Author + information
- Received June 27, 2003
- Revision received November 7, 2003
- Accepted November 13, 2003
- Published online May 5, 2004.
- Christopher E Buller, MD, FACC*,* (, )
- Jorge G Nogareda, MD*,
- Krishnan Ramanathan, MD*,
- Donald R Ricci, MD, FACC*,
- Ognjenka Djurdjev, MSc‡,
- Kathryn J Tinckam, MD†,
- Ian M Penn, MD, FACC*,
- Rebecca S Fox, MSc§,
- Lesley A Stevens, MD†,
- John A Duncan, MD† and
- Adeera Levin, MD†
- ↵*Reprint requests and correspondence:
Dr. Christopher E. Buller, The Sauder Family Heart and Stroke Foundation, Chair in Cardiology, Head–Division of Cardiology, University of British Columbia, St. Paul's Hospital, 1081 Burrard Street, Vancouver, B.C., Canada V6Z 1Y6.
Objectives We examined the prevalence and severity of renal artery stenosis (RAS) in patients undergoing cardiac catheterization who were deemed at risk for RAS based on clinical or laboratory criteria for study entry, but who had not previously been suspected of having RAS.
Background The diagnosis of atherosclerotic RAS remains problematic because its clinical manifestations are nonspecific.
Methods Consecutive patients undergoing non-emergent cardiac catheterization at a single institution during a 12-month period were evaluated using standardized clinical, laboratory, and angiographic criteria. Patients exhibiting at least one of four predefined selection criteria (severe hypertension, unexplained renal dysfunction, acute pulmonary edema with hypertension, or severe atherosclerosis) were prospectively registered and underwent coincident diagnostic renal angiography.
Results Renal angiography was performed in 851 patients and was diagnostic in 837. Angiographically evident renal atherosclerosis was present in 39% of the population, with RAS ≥50% in 120 (14.3%) and severe stenosis (≥70%) in 61 (7.3%). Severe stenosis was present in 48 (7%) patients with severe atherosclerosis, 38 (16%) with renal dysfunction, 25 (9%) with hypertension, and 2 (22%) with acute pulmonary edema with hypertension. The prevalence was higher in those exhibiting multiple selection criteria. In a multivariate model, severe RAS was associated with age, female gender, reduced creatinine clearance, increased systolic blood pressure, and peripheral or carotid artery disease.
Conclusions In a population at risk of, but not previously suspected of having RAS, severe RAS is associated with simple and readily determined clinical and laboratory patient characteristics. These data facilitate focused application of diagnostic renal angiography.
The clinical diagnosis of atherosclerotic renal artery stenosis (RAS) remains problematic. In contrast to myocardial ischemia, the pathophysiologic manifestations of RAS (hypertension, renal dysfunction, and acute left ventricular [LV] failure) are nonspecific and often attributed to other processes. This dilemma has hindered both the clinical detection of RAS in individuals at risk and the determination of RAS prevalence in populations at risk.
Contrast angiography is a standard criterion for RAS; it is readily performed in combination with coronary angiography. A correlation between coronary disease burden and the prevalence of RAS has already been established, but associations between patient demographic characteristics, coronary disease burden, extracoronary atherosclerosis, putative manifestations of RAS, and the prevalence of RAS have not been prospectively and rigorously examined.
To determine the overall prevalence of RAS in a cohort considered at risk, but who had not previously been suspected of having RAS, we performed renal angiography in 851 patients undergoing clinically indicated non-emergent coronary angiography over a 12-month period in a single institution. Subjects were required to meet at least one of the predefined selection criteria: 1) severe atherosclerosis; 2) severe or resistant hypertension; 3) unexplained renal dysfunction; or 4) acute pulmonary edema presumed due to diastolic LV dysfunction. We then examined in detail, as well as modeled, the associations between the presence of angiographically demonstrated RAS and baseline clinical, laboratory, and angiographic variables.
Screening and data collection
Between June 2001 and May 2002 (inclusive), we screened all patients undergoing non-emergent diagnostic cardiac catheterization at Vancouver Hospital and evaluated them for study inclusion according to predetermined criteria. Before catheterization, a protocol-based clinical examination was used to determine demographics, cardiac history, indications for cardiac catheterization, atherogenic risk factors, features of extracoronary vascular disease, and related comorbidities. We recorded current antihypertensive and cardiovascular drug therapy and normalized doses of agents affecting blood pressure (BP) using World Heath Organization (WHO)-standardized daily dose equivalents (1). We obtained noninvasive BP readings during usual therapy, before catheterization and after at least 5 min of rest, using a previously validated automated manometer (VSM MedTech Ltd., Vancouver, BC, Canada) employing multiple readings averaged over 5 min (2,3).
We categorized all patients according to the presence or absence of each of four selection criteria determined a priori (Table 1). The criteria were developed to identify patients with putative pathophysiologic manifestations of RAS (hypertension, renal dysfunction, or acute pulmonary edema) or advanced atherosclerosis in other vascular territories. Patients meeting at least one selection criterion before knowledge of the coronary anatomy were invited to participate and asked to provide written consent before their cardiac catheterization procedure. Patients not meeting a selection criterion before catheterization were asked to provide written consent to participate if cardiac catheterization demonstrated qualifying severe coronary atherosclerosis. We excluded patients for any of the following reasons: renal replacement therapy, known or suspected acute renal failure, history of contrast nephropathy, hemodynamic instability, physician preference, or refusal or inability to provide informed consent. Baseline serum creatinine was measured before the procedure; estimated creatinine clearance (CrCl) was calculated using the Cockcroft-Gault formula; and the glomerular filtration rate (GFR) was estimated using the Modification of Diet in Renal Disease formula (4). Renal dysfunction (as an inclusion criterion) was based on Cockcroft-Gault CrCl <50 ml/min or GFR <60 ml/min. Patients qualifying for inclusion on the basis of renal dysfunction alone were excluded if a well-documented cause of renal dysfunction existed. The study was approved by the Institutional Review Board of the University of British Columbia.
Renal artery angiography
Participating patients underwent either selective or nonselective renal angiography before completion of their cardiac catheterization procedure. Selective angiography was encouraged and generally employed a right Judkins coronary catheter with hand injection of 4 to 8 ml of contrast agent (ioversol; Optiray, Mallinckrodt, Pointe-Claire, Quebec, Canada) in each main and accessory renal artery and with supplementary semiselective injections if needed. Nonselective angiography, when utilized, was performed by powered injection of ≥20 ml Optiray over 1 s through a pigtail catheter positioned at the level of the L1 vertebral body in the postero-anterior projection. All images were recorded digitally at 30 frames/s. Digital subtraction was reserved for cases with poor visualization of the renal artery due to overlying gas or structures.
Each operator categorically graded main and proximal renal arteries as normal or abnormal (any roughening or stenosis consistent with atherosclerosis) and further categorized abnormal arteries according to visually estimated diameter stenosis severity (<50% [mild], 50% to 70% [significant], >70% to 99% [severe], or 100% [totally occluded]) and according to stenosis location (aorto-ostial or other). The consensus of at least two experienced angiographers was required in cases where stenosis severity was initially uncertain. When uncertainty remained and consensus could not be achieved, patients were excluded.
Administration of periprocedural intravenous fluids or N-acetyl cysteine was left to the discretion of the responsible physician. Repeat serum creatinine was obtained five to seven days after enrollment and evaluated centrally. We predefined procedure-related kidney dysfunction (PRKD) as an increase in serum creatinine of ≥0.5 mg/dl (44.2 μmol/l). Patients exhibiting PRKD had repeat serum creatinine determinations every two weeks until return to values within 0.5 mg/dl of baseline. We predefined chronic PRKD as dysfunction lasting at least 12 weeks. Requirement for renal replacement therapy within 12 weeks of enrollment was recorded and categorized as temporary or prolonged. For redundancy, we linked the data set to the British Columbia Renal Agency (a central registry of all dialysis patients in the Province of British Columbia) database to capture any incidence of dialysis within 12 weeks of enrollment.
Data management and statistical analysis
Registry data on all participating patients were entered in a central study database. Coronary anatomy and LV ejection fraction were obtained by linkage to the British Columbia Cardiac Registries (5).
Descriptive statistics are presented as the mean value ± SD or frequency (%). The clinical characteristics of patients were compared using analysis of variance for continuous variables and the chi-square test for categorical variables. Differences in lateral distribution of renal artery abnormalities were tested using Bowker's test of symmetry (6). A value of p < 0.05 for two-sided univariate tests was considered statistically significant. The logistic regression model was used to investigate the multivariate associations between severe RAS (≥70%) and patient characteristics. The variables included in the multivariate analysis were patient age and gender, diabetes, smoking, weight, GFR, systolic BP, diastolic BP, resistant hypertension, unexplained renal impairment, abdominal aortic/lower extremity disease, carotid artery disease, severe coronary artery disease (CAD), and acute pulmonary edema. The model fit was tested using the Hosmer-Lemeshow goodness-of-fit test (7). The backward elimination method was used to select the variables for the model. The area under the receiver-operating characteristics (ROC) curve for the selected model was determined by the trapezoidal rule.
During the screening period, 2,428 patients were evaluated. Of these patients, 47% (n = 1,149) met at least one selection criterion, of whom 298 met exclusion criteria because of non-renovascular kidney dysfunction (n = 14), renal replacement therapy (n = 16), known or suspected acute kidney failure (n = 5), pulmonary edema on current examination (n = 5), hemodynamic instability (n = 5), patient refusal (n = 109), or physician refusal (n = 144). Renal angiography was performed in 851 but was judged to be nondiagnostic in 14 (1.6%). The remaining 837 patients constituted the study cohort. Selective renal angiography alone was employed in 471 (56%), aortography in 314 (38%), and both techniques in 52 (6%) patients. Acute PRKD was observed in 17 patients (2%), but was transient in all but one. No patient required renal replacement therapy within 12 weeks of enrollment. No renovascular, aortic, or other catheter-related complications attributable to renal angiography were observed.
Prevalence of RAS in selected cohort
Table 2shows the prevalence of each category of RAS in the study cohort, stratified by the left or right renal artery. Angiographically evident renal artery atherosclerosis was common (n = 309 [36.9%]). A total of 120 (14.3%) patients had stenosis ≥50% in at least one proximal renal artery and 61 (7.3%) had severe stenosis ≥70%. Severe bilateral stenosis (≥70% in both arteries), however, was infrequent (n = 12 [1.4%]), and occlusions were rare (n = 8 [1%]). Significantly more abnormal renal arteries were observed on the left side (p = 0.028).
Selection criteria and RAS results
Table 3shows the proportion of patients meeting each selection criterion, together with each criterion's corresponding distribution of severity of RAS (the compound atherosclerosis/hypertension criteria are also shown according to their components). Severe atherosclerosis, predominantly due to the presence of severe coronary atherosclerosis, was the most frequent selection criterion and was present in 78% of patients (n = 651). Resistant or severe hypertension was present in 264 (32%) patients, unexplained kidney dysfunction in 232 (28%), and acute pulmonary edema in 9 (1%). Only 7 (2.2%) of 232 patients with unexplained kidney dysfunction were included on the basis of a documented history of acute renal dysfunction caused by angiotensin-converting enzyme inhibitors or angiotensin receptor blockers. Kidney dysfunction and pulmonary edema criteria were associated the highest prevalence of severe RAS (16% and 22%, respectively). However, these criteria identified fewer cases of severe stenosis than did the more commonly satisfied criterion of severe atherosclerosis.
Patients satisfying more than one selection criterion were common: 226 (27%) fulfilled two criteria and 46 (5%) fulfilled three or four criteria. Patients satisfying multiple inclusion criteria were significantly more likely to exhibit renal artery atherosclerosis and stenosis than those satisfying only one criteria. Requiring two or more selection criteria to qualify for renal angiography would have identified 41 (66%) of 61 cases of severe RAS from a cohort of 272 patients (15% prevalence).
Clinical and laboratory variables
Tables 4 and 5⇓⇓provide specific demographic, clinical, laboratory, and cardiac catheterization characteristics of the study cohort in aggregate and according to the findings at renal artery angiography. Univariate demographic and clinical variables associated with ≥70% RAS include age (73.2 vs. 66.5 years, p = 0.0001), female gender (13% vs. 6%, p = 0.001), lower body weight (72.7 vs. 79.9 kg, p = 0.0005), preprocedural systolic BP (146.4 vs. 135.4 mm Hg, p = 0.0084), and reduced CrCl (52.5 vs. 68.0 ml/min, p = 0.0001). No significant relationship was observed with diabetes, cigarette smoking history, diastolic BP, or LV ejection fraction.
Patients age <60 years accounted for 25% of the selected population, but only 2 (3%) of 61 cases of severe RAS and 9 (8%) of 120 cases of significant stenosis (Fig. 1). The prevalence of significant and severe RAS among women was approximately double that observed in men (significant: 22% vs. 12%, p = 0.001; severe: 13% vs. 6%, p = 0.001). This pattern was observed despite a similar prevalence of other peripheral or carotid artery disease and a significantly lower prevalence of severe coronary disease (50% vs. 75%, p = 0.001) (Fig. 2). The relationship between severe RAS and coronary disease burden, as expressed by vessel count, was complex and indirect (Table 5). The highest prevalence of severe RAS was observed with intermediate coronary disease burden.
Multivariate modeling showed age, female gender, clinically apparent carotid and peripheral arterial disease, kidney dysfunction, and systolic hypertension to be factors independently associated with severe RAS (Table 6). The area under the ROC curve for this model is 0.799. The Hosmer-Lemeshow goodness-of-fit statistic for the model is Ĉ = 10.02, with p = 0.26, indicating no evidence of a lack of fit in the selected model.
This study demonstrates the prevalence of angiographic RAS in a cohort of patients undergoing coronary angiography, who exhibited one or more risk factor criteria for RAS but who were not previously suspected of having RAS. Other investigators have also described unexpected RAS in patients undergoing cardiac catheterization, but they used abdominal aortography exclusively and did not employ prespecified criteria to select individuals at risk from all patients undergoing cardiac catheterization. In a seminal study, Harding et al. (8)found an 11% prevalence of RAS (defined as stenosis ≥50%) in an unselected cohort of 1,235 patients. Subsequent reports (9,10)derived from small series of clinically (rather than systematically) selected cohorts observed prevalence rates of 11% and 18%, respectively. These reports focused on the relationship between coronary disease burden and RAS prevalence but were limited by the absence of protocol-based patient selection and characterization.
The present study reflects a contemporary cardiac catheterization population and differs methodologically from previous reports in several important ways. First, we performed renal angiography only in patients manifesting a potential consequence of RAS (using predefined objective criteria of severe or resistant hypertension, unexplained kidney dysfunction, or pulmonary edema with preserved systolic function) or in whom objective evidence of severe atherosclerosis existed. This approach identified a prevalence of RAS higher than that observed in unselected cardiac catheterization-based studies. Second, rigorous screening for predefined enrollment criteria drove a detailed protocol-based and standardized collection of clinical and laboratory patient characteristics. The data generated provide a robust basis for modeling associations. Third, we encouraged the use of selective renal angiography, rather than aortography, and employed both techniques if the initial images were not diagnostic. The need for use of both techniques in 6% of registered patients speaks to the diagnostic superiority of this complementary approach over either technique alone. Fourth, we systematically screened for impairment of kidney function attributable to the procedures at regular intervals after the procedure. Finally, recognizing that no universally accepted angiographic definition for RAS exists, we distinguished between ≥50% and ≥70% stenosis (albeit semiquantitatively) and conservatively applied our modeling of associations only to stenoses ≥70%.
Our study extends and quantifies the relationships between RAS and extrarenal atherosclerosis burden. In our cohort, coronary disease burden was not observed to have an important association with RAS. Patients with two-vessel coronary disease (who, by protocol, required a noncoronary criterion for inclusion) displayed a 14% prevalence of severe RAS. Those with more advanced three-vessel or left main coronary disease (who most often qualified for inclusion on the basis of coronary disease burden alone) displayed an RAS prevalence of ≤8% (Table 4). We also found the prevalence of both renal atherosclerosis and significant or severe RAS was substantially higher in subgroups defined by peripheral or carotid disease than in groups defined by severe CAD. Renal artery stenosis remained strongly associated with clinically evident atherosclerosis in noncoronary territories, particularly the carotid arteries, after adjustment for all other known factors, including coronary disease burden.
The associations we observed between RAS and carotid or peripheral vascular disease may entirely be due to covariance and simply reflect a large total body plaque burden. However, there is accumulating evidence that the physiologic consequences of RAS may themselves drive atherosclerosis (11). For instance, disturbed intrarenal hemodynamics leading to inappropriate activation of the renin-angiotensin system with consequent long-term, excess stimulation of angiotensin type 1 receptors is hypothesized to have a pivotal pathogenetic role, and the converse (long-term antagonism of the renin-angiotensin system by angiotensin-converting enzyme inhibition) has been shown to reduce cardiovascular events and death (12). Furthermore, ischemic renal dysfunction itself can result not only in hypertension but also in other atherogenic disturbances, including increased oxidative stress, dyslipidemia, reduced excretion of homocysteine, abnormal phosphate/calcium metabolism, and anemia. In this light, it is possible that the associations previously noted could partly reflect a causal relationship between RAS and advanced generalized atherosclerosis.
Older age was strongly and independently associated with RAS, implying delayed development or slower progression of atherosclerosis in renal compared with coronary and other peripheral vascular territories. Despite meeting enrollment criteria, we detected significant RAS infrequently among individuals under 60 years of age. This observation has important practical implications for cardiac catheterization-based RAS screening.
The relationship between RAS and secondary hypertension mediated through inappropriate renin-angiotensin system activation has long been established (13). Although severe hypertension was associated with a 26% prevalence of severe RAS and an adjusted odds ratio of >4, it was an infrequent finding and therefore of limited potential clinical utility. The association between RAS and systolic BP, expressed as a continuous variable, holds greater promise for inclusion in formulae to determine the pretest likelihood of RAS. Notably, the association between RAS and diastolic BP did not persist in the multivariate model. Our data cannot distinguish whether this reflects abnormal vascular impedance typical of patients predisposed to RAS or impedance characteristics arising due to RAS-driven vascular hypertrophy, calcification, and sclerosis.
The strong association with female gender extends previous observations (14). Persistence of female gender in the multivariate model after adjustment for age and other factors is intriguing and remains unexplained. Moreover, our data suggest a gender-specific distribution of atherosclerosis characterized by greater coronary disease burden among men, greater RAS burden among women, and similar rates of clinically evident carotid and peripheral arterial disease in both genders. Heart failure with preserved systolic function has also recently been shown to be strongly associated with female gender (15). The pathogenetic basis for this association, however, remains enigmatic. The higher prevalence of RAS in women compared with men, which we observed, with a consequently greater burden of systolic hypertension, ischemic renal dysfunction, LV and vascular hypertrophy, and sodium retention could account, at least partly, for this observation.
Atherosclerotic RAS as a cause of kidney dysfunction and end-stage kidney disease is being increasingly recognized. Recent reports state that RAS is the most common potentially reversible disorder leading to renal replacement therapy (16,17). However, the degree to which RAS accounts for renal dysfunction in broadly defined cardiac populations is unknown. We observed a 24% prevalence of RAS (≥50%) in the 232 patients who met the enrollment criterion of unexplained kidney dysfunction. Furthermore, calculated CrCl demonstrated an independent and continuous relation to the likelihood of detecting RAS when examined in the entire cohort (including patients with normal renal function, mild dysfunction [CrCl >50 ml/min], or dysfunction attributed to known causes). Although this strong association does not necessarily imply causation, prospective studies of patients with identified RAS could be designed to measure the impact of RAS progression on progression of kidney disease. At a minimum, it appears that clinicians should consider RAS when kidney dysfunction and CAD coexist.
In addition to the substantial burden of end-stage disease of the kidney associated or attributable to RAS, a large body of evidence now links renal function to cardiovascular mortality in patients with cardiac disease (18–20). Furthermore, the severity of RAS has been shown to correlate to mortality in long-term follow-up of a cardiac catheterization population (11). Thus, well-designed, large-scale, randomized trials comparing current medical management alone to revascularization of the kidney need to be conducted to define the most appropriate therapy for RAS in cardiac populations. Key questions include whether kidney revascularization can preserve or improve kidney dysfunction, delay the need for renal replacement therapy, or perhaps, most importantly, reduce cardiovascular events and mortality. Methodologic limitations of existing trials have prompted the development of American Heart Association guidelines for reporting of renal artery revascularization in clinical trials (21). Efficient identification of patients with RAS will be needed to conduct such definitive trials. The findings reported herein could enable the identification of patients with a significant pretest likelihood of RAS based on easily captured clinical and laboratory data.
The observations and associations made in our analysis apply with certainty only to patients undergoing cardiac catheterization who met one or more of our predefined selection criteria. Although broad and clinically intuitive, these criteria apply to only one-half of all patients undergoing non-emergent cardiac catheterization at our institution. We did not employ quantitative or independent analysis of the severity of RAS. Quantitative analysis of aorto-ostial lesions is, in any event, problematic. We used categorical grading of severity, intended to minimize potential discrepancies and deemphasize small gradations of severity. Furthermore, the use of a single-plane projection for assessment of stenosis severity may fail to identify eccentric stenoses. Our measures of prevalence are therefore likely to be conservative. It is possible our method of safety surveillance failed to document new kidney impairment due to atheroembolism, which occurred after the five- to seven-day follow-up after the procedure. Because renal replacement therapy in British Columbia is registered centrally, we are confident that no patient in our cohort underwent dialysis within 12 weeks of enrollment.Finally, our analysis did not attempt to correlate abnormalities in kidney size or split renal function with RAS.
This study characterizes, in detail, patients at risk of RAS in whom scheduled cardiac catheterization afforded an opportunity to determine whether RAS was present by angiographic criteria. We found the overall prevalence of significant and severe angiographic RAS was substantial in this cohort and described independent associations between severe RAS and readily identifiable clinical and laboratory variables. We also found selective renal angiography coincident with cardiac catheterization to be safe in an otherwise high-risk group with prevalent peripheral vascular atherosclerosis and preexisting kidney dysfunction when performed by experienced cardiac angiographers. Long-term follow-up of this cohort to determine the significance of RAS, with respect to specific heart and kidney disease outcomes, is the focus of future research.
The authors thank the Interventional Cardiology Research Coordinators: Jaclyn Chow, Sandra MacLeod, and Brenda Mercier, at Vancouver Hospital, for their thorough data collection, as well as Anne Eichmann for database management.
☆ This study was funded internally by the Vancouver Hospital Interventional Cardiology Clinical Trials trust.
- blood pressure
- coronary artery disease
- creatinine clearance
- glomerular filtration rate
- left ventricular
- procedure-related kidney dysfunction
- renal artery stenosis
- Received June 27, 2003.
- Revision received November 7, 2003.
- Accepted November 13, 2003.
- American College of Cardiology Foundation
- ↵Norwegian Institute of Public Health. World Health Organization Collaborating Centre for Drug Statistics Methodology. Guidelines for ACC Class and DDD Assignment. 6th edition. Oslo, Norway, 2003.
- Hosmer D.W,
- Lemeshow S
- Harding M.B,
- Smith L.R,
- Himmelstein S.I,
- et al.
- Weber-Mzell D,
- Kotanko P,
- Schumacher M,
- et al.
- Conlon P.J,
- Athirakul K,
- Kovalik E,
- et al.
- Goldblatt H,
- Lynch J,
- Hanzal R.F,
- Summerville W.W
- Kuroda S,
- Nishida N,
- Uzu T,
- et al.
- Masoudi F.A,
- Havranek E.P,
- Smith G,
- et al.
- Mailloux L.U,
- Napolitano B,
- Bellucci A.G,
- et al.
- McCullough P.A,
- Sandeep S.S,
- Shah S.S,
- et al.
- Rundback J.H,
- Sacks D,
- Kent K.C,
- et al.