Renal resistive index in experimental partial and complete ureteral obstruction

Renal Resistive Index in Experimental Partial and Complete Ureteral Obstruction Brian D. Coley, MD 1, Ronald S. Arellano, MD 1, Lee B. Talner, MD 1,2, Kristine G. Baker 1, Tom Peterson t, Robert F. Mattrey, MD 1

Rationale and Objectives. Recent clinical work suggests that the Doppler resistive index (RI) may be useful in distinguishing obstructive from nonobstructive hydronephrosis. We evaluated the usefulness of the RI in a rabbit model of hydronephrosis. Methods. Unilateral partial ureteral obstruction was produced in nine rabbits and complete obstruction in another nine. Three sham operations were performed, and these animals served as control subjects. The RI was measured in all kidneys before and 6 hr after surgery and on days 1, 4, and 7 postoperatively. The RI and the difference in RI (ARI) between the obstructed and normal kidney were evaluated over time using a two-way analysis of variance. The intravenous urography and Whitaker tests served as gold standards. Results. Hydronephrosis was o b s e r v e d on sonograms in all obstructed kidneys. Comparing groups, there was no significant difference in mean RI or ARI between the three groups at any time point. Looking at individual groups over time, there was no significant change in mean ARI, whereas the change in mean RI was significantly elevated above baseline only in the complete obstruction group at 6 hr ( p = .002) and on days 4 ( p = .008) and 7 ( p = .006). In evaluating varying thresholds of RI and ARI, we could not consistently discriminate between normal and obstructed kidneys. Conclusion. Although complete obstruction caused a significant increase in RI, partial obstruction failed to do so. RI and ARI values proved to be insensitive predictors of obstruction in this rabbit model. Key Words. Resistive index; obstructive hydronephrosis; Doppler ultrasound; renal dilatation. From the 1Department of Radiology, University of California, San Diego; and 2current address: Department of Radiology, Harborview Medical Center, Seattle, WA. Address reprint requests to R. R Mattrey, MD, MRI Institute, Universityof California at San Diego Medical Center, 410 Dickinson St., San Diego, CA92103. Received June 27, 1994, and accepted for publication after revision January 8, 1995,

Acad Radio11995,2:373-378 © 1995, Association of University Radiologists

ilatation of the renal collecting system is readily identified by many' imaging techniques, including ultrasound. When the clinical suspicion of subacute or chronic renal obstruction is high, the detection of hydronephrosis by ultrasound is highly sensitive but only moderately specific [1, 2]. Because renal dilatation is not always caused by obstruction [3-5], the distinction between obstructive and nonobstructive dilatation often requires further tests. Recent clinical investigation of the renal arterial resistive index (RI) measured by Doppler ultrasound in patients with and without obstruction 373

D.

C O L E Y ET AL.

has yielded conflicting results, but results of several studies suggest that RI measurements can consistently distinguish obstructive from nonobstructive dilatation [6-9]. If reliable, this would greatly improve the specificity of ultrasound in the assessment of obstructive hydronephrosis. To be of m a x i m u m clinical use, however, renal Doppler ultrasound must yield consistent results in both total and partial obstruction [4, 10]. To this end, we evaluated the renal vascular resistive index in rabbits that had complete or partial ureteral obstruction.

MATERIALS AND METHODS Animal Preparation Twenty-one New Zealand White rabbits (weight = 2--4 kg) were divided into three groups: sham-operated normal controls (n = 3), partial ureteral obsm_tction (n = 9), and complete ureteral obstruction (n = 9). All animals underwent general anesthesia with subcutaneous ketamine hydrochloride (50 mg/kg; Ketaset, Aveco, Fort Dodge, IA) and xylazine hydrochloride (8 mg/kg; Anased, Lloyd Laboratories, Shenandoah, IA) and midline laparotomy. Sham-operated control animals were sewn back up without any other intervention. The remaining animals were randomly assigned to undergo partial or complete unilateral ureteral obstruction. The side of intervention also was randomized. Partial ureteral obstruction was produced by a method described by Claesson et al. [11] in which 1-2 cm of the ureter is embedded within the ipsilateral psoas muscle. The resting tension of the muscle produces a reliable and reproducible partial ureteral obstruction. Complete obstruction was produced by in situ double ligation of the ureter. Animals were allowed free access to food and water throughout the study.

Imaging M1 ultrasound examinations were performed on an Acuson 128 equipped with color Doppler (Acuson, Milpitas, CA). Using a 7-Mhz transducer, the same sonographer, unaware of animal assignment and side of obstruction, acquired gray-scale images and measured the RI from the main and second-order branches, and from the arcuate renal arteries using the Acuson software. Doppler waveform tracings were scaled to optimize RI measurements and calculations. The Doppler gate was kept to a minimum, and the minimum wall filter was used. Only the values from the second-order arteries (equivalent to segmental arteries in human kidneys) were used for analysis because they were the most reliably imaged vessels. Animals were anesthetized for the ultra374

Vol. 2, No. 5, May 1995

sound studies, which were performed the day before surgery, 6 hr after surgery, and then again at 1, 4, and 7 days after surgery. Intravenous urography @VU) followed the 1-, 4-, and 7-day ultrasound study in the partial obstruction group but was done only on day 1 in the complete obstruction group. For IVU, 2 ml/kg of nonionic contrast medium (Optiray 320; Mallinckrodt Medical, St. Louis, MO) was used. The sonographer was unaware of the IVU results until after the last sonogram on day 7. After completion of the imaging studies on day 7, Whitaker tests were performed to confirm the presence of partial obstruction. Two 22-gauge intravenous catheter needle units were inserted into the renal p e n i s above the level of obstruction. The plastic sheath of one was directed toward the kidney, through which dilute contrast material was infused at 3.1 ml per minute. The other was directed toward the obstructed ureter and was connected to a manometer. Tests were performed with fluoroscopic monitoring to ensure that there were no catheter or collecting system leaks. The bladder was decompressed.

Data Analysis T h e normal RI and ARI and their normal ranges (mean + two standard deviations [SDs]) for the rabbit kidney were derived by averaging the 42 RI measure= ments and the 21 ARI measurements acquired from the 21 rabbits scanned prior to surgical intervention. The ARI was calculated as the absolute value of the difference (higher RI minus lower RI value) to avoid the introduction of arithmetic artifacts. The Whitaker data among experimental groups were tested for significance using the unpaired Student's t test. The RI data were compared for the normal and obstructed kidneys in four ways. For the first analysis, the difference in RI (ARI) between the obstructed and nonobstructed kidney was calculated for each animal at each time point. The absolute value was used for the sham-operated control group. The ARI was then grouped for the control group (n = 3), the partial group (n = 9), and the complete obstruction group (n = 9). In the second analysis, the mean RI values for the kidneys with partial ( n = 9) or complete obstruction (n = 9) were compared with the mean RI values for both normal kidneys from the sham-operated control group (n = 6) at each time point. For these two analyses, the difference between experimental groups was tested for significance using a two-way analysis of variance (ANOVA) in which time and experimental group were the independent variables. When the ANOVA was significant, a one-way ANOVA was then performed, followed by an

Vol. 2, No. 5, May 1995

RENAL

RESISTIVE

I N D E X IN U R E T E R A L

OBSTRUCTION

unpaired Student's t test if a significant difference was detected between groups and the paired Student's t test if a significant difference was detected over time within each group. To be considered significant, the probability value had to be less than .05 for a two-tailed distribution. All values are shown as mean +_ standard error of the mean unless otherwise stated as +standard deviation (±SD). In the third and fourth analyses, the RI and kRI were used to calculate the true-positive and false-positive fractions, as their respective thresholds for distinguishing normal from obstructed kidneys was increased from the minimum to the mayJmum value recorded. In both analyses, a plot of true-positive to false-positive fraction was generated, analogous to a receiver operator characteristic curve.

obstructed kidneys reached an equilibrium m e a n pressure of 20.7 _+ 1.0 cm of water (p < .001). As would be expected, pressures within completely obstructed kidneys progressively increased during infusion until extravasation occurred. Imaging

Hydronephrosis was readily detected on gray-scale sonography in partially and completely obstructed kidneys at all time points (Fig. 1). The degree of collecting system dilatation was similar for both groups. IVU confirmed the existence of all experimentally induced partial ureteral obstructions as well as all complete obstructions (Fig. 2). Resistive Indexes

RESULTS Whitaker Tests

Normal control animals had a m e a n renal pelvic pressure of 4.0 _+ 0.2 cm of water during infusion. Partially

The normal RI and zXRIgenerated from the 21 normal rabbits were 0.45 -+ 0.05 (+SD) and 0.027 + 0.026 (±SD), respectively. The mean RI for each group of rabbits over time is shown in Figure 3. The preoperative RIs were

FIGURE 1. Gray-scale sonogram shows partial (A) and complete (B) obstructive hydronephrosis. Note the single medullary pyramid of the unipapillary rabbit kidney (arrow)surrounded by the dilated pelvis (p).

A

B

A

B

FIGURE 2, Intravenous urogram at 30 sec (A) and 5 min (B) in a rabbit with unilateral left partial ureteral obstruction showing the negative pyelogram at 30 sec (arrow in A) and the hydronephrosis with the constricted segment of ureter (arrowheadin B) that was embedded in the ipsilateral psoas.

375

COLEY

ET AL.

vol. 2, No. 5, May 1995

E f f e c t of T i m e on RI

E f f e c t o f T i m e on ARI

in Partial and C o m p l e t e Obstruction

in Partial and C o m p l e t e Obstruction

0,20

.•0,10 r

t

0.15

o.o,

44

=.

0.10

l

l

0.05 0.00 0.00

&

-0.05

L)

i

Pre-operative -o-

-0.05

r

6 hrs

1 day

Normal

-u-

4

-6-

Partial

days

Pre-operative

day

7

Complete

-o-

FIGURE 3. Mean change in resistive index ± standard error of the mean (SEM)are shown for each animal group at each of the 5 time points. No significant difference was detected between groups. * p < .05, relative to baseline value.

similar for all groups (0.46 -+ 0.01 for the control group, 0.43 -+ 0.02 for the partially obstructed group, and 0.46 _+ 0.01 for the completely obstructed group). The two-way ANOVA did not show any positive interaction between time and group when the R1 data were assessed. The two-way ANOVA failed to detect any significant difference in RI among groups, but it did detect a significant effect over time (p = .007). The only significant one-way ANOVA across time was for the complete obstruction group (p=.01), for which a paired t test showed a significant elevation at 6 hr (p = .002), 4 days (p = .008), and 7 days (p = .006). The change in ARI over time is shown in Figure 4. The two-way ANOVA of the ARI values did not show any positive interaction between time and group. There

'

'

4 days

7 day

Normal

-u-

Partial

-6-

Complete

was no significant difference detected between groups or over time within each group. The scatter plots of the RIs and kRIs over time are shown in Figure 5. The majority of the abnormal kidneys fell within the 2 SD limits of normal kidneys. The treepositive data plotted against the false-positive data as the thresholds of normal RI and ARI were increased are shown in Figure 6. Note that as sensitivity increased, specificity decreased markedly. DISCUSSION

Ultrasonography has long been recognized as a sensitive method of detecting dilatation of the renal collecting system and is used to infer the presence or absence of

ARI of Obstructive

Hydronephrosis

Scatter About the Normal Mean ± 2 SD

Scatter About the Normal Mean ± 2 SD 0.3

o o o

0.2

0.6 @ _ _

•~ 0.5

. . . . . . . . . . .

o ~ . . . . . . . . . . . . .

. . . . . . . . . . . . .

o

o@

0.1

-

o

~_ _6

o

. . . . . . . . . . . . .

o

*

o~

o 08

oO o

~ 0.4.

'

1 day

FIGURE 4. Mean change in ARI + standard error of the mean (SEM)is shown for each animal group at each of the five time points. No statistically significant change was detected between groups or relative to baseline within each group. ARI = difference in the resistive index.

RI o f O b s t r u c t i v e H y d r o n e p h r o s i s 0.7

'

6 hrs

0.0

~o

o o

o o o . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

-0.1 o

0.3 6 hrs

o Partial

A

1 day

4 days

7

-0.2

days

o Complete

i

i

i

i

6 hrs

1 day

4 days

7 days

o Partial

o Complete

B

FIGURE 5. Scatter plot of resistive index (A) and ARI (B) from partially and completely obstructed kidneys throughout the observation period are shown relative to the normal mean + 2 standard deviations (SDs) calculated from the 21 normal rabbits imaged prior to surgical intervention. RI = resistive index, ARI = difference in the RI.

376

Vol. 2, No. 5, May 1995

RENAL RESISTIVE

Roe Varying

Varying

f o r RI

o

for ARI

~o 0.6

o.s

RI-0.35-

o

RI-0.45

g 0.6

o

o

0#

o

•~ 0 . 2 o

v----"Rl'0"55 DooD~---RI=0.6

~

-~/RI'°'Tf

0.2

o

0.4

08

0.4

0.0

Threshold

•~ 1.o

0.8

b~

OBSTRUCTION

ROC C u r v e

Curve

Threshold

•~ 1.0

0.0

I N D E X IN U R E T E R A L

•~ 0.2 O

i

i

i

0.4

0.6

0.8

o.o o.o

1.0

Q

\

o

o ARI-O.02

ARI-0,04

o gv--.~ ARI-0.06 i 0.2

i

0.4

i

0.6

0.8

1.0

False Positive Fraction (1-Specificity)

False Positive Fraction (1-Specificity)

A

ooo

o

B

FIGURE 6. Plot of the true-positive fraction as a function of the false-positive fraction for RI (,4) and ARI (B) as the threshold for the normal kidney was varied from the minimum to the maximum value detected in all rabbits. ROC = receiver operating characteristic, RI = resistive index, ARI = difference in the RI.

renal obstruction [1]. Unfortunately, dilatation m a y be attributable to causes other than obstruction [3-5], and the definitive diagnosis may require invasive testing. Previous animal studies using direct blood flow and pressure measurements have shown that during the first 1-2 hr, acute complete ureteral occlusion produces a decrease in renal vascular resistance and an increase in renal blood flow [12, 13]. Complete ureteral occlusion of 1 day's duration or longer produces increased renal vascular resistance and a decrease in renal blood flow [10, 12, 14]. These observations are similar w h e n ureteral obstruction is partial or complete. Ureteral obstruction produces increased pressure within the collecting system, which in turn increases interstitial pressure. The rise in renal interstitial pressure, as well as local humoral responses, produces arterial vasoconstriction and decreased renal blood flow [13, 15]. Humoral factors, which are less well characterized, are likely very important in obstruction of long duration, because pressure within an obstructed collecting system returns to normal after several days to weeks [10, 13-15]. On the basis of this past work and prospective clinical trials, several authors have advocated Doppler RI measurement of the kidney as a useful discriminator between obstructive and nonobstructive renal dilatation [6-9], reporting high sensitivity and specificity. Published animal data on this subject are limited. Dodd et al. [16], using unilateral complete ureteral obstruction in dogs, reported a sensitivity of 74% and a specificity of 77%. They found that the renal RI significantly increased in the first week, but these increases were not consistent over the 4-week

study period. Conceding that there may be limitations to the animal model used, they suggested that the significant percentage of false-positive and false-negative studies may limit the clinical usefulness of this test. This specificity (74%) is similar to that reported by Ellenbogen et al. [1] using sonography without Doppler in patients with suspected subacute or chronic obstruction. Similar to the Dodd et al. study, the complete obstruction group in this study also had an inconsistent elevation in the RI. The RI was normal on day 1 but was elevated thereafter. Although we wanted to use the same RI criteria as that used clinically (normal RI < 0.70, normal ARI < 0.10), potential differences among species required establishing the normal value for rabbits. The mean normal rabbit RI and the standard deviation (0.45 -+ 0.05) were established from 42 measurements obtained from the 21 rabbits imaged prior to surgical intervention. When the RI was greater than 2 SDs over the normal rabbit mean (RI = 0.55), the kidneys were either normal ( n = 5), completely obstructed (n = 11), or partially obstructed ( n = 9). However, the remaining animals, and these same kidneys at other time points, had RI values within the 2 SD limit. We evaluated varying discriminatory RI thresholds in judging kidneys, but as is apparent from Figure 6A, no useful threshold for the RI value was found to discriminate obstructed from normal kidneys. Because clinical obstructive hydronephrosis is most commonly attributable to partial ureteral obstruction [4, 10], the assessment of this condition in a contl~olled experiment is important. Partial obstruction was consistently produced using the model reported here and was 377

COLEY

ET AL.

confirmed by both the Whitaker test [17] and IVU. It produced a degree of hydronephrosis that was not distinguishable from complete obstruction sonographically. The difference between the two obstruction groups was not statistically significant, as assessed by the ANOVA. Furthermore, the partial obstruction group was not significantly different from the control group, nor did it vary significantly from its baseline value. This is consistent with the clinical observation that partial renal obstruction does not appear to produce consistently elevated RIs: Chen et al. [18] reported a sensitivity of only 52% in 33 patients (an RI of less than 0.7 was considered normal). Moreover, of their 18 patients with mild obstruction, 83% had normal RIs, whereas 93% of the 15 patients with severe obstruction had abnormal RIs. In clinical practice, w h e n the RI value of the abnormal kidney is within normal limits, the use of the RI difference between the abnormal and contralateral normal kidney has reportedly improved accuracy and allowed detection of obstruction [6, 19]. This observation was not supported in this study w h e n the ARI was similar for all groups at all time points and did not significantly differ from baseline. In varying discriminatory thresholds for normal kidneys, the ARI suffered a similar disadvantage as the RI value (Fig. 6B). ARI thresholds yielding high sensitivity gave low specificity and thus had limited discriminatory value. One variable that was not controlled in this study and could have affected our results was systemic blood pressure. We attempted to keep the anesthesia level constant by monitoring physiologic responses (e.g., corneal reflex, respiratory and cardiac rate). Blood pressure was probably kept constant in this study, because in a subsequent study [20] in which blood pressure was monitored and a similar anesthetic protocol was followed, blood pressure remained constant throughout the anesthesia period. When comparing data from animal and clinical experiments, one must be aware of potentially different physiologic responses to the same process in different species. We believe that the rabbit model is reasonable because it has been shown that raising ureteral pressure in this species, which can occur with obstruction, will produce an elevation of the RI [21]. In summary, although the RI increased with complete obstruction at nearly all time points tested within the first week after obstruction, it was not significantly

378

Vol. 2, No. 5, May 1995

affected by partial obstruction, nor did not consistently discriminate between the normal and obstructed kidneys. Although the trend for the ARI was to increase with obstruction, it was not significantly affected by obstruction at any time point. Although it is possible that this rabbit model may not reflect the clinical situation, the results of our study and the conflicting clinical experience suggest that further research is needed.

REFERENCES 1. EIlenbogen PH, Scheible FW, Talner LB, Leopold GR. Sensitivity of gray scale ultrasound in detecting urinary tract obstruction. AJR 1978;130:731-733. 2. Kamholtz RG, Cronan J J, Dorfman GS. Obstruction and the minimally dilated renal collecting system: US evaluation. Radiology 1989; 170:51-53. 3. Amis ES, Cronan JJ, Pfister RC, Yoder IC. Ultrasonic inaccuracies in diagnosing renal obstruction. Urology 1982;29:101-105. 4. Talner LB. Obstructive uropathy. In: Pollack HM, ed. Clinical urography. Philadelphia: Saunders, 1990:1535-1628. 5. Cronan JJ. Contemporary concepts for imaging urinary tract obstruction. Urol Radio11992;14:8-12. 6. Platt JF, Rubin JM, Ellis JH. Distinction between obstructive and nonobstructive pyelocaliectasis with duplex Doppler sonography. AJR 1989;153:997-1000. 7. Plait JF, Rubin JM, Ellis JH, DiPrieto MA. Duplex Doppler US of the kidney: differentiation of obstructive from nonobstructive dilatation. Radiology 1969; 171:515-517. 8. Platt JF, Rubin JM, Ellis JH. Acute renal obstruction: evaluation with intrarenal duplex Doppler and conventional US. Radiology 1993;186:685-688. 9. Gilbert R, Garra B, Gibbons MD. Renal duplex Doppler ultrasound: an adjunct in the evaluation of hydronephrosis in the child. J Uro11993;150: 1192-1194. 10. Ryan PC, Maher KP, Murphy B, Hurley GD, Fitzpatrick JM. Experimental partial ureteric obstruction: pathophysiologic changes in upper traet pressures and renal blood flow. J Uro11987;138:674-678. 11. Claesson G, Josephson S, Robertson B. Experimental partial ureteric obstruction in newborn rats: IV. Do the morphological effects progress continuously? J Urol 1983; 130:1217-1222. 12. Murphy GP, Scott VVW. The renal hemodynamic response to acute and chronic ureteral occlusions. J Uro11966;95:636-657. 13. Klahr S. Pathophysiology of obstructive nephropathy. Kidney Int 1983;23: 414-426. 14. Vaughan ED, Sorenson EJ, Gillenwater JY. The renal hemodynamic response to chronic unilateral complete ureteral occlusion. Invest Uro/1970;8:78--90. 15. Walker RD, Richard GA, Bueschen A J, Retik AB. Pathophysiology and recoverability of function and structure in obstructed kidneys. Urol Clin North Am 1980;7:291-310. 16. Dodd GD III, Kaufman PN, Bracken RB. Renal arterial duplex Doppler ultrasound in dogs with urinary obstruction. J Uro11991;145:644-646. 17. Ryan PC, Maher K, Hurley GD, Fitzpatrick JM. The Whitaker test: experimental analysis in a canine model of partial ureteric obstruction. J Urol 1988; 141:387-390. 18. Chen JH, Pu YS, Liu SP, Chiu TY. Renal hemodynamics in patients with obstructive uropathy evaluated by duplex Doppler sonography. J Urol 1993;150:18-21. 19. Rodgers PM, Bates JA, Irving HC. Intrarenal Doppler ultrasound studies in normal and acutely obstructed kidneys. Br J Radio11992;65:207-212. 20. Arellano RS, Lira GM, Coley BD, Talner LB, Mattrey RF. Effect of diuresis on resistive index in normal rabbits and kidneys with partial or complete ureteral obstruction. Radiology 1992;185(P):287. 21. Argyropoulou M, Courtel J, Malgouyras A, BruneUe F. Color Doppler flow evaluation of renal artery blood flow and resistivity index as a function of ureteral pressure in experimental hydronephrosis in rabbits. Radiology 1992;185(P):287