| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Vascular and Interventional Radiology |
1 From the Department of Nephrology (J.H.M.S., W.A.M.A.v.d.M., P.J.B.), Department of Radiology, Image Sciences Institute (C.B., O.E.H.E., C.J.G.B., J.J.Z., W.P.T.M.M.), and Julius Center for Health Sciences and Primary Care (S.K.), University Medical Center Utrecht, Heidelberglaan 100, 3584 CX Utrecht, the Netherlands. Received May 2, 2001; revision requested June 18; final revision received April 5, 2002; accepted May 8. J.H.M.S. supported by grant C.97.1643 from the Dutch Kidney Foundation. Address correspondence to C.J.G.B. (e-mail: c.j.g.bakker@azu.nl).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Twenty-two accesses (14 arteriovenous grafts [AVGs], eight arteriovenous fistulas [AVFs]) in 18 consecutive patients were imaged with flow-interrupted contrast-enhanced MR angiography and subsequent conventional DSA. MR image quality was assessed as excellent, good, or nondiagnostic. Anastomotic diameters in AVGs and postanastomotic diameters in AVFs were measured in consideration of an adjacent normal segment. Reductions in the diameter of the lumen and interobserver differences were analyzed with method comparison as described by Bland and Altman and expressed as the mean difference with its 95% confidence limits (CLs) (mean ± 2 SDs).
RESULTS: Image quality obtained with flow-interrupted contrast-enhanced MR angiography was considered excellent in 73% (16 of 22) and good in 23% (5 of 22). Method comparison analysis between MR angiography and DSA indicated a mean difference of 3.2% (95% CLs: -26.7%, 33.1%) for observer 1 and 4.1% (95% CLs: -23.8%, 32.1%) for observer 2. Interobserver analysis at MR angiography indicated a mean difference of 3.2% (95% CLs: -28.8%, 35.2%), and that at DSA indicated a mean difference of 3.6% (95% CLs: -9.4%, 16.7%).
CONCLUSION: Image quality and anatomic depiction with flow-interrupted contrast-enhanced MR angiography and with DSA were comparable.
© RSNA, 2002
Index terms: Dialysis, shunts, 91.4539 • Digital subtraction angiography, comparative studies, 91.122, 91.12942 • Fistula, arteriovenous, 91.457 • Grafts, interventional procedures • Magnetic resonance (MR), vascular studies, 91.12942, 91.12943, 91.12944
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
With DSA, however, there are several drawbacks: the radiation load for the patient and radiology personnel, the use of potentially nephrotoxic iodinated contrast agents, and the limited spatial information (14). Magnetic resonance (MR) angiography does not have these disadvantages and could be an attractive alternative. In addition, MR angiography offers the possibility to measure access flow. Access flow has been suggested as a better parameter for impending vascular access failure than anatomic information alone (6,12,15).
However, a major problem with flow-based approaches to MR angiography, such as phase-contrast and time-of-flight imaging, has been the frequent occurrence of flow artifacts in regions with disturbed flow that complicate interpretation of the MR angiograms. Especially near stenoses but also near bends and cusps (16), the appearance of the lumen is distorted by a combination of dephasing, displacement, and, for time-of-flight methods, saturation artifacts (17). Application of phase-contrast and time-of-flight methods to imaging hemodialysis accesses (ie, both AVFs and AVGs) also is associated with the same types of inaccuracies (18,19), but interpretation of the angiograms is further complicated by the broad range of flow rates, roughly 100–3,000 mL/min, that occurs in these hemodialysis accesses (20). As flow disturbances tend to increase with flow rate, signal voids may easily arise at mild stenoses or sharp-angled anastomoses when a high flow rate is present (21).
Contrast material–enhanced MR angiography is less sensitive to these artifacts (22) and has been reported to improve hemodialysis access visualization when compared with time-of-flight and phase-contrast imaging (23). However, flow-related artifacts remain present under the extreme flow conditions that occur in hemodialysis accesses (24). To eliminate these artifacts, we recently reported the use of a cuff, as used with DSA, to obstruct the blood flow temporarily (24–26). The technical details of this flow-interrupted contrast-enhanced MR angiographic imaging technique have been described (24).
The purpose of our study was to compare flow-interrupted contrast-enhanced MR angiography with conventional DSA for hemodialysis access imaging.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Imaging Techniques
MR angiography.—All examinations were performed by one of two radiologists (O.E.H.E., J.J.Z.) with a 1.5-T clinical imager (Gyroscan ACS-NT; Philips Medical Systems, Best, the Netherlands). Patients were imaged in the supine position and entered the unit feet first to facilitate communication with the physician at the front of the imager. A 17-gauge plastic needle (Clampcath; Medikit, Tokyo, Japan) was inserted in the hemodialysis access and fixed with tape.
The needle was placed in the AVG at the usual location of the arterial needle for hemodialysis. The difference between the usual dialysis arterial needle and our plastic needle was that our plastic needle pointed toward the venous anastomosis. In the AVF, the plastic needle was located at the usual position of the arterial needle, and, similarly, pointed toward the arterial anastomosis. The function of the vascular access was evaluated by using nontriggered two-dimensional phase-contrast MR flow measurements (27). Access flow was measured in both limbs of the AVG and in the dilated vein of the AVF. The imaging plane at MR imaging for flow measurements was oriented perpendicular to the vessel of interest on the coronal and sagittal surveys. Access flow was measured in a single section, with section thickness of 8 mm, repetition time msec/echo time msec of 16/9.5, flip angle of 7.5°, field of view of 256 x 256 mm, with eight signals acquired, a matrix of 256 x 256, and velocity encoding of 250 cm/sec. In AVGs, volume flow was calculated as the average of arterial and venous flow measurements.
Then, two syringes containing 20 or 30 mL of gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) diluted in saline to a 20:1 concentration were connected to the needle through 100-cm Luer-Lok tubing. A cuff that would not cause artifacts with MR imaging was wrapped around the upper arm and fixed with a strap. A rectangular surface coil of 10 x 40 cm was positioned on the access of the patient for signal reception.
First, three orthogonal two-dimensional phase-contrast surveys were obtained. Filling of the access with contrast material was monitored by using a fast two-dimensional spoiled gradient-echo sequence with complex mask subtraction, and one image was obtained every 0.9 second. Care was taken to select an imaging plane that showed the arterial and venous limbs of the access well separated. The images were reconstructed immediately and were presented to the radiologist (O.E.H.E.) on a liquid crystal display screen in the imaging room. A three-dimensional (3D) image of the access was acquired by using a spoiled gradient-echo sequence with centric k-space order. Imaging parameters included the following: 6.1/1.8, a flip angle of 45°, a field of view of 400 x 100 mm, and an acquisition matrix of 512 x 100. The volume consisted of 64 1.0-mm partitions that were zero-filled and reconstructed every 0.5 mm. The volume had a sagittal orientation to prevent overlap from the trunk. More sections were added, if necessary, to include the left-right extent of the access. Imaging time was 38–48 seconds.
We first reduced access flow by inflating the cuff to diastolic blood pressure. Then, contrast material was injected manually until the access downstream of the puncture site was adequately filled. Next, the flow in the access was blocked by increasing the cuff pressure to well above the systolic pressure, which enabled retrograde filling of the upstream part of the access, the arterial anastomosis, and the native artery. When the filling of the access was seen as sufficient and stable on the complex subtraction images, we proceeded to the 3D acquisition. After completion of the acquisition, the cuff was released.
The cuff was inflated for an average of 82 seconds per series (range, 69–93 seconds), and this time included that for the filling of the access and the subsequent 3D imaging. Cuff inflation and contrast material injection were well tolerated in all patients.
In some early experiments with this new technique, we observed contrast material layering along the vessel wall. After adjusting the technique by injecting the contrast material solution more rapidly (at a rate of about 2 mL/sec), this artifact disappeared. The time required for an entire MR angiographic session was measured by adding the patient preparation time to the time between obtaining the first scout image and the last 3D volume image.
DSA.—DSA was performed with dedicated angiographic equipment (DVI-s and Integris 3000; Philips Medical Systems). The DSA studies were performed by experienced radiologists (O.E.H.E., J.J.Z.) who were aware of the clinical information regarding vascular access function and who had access to findings of previously performed DSA studies. Parameters included a matrix size of 1,024 x 1,024 and a field of view of 25–38 cm. Our standard DSA imaging strategy included at least four DSA series in AVFs and five in AVGs. The needle that was used for MR angiography also was used for DSA contrast material injection.
In one series, the loop of the graft was visualized. In the second series, the arterial anastomosis was visualized with injection of iodinated contrast agent (Ultravist 300; Schering, Berlin, Germany) into the arterial limb of the graft during compression of the cuff that was placed on the upper arm to approximately 200 mm Hg. This ensured adequate filling of the arterial anastomosis. Then the venous anastomosis and part of the upper arm veins were visualized. Two further series were performed to visualize the veins up to the superior vena cava. If the anastomoses were projected on top of each other, which made adequate diameter assessment impossible, additional projections of the anastomotic areas were obtained. A total of approximately 20 mL of contrast agent was needed for the procedure. The contrast agent was injected manually at a rate of approximately 3 mL/sec.
Image Analysis
MR image quality was assessed as excellent, good, or nondiagnostic. Excellent image quality referred to an image that had no layering artifacts. Good image quality referred to an image that had minimal layering artifacts but not to the extent that they affected the overall ability of the radiologist to analyze the image. Nondiagnostic image quality referred to an image that had severe layering artifacts that hampered image analysis. Both DSA and MR angiographic multiplanar reformatted images were reviewed on separate occasions in random order by two independent experienced vascular radiologists. For DSA images, all lumen diameters were measured on the hard copies by using a mechanical caliper with a digital display (PAV Electronics, Vaduz, Liechtenstein). For MR angiographic images, the data were presented on a console (EasyVision; Philips Medical Systems). The lumen diameters were measured from vessel cross sections that were created by multiplanar reformatting of the volume data perpendicular to the vessel axis.
For comparison, we considered the vascular access segments that were easily definable and were responsible for most of the clinical problems. For AVFs, we scored the narrowest diameter near the anastomosis (within 3 cm downstream) and compared it with the diameter of a normal segment of the vein downstream. This resulted in one ratio number for the AVFs, and it was from this ratio number that the relative degree of stenosis was calculated.
For AVGs, the narrowest diameter of the arterial anastomosis was divided by the diameter of a normal arterial graft segment near the anastomosis, and the diameter of the narrowest segment of the venous anastomosis was divided by the diameter of a normal segment of the graft that was close by. Therefore, two ratio measurements were calculated per AVG per observer per imaging method. The degree of stenosis was calculated from these ratios. The scoring method was used only for comparison purposes and not to assess whether a stenosis was pathophysiologic (eg, caused by intimal hyperplasia) or not (eg, caused by tapering of the arterial anastomosis in AVGs).
The maximum percentage of stenosis was determined with the following calculation:
|
Statistical Analysis
The imaging methods were compared by analyzing the difference between the measurements at each individual data point, as described by Bland and Altman (28). Similarly, interobserver agreement was analyzed by using this comparison method (28). We calculated the mean differences of each data point as an estimate of the average bias of one method or observer relative to the other. Additionally, for each comparison, we defined the 95% confidence limits (CLs) (ie, mean ± 2 SDs).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
For AVFs, reductions in seven lumina were measured for each imaging method per observer. For AVGs, MR angiographic stenotic ratios could be calculated in 21 segments (28 expected), and seven could not be measured because of the following reasons: three venous anastomoses were extended to the axilla and were therefore not imaged, three arterial anastomoses were not depicted because fill-up was not complete, and one venous anastomosis was compressed under the cuff during MR angiographic examination. DSA stenotic ratios could be calculated in 24 segments (28 expected); four ratios could not be calculated because the AVGs were extended to more proximal veins, and these segments were not adequately depicted. Mean access flow measured with MR angiography was 890 mL/min ± 449 in AVGs and 677 mL/min ± 384 in AVFs. Examples of the image quality for AVGs and AVFs are shown in Figures 1 and 2, respectively.
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Compared with DSA, flow-interrupted contrast-enhanced MR angiography had an excellent correlation when the most problematic areas (ie, the anastomoses) were evaluated. This finding was expected, since with both imaging methods the lumen of the graft filled up with contrast agent.
In previous studies in which non–contrast-enhanced MR angiography in vascular accesses (18,19) was compared with contrast-enhanced non–flow-interrupted MR angiography in other vascular beds (29), results indicated an overestimation of stenosis assessment. In this study, findings indicated that a slightly higher degree of percentage of stenosis (on average, 3.7%) that is measured with MR angiography, when compared with that measured with DSA, still exists. Possibly, reduction in the diameter of the lumen is underestimated with DSA; this underestimation is conceivable, since the percentage of stenosis can be assessed by using DSA in only one or two directions. Another explanation for this phenomenon could be that with MR angiography an overestimation of the real stenosis results because of the limited spatial resolution (30). However, the differences between both techniques were minimal.
Also, both the MR angiographic source image scores and the DSA image scores indicated a good interobserver agreement, with a mean interobserver difference of 3.2% at MR angiography and 3.6% at DSA. The 95% CLs of the interobserver DSA data range probably are smaller than those of the MR angiographic data range, because of the experience of the radiologists in examining and scoring DSA images rather than in scoring MR angiographic source images. However, with more experience in scoring source images, the 95% CLs of the interobserver MR angiographic data range will probably become smaller.
Findings of this study further suggest that direct injection strongly reduces the required contrast agent dose and, thus, offers the possibility that the examination may be repeated without the need for large contrast agent doses. The issue of synchronizing acquisition and peak contrast concentration, which is a problem of 3D contrast-enhanced MR angiography, is circumvented.
MR angiography has a number of benefits with respect to DSA, which is the present method of choice for anatomic hemodialysis access evaluation. The images acquired are 3D and can be reformatted to show the access from arbitrary projection angles. As stated previously in this article, this is useful, because it is difficult to define a standard view for these vascular constructs of varying geometry (26). A major asset of MR imaging is its capability of aiding in a functional evaluation of the hemodialysis access by facilitating measurement of the flow rate (27). Low values of the access flow rate are indicative of increased risk of thrombosis (5,31–33) and may prompt initiation of radiologic intervention in case of an underlying stenosis (6). Ideally, when a low value of access flow is measured and the underlying stenosis is located by using MR angiography, a PTA procedure is performed during the same procedure. Some attempts have been made to provide this all-inclusive approach with MR angiography (21).
The contrast agent used, gadopentetate dimeglumine, has an excellent safety profile and is considered at least as safe as iodinated contrast agents (34). Gadopentetate dimeglumine is not contraindicated for use in patients with impaired renal function (35,36) and can be removed by using hemodialysis (37). Finally, patients and physicians are not exposed to ionizing radiation.
A clear advantage of DSA compared with MR angiography is that the corrective intervention can be performed on the spot, should a hemodynamically significant stenosis be detected. Techniques for MR imaging–guided endovascular interventions still have to be developed further to provide a clinically attractive alternative (38). In addition, a complete angiographic evaluation of the graft or fistula at DSA includes visualization of the runoff vessels to the superior vena cava (26). Findings in several studies indicate that in a small number of patients stenoses are found in the venous outflow tract (7,39). Findings in a report (40) indicate that images of the runoff veins of the extremities can be obtained by using contrast-enhanced MR angiography as well. The acquisition is started immediately after intravenous injection of the contrast agent, instead of waiting for the venous phase following the first arterial passage.
More importantly, this study was performed primarily to determine whether MR angiography is comparable with DSA. Therefore, we chose to analyze certain predefined segments of the vascular accesses. MR angiography and DSA were compared with respect to measurement of these segments. Whether or not they were clinically important is not part of this study. In the K/DOQI guidelines for vascular access (1) of the National Kidney Foundation, it is recommended that only hemodynamically impaired accesses (ie, those with access flows lower than 600 mL/min) need to be subjected to further examination and intervention. Data from recent studies by Van der Linden et al (41) and by Ahya et al (42) suggest that there is a weak correlation between access flow and degree of stenosis. Therefore, the decision about whether or not to intervene by using PTA depends mostly on the hemodynamic significance of a stenosis (ie, on the level of access flow).
In conclusion, selective flow-interrupted contrast-enhanced MR angiography can be used to obtain images of hemodialysis AVFs and AVGs that are of excellent quality and free of flow artifacts. The high image quality of the MR angiograms is comparable with that of DSA images. In combination with MR imaging flow quantification and imaging of the outflow tract, our approach may provide a complete anatomic and functional evaluation of hemodialysis accesses with MR imaging.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, J.H.M.S.; study concepts, C.J.G.B., J.J.Z., W.P.T.M.M.; study design, J.H.M.S., P.J.B., C.J.G.B., J.J.Z., W.P.T.M.M.; literature research, J.H.M.S.; clinical studies, J.H.M.S., C.B., O.E.H.E., W.A.M.A.v.d.M., J.J.Z.; data acquisition, J.H.M.S., C.B., O.E.H.E.; data analysis/interpretation, J.H.M.S., C.B., O.E.H.E., J.J.Z., C.J.G.B.; statistical analysis, J.H.M.S., C.B., O.E.H.E., S.K.; manuscript preparation, J.H.M.S., C.B., O.E.H.E., W.A.M.A.v.d.M., P.J.B., C.J.G.B., J.J.Z., W.P.T.M.M.; manuscript definition of intellectual content, J.H.M.S., C.B., J.J.Z.; manuscript editing, J.H.M.S., C.B., O.E.H.E., C.J.G.B.; manuscript revision/review, J.H.M.S., C.B., C.J.G.B., P.J.B., S.K.; manuscript final version approval, J.H.M.S., C.B., O.E.H.E., W.A.M.A.v.d.M., P.J.B., C.J.G.B., J.J.Z., W.P.T.M.M.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  F. Stepansky, E. M. Hecht, R. Rivera, L. E. Hirsh, B. Taouli, M. Kaur, and V. S. Lee Dynamic MR Angiography of Upper Extremity Vascular Disease: Pictorial Review RadioGraphics, January 1, 2008; 28(1): e28 - e28. [Abstract] [Full Text]
|
|
|
|
 |
|
 C. Leon and A. Asif Arteriovenous Access and Hand Pain: The Distal Hypoperfusion Ischemic Syndrome Clin. J. Am. Soc. Nephrol., January 1, 2007; 2(1): 175 - 183. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Zhang, E. M. Hecht, T. Maldonado, and V. S. Lee Time-Resolved 3D MR Angiography with Parallel Imaging for Evaluation of Hemodialysis Fistulas and Grafts: Initial Experience. Am. J. Roentgenol., May 1, 2006; 186(5): 1436 - 1442. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 C. L. Froger, L. E. M. Duijm, Y. S. Liem, A. V. Tielbeek, A. B. Donkers-van Rossum, P. Douwes-Draaijer, P. W. M. Cuypers, J. Buth, and H. C. M. van den Bosch Stenosis Detection with MR Angiography and Digital Subtraction Angiography in Dysfunctional Hemodialysis Access Fistulas and Grafts Radiology, January 1, 2005; 234(1): 284 - 291. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
