Assessing outcomes to determine whether symptoms related to hypertension justify renal artery stenting

From the Peripheral Vascular Surgery Society Assessing outcomes to determine whether symptoms related to hypertension justify renal artery stenting J. Gregory Modrall, MD, a,b Eric B. Rosero, MD, a,b Carlos H. Timaran, MD, a,b Thomas Anthony, MD, a Jayer Chung, MD, b R. James Valentine, MD, b and Clayton Trimmer, DO, c Dallas, Tex Objective: The goal of the study was to determine the blood pressure (BP) response to renal artery stenting (RAS) for patients with hypertension urgency, hypertension emergency, and angina with congestive heart failure (angina/ congestive heart failure [CHF]). Methods: Patients who underwent RAS for hypertension emergencies (n 13), hypertension urgencies (n 25), and angina/chf (n 14) were included in the analysis. By convention, hypertension urgency was defined by a sustained systolic BP > 180 mm Hg or diastolic BP > 120 mm Hg, while the definition of hypertension emergency required the same BP parameters plus hypertension-related symptoms prompting hospitalization. Patient-specific response to RAS was defined according to modified American Heart Association reporting guidelines. Results: The study cohort of 52 patients had a median age of 66 years (interquartile range 58-72). The BP response to RAS varied significantly according to the indication for RAS. Hypertension emergency provided the highest BP response rate (85%), while the response rate was significantly lower for hypertension urgency (52%) and angina/chf (7%; P.03). Only 1 of 14 patients with angina/chf was a BP responder. Multivariate analysis showed that hypertension urgency or emergency were not independent predictors of BP response to RAS. Instead, the only independent predictor of a favorable BP response was the number of preoperative antihypertensive medications (odds ratio 7.5; 95% confidence interval 2.5-22.9; P.0004), which is another indicator of the severity of hypertension. Angina/CHF was an independent predictor of failure to respond to RAS (odds ratio 118.6; 95% confidence interval 2.8-999.9; P.013). Conclusions: Hypertension urgency and emergency are clinical manifestations of severe hypertension, but the number of preoperative antihypertensive medications proved to be a better predictor of a favorable BP response to RAS. In contrast, angina/chf was a predictor of failure to respond to stenting, providing further evidence against the practice of incidental stenting during coronary interventions. (J Vasc Surg 2012;55:413-20.) Recent studies have documented that 20% to 66% of patients had no improvement in blood pressure (BP) control after renal artery stenting (RAS). 1-3 A potential explanation for these treatment failures is the relative difficulty of identifying patients with renovascular hypertension within the vast milieu of hypertensive patients. The 2006 American Heart Association practice guidelines suggested that RAS may be considered in patients with accelerated hypertension, resistant hypertension, malignant hypertension, hypertension with an unexplained unilateral small From the Dallas Veterans Affairs Medical Center a ; and the Division of Vascular and Endovascular Surgery, Department of Surgery, b and Division of Interventional Radiology, Department of Radiology, c University of Texas Southwestern Medical School. Competition of interest: Dr Trimmer is a paid consultant for Cook Medical, Inc, holds stock in Boston Scientific Corporation, and has received NIH funding related to the CORAL Trial. Presented at the Thirty-sixth Annual Spring Meeting of the Peripheral Vascular Surgery Society, Chicago, Ill, June 15, 2011. Reprint requests: J. Gregory Modrall, MD, Division of Vascular Surgery, University of Arkansas for Medical Sciences, 4301 West Markham, #520, Little Rock, AR 72205 (e-mail: gmodrall@uams.edu). The editors and reviewers of this article have no relevant financial relationships to disclose per the JVS policy that requires reviewers to decline review of any manuscript for which they may have a competition of interest. 0741-5214/$36.00 Copyright 2012 by the Society for Vascular Surgery. Published by Elsevier Inc. All rights reserved. doi:10.1016/j.jvs.2011.08.056 kidney, and hypertension with intolerance to medication, which are clinical clues to the presence of renovascular hypertension. 4 In practice, these guidelines require a degree of clinical judgment or subjectivity in patient selection that may lead to errors in diagnosis and compromise clinical outcomes. To address this clinical dilemma, our efforts have focused on identifying preoperative predictors of outcome for RAS that may improve patient selection for stenting. 3,5 Resistant or malignant hypertension may present as a hypertension emergency or hypertension urgency. 6 Some clinicians have proposed that renovascular hypertension may also exacerbate cardiac symptoms, such as angina and congestive heart failure (CHF). 7 Ostensibly any patient with one of these clinical presentations would be an ideal candidate for RAS, although there are scant data on the outcomes for RAS performed for these indications. Therefore, the aim of this study was to report the outcomes of RAS performed as treatment for hypertension emergency, urgency, or angina/chf to determine if these hypertensive symptoms predict a favorable BP response to RAS. METHODS Study population and clinical data collection. This retrospective review examined the outcomes of 52 patients who underwent RAS at the University of Texas Southwestern Medical School and its affiliated hospitals over a 9-year 413

414 Modrall et al JOURNAL OF VASCULAR SURGERY February 2012 period (January 1, 2000 to July 1, 2009) for symptoms believed to be related to renovascular hypertension. Subjects were culled from billing databases from the Departments of Surgery and Radiology. The indications for RAS included hypertension emergency (n 13), hypertension urgency (n 25), or angina/chf (n 14). By convention, hypertension urgency was defined by a sustained systolic BP 180 mm Hg or diastolic BP 120 mm Hg, while the definition of hypertension emergency required the same BP parameters plus hypertension-related symptoms prompting hospitalization. 6 Exclusion criteria included nonatherosclerotic renal artery lesions, secondary renal artery stenoses after prior angioplasty or stenting, and insufficient postoperative follow-up ( 1 month) to assess the clinical outcomes for RAS. The Institutional Review Boards of the University of Texas Southwestern Medical School and the Dallas Veterans Affairs Medical Center approved the study. Preoperative demographic and clinical data were collected on each patient. Average preoperative and postoperative BP, serum creatinine, and estimated glomerular filtration rate (egfr) were calculated from a minimum of two clinic visits at each time point. Postoperative data were obtained from the last follow-up visits. Average BP and numbers of antihypertensive medications were quantified as reported previously. 3 When multiple BP measurements were obtained at a single clinic visit, the arm with the highest BP was used and all BP measurements for that day were averaged. All antihypertensive medications known to influence BP were tallied for each clinic visit, including diuretics and nitrates prescribed for cardiac purposes. The abbreviated Modification of Diet in Renal Disease formula was used to calculate egfr: 186 (serum creatinine) 1.154 (age) 0.203 (0.742 if female) (1.210 if African American). 8 Patients receiving dialysis were assigned an egfr of 10 ml/min/1.73 m 2. Definition of clinical response to RAS. The primary outcome measure was BP response to RAS. Patients were categorized as BP responders or nonresponders based on a modification of American Heart Association reporting guidelines, 7 as reported previously. 3,5 The current modification of these reporting guidelines utilized a more liberal target BP of 160/90 mm Hg since the target BP proposed by the American Heart Association changed during the years encompassed by the current series. To assess the BP response to RAS, the average post-stenting BP and number of antihypertensive medications at last follow-up were compared with the average pre-stenting BP and number of medications for each patient. BP responders were defined by an average postoperative BP 160/90 on a reduced number of antihypertensive medications or a reduction in average diastolic BP to 90 mm Hg on the same number of medications after RAS. All other patients were deemed BP nonresponders. Stented patients were categorized as renal function responders if the average egfr at last follow-up improved 20% over average pre-stenting egfr in accordance with the American Heart Association reporting guidelines and prior publications. 1,5,9 Patients with stable Table I. Cohort demographics (n 52) Risk factor Patients, No. (%) Age (years) 66 years (58-72) Female gender 34 (65.4) Race White 24 (46.2) African American 15 (28.9) Hispanic/other 13 (25.0) Hypertension 52 (100) Chronic renal insufficiency 26 (50.0) Diabetes 21 (40.4) Coronary artery disease 32 (61.5) Hyperlipidemia 24 (46.2) Tobacco history 48 (92.3) COPD 9 (17.3) COPD, Chronic obstructive pulmonary disease; IQR, interquartile range. For age, data are presented as median (IQR). For comorbid conditions, data are presented as frequency (%). ( 20% change in egfr) or worsened renal function ( 20% decrease in egfr) after RAS were categorized as renal function nonresponders. Statistical analysis. The primary endpoint of the study was BP response to RAS. Secondary endpoints included procedural complications, renal function response to stenting, and survival. Categorical data were reported as proportions using 2 or Fisher exact tests. Continuous data were reported as medians with interquartile ranges (IQR), and comparisons between groups were performed with Wilcoxon matched pairs tests and Mann-Whitney U tests. Survival was calculated according to the Kaplan-Meier technique. Survival curves were compared using the logrank test. Stepwise multivariate logistic regression analysis was used to identify predictors of BP response to stenting. The regression model included variables that were significant (P.05) on univariate analysis. Interaction between the variables in the regression model was tested in multiple stages. First, correlation analysis was performed between the variables in the model. Thereafter, a test for multicollinearity was performed. Finally, interaction between the variables was tested in the stepwise logistic regression analysis. For all statistical analyses, the threshold for significance was.05. Statistical analysis was performed using SAS, version 9.13 (SAS Institute, Inc, Cary, NC). RESULTS Baseline characteristics. The cohort was typical of a patient population with renovascular disease (Table I). At baseline, the median pre-stenting systolic BP was 184 mm Hg (IQR 154-192 mm Hg), and the median pre-stenting diastolic BP was 87 mm Hg (IQR 76-100 mm Hg). The median number of preoperative antihypertensive medications was 3 (IQR 2-4 medications). The median preoperative serum creatinine, excluding three patients (5.8%) receiving renal replacement therapy, was 1.4 mg/dl (IQR 1.0-2.0 mg/dl). The median pre-stenting egfr for the entire cohort was 48 ml/min/1.73 m 2 (IQR 26-60).

JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 Modrall et al 415 Table II. Preoperative clinical parameters, stratified by clinical presentation Preoperative parameter HTN emergency (n 13) HTN urgency (n 25) Angina/CHF (n 14) Age, years 68 (60-75) 68 (57-74) 65 (59-80) Gender, % females 53.8% 60.0% 85.7% Systolic blood pressure, mm Hg 188 (166-214) 188 (179-194) 151 (143-166) Diastolic blood pressure, mm Hg 103 (91-116) 88 (78-97) 79 (69-85) No. of antihypertensive MEDS 4 (3-5) 3 (2-5) 3 (2-4) Serum creatinine (mg/dl) a 1.6 (1.1-2.1) 1.7 (1.0-2.2) 1.0 (0.8-1.3) egfr (ml/min/1.73 m 2 ) 38 (19-60) 38 (27-60) 60 (45-60) egfr, Estimated glomerular filtration rate; HTN, hypertension; IQR, interquartile range; MEDS, medications. Categorical data are presented as frequency (%) and compared with 2 square or Fisher exact test as appropriate. Continuous data are presented as median (IQR) and compared with Kruskal-Wallis test. a Excluding patients receiving renal replacement therapy. Table III. Cohort outcomes after renal artery stenting (n 52) Parameter Preoperative Postoperative P value Systolic blood pressure, mm Hg 184 (154-192) 147 (132-164).0001 Diastolic blood pressure, mm Hg 87 (76-100) 74 (65-86).0001 No. of antihypertensive MEDS 3 (2-4) 3 (3-4).72 Serum creatinine (mg/dl) a 1.4 (1.0-2.0) 1.3 (1.0-2.0) 1.0 egfr (ml/min/1.73 m 2 ) 48 (33-69) 48 (26-66).82 Proportion on dialysis (%) 5.8% 13.5%.32 egfr, Estimated glomerular filtration rate; IQR, interquartile range; MEDS, medications. Cohort data are analyzed in aggregate. Data are presented as medians (IQR) and compared with Wilcoxon matched pairs test. a Excluding patients receiving renal replacement therapy. Indications for treatment. The indications for RAS included the presence of at least one renal artery stenosis and concurrent hypertension emergency (25%), hypertension urgency (48.1%), or angina/chf (26.9%). By definition, all 13 patients diagnosed with a hypertension emergency had symptomatic hypertension, including flash pulmonary edema in six patients and headache and blurred vision in seven patients. No patient with angina/chf had other symptoms of hypertension, such as headache or blurred vision. At baseline, there were some significant differences between patients with the different clinical presentations (Table II). Not surprisingly, patients presenting with a hypertension emergency or urgency arrived with a significantly higher average preoperative systolic BP (P.0006) and diastolic BP (P.0005) than patients with angina/ CHF. Patients with hypertension emergency or urgency also had worse renal function prior to stenting than patients with angina/chf (P.007). Procedural details. A total of 65 renal arteries were stented in 52 patients. Unilateral RAS was performed in 39 of 52 patients (75%), while bilateral stenting was performed in the remaining 13 patients. Three patients (5.8%) had solitary kidneys stented. Procedural complications occurred in three of 52 (5.8%) patients, including one groin hematoma and two renal artery dissections that required additional stents. Treated renal artery stenoses had a median stenosis of 70% (IQR 70%-80%) diameter narrowing, based on procedure reports. Embolic protection was employed in three of 52 (5.8%) cases. Predilatation of the culprit stenosis was performed in four patients (7.7%). All renal stents were balloon expandable stents, having a median diameter of 6 mm (IQR 5-6 mm) and length of 18 mm (IQR 15-18 mm). Cohort outcomes. The median follow-up for the entire cohort was 17 months (IQR 9-27 months). The outcomes for the cohort are summarized in Table III. Systolic and diastolic BPs were significantly lower after RAS, while the number of antihypertensive medications required to achieve BP control was unchanged. Neither serum creatinine nor egfr was significantly improved after RAS. A higher proportion of patients required renal replacement therapy after stenting, but the difference was not statistically significant. During follow-up, one patient developed in-stent stenosis that was heralded by an acute deterioration in renal function. That patient had a pre-stenting serum creatinine of 1.9 mg/dl that improved to 1.5 mg/dl post-stenting, corresponding to an improvement in egfr from 36 to 47 ml/min/1.73 m 2. After having stable renal function for several months, he returned 10 months post-stenting with worsened renal function (serum creatinine 2.4 mg/dl; egfr 47 ml/min/ 1.73 m 2 ). After treatment of the in-stent stenosis with repeat angioplasty, the serum creatinine returned to 1.6 mg/dl (egfr 44 ml/min/1.73 m 2 ). Patient-specific outcomes. Twenty-five of 52 patients (48.1%) had a favorable BP response to RAS and were

416 Modrall et al JOURNAL OF VASCULAR SURGERY February 2012 Table IV. Outcomes from renal artery stenting stratified by clinical indication Outcome variable a HTN emergency (n 13) HTN urgency (n 25) Angina/CHF (n 14) P value BP response rate 84.6% 52.0% 7.1%.0003 Change in systolic BP (mm Hg) 43 ( 31 to 71) 32 ( 55 to 0) 7 ( 23 to 21).001 Change in diastolic BP (mm Hg) 18 ( 46 to 12) 12 ( 20 to 3) 2 ( 14 to 8).0025 Change in number of MEDS 1 ( 1.5 to 0) 0 ( 1 to 1) 0 (0 to 1.3).046 Change in egfr (ml/min/1.73 m 2 ) 0( 9 to10) 3 ( 10 to 3.5) 0 ( 12 to 7).58 BP, Blood pressure; CHF, congestive heart failure; HTN, hypertension; IQR, interquartile range; MEDS, antihypertensive medications. Categorical data are presented as frequency (%) and compared with 2 or Fisher exact test as appropriate. Continuous data are presented as median (IQR) and compared with Kruskal-Wallis test. a Outcomes represent a comparison of postoperative and preoperative data. Fig. The Kaplan-Meier plot depicts survival after renal artery stenting for responders (continuous line) and nonresponders (discontinuous line). Ticks along the lines represent censored cases. The number at risk is provided for each time point. The standard error for responders and nonresponders exceeded 10% at 52 and 44 months, respectively. There was no significant difference in survival between the two groups (log-rank test, P.35). deemed BP responders, according to the criteria defined in the Methods section. The BP response to RAS varied significantly according to the indication for RAS (Table IV). Hypertension emergency provided the highest BP response rate (85%), while the response rate was significantly lower for hypertension urgency (52%) and angina/ CHF (7%; P.03). Only one of 14 patients with angina/ CHF was a BP responder. After stenting, responders enjoyed a median decrease in systolic BP of 43 mm Hg (IQR 31 to 61 mm Hg) from baseline, compared witha4mmhgdecrease (IQR 19 to 27 mm Hg) among nonresponders (P.0001). Diastolic BP was also decreased substantially more in responders ( 18 mm Hg, IQR 10 to 30 mm Hg) than nonresponders ( 2mmHg,IQR7to 14 mm Hg; P.0001) after stenting. This improvement in BP control occurred despite fewer antihypertensive medications for responders. The median number of antihypertensive medications required for BP control decreased by one for responders (IQR 1to 2 medications), while there was no change in the number of antihypertensive medications for nonresponders (IQR 0 to two medications; P.0001). Despite having improved BP control, survival was not significantly different between BP responders and nonresponders (Fig). Fewer patients derived any improvement in renal function from RAS, as only 10 of 52 patients (19.2%) had an improvement in postoperative egfr of at least 20% over baseline. Among patients with a pre-stenting serum creatinine of 1.5 mg/dl or greater, the response rate was higher (33.3%). A major focus of this analysis was to define preoperative differences between BP responders and nonresponders that may be clinically useful in patient selection. In Table V, responders and nonresponders were compared. The indication for stenting was different for responders and nonresponders. Specifically, responders were more likely to present with a hypertension emergency or urgency, whereas nonresponders were not likely to present with a hyperten-

JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 Modrall et al 417 Table V. Comparison of preoperative clinical parameters, responders in contrast to nonresponders Preoperative parameter Responders (n 25) Nonresponders (n 27) P value Age (years) 68 (61-73) 65 (57-72).34 Gender (% females) 48.0% 22.2%.08 Race White 52.0% White 44.4%.09 African American 36.0% African American 18.5% Hispanic/other 12.0% Hispanic/other 37.1% Indication for stenting HTN emergency 44.0% HTN emergency 7.4%.0003 HTN urgency 52% HTN urgency 44.5% Angina 4% Angina 48.1% Systolic blood pressure, mm Hg 188 (173-194) 175 (149-189).04 Diastolic blood pressure, mm Hg 94 (78-102) 82 (76-89).04 Proportion with diastolic blood pressure 90 mm Hg 68.0% 25.9%.005 No. of antihypertensive MEDS 4.0 (3.5-5.0) 3.0 (2.0-3.0).0001 Proportion requiring 76.0% 18.5%.0001 4 antihypertensive MEDS 52.0% 14.8%.007 Clonidine use egfr, ml/min/1.73 m 2 48 (34-68) 45 (25-69).71 Proportion on hemodialysis 0% 11.1%.24 Bilateral stenting 24.0% 25.9% 1.0 Solitary kidney 8.0% 3.7% 1.0 egfr, Estimated glomerular filtration rate; HTN, hypertension; IQR, interquartile range; MEDS, medications. Categorical data are presented as frequency (%) and compared with 2 or Fishers exact test as appropriate. Continuous data are presented as median (IQR) and compared with Mann-Whitney U test. sion urgency or angina/chf. In addition, responders had more severe hypertension than nonresponders prior to stenting. This was manifested as a higher median systolic BP, diastolic BP, and antihypertensive medication requirements among responders. Consequently, the proportion of stented patients with a diastolic BP 90 mm Hg, a requirement for 4 antihypertensive medications, and clonidine use was higher among responders than nonresponders. Renal function at baseline was not significantly different between BP responders and nonresponders. There were no procedural differences that distinguished responders from nonresponders. The proportion of bilateral stenting procedures and solitary kidneys was not significantly different between the responders and nonresponders (Table V). The stented lesions were not significantly different between responders and nonresponders, based on the degree of stenosis (responders in contrast to nonresponders 75% [IQR 70%-80%] in contrast to 70% [60%-80%]; P.29). Median stent diameter was not significantly different between responders and nonresponders (responders in contrast to nonresponders 6mm [IQR 5-6 mm] in contrast to 6 mm [IQR 5-6 mm]; P.40). Median stent length was also similar between responders and nonresponders (responders in contrast to nonresponders 18 mm [IQR 15-18 mm] in contrast to 18 mm [IQR 15-18 mm]; P.76). Multivariate analysis was used to identify independent predictors of a positive BP response to RAS. From the univariate analyses (Table V), five variables were included in the multivariate model: (1) indication for stenting (hypertension emergency in contrast to urgency in contrast to angina/chf); (2) systolic BP; (3) diastolic BP, (4) number of antihypertensive medications, and (5) clonidine use. Recognizing the potential for interaction between these variables, tests of correlation between these variables were performed. No strong correlations were noted between the variables (r.6 in all cases), suggesting a low probability of collinearity. A subsequent test for multicollinearity yielded tolerance values exceeding 0.40, confirming that these variables were not highly correlated. The multivariate analysis showed that hypertension urgency or emergency were not independent predictors of BP response to RAS. Indeed, the only independent predictor of a favorable BP response was the number of preoperative antihypertensive medications (odds ratio 7.5; 95% confidence interval 2.5-22.9; P.0004). Angina/CHF was an independent predictor of failure to respond to RAS (odds ratio 118.6; 95% confidence interval 2.8-999.9; P.013). There was no significant interaction between the variables in the model. The C statistic for the model was.94, indicating an excellent ability to discriminate responders from nonresponders. The Hosmer-Lemeshow goodness-of-fit test for the model was nonsignificant (P.96), which confirms that there was little departure from a perfect fit. DISCUSSION Among the vast milieu of hypertensive patients, there is a subset with renovascular hypertension for which RAS may remediate the hypertensive diathesis. The challenge is to correctly identify those patients who have renovascular hypertension, especially those who will benefit from stenting. Unfortunately, the most recent American Heart Association practice guidelines are too nonspecific to guide interventionalists with patient selection. 4 To address this challenge, our group recently identified three clinical predictors of a favorable BP response to stenting: (1) a requirement for 4 antihypertensive medications to treat the hypertension; (2) persistent diastolic BP 90 mm Hg; and (3) use of clonidine as an antihypertensive agent. In the current study, we extended these observations by describ-

418 Modrall et al JOURNAL OF VASCULAR SURGERY February 2012 ing the clinical outcomes for RAS performed to treat hypertension emergency, hypertension urgency, or angina/ CHF. Several important findings emerged from this analysis. First, we noted that hypertensive emergency and hypertension urgency were associated with a positive BP response in 85% and 52% of stented patients, respectively. Second, when these indications for intervention were included in a multivariate analysis of candidate predictors of BP response, the number of antihypertensive medications used for hypertension therapy was the sole independent predictor of a positive BP response to stenting. This result reinforces the findings of our prior study, which observed an incremental improvement in the BP response rate with increasing numbers of preoperative antihypertensive medications. Finally, we found that angina/chf is a predictor of failure to respond to RAS, suggesting that cardiac symptoms per se may not be useful in selecting patients for stenting if BP control is the primary goal of treatment. The finding that hypertension emergency and urgency were not independent predictors of a positive BP response to RAS appears to be incongruent with the high response rate observed for these clinical indications. However, a closer examination of the pattern of medication use in this cohort will clarify this finding. The majority (85%) of patients who presented with a hypertension emergency was taking at least four antihypertensive medications for hypertension control prior to admission, and the BP response rate in this group was proportionately high (85%). By comparison, the antihypertensive medication requirement for patients with hypertension urgency was more variable. Those patients with hypertension urgency who required 4 antihypertensive medications had an improvement in BP after stenting in 11 of 13 cases (85%), whereas those patients taking fewer than four antihypertensive medications were nonresponders in the majority (92%) of cases. The association between increasing preoperative antihypertensive medications and the BP response to RAS is even more compelling when one notes that our prior study observed a similar relationship in a significantly larger cohort of 149 patients. The sum of these observations is that clinicians should carefully consider the number of antihypertensive medications in selecting patients for stenting for BP control, regardless of the clinical presentation. It is equally important to identify those patients who are unlikely to respond to stenting. The current study found that only one of 14 patients with angina/chf experienced any improvement in BP after renal stenting. We interpret these data as an admonition that the presence of angina/chf should not be used as a primary indication for RAS if BP control is the primary goal of treatment. Even among the patients with angina/chf who required four or more antihypertensive medications, only one in five patients (20%) had any improvement in BP after stenting. There may be something fundamentally different about the patients with angina/chf, although we cannot offer any reasonable explanation for this observation. Severe hypertension is a classic indication for RAS, but we are aware of no articles that specifically report the outcomes for RAS in patients with hypertension emergency or urgency per se. It is likely that patients with these clinical manifestations of severe hypertension constituted an undefined proportion of those patients in multiple prior studies on RAS, 1,2 but we believe that the current study offers the first dedicated analysis of the outcomes this subset of patients. A small number of prior studies have investigated the relative benefits of RAS among patients with cardiac symptoms. Those studies touted the potential cardiac benefits of RAS, rather than the effects on BP control. Khosla and colleagues observed an improvement in angina and CHF symptoms after RAS in a cohort of 48 patients. 10 It is not entirely clear, however, that renal stenting was the primary reason for these improvements, as many of the patients in that series also underwent concurrent percutaneous coronary interventions. 10 In a cohort of 39 patients who underwent RAS for recurrent episodes of congestive heart failure or flash pulmonary edema, Gray and colleagues found that the average New York Heart Association Functional Class decreased from 2.9 to 1.6 after RAS. 11 Whether these improvements translate into improved survival over the long term is unknown. Certainly these studies raise the possibility that RAS may have benefits aside from the immediate effects on hypertension or renal function. The current study offers additional insight into the expected outcomes for the most severe clinical presentations of hypertension, but there are limitations to the study that are merit further discussion. First, the size of this cohort raises the question of whether the statistical power of the study is sufficient to identify all predictors. For those variables that were significant on multivariate analysis, a larger sample size would not have altered the outcome of the analysis. For the remaining variables, however, we cannot exclude the possibility of a type II error that would alter the conclusions of the study. This concern suggests that it may be prudent to reexamine these variables in a larger study cohort. Second, some clinicians may find the definitions of responders and nonresponders to be highly subjective and without obvious clinical relevance. Although we acknowledge this criticism, it is worthwhile noting that these definitions represent a modification of American Heart Association reporting guidelines. 7 Defining the success or failure of treatment is inherently subjective when the outcome variable is a continuous variable such as BP. Third, the primary endpoint of the study was BP response to stenting, which ignored other potential benefits of stenting, such as the effect on renal function and late cardiovascular morbidity and mortality. In the current series, only 19% of patients experienced an improvement in post-stenting egfr of 20% or greater over baseline. The outcomes for renal function may be misleading, though, since ischemic nephropathy was not an indication for stenting in this series. An additional limitation is the lack of duplex-derived resistive indexes for most patients in the cohort. The absence of those data precluded any comparisons with clinical predictors. Finally, the predictors of outcome identified in the current and prior studies must be validated prospectively in future studies.

JOURNAL OF VASCULAR SURGERY Volume 55, Number 2 Modrall et al 419 CONCLUSIONS Hypertension urgency and emergency are clinical manifestations of severe hypertension, but the number of preoperative antihypertensive medications proved to be a better predictor of a favorable BP response to RAS. In contrast, angina/chf was a predictor of failure to respond to stenting, providing further evidence against the practice of incidental RAS during coronary interventions. AUTHOR CONTRIBUTIONS Conception and design: JM, CT Analysis and interpretation: JM, ER, CHT, TA, JC, RV, CT Data collection: JM, CT Writing the article: JM, ER, CHT, TA, JC, RV, CT Critical revision of the article: JM, ER, CHT, TA, JC, RV, CT Final approval of the article: JM, ER, CHT, TA, JC, RV, CT Statistical analysis: JM, ER, CHT, TA Obtained funding: Not applicable Overall responsibility: JM REFERENCES 1. Nolan BW, Schermerhorn ML, Rowell E, Powell RJ, Fillinger MF, Rzucidlo EM, et al. Outcomes of renal artery angioplasty and stenting using low-profile systems. J Vasc Surg 2005;41:46-52. 2. Beck AW, Nolan BW, De Martino R, Yuo TH, Tanski WJ, Walsh DB, et al. Predicting blood pressure response after renal artery stenting. J Vasc Surg 2010;51:380-5. 3. Modrall JG, Rosero EB, Leonard D, Timaran CH, Anthony T, Arko FA III, et al. Clinical and kidney morphologic predictors of outcome for renal artery stenting: data to inform patient selection. J Vasc Surg 2011, 2011;53:1282-90. 4. Hirsch AT, Haska ZJ, Hertzer NR, Bakal CW, Creager MA, Halperin JL, et al. ACC/AHA 2006 practice guidelines for the management of patients with peripheral arterial disease (lower extremity, renal, mesenteric, and abdominal aortic). Cir 2006;113:e463-5. 5. Modrall JG, Timaran CH, Rosero EB, Chung J, Arko FA 3rd, Valentine RJ, et al. Predictors of outcome for renal artery stenting performed for salvage of renal function. J Vasc Surg 2011;54:1414-21.e1. 6. Chobanian AV, Bakris GL, Black HR, Cushman WC, Green LA, Izzo JL, et al. Seventh report of the joint national committee on prevention, detection, evaluation, and treatment of high blood pressure. Hypertension 2003;42:1206-52. 7. Rundback JH, Sacks D, Kent KC, Cooper C, Jones D, Murphy T, et al. Guidelines for the reporting of renal artery revascularization in clinical trials. Circulation 2002;106:1572-85. 8. Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461-70. 9. Edwards MS, Craven BL, Stafford J, Craven TE, Sauve KJ, Ayerdi J, et al. Distal embolic protection during renal artery angioplasty and stenting. J Vasc Surg 2006;44:128-35. 10. Khosla S, White CJ, Collins TJ, Jenkins JS, Shaw D, Ramee SR. Effects of renal artery stent implantation in patients with renovascular hypertension presenting with unstable angina or congestive heart failure. Am J Cardiol 1997;80:363-6. 11. Gray BH, Olin JW, Childs MB, Sullivan TM, Bacharach JM. Clinical benefit of renal artery angioplasty with stenting for the control of recurrent and refractory congestive heart failure. Vasc Med 2002;7: 275-9. Submitted Jun 19, 2011; accepted Aug 30, 2011. DISCUSSION Dr Jeffrey Indes (New Haven, Conn). I would like to thank the PVSS for the opportunity to discuss this paper and thank Dr Modrall for giving me the manuscript and for how well written it is. He and his colleagues report the outcomes of renal artery stenting performed as a treatment for hypertensive emergency, urgency, angina, or congestive heart failure (CHF) to determine if these hypertensive symptoms portend to a favorable blood pressure response to renal artery stenosis. Interestingly, in your article you found that four out of 27 patients who had no blood pressure response to stenting derived a significant improvement in estimated glomerular filtration rate (egfr) after stenting. Conversely, fewer patients derived any improvement in postop egfr after stenting, specifically 12 of 52 patients, while this increased to 42.3% if their preoperative creatinine was greater than 1.5. These factors in addition to your findings suggest that patient selection criteria is crucial to good outcomes in these patients, which leads me to my first question. Given these findings, do you think our current criteria for renal artery stenting should be modified in any way, and, if so, how? Number two, as you know, renal resistive index has been shown to be associated with success and failure of renal artery stenting in past studies. Do you think this could be looked at in your cohort of patients? Dr J. Gregory Modrall. Thank you for your questions and your comments. The focus of our research program has been trying to identify predictors of outcome. The randomized clinical trials have shown no benefit to stenting. The problem with those trials is that they were too inclusive, so that they should be viewed as an assessment of indiscriminate use of stenting. We believe that better patient selection will improve the outcomes. It is also our belief that better selection criteria would favor renal artery stenting if the same trials were repeated with more restrictive entry criteria. Our current criteria for selecting patients for renal artery stenting are a need for four or more antihypertensive medications, including diuretics and nitrates, diastolic blood pressure over 90 mm Hg, or the use of clonidine. For each additional criterion from among these three criteria, the response rate for renal artery stenting improves dramatically. For those patients who are being stented for renal insufficiency, the rate of decline in preoperative egfr over time is the sole predictor of improved renal function after stenting. Our data suggest that these criteria are superior to the presence of symptoms in predicting outcomes for renal artery stenting. I agree that renal resistive indexes are an important adjunct in patient selection. Because many of our patients are referred in with a duplex scan in hand, we often do not repeat the duplex. We tend to escalate the diagnostic workup to a more invasive study, such as a CT angiogram, or proceed to the operating room for an arteriogram. Unfortunately, the majority of the patients in this study did not have renal resistive indexes measured, but I would certainly advocate the routine use of resistive indexes in most practices. Dr Matthew Corriere (Atlanta, Ga). Your findings are very interesting, particularly the observation related to clonidine. Do you think the number of antihypertensive medications in your patient cohort truly reflects severity of hypertension, and do you have a protocol approach to antihypertensive agent management? Alternative reasons for antihypertensive agent use in this patient population would include risk factor reduction, such as routine use of ACE inhibitors or angiotensin receptor blockers, particularly in diabetic patients. Similarly, do you routinely utilize perioperative

420 Modrall et al JOURNAL OF VASCULAR SURGERY February 2012 beta blockade in these patients, and how did you account for combination medications? Taken together, it may be that number of antihypertensive medications may indicate severity of hypertension for some patients, but may also be indicative of appropriate pharmacologic risk factor reduction for many others. Dr Modrall. I think you bring up an important question. This is whether the number of antihypertensive medications a marker for better medical management and therefore better outcomes in the long-term? We have always interpreted that variable as an indication of the severity of hypertension. How you would distinguish the two explanations would be difficult. Our internists try to follow a standard protocol in the use of antihypertensive medications, but I cannot tell you how consistent that protocol is followed. CME Credit Now Available to JVS Readers Readers can now obtain CME credits by reading selected articles and correctly answering multiple choice questions on the Journal website (www.jvascsurg.org). Four articles are identified in the Table of Contents of each issue and 2 questions for each are posted on the website. After correctly answering the 8 questions, readers will be awarded 2 hours of Category I CME credit.