| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Vascular and Interventional Radiology |
1 From the Departments of Radiology (M.M., H.C., J.L., J.B.) and Vascular Surgery (P.L.), New York University School of Medicine, Suite HW 211, 560 First Ave, New York, NY 10016; and Siemens Medical Solutions, Malvern, Pa (B.S.). Received September 20, 2005; revision requested November 18; revision received December 20; accepted January 20, 2006; final version accepted March 1. Address correspondence to M.M. (e-mail: michael.macari{at}med.nyu.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Materials and Methods: This HIPAA-compliant study was exempt from institutional review board approval; informed patient consent was waived. Eighty-five patients (66 men, 19 women; mean age, 66 years; range, 45–81 years) underwent 110 multidetector CT examinations after endovascular repair of abdominal aortic aneurysms. Nonenhanced CT images were obtained. Intravenous contrast material was then injected at 4 mL/sec, and arterial and venous phase (60 seconds) CT images were obtained. The nonenhanced and venous phase images were evaluated to determine if an endoleak was present. Subsequently, arterial phase images were analyzed. The effective dose was calculated. Ninety-five percent confidence intervals as indicators of how often arterial phase imaging would contribute to the diagnosis of endoleak were determined.
Results: Twenty-eight type II endoleaks were detected by using combined nonenhanced and venous phase acquisitions. Twenty-five of the 28 endoleaks were also visualized during the arterial phase. Three type II endoleaks were seen only during the venous phase. The arterial phase images depicted no additional endoleaks. Seventy-eight CT examinations performed in 67 patients revealed no endoleak during the venous phase. The arterial phase images also depicted no endoleaks at these examinations. Thus, for no more than 3.1% of all examinations, there was 95% confidence that arterial phase imaging would depict an endoleak missed at venous phase imaging. Arterial phase imaging contributed to a mean of 36.5% of the effective dose delivered.
Conclusion: Study results indicate that arterial phase imaging may not be necessary for the routine detection of endoleaks. Radiation exposure can be decreased by eliminating this phase.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
Open repair of the aneurysm has been the standard therapy for the management of AAAs. However, open repair is associated with mortality rates as high as 5.6% in patients with average risk (3). Generally, an aneurysm diameter of 5 cm has been the criterion used to indicate that a patient should undergo repair. Results of a study suggest that patient survival is not improved when AAAs smaller than 5.5 cm are immediately repaired (2). In that study, more than 1000 patients with AAAs 4.0–5.4 cm in diameter were randomly assigned to undergo either immediate repair or surveillance. Overall, the survivals and mortality rates associated with AAA were not improved by using open repair rather than surveillance in that study.
In 1991, Parodi et al performed endovascular repair of an AAA (4). Since that time, the frequency of endovascular AAA repair has dramatically increased (5). Endovascular repair, as compared with conventional open repair, has been shown to facilitate decreased hospital and intensive care unit stays and increased rates of hospital discharge to home rather than to rehabilitation centers (5,6). A recent multicenter randomized trial of open versus endovascular repair of AAAs at least 5 cm in diameter revealed 30-day mortality rates with open and endovascular AAA repair of 4.6% and 1.2%, respectively (6). At centers where the personnel have expertise in this type of repair, the majority of AAAs that are amenable to endovascular therapy are treated this way. However, there are concerns regarding the treatment of AAAs with an endovascular approach.
After endovascular repair, patients must undergo noninvasive follow-up imaging for evaluation of potential associated complications for the remainder of their lives. Acute complications include arteriovenous fistula, atherosclerotic plaque embolus, endoleak, and graft thrombosis (7). Delayed complications include endoleak, thrombosis, and graft migration. Contrast material–enhanced computed tomography (CT) is the imaging modality of choice for evaluation of these potential complications. It has been suggested that CT should be performed before aortic enhancement, during the arterial phase of enhancement, and during the venous phase of enhancement to optimize the detection of these complications—especially endoleaks (8). However, the use of this protocol results in three CT data acquisitions per examination. Since an endoleak or thrombosis may be delayed and occur long after stent-graft placement, individuals who undergo stent-graft therapy need to be followed up indefinitely (9). Therefore, these patients will be exposed to a large cumulative amount of ionizing radiation over time.
The health risk associated with ionizing radiation from diagnostic CT is low but not 0% and has been linked to an increase in the lifelong risk of fatal cancers (10). This risk increases with multiple CT examinations. We have found that endoleaks occasionally are visualized only at venous phase imaging. Therefore, in an attempt to decrease patient exposure to radiation, we conducted this study to retrospectively determine if arterial phase imaging is necessary for the follow-up imaging of patients who have undergone endovascular stent-graft therapy for AAA.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
CT Technique
CT was performed by using a four–detector row (n = 21) or 16–detector row (n = 64) scanner (Siemens Medical Systems, Forchheim, Germany). Initially, nonenhanced CT images through the abdomen and pelvis were obtained. One hundred fifty milliliters (300 mg of iodine per milliliter) of a nonionic intravenous contrast material (iopromide, Ultravist 300; Berlex Laboratories, Wayne, NJ) was then administered, through a 20-gauge catheter inserted into an antecubital vein, at a rate of 4 mL/sec by using a power injector (Envision CT Injector; Medrad, Pittsburgh, Pa). Arterial and venous phase acquisitions were then performed.
Arterial phase imaging was performed by using bolus tracking. Arterial phase CT data acquisition was initiated 6 seconds after the attenuation of a region of interest positioned in the aorta at the level of the celiac artery reached 150 HU. The 6-second delay allowed the patient to receive breathing instruction and facilitated maximized aortic enhancement. Venous phase acquisition was initiated immediately after the arterial phase with the four–detector row scanner and 60 seconds after the arterial phase with the 16–detector row scanner.
The nonenhanced CT and arterial phase acquisitions were initiated at the level of the celiac artery and continued to the level of the lower symphysis. The venous phase acquisition encompassed the entire abdomen and pelvis. For the arterial phase acquisition, a 0.75- or 1.00-mm detector configuration was used, depending on whether a 16– or four–detector row scanner was used, respectively. For the nonenhanced CT and venous phase acquisitions, a 1.5- or 2.5-mm detector configuration was used, again depending on whether a 16– or four–detector row scanner was used, respectively. Venous phase data acquisition with use of the four– and 16–detector row scanners was performed in less than 20 and 12 seconds, respectively.
Four-millimeter-thick sections were reconstructed and downloaded to the picture archiving and communication system (Siemens Medical Systems). CT parameters were set at 120 kVp and 180 mAs (effective) for all (ie, nonenhanced, arterial, and venous phase) acquisitions and both the four– and 16–detector row scanners.
Data Interpretation
Two readers (H.C. and J.L., with 3 years and 1 year of vascular CT experience, respectively) retrospectively and in consensus evaluated the nonenhanced and venous phase data to determine if an endoleak was present. The data on continuous 4-mm-thick transverse sections were evaluated without patient identifiers on the picture archiving and communication system workstation. A leak was defined as a blush of contrast material that was seen in the aneurysm sac on the venous phase images but not on the nonenhanced images. The type of endoleak was determined: A type I endoleak was one that occurred at a site of stent attachment to the aortic wall. A type II endoleak was one that occurred in a back-bleeding vessel—specifically, the inferior mesenteric or lumbar artery. A type III endoleak was one that occurred in the stent-graft itself if the device was a nonarticulating stent-graft (Ancure; Guidant, Indianapolis, Ind) or a leak that occurred at an articulation of the stent if it was a modular device (Aneurex; Medtronic, Minneapolis, Minn). Type IV endoleaks, which occur at the time of stent-graft placement, were not evaluated, and endotension—that is, a type V endoleak—was considered to be present when the aneurysm sac increased in size but no leak was visualized.
Subsequently and in the same setting, the arterial phase images were analyzed by the same two readers to determine if the endoleaks seen during the venous phase were also visualized during the arterial phase. The arterial phase images were analyzed to also determine if additional endoleaks that were not seen during review of the venous phase images were present.
Calculation of Effective Dose
For each acquisition phase (ie, nonenhanced, arterial, and venous), the average CT dose index and average dose-length product (DLP) were calculated by using measurements on the CT console. To determine the effective dose (E), we used the following equation: E = EDLP · DLP (11), where EDLP is a region-specific factor that is normalized to E. For the abdominal region, the EDLP is 0.015 mSv · mGy–1 · cm–1, and for the pelvic region, the EDLP is 0.019 mSv · mGy–1 · cm–1 (11). Since in our study both abdominal and pelvic acquisitions were performed, the mean value for both regions (ie, EDLP = 0.017 mSv · mGy–1 · cm–1) was used to estimate the patient dose.
Statistical Analyses
To determine if the arterial phase acquisition was necessary for the diagnosis of endoleaks, we sought to ascertain whether this phase enabled the detection of endoleaks that were not visualized during venous phase imaging. Thus, arterial phase imaging was considered to have contributed to the diagnosis of endoleaks if it enabled the visualization of a leak that was not detected during the corresponding venous phase. As a result, the probability that the arterial phase would contribute to the diagnosis of any random CT study was equal to the probability that the study would not have a leak detected at the venous phase times the conditional probability that the arterial phase image would permit visualization of a leak given that no leak was seen at the venous phase.
The Blyth-Still-Casella procedure was used to compute an exact 95% upper confidence bound for the probability that the arterial phase would contribute to the diagnosis and the conditional probability that the arterial phase image would allow endoleak detection when no leak was seen at the venous phase. In this analysis, we accounted for correlations among the results of multiple CT examinations performed in the same patient by reducing the data to a single study per patient. That is, since for every patient who underwent multiple examinations there was literal perfect agreement among the studies with respect to whether or not arterial phase imaging contributed to the diagnosis, only the data from the first examination performed in each patient were used in the analysis. This approach yielded a valid confidence bound for each parameter in the sense that the corresponding confidence interval had a coverage probability of at least .95. This approach was also validated by the fact that the contribution from arterial phase imaging was constant across all 110 CT examinations. That is, it was not possible to explicitly model the arterial phase imaging contribution because the contribution was never observed to vary. Thus, it was not possible to use statistical procedures such as conditional logistic regression or generalized estimating equations as a means of applying data from all 110 examinations while accounting for correlations among the observations from multiple examinations performed in the same patient. All statistical computations were performed by using StatXact, version 6.2, software (Cytel Software, Cambridge, Mass).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
|
|
|
|
|
|
Dose Calculations
The average CT dose indexes and average dose-length products were 13.68 mGy and 491 mGy · cm, respectively, for nonenhanced CT; 15 mGy and 682 mGy · cm, respectively, for arterial phase CT; and 13.68 mGy and 699 mGy · cm, respectively, for venous phase CT. Calculated effective doses were 8.3, 11.6, and 11.9 mSv for the nonenhanced, arterial phase, and venous phase acquisitions, respectively. Therefore, arterial phase imaging contributed to a mean of 36.5% of the total absorbed dose to the patient.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
However, the long-term outcomes after endovascular repair are not as well documented as the long-term outcomes after open repair. Reported rates of graft failure requiring open intervention or secondary endovascular interventional procedures have been as high as 3% and 10%, respectively, after endovascular stent-graft therapy (13). There are many possible reasons for graft failure, including distal migration and thrombosis, but the primary reason is continued expansion of the aneurysm sac related to the endoleak(s) (14). Moreover, a recent study of the 2-year cumulative survivals of patients randomly assigned to undergo open versus endovascular repair revealed no difference in results between the two groups (89.6% versus 89.7%) (15).
Since complications of endovascular therapy may occur early or be delayed, lifelong imaging follow-up is required. CT is the primary imaging method used for patient surveillance after stent-graft therapy. Typical follow-up imaging management consists of a CT examination performed within the first month after stent-graft placement. CT is performed to detect complications and particularly to detect endoleaks and measure the residual aneurysm sac diameter. If the CT examination reveals a stable or decreasing aneurysm sac diameter and no evidence of endoleak, continued surveillance at 6–12 months is typically performed (9). If a small type II endoleak is seen and the aneurysm sac size is stable, more frequent follow-up—at 3–6 months—is indicated (9). If larger endoleaks or aneurysm sac expansion is seen, endovascular intervention may be warranted. Since complications such as endoleak, aneurysm sac expansion, and graft migration may occur long after the stent-graft is placed, the imaging surveillance is lifelong (9).
A major concern regarding the surveillance in patients who have undergone endovascular AAA repair is the cumulative amount of exposure to ionizing radiation (10). Since these patients are followed up on at least a yearly basis, and often more frequently depending on prior imaging findings, they will undergo many CT examinations during their lifetime. Moreover, the typical surveillance CT protocol for examining a patient after endograft placement consists of three separate acquisitions (8): a nonenhanced examination, an arterial phase acquisition, and a venous phase acquisition. It has been suggested that these three phases are necessary when evaluating stent-grafts (8).
When evaluating CT data after stent-graft therapy, a nonenhanced examination is essential because there is often calcifying thrombus within the aneurysm sac. If nonenhanced CT is not performed, the calcifying thrombus, which has high attenuation, may be mistaken for a leak on the contrast-enhanced images. However, in our study, the addition of the arterial phase data did not contribute to the detection, classification, or determination of etiology of the endoleaks. In three patients, a type II endoleak was identified during the venous phase only. In one patient, a type III endoleak was much better depicted during the venous phase. In the remaining patients, the arterial and venous phase images showed similar findings.
Results of prior studies (16,17) suggest that the early (ie, arterial) phase acquisition is essential because in some cases, endoleaks were detected only during this acquisition. However, in these studies, the data were obtained with a single–detector row CT scanner by using 5-mm-thick sections and the delayed (ie, venous) phase data were acquired 70–80 seconds after the initiation of the contrast material injection. These imaging parameters probably lead to suboptimal detection because the venous phase data are acquired too long after initiation of the contrast material injection. If data are obtained too long after initiation of the contrast material injection, contrast enhancement will be suboptimal. By using multidetector scanners, we were able to obtain the venous phase images 60 seconds after initiation of the contrast material injection and within a 12–20-second acquisition period.
We believe that the venous phase acquisition should not be delayed more than 60 seconds after the start of contrast material administration. However, the timing of the delayed phase that is optimal for the detection of endoleaks has not been established. If the aneurysm sac is noted to be getting larger at subsequent CT examinations and no endoleak is seen, further investigation with multiphase CT or conventional angiography could be performed. Subsequent noninvasive magnetic resonance (MR) imaging could be performed to take advantage of the higher contrast resolution and the multiple data acquisition time points that are possible.
When evaluating CT data after stent-graft therapy, it is important to determine the etiology and site of the endoleak (18). The use of thin-section, near-isotropic z-axis imaging with multidetector CT can facilitate this evaluation by allowing the radiologist to use coronal and sagittal multiplanar images in addition to transverse images. Type I endoleaks, which occur at proximal or distal attachment sites, are usually large and well visualized on both arterial and venous phase images. Type II endoleaks occur because of retrograde back bleeding of aortic side branches into the aneurysm sac. Because the circulation time for retrograde back bleeding into an aneurysm sac may be considerable, especially in patients with atherosclerotic disease or underlying cardiac insufficiency—as is often the case in patients with AAA—it is not surprising that some type II endoleaks are seen only during delayed acquisition.
There are several potential concerns about eliminating the routine use of arterial phase data acquisition. One concern is with regard to differentiating type II endoleaks from type I and III endoleaks (18). A type III endoleak is defined as a leak within the graft itself. Although these endoleaks are uncommon, the incidence of them may be increasing with the more widespread use of modular grafts that have multiple, separate, interlocking components. However, careful evaluation of the location of the endoleak should enable the differentiation between type II and type III endoleaks. A type III endoleak should be adjacent to and in contact with the graft. In the case of type II endoleaks, careful evaluation of the data will reveal a patent inferior mesenteric artery or lumbar artery, with the endoleak closely related to the back-bleeding vessel. It has been shown, however, that leaks may be misclassified with use of single– and four–detector row CT examinations (18). In the setting of an expanding aneurysm sac after endovascular repair, conventional angiography may be warranted to definitively treat the endoleak, regardless of its etiology.
A second potential concern about eliminating the routine use of arterial phase imaging is the possibility that common femoral arteriovenous fistulas and other complications such as pseudoaneurysms will be missed. These complications occur at the time of intervention or very shortly thereafter and are often clinically apparent. If there are local symptoms that suggest one of these complications, arterial phase acquisition through the region is mandatory. Because these are relatively acute procedure-related complications that generally do not occur after a delay, it may be prudent to perform an arterial phase acquisition during the initial follow-up after stent-graft therapy. Subsequent to this initial follow-up examination, imaging with nonenhanced and venous phase CT acquisitions can proceed. Other mid- and long-term complications such as graft thrombosis and migration and aortoenteric fistula should be easily detected during the venous phase. Moreover, if the follow-up imaging examinations are revealing that the aneurysm sac is getting larger and an endoleak is not identified, CT angiography or conventional angiography is warranted.
By eliminating the arterial phase, a reduction of slightly more than one-third of the total effective dose can be achieved. Although the additional lifelong risk of developing fatal cancer as a result of undergoing abdominal CT is low, it is not 0% (10). Moreover, since follow-up imaging after stent-graft therapy is performed at least annually, the radiation exposure is cumulative.
To reduce the radiation dose delivered to patients at follow-up imaging after stent-graft therapy, alternative imaging strategies have been applied. Both ultrasonography (US) and MR imaging can be used to detect complications after stent-graft therapy (17). US, however, is user dependent, and the ability to consistently measure the aneurysm sac size and detect and classify small endoleaks in the presence of a stent is difficult. As a result, US is not routinely used for this evaluation. MR imaging also does not involve the use of ionizing radiation and can depict endoleaks with use of dynamic gadolinium-enhanced T1-weighted gradient-echo sequences. MR imaging is the modality of choice for evaluating stent-grafts when the patient has renal insufficiency or is allergic to iodinated contrast material. However, because of the limited access to and expense of MR imaging, CT remains the imaging modality of choice for the follow-up of patients who have undergone AAA repair.
There were some limitations to our study. First, it was retrospective. Second, with the exception of the three type I endoleaks, the endoleaks detected at CT were not confirmed at conventional angiography. This is because CT has become the reference standard for detecting endoleaks. In the patient with the type III endoleak, since it was small and the aneurysm sac size was unchanged, the patient is being followed up clinically and has not undergone intervention.
In conclusion, with elimination of the routine use of the arterial phase at CT imaging in patients undergoing surveillance after endovascular therapy for AAA, no endoleaks were missed when data were acquired 60 seconds after initiation of the contrast material injection. Our results indicate that there is 95% confidence that arterial phase imaging will depict an endoleak that was missed at venous phase imaging in 3.1% or fewer of all CT examinations performed in patients with a stent-graft. By eliminating the arterial phase, an at least one-third reduction in radiation dose can be achieved.
|
ADVANCES IN KNOWLEDGE |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
|
FOOTNOTES |
|---|
Abbreviations: AAA = abdominal aortic aneurysm
Author contributions: Guarantor of integrity of entire study, M.M.; study concepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manuscript revision for important intellectual content, all authors; manuscript final version approval, all authors; literature research, M.M., H.C., B.S.; clinical studies, M.M., H.C., B.S., J.L., P.L.; statistical analysis, M.M.; and manuscript editing, M.M., P.L., J.B.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONADVANCES IN KNOWLEDGEReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. Stolzmann, T. Frauenfelder, T. Pfammatter, N. Peter, H. Scheffel, M. Lachat, B. Schmidt, B. Marincek, H. Alkadhi, and T. Schertler Endoleaks after Endovascular Abdominal Aortic Aneurysm Repair: Detection with Dual-Energy Dual-Source CT Radiology, November 1, 2008; 249(2): 682 - 691. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Chandarana, M. C. B. Godoy, I. Vlahos, A. Graser, J. Babb, C. Leidecker, and M. Macari Abdominal Aorta: Evaluation with Dual-Source Dual-Energy Multidetector CT after Endovascular Repair of Aneurysms--Initial Observations Radiology, November 1, 2008; 249(2): 692 - 700. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. Hong, J. P. Heiken, G. A. Sicard, T. K. Pilgram, and K. T. Bae Clinical Significance of Endoleak Detected on Follow-Up CT After Endovascular Repair of Abdominal Aortic Aneurysm Am. J. Roentgenol., September 1, 2008; 191(3): 808 - 813. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. W. Stavropoulos and S. R. Charagundla Imaging Techniques for Detection and Management of Endoleaks after Endovascular Aortic Aneurysm Repair Radiology, June 1, 2007; 243(3): 641 - 655. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
