(1/20/05 3:53 am)
Renovascular Hypertension: Can We Identify a Population at H|
Renovascular Hypertension: Can We Identify a Population at High Risk?
from Southern Medical Journal
Mitchell H. Rosner, Division of Nephrology, Department of Internal Medicine, University of Virginia Health Sciences Center, Charlottesville
Abstract and Introduction
Renovascular hypertension is a common cause of secondary hypertension. However, diagnostic tests are limited by lack of sensitivity and specificity, cost, or invasiveness. Selecting patients with hypertension for evaluation of renal artery stenosis can be challenging. This review focuses on the sensitivity and specificity of commonly used screening tests for renal artery stenosis and on the clinical variables that are most likely to distinguish patients with renal artery stenosis from patients with other causes of their hypertension. This approach allows for the rational screening of patients at high and moderate risk for renal artery disease.
Renovascular hypertension (RVHT) is one of the most common causes of secondary hypertension, with reports of its frequency varying from less than 1% to as high as 30% in highly selected referral populations. Its importance lies in the fact that it is a potentially curable disease that -- if left undiagnosed and untreated -- may lead to permanent end-stage renal disease, heart disease, retinopathy, and cerebrovascular disease.
Renovascular hypertension refers to hypertension that is caused by renal hypoperfusion due to a flow-limiting vascular stenosis and is corrected by either surgical or percutaneous revascularization. It is important to distinguish RVHT from the much more prevalent condition of renovascular disease (defined as significant stenosis of a renal artery), since one condition frequently exists without the other. Eyler et al found that in normotensive patients over the age of 60, the prevalence of renovascular disease was 45%. Furthermore, the presence of both renovascular disease and hypertension does not necessarily implicate causality. This is shown by work from Smith in 1956 who recognized that nephrectomy was able to relieve hypertension in only 35% of patients with documented coexisting renovascular disease and hypertension. However, these data may be partially explained by the fact that the longer the duration of RVHT, the less likely revascularization will be able to effect a cure of hypertension. Among 110 patients who had surgery for RVHT, hypertension improved in 78% of patients with hypertension for less than 5 years, but in only 25% with a longer duration of hypertension.
Renovascular hypertension results when a vascular stenosis leads to renal ischemia and elevation of renin release from the affected kidney.[5,6] Renin subsequently leads to angiotensin II formation, which serves as a powerful vasoconstrictor as well as a stimulator of aldosterone secretion with resulting salt and water retention. The central role of the renin-angiotensin-aldosterone axis in the development of hypertension is clearly shown in the profound response of both renal function and blood pressure to the introduction of angiotensin-converting enzyme (ACE) inhibitors in patients with RVHT. Numerous causes can lead to renovascular disease; the two most common are atherosclerosis and fibromuscular dysplasia. Fibromuscular dysplasia leads to renovascular disease and hypertension through medial fibroplasia of the renal artery. It is seen most commonly in young females and often involves multiple arteries originating from the aorta. This review will focus solely on RVHT due to atherosclerosis.
Since 1937, when the first successful report of treating hypertension with nephrectomy was published, clinicians have struggled to detect those patients with hypertension who have underlying renal artery stenosis (RAS), and in whom correction of RAS will improve blood pressure. Unfortunately, there are four major obstacles to designing a strategy for the accurate and efficient diagnosis of RVHT:
First, clinical features suggestive of RVHT, such as an abdominal bruit, are often infrequent and difficult to detect. Furthermore, the reported prevalence of RVHT in patients with these clinical features varies greatly. This lack of specific clinical findings dictates that diagnostic testing must play an important role.
Second, several tests are available for diagnostic evaluation of patients with suspected RVHT. However, the sensitivity, specificity, costs, and potential morbidities of these tests vary widely. In most cases, standardized, prospective comparisons among the available tests are not available.
Third, currently, only patients with clinical features suggestive of RVHT have diagnostic testing. However, the prevalence of RVHT in patients without these features is unknown. A sensitive, specific, and cost-effective screening test is not available.
Finally, as previously mentioned, although RAS is a necessary precondition for RVHT, it is frequently found coincidentally, and our ability to predict the hemodynamic consequence of a particular renal artery lesion is unclear. Functional tests that can predict the response of a patient's blood pressure to revascularization are not readily available.
Clinicians are concerned with the ability of a test result to discriminate between persons with and without a particular disease. This ability is determined by a test's sensitivity and specificity. Sensitivity refers to the probability that a test will be positive when it is applied to a person with a disease. Specificity refers to the probability that a test will be negative when applied to a person free of disease. Thus, a perfectly sensitive test, if negative, can rule out the disease, while a perfectly specific test, if positive, can rule in the disease. Although sensitivity and specificity are vital in selecting a test, they are not ordinarily what concerns clinicians in making a decision. Rather, it is the probability of a patient having a disease if the test returns positive or negative. The answer to this question guides clinical practice. These values are termed positive and negative predictive values. Jekel et al[1 thoroughly reviewed these concepts in 1996. Importantly, predictive values, unlike sensitivity and specificity, are dependent on the prevalence of disease in the population that is undergoing diagnostic testing. For a test with a sensitivity and specificity of 90%, if the prevalence of disease in the tested population is 5%, then the positive predictive value would be approximately 20%. However, if the prevalence of disease increases to 50%, then the positive predictive value is 80% to 90%. Thus, if one can identify a population with a high prevalence of disease before diagnostic testing, the predictive value of the test result becomes more powerful.
This review will focus on the diagnostic tests that are available for the diagnosis of RAS and will review clinical characteristics that can be used to identify a population of high prevalence. The goal is to identify a population with a high prevalence of RAS before further diagnostic testing, thus increasing the predictive value of any chosen diagnostic test. Before embarking on a potentially costly workup for renovascular disease, two primary questions should be considered: (1) Is the patient a suitable candidate for either surgical or percutaneous revascularization? and (2) What is the goal of treatment? Renal artery revascularization can be done for two main reasons: treatment of hypertension and improvement in renal function in ischemic nephropathy. Obviously, if a patient has well-controlled hypertension and stable renal function, it makes little sense to screen for renovascular disease. However, in the patient with difficult to control blood pressure and worsening renal function, diagnostic workup becomes more reasonable.
Diagnostic Tests for Renal Artery Stenosis
Two classes of diagnostic tests are used to investigate the presence of RAS and indirectly RVHT. These classes are tests that assess renin release and those that rely on imaging modalities. Reported sensitivities and specificities may vary according to the expertise of the individual institution.
Assessing Renin Release
As previously discussed, since hypersecretion of renin due to renal hypoperfusion is a primary event in the pathogenesis of RVHT, it is not surprising that patients with RVHT have been found to have elevated plasma renin activity (PRA). However, PRA is a complex variable affected by medications, volume status, and numerous other variables. Furthermore, the PRA may over time decrease in patients with RVHT. Thus, renin levels are generally higher in patients with fibromuscular dysplasia and less useful in elderly patients. For PRA measurements, the reported sensitivity is 57% and specificity is 66%.
To improve the sensitivity and specificity of PRA measurements, investigators have measured PRA after captopril administration. To perform this test, all patients must discontinue diuretics and ACE inhibitors for 2 weeks before testing. The PRA level is tested before and then 30 minutes after a captopril dose. A positive test requires three features: a stimulated PRA >12 ng/mL per hour, an absolute increase in PRA of at least 10 ng/mL per hour, and an increase in PRA levels of >/=150%(or >/=400% if the baseline PRA level was <3 ng/mL per hour). This test showed a 96% sensitivity but only a 55% specificity. Thus, when applied to a population with a high prevalence, a negative result has good predictive value, but a positive result does not aid clinical decision-making.
The standard for PRA measurement comes from renal vein sampling. The comparison of renin levels between renal veins (obtained by percutaneous catheterization) uses a ratio of renin levels greater than 1.5 to 1 as abnormal and lateralizing. In a large series, an abnormal ratio was 92% predictive of curability with revascularization; however, 65% of patients with non-lateralizing renin ratios also had curable disease. In an effort to improve the sensitivity and specificity of the test, the renal/systemic renin index has been used. This allows determination of the functional significance of bilateral lesions. The index is obtained by subtracting the systemic (infrarenal vena cava) PRA from the PRA levels in the renal veins and dividing by the systemic PRA. An index above 0.24 indicates excessive renin production from that kidney, while lower levels are indicative of renin suppression. However, given the invasive nature of these tests and the low specificity, renal vein renin sampling is usually reserved for diagnostic dilemmas.
Ultrasonography. Duplex ultrasonography combines direct visualization of the renal arteries (B-mode) with hemodynamic measurements in the renal arteries (Doppler). Furthermore, ultrasonography allows direct measurement of renal size. The procedure identifies the abdominal aorta at the level of the renal arteries and records blood velocity at this site, followed by identification and measurements of blood velocity in the renal arteries. Other measurements include those of the renal parenchymal velocities from the upper, middle, and lower poles of the kidneys, as well as renal size. The important parameter is the ratio of velocity in the renal artery to that of the aorta. If the ratio is >3.5, this is likely to be associated with a stenosis of >60%. If the renal artery velocity is >180 cm/second, this is also considered abnormal. The test is limited by the experience of the operator and by patient habitus (being more difficult in obese patients) and bowel gas. Thus, Doppler ultrasonography of the kidneys is best performed in the early morning after fasting. Reported sensitivity and specificity range from 90% to 95% and 60% to 90%, respectively. In one study, if renal arteriograms were performed on all patients with a positive Doppler ultrasound, a 2.7% false-positive rate was found. A further use of Doppler ultrasonography may be in predicting which patients would benefit from revascularization. The renal resistive index (RRI) is obtained from Doppler ultrasonography and represents [1 - (end-diastolic velocity/maximal systolic velocity) x l0 .
In a recent study, the RRIwas predictive in determining response of blood pressure to revascularization. An RRI >80 identified patients with RAS in whom angioplasty or surgery did not improve blood pressure or renal function. Finally, the presence of asymmetric kidney size may be a clue to underlying RAS and renal ischemia.
Captopril-Augmented Nuclear Renography. Renography is performed with radioactive agents that are excreted either by glomerular filtration (99Tc-DPTA) or mainly by tubular secretion combined with glomerular filtration (99Tc-MAG3). The latter agent gives a good representation of renal blood flow. It is best to withhold ACE inhibitors and diuretics for at least 48 hours before the test. During the test, blood pressure must be closely monitored, since systolic blood pressures <100 mm Hg or mean arterial pressures <70 mm Hg can invalidate the test.[2 Both scintigraphic images and time-activity curves are measured after the injection of the tracer. When used alone, isotopic renograms have a sensitivity and specificity of approximately 75%. However, given the observation that renal function in an ischemic kidney could be abruptly worsened after a single dose of an ACE inhibitor, captopril was subsequently used to augment the sensitivity of the scan.[2 Normally, there is rapid uptake, followed by excretion of the tracer. However, in the presence of RAS and captopril, there is delayed uptake and excretion. Studies have documented a sensitivity of approximately 83% and a specificity of 93% for detecting RAS >70%. Perhaps more importantly, the renogram had 93% sensitivity and 100% specificity for predicting blood pressure response to revascularization if the kidney function was normal. The sensitivity and specificity of renograms decline considerably in the face of renal impairment. Thus, patients with elevated creatinine levels (>2.0 mg/dL) are not ideal candidates for this test.
Magnetic Resonance Angiography. This relatively new technique shows great promise for the diagnosis of proximal (and thus largely atherosclerotic) RAS. Gadolinium, which has a low degree of nephrotoxicity and can thus be utilized in patients with renal insufficiency, is used as a contrast agent. Reconstructions of images are used to obtain detailed views of the renal arteries. Various small trials have reported sensitivities ranging from 83% to 100% and specificities of 92% to 97%.[25,26] Limitations include the high cost, limited availability, and substantial expertise needed to analyze images. Results for computed tomography with angiography are similar, but have the down-side of requiring contrast material with the attendant risk of nephropathy.
Renal Arteriography. This standard diagnostic test has the advantage of allowing for therapeutic intervention at the time of diagnosis. When the famous bank robber Willie Sutton was asked why he robbed banks, he replied, "I rob banks because that's where the money is." Many believe, following the axiom of Willie Sutton, "If you are looking for a lesion, look at the vessel." This axiom is particularly true for patients with highly suggestive clinical features in whom negative results on noninvasive tests would not be sufficient to exclude the possibility of RAS. Three main advantages are (1) arteriography provides an immediate answer; (2) angioplasty can be performed immediately, if needed; and (3) given new digital-subtraction techniques, small volumes of ionic-contrast or gadolinium-carbon dioxide contrast can be used to minimize the risk of contrast nephropathy. However, angiography has significant potential risks, including contrast-induced renal failure, atheroembolism, local puncture site bleeding, puncture site pseudoaneurysm formation, and anaphylaxis. Furthermore, angiography is of relatively little value in predicting blood pressure response to an intervention. Measurement of renal artery pressure gradients may hold some promise in this regard.[3
Table 1 lists the published sensitivities and specificities of these tests. All the diagnostic tests have limitations, ranging from low sensitivity and specificity to high cost and invasiveness. None of these tests could be used to screen an unselected population of hypertensive patients. For instance, assuming a prevalence of RAS in hypertensive patients of 5%, a test with a sensitivity and specificity of 90%, if positive, would give a positive predictive value of only 20% to 30%. This illustrates the importance of defining clinically a population with a high prevalence of RAS. If we could use clinical factors to identify a population of hypertensive patients who had a prevalence of RAS of 50%, then the positive predictive value of this same test would approach 90%.
Clinical Features Predictive of Renal Artery Stenosis
Numerous retrospective and prospective studies have attempted to identify clinical factors that could be used to predict whether patients had a high likelihood of atherosclerotic RAS. The first and largest prospective study that examined clinical and radiographic features of patients with RVHT was the Cooperative Study of Renovascular Hypertension. This study enrolled 2,442 patients with hypertension and subjected them to a battery of tests, including detailed histories and physical examinations, serum chemistries, and diagnostic and therapeutic tests, including renal arteriograms and revascularization, if warranted. From this population, 339 patients with essential hypertension were compared with 175 patients with RVHT, as defined by the presence of RAS and improvement in blood pressure with surgical revascularization. In this study, certain clinical characteristics stood out as highly prevalent in the population with RVHT. These included high-grade retinopathy, abdominal or flank bruits, the presence of peripheral vascular disease, absence of family history of hypertension, and recent onset of hypertension, especially after age 50. Laboratory findings that were more common in the population with RVHT included hypokalemia, metabolic alkalosis, and an elevated blood urea nitrogen level (>20 mg/dL).
The frequency of RVHT in patients with high-grade (grade III or IV) retinopathy has been further investigated. Of 123 patients identified with high-grade retinopathy, 93 had renal angiography, and renovascular disease was detected in 31% of these patients. Thus, severe retinopathy seems to identify a patient population at higher risk.
Numerous studies have examined the prevalence of abdominal and flank bruits in patients with hypertension. Proper ausculation requires that the patient be supine and in a quiet room. Since the kidneys are located retroperitoneally, with the renal arteries leaving the aorta just cephalad to the umbilicus, auscultation should begin in the epigastrium and should include all four quadrants. Auscultation should continue over the spine and flanks in the areas between the twelfth thoracic and second lumbar vertebrae. Finally, the bruit needs to be timed to the cardiac cycle and characterized as a systolic, systolic-diastolic, or venous hum. The differential diagnosis of an abdominal bruit other than RAS includes splenic arteriovenous fistula, hepatic cirrhosis, hepatoma, abdominal aortic aneurysm, celiac artery compression syndrome, intestinal ischemia, and pancreatic carcinoma. Abdominal bruits have a prevalence of 6.5% to 31% in the healthy population, and a prevalence of 28% in patients with all-cause hypertension. However, in patients with angiographically proven RAS, the prevalence ranges from 78% to 87%. Two studies have investigated the sensitivity and specificity of finding a systolic-diastolic abdominal bruit in the diagnosis of RAS. Sensitivity ranged from 39% to 63%, with specificity of 90% to 99%. Thus, the presence of a systolic-diastolic bruit is highly suggestive of RAS and should be screened for, while the absence of a bruit does not exclude RAS.
As expected, RAS is prevalent in patients with underlying peripheral vascular disease. In an analysis of 300 autopsies, in those cases with aortic atherosclerosis, the prevalence of renovascular disease was 40%. Furthermore, 33% of patients having abdominal aortography for either peripheral vascular disease or aortic aneurysm were found to have clinically unsuspected RAS. Finally, 1,235 patients having cardiac catheterization had concurrent angiography, and 187 patients (15.2%) had significant RAS, with 93 patients having lesions >76%. Only half of this population had hypertension, and none had clinically suspected RVHT. Factors in this study that were predictive of RAS included older age, peripheral vascular disease, coronary artery disease, congestive heart failure, serum creatinine > 1.4 mg/dL, female sex, and smoking.
Few prospective studies use clinical criteria alone to select patients for renal arteriography. Svetky et al, in 1990, studied 100 hypertensive patients with renal arteriography who met any one of six entry criteria. These included abdominal or flank bruit, refractory hypertension (blood pressure >160/95 mm Hg while receiving three or more medications), accelerated hypertension (increase in blood pressure of >15% in 6 months), severe hypertension (diastolic blood pressure >115 mm Hg or high-grade retinopathy), recent-onset (within the last 2 years) hypertension, and onset of hypertension before age 25 or after age 45. In this study, the only prospective feature that was predictive of RAS that met statistical significance was the presence of an abdominal bruit. There was a trend toward refractory hypertension being statistically significant.
In 1998, Dutch investigators analyzed the clinical characteristics of 477 patients with either drug-resistant hypertension (mean diastolic blood pressure >95 mm Hg on three visits while treated with either amlodipine/atenolol or enalapril/hydrochlorothiazide, or the requirement of three antihypertensives) or an increase in the serum creatinine level to >0.23 mg/dL while receiving an ACE inhibitor.[4 The patients were subjected to renal angiography, and the clinical characteristics of the population were analyzed to determine a clinical predictive score for the probability of renovascular disease. Renal artery stenosis was found in 22% of these patients, 84% of whom had atherosclerotic disease and 25% of whom had bilateral RAS. Univariant and multivariant analyses were done on clinical variables and their association with RAS and a scoring index was formed. Not surprisingly, the most informative variables were the presence of an abdominal bruit, an elevated serum creatinine concentration, obesity, smoking, and advanced age. This scoring index was used to determine a predicted probability of RAS. A 70-year-old female smoker with an abdominal bruit and a body mass index >25 kg/m2 who had drug-resistant hypertension would have a predicted probability of RAS of approximately 50%. Unfortunately, this scoring system has not yet been validated in a large number of patients. However, its potential usefulness can be seen in Table 2. By increasing the pretest probability, analysis of clinical variables allows a positive captopril renogram to be more informative. Furthermore, if validation of a scoring system can be determined, then clinicians may be able to reliably determine a level of pretest probability in which it is both safe and cost-effective to begin with invasive diagnostic testing.
Renovascular hypertension is a common cause of secondary hypertension. Its importance lies in the fact that it is potentially correctable, but the diagnosis requires invasive and expensive tests. Thus, it is imperative that patients at highest risk for RVHT be selected from the hypertensive population. Doing so renders noninvasive diagnostic testing more informative. Kaplan published an algorithm for the diagnosis of RVHT that allows for clinical suspicion of RVHT to guide diagnostic testing. In his algorithm, patients at highest risk should have angiography, while those with moderate risk should have captopril renography or duplex ultrasonography (depending on the expertise of the institution). How should we best define patients who are at high or moderate risk for RVHT? Clinical studies indicate the best predictors are abdominal or flank bruit, high-grade retinopathy, peripheral vascular disease elsewhere, hypertension refractory to treatment with three antihypertensive agents, and elevated serum creatinine level. Patients with any of these features should be screened for RVHT using a noninvasive test. The best test often depends on the expertise of the institution, but the best options are duplex ultrasonography, captopril-renogram, or magnetic resonance angiography. Patients with more than three of these clinical features are at high risk and represent a group that may benefit from immediate renal angiography (Table 3). No diagnostic test should be offered without a thorough clinical assessment and determination of risk. The lack of sensitivity and specificity of these diagnostic tests yields little predictive information in unselected hypertensive patients. Unfortunately, we still are unable to accurately predict which patients with hypertension will benefit from revascularization, though measurement of renal artery pressure gradients and renal resistive indices may prove useful in the future.
Reprint requests to Mitchell H. Rosner, MD, University of Virginia Health Sciences Center, Division of Nephrology, Box 800133, Charlottesville, VA 22908.