Acute Kidney Injury after Cardiac Surgery: Prediction, Prevention, and Management

Accurate prediction of cardiac surgery–associated AKI requires a consistent definition of the AKI outcome that can be compared between studies. Historical definitions used for cardiac surgery–associated AKI have ranged from subtle increases in serum creatinine to the need for new postoperative renal replacement therapy. The Society of Thoracic Surgeons defines AKI as a 3-fold increase in serum creatinine, creatinine more than 4 mg/dL, or initiation of dialysis after cardiac surgery. Eventually, clinicians and researchers began to adopt more uniform definitions of AKI, with the creation of the Risk, Injury, Failure, Loss, and End-stage renal failure (RIFLE) and Acute Kidney Injury Network (AKIN) classifications. Both classifications use elevations in serum creatinine during different periods of time (7 days in RIFLE vs. 48 h in AKIN) and decreases in urine output to categorize AKI into stages. In 2012, the Kidney Disease: Improving Global Outcomes (KDIGO)¹⁷  became the newest classification for AKI, integrating components of both RIFLE and AKIN criteria, and including rises in serum creatinine for both 48 h and 7 days after cardiac surgery. Table 1 reviews the elements of the RIFLE, AKIN, and KDIGO criteria. During the past 10 plus years, the KDIGO criteria have become an established standard for defining cardiac surgery–associated AKI both in clinical practice and in clinical research. However, there are some discrepancies in how the KDIGO criteria are used in cardiac surgery studies. Some of these studies diagnosed AKI using serum creatinine increases alone, while others used both the serum creatinine criteria and the KDIGO criteria for oliguria.¹⁸  These discrepancies exist because some investigators find urine output to be unreliable in the setting of diuretic use.

A meta-analysis that included 32 observational studies of cardiac surgery patients published between 2006 and 2014 reported a median pooled incidence for AKI of 22.1%.¹⁹  The incidence of stage 1, 2, and 3 AKI defined by either RIFLE, AKIN, or KDIGO criteria among 19 studies that reported these data was 17.9%, 4.4%, and 3.5%, respectively. Another meta-analysis of 91 observational studies of cardiac surgery patients published between 2004 and 2014 reported a similar pooled incidence rate of AKI of 22.3%.³  The pooled AKI rate in this study was lower in patients with coronary artery bypass graft (CABG) surgery (19.0%) compared to those who underwent valve surgery (27.5%) or aortic surgery (29.0%). A large epidemiologic study used to create the Mehta score reported that type of surgery is an independent predictor of dialysis-dependent AKI after cardiac surgery with combined CABG-mitral valve surgery (odds ratio 2.57) and mitral valve surgery (odds ratio 2.01) having the highest risk.²⁰  AKI was also associated with significantly higher length of stay in the intensive care unit (ICU; 5.4 days vs. 2.2 days) and hospital (15.0 days vs. 10.5 days) when compared to patients without AKI.²¹ 

Acute kidney injury–associated in-hospital and long-term (1 to 5 yr) mortality in the latter meta-analysis was 10.7% and 30.0%, respectively.³  Duration of cardiac surgery–associated AKI has been found to predict mortality in a number of published studies. One meta-analysis of nine observational studies of cardiac surgery patients reported that patients who recovered renal function before hospital discharge had a significantly lower long-term mortality risk compared to patients discharged with persistent abnormal renal function.²¹  However, even patients who recovered their renal function after developing AKI in this study had a significantly higher mortality than those who did not develop AKI. A 90-day follow-up of participants of the Randomized Evaluation of Normal versus Augmented Level Replacement Therapy (RENAL) Study, which randomly assigned critically ill patients (including cardiac surgery patients) between receiving higher and lower levels of renal replacement therapy dose intensity, reported an overall mortality of 62.3% among all patients with AKI requiring renal replacement.²²  This was confirmed in a retrospective study of cardiac surgery patients that reported a mortality of 58.6% in patients requiring new postoperative renal replacement therapy.²³ 

The pathogenesis of cardiac surgery–associated AKI is complex, resulting from a multitude of potential perioperative insults to the kidney. The commonality of these insults is lower oxygen delivery to the kidneys relative to the kidneys’ oxygen demand, resulting in renal tubular injury. The hypothesized mechanisms that contribute to cardiac surgery–associated AKI include hypoperfusion, atheroembolic events, exposure to nephrotoxins, inflammation, and oxidative stress (fig. 1).²⁴ 

The kidneys receive 20% of the total cardiac output; however, intrarenal oxygen availability requires an orchestrated balance between supply (i.e., blood flow) and demand, which is determined by basal metabolism and tubule-glomerular functions. For example, the renal glomeruli extract only 10 to 20% of the oxygen delivered to them and maintain a low oxygen tension at baseline,²⁵  which makes them particularly vulnerable to hypoxia during periods of hypoperfusion. Renal hypoperfusion can occur throughout the perioperative period due to hypotension, decreased cardiac output, sympathetic stimulation, the administration of vasoconstrictive medications, and activation of the renin-angiotensin-aldosterone system. Each of these events can interfere with renal autoregulation and reduce glomerular filtration rate.²⁶ 

Cardiopulmonary bypass is associated with nonpulsatile flow, altered hemodynamics, decreased oxygen delivery, inflammation, and oxidative stress, each of which can contribute to AKI.²⁶  Renal perfusion while on cardiopulmonary bypass is directly proportional to mean arterial pressure (MAP) and pump flow rate, suggesting that normal autoregulatory mechanisms may be impaired during this period.²⁷  Lannemyer et al.²⁸  found that cardiopulmonary bypass induces renal vasoconstriction and redistribution of blood from the kidneys in the setting of unchanged glomerular filtration rate and renal oxygen consumption. This oxygen supply–demand mismatch may explain why the duration of time,²⁹  degree of hemodilution,^30,31  and hypotension³²  on cardiopulmonary bypass have all been linked to postoperative AKI. Atherosclerotic disease is also prevalent in patients undergoing cardiac surgery and elements of the surgical procedure are associated with atheroembolic events, including placement of intra-aortic balloon pumps, manipulation of the left atrium, and both clamping and declamping of the aorta. One autopsy-based study found that, among patients who died in the hospital after cardiac surgery, 10.4% had atheroembolic findings in the kidneys.³³  Rewarming from cardiopulmonary bypass provides a period of time when the kidney is susceptible to hypoxic injury. During this time, oxygen consumption increases with temperature in the renal medulla and can exceed available supply.³⁴ 

Perioperative medications associated with nephrotoxicity include antibiotics, angiotensin-converting enzyme inhibitors, angiotensin receptor blockers, nonsteroidal anti-inflammatory drugs (NSAIDs), diuretics, and intravascular contrast agents (table 2). Commonly used antibiotics, including aminoglycosides, β-lactams, and vancomycin, mediate AKI through immune and nonimmune mechanisms. Aminoglycosides accumulate in proximal tubular cells and cause apoptosis and acute tubular necrosis.³⁵  β-Lactam antibiotics can cause a hypersensitivity reaction resulting in acute interstitial nephritis and can also mediate crystal development and obstructive injury to the tubules.³⁵  Vancomycin is thought to cause AKI by promoting free radical formation and oxidative stress resulting in acute tubular necrosis, causing a hypersensitivity reaction resulting in acute interstitial nephritis, and through intratubular crystal obstruction. Among these antibiotics, KDIGO guidelines recommend avoiding aminoglycosides unless a suitable alternative is unavailable.³⁶  Intravascular contrast agents are associated with acute tubular necrosis due to their cytotoxic effect on tubular cells and by mediating renal vasoconstriction. Because this association is based on observational studies, there is controversy regarding the true incidence of contrast-associated AKI and effective interventions in high-risk patients. Even studies utilizing propensity matching to control for non–contrast-associated etiologies of AKI reported disparate results with one study finding a higher incidence of contrast-associated AKI in patients with impaired baseline kidney function³⁷  and one study finding no association between contrast exposure and AKI regardless of baseline kidney function.³⁸  Observational studies of cardiac surgical patients suggest that exposure to intravascular iodinated contrast agents less than 7 days before surgery may result in a higher incidence of postoperative AKI.^39–41  Until data from randomized clinical trials are available, the decision to postpone cardiac surgery after contrast exposure should be based on an individual evaluation of the patient and other risk factors for postoperative AKI.⁴² 

Although angiotensin-converting enzyme inhibitors and angiotensin receptor blockers are associated with improved renal blood flow, their preoperative administration is associated with an increased risk of AKI in patients undergoing cardiac surgery. The mechanism for this injury may be functional renal insufficiency, or hypotension resulting in decreased renal perfusion pressure.⁴³  For this reason, the KDIGO “bundle of care” recommends discontinuing these medications for 48 h after cardiac surgery.³⁶  Diuretics, including the loop diuretic furosemide and mannitol, have not been found to reduce cardiac surgery–associated AKI, and furosemide may actually increase the risk of AKI by inducing hypovolemia in some patients.¹⁵ 

Cardiac surgery patients are also susceptible to pre-renal and postrenal causes of AKI throughout the perioperative period. Aggressive preoperative diuresis, hemofiltration during cardiopulmonary bypass, hemorrhage, and vasoplegia can contribute to an absolute or relative hypovolemia and impaired renal perfusion. Postrenal AKI can result from urinary catheter obstruction (e.g., from blood clots or kinking of the catheter tubing) or from urinary retention after the catheter is removed in the postoperative period. A thorough review of historical details (e.g., diuretic administration, prostatic hyperplasia), physical examination (e.g., flushing the urinary catheter, bladder ultrasonography), and laboratory tests (e.g., fractional excretion of sodium, fractional excretion of urea) are necessary to distinguish between pre-renal, postrenal, and intrinsically renal etiologies of AKI after cardiac surgery.

Identifying patients with modifiable risk factors for cardiac surgery–associated AKI may facilitate interventions to mitigate their risk throughout the perioperative period. Numerous articles have assessed clinical and surgical risk factors for cardiac surgery–associated AKI, but, to date, the variables identified for risk prediction are most accurate for predicting stage 3 AKI, which is renal failure associated with a 3-fold increase in serum creatinine or requiring renal replacement therapy. These models include the Cleveland Clinic score,⁴⁴  the Mehta score,²⁰  and the Simplified Renal Index.⁴⁵  Each of these uses preoperative risk factors to predict the risk of severe cardiac surgery–associated AKI, which are listed in table 3. The Cleveland Clinic score has been validated in a cohort of 12,096 to have the best discriminatory value (area under the receiver operating characteristics curve [AUC]: 0.86) to predict the need for renal replacement therapy after cardiac surgery compared to the Mehta score and the Simplified Renal Index (AUC: 0.81 and 0.79, respectively).⁴⁶  Despite this, it has not been widely implemented in clinical practice. Research is being done to improve prediction of cardiac surgery–associated AKI using biomarkers of renal function and renal injury, because predictive models have integrated clinical variables with these biomarkers⁴⁷  and incorporated machine learning algorithms⁴⁸  to improve clinical risk prediction for cardiac surgery–associated AKI. The greatest limitation of these newer machine learning models is their use of data obtained at static intervals to predict an outcome that is impacted by the multimodal dynamic data during the perioperative period. Future models that incorporate these multimodal dynamic data (e.g., perioperative hemodynamics and oxygenation variables, among others) and key biomarkers of kidney health obtained at various time points may be able to better predict all stages of AKI with greater accuracy and within actionable time points to intervene.

In terms of identifying demographic and preoperative clinical characteristics that predict overall AKI after cardiac surgery, multiple parameters have been identified from observational studies, although not always consistently. Preoperative risk factors frequently identified include advanced age,^49,50  female sex,⁴⁹  higher body mass index,^49,51  proteinuria⁵²  and the presence of systemic comorbidities including hypertension,⁴⁹  diabetes,^49,50,53  chronic kidney disease,^42,49,50,53  chronic obstructive pulmonary disease,⁴⁹  left ventricular dysfunction,^49,53  and perioperative anemia.⁴⁹  Patients with chronic kidney disease have lower kidney reserve; hence, during cardiac surgery and cardiopulmonary bypass, the kidney can be overwhelmed by surgical stressors, inflammation, and oxidative stress, resulting in further decline in glomerular filtration rate. Husain-Syed et al.⁵⁴  studied patients who underwent CABG surgery at a single center and calculated their preoperative renal functional reserve using a high oral protein load. The authors compared patients who had an in-hospital postoperative increase in serum creatinine and returned to baseline to a group who did not experience a rise in postoperative creatinine. The patients who experienced a transient increase in postoperative serum creatinine that returned to preoperative baseline concentration were subsequently found to have significantly lower renal functional reserve when compared to patients who did not have any rise in serum creatinine after surgery. This illustrates that AKI leaves patients with reduced long-term function and less renal functional margin in the setting of additional subsequent AKI events. Hypertension and diabetes are highly prevalent in patients undergoing cardiac surgery. Chronic hypertension can lead to autoregulatory vasoconstriction of the preglomerular vasculature, which can impair autoregulatory mechanisms and make the kidneys vulnerable to injury when the perfusion pressure is inadequate.⁵⁵  Diabetes is associated with vasculopathy and nephropathy resulting from tissue hypoxia, inflammation, endothelial damage, and other mechanisms. Proteinuria is a marker of tubulointerstitial damage, and the resorption of filtered protein in the proximal tubules is associated with upregulation of inflammatory mediators. One study found that preoperative proteinuria was an independent predictor of cardiac surgery–associated AKI requiring renal replacement therapy by AKIN criteria.⁵² 

Urgent surgery, reoperative surgery, and the preoperative placement of an intra-aortic balloon pump are also associated with cardiac surgery–associated AKI diagnosed using RIFLE criteria. Compared to patients undergoing CABG surgery, those undergoing isolated valve or aortic surgery have a higher incidence of cardiac surgery–associated AKI.^49,53 

Intra-operative and early postoperative risk factors include cardiopulmonary bypass duration,^42,49  anemia while on cardiopulmonary bypass, blood transfusion,⁴⁹  and development of early postoperative proteinuria.⁵⁶  While prolonged cardiopulmonary bypass time has been associated with cardiac surgery–associated AKI, large trials comparing outcomes after on-pump and off-pump CABG surgery have failed to show a reduction in cardiac surgery–associated AKI among patients who underwent off-pump procedures.^57,58  Postoperative tissue hypoperfusion defined by elevated serum lactate levels⁵⁹  and the use of inotropes^59,60  have been associated with cardiac surgery–associated AKI. Additionally, re-exploration after surgery is a strong, independent predictor of cardiac surgery–associated AKI.^49,61 

Current diagnostic criteria for cardiac surgery–associated AKI utilize increases in serum creatinine and decreases in urine output. While creatinine provides a good approximation of glomerular filtration rate when kidney function is normal, its accuracy is diminished in non–steady state conditions such as the perioperative period. The increased volume of distribution in postoperative cardiac surgery patients or in other critically ill populations who are exposed to aggressive fluid resuscitation⁶²  can lead to dilution of serum creatinine, delaying the diagnosis and underestimating the degree of kidney injury.⁶³  The inclusion of other AKI-related biomarkers may enable earlier detection and better prediction of cardiac surgery–associated AKI (greater sensitivity and/or specificity) than with clinical and surgical risk factors alone (fig. 2).⁴⁷  This may allow discovery of mitigating therapies to prevent the incidence and further progression of AKI. While a number of blood and urine biomarkers have been assessed for prediction of cardiac surgery–associated AKI, many of these perform inconsistently between subpopulations of cardiac surgery patients and only one is approved for clinical use by the U.S. Food and Drug Administration. Additionally, evaluating biomarkers of kidney injury using functional biomarkers such as serum creatinine as clinical endpoints poses a challenge. Biomarker discovery and development is a vast area of research that is beyond the scope of this review, but we include a brief summary of some of the most relevant AKI biomarkers to date. These biomarkers are briefly reviewed in table 4.

• Neutrophil gelatinase–associated lipocalin is part of the innate immune system and functions by binding to bacterial iron-binding proteins. It is a sensitive and specific marker of injury to the kidneys and can be detected in the blood and urine after both nephrotoxic and ischemic injuries. While increases in blood and urine neutrophil gelatinase–associated lipocalin concentration have been associated with cardiac surgery–associated AKI in children,⁶⁴  a meta-analysis of observational studies found that it had a much lower predictive value in adults undergoing cardiac surgery (AUC 0.83 in adults vs. 0.89 in children) and those with chronic kidney disease (AUC 0.81 in patients with chronic kidney disease vs. 0.87 in patients without chronic kidney disease).⁶⁵ 

• Cystatin C is a low molecular weight protein that undergoes complete reabsorption by the renal tubules. Unlike creatinine which is produced by muscle cells and undergoes tubular filtration, cystatin C is produced by all nucleated cells and undergoes glomerular filtration. It has therefore been investigated as a functional biomarker for kidney clearance. A post hoc analysis of a randomized controlled trial of cardiac surgery patients found that an early postoperative rise in serum cystatin C was associated with AKI requiring renal replacement therapy.⁶⁶  A meta-analysis evaluating the diagnostic accuracy of cystatin C in predicting AKI in adults in a variety of settings also found that it demonstrated high sensitivity (0.82) and specificity (0.82), with an AUC of 0.89.⁶⁷ 

• Urinary interleukin-18 is specific to the renal tubules and is thought to mediate ischemic and inflammatory injury to the kidneys. Urinary levels of interleukin-18 were found to be significantly elevated in patients with cardiac surgery–associated AKI, as early as 4 to 6 h after cardiopulmonary bypass.⁶⁶  In cardiac surgery patients, one study found that patients with the highest concentration of urinary interleukin-18 (5th quintile) had a nearly 7-fold incidence of postoperative AKI. The addition of interleukin-18 to a clinical prediction model that included perioperative variables improved its receiver-operating characteristic curve (AUC) from 0.69 to 0.76.⁴⁷ 

• Kidney Injury Molecule-1 is a transmembrane protein with two extracellular domains that separate from the cell surface and enter the urine after kidney injury.⁶⁸  The urinary concentration of kidney injury molecule-1 peaks as early as 3 h after kidney injury, making it an earlier marker of cardiac surgery–associated AKI compared to serum creatinine.⁶⁹  One study evaluating biomarkers in post–cardiac surgery patients found that the combination of kidney injury molecule-1 and interleukin-18 had excellent predictive value (AUC 0.93) for stage 3 AKIN and mortality.⁷⁰ 

• Tissue inhibitor of metalloproteinase and insulin-like growth factor-binding protein are inducers of cell cycle arrest in renal tubular cells, which occurs in the very early phases of cellular injury. Urinary concentrations of these biomarkers were found to have a superior sensitivity and specificity in predicting KDIGO stage 2 to 3 AKI when compared to other novel biomarkers.⁷¹  The product of the urinary concentrations of these biomarkers has been approved by the U.S. Food and Drug Administration for early clinical AKI prediction.

• C-C motif chemokine ligand 14 is a chemokine that is released from tubular epithelial cells in response to injury. It mediates the renal inflammatory response that results from AKI by binding to receptors on monocytes and T cells and promoting chemotaxis in these cells.⁷²  An observational study in cardiac surgery patients found that elevated serum C-C motif chemokine ligand 14 was predictive of KDIGO stage 3 AKI (AUC 0.93). Because C-C motif chemokine ligand 14 is a biomarker of kidney disease progression, it has also been investigated as a predictor of chronic kidney disease. One study of ICU patients found that urinary C-C motif chemokine ligand 14 levels were associated with renal nonrecovery.⁷³ 

This section highlights results from some key areas that clinical trials and studies have focused on to date for the purposes of preventing or mitigating cardiac surgery–associated AKI.

A study that included two separate randomized controlled trials assigned patients undergoing CABG surgery to rewarming from 32° to 34°C versus 32° to 37°C in one trial (n = 233) and a strategy of mild sustained hypothermia (34°C) versus sustained normothermia (37°C) without rewarming in the second trial (n = 267).⁷⁴  These two randomized trials found that rewarming from 32° to 37°C in a 10- to 15-min period resulted in an increased incidence of AKI compared to rewarming to 34°C; however, that sustained mild hypothermia did not have a nephroprotective effect. Newland et al.⁷⁵  analyzed a large multicenter registry of cardiac surgery patients using propensity scoring to adjust for confounders and found that the duration of hyperthermic perfusion (greater than 37°C) was an independent predictor of increased cardiac surgery–associated AKI.

Clinical trials have recently sought to determine the impact of a goal-directed oxygen delivery while on cardiopulmonary bypass on cardiac surgery–associated AKI. The Goal-Directed Perfusion Trial (GIFT) evaluated maintaining oxygen delivery greater than 280 ml·min^–1·m^–2versus conventional perfusion on the rate of AKI in cardiac surgery patients.⁷⁶  In this multicenter randomized controlled trial that enrolled 350 patients from nine European institutions, patients in the intervention group had a lower incidence of AKIN stage 1 AKI while no difference in the incidence of AKIN stage 2 or 3 AKI. A similar study that randomly assigned 300 patients undergoing cardiac surgery between targeting an oxygen delivery greater than 300 ml·min^–1·m^–2versus conventional perfusion while on cardiopulmonary bypass found a higher rate of AKI (defined by KDIGO) in the conventional perfusion group.⁷⁷  A subgroup analysis of the study found that the goal-directed oxygen delivery strategy was superior in patients with lower hematocrits and lower body surface area. The results of these two trials suggest the need to individualize oxygen delivery while on cardiopulmonary bypass rather than use conventional body surface area–derived perfusion targets.

Vasoplegia is common in cardiac surgery patients.⁷⁸  Because endogenous vasopressin is thought to be reduced after cardiac surgery and because vasopressin preferentially binds receptors in the efferent arterioles to potentially increase glomerular filtration rate, vasopressin use should be considered before using norepinephrine to treat post–cardiopulmonary bypass vasoplegia.¹⁵  The Vasopressin versus Norepinephrine in Patients with Vasoplegic Shock after Cardiac Surgery (VANCS) trial randomly assigned 300 patients who experienced post–cardiopulmonary bypass vasoplegia to receive vasopressin versus norepinephrine as a first-line agent.⁷⁹  Patients who received vasopressin as a first-line agent had a significantly lower incidence of moderate to severe AKI and a lower mortality.

Cardiac surgery is associated with potentially turbulent hemodynamics related to anesthetics, surgical manipulation of the heart and great vessels, comorbidities including cardiovascular and coronary artery disease, nonpulsatile flow on cardiopulmonary bypass, anemia, vasoplegia, and pump failure. Systemic hypotension due to any of these etiologies for a prolonged period can lead to impaired renal perfusion, renal tubular ischemia, and a consequent decrease in glomerular filtration rate. While evidence of an association between intraoperative hypotension and AKI has been established in the noncardiac surgery population,⁸⁰  there is controversy regarding this association in cardiac surgery patients. Several studies have attempted to evaluate the impact of intraoperative hypotension on cardiac surgery–associated AKI. One observational study of 6,523 patients reported an association between a MAP less than 65 for 10 min or more after cardiopulmonary bypass and the need for new postoperative renal replacement therapy.⁸¹  Another study of 4,984 patients found an association between each 10-min period of hypotension throughout the cardiac surgical period and a composite outcome of stroke, AKI, or death after cardiac surgery.⁸²  Two randomized controlled trials^83,84  found that targeting higher MAPs on cardiopulmonary bypass did not reduce the incidence of cardiac surgery–associated AKI, while another clinical trial⁸⁵  showed that, in the higher MAP group, there was a significantly larger number of patients who doubled their postoperative creatinine. These findings suggest the need for further investigation on the role of intraoperative blood pressure optimization as a modality to avoid cardiac surgery–associated AKI.

Both anemia^86,87  and blood transfusion⁸⁸  have been linked to cardiac surgery–associated AKI in multiple cohort studies. However, a large retrospective study found that combined exposure to anemia and blood transfusion increased the risk of post–cardiac surgery AKI more than either exposure alone.⁸⁹  The Transfusion Requirements in Cardiac Surgery (TRICS) III study, a multicenter randomized controlled trial, found no difference in the incidence of new-onset AKI requiring renal replacement therapy in patients who were transfused to a restrictive (7.5 g/dL) versus a liberal transfusion (9.5 g/dL) threshold.⁹⁰  A prespecified substudy of this trial also found no difference in the incidence of milder forms of AKI between patients randomly assigned to restrictive versus liberal transfusion thresholds.⁹¹  Avoiding anemia and transfusion during the cardiac surgery perioperative period may necessitate multimodal preoperative anemia management with oral iron therapy, erythropoietin administration in patients with iron deficiency anemia, and supplementation with vitamin B₁₂ and folate for B₁₂ and folate deficiency anemia, respectively.⁹² 

Hemolysis occurs frequently in cardiac surgery because of the shear stress applied on the red blood cells, which is invoked by blood flowing through the cardiopulmonary bypass circuit and the use of venting and suction devices. Erythrocyte washing and the transfusion of cell-salvaged blood are additional sources of hemolysis products during cardiac surgery.^93,94  Injured erythrocytes release cell-free hemoglobin and other heme-derived products, which normally bind to the hemoglobin and heme scavengers, haptoglobin and hemopexin, respectively, and then cleared by splenic and liver macrophages.⁹⁵  However, if these scavenging systems are overwhelmed during prolonged or excessive hemolysis, plasma levels of cell-free hemoglobin and heme increase, and their clearance relies on renal filtration. Renal cell exposure to cell-free hemoglobin and other hemolysis products results in increased oxidative stress, inflammation, and reduced bioavailability of NO in the kidneys, all of which may exacerbate renal ischemia-reperfusion injury and endothelial dysfunction, thus promoting tubular renal cell death.^96–100 

Observational studies in patients undergoing cardiac surgery with cardiopulmonary bypass report that elevated levels of cell-free hemoglobin, as well as low levels of its scavenger, haptoglobin, are associated with increased cardiac surgery–associated AKI and increased mortality.^101–106  Furthermore, it has been demonstrated that elevated levels of catalytic iron, another by-product of hemolysis, are associated with increased AKI and mortality, both after cardiac surgery and in other critically ill patient groups.^107,108 

Extracorporeal cytokine removal to decrease the cardiopulmonary bypass–related inflammatory response via hemoadsorption has been attempted for some time. The technology is based on highly porous, biocompatible nonpolar polymer beads that bind and remove inflammatory mediators through size exclusion and nonspecific surface adsorption. The technology of polymer bead-based cytokine hemoadsorption has shown to rapidly eliminate cytokines in vitro and in vivo.^109,110  The Cytosorb hemoadsorption filter device (Cytosorbents Corporation, Princeton, New Jersey) has been designed to directly capture and reduce mid molecular weight inflammatory mediators (approximately 10 to 60 kDa). The substances are adsorbed due to physiochemical binding depending on their concentration. Several studies have examined the ability of the Cytosorb filter to remove cell-free hemoglobin and other hemolysis products during cardiopulmonary bypass, but the results were not promising, showing a very limited ability to remove cell-free hemoglobin.^111,112  Furthermore, even if effective, it is important to remember that the use of hemoadsorption requires patients to be on extracorporeal circulation such as during cardiopulmonary bypass, extracorporeal membrane oxygenation, or renal replacement therapy.¹¹³ 

The KDIGO Clinical Practice Guidelines includes recommendations for the prevention and treatment of AKI.¹¹⁴  This section discusses the implementation of these recommendations in cardiac surgery as a “Bundle of Care.”

Cardiac output and stroke volume monitoring throughout the intraoperative and postoperative period is necessary to evaluate and optimize kidney perfusion. The PrevAKI single-center trial evaluated the impact of adherence to a bundle of interventions recommended by KDIGO to reduce AKI in high-risk patients undergoing cardiac surgery and reported a significant reduction in all stages of AKI in the KDIGO bundled care group as compared with the control group of routine care delivered in the post–cardiac surgery ICU.¹¹⁵  Patients in the intervention group underwent monitoring with a Pulse index Continuous Cardiac Output monitor that provided stroke volume variation and cardiac index to guide the administration of fluids and inotropes, respectively. The multicenter PrevAKI trial, conducted by the same group of investigators, also utilized functional hemodynamic monitoring and optimized MAP and cardiac index in their intervention group.¹¹⁶ 

Perioperative strategies for preventing cardiac surgery–associated AKI aim to optimize kidney perfusion—primarily by maintaining blood pressure and cardiac output within physiologic limits. The multicenter PrevAKI study reported that cardiac surgery patients at high risk for AKI who were randomly assigned to receive the KDIGO bundle of interventions had a significantly lower incidence of post–cardiac surgery stage 2 or stage 3 AKI (14% vs. 24%). However, there were no significant between-group differences in the multicenter study when stage 1 AKI was also assessed. This bundle included an algorithmic approach to hemodynamic optimization to maintain MAP > 65 and cardiac index > 2.5 l·min^–1·m^–2, resulting in intervention group patients receiving significantly more fluids and being more likely to receive dobutamine.¹¹⁶ 

Changes in autoregulation due to comorbidities and medications such as angiotensin-converting enzyme inhibitors and angiotensin receptor blockers may potentiate renal hypoperfusion even when the systemic blood pressure is within the normal range. Patients with chronic hypertension, for example, have their autoregulatory curves shifted rightward, resulting in higher MAP requirement to ensure adequate renal perfusion. Both the single-center and multicenter PrevAKI trials discontinued angiotensin-converting enzyme inhibitors and angiotensin receptor blockers during the first 48 h after surgery as part of the KDIGO “bundle of care.”^115,116  However, a systematic review of six studies (three randomized controlled trials and three prospective cohort studies) evaluating the discontinuation of these agents before coronary angiography and cardiac surgery found no evidence that drug cessation impacted the incidence of AKI, citing a low quality of Grading of Recommendations, Assessment, Development and Evaluation (GRADE) evidence.¹¹⁷ 

Dobutamine has been shown to reduce cardiac surgery–associated AKI when implemented as part of the KDIGO “bundle of care” in both the single-center and the multistudy Prev-AKI studies, suggesting that it should be used as part of a goal-directed strategy to maintain kidney perfusion in the postoperative ICU. Significantly more patients in the goal-directed management groups received dobutamine in both the single-center (31% vs. 9%) and multicenter (32.6% vs. 18.8%) trials, resulting in higher MAPs at various time points.^115,116 

Compared to 0.9% saline, balanced crystalloid solutions such as Ringer’s lactate have been associated with a lower risk of AKI in critically ill patients, although the effect size was low in one meta-analysis of six studies (relative risk of 0.96).¹¹⁸  This may be due to the effect of chloride-mediated vasoconstrictions of the kidneys as well as the detrimental effect of chloride-associated metabolic acidosis.¹¹⁹  The Acute Disease Quality Initiative international consensus guidelines recommend resuscitation with a balanced crystalloid solution rather than 0.9% saline.¹¹⁶  A meta-analysis evaluating the role of urinary alkalinization with sodium bicarbonate to prevent cardiac surgery–associated AKI found no benefit in the general cardiac surgery population.¹¹⁷  However, there was a significantly lower incidence of renal replacement therapy in patients undergoing elective CABG surgery. Further studies are needed before the routine use of sodium bicarbonate can be recommended. The Albumin in Cardiac Surgery trial randomly assigned cardiac surgery patients to receive Ringer’s acetate solution or 4% albumin both in the cardiopulmonary bypass prime and used for intraoperative fluid resuscitation and found no significant difference in the incidence of AKI between the two groups.¹²⁰  However, the higher incidence of bleeding and infection seen in patients who received albumin may limit its use in cardiac surgery. Hydroxyethyl starch should be avoided throughout the perioperative period because of their association with AKI in critically ill patients,¹²¹  including cardiac surgery patients.¹²² 

The indications, clinical timing of initiation, and different modalities of renal replacement therapy have not been specifically studied in the context of cardiac surgery–associated AKI. In the general critically ill patient population, indications of renal replacement therapy are focused on solute control (e.g., electrolyte or acid-base imbalances) and volume control to restore euvolemia when patients are no longer diuretic-responsive.^123,124  For timing, data from recently completed clinical trials suggest that “early” initiation of renal replacement therapy in the absence of emergent indications of solute/volume control does not significantly improve mortality outcomes and is associated with more episodes of hypotension and hypophosphatemia.¹²⁵  Further, in secondary outcome analysis of the Standard versus Accelerated Initiation of Renal Replacement Therapy in Acute Kidney Injury (STARRT-AKI) trial, survivors of AKI-renal replacement therapy had higher risk of being renal replacement therapy-dependent at 90 days when compared to patients randomly assigned to the “delayed” renal replacement therapy initiation strategy.¹²⁶  Therefore, the current consensus recommendation is that renal replacement therapy should be provided for specific solute or volume control indications that are refractory to medical management, which applies to patients with cardiac surgery–associated AKI. In this context, it is critical to carefully monitor the perioperative clinical trajectory of patients at risk for or with cardiac surgery–associated AKI to evaluate the need of renal replacement therapy.¹²⁷  When renal replacement therapy is indicated, both intermittent and continuous renal replacement therapy are viable options for patients with cardiac surgery–associated AKI. Typically, renal replacement therapy is reserved for hemodynamically unstable patients given its lower net ultrafiltration rate, lower rate of osmotic shift, and slower change in extracellular fluid electrolyte concentrations with less impact on resting membrane potentials when compared to intermittent hemodialysis.^123,124 

Noninvasive urine oximetry is a novel concept for intraoperative and postoperative ICU use in cardiac surgery patients to assess the adequacy of kidney perfusion and oxygen delivery. This can be potentially done by continuously measuring urine oxygen partial pressure. When measured from urine as it is excreted, urine oxygen partial pressure is thought to approximate the partial pressure of oxygen in the renal medulla.¹²⁸  For this reason, urine oxygen partial pressure measurement is being investigated in cardiac surgery patients as a real-time monitor of oxygen delivery to the kidneys. Zhu et al.¹²⁹  used fiberoptic probes attached to the tip of urinary catheters to measure the partial pressure of urine in the bladders of patients undergoing cardiac surgery. The authors reported that low urine oxygen partial pressure predicted postoperative AKI and both the duration of low partial pressure as well as the nadir value increased the odds of AKI. Another study that measured urine oxygen partial pressure using a prototype device placed between the urinary catheter and collection bag in cardiac surgery found that mean oxygen partial pressure in the post–cardiopulmonary bypass period was associated with postoperative AKI.¹³⁰  These studies suggest that future studies and technical modifications may ultimately result in clinically reliable real-time monitoring of kidney oxygen delivery, allowing early interventions to mitigate risk of cardiac surgery–associated AKI.

Animal studies in models of hemolysis-associated AKI have shown that administration of haptoglobin prevented circulating free hemoglobin-induced kidney injury.^131–133  In humans, a propensity-matched retrospective study reported that intraoperative administration of human haptoglobin (to bind and remove excessive cell-free hemoglobin) to patients exhibiting macrohemoglobinuria during cardiopulmonary bypass, was associated with a reduced incidence of cardiac surgery–associated AKI.¹³⁴  Because haptoglobin is the natural scavenger of hemoglobin in the body, this approach seems clinically promising; however, human haptoglobin is only available for use in Japan and is not available for clinical use in other countries.

One of the consequences of intravascular hemolysis is depletion in plasma NO,^135,136  which may result in vasoconstriction and impaired tissue perfusion.¹⁰⁶  Cell-free hemoglobin is released in the form of ferrous oxyhemoglobin and quickly reacts with plasma NO to form ferric methemoglobin, hence reducing the level of plasma NO.¹³⁷  A second process contributing to reduced NO bioavailability during hemolysis is the increase in the level of the enzyme arginase-1, which is released from hemolyzed erythrocytes.¹³⁸  This enzyme metabolizes l-arginine, the main precursor of NO, therefore decreasing the available substrate for NO production.¹³⁷  Finally, injured erythrocytes also release asymmetric dimethylarginine, which directly inhibits endothelial NO synthase.¹³⁹  Several studies have explored the effects of intraoperative administration of NO on postoperative outcomes in cardiac surgery patients. In a prospective study of 244 adult patients undergoing multiple valve surgeries, Lei et al.¹⁰⁰  demonstrated that administration of NO (intraoperatively during cardiopulmonary bypass [80 ppm via the cardiopulmonary bypass circuit] and postoperatively for 24 h) compared to controls who received nitrogen, was associated with a significant reduction in cardiac surgery–associated AKI and major adverse kidney events up to 1 yr after surgery. Similarly, Kamenshchikov et al.,¹⁴⁰  reported that intraoperative administration of NO (40 ppm via the cardiopulmonary bypass circuit) was associated with a significant decrease in the incidence of cardiac surgery–associated AKI. Interestingly, there was no difference in the concentration of cell-free hemoglobin between treated patients and the controls in the latter study. Please note that using NO to prevent AKI is still investigational and not an approved indication.

Acetaminophen has anti-inflammatory and antioxidative properties, specifically reducing the oxidation of cell-free hemoglobin by preventing the conversion of iron from the ferric ion to its more inflammatory and nephrotoxic ferryl ion form.^141,142  In an animal model of rhabdomyolysis-induced kidney injury, acetaminophen significantly decreased markers of lipid peroxidation and preserved kidney function.¹⁴²  Several observational retrospective studies have reported that exposure to acetaminophen was associated with decreased incidence of cardiac surgery–associated AKI in both adults and children undergoing cardiac surgery with cardiopulmonary bypass.^143,144  Furthermore, Janz et al.¹⁴⁵  reported a phase IIa randomized placebo-controlled trial of acetaminophen in patients with severe sepsis and elevated circulating cell-free hemoglobin. Patients who received acetaminophen had lower levels of markers of lipid peroxidation as well as lower serum creatinine. In addition, two small clinical trials in adults and children undergoing cardiac surgery demonstrated that acetaminophen decreased markers of lipid peroxidation in the plasma after cardiopulmonary bypass. Although no differences in the incidence of AKI between the acetaminophen and placebo groups were reported, those studies were limited by small sample sizes and low event rates.^146,147  Taken together, it seems that acetaminophen may have a kidney-protective effect in cardiac surgery; however, further studies, including appropriately powered prospective trials, are needed to fully understand the role of acetaminophen as a kidney-protective medication.

Published guidelines by the Society of Thoracic Surgeons, Society of Cardiovascular Anesthesiologists, American Society for Extracorporeal Technology,¹⁵  and a practice update by the Society of Cardiovascular Anesthesiologists AKI Working group¹⁶  summarize the available literature and provide recommendations for the management of cardiac surgery–associated AKI. The Society of Thoracic Surgeons, Society of Cardiovascular Anesthesiologists, American Society for Extracorporeal Technology guidelines graded the available evidence using the American College of Cardiology and American Heart Association Recommendation System. The Society of Cardiovascular Anesthesiologists practice update assessed published randomized controlled trials using the GRADE methodology.

The Society of Thoracic Surgeons, Society of Cardiovascular Anesthesiologists, American Society for Extracorporeal Technology guidelines recommended avoidance of hyperthermic perfusion on cardiopulmonary bypass and recommended goal-directed oxygen delivery on cardiopulmonary bypass as a Class I recommendations-evidence of benefit¹⁵  to prevent cardiac surgery–associated AKI. Adopting the KDIGO “bundle of care” for high-risk postoperative patients is also recommended to reduce the risk of cardiac surgery–associated AKI (Class IIa recommendation: should be considered). The use of minimally invasive extracorporeal circulation techniques was recommended as a Class IIb recommendation: may be considered. The guidelines recommend the use of fenoldopam in patients who tolerate the drug without hypotension (Class IIb recommendation: may be considered), but use of this medication has not been widely accepted in clinical settings. The guidelines recommend against the perioperative use of dopamine and mannitol, citing evidence that neither drug has shown to protect against the development of cardiac surgery–associated AKI (Class III recommendation: not recommended).

The Society of Cardiovascular Anesthesiologists Practice Update endorses the use of goal-directed oxygen delivery and the KDIGO “bundle of care” in high-risk patients, citing moderate level of GRADE evidence.¹⁶  This bundle, described in greater detail previously, includes volume and hemodynamic optimization, and avoidance of nephrotoxins and hypoglycemia. They also recommend the use of vasopressin over norepinephrine for treatment of vasoplegic shock after cardiac surgery, citing a low level of GRADE evidence. The practice update also reported that targeting higher MAP goals during cardiopulmonary bypass, and the perioperative use of dopamine and dexmedetomidine did not reduce cardiac surgery–associated AKI, citing a low level of GRADE evidence.

Acute kidney injury is common after cardiac surgery and is associated with worse outcomes than patients who do not develop post–cardiac surgery AKI. Although the pathogenesis of cardiac surgery–associated AKI is complex and multifactorial, the common consequence is renal tubular injury with a decline in glomerular filtration rate. The risk factors for cardiac surgery–associated AKI have been well characterized and used to develop predictive models. These models have performed well in predicting the need for renal replacement therapy but have less discriminatory ability to predict milder forms of AKI that are also associated with worse outcomes. Novel biomarkers of kidney function, inflammation, and injury may allow earlier detection of cardiac surgery–associated AKI, better prediction models for developing cardiac surgery–associated AKI and to enable clinical trialists to study potential preventive and treatment strategies in the patients who are truly at risk for developing cardiac surgery–associated AKI. Preventing AKI requires the identification of patients with modifiable risk factors and mitigating this risk using evidence-based interventions throughout the perioperative period. Discontinuing nephrotoxic medications, optimizing chronic medication conditions, and treating anemia are recommended before elective cardiac surgery. During the intraoperative and postoperative phases, utilizing balanced crystalloids for resuscitation and using a goal-directed hemodynamic strategy to optimize cardiac output and kidney perfusion pressure have been shown to improve kidney outcomes. Research-driven advances in monitoring, real-time data capture, and subphenotyping analyses, as well as discovery of novel detection and prediction biomarkers of kidney function and injury may ultimately allow real-time detection of kidney injury in the operating room or in the early postoperative period.

This work was supported by National Institutes of Health, National Heart, Lung, and Blood Institute (Bethesda, Maryland) grant R01HL148448 (to Dr. Fox) and National Institute of Diabetes and Digestive and Kidney Diseases (Bethesda, Maryland) grants R01DK128208, R01DK133539, U01DK129989, and P30 DK079337 (to Dr. Neyra).

Dr. Raphael has consulted for Octapharma USA, Inc. (Paramus, New Jersey). Dr. Neyra has received consulting honoraria from Baxter (Deerfield, Illinois), Outset (San Jose, California), Vifor (Zurich, Switzerland), and AcelRx (San Mateo, California). The other authors declare no competing interests.