Abstract
Abstract. Acute renal failure (ARF) requiring dialysis occurs in up to 4% of patients after cardiopulmonary bypass (CPB). CPB leads to the generation of intravascular free hemoglobin, resulting in increased endothelial and renal tubular cell free iron, which is associated with renal injury. Conversely, renoprotection is conferred by processes that upregulate heme and iron sequestration pathways, such as ferritin. This study evaluates the influence of free hemoglobin generation during CPB and the capacity to sequester free iron on the occurrence of post-CPB renal insufficiency. Thirty consecutive patients undergoing CPB were enrolled in the study. Serum creatinine, free hemoglobin, and ferritin were measured preoperatively, at the end of bypass, and 24 and 48 h after surgery. Renal injury, as determined by an increase in the serum creatinine of ≥25% (ARF) by 48 h after surgery, occurred in 40% (12 of 30) of patients, and dialysis was necessary in 6.6% (2 of 30). Free hemoglobin levels increased in all patients but did not correlate with postoperative ARF. However, patients with preoperative serum ferritin levels ≤130 μg/L, the median value for the group, had a sixfold greater likelihood of developing ARF compared to patients with levels above this value (P = 0.03). Lower serum ferritin levels appear to be associated with the development of ARF. Serum ferritin levels may signify intravascular as well as endothelial and renal epithelial cell ability to bind free iron generated during CPB-induced hemolysis, and thus may help provide information regarding the risk for ARF.
Acute renal insufficiency or renal failure (ARF) complicates up to 40% of cardiopulmonary bypass (CPB) procedures with 1 to 4% of patients requiring dialysis (1,2,3). ARF has recently been independently associated with postoperative mortality (4). Multiple patient characteristics (age, diabetes mellitus, renal function) and surgical factors (type of surgery, clamp time, bypass time) have been associated with post-CPB ARF but none has accurately forecast its occurrence (2,5, 6, 7, 8, 9, 10, 11, 12). In addition, the precise etiology of CPB-associated ARF is not known. The byproducts of traumatic intravascular hemolysis, such as free hemoglobin, may be possible contributors to the development of ARF (13). Free hemoglobin may precipitate within the tubular lumen causing obstruction; directly injure renal epithelial cells through iron-catalyzed free radical oxidation of cellular membranes; and induce renal vasoconstriction (13, 14, 15, 16, 17). Experimental models of heme-related injury have shown that the capacity to sequester free iron conveys a renoprotective effect. Desferoxamine administration (iron chelation) or augmentation of the heme oxygenase-ferritin system (iron sequestration) before a large release of heme-containing pigments appears to ameliorate functional and structural renal damage, while an inhibition of these sequestration pathways results in more pronounced injury (18, 19, 20, 21, 22).
The generation of free hemoglobin during CPB may result in increased delivery of toxic free iron to the renal endothelium and tubular epithelium providing a major pathway for the induction of post-CPB ARF. We therefore evaluated the relationship between the generation of free hemoglobin, the capacity to bind free iron, and the development of ARF. Additionally, the ability to detect subtle tubular injury by measuring the urinary excretion of proximal (N-acetyl-β-D-glucosaminidase [NAG] and α-glutathione S-transferase [α-GST]) and distal (π-glutathione S-transferase [π-GST]) tubular enzymes was evaluated. NAG (molecular weight 140,000) is a lysosomal enzyme found in tubular cells (23). GST (molecular weight 25,000) is a cytosolic enzyme found in the kidney and other organs where it detoxifies compounds by catalyzing their conjunction with glutathione and promoting renal excretion. There are several isoforms of GST within the kidney; α-GST is located in the proximal tubule and π-GST is found in the distal convoluted tubule, loop of Henle, and collecting duct (24).
Materials and Methods
Patients and Operative Management
Thirty consecutive patients undergoing CPB were evaluated after giving informed consent. All patients had been electively scheduled for coronary artery bypass or valvular replacement by one cardiothoracic surgeon. Aprotinin use during surgery was reserved for patients with an increased risk for bleeding. The use of cardiac suction, which is traumatic to red blood cells, was minimized, and the target hematocrit during bypass was 25%. Blood products (platelets, coagulation factors, red blood cells) were administered as needed to stop excessive bleeding and to increase oxygen-carrying capacity. Mannitol (12.5 g) was used in the CPB prime. CPB was performed using a Medtronic centrifugal pump (Biomedicus, Minneapolis, MN), which provided nonpulsatile flow averaging 2.5 L/m2 per min. Cardioactive drugs were administered at the discretion of the attending anesthesiologist to maintain systolic BP or as inotropic support for weaning from CPB. Ultrafiltration was performed using a cellulose diacetate filter as needed for volume control (25).
Postoperative care was managed by the attending cardiovascular surgeon. Red blood cells were transfused for a hematocrit of <20% or as needed to increase oxygen-carrying capacity. Platelets were infused for postoperative bleeding if the thromboelastograph pattern indicated platelet abnormalities. Pressor support was started, maintained, or increased as needed to keep the mean arterial pressure above 60 mmHg with a good cardiac index. Prophylactic dopamine and diltiazem were not used (3).
Clinical Data
Prospectively collected data included: history of renal disease, diabetes, prior myocardial infarction, claudication or cerebral vascular event, use of antihypertensive medications, number of days between cardiac catheterization and surgery, and the type of surgery performed. The intraoperative factors recorded included: duration of surgery, length of CPB, clamp time, mean CPB arterial pressure, mean CPB flow/m2, urine output, oxygen extraction ratio during bypass, volume of cell-saved red blood cells reinfused, volume of red blood cells transfused, use of ultrafiltration, aprotinin use, placement of a balloon pump, and use of pressors. Postoperative factors monitored during the first 48 h included: cardiac index, number of hypotensive episodes (15 min of BP <90/60), temperature, milrinone use, and number of transfusions.
Laboratory Data
Blood was sampled for the measurement of serum creatinine, free hemoglobin (normal range, 0 to 5 mg/dl), and ferritin (normal range, 15 to 160 μg/L) at the induction of anesthesia, just before the onset of CPB, at the conclusion of CPB, and 24 and 48 h after surgery. Serum albumin was obtained preoperatively. Timed urine samples were obtained before bypass, at the end of bypass, and 24 and 48 h postoperatively, for the measurement of urinary albumin to creatinine ratios and the urinary excretion of NAG, α-GST, and π-GST. Urinary samples for NAG and GST were collected and immediately stored at -70°C until analysis.
Urinary NAG was measured by the production of p-nitrophenol determined colorimetrically (Boehringer Mannheim Biochemicals, Indianapolis, IN). Urinary GST was measured by an enzyme-linked immunosorbent assay (Biotrin, Cedar Knolls, NJ). All samples were processed concurrently. Urinary albumin was measured by an immunoassay and determined by nephelometry (Behring, San Jose, CA). Urinary albumin and creatinine, and serum creatinine, ferritin, and hemoglobin levels were processed immediately in the clinical laboratory.
Statistical Analyses
ARF was defined as a 25% or greater increase in the serum creatinine over baseline. t test and χ2 or Fisher exact test were used as appropriate to evaluate differences in continuous and categorical variables, respectively, between patients who developed ARF after cardiac surgery and those who did not. Continuous variables without approximate normal distribution were analyzed with nonparametric tests (Wilcoxon rank sum). Logistic regression was used to evaluate the independent association of individual covariates with the development of ARF. The primary covariate of interest was the preoperative ferritin value. Ferritin was explored as a continuous, dichotomous (with the median ferritin selected as the cutoff: ≤130 or >130 μg/L) and categorical (ferritin quartiles) covariate. A limited multivariate analysis was performed (although the number of events did not meet strict criteria for this analysis) to evaluate the possibility of confounding by age and the presence of diabetes. A secondary analysis of the relationship between urinary albumin excretion (albumin:creatinine ratio) and the change in creatinine from preoperative to 48 h postoperatively was performed using linear regression.
Results
Acute renal insufficiency developed in 40% (12 of 30) of patients as determined by an increase in the serum creatinine of ≥25% by 48 h after surgery. Patient demographic characteristics are presented in Table 1. There were no significant differences between patients who developed ARF and those who did not. Specifically, there was no difference between the groups in age, gender, presence of diabetes, antihypertensive medication use, or length of time between cardiac catheterization and surgery. The 48-h postoperative serum creatinine in patients with ARF was 1.6 ± 0.66 mg/dl (preoperative creatinine, 0.89 ± 0.05 mg/dl) compared to 0.95 ± 0.23 mg/dl (preoperative creatinine 0.95 ± 0.07) in those without ARF (C, control) (P < 0.001). Two patients (6.6%) required dialysis. The levels of free hemoglobin increased in all patients during CPB and returned to within the normal range by 24 h after surgery (Table 2). The hematocrit declined to similar levels in both groups postoperatively, from 40 ± 4.9 to 27.1 ± 2.9% among ARF patients and from 40 ± 5.5 to 28.1 ± 4.1% among control subjects.
Characteristics of patients with and without postoperative renal failurea
Plasma-free hemoglobin before, immediately after, and 24 h after CPBa
Association of Preoperative Factors with ARF
The distribution of ferritin levels before CPB for the entire patient cohort is shown in Figure 1. The mean ± SD and median ferritin levels for patients without ARF were 196 ± 137 and 177 μg/L, respectively, and for patients with ARF they were 133 ± 138 and 68 μg/L, but the differences were not statistically significant. However, there was a greater proportion of patients with ARF who had ferritin levels in the lower two quartiles compared to those without ARF, and no patient with ARF had a ferritin level in the third quartile (Table 3). Additionally, patients with a serum ferritin below the median (130 μg/L) had a sixfold greater likelihood of developing ARF compared to patients with a ferritin higher than the median (P = 0.03) (Table 4). This difference in odds of ARF between the two groups held after adjusting for diabetes and age (P = 0.04). Furthermore, patients with the lowest ferritin levels (<100 μg/L) had a 9.9-fold greater odds of developing ARF compared to patients with a ferritin level above this value (P = 0.009).
The distribution of preoperative serum ferritin levels in patients undergoing cardiopulmonary bypass (n = 30).
Comparison of proportions of patients with and without postoperative ARF in ferritin quartile levels before cardiopulmonary bypassa
Odds of developing ARF after CPB
Ferritin explored as a continuous or categorical (quartiles) covariate in the logistic regression analysis did not significantly influence the odds of developing ARF. This was most likely due to the relatively small sample size, which precluded finding differences between categories containing relatively few data points. Additionally, no correlation between preoperative ferritin and serum albumin levels was found (r = -0.06, P 0.8), making nutritional deficiency an unlikely explanation for these findings. Although previous studies have shown an association between elevated serum ferritin levels and the occurrence of coronary artery disease, recent data from large population-based studies and our current study do not support a specific range of ferritin values with the presence of cardiac disease (26, 27, 28). Instead, ferritin levels in patients with coronary artery disease show a distribution comparable with the general population. There was likewise no correlation between urinary albumin excretion and the change in creatinine from baseline to 48 h postoperatively (Table 5).
Urinary albumin : creatinine ratio, NAG, and GST excretion before, immediately after, and 24 and 48 h after cardiopulmonary bypassa
Association of Intra- and Postoperative Factors with the Development of ARF
Intra- and postoperative factors that did not appear to be associated with postoperative renal impairment are shown in Table 1. Additional factors not associated with ARF included: number of transfusions, use of the cell saver, urine output during surgery, mean arterial pressure during bypass, bypass flow rate, pressor use during or after surgery, milrinone administration, placement of a balloon pump, use of aprotinin, intraoperative hemofiltration, and the postoperative cardiac index. Likewise, the number of postoperative hypotensive events was similar in both groups within the first 12 postoperative hours, and there was no difference between the groups in the occurrence of cardiac dysrhythmias. Urinary NAG and GST excretion increased in most patients after CPB; increased excretion, however, was not associated with the development of clinically apparent ARF (Table 5, Figure 2).
The excretion of α- and π-glutathione S-transferase (GST) expressed as ng/mg of creatinine before bypass, after bypass, and 24 and 48 h after surgery.
Discussion
Previous investigators have concentrated on demographic characteristics and surgical parameters when attempting to develop predictors of the development of post-CPB ARF. These factors do not predict acute renal failure based on the mechanism(s) of injury but concentrate on nonspecific markers that may identify patients at risk. The current study evaluates the ability of a serum marker of a possible renoprotective mechanism to determine the risk for postoperative ARF. Ferritin was chosen for study because of its ability to sequester iron, a nephrotoxin released from free hemoglobin produced during CPB (29). Ferritin has a large capacity to sequester iron, each molecule binding 4500 molecules of free iron (30). Animal studies evaluating the nephrotoxic effect of infusing free hemoglobin or the generation of other heme-containing compounds, such as myoglobin during rhabdomyolysis, have shown decreased renal injury if free iron-sequestering mechanisms are upregulated (18,31).
Free hemoglobin and free iron are generated during CPB secondary to mechanical trauma to red cells within the bypass system and suction devices (29). Suction devices are used to salvage blood from the operative field, which is then reinfused into the patient. The mechanisms of hemoglobin-mediated renal failure are thought to be intratubular hemoglobin precipitation causing obstruction, iron-facilitated oxidant damage to tubular epithelial cells, and induction of vasoconstriction (13,32, 33, 34). Within the kidney, filtered hemoglobin is sequestered into endosomes; the rate and total amount taken into the cell is a linear function of the concentration within the tubular lumen (32). Once inside the cell, heme oxygenase opens the heme ring releasing iron and carbon monoxide and starts the process of bilirubin production. The iron is then bound by ferritin, carbon monoxide may promote renal vascular dilation, and bilirubin may act as an antioxidant (21,31,35).
ARF caused by insults that increase endothelial and renal tubular epithelial cell free iron (i.e., cisplatin, glycerol-induced rhabdomyolysis, intravascular hemolysis, and hemoglobin infusion) may be partially abrogated by preinjury administration of heme-binding proteins (DFO), induction of endothelial and tubular cell heme oxygenase and ferritin, or blocking the release of sequestered intracellular iron (18, 19, 20, 21,30,31,36,37). Heme oxygenase activity is not normally expressed in the kidney, but is induced along with ferritin within 3 to 6 h of oxidant stress (18,20,21,31,38, 39, 40). In fact, the promoter for heme oxygenase-1 contains a heat shock consensus element, nuclear factor-κB and AP-1 binding sites, indicating induction by inflammatory stimuli, including cytokines, oxidative stress, and heat shock (41, 42, 43, 44, 45). In addition, transfection of endothelial cells with the heme oxygenase gene markedly increases blood vessel formation in conjunction with increased enzymatic activity (46). Thus, heme oxygenase may moderate the severity of cellular injury by decreasing the damage caused by heme and other prooxidant species, and by activating new vessel formation in areas of ischemic damage (31,46,47). The end result may be a major limitation to both direct tubular damage and the impairment of blood flow. Alternatively, increased heme oxygenase activity may also induce other pathways of membrane protection. Heme oxygenase and ferritin in this setting may not be the primary protectants; higher or lower ferritin levels may simply denote the presence or absence of other cytoprotective mechanisms that have been stimulated through classic inflammatory pathways (15,19,48,49).
The present study suggests that serum ferritin levels may be associated with the development of ARF after CPB. The association of ferritin but not plasma-free hemoglobin with renal injury may be explained by the fact that injury is not determined by generation of hemoglobin and thus free iron, but instead by the ability to handle free iron once it has been generated. Serum ferritin levels may indicate the capacity to chelate intravascular heme and thus mitigate oxidant injury to the renal endothelium, which leads to decreased renal blood flow. Furthermore, serum ferritin may be a gauge of intracellular heme-sequestering ability, as it is likely related to the intracellular heme oxygenase activity and ferritin concentration (50). Low intracellular levels may increase the risk of renal tubular injury and endothelial damage from the free iron generated during CPB, whereas higher levels may protect the patient from developing renal failure. The final result would depend on the interaction between this protective mechanism and the degree of other injury sustained by the patient.
Pre-, intra-, and postoperative injuries other than the generation of free iron, although not identified in this study, also likely contribute to the generation of ARF after CPB. This fact and the relatively small sample size may explain the failure to find an inverse linear relationship between the preoperative serum ferritin level and the odds of developing ARF. Furthermore, high serum ferritin levels may denote significant preoperative damage and inflammation that may represent established injury and thus a greater risk of ARF secondary to the effects of CPB. In this setting, the protective abilities of the heme oxygenase system could have already been saturated, affording no reserve for the injuries sustained during CPB. Alternatively, other injuries not influenced by the heme oxygenase pathway could be sustained after CPB. In those instances, serum ferritin levels would not effect the outcome. Thus, at this time, it appears that low serum ferritin levels are the important associated factor for the development of post-CPB ARF. Low serum ferritin levels appear to identify patients at increased risk for the development of post-CPB ARF, perhaps by indicating their lack of protection against one of several pathways of renal injury.
Footnotes
-
American Society of Nephrology
- © 1999 American Society of Nephrology