If only indirect signs are present, such as a mediastinal hematoma, then aortography is not necessary. Patients with equivocal direct signs of TAI require a confirmatory aortogram. Finally, patients with obvious direct signs of TAI on CTA should save time by going directly to surgery. ^{9,16,17,25,40,45} Adherence to this pathway will hopefully save more lives and reduce unnecessary aortography (Figure 11).

Conclusion

Traumatic aortic injury is a deadly disease, and patient outcome is highly dependent on its early diagnosis and treatment. Helical CT angiography has proven to be a very effective screening tool for aortic injury. CT angiography is fast, accurate, and in certain aspects, superior to aortography. Optimal care of patients with traumatic aortic injury depends on the identification of direct and indirect signs of aortic injury on CTA and understanding the appropriate next step in management.

Acknowledgments

The author thanks Dr. Sanjay Saluja for guidance throughout this project and Cathy Camputaro for assistance in generating the 3D images.