Renal Artery Stenosis is Not Associated with the Development of Acute Renal Failure Following Coronary Artery Bypass Grafting

Abstract

Background: Acute renal failure (ARF) is a frequent complication of coronary artery bypass grafting (CABG) surgery and is strongly associated with perioperative morbidity and mortality. We hypothesized that renal artery stenosis (RAS), causing occult renal ischemia, may be an important factor contributing to development of ARF after CABG surgery. Methods: Preoperative and intraoperative data on 798 consecutive adult patients undergoing CABG surgery with cardiopulmonary bypass from February 1, 1995 to February 1, 1997 (who had also undergone an abdominal aortogram for the evaluation of RAS) were recorded and entered into a computerized database. The development of ARF was defined as a rise in serum creatinine of 1 mg/dL (88.4 µmol/L) above baseline postoperatively. The association between the presence of renal artery stenosis together with preoperative and intraoperative variables and the development of ARF was assessed by multivariate logistic regression. Results: A total of 798 patients underwent isolated coronary bypass grafting, of which 18.7% demonstrated 50% or more RAS. ARF developed in 82 patients (10.2%), of which three (0.3%) required dialysis support. The mortality for patients who developed ARF was 14% (OR 15, P = 0.0001) compared to 0.2% among those who did not develop ARF. The presence of renal artery stenosis of any severity ranging from unilateral 50% RAS to bilateral 95% RAS was not associated with the subsequent development of ARF. Conclusions: The development of ARF following CABG surgery is associated with high mortality. The presence of RAS does not appear to increase the risk for developing ARF.

• Cardiac surgery
• Renal failure
• Hemodialysis
• Renal artery stenosis

Introduction

Acute renal failure (ARF) remains a frequent and serious complication of coronary artery bypass grafting (CABG) surgery. The incidence of ARF following cardiac surgery has been reported to vary between 1 and 30%.[1-4] When renal failure develops following cardiac surgery, it is associated with increased mortality, hospitalstay, and cost. Despite improvements in the results of cardiac surgery, there has been a trend toward operating on higher-risk patients, which inevitably leads to increased morbidity and mortality. The etiology of acute renal failure following cardiac surgery is poorly understood, but it is believed that ischemic injury of the kidney, resulting from inadequate perfusion, is a major factor. Chertow et al.lsqb;5rsqb; speculated that ARF following cardiac surgery was strongly associated with two major factors: 1) occult renal ischemia (associated with poor cardiac performance, fixed atherosclerotic disease of the renal arteries, and prolonged hypoxemia) and 2) reduced renal functional reserve. Previous analysis from our institution has demonstrated a relationship between factors associated with decreased renal perfusion, including the presence of congestive heart failure and decreased ejection fraction and the presence of systemic arterial disease (as manifested by the presence of a carotid artery bruit), and the development of ARF post-CABG.[6]

Recently, there has been increased interest in performing an abdominal aortogram at the time of coronary angiography to assess for the presence of renal vascular disease. At the same time, developments in surgical and endoluminal technologies have dramatically improved the technical success of treatment of renal artery stenosis.[7&8]

We hypothesized that ARF following CABG was related to renal artery stenosis, and that the association would become stronger with increasing severity of stenosis (i.e., from no RAS to bilateral RAS > 95%).

Subjects and Methods

Between February 1, 1995, to February 1, 1997, 2,672 adult patients underwent cardiac surgery with total cardiopulmonary bypass in Duke University Medical Center. Of these, 798 patients underwent an abdominal aortogram at the time of their cardiac catheterization, and all of these patients were included in this analysis. The decision with regard to which patients underwent aortograms at catheterization was at the discretion of the operator, with no selection bias. ARF was defined by a rise in serum creatinine (Cr) from baseline of 1 mg/dL (88.4 µmol/L) or greater following CABG.

Demographic Data

Demographic data, medical history, physical findings, and blood chemistries were prospectively entered into a computerized medical information system prior to cardiac catheterization. Peripheral vascular disease was defined as a history of claudication, previous vascular procedure, or physical exam evidence of femoral or abdominal bruits. Symptoms of congestive heart failure were classified in accordance with the New York Heart Association criteria. Chronic obstructive lung disease was defined as functional disability or hospitalization or that requiring chronic bronchodilator therapy. Patients were considered to have diabetes if they were taking insulin or oral hypoglycemic agents at the time of cardiac surgery. The left ventricular ejection fraction was assessed preoperatively by either contrast ventriculography, radionucleotide ventriculography, or echocardiography. Serum creatinine concentration was assessed within 48 h of cardiac surgery. Operative mortality was defined as death during the index hospitalization.

Assessment of Renal Artery Stenosis

After coronary angiography and left ventriculography, the pigtail catheter was withdrawn into the abdominal aorta and positioned a few centimeters superior to the renal arteries. Aortography was performed in the anterior–posterior projection with nonionic contrast, power injected at a rate of 20 mL/min to a total volume of 30–40 mL. The injection was recorded on 35 mm cine film at 30 frames per second for later analysis. Renal artery stenosis was considered to be present if one or more renal arteries had 50% or more luminal narrowing on abdominal aortography. In order to fully explore the relationship between the development of ARF post-CABG and degrees of stenosis, six different levels of severity of RAS were defined: unilateral ≥ 50% RAS, bilateral ≥ 50% RAS, unilateral ≥ 75% RAS, bilateral ≥ 75% RAS, unilateral ≥ 95% RAS, and bilateral ≥ 95% RAS.

Operative Procedure

Patients were premedicated with oral diazepam or lorazepam and methadone hydrochloride 90 min before anesthesia induction. Anesthesia was induced with 75–100 µg/kg midazolam and 5–10 µg/kg fentanyl intravenously. Anesthesia maintenance consisted of 0.8 µg/kg/min of midazolam and 0.08 µg/kg/min of fentanyl by continuous infusion. Vecuronium was given as required to maintain complete neuromuscular blockade.

The perfusion apparatus consisted of a Cobe CML oxygenator (Cobe Laboratories, Lakewood, CO), a Sarns 7000 MDX pump (Sarns Inc., Ann Arbor, MI), and a Pall SP 3840 arterial line filter (Pall Biomedical Products Co., Glencove, NY). Nonpulsatile perfusion was used throughout the study, with perfusion maintained between 2 and 2.8 L/min/m². The pumps were primed with crystalloid (a sanguinous) solution formulated to achieve a hematocrit of 18% or more during extracorporeal circulation.Packed red blood cells were added to achieve the desired hematocrit and as required by the clinical circumstance. Measurements of oxygen saturation and hemoglobin were performed with an IL482 co-oximeter (Instrumentation Laboratories, Lexington, MA). α-Stat blood gas management was used during CPB, with PaCO₂ maintained at 35–40 mm Hg and PaO₂ maintained at 150–250 mm Hg.

Statistical Analysis

The association between variables measured at baseline and intraoperative variables and the development of ARF was assessed by logistic regression. Variables measured at baseline included presence or absence of RAS of any severity, age, sex, race, presence of a carotid artery bruit, presence of congestive heart failure, presence of cardiogenic shock, history of previous cerebrovascular accident, presence of diabetes, left ventricular ejection fraction, peripheral arterial disease, serum creatinine, body weight, and presence of COPD. In addition, the relationship between the following intraoperative variables and the development of ARF was assessed: length of time on bypass, number of coronary artery grafts performed, the use of intra-aortic balloon pump to assist in separation off bypass, the lowest in-flow temperature, and the lowest nasal temperature.

Variables that were significantly associated (at the p < 0.1 level of significance) with the development of ARF were also included in a multivariate logistic model. Backward variable selection was used to serially remove nonsignificant factors, until only significant (p < 0.05) factors remained in the model.

Model cross-validation was also performed using a 100 sample, bootstrap analysis. Means are listed as ± 1 standard deviation, and odds ratios (OR) are expressed together with their 95% confidence intervals (CI) and associated p values. The difference in mean peak postoperative serum creatinine between patients with and without RAS was tested using the student's t-test.

Results

The study population consisted of 798 patients; 150 (18.7%) of those 798 had documented RAS (Table 1). Nopatients fell into the category of unilateral 95% RAS, and this category was thus omitted. Acute renal failure developed in 82 patients (10.2%), and three patients (0.3%) required dialysis support. The mortality of patients who developed ARF was 14% (OR = 15.3, P = 0.0001) compared to 0.2% among those who did not develop ARF.

Table 1. Prevalence of RAS in Study Population.

	N	%
Less than 50% RAS	648	81.2
Unilateral 50% RAS or greater	150	18.7
Bilateral 50% RAS or greater	35	3.6
Unilateral 75% RAS or greater	82	8.4
Bilateral 75% RAS or greater	18	1.8
Bilateral 95% RAS or greater	9	0.9

The baseline characteristics for study subjects are displayed in Table 2, according to whether or not they developed ARF. Preoperative variables that were significantly associated with the development of ARF included black race, presence of a carotid bruit, presence of congestive heart failure, history of a previous cerebrovascular event, diabetes, decreased left ventricular ejection fraction, and increased preoperative serum creatinine. The risk of developing acute renal failure for subjects with normal renal function preoperatively (Cr < 1.4 mg/dL) was 8.1%, and that among patients with moderate renal insufficiency with a serum Cr of between 2.5 and 3.5 mg/dL was 37%. Intraoperative variables that were significantly associated with the development of ARF included the length of time on bypass and insertion of an intra-aortic balloon pump to assist with separation from bypass.

Table 2. Baseline and Intraoperative Variables Associated with the Development of Acute Renal Failure Following CABG.

	No renal failure	Renal failure	OR (95% CI)	P-value
	(N = 716)	(n = 82)
Agea	65 ± 10.5	66.2 ± 11.03	1.08 (0.9–1.1)	0.5
Sex % (male/female)	65.5%/34.5%	66%/34%	1.01 (0.7–1.2)	0.3
Race (white/black)	89%/11%	80%/20%	1.8 (1.8–2.3)	0.008
Carotid bruit	14.4%	31%	2.7 (2.1–3.7)	0.0001
Congestive heart failure	29%	40%	1.6 (1.3–2.1)	0.03
Cardiogenic shock	0.4%	2.4%	5.9 (2.3–14.5)	0.06
Previous CVA	12.6%	23%	2.0 (1.6–2.7)	0.01
Diabetes	27%	40%	1.8 (1.4–2.2)	0.0001
Ejection fraction	55.2 ± 13.8	51.5 ± 12.4	0.8 (0.7–0.9)	0.02
Peripheral arterial disease	14.6%	21%	1.6 (1.2–2.1)	0.2
Weight (kg)a	80.3 ± 16.5	83.1 ± 15.3	1.1 (1.0–1.2)	0.1
COPD	14%	15%	1.1 (0.8–1.6)	0.7
Serum creatinine	1.1 ± 0.5	1.5 ± 0.7	2.5 (2.0–3.0)	0.0001
Bypass timea	119.3 ± 46.1	124.4 ± 52.3	1.07 (1.05–1.08)	0.0001
Number of grafts	3.0 ± 0.9	2.9 ± 0.9	0.8 (0.7 –1.0)	0.2
IABP inserted in OR	1%	4.9%	5.0 (2.6–9.5)	0.01

[OR = Odds ratio ± 95% confidence interval.]

^aOdds ratio expressed for a 10-unit change.

When both preoperative and intraoperative variables (Table 3) were included in a multivariate model, preoperative serum creatinine (OR = 2.5, P = 0.0001), duration of bypass (OR = 1.1, P = 0.0001), presence of a carotid bruit (OR = 3.0, P = 0.0001), and increased body weight (OR = 1.2, P = 0.02) were each independently associated with the probability of developing a 1 mg/dL rise in serum creatinine postoperatively.

Table 3. Multivariate Analysis of Risk Factors Associated with the Development of Acute Renal Failure Following CABG.

	Odds ratio (95% CI)	P-value	Bootstrap
Preoperative serum creatinine (mg/dL)	2.5 (2.1–3.2)	0.0001	100
Bypass time (minutes)a	1.1 (1.09–1.14)	0.0001	100
Carotid artery bruit	3.0 (2.2–3.9)	0.0001	95
Weight (kg)a	1.2 (1.09–1.3)	0.02	67

Results derived from bootstrap procedure display the variables most frequently significantly associated with the development of acute renal failure in patients undergoing CABG.

^aOdds ratio expressed for a 10-unit change.

Model cross-validation using a 100-sample, 798-observation bootstrap analysis was performed. The frequency that these variables were significantly associated with the development of ARF is outlined in Table 2. It can be seen that preoperative serum creatinine, length of time on bypass, and presence of a carotid artery bruit were almost always significantly associated with the risk of developing ARF. Increased body weight was associated with the development of ARF in 67% of cases.

The presence of renal artery stenosis of any severity was not associated with either the development of ARF or the mean peak postoperative serum creatinine (Table 4).

Table 4. Probability of Developing ARF and Mean Peak Change in Serum Creatinine (ΔCr in mg/dL) Based on Different Degrees of RAS.

Degree of RAS	N	% Showing ARF	Mean ΔCr	P-value
Unilateral 50% RAS	150	12%	0.437 ± 0.61	0.09
< 50% RAS	820	7.8%	0.52 ± 0.76
Bilateral 50%	35	11.4%	0.48 ± 0.64	0.5
< 50% bilateral RAS	935	8.6%	0.48 ± 45
Unilateral 75% RAS	82	12%	0.5 ± 0.4	0.3
< 75% bilateral RAS	887	7.4%	0.4 ± 0.6
Bilateral 75% RAS	18	11%	0.52 ± 0.41	0.6
< 75% bilateral RAS	952	8.4%	0.44 ± 0.64
Bilateral 95%	9	11	0.54 ± 1.3	0.7
< 95 bilateral RAS	961	8.4%	0.44 ± 0.60

P-value is for chi-square test comparing probability of developing ARF with or without different degrees of RAS.

The number of patients in this data set developing ARF requiring dialysis (n = 3) was too small to develop amodel associating preoperative and intraoperative variables with need for dialysis.

Discussion

In this study, we clearly demonstrated that the presence of asymptomatic RAS does not increase the risk of developing acute renal failure after CABG surgery. We also confirmed our previous findings of the frequent association of RAS among patients with CAD.

The diagnosis of renal artery stenosis was based on a plain abdominal aortogram at the time of coronary angiography to assess for the presence of renal vascular disease, and selective angiography was not used. While plain abdominal aortogram is not the gold standard, it is certainly a reasonable estimation for assessment of renal artery stenosis and is a standard used in many studies of this type.

Because of the limitation of relatively few outcome events imposed on our analysis, we used the bootstrapping technique to further explore the association of the independent variables with the development of acute renal failure and the development of dialysis requiring acute renal failure. Bootstrapping is a methodology used in regression to adjust for the effect of outliers, and it decreases the effect of variability in a given sample. This is done through the creation of multiple subsamples by random selection of subjects from the original sample. The selected subject is immediately returned to the original sample and, thus, can be selected more than once in a given subsample (i.e., sampling with replacement). The size of each subsample is the same as the original (798 in our case), and therefore, some subjects will be selected into a given subsample more than once and others not at all. Repeated samples are then used to examine the relationship between the variables of interest. We generated 100 bootstrap samples to which a multivariable regression analysis was performed using all independent variables (Table 3) with the development of ARF.

We confirmed our previous findings of the importance of preoperative serum creatinine, length of time on bypass, presence of carotid artery disease, and body weight in this data set.[9] The development of acute renal failure following CABG increases the odds of death 20-fold.[10] Recently, there has been a trend toward surgical repair or angioplasty of renal artery disease prior to CABG in an effort to reduce the probability of developing ARF, although there are no data in the literature to support this intervention. Indeed, although the technical success of surgical repair and angioplasty has improved the treatment of renal artery stenosis, there is a significant risk in carrying out these procedures, with mortality rates of 2–9% and a worsening of renal function seen in approximately 30% of patients.[7&8] It should also be noted that the patient groups most likely to develop renal impairment after CABG, in particular, those patients with impaired renal function and extensive peripheral vascular disease, are those who are least likely to benefit from RAS repair.[4], [11] Therefore, this data would not support the prophylactic repair of RAS prior to CABG to reduce the risk of developing ARF post-CABG.

Some recent studies have suggested that the process causing renal dysfunction in patients with RAS may not be solely due to renal artery narrowing, and that other processes such as hypertension and atheroemboli may have an important role.[12&13] As our knowledge of the progression of RAS expands, we may develop a further understanding of the role of RAS in postoperative renal impairment.[14]

In summary, we have demonstrated no association between the presence of RAS, even severe RAS, and the risk of developing ARF after CABG.