ORIGINAL ARTICLES From the Southern Association for Vascular Surgery
Renal duplex sonography: Evaluation of clinical utility K i m b e r l e y J. Hansen, M D , Reid W. Tribble, M D , Scott W. Reavis, Vincent J. Canzanello, M D , T i m o t h y E. Craven, M S P H , George W. Plonk, Jr., M D , and R i c h a r d H . Dean, M D , Winston-Salem, N.C. ~¥ith the exception of conventional angiography, no previously proposed screening test has the necessary sensitivity/specificity to guide further evaluation for correctable renovascular disease. Recently, renal duplex sonography has been suggested as a usefial substitute in such screening for renovascular disease. This report analyzes our data collected over the past 10 months in evaluation of renal duplex sonography to examine its diagnostic x~~]ue. The study population for renal duplex sonography validity analysis consisted of 74 ~:onsecutive patients who had 77 comparative renal duplex sonography and standard angiographic studies of the arterial anatomy to 148 kidneys. Renal duplex sonography results from sLx kidneys (4%) were considered inadequate for interpretation. This study population contained 26 patients (35%) with severe renal insufficiency (mean 3.6 rag/dl) and 67 hypertension (91%). Fourteen patients (19%) had 20 kidneys with multiple 'renal arteries. Bilateral disease was present in 22 of the 44 patients with significant renovascular disease. Renal duplex sonography correctly identified the presence of renovascular disease in 41 of 44 patients with angiographically proven lesions, and renovas~atlar disease was not identified in any patient free of disease. When single renal arteries were present (122 kidneys), renal duplex sonography provided 93% sensitivity, 98% specificity, 98% positive predictive value, 94% negative predictive value, and an overall accuracy of 96%. These results were adversely affected when kidneys with multiple (polar) renal arteries were examined. Although the end diastolic ratio was inversely correlated with serum creatinine (r = - 0.3073, p = 0.009), low end diastolic ratio in 35 patients submitted to renovascular reconstruction did not preclude beneficial blood pressure or ~renal function response. We conclude from this analysis that renal duplex sonography can be a valuable screening test in the search for correctable renovascular disease causing global renal ischemia and secondary renal insufficiency (ischemic nephropathy). Renal c" ~plex sonography does not, however, exclude polar vessel renovas~axlar disease causing hypertension alone nor does it predict hypertension or renal function response after correction of renovascular disease. (J VASe SURG 1990;12:227-36.)
Whether present as an isolated lesion, existing with hypertension alone, or in combination with ren d insufficiency (ischemic nephropathy), the diagnosis of critical stenosis of the main renal artery • (RVD) has required angiography? A number of alFrom the Division of Surgical Sciences,Departmentof Medicine and Department of Public Health Sciences, Bowman Gray School of Medicine,Winston-Salem. Presented at the Fourteenth Annual Meeting of the Southern Association for Vascular Surgery,Acapulco,Mexico, Jan. 2427, 1990. Reprint requests: KimberleyJ. Hansen, MD, AssistantProfessor of Surgery, BowmanGray Schoolof Medicineof Wake Forest University, 300 South Hawthorne Rd., Winston-Salem,NC 27103 ~/6/22791
ternative screening tests have been advocated for the detection of RVD; however, these screening tests lack the necessary sensitivity to guide further management of correctable RVD. 1,2 As a result of the expense and perceived risk of angiography combined with the lack of an accurate diagnostic alternative, the prevalence of RVD remains poorly defined? ,~ Although we have proceeded with angiography in any patient who would be a candidate for renovascular reconstruction if correctable RVD were found, an inexpensive, noninvasive method that accurately detected RVD could have important clinical utility. Recently, renal duplex sonography (RDS) combining a B-scan image with fast-Fourier transform (FFT) analysis of the aortic and renal artery Doppler227
228
Journal of VASCULAR SURGERY
Hansen et al.
shifted signals has been reported to demonstrate improved sensitivity/specificity for detection of RVD compared to previously proposed screening methods. s7 These reports suggest that agreement of RDS with angiography is such that a negative RDS examination finding eliminates the need for further study. Furthermore, experience with RDS as a diagnostic tool to diagnose renal transplant rejection has demonstrated a correlation between features of the transplant renal artery Doppler-shifted diastolic spectrum and vascular rejection. 81° If these diastolic features of the Doppler spectra reflect parenchyrnal renovascular disease, RDS might also prove useful in determining the relative contributions of main renal artery stenosis versus intrinsic RVD in patients with hypertension and patients with associated renal insufficiency. This report reviews our experience with RDS compared with conventional angiography during a recent 10-month period to detect the presence or absence of RVD and to predict the clinical response to renovascular reconstruction. PATIENT MATERIAL During the 10-month period, from November 20, 1988, through August 10, 1989, 213 patients and 20 healthy volunteers were studied with RDS at the Bowman Gray School of Medicine of Wake Forest University. Seventy-four patients had 77 comparative angiograms to 142 kidneys within 3 weeks of their ultrasound examination. This study population having both RDS and angiography forms the basis for prospective RDS validity analysis. Thirty-three men and 41 women, ranging in age from 28 to 90 years (mean age, 61 years) were in the study population. Significant hypertension was present in 67 patients and ranged from 182/90 to 280/130 m m Hg (mean, 198/102 m m Hg.). Hypertension had been present 1 to 28 years in these patients (mean duration, 9.2 _ 6 years), and antihypertension treatment included 1 to 7 drugs (mean 2.9 -+ 1:1 drugs). In addition to hypertension, all but five patients demonstrated one or more of the following additional atherosclerotic risk factors: tobacco use (69 patients), diabetes mellims (16 patients), and hypercholesterolemia (22 patients). Based on serum creatinine values obtained before angiography, 26 patients (35%) were considered to have renal insufficiency (serum creatinine ---2.0 mg/dl, mean, 3.6 mg/dl). Eleven patients had moderate renal insufficiency (serum creatinine 2.0 to 2.9 mg/dl), 12 patients had severe renal insufficiency, yet were dialysis independent (serum creatinine ->3.0 mg/dl), and three patients were dialysis dependent at the time of their study.
TECHNIQUE OF STUDIES Renal duplex sonography Renal duplex sonography studies were performed with either an Ultramark-8 or an Ultramark-9 Ultrasound System (Advanced Technology Laboratories, Bothell, Wash.) with either a 3.0 MHz. mechanical long focus probe or a 2.25 MHz. phased array probe with Doppler color-flow capability. After informed consent was obtained, patients fasted overnight and received 10 mg bisacodyl by mouth to minimize bowel gas interference. With the subject supine, the ultrasound probe was positioned in the abdominal midline 2 to 3 cm inferior to the xiphoid process. First, a sagittal B-scan image was obtained of the upper abdominal aorta at the level of the superior mesenteric artery. A center stream aortic peak systolic velocity (Ao-PSV) derived f r o ~ the real-time FFT analysis of the Doppler-shifted signal and angle ofinsonation was obtained at this level. The probe was then rotated 90 degrees to obtain a B-scan image of aorta and proximal superior mesenteric artery in cross section. By use of the left renal vein as a visual reference, each renal artery origin was identified. Doppler velocity waveforms were obtained from multiple sites in both renal arteries (approximately 10 sites per artery) from aortic origin to renal hilum. During main renal artery Doppler interrogation, the angle of insonation was maintained at 60 degrees or less for calculation of peak systolic velocity (RA-PSV). Intrarenal Doppler-shifted signals were also obtained from regions of the interlobar and arcuate arteries. This process was then repeated from a flank approach with the patient in the left and right decubitus positions: images and Doppler signals were obtained from each renal artery and kidneF From this flank approach, length, width, and thickness measurements of each kidney were recorded from the B-scan image. When measured Doppler spectral values differed between values obtained by supine or flank approaches, values obtained with the least angle of insonation were recorded. B-scan images and spectra from the FFT spectrum analyzer were recorded on videotape and hard copy processor for independent review.
Angiography Conventional cut-film angiography (74 studies in 70 patients) or intraarterial digital subtraction angiography (3 studies) were obtained in each patient. The patients underwent angiographic study to determine the suspected presence of KVD (39 patients), aortoiliac disease (33 patients), or visceral artery disease (2 patients). Seventeen patients had special oblique projections when anterior-posterioe'
Volume 12 Number 3 September 1990
views were insufficient to delineate renal artery anatomy. DATA ANALYSIS Renal duplex sonography From the recorded FFT analysis of the Doppler-shifted signals and estimated angle of insonation, calculations ofAo-PSV, RA-PSV, and renal artery enddiastolic velocity (RA-EDV) were determined from five consectutive frequency spectra. From these values, the ratio of the RA-PSV to Ao-PSV (the renal aortic ratio [RAR]) and the ratio of the RA-EDV to RA-PSV (the end-diastolic ratio [EDR]) were determined. Without knowledge of the angiogram results, these measured and calculated values were used to determine the presence or absence of critical renal artery occlusive disease based on criteria of Taylot et al. 7 (Table I). In brief, these criteria are (1) ~negative study (3.5.
Table III. ANOVA for RA-EDR versus hypertension response after renal revascularization (N = 35 patients) Responsea~ Cured Improved N o change
No.
Mean RA-EDR (+_ SD)
p value
8 22 5
0.25 + 0,04 0.29 + 0.02 0.22 _+ 0.04
0.225
with multiple renal arteries contributed three falsenegative studies. Only 21 of 43 arteries to these 20 kidneys were identified, providing 67% sensitivity, 100% specificity, 79% negative predictive value, and 85% overall accuracy. Failure to recognize the presence of multiple polar arteries was not reduced by use of color Doppler interrogation. To assess the impact of multiple polar arteries on the clinical utility of RDS, a patient was considered to have significant RVD if one or all renal arteries demonstrated a ->60% renal artery stenosis or occlusion. Eighteen of 60 patients with bilateral single renal arteries had significant RVD to one kidney, and 17 patients had Significant disease to both. Renal duplex sonography c0rrectly identified all 35 patients with single renalarteries and significant RVD. Of
the 14 patients with multiple renal arteries to one (eight patients) or to both kidneys (six patients), RDS identified six of nine patients with significant disease (Fig. 1). Renal duplex sonography did not identify a patient as having RVD in the absence of significant disease. Overall, RDS was 93% sensitive and 100% specific for detection of patients with si~ nificant RVD. As described by other investigators, Ao-PSV varied widely among patients. The RAR criteria for critical renal artery stenosis presumes that RA-PSV increases relative to Ao-PSV, however in this patient group, RA-PSV had no apparent relationship to AoPSV (Fig. 2). We therefore have retrospectively examined RA-PSV alone as a predictor of critical renal artery stenosis. Analysis of scatter plots demonstrated that renal artery PSV of ->2.0 m/see was equivalent to an RAR of ->3.5 m/see as a predictor of ->60% diameter-reducing renal artery stenosis (Figs. 3 and 4). Although two kidneys demonstrated PSV >2.0 and RAR >3.5 m/see, each were considered free of significant RVD since neither demonstrated a focal velocity increase or turbulent distal velocity waveforms. Thirty-five of the 74 patients underwent surgical
"Volume 12 Number 3 September 1990
Evaluation of clinical utility 233
7"
r= 0.7118 p < 0.001
*
B"
*
> 5" o3 O_
h
C 3"
2-
,
iO
I
I
I
I
20
30
40
50
BO
~
,
I
I
I
I
70
80
90
iO0
r, S t e n o s i s in Renal Artery
Fig. 4. Scatter plot and Pearson correlation coefficient for renal artery peak systolic velocity (PSV) versus percent renal artery stenosis. ---60% diameter-reducing stenosis indicated by renal artery PSV ->2.0 m/see. renal revascularization (33 patients) or percutaneous balloon angioplasty (2 patients) for treatment of presumed renovascular hypertension. Thirteen of the 33 °urgical procedures included bilateral renal revascularization (including one contralateral nephrectomy). Applying a previously described algorithm for blood pressure and renal function response to renal revascularization, hypertension was considered cured in 8 patients, improved in 22 patients, and unchanged in 5 patients (at a mean follow-up of 37 weeks). 11Serum creatinine was ---1.5 mg/dl in 22 of these 35 patients (mean, 3.4 mg/dl). Renal function was considered improved in 9 patients, unchanged in 10 patients, and worsened in 3 patients. As reported by others, RA-EDR was negatively correlated with prerepair serum creatinine in the presence and absence of RVD (Fig. 5; r = - 0.3073,p = 0.009).6,s This negative correlation between RA-EDR and serum creatinine has been suggested to reflect parenchymal renovascular disease. Although RA-EDR was significandy decreased compared with healthy volunteers (p 2 0 0~ change in serum creatinine.
repair RA-EDR and blood pressure (p = 0.225) or renal function response (p = 0.278) in paticnts submitted to renovascular repair (Tables III and IV). DISCUSSION
This report describes our experience with RDS to identify significant RVD during a recent 10month period. Renovascular disease defined by RDS criteria compared prospectively with angiography formed the basis of our validity analysis. Angiograplay was obtained in these patients for evaluation of
234
Journal of VASCULAR SURGERY
H a n s e n et al.
0.55
0.50 r = -0.3073 p = 0.009
~0.45 c-
k~ 0 . 4 0
**~
c-
*
:#*~
,
*
F7 0 . 3 5 "0 cO
, 0.30
*~*
~,
.+.a
,
*
,
*
_J
~_ 0 . 2 5 0
0~0.20
~-,~,,0 . 1 5 n-" rm laJ
0.10 0.05
O.
I
I
1
I
I
I
I
I
I
I
I
I
I
I
I
0.5
~.0
~.5
2.0
2.5
3,0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
Baseline Creatinine
Fig. 5. Scatter plot and Pearson correlation coefficient for average end diastolic ratio (EDR) versus baseline serum creatinine. renovascular (52%) or aortoiliac (45%)disease. Prevalance of angiographic disease was high in this patient group--58% of patients and 44% of kidneys demonstrated significant RVD, whereas 71% of patients demonstrated occlusive (51%) or aneurysmal (20%) aortoiliac disease. Renal duplex sonography accurately defined the presence or absence of RVD in patients (100% accuracy) and in kidneys (96% accuracy) with single renal arteries. These results were not influenced by associated aortoiliac disease or renal insufficiency. However, RDS did not identify multiple (polar) renal arteries, failing to identify 22 of 23 polar arteries (in 17 with disease) in three of nine patients with significant RVD, accounting for all three patients incorrectly interpreted as free of disease. Despite the reported association between RA-EDR and primary parenchymal disease, low RAEDR did not preclude favorable clinical response to renal revascularizafion.6,s,12 Our results support those obtained by other groups using similar RDS criteria to define RVD. Derived from experimental and retrospective clinical experience, Taylor et al.7 prospectively applied these criteria in a patient group with 37% prevalence of RVD in 58 kidneys and demonstrated 84% sensitiv-
ity and 97% specificity compared with angiography. Only one of three kidneys with multiple renal arteries was identified. Although comparable with these results, our sensitivity in 142 kidneys (88%) may be enhanced by the high prevalence of severe renal artery stenosis (53 of 63 diseased renal arteries >80% stenosis) and the high prevalence of atherosclerotic ostial disease (87%) in our patient group. As reported by other investigators, RA-EDV relative to RA-PSV demonstrated a significant inverse correlation with serum creatinine.6, s Using a canine model of sequential microsphere embolization of the kidney, Norris et al.~' 8 demonstrated that RA-EDR identified increasing grades of renal parenchymal vascular resistance. When they applied them clinically, these investigators demonstrated significant differences in RA-EDR among healthy, hypertensive, and atherosclerotic patients. Other investigators have reported that decreased RA-EDR identifies increased parenchymal vascular resistance associated with vascular rejection in renal transplants. 8,1°,1a In patients with hypertension because of RVD, intrinsic renal parenchymal disease might be expected to decrease the oppommity for a favorable blood pressure or renal function response to renal revascularization. 1
Volume 12 Number 3 September 1990
Evaluation of clinical utility
235
Table V. Comparative analysis parameter estimates and their 95% confidence intervals Group All kindeys
Measure
Estimate
142 (kidneys)
Sensitivity Specificity PPV NPV Accuracy
0.88 0.99 0,98 0,92 0.91
(0.84, (0.97, (0.96, (0.89, (0.87,
0.92) 1.00) 0.99) 0.95) 0.95)
Sensitivity Specificity PPV NPV Accuracy
0.93 0.98 0.98 0.94 0.91
(0.90, (0.96, (0.96, (0.91, (0.87,
0.96) 1.00) 1.00) 0.97) 0.95)
148 (kidneys) Kidneys with single renal artery
95% Confidence interval
No.
122 (kidneys)
148 (kidneys) Kidneys with multiple renal arteries
21 (arteries)
Sensitivity Specificity PPV NPV Accuracy
0.67 1.00 ~ 1.00 ~ 0.79 0.86
(0.53, 0.81) --(0.68, 0.90) (0.76, 0.96)
All patients
74 (subjects)
Sensitivity Specificity PPV NPV Accuracy
0.93 1.00 ~ 1.00 ~ 0.91 0.96
(0.87, 0.99) --(0.84, 0.98) (0.91, 1.00)
~Estimated SEM, is zero, confidence level is inestimable. PPV, Positive predictive value; NPV, negative predictive value.
Despite this, RA-EDR did not correlate with improvement in hypertension or change in. serum creatinine in the 35 patients submitted to correction of RVD in our study. Although RDS has clinical utility as a screening test for anatomic RVD, we feel that standard tests of physiologic significance (renal vein renin assays, split renal function tests) remain most useful to guide further management once correctable RVD is identified. Our results and the results of others suggest that the accuracy of RDS exceeds other available screening tests for detection of RVD, however a few points deserve special comment. These results can be anticipated only when an experienced and skilled sonologist is capable of complete renal artery interrogation from aorta to renal hilum. Increased RA-PSV associated with a critical stenosis may decrease to within normal values within two vessel diameter's distance distal to the lesion) 4 Incomplete Doppler sampling of the main renal artery may contribute to unacceptably high false-negative rates. Despite extensive experience with deep abdominal Doppler interrogation, most multiple (polar) renal arteries will escape detection by RDS and the prevalence of these vessels and their rate of disease will adversely affect the results of the examination. Given these limitations, we do not think that
RDS can replace angiography as a method of detection for RVD in all patients. In contrast to those who propose that a negative RDS outcome negates the need for further angiographic study, we would proceed with additional screening tests for secondary hypertension, including conventional angiography in young patients (
Renal duplex sonography: Evaluation of clinical utility K i m b e r l e y J. Hansen, M D , Reid W. Tribble, M D , Scott W. Reavis, Vincent J. Canzanello, M D , T i m o t h y E. Craven, M S P H , George W. Plonk, Jr., M D , and R i c h a r d H . Dean, M D , Winston-Salem, N.C. ~¥ith the exception of conventional angiography, no previously proposed screening test has the necessary sensitivity/specificity to guide further evaluation for correctable renovascular disease. Recently, renal duplex sonography has been suggested as a usefial substitute in such screening for renovascular disease. This report analyzes our data collected over the past 10 months in evaluation of renal duplex sonography to examine its diagnostic x~~]ue. The study population for renal duplex sonography validity analysis consisted of 74 ~:onsecutive patients who had 77 comparative renal duplex sonography and standard angiographic studies of the arterial anatomy to 148 kidneys. Renal duplex sonography results from sLx kidneys (4%) were considered inadequate for interpretation. This study population contained 26 patients (35%) with severe renal insufficiency (mean 3.6 rag/dl) and 67 hypertension (91%). Fourteen patients (19%) had 20 kidneys with multiple 'renal arteries. Bilateral disease was present in 22 of the 44 patients with significant renovascular disease. Renal duplex sonography correctly identified the presence of renovascular disease in 41 of 44 patients with angiographically proven lesions, and renovas~atlar disease was not identified in any patient free of disease. When single renal arteries were present (122 kidneys), renal duplex sonography provided 93% sensitivity, 98% specificity, 98% positive predictive value, 94% negative predictive value, and an overall accuracy of 96%. These results were adversely affected when kidneys with multiple (polar) renal arteries were examined. Although the end diastolic ratio was inversely correlated with serum creatinine (r = - 0.3073, p = 0.009), low end diastolic ratio in 35 patients submitted to renovascular reconstruction did not preclude beneficial blood pressure or ~renal function response. We conclude from this analysis that renal duplex sonography can be a valuable screening test in the search for correctable renovascular disease causing global renal ischemia and secondary renal insufficiency (ischemic nephropathy). Renal c" ~plex sonography does not, however, exclude polar vessel renovas~axlar disease causing hypertension alone nor does it predict hypertension or renal function response after correction of renovascular disease. (J VASe SURG 1990;12:227-36.)
Whether present as an isolated lesion, existing with hypertension alone, or in combination with ren d insufficiency (ischemic nephropathy), the diagnosis of critical stenosis of the main renal artery • (RVD) has required angiography? A number of alFrom the Division of Surgical Sciences,Departmentof Medicine and Department of Public Health Sciences, Bowman Gray School of Medicine,Winston-Salem. Presented at the Fourteenth Annual Meeting of the Southern Association for Vascular Surgery,Acapulco,Mexico, Jan. 2427, 1990. Reprint requests: KimberleyJ. Hansen, MD, AssistantProfessor of Surgery, BowmanGray Schoolof Medicineof Wake Forest University, 300 South Hawthorne Rd., Winston-Salem,NC 27103 ~/6/22791
ternative screening tests have been advocated for the detection of RVD; however, these screening tests lack the necessary sensitivity to guide further management of correctable RVD. 1,2 As a result of the expense and perceived risk of angiography combined with the lack of an accurate diagnostic alternative, the prevalence of RVD remains poorly defined? ,~ Although we have proceeded with angiography in any patient who would be a candidate for renovascular reconstruction if correctable RVD were found, an inexpensive, noninvasive method that accurately detected RVD could have important clinical utility. Recently, renal duplex sonography (RDS) combining a B-scan image with fast-Fourier transform (FFT) analysis of the aortic and renal artery Doppler227
228
Journal of VASCULAR SURGERY
Hansen et al.
shifted signals has been reported to demonstrate improved sensitivity/specificity for detection of RVD compared to previously proposed screening methods. s7 These reports suggest that agreement of RDS with angiography is such that a negative RDS examination finding eliminates the need for further study. Furthermore, experience with RDS as a diagnostic tool to diagnose renal transplant rejection has demonstrated a correlation between features of the transplant renal artery Doppler-shifted diastolic spectrum and vascular rejection. 81° If these diastolic features of the Doppler spectra reflect parenchyrnal renovascular disease, RDS might also prove useful in determining the relative contributions of main renal artery stenosis versus intrinsic RVD in patients with hypertension and patients with associated renal insufficiency. This report reviews our experience with RDS compared with conventional angiography during a recent 10-month period to detect the presence or absence of RVD and to predict the clinical response to renovascular reconstruction. PATIENT MATERIAL During the 10-month period, from November 20, 1988, through August 10, 1989, 213 patients and 20 healthy volunteers were studied with RDS at the Bowman Gray School of Medicine of Wake Forest University. Seventy-four patients had 77 comparative angiograms to 142 kidneys within 3 weeks of their ultrasound examination. This study population having both RDS and angiography forms the basis for prospective RDS validity analysis. Thirty-three men and 41 women, ranging in age from 28 to 90 years (mean age, 61 years) were in the study population. Significant hypertension was present in 67 patients and ranged from 182/90 to 280/130 m m Hg (mean, 198/102 m m Hg.). Hypertension had been present 1 to 28 years in these patients (mean duration, 9.2 _ 6 years), and antihypertension treatment included 1 to 7 drugs (mean 2.9 -+ 1:1 drugs). In addition to hypertension, all but five patients demonstrated one or more of the following additional atherosclerotic risk factors: tobacco use (69 patients), diabetes mellims (16 patients), and hypercholesterolemia (22 patients). Based on serum creatinine values obtained before angiography, 26 patients (35%) were considered to have renal insufficiency (serum creatinine ---2.0 mg/dl, mean, 3.6 mg/dl). Eleven patients had moderate renal insufficiency (serum creatinine 2.0 to 2.9 mg/dl), 12 patients had severe renal insufficiency, yet were dialysis independent (serum creatinine ->3.0 mg/dl), and three patients were dialysis dependent at the time of their study.
TECHNIQUE OF STUDIES Renal duplex sonography Renal duplex sonography studies were performed with either an Ultramark-8 or an Ultramark-9 Ultrasound System (Advanced Technology Laboratories, Bothell, Wash.) with either a 3.0 MHz. mechanical long focus probe or a 2.25 MHz. phased array probe with Doppler color-flow capability. After informed consent was obtained, patients fasted overnight and received 10 mg bisacodyl by mouth to minimize bowel gas interference. With the subject supine, the ultrasound probe was positioned in the abdominal midline 2 to 3 cm inferior to the xiphoid process. First, a sagittal B-scan image was obtained of the upper abdominal aorta at the level of the superior mesenteric artery. A center stream aortic peak systolic velocity (Ao-PSV) derived f r o ~ the real-time FFT analysis of the Doppler-shifted signal and angle ofinsonation was obtained at this level. The probe was then rotated 90 degrees to obtain a B-scan image of aorta and proximal superior mesenteric artery in cross section. By use of the left renal vein as a visual reference, each renal artery origin was identified. Doppler velocity waveforms were obtained from multiple sites in both renal arteries (approximately 10 sites per artery) from aortic origin to renal hilum. During main renal artery Doppler interrogation, the angle of insonation was maintained at 60 degrees or less for calculation of peak systolic velocity (RA-PSV). Intrarenal Doppler-shifted signals were also obtained from regions of the interlobar and arcuate arteries. This process was then repeated from a flank approach with the patient in the left and right decubitus positions: images and Doppler signals were obtained from each renal artery and kidneF From this flank approach, length, width, and thickness measurements of each kidney were recorded from the B-scan image. When measured Doppler spectral values differed between values obtained by supine or flank approaches, values obtained with the least angle of insonation were recorded. B-scan images and spectra from the FFT spectrum analyzer were recorded on videotape and hard copy processor for independent review.
Angiography Conventional cut-film angiography (74 studies in 70 patients) or intraarterial digital subtraction angiography (3 studies) were obtained in each patient. The patients underwent angiographic study to determine the suspected presence of KVD (39 patients), aortoiliac disease (33 patients), or visceral artery disease (2 patients). Seventeen patients had special oblique projections when anterior-posterioe'
Volume 12 Number 3 September 1990
views were insufficient to delineate renal artery anatomy. DATA ANALYSIS Renal duplex sonography From the recorded FFT analysis of the Doppler-shifted signals and estimated angle of insonation, calculations ofAo-PSV, RA-PSV, and renal artery enddiastolic velocity (RA-EDV) were determined from five consectutive frequency spectra. From these values, the ratio of the RA-PSV to Ao-PSV (the renal aortic ratio [RAR]) and the ratio of the RA-EDV to RA-PSV (the end-diastolic ratio [EDR]) were determined. Without knowledge of the angiogram results, these measured and calculated values were used to determine the presence or absence of critical renal artery occlusive disease based on criteria of Taylot et al. 7 (Table I). In brief, these criteria are (1) ~negative study (3.5.
Table III. ANOVA for RA-EDR versus hypertension response after renal revascularization (N = 35 patients) Responsea~ Cured Improved N o change
No.
Mean RA-EDR (+_ SD)
p value
8 22 5
0.25 + 0,04 0.29 + 0.02 0.22 _+ 0.04
0.225
with multiple renal arteries contributed three falsenegative studies. Only 21 of 43 arteries to these 20 kidneys were identified, providing 67% sensitivity, 100% specificity, 79% negative predictive value, and 85% overall accuracy. Failure to recognize the presence of multiple polar arteries was not reduced by use of color Doppler interrogation. To assess the impact of multiple polar arteries on the clinical utility of RDS, a patient was considered to have significant RVD if one or all renal arteries demonstrated a ->60% renal artery stenosis or occlusion. Eighteen of 60 patients with bilateral single renal arteries had significant RVD to one kidney, and 17 patients had Significant disease to both. Renal duplex sonography c0rrectly identified all 35 patients with single renalarteries and significant RVD. Of
the 14 patients with multiple renal arteries to one (eight patients) or to both kidneys (six patients), RDS identified six of nine patients with significant disease (Fig. 1). Renal duplex sonography did not identify a patient as having RVD in the absence of significant disease. Overall, RDS was 93% sensitive and 100% specific for detection of patients with si~ nificant RVD. As described by other investigators, Ao-PSV varied widely among patients. The RAR criteria for critical renal artery stenosis presumes that RA-PSV increases relative to Ao-PSV, however in this patient group, RA-PSV had no apparent relationship to AoPSV (Fig. 2). We therefore have retrospectively examined RA-PSV alone as a predictor of critical renal artery stenosis. Analysis of scatter plots demonstrated that renal artery PSV of ->2.0 m/see was equivalent to an RAR of ->3.5 m/see as a predictor of ->60% diameter-reducing renal artery stenosis (Figs. 3 and 4). Although two kidneys demonstrated PSV >2.0 and RAR >3.5 m/see, each were considered free of significant RVD since neither demonstrated a focal velocity increase or turbulent distal velocity waveforms. Thirty-five of the 74 patients underwent surgical
"Volume 12 Number 3 September 1990
Evaluation of clinical utility 233
7"
r= 0.7118 p < 0.001
*
B"
*
> 5" o3 O_
h
C 3"
2-
,
iO
I
I
I
I
20
30
40
50
BO
~
,
I
I
I
I
70
80
90
iO0
r, S t e n o s i s in Renal Artery
Fig. 4. Scatter plot and Pearson correlation coefficient for renal artery peak systolic velocity (PSV) versus percent renal artery stenosis. ---60% diameter-reducing stenosis indicated by renal artery PSV ->2.0 m/see. renal revascularization (33 patients) or percutaneous balloon angioplasty (2 patients) for treatment of presumed renovascular hypertension. Thirteen of the 33 °urgical procedures included bilateral renal revascularization (including one contralateral nephrectomy). Applying a previously described algorithm for blood pressure and renal function response to renal revascularization, hypertension was considered cured in 8 patients, improved in 22 patients, and unchanged in 5 patients (at a mean follow-up of 37 weeks). 11Serum creatinine was ---1.5 mg/dl in 22 of these 35 patients (mean, 3.4 mg/dl). Renal function was considered improved in 9 patients, unchanged in 10 patients, and worsened in 3 patients. As reported by others, RA-EDR was negatively correlated with prerepair serum creatinine in the presence and absence of RVD (Fig. 5; r = - 0.3073,p = 0.009).6,s This negative correlation between RA-EDR and serum creatinine has been suggested to reflect parenchymal renovascular disease. Although RA-EDR was significandy decreased compared with healthy volunteers (p 2 0 0~ change in serum creatinine.
repair RA-EDR and blood pressure (p = 0.225) or renal function response (p = 0.278) in paticnts submitted to renovascular repair (Tables III and IV). DISCUSSION
This report describes our experience with RDS to identify significant RVD during a recent 10month period. Renovascular disease defined by RDS criteria compared prospectively with angiography formed the basis of our validity analysis. Angiograplay was obtained in these patients for evaluation of
234
Journal of VASCULAR SURGERY
H a n s e n et al.
0.55
0.50 r = -0.3073 p = 0.009
~0.45 c-
k~ 0 . 4 0
**~
c-
*
:#*~
,
*
F7 0 . 3 5 "0 cO
, 0.30
*~*
~,
.+.a
,
*
,
*
_J
~_ 0 . 2 5 0
0~0.20
~-,~,,0 . 1 5 n-" rm laJ
0.10 0.05
O.
I
I
1
I
I
I
I
I
I
I
I
I
I
I
I
0.5
~.0
~.5
2.0
2.5
3,0
3.5
4.0
4.5
5.0
5.5
6.0
6.5
7.0
7.5
Baseline Creatinine
Fig. 5. Scatter plot and Pearson correlation coefficient for average end diastolic ratio (EDR) versus baseline serum creatinine. renovascular (52%) or aortoiliac (45%)disease. Prevalance of angiographic disease was high in this patient group--58% of patients and 44% of kidneys demonstrated significant RVD, whereas 71% of patients demonstrated occlusive (51%) or aneurysmal (20%) aortoiliac disease. Renal duplex sonography accurately defined the presence or absence of RVD in patients (100% accuracy) and in kidneys (96% accuracy) with single renal arteries. These results were not influenced by associated aortoiliac disease or renal insufficiency. However, RDS did not identify multiple (polar) renal arteries, failing to identify 22 of 23 polar arteries (in 17 with disease) in three of nine patients with significant RVD, accounting for all three patients incorrectly interpreted as free of disease. Despite the reported association between RA-EDR and primary parenchymal disease, low RAEDR did not preclude favorable clinical response to renal revascularizafion.6,s,12 Our results support those obtained by other groups using similar RDS criteria to define RVD. Derived from experimental and retrospective clinical experience, Taylor et al.7 prospectively applied these criteria in a patient group with 37% prevalence of RVD in 58 kidneys and demonstrated 84% sensitiv-
ity and 97% specificity compared with angiography. Only one of three kidneys with multiple renal arteries was identified. Although comparable with these results, our sensitivity in 142 kidneys (88%) may be enhanced by the high prevalence of severe renal artery stenosis (53 of 63 diseased renal arteries >80% stenosis) and the high prevalence of atherosclerotic ostial disease (87%) in our patient group. As reported by other investigators, RA-EDV relative to RA-PSV demonstrated a significant inverse correlation with serum creatinine.6, s Using a canine model of sequential microsphere embolization of the kidney, Norris et al.~' 8 demonstrated that RA-EDR identified increasing grades of renal parenchymal vascular resistance. When they applied them clinically, these investigators demonstrated significant differences in RA-EDR among healthy, hypertensive, and atherosclerotic patients. Other investigators have reported that decreased RA-EDR identifies increased parenchymal vascular resistance associated with vascular rejection in renal transplants. 8,1°,1a In patients with hypertension because of RVD, intrinsic renal parenchymal disease might be expected to decrease the oppommity for a favorable blood pressure or renal function response to renal revascularization. 1
Volume 12 Number 3 September 1990
Evaluation of clinical utility
235
Table V. Comparative analysis parameter estimates and their 95% confidence intervals Group All kindeys
Measure
Estimate
142 (kidneys)
Sensitivity Specificity PPV NPV Accuracy
0.88 0.99 0,98 0,92 0.91
(0.84, (0.97, (0.96, (0.89, (0.87,
0.92) 1.00) 0.99) 0.95) 0.95)
Sensitivity Specificity PPV NPV Accuracy
0.93 0.98 0.98 0.94 0.91
(0.90, (0.96, (0.96, (0.91, (0.87,
0.96) 1.00) 1.00) 0.97) 0.95)
148 (kidneys) Kidneys with single renal artery
95% Confidence interval
No.
122 (kidneys)
148 (kidneys) Kidneys with multiple renal arteries
21 (arteries)
Sensitivity Specificity PPV NPV Accuracy
0.67 1.00 ~ 1.00 ~ 0.79 0.86
(0.53, 0.81) --(0.68, 0.90) (0.76, 0.96)
All patients
74 (subjects)
Sensitivity Specificity PPV NPV Accuracy
0.93 1.00 ~ 1.00 ~ 0.91 0.96
(0.87, 0.99) --(0.84, 0.98) (0.91, 1.00)
~Estimated SEM, is zero, confidence level is inestimable. PPV, Positive predictive value; NPV, negative predictive value.
Despite this, RA-EDR did not correlate with improvement in hypertension or change in. serum creatinine in the 35 patients submitted to correction of RVD in our study. Although RDS has clinical utility as a screening test for anatomic RVD, we feel that standard tests of physiologic significance (renal vein renin assays, split renal function tests) remain most useful to guide further management once correctable RVD is identified. Our results and the results of others suggest that the accuracy of RDS exceeds other available screening tests for detection of RVD, however a few points deserve special comment. These results can be anticipated only when an experienced and skilled sonologist is capable of complete renal artery interrogation from aorta to renal hilum. Increased RA-PSV associated with a critical stenosis may decrease to within normal values within two vessel diameter's distance distal to the lesion) 4 Incomplete Doppler sampling of the main renal artery may contribute to unacceptably high false-negative rates. Despite extensive experience with deep abdominal Doppler interrogation, most multiple (polar) renal arteries will escape detection by RDS and the prevalence of these vessels and their rate of disease will adversely affect the results of the examination. Given these limitations, we do not think that
RDS can replace angiography as a method of detection for RVD in all patients. In contrast to those who propose that a negative RDS outcome negates the need for further angiographic study, we would proceed with additional screening tests for secondary hypertension, including conventional angiography in young patients (
