Because of these limitations of arteriography, investigators are interested in noninvasive screening for renal artery disease [3, 4]. Many of the noninvasive screening tests (such as intravenous urography, renal scintigraphy, determination of plasma renin activity, and the captopril test) have an unacceptably low sensitivity and specificity. Renal scintigraphy with the administration of captopril is safe and noninvasive and has a reasonable sensitivity and specificity in many circumstances. Captopril renography is not as sensitive in patients with bilateral renal artery stenosis or in patients with stenosis to a solitary functioning kidney [4, 5]. Renal scintigraphy with captopril may also not be as sensitive in patients with substantial azotemia, the subgroup of patients most in need of a safe screening test for renovascular disease. The captopril-stimulated renal flow scan does not provide an anatomical assessment of the degree of stenosis, but it is an excellent test for assessing the functional significance of a stenotic lesion.

Magnetic resonance angiography is an attractive noninvasive screening method for assessing the renal arteries [6]. Most segments of the renal artery can be adequately visualized, and it is now possible to view the renal arteries three-dimensionally through post-processing reconstruction [7]. Currently, this technique has two major drawbacks. It is expensive and may overestimate the degree of stenosis. One study of three-dimensional, spiral computed tomographic angiography showed excellent imaging of the mesenteric and renal arteries [8], but the procedure required a large amount of contrast material administered in a bolus, thus limiting the utility of this test in patients with azotemia.

Duplex ultrasound scanning of the renal arteries is a noninvasive screening test for the detection of renal artery stenosis. It combines direct visualization of the renal arteries (B-mode imaging) with measurement of various hemodynamic factors in the main renal artery and within the kidney (Doppler), thus providing both an anatomical and a functional assessment [9, 10]. Duplex scanning also allows the measurement of kidney size at the same time that the examination is being done. Unlike other noninvasive screening tests, duplex scanning is not affected by medications that the patient may be taking; the level of renal function; or whether the disease is unilateral or bilateral or affects a solitary functioning kidney. Improvements in image resolution, Doppler technology, and processor software have led to the ability to better scan the visceral circulation [11].

We determined the utility of duplex ultrasound of the renal arteries for identifying patients with renal artery stenosis of 60% or more and for excluding patients with either normal renal arteries or renal artery stenosis of less than 60%.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

We prospectively entered patients into our study who were being evaluated for possible renal artery stenosis. One hundred and two consecutive patients who had both duplex ultrasound scanning of the renal arteries and renal arteriography were included for analysis. Duplex scanning was done before renal arteriography in 60 (58.8%) patients. Only main renal arteries were analyzed in our study. Only 9 patients had accessory renal arteries on arteriography, which was a smaller number than we expected [5]. The cohort of patients that we studied had a high pretest likelihood of disease. Most patients who were studied had either hypertension that was difficult to control or unexplained azotemia (or both), giving them a pretest likelihood of renovascular disease of 70% [1]. A smaller number of patients was evaluated because of associated peripheral vascular disease or abdominal aortic aneurysm. The pretest likelihood in this group is approximately 40% [1].

The persons who did the duplex ultrasound examinations did not have access to the patient's chart or to the results of renal arteriography. All renal arteriograms had at least two views of the renal artery, and the degree of stenosis was estimated by visual examination. Renal arteriograms were both formally interpreted by the person doing the study and reviewed by one of the authors. If a difference of opinion existed between the formal reading and the review, a third investigator evaluated the arteriogram without knowledge of the two previous evaluations and the majority ruled. Those interpreting the arteriograms were blinded to the results of the person doing the study and to the results of the duplex ultrasound.

Renal Duplex Scanning Technique

All patients were scanned with an Ultramark 9 HDI ultrasound machine (Advanced Technology Laboratories, Bothell, Washington) using a C3 40R transducer. This curved array transducer has a 3.0-MHz imaging probe and a 2.5-MHz pulsed Doppler with a focal zone in the vicinity of the depth of the renal arteries.

Patients were studied after a 12-hour fast. Ninety percent of the patients had an adequate examination when studied in the fasting state. If excess bowel gas was present, the patients were studied the following morning or at the next available convenient time for the patient. Patients were studied in the anterior, lateral decubitus, and, at times, the prone position so that all portions of the renal artery from the takeoff of the aorta to where the renal artery enters the kidney were visualized. The hilar approaches were particularly useful in patients with excess bowel gas and in obese persons.

The renal arteries were first visualized and identified. The ability to use color duplex imaging may allow the ultrasonographer to directly visualize turbulent blood flow if it is present. Once the renal arteries were correctly identified, the Doppler signature was taken at as close to a 60-degree angle as possible. Under no circumstances was the angle more than 60 degrees. Doppler samples were taken at the origin of the renal artery, the proximal renal artery, the midportion of the renal artery, and the distal renal artery.

The renal-aortic ratio was calculated by dividing the peak systolic velocity in the renal artery by the peak systolic velocity in the aorta. The peak systolic velocity is a measure of the maximum velocity of blood flow during systole. Using Doppler principles, as an artery narrows, the velocity of blood flow increases. If the renal-aortic ratio was 3.5 or more, it indicated 60% to 99% stenosis of the renal artery. End-diastolic velocities were also measured in all arterial segments. The end-diastolic velocity is a measure of the velocity of blood flow at the end of diastole. In patients with severe degrees of arterial narrowing, the velocity of diastolic flow may be increased. After studies were done using an anterior approach, the patient was placed in the lateral decubitus or prone position (or both), and renal artery velocities were again sampled. The kidney size was measured at its longest axis. Parenchymal flow samples from the kidney itself were obtained, and the resistive index was calculated by using the formula of the peak systolic velocity minus the end-diastolic velocity divided by the peak systolic velocity. The average time for completion of a renal duplex scan was approximately 1 hour. The most common reasons for technical failure were excess bowel gas and obesity. Identification of accessory renal arteries was sometimes difficult.

The criteria used for classifying patients were as follows: 1) A renal-aortic ratio of less than 3.5 and a peak systolic velocity of less than 200 cm/s identified patients with 0% to 59% renal artery stenosis; 2) a renal-aortic ratio of 3.5 or more or a peak systolic velocity of more than 200 cm/s [or both] identified patients with 60% to 99% renal artery stenosis; and 3) occlusion of the renal artery was diagnosed by the absence of a flow signal in the renal artery and by a low-amplitude parenchymal signal.

The renal artery duplex scans were done by one of four technicians and were reviewed by one of the attending staff in the vascular laboratory. Technician A did 39 (38%) studies, technician B did 41 (40%) studies, technician C did 12 (12%) studies, and technician D did 10 (10%) studies. The technicians often sought the help of their colleagues for patients who were difficult to study. No difference was noted in the accuracy of the results or in the quality of the studies among the four technicians who participated in our study.

Statistical Analysis

The sensitivity of ultrasound relative to arteriography was calculated as the proportion of positive arteriograms that were positive by ultrasound. The specificity was calculated as the proportion of negative arteriograms that were negative by ultrasound. The positive predictive value was defined as the proportion of positive ultrasounds that were positive by arteriography, and the negative predictive value was defined as the portion of negative ultrasounds that were negative by arteriography. Comparison of hemodynamic variables between the various levels of stenosis was done using analysis of variance techniques. Values are expressed as mean ±SD. When the overall difference was significant (P < 0.05), this significance level was decreased to adjust for multiple comparisons [12].

Comparisons of nominal variables were done using chi-square tests unless any of the expected cell frequencies were less than five, in which case the Fisher exact test was used. All statistical tests were two-sided. For the primary analyses, we assumed that evaluations of arteriograms and duplex ultrasounds of multiple arteries within a single patient were made independently of each other. To confirm the validity of our findings, we also analyzed the data using only one artery per patient. All results reported were consistent in both analyses.

Results

Top Methods Results Discussion Author & Article Info References

Our study included 44 men and 58 women with a mean age (±SD) of 63.3 ±13.4 years. We studied 187 arteries in 102 patients. Ninety-eight (96%) patients had atherosclerosis, and 4 (4%) patients had medial fibroplasia of the renal artery. Duplex ultrasound scanning correlated extremely well with arteriography (Table 1). Sixty-two of 63 arteries with less than 60% stenosis on renal arteriography showed less than 60% stenosis on duplex scanning. Thirty-one of 32 arteries with 60% to 79% stenosis by arteriography showed 60% to 99% stenosis on duplex scanning. Sixty-seven of 69 arteries with 80% to 99% stenosis by arteriography showed 60% to 99% stenosis by duplex ultrasound. Twenty-two of 23 arteries with total occlusion of the renal artery on arteriography showed total occlusion on duplex ultrasound. Using the cut points of 60% or more stenosis and less than 60% stenosis, we found that duplex ultrasound scanning had an overall sensitivity of 0.98 (122 of 124 arteries), a specificity of 0.98 (62 of 63), a positive predictive value of 0.99 (122 of 123), and a negative predictive value of 0.97 (62 of 64).

Ultrasound results differed from the arteriographic results for five arteries. In two of the studies in which ultrasound wrongly showed total occlusion, the technician noted that the study had been technically difficult. This generally meant that bowel gas had interfered with obtaining an adequate image and with Doppler interrogation. One patient was diagnosed as having 60% to 99% stenosis on ultrasound but actually had 0% to 59% stenosis on arteriography. The reviewers of the arteriogram were uncertain whether the arteriogram should have been classified as less than or more than 60% but the patient was placed in the 0% to 59% category by a majority opinion. Of two arteries classified as having 0% to 59% stenosis on ultrasound, one had 80% to 99% stenosis and one was occluded on arteriography.

The distribution of the renal-aortic ratios grouped by the percentage of stenosis is shown in Figure 1. The mean renal-aortic ratio (±SD) was 2.1 ±0.8 in the group with 0% to 59% stenosis, 5.1 ±0.6 in the group with 60% to 79% stenosis, and 6.1 ±1.9 in the group with 80% to 99% stenosis. The mean renal-aortic ratio differed significantly in the 0% to 59% group compared with the 60% to 79% and the 80% to 99% groups (P = 0.0001). A significant difference was also noted between the mean renal-aortic ratios in the 60% to 79% group compared with the 80% to 99% group (P = 0.016). This difference remained significant after adjustments for multiple comparisons were made (P = 0.017). However, so much overlap was noted between these two groups that the renal-aortic ratio was not clinically useful in separating these two categories. All four patients in the 60% to 99% category who had renal-aortic ratios of less than 3.5 had peak systolic velocities of more than 250 cm/s.

View larger version (13K): [in this window] [in a new window]   Figure 1. Renal-aortic ratios of patients with stenosis of the renal artery determined by arteriography. A renal-aortic ratio of 3.5 or more (solid line) separates patients whose renal arteries have less than 60% stenosis from those with 60% or more stenosis. The renal-aortic ratio differed among the three groups: 0% to 59% compared with 60% to 79% (P = 0.0001); 0% to 59% compared with 80% to 99% (P = 0.0001); and 60% to 79% compared with 80% to 99% (P = 0.016).

Figure 2 shows the distribution of the peak systolic velocity according to the degree of stenosis by arteriography. For the 0% to 59% group, the mean peak systolic velocity (±SD) was 146 ±49 cm/s; for the 60% to 79% group, it was 363 ±101 cm/s; and for the 80% to 99% group, it was 380 ±103 cm/s. The mean peak systolic velocity was lower in the 0% to 59% group compared with the 60% to 79% group or with the 80% to 99% group (P = 0.0001). No statistically significant difference was noted in the mean peak systolic velocities between the 69% to 79% and the 80% to 99% groups. Only two patients in the 80% to 99% group had a peak systolic velocity of less than 200 cm/s, and all patients in the 60% to 79% group had peak systolic velocities of more than 200 cm/s.

View larger version (11K): [in this window] [in a new window]   Figure 2. Peak systolic velocities of patients with stenosis of the renal artery determined by arteriography. A peak systolic velocity of 200 cm/s or more (solid line) separates patients whose renal arteries have less than 60% stenosis from those with 60% or more stenosis. The peak systolic velocity differed among the groups: 0% to 59% compared with 60% to 79% (P = 0.0001); 0% to 59% compared with 80% to 99% (P = 0.0001). No significant difference was noted in the peak systolic velocity between the 60% to 79% and the 80% to 99% groups.

Figure 3 shows the distribution of the end-diastolic velocity according to the percentage of stenosis by arteriography. The mean end-diastolic velocity (±SD) for the 0% to 59% group was 37 ±20 cm/s; for the 60% to 79% group, it was 132 ±78 cm/s; and for the 80% to 99% group, it was 143 ±80 cm/s. A difference was noted in the mean end-diastolic velocity between the 0% to 59% and the 60% to 79% groups or between the 0% to 59% and the 80% to 99% groups (P = 0.0001). No statistically significant difference was noted in the mean end-diastolic velocity between the 60% to 79% and the 80% to 99% groups.

View larger version (11K): [in this window] [in a new window]   Figure 3. End-diastolic velocities of patients with stenosis of the renal artery determined by arteriography. The end-diastolic velocity differed significantly among the groups: 0% to 59% compared with 60% to 79% (P = 0.0001); 0% to 59% compared with 80% to 99% (P = 0.0001). No significant difference was noted in the end-diastolic velocity between the 60% to 79% and the 80% to 99% groups.

When the end-diastolic velocity was 150 cm/s or more, 81% of the arteries (38 of 47) had 80% to 99% stenosis and 19% of the arteries (9 of 47) had 60% to 79% stenosis by arteriography (P < 0.0001). Twenty-seven of 34 arteries (79%) with 80% to 99% stenosis on arteriography and with an end-diastolic velocity of 150 cm/s or more had a resistive index of less than 0.70, whereas 25 of 27 arteries (93%) with an end-diastolic velocity of less than 150 cm/s and with 80% to 99% stenosis had a resistive index of 0.70 or more (P = 0.0001) (Table 2).

Discussion

Top Methods Results Discussion Author & Article Info References

In 1986, Kohler and colleagues [13] retrospectively evaluated renal artery duplex scans in 43 arteries and compared the results with those obtained using arteriography. Using the renal-aortic ratio to separate patients with renal artery stenosis of less than 60% from patients with stenosis of 60% or more, they showed a sensitivity of 91% (20 of 22 diseased arteries were correctly identified) and a specificity of 95% (20 of 21 normal or insignificantly diseased arteries were correctly identified). In this early study, one of four occluded arteries was incorrectly interpreted as being patent because of misidentification of a collateral. Several years later, these investigators prospectively evaluated 58 renal arteries in 29 patients in whom arteriograms and duplex scans were available [14]. Renal duplex scanning accurately diagnosed 38 of 39 patients with 0% to 59% stenosis, 11 of 14 patients with 60% to 99% stenosis, and 4 of 5 patients with occluded renal arteries, yielding a sensitivity of 84%, a specificity of 97%, and a positive predictive value of 94%.

In one of the largest series reported to date, Hansen and colleagues [15] analyzed data from 74 consecutive patients (148 kidneys) and compared 77 renal duplex scans with standard angiographic studies. Duplex scanning correctly identified the presence of renovascular disease in 41 of 44 patients with angiographically proven lesions, and renovascular disease was not identified in any patient free of disease. In patients with single renal arteries (no accessory arteries), duplex scanning in this series provided a sensitivity of 93%, a specificity of 98%, a positive predictive value of 98%, a negative predictive value of 98%, and an overall accuracy of 96%. However, when accessory renal arteries were present, the sensitivity decreased to 67%, whereas the specificity was 100%. These and other early studies [11, 16, 17] showed that duplex scanning is a reasonable noninvasive screening test for the detection of renal artery stenosis.

In our series, we studied 187 renal arteries from 102 patients; ultrasound results differed from arteriographic results in only 5 arteries. The sensitivity of 98% and the specificity of 98% are similar to findings in some of the other smaller series. However, our study differs from the other published series because we showed that if the end-diastolic velocity was 150 cm/s or more, patients had a high likelihood of having renal artery stenosis of 80% or more. In addition, we hypothesized that the end-diastolic velocity may not be increased in some patients with 80% or more stenosis because the resistance within the kidney (resistive index) is high.

Captopril renography, magnetic resonance angiography, spiral computed tomographic scanning, and duplex scanning are four of the better noninvasive screening tests for detecting renal artery stenosis. Renal artery duplex scanning has several advantages over these other tests. Unlike captopril renography, it shows the anatomy of the renal artery, provides hemodynamic information, and indicates kidney size [10]. The sensitivity of duplex scanning is not affected in the presence of bilateral renal artery disease or azotemia, and medications do not need to be discontinued before the test is done. Several large prospective trials [14, 15] showed that duplex scanning of the renal arteries is a sensitive and specific method for detecting renal artery stenosis. This is not the case for magnetic resonance angiography or spiral computed tomographic scanning. Magnetic resonance angiography using three-dimensional phase-contrast sequences produces visually pleasing images, but signal loss in areas of turbulence may cause overestimation of the degree of stenosis. Cost, availability, artifacts by metal clips, and the necessity for some breath holding may limit the usefulness of magnetic resonance angiography in some patients. A bolus of contrast material is required during spiral computed tomographic scanning; thus, this type of scanning has limited usefulness as a screening test in patients with azotemia [8]. Duplex scanning does not have such limitations.

In our series, the results were consistent when we analyzed the data using only one renal artery per patient. Depending on whether the first or second artery was used in the analysis, the sensitivity was 97% to 100%, the specificity was 97% to 100%, the positive predictive value was 98% to 100%, and the negative predictive value was 94% to 100%. Similar results were found in the analyses of peak systolic velocity, end-diastolic velocity, and renal-aortic ratio.

Renal artery duplex scanning is time consuming and technically demanding and has a steep learning curve. When all segments of the renal artery could not be adequately visualized, the technician sought the help of a colleague. For two of the examinations in which the ultrasound did not agree with the arteriogram, the technician noted that the study was technically difficult. Thus, for technically difficult studies or if all segments of the renal artery cannot be visualized because of excess bowel gas, the patient should have a complete examination at another time before the ultrasound results are interpreted.

One patient had 60% to 99% stenosis on duplex scanning and 0% to 59% stenosis by renal arteriography, but the degree of stenosis by arteriography was close to 60% by visual examination. This example shows some of the pitfalls of using standard arteriography as the "gold standard" in assessing the exact degree of stenosis. Perhaps computer-assisted quantitative angiography [18] or intra-arterial ultrasound determination of luminal stenosis would be more accurate in estimating the degree of stenosis.

Although renal artery duplex scanning is effective in identifying renal arteries with less than 60% stenosis, renal arteries with 60% to 99% stenosis, and occluded renal arteries, it has not been useful in further subdividing the category of 60% to 99% stenosis. Duplex scanning of the carotid artery can accurately determine those patients with stenosis of 60% to 79% (peak systolic velocity of 150 to 240 cm/s) or 80% to 99% (peak systolic velocity more than 240 cm/s and end-diastolic velocity more than 135 cm/s) [19]. Therefore, we tried to determine if similar variables could be defined in the renal arteries. Because of the overlap in the renal-aortic ratios and the peak systolic velocities between the 60% to 79% and the 80% to 99% groups (Figures 1 and 2), these tests were useless in separating these two categories.

Because of limitations in the pulse repetition frequency that occur when examining deeper abdominal structures such as the renal arteries, it may not be possible to obtain an accurate peak systolic velocity. In some patients, therefore, the recorded peak systolic velocity is actually less than the true peak systolic velocity because of limitations of the technology. This is not a problem with carotid artery duplex scanning because the carotid arteries are much more superficial and thus the capabilities of pulse repetition frequency are not exceeded. Mean end-diastolic velocities did not differ in the 60% to 79% group compared with the 80% to 99% group. However, when the end-diastolic velocity was 150 cm/s or more, 81% of the arteries had 80% to 99% stenosis. Therefore, an increased end-diastolic velocity is suggestive, but not diagnostic, of high-grade stenosis.

Some patients with high-grade stenosis do not have an increased end-diastolic velocity. This may be related to the resistance within the kidney. The resistive index is defined as the peak systolic velocity minus the end-diastolic velocity divided by the peak systolic velocity. It is an indication of the amount of renal arterial impedance [20]. The end-diastolic velocity may not have been increased in some patients with 80% to 99% stenosis because the increased resistance (resistive index) within the renal circulation may have prevented an increase in the end-diastolic velocity.

The resistive index may also be an indicator of the degree of renal impairment in patients with medical renal disease [21]. Whether this resistance is fixed or reversible is unknown. Veglio and colleagues [22] showed that the baseline resistive index correlated well with the duration of hypertension and that, in patients with moderate-to-severe hypertension, the resistive index did not change after the administration of an angiotensin-converting-enzyme inhibitor. No good information exists about the effects that other drugs (vasodilators, calcium channel blocking agents, nitrates) have on the renal resistive index. It is also not known whether the resistive index changes from one portion of the kidney to another. Additional prospective studies need to be done to determine if subdividing the category of 60% to 99% stenosis into categories of 60% to 79% and 80% to 99% is possible with the technology that is currently available.

Duplex ultrasound scanning can predict the presence or absence of renal artery stenosis with a high degree of accuracy. The procedure is technically difficult and is associated with a steep learning curve. By insonating the renal artery from the anterior, lateral decubitus, and posterior approaches, all segments of the renal artery can be adequately visualized. The lateral and posterior approaches are particularly useful for studying obese patients with excess bowel gas.