- Email: [email protected]
Screening of renal artery stenosis: Assessment with magnetic resonance angiography at 1.0 T
Magnetic
ELSEVIER
l
Resonance
Imaging,
Vol. 14, No. 9. pp. 1033-1041, 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0730-725X/96 $15.00 + .OO
PI1 SO730-725X( 96) 00097-X
Original Contribution SCREENING OF RENAL ARTERY STENOSIS: ASSESSMENT WITH MAGNETIC RESONANCE ANGIOGRAPHY AT 1.0 T
JEAN-PIERREI,AISSY,* MOURAD BENYOUNES," OLIVIER LIMOT, * ANNE CINQUALBRE,* HAKIM BENAMER,? SABINE KENOUCH,$ MARIE-CECILE HENRY-FEUGEAS,* BEATRICE FALISE,~ SYLVIE CHILLON,* PAUL E. VALERE,~ AND ELTSABETHSCHOUMAN-CLAEYS* Departments of *Radiology, tcardiology, and $Nephrology, HGpital Bichat, 46 rue Henri Huchard, 75877 Paris Cedex 18, France; and §Siemens Medical Systems, 39-47 Boulevard d’Omano, 93527 Saint-Denis CCdex 2 France The results of MR angiography at 1.0 T with digital intraarterial angiography in the screening of patients with suspected renal hypertension were compared. In the first phase of the study, 10 volunteers underwent examination with both two-dimensional (2D) with traveling saturation time-of-flight (TOF) magnetic resonance angiography (MRA) with various parameters to develop a protocol for evaluation of the renal arteries. In the second phase, 36 patients with suspected renovascular hypertension underwent both 2D TOF MRA and intraarterial digital angiography to evaluate the clinical value of MRA. The degree of stenosis was graded with a two-point scale. In volunteers, using 2D acquisitions, C/N ratios indicated the best flip angle as being 55” (p = .02). MRA showed 100% (70/70) of all main arteries and 86% (6/7) of all accessory renal arteries seen on angiography. MRA had a sensitivity of 94% (15116) and a specificity of 98% (60/61) for detection of stenoses of greater than 50% present in 14 patients. 2D-TOF MRA at 1.0 T shows promise in the noninvasive diagnosis of patients with suspected renovascular hypertension. Copyright 0 1996 Elsevier Science Inc. Keywords: Renal arteries; MR studies; Renal arteries, Stenosis or obstruction; Magnetic resonance (MR), Comparative studies; Magnetic resonance (MR), vascular studies.
INTRODUCTION
trast-enhanced angiography is associated with risks emerging from contrast media and catheterization, a noninvasive imaging technique that would depict renal artery stenosis with a high level of accuracy would be of great clinical value. Duplex Doppler ultrasonography, I-3 captopril radionuclide studies, ‘*’ spiral CT,4,5 and magnetic resonance imaging have been in the field of several investigations. The capability of magnetic resonance angiography (MRA) to depict renal artery disease has been emphasized in several studies, these studies being performed at a field strength of 1.5 T.h-22 The purpose of this study was to evaluate the clinical usefulness of MRA at 1.0 T by direct comparison of MR angiograms and intraarterial digital arteriograms from patients suspected for renovascular hypertension.
Renovascular hypertension (RVH) is defined as hypertension caused by renal artery stenosis (RAS). In addition to hypertension, RAS may cause renal insufficiency. Renal vein sampling is widely considered to be an unreliable indicator of RVH.’ Renal arteriography is the definitive anatomic gold standard, whereas captopril renal scintigraphy is considered to be the definitive functional measure at this time.“’ Improved therapeutic efficacy of this disease has been demonstrated with the large use of renal percutaneous transluminal angioplasty, percutaneous metallic stent placement, and surgery. 1-Z Because RAS is a readily treatable cause of hypertension, an early diagnosis of this condition is essential in order that the patient may benefit from any revascularization procedure. However, because con1019195;ACCEFTED 4115196. Address correspondence to J. P. Laissy, Department of Radiology, Centre Hospitalier et Universitaire Bichat-Claude
Bernard, 46 rue Henri Huchard, 75877 Paris cedex 18, France.
RECEIVED
1033
Magnetic Resonance Imaging l Volume 14, Number 9, 1996
1034
MATERIALS
AND METHODS
Between November 1993 and June 1995,36 consecutive patient (24 males and 12 females, aged 32-67 years [mean, 43 years] ) with strong suspicion of renal hypertension on the basis of clinical grounds, usually because of a baseline diastolic blood pressure above 110 mm Hg or the presence of treatment-resistant hypertension, were studied with both renal angiography and MRA. Subjects were eligible patients who were scheduled for renal angiography, were willing to participate in the study, and could be scheduled for MRA on the week of their angiographic examination. Nine patients had renal insufficiency, with serum creatinine exceeding 150 pmollliter. Imaging Techniques
MRA was performed within l-4 days from diagnostic angiography. The final diagnoses were based on results of intraarterial digital angiography used as the gold standard in all patients. All angiograms consisted of intraarterial flush aortograms, and digital recording was used exclusively. Intraarterial DSA of the abdominal aorta was performed via a 5-French pigtail catheter positioned at the level of renal arteries; contrast material (Hexabrix 320, laboratoire Guerbet, Aulnay sous bois, France) was administered at 20 ml/s for 2.5 seconds. DSA examinations systematically included both frontal and oblique views to clearly analyze renal arteries and renal ostia. No selective angiograms were performed. MR studies were performed using a superconducting magnet (Magnetom, Siemens, Erlangen, Germany) operating at 1.0 T. In all cases, examinations were performed using the body coil. To visualize the origin of renal arteries, all subjects were first examined in the coronal plane, by means of SE T’weighted sequences (TR/TE 450/15 ms, g-mm section thickness). Volunteers In the first phase of the study, 10 volunteers underwent two-dimensional (2D) MRA using a gradient-recalled echo (GRE) time-of-flight (TOF) technique with low angle shot, gradient spoiling and first-order motion flow compensation in slice select and readout direction (FLASH), and a traveling venous saturation pulse. The following imaging parameters were systematically varied: field-of-view (FOV) , 3035 cm; section thickness, 3-5 mm; flip angle, 35-60”; repetition time, 32-50 ms; number of phase encoding steps, 128-192; and number of signal averaged, 2-4. The TE was the shortest possible authorized by our machine (9 ms). Flow-compensated gradients and a traveling venous saturation pulse (with a saturation slab eight times as thick as the imaged section, the center position of the sat band being shifted from the
imaging section position by eight times the section thickness) were used in all studies. An additional stationary oblique-coronal presaturation pulse was placed posteriorly on the left kidney to exclude the signal intensity from the left renal vein. Patients In the second phase of the study, 36 patients underwent MRA examinations using the same 2D GRE sequence than that used in volunteers. Acquisitions consisted in multisection sequential 2D FLASH sequences in the axial plane from the celiac abdominal aorta through the infrarenal aorta. One series of multisection axial sequences used a fixed TRITE of 3519 ms and 55” flip angle. Twenty-four successive sections were acquired, using a FOV of 350 mm, a 140 X 256 acquisition matrix, a 4-mm section thickness with a 2mm overlap, and two excitations, for an average acquisition time of 4 mitt, 08 s. In case of insufficient flow signal within renal arteries, the same 2D TOF sequences were repeated successively in two separate axial oblique planes (i.e., axial toward sagittal) in order that each sequence was the most perpendicular to each renal artery (Fig. 1) . Individual MRAs were then postprocessed using a maximum intensity projection (MIP) algorithm. Five to six projection images with angles incremented by steps of 15” from right lateral to left lateral were displayed on film for interpretation using multiplanar reformation. Data Analysis
In the first phase of the study, two radiologists (JPL, MBY) evaluated the 2D techniques in volunteers by measuring the length of arteries exhibited and the contrast-to-noise (C/N) ratio of aorta and renal arteries by means of regions-of-interest of at least 25 pixels. Because the imaged volume was entirely within the patient, background noise could not be measured outside the patient. Thus, background noise was measured on a homogeneous region laterally within the image volume, as previously recommended.2’ In the second phase of the study, three radiologists (JPL, OL, MBY) reviewed separately the MRAs without knowledge of the angiographic results. Digital arteriograms were also interpreted without knowledge of the MRA results in a blinded fashion by the same radiologists at a different time. Readers nos. 1 and 2 were experienced radiologists, whereas reader no. 3 was a resident in training. Both the axial images and MIP reconstructions were reviewed before a final opinion was rendered. In cases of discrepancies between the readers, the final MRA diagnosis was obtained by consensus. Stenoses were assessed subjectively. Be-
RenalMRA at 1.0 T 0 J.-P.
LAISSY ETAL.
1035
Fig. 1. Patient with aortic atheroma and proximal right artery stenosis. (A) Poor-quality MRA from 2D axial acquisitions precludes accurate assessmentof renal arteries. (B and C) MFLA from 2D right (B) and left (C) oblique axial acquisitions improve visibility of stenosed (arrow) right (B) and irregular left (C) renal arteries. Stair-step effect and signal drop-out from inplane artifact (C) are seen. (D) The corresponding angiographic image confirms a significant right artery stenosis and widely patent left renal artery.
cause the objective was merely to screen significant renal artery stenoses, stenosis was graded with a two-point scale on MRA and angiography: normal and nonsignificant proximal renal artery stenoses of less than 50% and significant stenoses of more than
50% in diameter, as well as multiple stenoses and thrombosis. Accessory arteries were assessed as normal or involved. On MRA, significant stenosis corresponded to a localized luminal narrowing with or without
flow signal inhomogeneities
flow void; thrombosis
or localized
was diagnosed as a complete
absence of flow signal regardless of the diameter of the artery. Presence or absence of accessory arteries was assessed.
Statistical Analysis In volunteers, signal-to-noise ratios expressed as a function of angle and NEX were presented as mean values + SD. Differences
in the aortic and renal SI
and C/N ratios for each of the sequences were evaluated with the Student’s t-test. For the comparison between MRA and angiographic data in patients, sensitivity, specificity, and predictive accuracy of MRA for the detection of significant renal artery stenosis were based on standard definitions. The null hypothesis was rejected at the 95% confidence level, considering p < .05 to be significant.
Interobserver variability between the readers was determined by using the kappa statistic.23
Magnetic Resonance Imaging l Volume 14, Number 9, 1996
1036
RESULTS Volunteers The first phase of our study was directed at development of a clinically effective MRA protocol for evaluation of renal arteries. C/N ratios indicated the best flip angle as 55” (p = .02). The number of excitations (4 vs 2) did not modify significantly C/N ratio (p = .08). Section thickness (3 vs 5 mm) did not influence significantly C/N ratio. However, even when overlap of 2 mm was used, a “stair-step” effect was still seen in two cases by using 5-mm-section thickness. Hence, using the 2D technique, the best imaging parameters were TR/TE 35/9 ms, flip angle 55”, NEX=2, FOV=350 mm, voxel size = 4 X 1.37 x 1.87 mm, 24 overlapping sections for an overall acquisition time of 4.08 minutes. With these parameters, maximal lengths measured on 2D MRAs were 1.5-7 cm (mean + SD; 3.8 t 2 cm) for the right renal artery and 2-6 cm (3.7 t 1.8 cm) for the left renal artery. Intrarenal branches were seen in 70% (7/10) in the right renal artery and in 80% (8/ 10) in the left renal artery (Fig. 2). Patients All MRAs were available for interpretation. A total of 77 renal arteries was detected by using digital angiography. MRAs showed 100% (70/70) of the main renal arteries and 86% (6/7) of the accessory renal arteries seen at angiography. Stenoses of the proximal main renal artery greater than 50% were visualized in 14 vessels in 12 patients at angiography (Table 1)
and two significant accessory artery stenoses in two other patients, yielding a prevalence of RAS of 39%. Renal artery involvement was related to fibromuscular dysplasia in one patient. Of the two isolated accessory renal artery stenoses identified on angiography, only one was seen on MRAs (Table 2). No stenosis was present in the accessory renal artery missed by MRA. The sensitivity of detection of main renal artery and accessory artery stenoses was 100% and 50%, respectively (14/14, CI 1.0, 0.92; and l/2, CI 0.8, 0.12) and the specificity was 98% and lOO%, respectively (55/56, CI 0.99, 0.83; and 4/4, CI, 1.0, 0.46). MRA had a sensitivity of 94% ( 15/ 16) and a specificity of 98% (60/61) in the overall detection of significant stenoses, for a positive and negative predictive values of 98% and 94%, respectively. MRA was 93% sensitive (13/14, CI 0.94, 0.85) and 95% specific (21/22, CI 0.97, 0.89) for identifying patients with at least one significant renal artery stenosis (Figs. 3-5). Discrepancies between readers were observed in two cases (one stenosis considered normal by one reader and with a significant stenosis by the two others, and one stenosis judged nonsignificant by one reader and significant by the two others). The interobserver variability between the readers was 0.9 1. Reader experience involved significant greater accuracy only between reader no. 1 and reader no. 3 (p = 0.04). DISCUSSION Ultrasound and radionuclide studies are established noninvasive screening tests in renovascular
Fig. 2. MIP reconstruction obtained from 2D axial acquisitions in a volunteer. Renal arteries are well seen until renal pelvis. An area of signal loss is seen in the right middistal renal artery where the renal pelvis crosses (arrow).
Renal MRA at 1.O T 0 J.-P. LAISSY
Table 1. Sensitivity and specificity of MRA in the diagnosis of renal artery stenosis, according to the readers Reader Reader Reader Concurrent 1 2 3 interpretation Main renal artery stenosis Sensitivity (%) Specificity (%) Main plus accessory renal artery stenosis Sensitivity (%) Specificity (%)
100 95
91 98
91 98
100 98
‘94 ‘97
85 98
79 98
94 98
hypertension. Duplex Doppler is affected of a broad range of accuracies, with accuracy estimated between 79 and 91% according to the different studies.3 However, this technique is susceptible to a variety of problems such as operator skill, interference by abdominal bowel gas, and poor reproducibility between institutions, and nearly one-third of examinations are noncontributive.2,3 Renal scintigraphy needs to be performed in conjunction with angiotensin-converting enzyme inhibitors (e.g., captopril) administration. Captopril renal scintigraphy provide accurate functional information upon renal function, mainly relative to its local renal effect.‘,* Captopril decreases postglomerular efferent tone and thus causes an important reduction in the glomerular filtration rate of kidneys with renal artery stenosis.’ Captopril renography has respective sensitivity and specificity rates of approximately 90%.’ Moreover, the value of captopril renography is limited in patients with renal insufficiency, due to decreased renal uptake and increased background activity.‘4 More accurate noninvasive tests are a major issue in the screening of RVH. Early promising results have been obtained with spiral CT.4.5 The sensitivity of spiral CT for detection of RAS?SO% has recently been reported as high as 98% and the specificity as high as 94%.25 However, this technique involves large amounts of contrast medium, which can have deleterious effects in patients with renal insufficiency. Several investigators focused on the diagnostic capabilities of vascular MRI in the study of aorta and renal arteries.6-24 These studies were mainly obtained on MR devices operating at 1.5 T. In these studies, abdominal acquisitions were performed by means of either three-dimensional (3D) phase-contrast angi2D contiguous overlapping slice TOF -graphy, 6.9.‘2~‘s~‘7 acquisitions, 7,10’62D breath-held TOF acquisitions, *- “Z
1037
ET AL.
or, more recently, 3D acquisitions with use of multiple overlapping slab acquisitions, ‘8,22with fat saturation 2o or with incremented flip angles.*’ Our results obtained at 1 .O T using 2D overlapping techniques 5,9~1o compare favorably with the 87-100% sensitivity and 92-97% specificity of MRA for detection of significant stenoses,8-10~‘2,‘6~‘8~24 even if our results are limited because of the small number of diseased vessels. Moreover, compared with other MRA techniques, the time provided by 2D acquisition is short.9*16 The potential for improving on the results here by use of more state of the art techniques includes the development of new sequences,“0B22such as fat suppressed 2D TOF MRA done immediately after intravenous administration of an MR contrast agent*’ and hardware implementation of echo planar imaging.7 Detection of accessory renal arteries with MRA remains problematic. The reported range of accessory arteries depicted by MRA is comprised between 50 and 100~~.8.9,'2,'8,'4,26 Our sensitivity in detecting significant accessory artery stenosis was 50%, when compared with the 100% sensitivity in detecting significant main renal artery stenosis. In our study, the acquisitions did not include the aortic bifurcation and proximal common iliac arteries, despite the fact that accessory renal arteries can arise quite low, even from iliac arteries. This is a difficult diagnostic problem in cases of RVH and in the preoperative evaluation of kidney donors.27 This study has some other limitations. A small number of stenoses is included, as well as a small number of accessory arteries. A 39% prevalence favors high sensitivity and high specificity, compared with the expected varying estimates of prevalence, from 1% to 10% of patients with hypertension screened.28 Another drawback is that stenotic lesions are often overestimated; stenoses are more difficult to be evaluated in the MIP technique that in a single slice technique. Phase dispersion due to turbulent flow has been described as a potential pitfall in the estimation of stenosis, in particular near renal ostia. 7,9B1’,21 Although our work only tried to identify nonsignificant from significant artery stenoses, our results demonstrated that phase dispersion in case of turbulences was a good indicator of the severity of stenoses, despite the fact that renal artery sinuosities Table 2. Analysis of accessory arteries by MRA (concurrent interpretation)
MRA Normal Significant
Normal
Significant stenosis
1
4 stenosis
1
Not seen
1
Magnetic ResonanceImaging 0 Volume 14, Number9, 1996
1038
(4
Fig. 3. Patient with suspected RVH. (A) Localized flow void 2 cm distal to the origin of left renal artery (arrow) on frontal MIP image (false-positive finding). (B and C) Axial MIP image (B) and frontal multiplanar reformatted image (C) show normal renal arteries. A flow void is present at the origin of the left renal artery on MIP image, corresponding to a reconstruction artifact (axial source images, not shown, displayed a patent left renal artery). (D) Angiography identifies no significant stenosis. Renin venous sampling was normal. by themselves can involve flow dephasing. This sign unfortunately lacks sensitivity. RAS may be caused by either atherosclerosis or fibromuscular dysplasia (FMD) . In these cases, successive stenoses may lead to a difficult evaluation of the degree of renal perfusion involvement. FMD can also involve the distal part of renal arteries, including intrarenal stenosis?’ Signal within renal arteries is reduced in patients with renal insufficiency who have reduced renal artery blood flow.“j Indeed, RAS has also been recognized as an important cause of impaired renal function.
The purpose of this study was only to evaluate MRA to assess significant from nonsignificant stenoses. Compared with carotid artery disease where the exact degree and morphology of the stenosis is the critical information to be obtained, the main information useful for treatment of RAS is the severity of RAS. The diagnosis of RAS currently relies on renal angiography. On the basis of this technique ascertained as a diagnostic standard, the criterion of a significant stenosis is yet polemic. Some investigators considered a 50% stenosis to be significant, yet
Renal MRA at 1.0 T 0 J.-P. LAISSY ETAL.
(4
Fig. 4. True positive findings of a significant left renal artery stenosis. (A) MRA suggests left renal artery occlusion (right renal artery better seen on other projections not shown). Source images (not shown) displayed a small patent left renal artery. (B) Angiography shows a small patent but severely diseased left renal artery.
perfusion pressure in a large artery is generally not reduced until the stenosis exceeds 70%.* Renal flow quantification by phase-contrast angiography has also been used l7 to overcome the limitation of TOFMRA by its inherent inability to predict the hemodynamic importance of a particular lesion. In conclusion, 2D TOF MRA at 1.0 T appears to be a simple accurate means in the screening of patients with suspected RVH. It could also be useful
in patients with renal disease who are at risk for contrast media toxicity. Among previous studies reported, the present one demonstrates that this field strength is not an obstacle in diagnosing renal artery disease. Acknowledgements-We thank Geraldine Patient and Marie-France Cahtrec for their secretarial assistance and Christine Besnard for her pictorial assistance.
Fig. 5. False-negative MRA in a case of accessory artery stenosis. (A) MRA displays a normal left accessory artery (arrow). (B) Angiography shows significant (arrowhead) left renal accessory artery stenosis (70%), with decreased left inferior polar nephrogram (large arrowheads).
REFERENCES 1. Hillman, B.J. Imaging advances in the diagnosis of renovascular hypertension. AJR 153:5-14; 1989. 2. Davidson, R.A.; Wilcox, C.S. Newer tests for the diagnosis of renovascular disease. JAMA 2683353-3358; 1992. 3. Middleton, W.D. Doppler US evaluation of renal artery stenosis: past, present, future. Radiology 184:307-308; 1992. 4. Galanski, M.; Prokop, M.; Chavan, A.; Schaefer, CM.; Jandeleit, K.; Nischelsky, J.E. Renal arterial stenoses: spiral CT angiography. Radiology 189: 185- 192; 1993. 5. Rubin, G.D.; Dake, M.D.; Napel, S.; Brooke Jeffrey
R. Jr.; McDonnell, C.H.; Sommer, F.G.; Wexler, L.; Williams, D.M. Spiral CT of renal artery stenosis: comparison of three-dimensional rendering techniques. Radiology 190: 181- 189; 1994. 6. Dumoulin, CL.; Yucel, E.K.; Vock, P.; Souza, S.P.; Terrier, F.; Steinberg, F.L.; Wegmuller, H. Two- and three-dimensional phase contrast MR angiography of the abdomen. J. Comput. Assist. Tomogr. 14:779-784; 1990. 7. Edelman, R.R. MR angiography: present and future. AJR 161:1-11; 1993. 8. m D.; Edelman, R.R.; Kent, K.C.; Porter, D.H.; Skillman, J.J. Abdominal aorta and renal artery stenosis evaluation with MR angiography. Radiology 174727-731; 1990.
Renal MRA at 1.0 T 0 J.-P. 9. Debatin, J.F.; Spritzer, C.E.; Grist, T.M.; Beam, C.; Svetkey, L.P.; Newman, G.E.; Sostman, H.D. Imaging of the renal arteries: value of MR angiography. AJR 157:981-990; 1991. 10. Kent, K.C.; Edelman, R.R.; Kim, D.; Steinman, T.I.; Porter, D.H.; Skillman, J.J. Magnetic resonance imaging : a reliable test for the evaluation of proximal atherosclerotic renal arterial stenosis. J. Vast. Surg. 13:311-318; 1991. 11. Lewin, J.S.; Laub, G;.; Hausmann, R. Three-dimensional time-of-flight MR angiography: applications in the abdomen and thorax. Radiology 179:261-264; 1991. 12. Vock, P.; Terrier, F.; Wegmtiller H.; Mahler, F.; Gertsch, P.: Souza, S.P.; Domoulin, C.L. Magnetic resonance angiography ‘of abdominal vessels: early experience using the three-dimensional phase-contrast technique. Br. J. Radiol. 64:10-16; 1991. 13. Arlart, I.P.; Guhl, L.; Edelman, R.R. Magnetic resonance angiography of the abdominal aorta. Cardiovasc. Intervent. Radiol. 15:43-50; 1992. 14. Yucel, E.K.; Magnsetic resonance angiography of the lower extremity and renal arteries. Semin. Ultrasound CT MRI 13:291-302; 1992. 15. Wolf, R.L.; King, B.F.; Torres, V.E.; Wilson, D.M.; Ehman, R.L. Measurements of normal renal artery blood flow: tine phase-contrast imaging vs clearance of paminohippurate. AJR 161:995- 1002; 1993. 16. Grist, T.M.; Kennel, T.W.; Sproat, I.A.; McDermott, J.C.; Wojtowycz, h4.M.; Flath, E.M.; Crummy, A.B. Prospective evaluation of renal MR angiography: comparison with conventional angiography in 35 patients. Radiology 189:190--196; 1993. 17. Debatin, J.F.; Ting, R.H.; Wegmtiller, H.; Sommer, F.G.; Fredrickson, J.O.; Brosnan, T.J.; Bowman, B.S.; Myers, B.D.; Herf kens, R.J.; Pelt. N.J. Renal artery blood flow: quantitation with phase-contrast imaging with and without breath holding. Radiology 190:371378; 1994. 18. Kaufman, J.A.; Geller, S.C.; Petersen, M.J.; Cambria, R.P.; Prince, M.R.; Waltman, A.C. MR imaging (including MR angiography) of abdominal aortic aneurysms: comparison with conventional angiography. AJR 163: 203-210; 1994.
LAISSV
ET AL.
1041
19. Hertz, S.M.; Holland, G.A.; Baum, R.A.; Haskal, Z.J.; Carpenter, J.P. Evaluation of renal artery stenosis by magnetic resonance angiography. Am. J. Surg. 168: 140-- 143; 1994. 20. Li, D.; Haacke, E.M.; Mugler III, J.P.; Berr, S.; Brookeman, J.P.; Hutton, M.C. Three-dimensional time-offlight MR angiography using selective inversion recovery RAGE with fat saturation and ECG-triggering : application to renal arteries. Magn. Res. Med. 31:414422; 1994. 21 Loubeyre, P.; Revel, D.; Garcia, P.; Delignette, A.; Canet, E.; Chirossel, P.; Genin, G.; Amiel, M. Screening patients for renal artery stenosis: value of three-dimensional time-of-flight MR angiography. AJR 162:847852; 1994. 22. Prince, M.R. Gadolinium-enhanced MR aortography. Radiology 191:155-164; 1994. 23. Cohen, J. A coefficient of agreement for nominal scales. Educ. Psychol. Meas. 20:37-46; 1960. 24. Yucel, E.K.; Kaufman, J.A.; Prince, M.R.; Bazari, H.; Fang, L.S.T.; Waltman, A.C. Time-of-flight renal MR angiography in patients with renal insufficiency. Magn. Reson. Imaging 11:925-930; 1993. 25. Galanski, M.; Prokop, M.: Chavan, A.; Schaefer, CM.; Jandeleit, K.; Nischelsky. J.E. Minimally invasive diagnosis of renal artery stenosis by spiral computed tomography angiography. Kidney Int. 48:1332- 1337; 1995. 26. Servois, V.; Laissy, J.P.; Feger, C.; Sibert, A.; Delahousse, M.; Baleynaud, S.; M&y, J. Ph.; Menu, Y. 2D MR TOF angiography without MIP reconstruction of renal arteries: a prospective comparison with angiography in 21 patients. J. Cardiovasc. Inter-vent. Radiol. 17:138-142; 1994. 27. Debatin, J.F.; Sostman, H.D.; Knelson, M.; Argabright, M.; Spritzer, C.E. Renal magnetic resonance angiography in the preoperative detection of supernumerary renal arteries in potential kidney donors. Invest. Radiol. 28:882-889; 1993. 28. Dean, R.H. Renovascular hypertension: an overview. In: R.B. Rutherford (Ed). Vascular Surgery, 3rd ed. Philadelphia: W.B. Saunders, pp. 1211- 1218; 1989.
ELSEVIER
l
Resonance
Imaging,
Vol. 14, No. 9. pp. 1033-1041, 1996 Copyright 0 1996 Elsevier Science Inc. Printed in the USA. All rights reserved 0730-725X/96 $15.00 + .OO
PI1 SO730-725X( 96) 00097-X
Original Contribution SCREENING OF RENAL ARTERY STENOSIS: ASSESSMENT WITH MAGNETIC RESONANCE ANGIOGRAPHY AT 1.0 T
JEAN-PIERREI,AISSY,* MOURAD BENYOUNES," OLIVIER LIMOT, * ANNE CINQUALBRE,* HAKIM BENAMER,? SABINE KENOUCH,$ MARIE-CECILE HENRY-FEUGEAS,* BEATRICE FALISE,~ SYLVIE CHILLON,* PAUL E. VALERE,~ AND ELTSABETHSCHOUMAN-CLAEYS* Departments of *Radiology, tcardiology, and $Nephrology, HGpital Bichat, 46 rue Henri Huchard, 75877 Paris Cedex 18, France; and §Siemens Medical Systems, 39-47 Boulevard d’Omano, 93527 Saint-Denis CCdex 2 France The results of MR angiography at 1.0 T with digital intraarterial angiography in the screening of patients with suspected renal hypertension were compared. In the first phase of the study, 10 volunteers underwent examination with both two-dimensional (2D) with traveling saturation time-of-flight (TOF) magnetic resonance angiography (MRA) with various parameters to develop a protocol for evaluation of the renal arteries. In the second phase, 36 patients with suspected renovascular hypertension underwent both 2D TOF MRA and intraarterial digital angiography to evaluate the clinical value of MRA. The degree of stenosis was graded with a two-point scale. In volunteers, using 2D acquisitions, C/N ratios indicated the best flip angle as being 55” (p = .02). MRA showed 100% (70/70) of all main arteries and 86% (6/7) of all accessory renal arteries seen on angiography. MRA had a sensitivity of 94% (15116) and a specificity of 98% (60/61) for detection of stenoses of greater than 50% present in 14 patients. 2D-TOF MRA at 1.0 T shows promise in the noninvasive diagnosis of patients with suspected renovascular hypertension. Copyright 0 1996 Elsevier Science Inc. Keywords: Renal arteries; MR studies; Renal arteries, Stenosis or obstruction; Magnetic resonance (MR), Comparative studies; Magnetic resonance (MR), vascular studies.
INTRODUCTION
trast-enhanced angiography is associated with risks emerging from contrast media and catheterization, a noninvasive imaging technique that would depict renal artery stenosis with a high level of accuracy would be of great clinical value. Duplex Doppler ultrasonography, I-3 captopril radionuclide studies, ‘*’ spiral CT,4,5 and magnetic resonance imaging have been in the field of several investigations. The capability of magnetic resonance angiography (MRA) to depict renal artery disease has been emphasized in several studies, these studies being performed at a field strength of 1.5 T.h-22 The purpose of this study was to evaluate the clinical usefulness of MRA at 1.0 T by direct comparison of MR angiograms and intraarterial digital arteriograms from patients suspected for renovascular hypertension.
Renovascular hypertension (RVH) is defined as hypertension caused by renal artery stenosis (RAS). In addition to hypertension, RAS may cause renal insufficiency. Renal vein sampling is widely considered to be an unreliable indicator of RVH.’ Renal arteriography is the definitive anatomic gold standard, whereas captopril renal scintigraphy is considered to be the definitive functional measure at this time.“’ Improved therapeutic efficacy of this disease has been demonstrated with the large use of renal percutaneous transluminal angioplasty, percutaneous metallic stent placement, and surgery. 1-Z Because RAS is a readily treatable cause of hypertension, an early diagnosis of this condition is essential in order that the patient may benefit from any revascularization procedure. However, because con1019195;ACCEFTED 4115196. Address correspondence to J. P. Laissy, Department of Radiology, Centre Hospitalier et Universitaire Bichat-Claude
Bernard, 46 rue Henri Huchard, 75877 Paris cedex 18, France.
RECEIVED
1033
Magnetic Resonance Imaging l Volume 14, Number 9, 1996
1034
MATERIALS
AND METHODS
Between November 1993 and June 1995,36 consecutive patient (24 males and 12 females, aged 32-67 years [mean, 43 years] ) with strong suspicion of renal hypertension on the basis of clinical grounds, usually because of a baseline diastolic blood pressure above 110 mm Hg or the presence of treatment-resistant hypertension, were studied with both renal angiography and MRA. Subjects were eligible patients who were scheduled for renal angiography, were willing to participate in the study, and could be scheduled for MRA on the week of their angiographic examination. Nine patients had renal insufficiency, with serum creatinine exceeding 150 pmollliter. Imaging Techniques
MRA was performed within l-4 days from diagnostic angiography. The final diagnoses were based on results of intraarterial digital angiography used as the gold standard in all patients. All angiograms consisted of intraarterial flush aortograms, and digital recording was used exclusively. Intraarterial DSA of the abdominal aorta was performed via a 5-French pigtail catheter positioned at the level of renal arteries; contrast material (Hexabrix 320, laboratoire Guerbet, Aulnay sous bois, France) was administered at 20 ml/s for 2.5 seconds. DSA examinations systematically included both frontal and oblique views to clearly analyze renal arteries and renal ostia. No selective angiograms were performed. MR studies were performed using a superconducting magnet (Magnetom, Siemens, Erlangen, Germany) operating at 1.0 T. In all cases, examinations were performed using the body coil. To visualize the origin of renal arteries, all subjects were first examined in the coronal plane, by means of SE T’weighted sequences (TR/TE 450/15 ms, g-mm section thickness). Volunteers In the first phase of the study, 10 volunteers underwent two-dimensional (2D) MRA using a gradient-recalled echo (GRE) time-of-flight (TOF) technique with low angle shot, gradient spoiling and first-order motion flow compensation in slice select and readout direction (FLASH), and a traveling venous saturation pulse. The following imaging parameters were systematically varied: field-of-view (FOV) , 3035 cm; section thickness, 3-5 mm; flip angle, 35-60”; repetition time, 32-50 ms; number of phase encoding steps, 128-192; and number of signal averaged, 2-4. The TE was the shortest possible authorized by our machine (9 ms). Flow-compensated gradients and a traveling venous saturation pulse (with a saturation slab eight times as thick as the imaged section, the center position of the sat band being shifted from the
imaging section position by eight times the section thickness) were used in all studies. An additional stationary oblique-coronal presaturation pulse was placed posteriorly on the left kidney to exclude the signal intensity from the left renal vein. Patients In the second phase of the study, 36 patients underwent MRA examinations using the same 2D GRE sequence than that used in volunteers. Acquisitions consisted in multisection sequential 2D FLASH sequences in the axial plane from the celiac abdominal aorta through the infrarenal aorta. One series of multisection axial sequences used a fixed TRITE of 3519 ms and 55” flip angle. Twenty-four successive sections were acquired, using a FOV of 350 mm, a 140 X 256 acquisition matrix, a 4-mm section thickness with a 2mm overlap, and two excitations, for an average acquisition time of 4 mitt, 08 s. In case of insufficient flow signal within renal arteries, the same 2D TOF sequences were repeated successively in two separate axial oblique planes (i.e., axial toward sagittal) in order that each sequence was the most perpendicular to each renal artery (Fig. 1) . Individual MRAs were then postprocessed using a maximum intensity projection (MIP) algorithm. Five to six projection images with angles incremented by steps of 15” from right lateral to left lateral were displayed on film for interpretation using multiplanar reformation. Data Analysis
In the first phase of the study, two radiologists (JPL, MBY) evaluated the 2D techniques in volunteers by measuring the length of arteries exhibited and the contrast-to-noise (C/N) ratio of aorta and renal arteries by means of regions-of-interest of at least 25 pixels. Because the imaged volume was entirely within the patient, background noise could not be measured outside the patient. Thus, background noise was measured on a homogeneous region laterally within the image volume, as previously recommended.2’ In the second phase of the study, three radiologists (JPL, OL, MBY) reviewed separately the MRAs without knowledge of the angiographic results. Digital arteriograms were also interpreted without knowledge of the MRA results in a blinded fashion by the same radiologists at a different time. Readers nos. 1 and 2 were experienced radiologists, whereas reader no. 3 was a resident in training. Both the axial images and MIP reconstructions were reviewed before a final opinion was rendered. In cases of discrepancies between the readers, the final MRA diagnosis was obtained by consensus. Stenoses were assessed subjectively. Be-
RenalMRA at 1.0 T 0 J.-P.
LAISSY ETAL.
1035
Fig. 1. Patient with aortic atheroma and proximal right artery stenosis. (A) Poor-quality MRA from 2D axial acquisitions precludes accurate assessmentof renal arteries. (B and C) MFLA from 2D right (B) and left (C) oblique axial acquisitions improve visibility of stenosed (arrow) right (B) and irregular left (C) renal arteries. Stair-step effect and signal drop-out from inplane artifact (C) are seen. (D) The corresponding angiographic image confirms a significant right artery stenosis and widely patent left renal artery.
cause the objective was merely to screen significant renal artery stenoses, stenosis was graded with a two-point scale on MRA and angiography: normal and nonsignificant proximal renal artery stenoses of less than 50% and significant stenoses of more than
50% in diameter, as well as multiple stenoses and thrombosis. Accessory arteries were assessed as normal or involved. On MRA, significant stenosis corresponded to a localized luminal narrowing with or without
flow signal inhomogeneities
flow void; thrombosis
or localized
was diagnosed as a complete
absence of flow signal regardless of the diameter of the artery. Presence or absence of accessory arteries was assessed.
Statistical Analysis In volunteers, signal-to-noise ratios expressed as a function of angle and NEX were presented as mean values + SD. Differences
in the aortic and renal SI
and C/N ratios for each of the sequences were evaluated with the Student’s t-test. For the comparison between MRA and angiographic data in patients, sensitivity, specificity, and predictive accuracy of MRA for the detection of significant renal artery stenosis were based on standard definitions. The null hypothesis was rejected at the 95% confidence level, considering p < .05 to be significant.
Interobserver variability between the readers was determined by using the kappa statistic.23
Magnetic Resonance Imaging l Volume 14, Number 9, 1996
1036
RESULTS Volunteers The first phase of our study was directed at development of a clinically effective MRA protocol for evaluation of renal arteries. C/N ratios indicated the best flip angle as 55” (p = .02). The number of excitations (4 vs 2) did not modify significantly C/N ratio (p = .08). Section thickness (3 vs 5 mm) did not influence significantly C/N ratio. However, even when overlap of 2 mm was used, a “stair-step” effect was still seen in two cases by using 5-mm-section thickness. Hence, using the 2D technique, the best imaging parameters were TR/TE 35/9 ms, flip angle 55”, NEX=2, FOV=350 mm, voxel size = 4 X 1.37 x 1.87 mm, 24 overlapping sections for an overall acquisition time of 4.08 minutes. With these parameters, maximal lengths measured on 2D MRAs were 1.5-7 cm (mean + SD; 3.8 t 2 cm) for the right renal artery and 2-6 cm (3.7 t 1.8 cm) for the left renal artery. Intrarenal branches were seen in 70% (7/10) in the right renal artery and in 80% (8/ 10) in the left renal artery (Fig. 2). Patients All MRAs were available for interpretation. A total of 77 renal arteries was detected by using digital angiography. MRAs showed 100% (70/70) of the main renal arteries and 86% (6/7) of the accessory renal arteries seen at angiography. Stenoses of the proximal main renal artery greater than 50% were visualized in 14 vessels in 12 patients at angiography (Table 1)
and two significant accessory artery stenoses in two other patients, yielding a prevalence of RAS of 39%. Renal artery involvement was related to fibromuscular dysplasia in one patient. Of the two isolated accessory renal artery stenoses identified on angiography, only one was seen on MRAs (Table 2). No stenosis was present in the accessory renal artery missed by MRA. The sensitivity of detection of main renal artery and accessory artery stenoses was 100% and 50%, respectively (14/14, CI 1.0, 0.92; and l/2, CI 0.8, 0.12) and the specificity was 98% and lOO%, respectively (55/56, CI 0.99, 0.83; and 4/4, CI, 1.0, 0.46). MRA had a sensitivity of 94% ( 15/ 16) and a specificity of 98% (60/61) in the overall detection of significant stenoses, for a positive and negative predictive values of 98% and 94%, respectively. MRA was 93% sensitive (13/14, CI 0.94, 0.85) and 95% specific (21/22, CI 0.97, 0.89) for identifying patients with at least one significant renal artery stenosis (Figs. 3-5). Discrepancies between readers were observed in two cases (one stenosis considered normal by one reader and with a significant stenosis by the two others, and one stenosis judged nonsignificant by one reader and significant by the two others). The interobserver variability between the readers was 0.9 1. Reader experience involved significant greater accuracy only between reader no. 1 and reader no. 3 (p = 0.04). DISCUSSION Ultrasound and radionuclide studies are established noninvasive screening tests in renovascular
Fig. 2. MIP reconstruction obtained from 2D axial acquisitions in a volunteer. Renal arteries are well seen until renal pelvis. An area of signal loss is seen in the right middistal renal artery where the renal pelvis crosses (arrow).
Renal MRA at 1.O T 0 J.-P. LAISSY
Table 1. Sensitivity and specificity of MRA in the diagnosis of renal artery stenosis, according to the readers Reader Reader Reader Concurrent 1 2 3 interpretation Main renal artery stenosis Sensitivity (%) Specificity (%) Main plus accessory renal artery stenosis Sensitivity (%) Specificity (%)
100 95
91 98
91 98
100 98
‘94 ‘97
85 98
79 98
94 98
hypertension. Duplex Doppler is affected of a broad range of accuracies, with accuracy estimated between 79 and 91% according to the different studies.3 However, this technique is susceptible to a variety of problems such as operator skill, interference by abdominal bowel gas, and poor reproducibility between institutions, and nearly one-third of examinations are noncontributive.2,3 Renal scintigraphy needs to be performed in conjunction with angiotensin-converting enzyme inhibitors (e.g., captopril) administration. Captopril renal scintigraphy provide accurate functional information upon renal function, mainly relative to its local renal effect.‘,* Captopril decreases postglomerular efferent tone and thus causes an important reduction in the glomerular filtration rate of kidneys with renal artery stenosis.’ Captopril renography has respective sensitivity and specificity rates of approximately 90%.’ Moreover, the value of captopril renography is limited in patients with renal insufficiency, due to decreased renal uptake and increased background activity.‘4 More accurate noninvasive tests are a major issue in the screening of RVH. Early promising results have been obtained with spiral CT.4.5 The sensitivity of spiral CT for detection of RAS?SO% has recently been reported as high as 98% and the specificity as high as 94%.25 However, this technique involves large amounts of contrast medium, which can have deleterious effects in patients with renal insufficiency. Several investigators focused on the diagnostic capabilities of vascular MRI in the study of aorta and renal arteries.6-24 These studies were mainly obtained on MR devices operating at 1.5 T. In these studies, abdominal acquisitions were performed by means of either three-dimensional (3D) phase-contrast angi2D contiguous overlapping slice TOF -graphy, 6.9.‘2~‘s~‘7 acquisitions, 7,10’62D breath-held TOF acquisitions, *- “Z
1037
ET AL.
or, more recently, 3D acquisitions with use of multiple overlapping slab acquisitions, ‘8,22with fat saturation 2o or with incremented flip angles.*’ Our results obtained at 1 .O T using 2D overlapping techniques 5,9~1o compare favorably with the 87-100% sensitivity and 92-97% specificity of MRA for detection of significant stenoses,8-10~‘2,‘6~‘8~24 even if our results are limited because of the small number of diseased vessels. Moreover, compared with other MRA techniques, the time provided by 2D acquisition is short.9*16 The potential for improving on the results here by use of more state of the art techniques includes the development of new sequences,“0B22such as fat suppressed 2D TOF MRA done immediately after intravenous administration of an MR contrast agent*’ and hardware implementation of echo planar imaging.7 Detection of accessory renal arteries with MRA remains problematic. The reported range of accessory arteries depicted by MRA is comprised between 50 and 100~~.8.9,'2,'8,'4,26 Our sensitivity in detecting significant accessory artery stenosis was 50%, when compared with the 100% sensitivity in detecting significant main renal artery stenosis. In our study, the acquisitions did not include the aortic bifurcation and proximal common iliac arteries, despite the fact that accessory renal arteries can arise quite low, even from iliac arteries. This is a difficult diagnostic problem in cases of RVH and in the preoperative evaluation of kidney donors.27 This study has some other limitations. A small number of stenoses is included, as well as a small number of accessory arteries. A 39% prevalence favors high sensitivity and high specificity, compared with the expected varying estimates of prevalence, from 1% to 10% of patients with hypertension screened.28 Another drawback is that stenotic lesions are often overestimated; stenoses are more difficult to be evaluated in the MIP technique that in a single slice technique. Phase dispersion due to turbulent flow has been described as a potential pitfall in the estimation of stenosis, in particular near renal ostia. 7,9B1’,21 Although our work only tried to identify nonsignificant from significant artery stenoses, our results demonstrated that phase dispersion in case of turbulences was a good indicator of the severity of stenoses, despite the fact that renal artery sinuosities Table 2. Analysis of accessory arteries by MRA (concurrent interpretation)
MRA Normal Significant
Normal
Significant stenosis
1
4 stenosis
1
Not seen
1
Magnetic ResonanceImaging 0 Volume 14, Number9, 1996
1038
(4
Fig. 3. Patient with suspected RVH. (A) Localized flow void 2 cm distal to the origin of left renal artery (arrow) on frontal MIP image (false-positive finding). (B and C) Axial MIP image (B) and frontal multiplanar reformatted image (C) show normal renal arteries. A flow void is present at the origin of the left renal artery on MIP image, corresponding to a reconstruction artifact (axial source images, not shown, displayed a patent left renal artery). (D) Angiography identifies no significant stenosis. Renin venous sampling was normal. by themselves can involve flow dephasing. This sign unfortunately lacks sensitivity. RAS may be caused by either atherosclerosis or fibromuscular dysplasia (FMD) . In these cases, successive stenoses may lead to a difficult evaluation of the degree of renal perfusion involvement. FMD can also involve the distal part of renal arteries, including intrarenal stenosis?’ Signal within renal arteries is reduced in patients with renal insufficiency who have reduced renal artery blood flow.“j Indeed, RAS has also been recognized as an important cause of impaired renal function.
The purpose of this study was only to evaluate MRA to assess significant from nonsignificant stenoses. Compared with carotid artery disease where the exact degree and morphology of the stenosis is the critical information to be obtained, the main information useful for treatment of RAS is the severity of RAS. The diagnosis of RAS currently relies on renal angiography. On the basis of this technique ascertained as a diagnostic standard, the criterion of a significant stenosis is yet polemic. Some investigators considered a 50% stenosis to be significant, yet
Renal MRA at 1.0 T 0 J.-P. LAISSY ETAL.
(4
Fig. 4. True positive findings of a significant left renal artery stenosis. (A) MRA suggests left renal artery occlusion (right renal artery better seen on other projections not shown). Source images (not shown) displayed a small patent left renal artery. (B) Angiography shows a small patent but severely diseased left renal artery.
perfusion pressure in a large artery is generally not reduced until the stenosis exceeds 70%.* Renal flow quantification by phase-contrast angiography has also been used l7 to overcome the limitation of TOFMRA by its inherent inability to predict the hemodynamic importance of a particular lesion. In conclusion, 2D TOF MRA at 1.0 T appears to be a simple accurate means in the screening of patients with suspected RVH. It could also be useful
in patients with renal disease who are at risk for contrast media toxicity. Among previous studies reported, the present one demonstrates that this field strength is not an obstacle in diagnosing renal artery disease. Acknowledgements-We thank Geraldine Patient and Marie-France Cahtrec for their secretarial assistance and Christine Besnard for her pictorial assistance.
Fig. 5. False-negative MRA in a case of accessory artery stenosis. (A) MRA displays a normal left accessory artery (arrow). (B) Angiography shows significant (arrowhead) left renal accessory artery stenosis (70%), with decreased left inferior polar nephrogram (large arrowheads).
REFERENCES 1. Hillman, B.J. Imaging advances in the diagnosis of renovascular hypertension. AJR 153:5-14; 1989. 2. Davidson, R.A.; Wilcox, C.S. Newer tests for the diagnosis of renovascular disease. JAMA 2683353-3358; 1992. 3. Middleton, W.D. Doppler US evaluation of renal artery stenosis: past, present, future. Radiology 184:307-308; 1992. 4. Galanski, M.; Prokop, M.; Chavan, A.; Schaefer, CM.; Jandeleit, K.; Nischelsky, J.E. Renal arterial stenoses: spiral CT angiography. Radiology 189: 185- 192; 1993. 5. Rubin, G.D.; Dake, M.D.; Napel, S.; Brooke Jeffrey
R. Jr.; McDonnell, C.H.; Sommer, F.G.; Wexler, L.; Williams, D.M. Spiral CT of renal artery stenosis: comparison of three-dimensional rendering techniques. Radiology 190: 181- 189; 1994. 6. Dumoulin, CL.; Yucel, E.K.; Vock, P.; Souza, S.P.; Terrier, F.; Steinberg, F.L.; Wegmuller, H. Two- and three-dimensional phase contrast MR angiography of the abdomen. J. Comput. Assist. Tomogr. 14:779-784; 1990. 7. Edelman, R.R. MR angiography: present and future. AJR 161:1-11; 1993. 8. m D.; Edelman, R.R.; Kent, K.C.; Porter, D.H.; Skillman, J.J. Abdominal aorta and renal artery stenosis evaluation with MR angiography. Radiology 174727-731; 1990.
Renal MRA at 1.0 T 0 J.-P. 9. Debatin, J.F.; Spritzer, C.E.; Grist, T.M.; Beam, C.; Svetkey, L.P.; Newman, G.E.; Sostman, H.D. Imaging of the renal arteries: value of MR angiography. AJR 157:981-990; 1991. 10. Kent, K.C.; Edelman, R.R.; Kim, D.; Steinman, T.I.; Porter, D.H.; Skillman, J.J. Magnetic resonance imaging : a reliable test for the evaluation of proximal atherosclerotic renal arterial stenosis. J. Vast. Surg. 13:311-318; 1991. 11. Lewin, J.S.; Laub, G;.; Hausmann, R. Three-dimensional time-of-flight MR angiography: applications in the abdomen and thorax. Radiology 179:261-264; 1991. 12. Vock, P.; Terrier, F.; Wegmtiller H.; Mahler, F.; Gertsch, P.: Souza, S.P.; Domoulin, C.L. Magnetic resonance angiography ‘of abdominal vessels: early experience using the three-dimensional phase-contrast technique. Br. J. Radiol. 64:10-16; 1991. 13. Arlart, I.P.; Guhl, L.; Edelman, R.R. Magnetic resonance angiography of the abdominal aorta. Cardiovasc. Intervent. Radiol. 15:43-50; 1992. 14. Yucel, E.K.; Magnsetic resonance angiography of the lower extremity and renal arteries. Semin. Ultrasound CT MRI 13:291-302; 1992. 15. Wolf, R.L.; King, B.F.; Torres, V.E.; Wilson, D.M.; Ehman, R.L. Measurements of normal renal artery blood flow: tine phase-contrast imaging vs clearance of paminohippurate. AJR 161:995- 1002; 1993. 16. Grist, T.M.; Kennel, T.W.; Sproat, I.A.; McDermott, J.C.; Wojtowycz, h4.M.; Flath, E.M.; Crummy, A.B. Prospective evaluation of renal MR angiography: comparison with conventional angiography in 35 patients. Radiology 189:190--196; 1993. 17. Debatin, J.F.; Ting, R.H.; Wegmtiller, H.; Sommer, F.G.; Fredrickson, J.O.; Brosnan, T.J.; Bowman, B.S.; Myers, B.D.; Herf kens, R.J.; Pelt. N.J. Renal artery blood flow: quantitation with phase-contrast imaging with and without breath holding. Radiology 190:371378; 1994. 18. Kaufman, J.A.; Geller, S.C.; Petersen, M.J.; Cambria, R.P.; Prince, M.R.; Waltman, A.C. MR imaging (including MR angiography) of abdominal aortic aneurysms: comparison with conventional angiography. AJR 163: 203-210; 1994.
LAISSV
ET AL.
1041
19. Hertz, S.M.; Holland, G.A.; Baum, R.A.; Haskal, Z.J.; Carpenter, J.P. Evaluation of renal artery stenosis by magnetic resonance angiography. Am. J. Surg. 168: 140-- 143; 1994. 20. Li, D.; Haacke, E.M.; Mugler III, J.P.; Berr, S.; Brookeman, J.P.; Hutton, M.C. Three-dimensional time-offlight MR angiography using selective inversion recovery RAGE with fat saturation and ECG-triggering : application to renal arteries. Magn. Res. Med. 31:414422; 1994. 21 Loubeyre, P.; Revel, D.; Garcia, P.; Delignette, A.; Canet, E.; Chirossel, P.; Genin, G.; Amiel, M. Screening patients for renal artery stenosis: value of three-dimensional time-of-flight MR angiography. AJR 162:847852; 1994. 22. Prince, M.R. Gadolinium-enhanced MR aortography. Radiology 191:155-164; 1994. 23. Cohen, J. A coefficient of agreement for nominal scales. Educ. Psychol. Meas. 20:37-46; 1960. 24. Yucel, E.K.; Kaufman, J.A.; Prince, M.R.; Bazari, H.; Fang, L.S.T.; Waltman, A.C. Time-of-flight renal MR angiography in patients with renal insufficiency. Magn. Reson. Imaging 11:925-930; 1993. 25. Galanski, M.; Prokop, M.: Chavan, A.; Schaefer, CM.; Jandeleit, K.; Nischelsky. J.E. Minimally invasive diagnosis of renal artery stenosis by spiral computed tomography angiography. Kidney Int. 48:1332- 1337; 1995. 26. Servois, V.; Laissy, J.P.; Feger, C.; Sibert, A.; Delahousse, M.; Baleynaud, S.; M&y, J. Ph.; Menu, Y. 2D MR TOF angiography without MIP reconstruction of renal arteries: a prospective comparison with angiography in 21 patients. J. Cardiovasc. Inter-vent. Radiol. 17:138-142; 1994. 27. Debatin, J.F.; Sostman, H.D.; Knelson, M.; Argabright, M.; Spritzer, C.E. Renal magnetic resonance angiography in the preoperative detection of supernumerary renal arteries in potential kidney donors. Invest. Radiol. 28:882-889; 1993. 28. Dean, R.H. Renovascular hypertension: an overview. In: R.B. Rutherford (Ed). Vascular Surgery, 3rd ed. Philadelphia: W.B. Saunders, pp. 1211- 1218; 1989.