- Email: [email protected]
Induced renal artery stenosis in rabbits: Magnetic resonance imaging, angiography, and radionuclide determination of blood volume and blood flow
Magnetic Resonance Printed in rhe USA.
Imaging. Vol. 6, pp. All rights reserved.
113-124,
1988 Copyright
0730-725X/88 $3.00 + .oO El 1988 Pergamon Press plc
l Original Contribution
INDUCED RENAL ARTERY STENOSIS IN RABBITS: MAGNETIC RESONANCE IMAGING, ANGIOGRAPHY, AND RADIONUCLIDE DETERMINATION OF BLOOD VOLUME AND BLOOD FLOW DONALD G. MITCHELL,* MICHAEL TOBIN, * ROBERT LEVEEN,~ JOHN TOMACZEWSKI, ** ABASS ALAVI,* MUNI STAUM,* AND HAROLD KUNDEL* Departments of Radiology* and Pathology, ** Hospital of the University of Pennsylvania, Philadelphia, PA 19104, TMedical Research Service, Veterans Administration Hospital, Philadelphia, PA 19104. To investigate the ability of MRI to detect alterations due to renal ischemin, a rabbit renal artery stenosis (RAS) model was developed. Seven rabbits had RAS induced by surgically encircling the artery with a polyethylene band which had a lumen of 1 mm, 1 to 2 weeks prior to imaging. The stenosis was confirmed by angiography, and the rabbits were then imaged in a 1.4 T research MRI unit. T, was calculated using four inversion recovery sequences with different inversion times. Renal blood flow, using “3Sn-microspheres, and regional water content by drying were then measured. The average Tl of the inner medulla was shorter for the ischemic (1574 msee) than for the contralateral kidney (1849 msec), while no change was noted in the cortex. lschemic kidneys had less distinct outer medullary zones on IR images with TI = 600 msec than did contralateral or control kidneys. Blood flow to both the cortex and medulla were markedly reduced in ischemic kidneys compared with contralateral kidneys (119.5 vs. 391 ml/min/lOO gm for cortex and 19.8 vs. 50.8 ml/min/lOti gm for medulla). Renal water and blood content were less affected. Our rabbit model of renal artery stenosis with MRI, radionuclide, and angiographic correlation has the potential to increase our understanding of MR imaging of the rabbit kidney. Keywords: Kidney; Vessels; Bloodflow.
relaxation and to better understand its underlying pathophysioiogy, we correlated renal MRI appearance and regional tissue Tr relaxation times with measured reductions of renal blood flow, blood volume, and water content in a rabbit RAS model.
INTRODUCTION A non-invasive method for accurate detection and evaluation of renovascular disease has not yet been found. Magnetic resonance imaging (MRI) has potential in this area because of the effect on tissue relaxation of water content, water structure, blood flow, and other parameters which may be influenced by renovascular disease. Preliminary investigations on MRI of the rabbit kidney reveal that T, and T2relaxation times reflect the amount of water present*,15 and that complete vascular or ureteral obstruction alters these relaxation times.” The effect of renal artery stenosis (RAS) remains undetermined. In order to determine the effect of RAS on tissue
MATERIALS
AND METHODS
Surgery
Nine male New Zealand White rabbits weighing approximately 3 kilograms each were studied. Using sterile technique and ketamine (40 mg/kg) and xylazine (5 mg/kg) anesthesia, we induced a stenosis of the left (n = 6) or right (n = I) renal artery by encircling it with a 1 cm long polyethylene band. The band
RECENED 7/2/87; ACCEPTED 8/27/87. Acknowledgments-We are indebted to EIanor Kasab,
Address correspondence and reprint requests to Donald G. Mitchell, M.D., Department of Radiology, Division of
Bethanne Moore, and Deborah DeSimone for technical expertise, and to Rachel Knowles and Emily N. Pompetti for manuscript preparation.
MRI, Thomas Jefferson University Hospital, Main Building, 10th Floor, Tenth and Sansom Streets, Philadelphia, PA 19107. 113
114
Magnetic
Resonance
Imaging
had a lumen of 1 mm, and had a longitudinal slit to allow its placement around the artery. A suture was placed around the band to assure a permanent stenosis. The band was left in place 1 to 2 weeks prior to imaging. The remaining two animals had no surgery. Angiography Immediately before MR imaging, an abdominal aortogram was performed from a carotid catheterization to confirm the stenosis. Approximately 5 cc of 60% meglumine diatrizoate were used in each case. In one non-surgerized rabbit, unintentional intimal dissection resulted in an acute right renal artery stenosis. Thus, only one rabbit was a true control. Catheters were also placed in a common iliac artery via a femoral approach to allow sampling of distal arterial blood, and in a marginal ear vein to administer medications. Magnetic Resonance Imaging The rabbits were imaged in a 1.4 Tesla research unit that has a 6-inch imaging bore. The appropriate axial plane was located using spin-echo images with repetition time (TR) of 200 msec and echo time (TE) of 10 msec. Tr was calculated in eight rabbits using four inversion recovery (IR) images with TR of 3000 msec, TE of 10 msec, and inversion times (TI) of 100, 600, 1200, and 2000 msec (Fig. 1). The method of calculation was similar to that used in a previous report from this institution, and involved sampling seven to nine contiguous regions from the cortex to medulla consisting of 9 pixels each.8 In six rabbits predominately Tz-weighted spin-echo images of each kidney were obtained using TR of 2500 msec and TE of 80 msec. Corticomedullary intensity ratios were obtained using 9 pixel boxes placed over cortex and medulla in each image. Glass tubes containing 0.5, 1.O, and 2.0 millimolar copper sulfate were included with each scan for calibration. Using IR images with TR/TE/TI = 3000/10/600 msec, renal area was estimated (0.785 times anteroposterior and transverse diameters). These were compared with in vivo images of five normal left kidneys in similar rabbits, as well as one high resolution image of a normal rabbit kidney ex vivo (IR: TR/TE/TI = 3000/10/600 msec), which were obtained in a previous experiment in this laboratory.8 Radionuclide Blood Volume Measurement Renal blood volume was determined in six rabbits using 99mT~labeled red blood cells (99mTc-RBC’s).5 During MRI, approximately 1.5 hours prior to nuclear imaging, 2 cc of the rabbit’s blood were withdrawn and labeled with approximately 10 mCi of double-eluted 99mT~04- using a commercially avail-
0 Volume
6, Number
2, 1988
able kit (Cadema Medical Products, Inc.; Red Blood Cell Kit for preparing 99mT~04- labeled RBC’s). Following a single washing with normal saline, radiotracer tagging of RBC’s was in excess of 98%. Following MRI, the aortic catheter was repositioned into the left ventricle (LV) using fluoroscopic guidance. With the dorsal surface of the rabbit positioned on top of a gamma camera fitted with a low energy all-purpose collimator, 3-5 mCi of 99mT~RBC were injected as a bolus into the LV. Analogue and digital images were collected at a rate of 2 set per image for 60 sec. Sixty set static images were then obtained for 10 min. Selected animals were also imaged in the lateral and ventral positions and based upon these images, it was decided not to depth correct renal activity for soft tissue attenuation. Approximately 4 cc of blood were then withdrawn from the femoral artery catheter. This blood was imaged for 1 min for calibration and was then used to obtain a central hematocrit. The blood was then centrifuged to check the stability of the 99mTc-RBC bond. Regions of interest were drawn around each kidney using the final dorsal static images. Background corrected renal counts were then compared with the counts from the 4 cc blood samples to obtain absolute renal blood volumes. Radionuclide Blood Flow Measurement In five rabbits, quantitative cortical and medullary blood flow was determined with microspheres. FiftyuCi ‘13Sn-labeled 15 A microspheres were ultrasonicated until the time of administration, and were hand injected over 60 set into the LV catheter. Beginning at the time of injection, blood was withdrawn via the iliac artery catheter at 1 ml per min with a Harvard pump. The animal was then sacrificed. Regional blood flow was determined 2 days following autopsy (see below). Total anesthesia time (angiography, MRI, radionuclide) averaged 6-7 hours. Post Mortem Analysis Following MRI and radionuclide studies, the kidneys were exposed via midline incision. The vascular pedicle of each kidney was located and a suture was placed around each. The sutures were then tied, preventing blood or urine loss, and the animal was sacrificed using intravenous pentobarbital. The kidneys were removed and then weighed. For five of the first six animals, a core of renal tissue was obtained via a sharply beveled thin-walled 8 mm diameter tube inserted into the renal hilum. This core was sectioned into cortex, outer and inner stripe of the outer medulla, and inner medulla.8 These sections of renal tissue were then placed in tared bottles,
Induced renal artery stenosis 0
DONALD
G.
MITCHELL ET AL.
115
Fig. 1. MRI appearance of the left kidney in the control animal, axial images. (a)-(d) IR images with TR 3000 msec, TE 10 msec. (a) Tl = 2000 msec. The medulla (arrow) has slightly lower intensity than the cortex (arrowheads), due to its longer r,. (b) TI = 1200 msec. The contrast between cortex and medulla is increased. The outer medulla (arrowheads) has intermediate intensity. (c) TI = 600 msec. The inner medulla now has higher signal intensity than the cortex. A low signal junctional band (arrows) is now visible between the inner medulla and cortex, presumably representing the outer medulla. (d) TI = 100 msec. The cortex and medulla are both higher intensity than fat. (e) Spin-echo image (TR 2500 msec, TE 80 msec). The medulla has higher signal intensity than the cortex due to its higher T2. e
weighed, and then dried to constant weight in a vacuum oven at 50” to determine water content. The kidneys in the last three rabbits studied were frozen in liquid nitrogen and sectioned in an axial plane. Slices were retained for autoradiography and histologic analysis. The remaining tissue was divided into cortex, outer medulla, and inner medulla. The specimens were weighed and placed into tubes. After allowing 2 days for 99mTC to decay, renal cortical and inner medullary samples were counted in a well counter using a 320-440 kev window. (Outer
medulla was not individually counted due to uncertainty of whether parts of cortex or inner medulla were included.) Counts were obtained for cortex, medulla, and total kidney as well as for the blood withdrawn during the microsphere injection. These values, along with the rate of withdrawal, allowed calculation of the cortical, medullary, and total blood flow per gram for each kidney. Autoradiographs were obtained by placing a section of renal tissue on a single emulsion nuclear medicine blue-based radiographic film for 3 days to 4 weeks.
116
Magnetic Resonance Imaging 0 Volume 6, Number 2, 1988
In three rabbits, sections were saved for histologic analysis. These were fixed in formalin and stained with hemotoxylin and eosin. Note was made of any pathologic alterations, as well as the location of microspheres. The number of microspheres per 100 glomeruli in both cortex and medulla was recorded. Statistical significance was determined using a paired t-test. RESULTS Angiography
The induced renal artery stenosis ranged from 80% (Fig. 2a) to total occlusion with collateral supply in one animal. In three rabbits, decreased nephrogram density and a delayed pyelogram were noted as well. In one of the non-surgerized animals a stenosis of the right renal artery due to intimal dissection was seen, accompanied by delayed filling of intrarenal vessels. Thus, this animal had an acute right renal artery stenosis and was not a control animal. MR Images
Renal area varied between 3.3 to 5.3 cm2 for the non-ischemic kidney, with a mean of 4.3 + 0.7 cm2. This was smaller than the normal left kidneys from the earlier experiment,8 which varied between 4.2 and 6.9 cm2 (mean 5.6 f 1.0 cm2). The ischemic kidneys were smaller than the contralateral kidneys (p < O.Ol), with a size ranging from 2.7 to 4.2 cm2 with a mean of 3.5 f 0.9 cm2 (Table 1). Cortex and medulla could best be distinguished based on relative intensities on IR images with TI of 1200 msec. With this sequence, the medulla was less intense than the cortex in both ischemic and non-
Table 1. Cross-sectional
area and width of junctional
Cross sectional No.
Ischemic
kidney
R L L L L L L* R** Control
Contralateral 4.2 3.3 5.3 3.6 4.2 5.1 3.4 ::;
ischemic kidneys (Figs. 2d,g). Mean corticomedullary intensity ratio was 2.0 f 0.8 for ischemic and 1.8 f 0.23 for non-ischemic kidneys (not significantly different). Intrarenal zones were best depicted in IR images with TI of 600 msec. In these images the cortex and inner medulla both had intermediate intensity. A low signal “junctional band” separated them, corresponding primarily to the outer stripe of the outer medulla (Fig. lc). 8*9The inner stripe of the outer medulla was less well defined. In four of the five normal left kidneys in the previous experiments and in both kidneys in the control animal in this experiment, the low signal junctional band measured 3 mm in diameter. In the image of a normal resected kidney, this band corresponded to the outer stripe of the outer medulla, which also measured 3 mm. In non-stenotic kidneys in experimental animals this band was well seen, ranging in thickness from 4 to 7 mm with a mean thickness of 5.7 f 0.9 mm (Fig. 2e). In ischemic kidneys, however, this band was less well seen (Fig. 2h), and in two of the six kidneys with surgically induced stenosis it could not be identified. In the four ischemic kidneys in which the junctional band could be seen, its thickness was decreased (p < 0.05), ranging from 1 to 2 mm, with a mean thickness of 1.8 f 0.4 mm. In the rabbit with RAS due to intimal dissection, the junctional band in the affected kidney was obscured by low signal of the medulla. With long TR/TE spin-echo images (predominately T2 weighted), the medulla had higher intensity than cortex (Figs. le, 2f, 2i). Mean corticomedullary intensity ratio (Table 2) was 0.8 1 f 0.19 for ischemic and 0.44 f 0.17 for nonischemic kidneys (not significantly different).
(R)
I/C = Ratio of ischemic to contralateral kidneys. *Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis.
area
(CM2)
band Stripe width (MM2)
Ischemic
I/C
3.9 3.2 4.2 3.0 2.7 4.2 1.9 4.6 4.2 (L)
0.95 0.96 0.79 0.82 0.62 0.82 0.56 0.80 0.89
Contralateral 6 No IR 3000/10/600 6 7 4 5 6 3 3
Ischemic Obscured images 2 2 2 Obscured 1 Obscured 3
Induced renal artery stenosis 0
DONALD G. MITCHELL ET AL.
. .
P
117
118
Magnetic
Resonance
Imaging
0 Volume
I a
6, Number
2, 1988
i
Fig. 2. (g)-(i) Axial MR images of the ischemic left kidney. (g) IR image (TR 3000 msec, TI 1200 msec, TE 10 msec). Contrast between cortex and medulla is decreased in comparison to Fig. l(b). (h) IR image (TR 3ooOmsec, Tl 600 msec, TE 10 msec. A thin low signal line (arrows) separates the cortex from medulla. (i) Spin-echo image (TR 2500 msec, TE 80 msec). Contrast between cortex and medulla is decreased in comparison with Figs. l(e) and 2(d). Contrast between cortex and medulla in (d) and (f) is greater than seen in (g) and (i). The junctional band in (e) appears thicker than that seen in Fig. 1 (c), but the images are otherwise similar to the control kidney in Figs. l(b)-(e).
120
Magnetic Resonance Imaging 0 Volume 6, Number 2, 1988 Table 2. Summary
of MRI intensity
measurements
Calculated No.
Ischemic
kidney
R L L L L L L* R** Control
expressed
as cortico-medullary
T, C/M ratio
Contralateral
C/M intensity
Ischemic
0.50
ratio (SE 2500/80)
Contralateral
Ischemic
0.68
0.48 0.48
0.46 0.84 0.48 0.94 0.45 0.40 0.30
No SE 2500/80 images 0.64 0.94 No SE 2500/80 0.63 0.18 1.00 0.65 0.47 0.58 0.43 No SE 2500180 0.57
Not calculated 0.45 0.43 0.51 0.66 0.50 0.48 0.46
ratios
0.50 0.79
I/C = Ratio of ischemic to contralateral kidneys. *Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis.
T, Calculation (Table 2) For all six rabbits with surgically induced RAS which had T, calculated, a gradient of increasing T, from cortex to medulla was noted in both ischemic and nonischemic kidneys. The mean T, gradient from cortex to medulla of non-ischemic kidneys was 914 msec (935 to 1849 msec), compared with 588 msec (986 to 1574 msec) for ischemic kidneys (Fig. 3). This is less than the Tl gradient for the control rabbit (mean of two kidneys), which was 1230 (720 to 1950 msec). The relaxation times of all the rabbits were longer than those previously reported from this laboratory for normal kidneys of 600 to 1200 msec.’ T, of cortex was similar for both ischemic and non-ischemic kidneys, but both the outer (junctional band) and inner medulla of the ischemic kidneys had lower T, than the corresponding zones in non-ischemic kidneys. Radionuclide Blood Volume Measurement In vivo blood volume was determined in five rabbits (Table 3). Dynamic images following a bolus of 99”TC-RBC’s revealed delayed flow to the ischemic kidney, but RBC volume following equilibration was similar in both kidneys (Figs. 2b, 2~). Blood volume determined from the final static image ranged from 1.7 to 4.6 ml with a mean of 2.8 ml in non-stenotic kidneys, compared to a range of 0.8 to 4.3 ml with a mean of 1.9 ml in ischemic kidneys (not significantly different). Radionuclide Blood Flow Measurement (Table 4) Blood flow to the cortex and medulla of the ischemic kidneys was markedly reduced. In non-stenotic kidneys, the average flow to cortex, medulla, and total
o = Tt x = % H20 by weight
I = average standard deviation 30%
20%
10%
0%
RENAL ZONE FROM CORTEX TO INNER MEDULLA
-10%
I
/
1
2
I
/
I
I
,
4
5
6
7
I
3
Fig. 3. Percent difference between stenotic and non-stenotic kidney of mean T, and water content from cortex to medulla. The difference in T,was largest in the outer medulla, less so in the inner medulla, and no difference was seen in the cortex. These differences did not parallel water content, which was similar in stenotic and non-stenotic kidneys.
Induced renal artery stenosis 0 DONALD G. MITCHELL ET AL.
121
Table 3. Weight, cortico-medullary water gradient and blood volume Weight (gm) No.
Ischemic kidney
1 2 3 4 5 6 7 8
R L L L
Contralateral
= contralateral;
L L L* L**
Contralateral
C/M percent HZ0 Gradient
Ischemic
Not measured Not measured Not measured 13.7 10.4 16.4 16.0 12.6
10.1 10.1 10.8 8.6 10.0
Contralateral 78-85 79-87 81-87 80-87 80-83 81-84 80-82 93-98
(0.92) (0.90) (0.93) (0.93) (0.97) (0.96)+ (0.97) + (0.95)+
Ischemic 81-89 77-88 81-89 85-88 77-91 83-84 81-82 94-88
Blood volume (ml) Contralateral
(0.86) (0.88) (0.91) (0.97) (0.84) (0.99)+ (OH)+ (1.07) +
Ischemic
Not measured Not measured 1.7 1.9 3.7 4.3 2.2 1.2 4.6 1.2 1.7 0.8 Not measured
C/M = cortico-medullary.
*Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis. +Kidneys frozen in liquid nitrogen prior to measurement of percent water.
was 391.0, 50.8, and 312.0 ml per min per 100 gm, respectively. Flow was markedly decreased to the ischemic kidney, being 119.5, 19.8, and 63.0 ml per min per 100 gm to cortex, medulla, and total kidney, respectively. Due to the small size of this sample, however, these results were not significant at the p < 0.05 level. Relative to the contralateral kidney, cortical blood flow was reduced in all four ischemic kidneys, but medullary flow was increased in two of four ischemic kidneys. Autoradiographs of renal sections were obtained in three rabbits, which confirmed the observation that most flow was to the cortex, which was 3 mm thick (Fig. 4a), and that flow was markedly reduced to the ischemic kidneys. Microspheres were counted histologically in two rabbits. Microspheres were located primarily in juxtaglomerular efferent arterioles (Fig. 4b). The average number counted per 100 glomeruli was 26 and 15 for cortex and medulla, respectively, in non-ischemic kidneys, compared with 5.5 and 8.5 for ischemic kidneys. kidney
Renal Weight In four rabbits with surgically induced RAS, post mortem renal weight was determined (Table 3). It varied from 10.4 to 16.4 gm in non-ischemic kidneys, with a mean of 14.1 gm. The weight of ischemic kidneys was reduced (p < O.Ol), ranging from 8.6 to 10.8 gm, with a mean of 9.9 gm. Water Content A gradient of increasing water content from cortex to medulla was noted in the five rabbits whose kidneys were not frozen prior to sectioning (Table 3). It
was similar in non-ischemic and ischemic kidneys (Fig. 3), ranging from 79.6 to 86.3% in the former and 79.4 to 89.0% in the latter. The gradient was abolished in the three rabbits whose kidneys were frozen. Histologic Analysis The kidneys in four rabbits were examined histologically. In the kidney with an occluded renal artery, large areas of infarction were seen. In one kidney with RAS, mild juxtaglomerular hyperplasia was noted. In all kidneys which were frozen, areas of cortical infarction were seen. DISCUSSION
In this experiment, we surgically induced unilateral renal artery stenosis (RAS) in seven rabbits, and used angiography for anatomic confirmation. Quantitative measurement of blood flow by labeled microspheres was accomplished in four rabbits with surgically induced subacute RAS and in one rabbit with acute RAS, secondary to unintentional catheter-induced intimal dissection. In these rabbits, decreased flow to the ischemic kidney was documented. Further evidence of physiologic significance of the stenosis included decreased weight of the ischemic kidney, as well as decreased size as measured from MR images. The basis for the MRI appearance of kidneys is complex and only partially understood. A prior report from this laboratory demonstrated a gradient of increasing T, relaxation in normal rabbit kidneys from cortex to medulla, correlating with a gradient of increasing tissue water content.* This gradient is pre-
122
Magnetic Resonance Imaging 0 Volume 6, Number 2, 1988 Table 4. Blood flow to cortex and inner medulla
Cortical flow (ml/min/ 100 gm) No. 4 5 6 7 8
Ischemic kidney L L L L* L**
Contralateral 341 811 263 149 76
Medullary flow (ml/min/lOO
Ischemic
I/C
240 10 176 32 7
0.70 0.01 0.67 0.21 0.09
Contralateral = contralateral; I/C = ratio of ischemic to contralateral *Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis.
sumably
T, values (gradient between 600 and 1200 msec) than we noted, even in the control rabbit. The longer T, values and smaller size of the kidneys in the current experiment may be a consequence of hypertension, tubulotoxic effects of urographic contrast media administered during angiography preceding MR imaging, the longer anesthesia time required for the current investigation, and/or possible dehydration during the experiment. We hypothesized that renal ischemia might lead to decreased urine and/or blood content and therefore to faster Tl relaxation. Total renal urine volume may have been reduced in the ischemic kidneys, which had decreased size and weight relative to the contralateral non-ischemic kidney. However, the differences in renal blood volume between kidneys, at most, corresponded to differences in size. Thus, significant differences in blood content per gram are unlikely. In addition, we found no significant difference in water
Contralateral 13 41 30 19 23
mg)
Ischemic
I/C
37 6 11 25 2
2.80 0.15 0.37 1.31 0.09
kidneys.
content per gram in any portion of ischemic versus non-ischemic kidneys. Therefore, the decreased T, relaxation time of the ischemic kidney relative to the contralateral kidney cannot be accounted for on the basis of differences in blood or water content. In the absence of significant differences in tissue content of blood or water, the explanation for the observed differences in T, are unclear. Alterations in blood flow may account for observed differences in MR relaxation times. Capillary blood flow is thought to play a role in MR image appearance,’ although its potential effect on IR images and on Tl relaxation is unclear. Another possible explanation is differences in the proportion of intracellular and extracellular water, or altered intracellular water structure, with total water content remaining similar. By visual analysis of IR images with T, of 600 msec, we noted a pattern indicative of decreased renal perfusion. In the contralateral kidney, a 5-lo-mm low signal band corresponding primarily to the outer stripe of the outer medulla was noted, similar to that previously described for normal kidneys.8 This band was slightly thicker than the 3 mm bands we noted in four of five normal rabbits from the previous experiment, however. In ischemic kidneys, thickness of this band was reduced, and it could not be identified in two kidneys. This band was not seen in the rabbit with acute RAS secondary to catheter-induced intimal dissection. This finding may therefore have predictive value regarding decreased renal perfusion. It should not be inferred, however, that the actual thickness of the outer stripe decreases to this extent with ischemia. More likely, decreased renal blood flow alters the T, of the outer medulla so that portions of it no longer exhibit relaxation to the “null point” of the IR relaxation curve. The abundance of vessels in the outer stripe7,‘4 may account for the more apparent changes in imaging characteristics of this zone. Further in-
Induced renal artery stenosis 0
DONALD G. MITCHELL ET AL
123
Fig. 4. Autoradiographic and histologic detection of microspheres (rabbit #7, non-ischemic kidney). (a) Autoradiograph depicting activity from 113 sn-microspheres restricted primarily to the cortex (arrows). Little activity was present in the ischemic kidney. (b) Histologic section (Hematoxalyn and Eosin). Microspheres can be seen lodged in efferent glomerular arterioles (arrows). Few microspheres were seen in the ischemic kidney.
a
b
Volume 6, Number
vestigation is needed to elucidate the physiologic alterations accounting for these changes in relaxation parameters and image appearance. Our method of measuring renal blood flow by labeled microspheres detects glomerular blood flow by embolization of afferent arterioles. Results by this method are, therefore, expected to differ from measurement of blood flow by radioactive krypton13 or xenon.’ Post-glomerular blood flow is not measured by labeled microspheres. Thus, the abundance of vessels in the outer stripe of the outer medulla, which may play an important role in the MRI appearance of this zone, is not reflected in the results of flow measurement by microspheres. Our apparently anomalous observation of increased microsphere delivery to the medulla in two of four ischemic kidneys may be secondary to vasoconstriction of cortical arteries, resulting in shunting of blood to the medulla (Trueta phenomenon), a reflex reaction which is particularly prominent in the rabbit.3,14 Investigations in the human kidney have disclosed that decreased corticomedullary contrast is indicative of renal pathology. I2 We observed this in our rabbit model of RAS, but, presumably due to the small size of our series, these results were not significant. We caution against direct extrapolation of our results to the human kidney, however, which differs in several respects regarding regulation of blood flow and urinary concentration.3*4*‘0*14 Certain limitations in the design of our study must be noted. Most important is the small size of our series, reducing the potential significance of negative observations. We only had one control rabbit, since an acute RAS was unintentionally produced by catheterrelated intimal dissection. To partially compensate for this deficiency, we have analyzed images of five normal rabbits obtained in a prior experiment in this laboratory. We are not able to determine from our data whether innate differences in size between left and right kidneys, such as exist in humans,2 might have influenced our results. This is not likely, however, because the six left and two right ischemic kidneys were all smaller than the contralateral kidneys. We attempted to minimize the effect of redistribution of renal water by ligating the renal pedicle (artery, vein, and ureter) prior to removing the kidneys and by using a sharply beveled instrument to obtain a core of renal tissue. It is possible, however, that water content of the kidneys may have changed during the imaging experiment or euthanasia of the animal. In some rabbits, we attempted to prevent intrarenal redistribution of water by freezing. The cortico-medullary gradient in water content in these kidneys was abolished, how-
2, 1988
ever, presumably indicating that intrarenal distribution of water was augmented. Future investigations involving alternative pulse sequences and MR contrast agents, using our model of RAS validated by angiography and radionuclide methods, should increase our understanding of renal pathophysiology and its capacity to be evaluated by MRI. REFERENCES 1. Blaufox, M.D. Methods for measurement blood flow. Prop. Nucl. Med. 2:71-84;
of the renal 1972.
2. Brandt, T.B.; Neiman, H.L.; Dragowski, M.J.; Bulawa, W.; Claycamp, G. Ultrasound assessment of normal renal dimensions. J. Ultrasound Med. 1:49-52; 1982.
3. Brewer, N.R. Some oddities of the rabbit kidney. Syn4.
5.
6.
7. 8.
9.
10.
II.
12.
13.
14.
15.
apse (American Society of Laboratory Animal Practitioners) 10:19-22; 1977. Dicker, S.E.; Weller, W. Relationship between glomerular filtration rate and urine flow in the rabbit. Science 112:340-341; 1950. Front, D.; Israel, 0.; Groshar, D.; Weininger, J. Technetium - 99m-labeled red blood cell imaging. Sem. Nucl. Med. 14:226; 1984. Heptinstall, R.H. Hypertension: I. Mechanisms of production. In: Pathology of the Kidney, Hepinstall, R.H., ed. 3rd Ed. Boston: Little, Brown and Company; 1984: 141-180. Jamison, R.L.; Kriz, W. Urinary concentrating mechanism. New York: Oxford University Press; 1982. Kundel, H.L.; Schlakman, B.; Joseph, P.M.; Fishman, J.E.; Summers, R. Water content and NMR relaxation time gradients in the rabbit kidney. Invest. Radiol. 21:12-17; 1986. LeBihan, D.; Breton, E.; Lallemond, D.; Grenier, P.; Cabanis, E.; Laval-Jeantet, M. MR imaging of intravoxel incoherent motions: Application to diffusion and perfusion in neurologic disorders. Radiology 161:401407; 1986. McAfee, J.G.; Subramanian, G. Experimental models and evaluation of animal data for renal radiodiagnostic agents. In: Lambrecht, R.M.; Eckleman, W.C., Eds. Animal Models in Radiotracer Design, New York: Springer-Verlag, Inc.; 1983:21 l-227. Osada, Y.; Masaki, Z.; Momose, S. Bilateral renal artery stenosis and renovascular hypertension in rabbits. J. Ural. 127:811-815; 1982. Terrier, F.; Hricak, H.; Dooms, G.C.; Justich, E. Diagnostic value of the renal corticomedullary contrast on MR images. Radiology 157(P):727; 1985. Thorburn, G.D.; Kopald, H.H.; Herd, J.A.; Hollenberg, M.; O’Morchoe, C.C.C.; Barger, A.C. Intrarenal distribution of nutrient blood flow determined with krypton 85 in the unanesthetized dog. Circ. Res. 13: 290-307; 1963. Trueta, J.; Barclay, A.F.; Daniel, P.M.; Franklin, K. J.; Prichard, M.M.L. Studies of the renal circulation. London: Blackwell Press; 1947. Yuasa, Y.; Kundel, H.L. Magnetic resonance imaging following unilateral occlusion of the renal circulation in rabbits. Radiology 154:151-156; 1985.
Imaging. Vol. 6, pp. All rights reserved.
113-124,
1988 Copyright
0730-725X/88 $3.00 + .oO El 1988 Pergamon Press plc
l Original Contribution
INDUCED RENAL ARTERY STENOSIS IN RABBITS: MAGNETIC RESONANCE IMAGING, ANGIOGRAPHY, AND RADIONUCLIDE DETERMINATION OF BLOOD VOLUME AND BLOOD FLOW DONALD G. MITCHELL,* MICHAEL TOBIN, * ROBERT LEVEEN,~ JOHN TOMACZEWSKI, ** ABASS ALAVI,* MUNI STAUM,* AND HAROLD KUNDEL* Departments of Radiology* and Pathology, ** Hospital of the University of Pennsylvania, Philadelphia, PA 19104, TMedical Research Service, Veterans Administration Hospital, Philadelphia, PA 19104. To investigate the ability of MRI to detect alterations due to renal ischemin, a rabbit renal artery stenosis (RAS) model was developed. Seven rabbits had RAS induced by surgically encircling the artery with a polyethylene band which had a lumen of 1 mm, 1 to 2 weeks prior to imaging. The stenosis was confirmed by angiography, and the rabbits were then imaged in a 1.4 T research MRI unit. T, was calculated using four inversion recovery sequences with different inversion times. Renal blood flow, using “3Sn-microspheres, and regional water content by drying were then measured. The average Tl of the inner medulla was shorter for the ischemic (1574 msee) than for the contralateral kidney (1849 msec), while no change was noted in the cortex. lschemic kidneys had less distinct outer medullary zones on IR images with TI = 600 msec than did contralateral or control kidneys. Blood flow to both the cortex and medulla were markedly reduced in ischemic kidneys compared with contralateral kidneys (119.5 vs. 391 ml/min/lOO gm for cortex and 19.8 vs. 50.8 ml/min/lOti gm for medulla). Renal water and blood content were less affected. Our rabbit model of renal artery stenosis with MRI, radionuclide, and angiographic correlation has the potential to increase our understanding of MR imaging of the rabbit kidney. Keywords: Kidney; Vessels; Bloodflow.
relaxation and to better understand its underlying pathophysioiogy, we correlated renal MRI appearance and regional tissue Tr relaxation times with measured reductions of renal blood flow, blood volume, and water content in a rabbit RAS model.
INTRODUCTION A non-invasive method for accurate detection and evaluation of renovascular disease has not yet been found. Magnetic resonance imaging (MRI) has potential in this area because of the effect on tissue relaxation of water content, water structure, blood flow, and other parameters which may be influenced by renovascular disease. Preliminary investigations on MRI of the rabbit kidney reveal that T, and T2relaxation times reflect the amount of water present*,15 and that complete vascular or ureteral obstruction alters these relaxation times.” The effect of renal artery stenosis (RAS) remains undetermined. In order to determine the effect of RAS on tissue
MATERIALS
AND METHODS
Surgery
Nine male New Zealand White rabbits weighing approximately 3 kilograms each were studied. Using sterile technique and ketamine (40 mg/kg) and xylazine (5 mg/kg) anesthesia, we induced a stenosis of the left (n = 6) or right (n = I) renal artery by encircling it with a 1 cm long polyethylene band. The band
RECENED 7/2/87; ACCEPTED 8/27/87. Acknowledgments-We are indebted to EIanor Kasab,
Address correspondence and reprint requests to Donald G. Mitchell, M.D., Department of Radiology, Division of
Bethanne Moore, and Deborah DeSimone for technical expertise, and to Rachel Knowles and Emily N. Pompetti for manuscript preparation.
MRI, Thomas Jefferson University Hospital, Main Building, 10th Floor, Tenth and Sansom Streets, Philadelphia, PA 19107. 113
114
Magnetic
Resonance
Imaging
had a lumen of 1 mm, and had a longitudinal slit to allow its placement around the artery. A suture was placed around the band to assure a permanent stenosis. The band was left in place 1 to 2 weeks prior to imaging. The remaining two animals had no surgery. Angiography Immediately before MR imaging, an abdominal aortogram was performed from a carotid catheterization to confirm the stenosis. Approximately 5 cc of 60% meglumine diatrizoate were used in each case. In one non-surgerized rabbit, unintentional intimal dissection resulted in an acute right renal artery stenosis. Thus, only one rabbit was a true control. Catheters were also placed in a common iliac artery via a femoral approach to allow sampling of distal arterial blood, and in a marginal ear vein to administer medications. Magnetic Resonance Imaging The rabbits were imaged in a 1.4 Tesla research unit that has a 6-inch imaging bore. The appropriate axial plane was located using spin-echo images with repetition time (TR) of 200 msec and echo time (TE) of 10 msec. Tr was calculated in eight rabbits using four inversion recovery (IR) images with TR of 3000 msec, TE of 10 msec, and inversion times (TI) of 100, 600, 1200, and 2000 msec (Fig. 1). The method of calculation was similar to that used in a previous report from this institution, and involved sampling seven to nine contiguous regions from the cortex to medulla consisting of 9 pixels each.8 In six rabbits predominately Tz-weighted spin-echo images of each kidney were obtained using TR of 2500 msec and TE of 80 msec. Corticomedullary intensity ratios were obtained using 9 pixel boxes placed over cortex and medulla in each image. Glass tubes containing 0.5, 1.O, and 2.0 millimolar copper sulfate were included with each scan for calibration. Using IR images with TR/TE/TI = 3000/10/600 msec, renal area was estimated (0.785 times anteroposterior and transverse diameters). These were compared with in vivo images of five normal left kidneys in similar rabbits, as well as one high resolution image of a normal rabbit kidney ex vivo (IR: TR/TE/TI = 3000/10/600 msec), which were obtained in a previous experiment in this laboratory.8 Radionuclide Blood Volume Measurement Renal blood volume was determined in six rabbits using 99mT~labeled red blood cells (99mTc-RBC’s).5 During MRI, approximately 1.5 hours prior to nuclear imaging, 2 cc of the rabbit’s blood were withdrawn and labeled with approximately 10 mCi of double-eluted 99mT~04- using a commercially avail-
0 Volume
6, Number
2, 1988
able kit (Cadema Medical Products, Inc.; Red Blood Cell Kit for preparing 99mT~04- labeled RBC’s). Following a single washing with normal saline, radiotracer tagging of RBC’s was in excess of 98%. Following MRI, the aortic catheter was repositioned into the left ventricle (LV) using fluoroscopic guidance. With the dorsal surface of the rabbit positioned on top of a gamma camera fitted with a low energy all-purpose collimator, 3-5 mCi of 99mT~RBC were injected as a bolus into the LV. Analogue and digital images were collected at a rate of 2 set per image for 60 sec. Sixty set static images were then obtained for 10 min. Selected animals were also imaged in the lateral and ventral positions and based upon these images, it was decided not to depth correct renal activity for soft tissue attenuation. Approximately 4 cc of blood were then withdrawn from the femoral artery catheter. This blood was imaged for 1 min for calibration and was then used to obtain a central hematocrit. The blood was then centrifuged to check the stability of the 99mTc-RBC bond. Regions of interest were drawn around each kidney using the final dorsal static images. Background corrected renal counts were then compared with the counts from the 4 cc blood samples to obtain absolute renal blood volumes. Radionuclide Blood Flow Measurement In five rabbits, quantitative cortical and medullary blood flow was determined with microspheres. FiftyuCi ‘13Sn-labeled 15 A microspheres were ultrasonicated until the time of administration, and were hand injected over 60 set into the LV catheter. Beginning at the time of injection, blood was withdrawn via the iliac artery catheter at 1 ml per min with a Harvard pump. The animal was then sacrificed. Regional blood flow was determined 2 days following autopsy (see below). Total anesthesia time (angiography, MRI, radionuclide) averaged 6-7 hours. Post Mortem Analysis Following MRI and radionuclide studies, the kidneys were exposed via midline incision. The vascular pedicle of each kidney was located and a suture was placed around each. The sutures were then tied, preventing blood or urine loss, and the animal was sacrificed using intravenous pentobarbital. The kidneys were removed and then weighed. For five of the first six animals, a core of renal tissue was obtained via a sharply beveled thin-walled 8 mm diameter tube inserted into the renal hilum. This core was sectioned into cortex, outer and inner stripe of the outer medulla, and inner medulla.8 These sections of renal tissue were then placed in tared bottles,
Induced renal artery stenosis 0
DONALD
G.
MITCHELL ET AL.
115
Fig. 1. MRI appearance of the left kidney in the control animal, axial images. (a)-(d) IR images with TR 3000 msec, TE 10 msec. (a) Tl = 2000 msec. The medulla (arrow) has slightly lower intensity than the cortex (arrowheads), due to its longer r,. (b) TI = 1200 msec. The contrast between cortex and medulla is increased. The outer medulla (arrowheads) has intermediate intensity. (c) TI = 600 msec. The inner medulla now has higher signal intensity than the cortex. A low signal junctional band (arrows) is now visible between the inner medulla and cortex, presumably representing the outer medulla. (d) TI = 100 msec. The cortex and medulla are both higher intensity than fat. (e) Spin-echo image (TR 2500 msec, TE 80 msec). The medulla has higher signal intensity than the cortex due to its higher T2. e
weighed, and then dried to constant weight in a vacuum oven at 50” to determine water content. The kidneys in the last three rabbits studied were frozen in liquid nitrogen and sectioned in an axial plane. Slices were retained for autoradiography and histologic analysis. The remaining tissue was divided into cortex, outer medulla, and inner medulla. The specimens were weighed and placed into tubes. After allowing 2 days for 99mTC to decay, renal cortical and inner medullary samples were counted in a well counter using a 320-440 kev window. (Outer
medulla was not individually counted due to uncertainty of whether parts of cortex or inner medulla were included.) Counts were obtained for cortex, medulla, and total kidney as well as for the blood withdrawn during the microsphere injection. These values, along with the rate of withdrawal, allowed calculation of the cortical, medullary, and total blood flow per gram for each kidney. Autoradiographs were obtained by placing a section of renal tissue on a single emulsion nuclear medicine blue-based radiographic film for 3 days to 4 weeks.
116
Magnetic Resonance Imaging 0 Volume 6, Number 2, 1988
In three rabbits, sections were saved for histologic analysis. These were fixed in formalin and stained with hemotoxylin and eosin. Note was made of any pathologic alterations, as well as the location of microspheres. The number of microspheres per 100 glomeruli in both cortex and medulla was recorded. Statistical significance was determined using a paired t-test. RESULTS Angiography
The induced renal artery stenosis ranged from 80% (Fig. 2a) to total occlusion with collateral supply in one animal. In three rabbits, decreased nephrogram density and a delayed pyelogram were noted as well. In one of the non-surgerized animals a stenosis of the right renal artery due to intimal dissection was seen, accompanied by delayed filling of intrarenal vessels. Thus, this animal had an acute right renal artery stenosis and was not a control animal. MR Images
Renal area varied between 3.3 to 5.3 cm2 for the non-ischemic kidney, with a mean of 4.3 + 0.7 cm2. This was smaller than the normal left kidneys from the earlier experiment,8 which varied between 4.2 and 6.9 cm2 (mean 5.6 f 1.0 cm2). The ischemic kidneys were smaller than the contralateral kidneys (p < O.Ol), with a size ranging from 2.7 to 4.2 cm2 with a mean of 3.5 f 0.9 cm2 (Table 1). Cortex and medulla could best be distinguished based on relative intensities on IR images with TI of 1200 msec. With this sequence, the medulla was less intense than the cortex in both ischemic and non-
Table 1. Cross-sectional
area and width of junctional
Cross sectional No.
Ischemic
kidney
R L L L L L L* R** Control
Contralateral 4.2 3.3 5.3 3.6 4.2 5.1 3.4 ::;
ischemic kidneys (Figs. 2d,g). Mean corticomedullary intensity ratio was 2.0 f 0.8 for ischemic and 1.8 f 0.23 for non-ischemic kidneys (not significantly different). Intrarenal zones were best depicted in IR images with TI of 600 msec. In these images the cortex and inner medulla both had intermediate intensity. A low signal “junctional band” separated them, corresponding primarily to the outer stripe of the outer medulla (Fig. lc). 8*9The inner stripe of the outer medulla was less well defined. In four of the five normal left kidneys in the previous experiments and in both kidneys in the control animal in this experiment, the low signal junctional band measured 3 mm in diameter. In the image of a normal resected kidney, this band corresponded to the outer stripe of the outer medulla, which also measured 3 mm. In non-stenotic kidneys in experimental animals this band was well seen, ranging in thickness from 4 to 7 mm with a mean thickness of 5.7 f 0.9 mm (Fig. 2e). In ischemic kidneys, however, this band was less well seen (Fig. 2h), and in two of the six kidneys with surgically induced stenosis it could not be identified. In the four ischemic kidneys in which the junctional band could be seen, its thickness was decreased (p < 0.05), ranging from 1 to 2 mm, with a mean thickness of 1.8 f 0.4 mm. In the rabbit with RAS due to intimal dissection, the junctional band in the affected kidney was obscured by low signal of the medulla. With long TR/TE spin-echo images (predominately T2 weighted), the medulla had higher intensity than cortex (Figs. le, 2f, 2i). Mean corticomedullary intensity ratio (Table 2) was 0.8 1 f 0.19 for ischemic and 0.44 f 0.17 for nonischemic kidneys (not significantly different).
(R)
I/C = Ratio of ischemic to contralateral kidneys. *Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis.
area
(CM2)
band Stripe width (MM2)
Ischemic
I/C
3.9 3.2 4.2 3.0 2.7 4.2 1.9 4.6 4.2 (L)
0.95 0.96 0.79 0.82 0.62 0.82 0.56 0.80 0.89
Contralateral 6 No IR 3000/10/600 6 7 4 5 6 3 3
Ischemic Obscured images 2 2 2 Obscured 1 Obscured 3
Induced renal artery stenosis 0
DONALD G. MITCHELL ET AL.
. .
P
117
118
Magnetic
Resonance
Imaging
0 Volume
I a
6, Number
2, 1988
i
Fig. 2. (g)-(i) Axial MR images of the ischemic left kidney. (g) IR image (TR 3000 msec, TI 1200 msec, TE 10 msec). Contrast between cortex and medulla is decreased in comparison to Fig. l(b). (h) IR image (TR 3ooOmsec, Tl 600 msec, TE 10 msec. A thin low signal line (arrows) separates the cortex from medulla. (i) Spin-echo image (TR 2500 msec, TE 80 msec). Contrast between cortex and medulla is decreased in comparison with Figs. l(e) and 2(d). Contrast between cortex and medulla in (d) and (f) is greater than seen in (g) and (i). The junctional band in (e) appears thicker than that seen in Fig. 1 (c), but the images are otherwise similar to the control kidney in Figs. l(b)-(e).
120
Magnetic Resonance Imaging 0 Volume 6, Number 2, 1988 Table 2. Summary
of MRI intensity
measurements
Calculated No.
Ischemic
kidney
R L L L L L L* R** Control
expressed
as cortico-medullary
T, C/M ratio
Contralateral
C/M intensity
Ischemic
0.50
ratio (SE 2500/80)
Contralateral
Ischemic
0.68
0.48 0.48
0.46 0.84 0.48 0.94 0.45 0.40 0.30
No SE 2500/80 images 0.64 0.94 No SE 2500/80 0.63 0.18 1.00 0.65 0.47 0.58 0.43 No SE 2500180 0.57
Not calculated 0.45 0.43 0.51 0.66 0.50 0.48 0.46
ratios
0.50 0.79
I/C = Ratio of ischemic to contralateral kidneys. *Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis.
T, Calculation (Table 2) For all six rabbits with surgically induced RAS which had T, calculated, a gradient of increasing T, from cortex to medulla was noted in both ischemic and nonischemic kidneys. The mean T, gradient from cortex to medulla of non-ischemic kidneys was 914 msec (935 to 1849 msec), compared with 588 msec (986 to 1574 msec) for ischemic kidneys (Fig. 3). This is less than the Tl gradient for the control rabbit (mean of two kidneys), which was 1230 (720 to 1950 msec). The relaxation times of all the rabbits were longer than those previously reported from this laboratory for normal kidneys of 600 to 1200 msec.’ T, of cortex was similar for both ischemic and non-ischemic kidneys, but both the outer (junctional band) and inner medulla of the ischemic kidneys had lower T, than the corresponding zones in non-ischemic kidneys. Radionuclide Blood Volume Measurement In vivo blood volume was determined in five rabbits (Table 3). Dynamic images following a bolus of 99”TC-RBC’s revealed delayed flow to the ischemic kidney, but RBC volume following equilibration was similar in both kidneys (Figs. 2b, 2~). Blood volume determined from the final static image ranged from 1.7 to 4.6 ml with a mean of 2.8 ml in non-stenotic kidneys, compared to a range of 0.8 to 4.3 ml with a mean of 1.9 ml in ischemic kidneys (not significantly different). Radionuclide Blood Flow Measurement (Table 4) Blood flow to the cortex and medulla of the ischemic kidneys was markedly reduced. In non-stenotic kidneys, the average flow to cortex, medulla, and total
o = Tt x = % H20 by weight
I = average standard deviation 30%
20%
10%
0%
RENAL ZONE FROM CORTEX TO INNER MEDULLA
-10%
I
/
1
2
I
/
I
I
,
4
5
6
7
I
3
Fig. 3. Percent difference between stenotic and non-stenotic kidney of mean T, and water content from cortex to medulla. The difference in T,was largest in the outer medulla, less so in the inner medulla, and no difference was seen in the cortex. These differences did not parallel water content, which was similar in stenotic and non-stenotic kidneys.
Induced renal artery stenosis 0 DONALD G. MITCHELL ET AL.
121
Table 3. Weight, cortico-medullary water gradient and blood volume Weight (gm) No.
Ischemic kidney
1 2 3 4 5 6 7 8
R L L L
Contralateral
= contralateral;
L L L* L**
Contralateral
C/M percent HZ0 Gradient
Ischemic
Not measured Not measured Not measured 13.7 10.4 16.4 16.0 12.6
10.1 10.1 10.8 8.6 10.0
Contralateral 78-85 79-87 81-87 80-87 80-83 81-84 80-82 93-98
(0.92) (0.90) (0.93) (0.93) (0.97) (0.96)+ (0.97) + (0.95)+
Ischemic 81-89 77-88 81-89 85-88 77-91 83-84 81-82 94-88
Blood volume (ml) Contralateral
(0.86) (0.88) (0.91) (0.97) (0.84) (0.99)+ (OH)+ (1.07) +
Ischemic
Not measured Not measured 1.7 1.9 3.7 4.3 2.2 1.2 4.6 1.2 1.7 0.8 Not measured
C/M = cortico-medullary.
*Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis. +Kidneys frozen in liquid nitrogen prior to measurement of percent water.
was 391.0, 50.8, and 312.0 ml per min per 100 gm, respectively. Flow was markedly decreased to the ischemic kidney, being 119.5, 19.8, and 63.0 ml per min per 100 gm to cortex, medulla, and total kidney, respectively. Due to the small size of this sample, however, these results were not significant at the p < 0.05 level. Relative to the contralateral kidney, cortical blood flow was reduced in all four ischemic kidneys, but medullary flow was increased in two of four ischemic kidneys. Autoradiographs of renal sections were obtained in three rabbits, which confirmed the observation that most flow was to the cortex, which was 3 mm thick (Fig. 4a), and that flow was markedly reduced to the ischemic kidneys. Microspheres were counted histologically in two rabbits. Microspheres were located primarily in juxtaglomerular efferent arterioles (Fig. 4b). The average number counted per 100 glomeruli was 26 and 15 for cortex and medulla, respectively, in non-ischemic kidneys, compared with 5.5 and 8.5 for ischemic kidneys. kidney
Renal Weight In four rabbits with surgically induced RAS, post mortem renal weight was determined (Table 3). It varied from 10.4 to 16.4 gm in non-ischemic kidneys, with a mean of 14.1 gm. The weight of ischemic kidneys was reduced (p < O.Ol), ranging from 8.6 to 10.8 gm, with a mean of 9.9 gm. Water Content A gradient of increasing water content from cortex to medulla was noted in the five rabbits whose kidneys were not frozen prior to sectioning (Table 3). It
was similar in non-ischemic and ischemic kidneys (Fig. 3), ranging from 79.6 to 86.3% in the former and 79.4 to 89.0% in the latter. The gradient was abolished in the three rabbits whose kidneys were frozen. Histologic Analysis The kidneys in four rabbits were examined histologically. In the kidney with an occluded renal artery, large areas of infarction were seen. In one kidney with RAS, mild juxtaglomerular hyperplasia was noted. In all kidneys which were frozen, areas of cortical infarction were seen. DISCUSSION
In this experiment, we surgically induced unilateral renal artery stenosis (RAS) in seven rabbits, and used angiography for anatomic confirmation. Quantitative measurement of blood flow by labeled microspheres was accomplished in four rabbits with surgically induced subacute RAS and in one rabbit with acute RAS, secondary to unintentional catheter-induced intimal dissection. In these rabbits, decreased flow to the ischemic kidney was documented. Further evidence of physiologic significance of the stenosis included decreased weight of the ischemic kidney, as well as decreased size as measured from MR images. The basis for the MRI appearance of kidneys is complex and only partially understood. A prior report from this laboratory demonstrated a gradient of increasing T, relaxation in normal rabbit kidneys from cortex to medulla, correlating with a gradient of increasing tissue water content.* This gradient is pre-
122
Magnetic Resonance Imaging 0 Volume 6, Number 2, 1988 Table 4. Blood flow to cortex and inner medulla
Cortical flow (ml/min/ 100 gm) No. 4 5 6 7 8
Ischemic kidney L L L L* L**
Contralateral 341 811 263 149 76
Medullary flow (ml/min/lOO
Ischemic
I/C
240 10 176 32 7
0.70 0.01 0.67 0.21 0.09
Contralateral = contralateral; I/C = ratio of ischemic to contralateral *Total occlusion with colateral perfusion of kidney. **Unintentional catheter-induced acute renal artery stenosis.
sumably
T, values (gradient between 600 and 1200 msec) than we noted, even in the control rabbit. The longer T, values and smaller size of the kidneys in the current experiment may be a consequence of hypertension, tubulotoxic effects of urographic contrast media administered during angiography preceding MR imaging, the longer anesthesia time required for the current investigation, and/or possible dehydration during the experiment. We hypothesized that renal ischemia might lead to decreased urine and/or blood content and therefore to faster Tl relaxation. Total renal urine volume may have been reduced in the ischemic kidneys, which had decreased size and weight relative to the contralateral non-ischemic kidney. However, the differences in renal blood volume between kidneys, at most, corresponded to differences in size. Thus, significant differences in blood content per gram are unlikely. In addition, we found no significant difference in water
Contralateral 13 41 30 19 23
mg)
Ischemic
I/C
37 6 11 25 2
2.80 0.15 0.37 1.31 0.09
kidneys.
content per gram in any portion of ischemic versus non-ischemic kidneys. Therefore, the decreased T, relaxation time of the ischemic kidney relative to the contralateral kidney cannot be accounted for on the basis of differences in blood or water content. In the absence of significant differences in tissue content of blood or water, the explanation for the observed differences in T, are unclear. Alterations in blood flow may account for observed differences in MR relaxation times. Capillary blood flow is thought to play a role in MR image appearance,’ although its potential effect on IR images and on Tl relaxation is unclear. Another possible explanation is differences in the proportion of intracellular and extracellular water, or altered intracellular water structure, with total water content remaining similar. By visual analysis of IR images with T, of 600 msec, we noted a pattern indicative of decreased renal perfusion. In the contralateral kidney, a 5-lo-mm low signal band corresponding primarily to the outer stripe of the outer medulla was noted, similar to that previously described for normal kidneys.8 This band was slightly thicker than the 3 mm bands we noted in four of five normal rabbits from the previous experiment, however. In ischemic kidneys, thickness of this band was reduced, and it could not be identified in two kidneys. This band was not seen in the rabbit with acute RAS secondary to catheter-induced intimal dissection. This finding may therefore have predictive value regarding decreased renal perfusion. It should not be inferred, however, that the actual thickness of the outer stripe decreases to this extent with ischemia. More likely, decreased renal blood flow alters the T, of the outer medulla so that portions of it no longer exhibit relaxation to the “null point” of the IR relaxation curve. The abundance of vessels in the outer stripe7,‘4 may account for the more apparent changes in imaging characteristics of this zone. Further in-
Induced renal artery stenosis 0
DONALD G. MITCHELL ET AL
123
Fig. 4. Autoradiographic and histologic detection of microspheres (rabbit #7, non-ischemic kidney). (a) Autoradiograph depicting activity from 113 sn-microspheres restricted primarily to the cortex (arrows). Little activity was present in the ischemic kidney. (b) Histologic section (Hematoxalyn and Eosin). Microspheres can be seen lodged in efferent glomerular arterioles (arrows). Few microspheres were seen in the ischemic kidney.
a
b
Volume 6, Number
vestigation is needed to elucidate the physiologic alterations accounting for these changes in relaxation parameters and image appearance. Our method of measuring renal blood flow by labeled microspheres detects glomerular blood flow by embolization of afferent arterioles. Results by this method are, therefore, expected to differ from measurement of blood flow by radioactive krypton13 or xenon.’ Post-glomerular blood flow is not measured by labeled microspheres. Thus, the abundance of vessels in the outer stripe of the outer medulla, which may play an important role in the MRI appearance of this zone, is not reflected in the results of flow measurement by microspheres. Our apparently anomalous observation of increased microsphere delivery to the medulla in two of four ischemic kidneys may be secondary to vasoconstriction of cortical arteries, resulting in shunting of blood to the medulla (Trueta phenomenon), a reflex reaction which is particularly prominent in the rabbit.3,14 Investigations in the human kidney have disclosed that decreased corticomedullary contrast is indicative of renal pathology. I2 We observed this in our rabbit model of RAS, but, presumably due to the small size of our series, these results were not significant. We caution against direct extrapolation of our results to the human kidney, however, which differs in several respects regarding regulation of blood flow and urinary concentration.3*4*‘0*14 Certain limitations in the design of our study must be noted. Most important is the small size of our series, reducing the potential significance of negative observations. We only had one control rabbit, since an acute RAS was unintentionally produced by catheterrelated intimal dissection. To partially compensate for this deficiency, we have analyzed images of five normal rabbits obtained in a prior experiment in this laboratory. We are not able to determine from our data whether innate differences in size between left and right kidneys, such as exist in humans,2 might have influenced our results. This is not likely, however, because the six left and two right ischemic kidneys were all smaller than the contralateral kidneys. We attempted to minimize the effect of redistribution of renal water by ligating the renal pedicle (artery, vein, and ureter) prior to removing the kidneys and by using a sharply beveled instrument to obtain a core of renal tissue. It is possible, however, that water content of the kidneys may have changed during the imaging experiment or euthanasia of the animal. In some rabbits, we attempted to prevent intrarenal redistribution of water by freezing. The cortico-medullary gradient in water content in these kidneys was abolished, how-
2, 1988
ever, presumably indicating that intrarenal distribution of water was augmented. Future investigations involving alternative pulse sequences and MR contrast agents, using our model of RAS validated by angiography and radionuclide methods, should increase our understanding of renal pathophysiology and its capacity to be evaluated by MRI. REFERENCES 1. Blaufox, M.D. Methods for measurement blood flow. Prop. Nucl. Med. 2:71-84;
of the renal 1972.
2. Brandt, T.B.; Neiman, H.L.; Dragowski, M.J.; Bulawa, W.; Claycamp, G. Ultrasound assessment of normal renal dimensions. J. Ultrasound Med. 1:49-52; 1982.
3. Brewer, N.R. Some oddities of the rabbit kidney. Syn4.
5.
6.
7. 8.
9.
10.
II.
12.
13.
14.
15.
apse (American Society of Laboratory Animal Practitioners) 10:19-22; 1977. Dicker, S.E.; Weller, W. Relationship between glomerular filtration rate and urine flow in the rabbit. Science 112:340-341; 1950. Front, D.; Israel, 0.; Groshar, D.; Weininger, J. Technetium - 99m-labeled red blood cell imaging. Sem. Nucl. Med. 14:226; 1984. Heptinstall, R.H. Hypertension: I. Mechanisms of production. In: Pathology of the Kidney, Hepinstall, R.H., ed. 3rd Ed. Boston: Little, Brown and Company; 1984: 141-180. Jamison, R.L.; Kriz, W. Urinary concentrating mechanism. New York: Oxford University Press; 1982. Kundel, H.L.; Schlakman, B.; Joseph, P.M.; Fishman, J.E.; Summers, R. Water content and NMR relaxation time gradients in the rabbit kidney. Invest. Radiol. 21:12-17; 1986. LeBihan, D.; Breton, E.; Lallemond, D.; Grenier, P.; Cabanis, E.; Laval-Jeantet, M. MR imaging of intravoxel incoherent motions: Application to diffusion and perfusion in neurologic disorders. Radiology 161:401407; 1986. McAfee, J.G.; Subramanian, G. Experimental models and evaluation of animal data for renal radiodiagnostic agents. In: Lambrecht, R.M.; Eckleman, W.C., Eds. Animal Models in Radiotracer Design, New York: Springer-Verlag, Inc.; 1983:21 l-227. Osada, Y.; Masaki, Z.; Momose, S. Bilateral renal artery stenosis and renovascular hypertension in rabbits. J. Ural. 127:811-815; 1982. Terrier, F.; Hricak, H.; Dooms, G.C.; Justich, E. Diagnostic value of the renal corticomedullary contrast on MR images. Radiology 157(P):727; 1985. Thorburn, G.D.; Kopald, H.H.; Herd, J.A.; Hollenberg, M.; O’Morchoe, C.C.C.; Barger, A.C. Intrarenal distribution of nutrient blood flow determined with krypton 85 in the unanesthetized dog. Circ. Res. 13: 290-307; 1963. Trueta, J.; Barclay, A.F.; Daniel, P.M.; Franklin, K. J.; Prichard, M.M.L. Studies of the renal circulation. London: Blackwell Press; 1947. Yuasa, Y.; Kundel, H.L. Magnetic resonance imaging following unilateral occlusion of the renal circulation in rabbits. Radiology 154:151-156; 1985.