| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University Hospital of Mainz, Langenbeckstrasse 1,
Mainz 55131, Germany.
2 Department of Cardiovascular Surgery, University Hospital of Mainz, Mainz,
Germany.
Received May 7, 2004;
accepted after revision November 29, 2004.
Address correspondence to M. B. Pitton
(pitton{at}radiologie.klinik.uni-mainz.de).
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Fifty-two patients underwent endovascular
aneurysm repair with nitinol stent-grafts. Follow-up data sets included
contrast-enhanced biphasic CT and MRI within 48 hr after the intervention; at
3, 6, and 12 months; and yearly thereafter. The endoleak size was categorized
as 
3%, > 3% 10%, > 10% 30%, or > 30% of the maximum
cross-sectional aneurysm area. A consensus interpretation of CT and MRI was
defined as the standard of reference.
RESULTS. Of 252 data sets, 141 showed evidence for endoleaks. The incidence of types I, II, and III endoleaks and complex endoleaks was 3.2%, 40.1%, 8.7%, and 4.0%, respectively. The sensitivity for endoleak detection was 92.9%, 44.0%, 34.8%, and 38.3% for MRI, biphasic CT, uniphasic arterial CT, and uniphasic late CT, respectively. The corresponding negative predictive values were 91.7%, 58.4%, 54.7%, and 56.1%, respectively. The overall accuracy of endoleak detection and correct sizing was 95.2%, 58.3%, 55.6%, and 57.1% for MRI, biphasic CT, uniphasic arterial CT, and uniphasic late CT, respectively.
CONCLUSION. MRI is significantly superior to biphasic CT for endoleak detection and rating of endoleak size, followed by uniphasic late and uniphasic arterial CT scans. MRI shows a significant number of endoleaks in cases with negative CT findings and may help illuminate the phenomenon of endotension. Endoleak rates reported after endovascular aneurysm repair substantially depend on the imaging techniques used.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Because a favorable clinical outcome depends on the reliable detection of endoleaks, the choice of the right imaging method for follow-up is crucial. However, published data for different methods vary greatly in terms of detection rates [9–16]. MDCT has been the imaging technique of choice in many centers but often shows only sac stability or sometimes sac enlargement without clear evidence of endoleak (type V endoleak); the definitions of the types of endoleaks are provided in Table 1. A technique with increased sensitivity for leak detection could be helpful for further stratification of those patients. This study analyzes the sensitivity of MRI compared with MDCT for endoleak detection and thereby shows the potential clinical impact of performing MRI for follow-up.
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
For helical CT, we used a single-detector scanner (PQ 6000, Picker International) from 1998 to 2001 for 27 studies (10.7%) and a 4-MDCT scanner (Somatom Volume Zoom, Siemens Medical Solutions) from December 2001 to 2003 for 225 studies (89.3%) after it was installed in December 2001. Images were obtained using 120 kV, 130 mAs, a pitch of 1.5, a reconstruction thickness of 3 mm (for MDCT, collimation was 4 x 2.5 mm), and a reconstruction interval of 2 mm. The CT data set included unenhanced CT scans and biphasic contrast-enhanced scans. Contrast material—120 mL of iomeprol (Iomeron 300, Altana Pharma AG)—was administered at a flow rate of 4 mL/sec via a 22-gauge cannula in the antecubital vein using a power injector (LF CT 9000, Liebel Flarsheim) and was followed by a saline chaser (50 mL) administered at the same flow rate.
Data acquisition was started using a bolus trigger for the early arterial contrast scans (automatic start of acquisition depending on the aortic contrast increase). Late scans were made 90 sec after the arterial phase. All studies were transferred to a separate workstation (Magic View, Siemens Medical Solutions). For sophisticated analysis, additional multiplanar reformations (slab thickness, 3 mm) were interactively created for endoleak evaluation and were assessed by two experienced radiologists. Contrast enhancement outside the stent-graft but inside the aneurysm sac was defined as an endoleak.
MRI studies were performed using a 1.5-T MR imager (Magnetom Vision, Siemens Medical Solutions) during inspiratory breath-holding. The study protocol included T2-weighted turbo spin-echo axial (TR/TE, 5,572/99), precontrast T1-weighted FLASH 2D axial (173.4/4.1; flip angle, 80°; fat saturation; slice thickness, 5 mm) and coronal 3D FLASH angiography (4.0/1.6; flip angle, 30°; fat saturation; 6/8–7/8 rectangular field of view, 400 mm; matrix size, 185 x 256 pixels; reconstruction thickness, 1.75 mm; acquisition time, 22 sec). The test bolus method was used to time acquisition delay [17]. Two acquisitions, one early arterial phase and a second approximately 30 sec after the beginning of the first acquisition, were performed. The study was completed with an axial postcontrast T1-weighted FLASH 2D sequence, which was comparable to late CT scans in terms of acquisition delay.
The contrast bolus (0.2 mL/kg of body weight gadopentetate dimeglumine [Magnevist, Schering]) was administered at a flow rate of 2 mL/sec via a 22-gauge cannula in the antecubital vein using a power injector (Spectris, Medrad) and was followed by a saline chaser administered at the same flow rate. Endoleak was defined as extraluminal contrast enhancement on 3D fast low-angle shot (FLASH) angiography and postcontrast T1-weighted flash 2D images of at least 100% compared with precontrast images. A smaller signal increase without flow detected on 3D MR angiography was not classified as an endoleak, because it could represent tissue organization rather than endoleak blood flow [18]. All studies were transferred to a separate workstation (Magic View) and were assessed by two experienced radiologists.
All studies were blinded and analyzed independently without knowledge of
prior or subsequent endoleaks detected by the same or the other imaging
method. The database therefore included a total of 252 CT and MRI
examinations. After a randomized stack of 252 CT and MRI data sets were
independently interpreted, the final diagnostic reference standard was made by
consensus interpretation of the respective images of both techniques in each
individual case. The reference diagnosis of endoleak was made in all cases
with positive endoleak findings obtained on CT, MRI, or both. All endoleaks
were classified as type I–V or as complex endoleaks
(Table 1). Primary endoleaks
were defined to be endoleaks identified in data obtained immediately after the
intervention or within the first 30 days after the intervention. Endoleaks
identified later were defined as secondary leaks. The size of each endoleak
(primary or secondary) was categorized in one of the following four groups: A,
3% of the cross-sectional area of the aneurysm sac; B, > 3% but
10%; C, > 10% but 30%; or D, > 30%
(Table 3).
|
The area of the endoleak was selected with the cursor in axial planes (region of interest [mm2]) and was compared with the maximum cross-sectional area of the aneurysm sac in axial planes. The percentage given is the ratio between the endoleak size and the maximum aneurysm size. For follow-up of aneurysm shrinkage, the maximum aneurysm diameters were measured.
The study protocol was approved by the institutional review board. Informed consent was obtained from all patients participating in the study.
Statistical Analysis
To determine the accuracy of endoleak detection, we calculated sensitivity
and negative predictive values. The accuracy of endoleak sizing was tested
using contingency tables to describe the coherency between the diagnostic
findings. The Fisher's exact test was used to compare the endoleak detection
rate of MRI and biphasic CT for each follow-up study. The Wilcoxon's test was
used to compare the overall endoleak detection rate of the imaging protocols.
Statistical significance was defined as a p value of less than
0.05.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The reference diagnosis classified endoleaks as type I, II, or III or as complex endoleaks in eight (3.2%), 101 (40.1%), 22 (8.7%), and 10 (4.0%) data sets (Table 4 and Figs. 1A, 1B, 1C, 1D, 2A, 2B, 2C, 3A, 3B, and 3C). According to the reference standard, primary endoleaks occurred in 22 cases. Spontaneous endoleak closure during follow-up was found in seven of these cases. Secondary endoleaks were identified in 14 of 30 patients. Sixteen patients were without an endoleak over the entire study period.
|
|
|
|
|
|
|
|
|
|
|
On MRI, the number (incidence) of type I, II, III, and complex endoleaks were seven (2.8%), 93 (36.9%), 21 (8.3%), and 10 (4.0%) compared with eight (3.2%), 42 (16.7%), 10 (4.0%), and two (0.8%) on biphasic CT. Seven proximal type I endoleaks (type Ia) were detected on both CT and MRI. However, one distal type I endoleak (type Ib) was missed using MRI. In this case, the small distal leak was masked by calcification of the vessel wall and the platinum markers of the distal limb of the stent-graft.
From a total of 101 type II endoleaks, 42 were recognized on biphasic CT compared with 93 on MRI. These endoleaks were in most cases smaller than type I endoleaks. MRI was also superior to CT for the detection of type III endoleaks and complex endoleaks. In fact, the superiority of MRI was statistically significant in all endoleak categories other than type I endoleaks. All proximal type I endoleaks were large, but no significant difference between CT and MRI was found. In cases of small endoleaks with subtle changes seen on CT images, MRI was required to confirm the diagnosis. Because the majority of type II endoleaks were small or medium in size, the difference between MRI and CT was particularly large for this category.
Contingency tables describe the coherency of MRI and biphasic CT compared with the reference standard (Table 3). The overall detection rate of endoleaks was significantly better on MRI compared with biphasic CT (p < 0.0001). At each follow-up, the endoleak detection rate was significantly greater on MRI (p < 0.05, from the first to fifth follow-up study; not statistically significant from the sixth to ninth follow-up study because of smaller sample size).
The reference standard identified a total of 141 endoleaks (56.0% of all data sets). The number of endoleaks detected was 131 (52.0%) for MRI, 62 (24.6%) for biphasic CT, 49 (19.4%) for uniphasic arterial CT, and 54 (21.4%) for uniphasic late CT (Table 5). Endoleaks were missed or underrated in all size groups.
|
Because CT and MRI protocols included both arterial phase and late contrast-enhanced images, these findings cannot be explained by flow dynamics. The overall sensitivity for endoleak detection was 92.9%, 44.0%, 34.8%, and 38.3% for MRI, biphasic CT, uniphasic arterial CT, and uniphasic late CT (Table 6). The negative predictive values were 91.7%, 58.4%, 54.7%, and 56.1%, respectively. With respect to the reference standard, endoleak detection and sizing were best with MRI, followed by biphasic CT, uniphasic arterial CT, and uniphasic late CT. On MRI, 10 endoleaks were not detected (eight size A, one size B, and one size C). The size of two other endoleaks was underrated (one size A instead of B and one size A instead of C) (Table 3). Thus, 12 (8.5%) of 141 endoleaks were underdiagnosed on the basis of MRI.
|
All CT techniques used produced a considerable number of false-negative findings. This number was particularly high for uniphasic CT protocols alone (Table 5). Using biphasic CT, 79 endoleaks were not detected (40 size A, 33 size B, two size C, and four size D endoleaks). Twenty-six additional endoleaks were undersized (15 leaks underrated by one size category and 11 leaks by two size categories), meaning that 105 (74.5%) were underrated in total (Table 3).
Endoleak exclusion with both CT and MRI was associated with slightly improved aneurysm shrinkage. In these cases at 1, 2, 3, and 4 years, the mean diameter reduction of the aneurysm sac was -4.3 ± 1.5 (SD), -7.9 ± 2.9, -7.6 ± 4.5, and -9.4 ± 6.0 mm compared with -3.1 ± 1.2, -5.8 ± 2.6, -5.9 ± 4.0, and -7.3 ± 3.6 mm in cases with negative CT findings but positive MRI findings. However, this difference did not reach statistical significance. During follow-up, 16 reinterventions were performed (Table 7). One proximal cuff was implanted 7 days after intervention in a patient without evidence of endoleak because of an inadequate fixation of the stent-graft.
|
A total of 15 reinterventions were performed because of endoleaks, including two cases with negative CT but positive MRI findings. In one of this cases, selective angiography was performed to embolize collateral arteries; however, the respective vessel could not be occluded and the patient had a late conversion operation. The other case showed significant kinking of the stent-graft and moderate aneurysm growth. This case also required late conversion.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
In the studies discussed in the literature, Doppler sonography, CT, and MRI have all been used for follow-up. However, different endoleak sensitivity and detection rates have been reported using different diagnostic gold standards [9–12]. The analysis of endoleak rates reported in clinical studies may for this reason be affected by the use of methods of different diagnostic validity. The purposes of this study were to compare the diagnostic sensitivity of CT and MRI for endoleak detection and to identify the most sensitive imaging technique. These findings may help to clarify cases with failed aneurysm shrinkage and the phenomenon of endotension.
The data in this article compare contrast-enhanced biphasic CT (arterial and late contrast) and contrast-enhanced MRI. As compared with the reference standard, MRI was significantly better than biphasic CT and biphasic CT was better than uniphasic CT. The few false-negative MRI findings were retrospectively explained by calcifications of the vessel wall or artifacts by platinum markers of the stent-grafts masking the endoleak on MRI (as shown by unenhanced CT scans). MRI is well known to provide better tissue contrast and contrast sensitivity than CT. This difference could explain the improved endoleak visibility in our study. Because the CT and MRI protocols included both early arterial phase and late contrast-enhanced images, this effect was, to some extent, independent of flow dynamics. Endoleak size may be another important factor influencing endoleak visibility; however, a comparison of endoleak sizes as determined on CT and MRI may be a debatable issue. If CT is less sensitive for extraluminal contrast detection, the endoleak will be less visible when the contrast agent becomes more diluted in the endoleak cavity away from the source, suggesting the endoleak is smaller than it really is. In the same endoleak on MRI, due to improved contrast sensitivity, smaller volumes of contrast agent may be detected after diffusing farther into the cavity, suggesting larger endoleaks.
The proportion of studies performed on a single-detector scanner was relatively low. Most cases were investigated using the more sophisticated MDCT, allowing a valid comparison of current MDCT and MRI techniques. Angiography has already been shown to be less sensitive than biphasic CT and was, therefore, not included in this analysis [1]. The absence of a reliable gold standard for endoleak detection is a limitation of this study, and histologic correlation is not feasible. The reference standard used was a consensus interpretation of CT and MRI findings, resulting by definition in a 100% specificity and 100% positive predictive value for MRI and CT. For false-positive MRI findings to be avoided, significant contrast enhancement was required for positive endoleak diagnosis on both 3D FLASH angiography and postcontrast T1-weighted FLASH sequences.
As for CT on its own, diagnostic performance was improved by the biphasic technique, which identified a higher rate of late enhancing endoleaks and achieved more reliable endoleak sizing. In a recent article, Rozenblit et al. [16] reported endoleak detection rates of 91%, 97%, and 100%, respectively, for uniphasic arterial CT (including unenhanced phase), biphasic CT (arterial and late enhanced CT without unenhanced phase), and biphasic CT with unenhanced phase [16]. These data concur with our data that confirm an improved endoleak detection rate with biphasic CT protocols versus uniphasic CT protocols. However, the increased diagnostic potential of MRI was not considered in this article.
Some authors have presented the benefits of MRI in endoleak detection [9–15]. Haulon et al. [15] reported a significantly higher sensitivity for endoleak detection on contrast-enhanced MRI compared with CT (0.94 vs 0.50, respectively) and an improved negative predictive value (0.91 vs 0.55, respectively). These data are consistent with our findings. In the study by Haulon et al., some false-positive MRI findings were interpreted as angiographic shortcomings because digital subtraction angiography did not detect all type II endoleaks. In fact, for endoleak detection with selective angiography, a certain minimum blood flow is required to achieve perceptible contrast enhancement during digital subtraction angiography. Low-flow endoleaks may therefore be detected more easily on MRI. However, some contrast enhancement in the aneurysm sac has also been shown to be associated with angiogenesis during the histologic organization of the aneurysm sac [18]. In that experimental study, the mean signal increase after gadolinium DTPA administration was as much as 70% of the unenhanced image [18], in keeping with increasing capillary spreading. Endoleaks identified by positive MRI findings, on the other hand, produce a considerably larger signal increase and a positive identification of blood flow on 3D MR angiography. Each signal increase on contrast-enhanced T1-weighted sequences was therefore reviewed for a corresponding positive finding on 3D MR angiography to differentiate endoleaks from contrast enhancement due to tissue reorganization. Moreover, some of the high-flow endoleaks produce a flow void on T2-weighted images.
Color Doppler sonography has also been proposed as a noninvasive diagnostic tool for endoleak detection [11, 19]. Because it is a nonionizing low-cost technique, it might offer significant advantages over CT and MRI. Sato et al. [19] reported sensitivity and negative predictive value for color Doppler sonography that are similar to those for helical CT. However, considering the large number of false-negative CT findings in our study, color Doppler sonography is likely to miss a significant number of endoleaks.
Apart from its superiority in diagnostic accuracy, MRI offers advantages in terms of reduced nephrotoxicity and in patients with an allergy to iodinated agents. Moreover, the prevention of radiation exposure might be an important benefit in the follow-up of young patients after endovascular treatment of thoracic aortic injuries and dissections. Hilfiker et al. [20] disproved some initial concerns regarding potential nitinol graft migration or heating due to MRI. Our clinical follow-up protocol includes radiographs in all cases. We have not yet seen any graft migration induced by the magnetic field.
Nitinol stent-grafts cause limited artifacts on MRI and did not influence the assessment of the aneurysm. The nitinol frame is represented clearly by a well-defined signal void of the stent struts. Some radiopaque landmarks of the stentgraft may cause typical artifacts on both CT and MRI and may interfere with an assessment of the immediate vicinity of these artifacts. In our study, these areas were limited to small platinum markers at the connection sites and at both ends of the stent-graft, which might result in false-negative MRI findings. About half of our patients had been treated by coil embolization of aortic side branches before stent-grafting or during follow-up. With the use of fibered platinum coils, these embolizations caused limited MRI signal void artifacts at some distance from the aneurysm sac and did not influence the assessment of the adjacent aneurysm sac. Non-nitinol stent-grafts may produce more artifacts, and there is not sufficient safety information; however, we assume that some non-nitinol grafts may be more MR-compatible than others. Therefore, knowledge about the safety aspects and possible artifacts of stent-grafts is essential for successful endoleak detection. Artifacts should be analyzed systematically to prevent false-negative ratings.
The extensive imaging protocol of this study might not seem an appropriate follow-up procedure for a large number of patients. Because shrinkage of the aneurysm sac is an indicator of effective endovascular repair, biphasic CT protocols may be sufficient in most cases. However, if no shrinkage occurs or the size of the aneurysm sac even increases, MRI provides the best endoleak detection. In a considerable number of such cases, MRI may identify endoleaks that would remain undetected by CT alone. In future clinical studies, MRI will therefore play an important role in the investigation of endotension, which may at least partly be a shortcoming of present CT rather than an unexplained pathophysiologic phenomenon.
In conclusion, this comparative study showed the superiority of contrast-enhanced MRI over CT for endoleak detection. On MRI, the number of endoleaks detected and the appearance of endoleak size are considerably increased. The diagnostic yield of CT might be only slightly improved by using a biphasic CT protocol. Therefore, in clinical studies, different results of endoleak rates might be explained to a certain degree by the different sensitivity of the imaging methods used for endoleak detection.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  G. N. Levine, A. S. Gomes, A. E. Arai, D. A. Bluemke, S. D. Flamm, E. Kanal, W. J. Manning, E. T. Martin, J. M. Smith, N. Wilke, et al. Safety of Magnetic Resonance Imaging in Patients With Cardiovascular Devices: An American Heart Association Scientific Statement From the Committee on Diagnostic and Interventional Cardiac Catheterization, Council on Clinical Cardiology, and the Council on Cardiovascular Radiology and Intervention: Endorsed by the American College of Cardiology Foundation, the North American Society for Cardiac Imaging, and the Society for Cardiovascular Magnetic Resonance Circulation, December 11, 2007; 116(24): 2878 - 2891. [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]
|
|
|
|
 |
|
 R. B. Schwope, H. J. Alper, A. D. Talenfeld, E. I. Cohen, and R. A. Lookstein MR Angiography for Patient Surveillance After Endovascular Repair of Abdominal Aortic Aneurysms Am. J. Roentgenol., April 1, 2007; 188(4): W334 - W340. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
