Association Between Postoperative Acute Kidney Injury and Duration of Cardiopulmonary Bypass: A Meta-Analysis

Association Between Postoperative Acute Kidney Injury and Duration of Cardiopulmonary Bypass: A Meta-Analysis Avinash B. Kumar, MD, FCCP,* Manish Suneja, MD,† Emine O. Bayman, PhD,*‡ Garry D. Weide, DO,* and Michele Tarasi, MD* Objective: This meta-analysis examined the association between cardiopulmonary bypass (CPB) time and acute kidney injury (AKI). Design: Meta-analysis of previously published studies. Setting: Each single-center study was conducted in a surgical intensive care unit and/or academic or university hospital. Participants: Adult patients undergoing heart surgery with CPB. Interventions: A systematic literature review was conducted using PubMed, EMBASE, and Cochrane Library databases and Google Scholar from January 1980 through September 2009. Initial search results were refined to include human subjects, age >18 years, randomized controlled trials, and prospective and retrospective cohort studies, meet the Acute Kidney Injury Network definition of renal failure, and report times on CPB. Measurements and main results: The length of time on CPB has been implicated as an independent risk factor for

C

ARDIAC SURGERY is viewed as one of the great medical advances of the 20th century. The key to the success and advancement of the procedure has been the development of cardiopulmonary bypass (CPB) by extracorporeal circulation. However, the procedure is not without complications. The incidence of acute kidney injury (AKI), formerly known as acute renal failure after CPB (AKI-CPB), varies from 5% to 30% depending on the definition of acute renal failure used and on the postoperative period studied.1,2 Patients with AKI-CPB who require any form of renal replacement therapy (RRT) have an unacceptably high mortality rate that approaches 68%, with most survivors remaining dependent on dialysis.3 The development of AKI-CPB also is associated with a significant increase in infectious complications, an increase in length of hospital stay, and an increase in mortality.4 Multivariate risk analyses from previously published studies have assigned an increased risk of AKI-CPB with advanced age, preexisting renal dysfunction, emergency surgery, low-cardiac-output states, and length of CPB.1 Despite the deleterious impact of AKI-CPB on outcome, its pathophysiology remains incompletely understood. It is plausible that the pathophysiologic changes associated with CPB are accentuated as the duration of CPB increases, which subsequently increases the risk of developing AKI-CPB. The aim of the present meta-analysis was to evaluate the duration of CPB as a risk factor for AKI-CPB.

From the Departments of *Anesthesia and †Nephrology, University of Iowa Hospitals and Clinics, Iowa City, IA; and ‡Department of Biostatistics, College of Public Health, Iowa City, IA. Address reprint requests to Avinash B. Kumar, MD, FCCP, Department of Anesthesia, 200 Hawkins Drive, University of Iowa Hospitals and Clinics, Iowa City, IA 52242. E-mail: [email protected] © 2012 Elsevier Inc. All rights reserved. 1053-0770/2601-0011$36.00/0 doi:10.1053/j.jvca.2011.07.007 64

development of AKI after CPB (AKI-CPB). The 9 independent studies included in the final meta-analysis had 12,466 patients who underwent CPB. Out of these, 756 patients (6.06%) developed AKI-CPB. In 7 of the 9 studies, the mean CPB times were statistically longer in the AKI-CPB cohort compared with the control group (cohort without AKI). The absolute mean differences in CPB time between the 2 groups were 25.65 minutes with the fixed-effects model and 23.18 minutes with the random-effects model. Conclusions: Longer CPB times are associated with a higher risk of developing AKI-CPB, which, in turn, has a significant effect on overall mortality as reported by the individual studies. © 2012 Elsevier Inc. All rights reserved.

KEY WORDS: meta-analysis, systematic review, acute kidney injury, acute renal failure, duration of cardiopulmonary bypass, cardiac surgery, mortality METHODS

Search, Selection, and Validity Assessment Defining Acute Kidney Injury The primary outcome was the development of new-onset AKI-CPB as defined by the Acute Kidney Injury Network (AKIN) criteria.5 Before the use of the AKIN classification, there had been up to 37 different definitions of renal failure in the medical literature, which made these studies very heterogenous and difficult to compare. These definitions varied from a 25% increase over baseline creatinine to the need for dialysis.6 The need for a consensus definition that also could be used to validate outcomes across multiple pathophysiologic diagnoses led to the establishment of the risk/injury/failure/loss/end-stage kidney disease (RIFLE) criteria and the AKIN criteria.1,5,6 The AKIN classification for AKI was used for this meta-analysis because it offers a simple reproducible definition and, unlike the RIFLE criteria, does not include chronic stages of kidney injury. Although the 2 systems have limitations, both have been validated in the setting of AKI-CPB.7 Four unblinded authors (A.K., M.S., M.T., G.W.) conducted a systematic literature review using the PubMed, EMBASE, Cochrane Library databases and Google Scholar from January 1980 through September 2009 for randomized, controlled studies and quasi-randomized trials that listed patients undergoing cardiovascular surgery supported by CPB and reported on acute renal failure. The search terms included “acute renal failure,” “cardiopulmonary bypass,” “acute kidney injury,” and “AKI and heart surgery.” In addition, references from previously published studies and bibliographic data from retrieved articles were included. The search strategy is outlined in Fig 1. In total, 486 articles were identified by the primary snowball search. The initial search results then were filtered to include age ⬎18 years, randomized, controlled trials, and prospective and retrospective cohort studies. Two reviewers independently viewed all citations and any citation considered significant was retrieved. This narrowed the search to 43 studies. These studies also were screened to include only those that had defined renal dysfunction that met the AKIN definition of renal failure6 and had reported times on CPB. This narrowed the present search to 20 studies. These 20 studies were reviewed carefully and another 11 were eliminated (5 studies were excluded for using RRT only as the endpoint for postoperative AKI and/or on closer evaluation

Journal of Cardiothoracic and Vascular Anesthesia, Vol 26, No 1 (February), 2012: pp 64-69

POSTOPERATIVE AKI AND DURATION OF CPB

65

noncardiac surgery, and an “off-pump” technique for surgery were excluded.

Data Abstraction and Study Characteristics

Fig 1. Search strategy for the final studies included in the metaanalysis. AKIN, Acute Kidney Injury Network; CPB, cardiopulmonary bypass; RCT, randomized controlled trial. (Color version of figure is available online.)

used a definition of postoperative renal insufficiency that did not meet AKIN criteria, and 6 studies were eliminated for incomplete data on CPB times). Of the 486 initial search results, 9 studies were selected.8-16 Within the final cohort of 9 filtered studies, patients with preoperative mechanical circulatory support devices, solid organ transplantation, CPB for

Two authors (A.K. and M.S.) independently reviewed the full text of each included study and extracted the following data using a selfdeveloped, standardized data extraction form: (1) general information, including title, lead author, journal, and year of publication; (2) study characteristics, including verification of study eligibility, patient demographics, and study design (prospective or retrospective); (3) the definition of AKI used in the study; (4) the number of patients with preexisting renal dysfunction, emergency surgery, and preoperative circulatory support devices (eg, intra-aortic balloon pump) were defined; (5) results of multivariate and univariate analyses of risk factors from the studies were compiled; and (6) outcome measurements, including overall hospital mortality and need for RRT, were noted. Differences in the data extracted were resolved by discussion among the authors. All data were converted to uniform measurements; thus, serum creatinine is presented as milligrams per deciliter and creatinine clearance as milliliters per minute. Some outcomes reported may differ slightly from the original published reports because data analyses and outcomes definitions were standardized. Once the cohort of studies was finalized, the corresponding authors listed in the original publications were contacted and original anonymous patient data were obtained from 2 of the authors. A summary of the final selected studies is presented in Table 1.

Quality Assessment and Data Analysis of Included Studies The quality assessment of the 9 studies, comprising 12,466 patients, was carried out as outlined by the Meta-Analysis of Observational Studies in Epidemiology (MOOSE) statement.17

Statistical Analysis Study-specific estimates of differences in mean CPB times between the AKI cohort and the no-AKI cohort are provided with their

Table 1. Brief Description of the Final Selected Studies Study

Landoni et

al13

Conlon et al8

Bove et al11 Fischer et al9 Tuttle et al12 Provenchère et al10 Kincaid et al16 Santos et al15 Sirvinskas et al14

Publication Source

Type of Study

Minerva Anestesiol 73:559565, 2007 Nephrol Dial Transplant 14: 1158-1162, 1999

Prospective observational Prospective cohort study

J Cardiothorac Vasc Anesth 18:442-445, 2004 Perfusion 17:401-406, 2002

Observational study

Am J Kidney Dis 41:76-83, 2003 Anesth Analg 96:1258-1264, 2003 Ann Thorac Surg 80:13881393, 2005 Arq Bras Cardiol 83:145154, 2004 Perfusion 23:323-327, 2008

Retrospective chart review Prospective trial Prospective trial Retrospective chart review Prospective cohort study Prospective randomized trial

Definition of AKI

Subjects on CPB

100% increase in sCr or 2 times baseline Cr 1.0-mg/dL increase above baseline sCr or need for dialysis 100% increase in sCr or 2 times baseline sCr 50% decrease in GFR and/or need for dialysis ⬎25% increase in sCr in 48 h

2,255

⬎30% increase in sCr or need for dialysis ⬎25% increase in baseline Cr or Cr ⬎2.0 mg/dL within 72 h ⬎0.5-mg/dL increase in sCr or ⬎1.8-mg/dL in sCr RIFLE criteria

649

2,672

5,068 99 112

1,209 223 179 12,466 total

Abbreviations: AKI, acute kidney injury; CPB, cardiopulmonary bypass; Cr, creatinine; GFR, glomerular filtration rate; RIFLE, risk/injury/failure/ loss/end-stage kidney disease; sCr, serum creatinine.

66

KUMAR ET AL

Table 2. Mean Cardiopulmonary Bypass Times and Number of Patients in the Acute Kidney Injury and Control Groups Study

Subjects With CPB

No-AKI Cohort

Landoni et al,13 2007 Conlon et al,8 1999 Bove et al,11 2004 Fischer et al,9 2002 Tuttle et al,12 2003 Provenchère et al,10 2003 Kincaid et al,16 2005 Santos et al,15 2004 Sirvinskas et al,14 2008 Total

2,255 2,672 5,068 143 120 649 1,209 223 179 12,518

2,187 2,465 4,897 48 61 537 1,168 187 160 11,710

CPB Time (min) ⫾ Time Range

82 ⫾ 28 113 ⫾ 48 81 ⫾ 31.1 107 ⫾ 40 106 ⫾ 44 93 ⫾ 33 130 ⫾ 46 87.4 ⫾ 24.8 100.59 ⫾ 43.9

AKI Cohort

68 207 171 51 51 112 41 36 19 756

CPB Time (min) ⫾ CPB Range

107 ⫾ 46.3 139 ⫾ 81.2 114 ⫾ 51.8 115 ⫾ 41 109 ⫾ 44 114 ⫾ 47 171 ⫾ 55 101.1 ⫾ 25.6 134.74 ⫾ 62.02

Abbreviations: AKI, acute kidney injury; CPB, cardiopulmonary bypass.

95% confidence intervals (CIs). Overall study estimates of differences in mean CPB times are presented by fixed- and randomeffects models. The presence of publication bias is examined by the funnel plot and fail-safe N measurements.18 Statistical analyses were performed in Comprehensive Meta Analysis 2.2.050 (Biostat Inc, Englewood, NJ). RESULTS

CPB Time as Risk Factor for AKI The CPB times for subjects with AKI-CPB and controls in all 9 studies are reported in minutes (mean ⫾ standard deviation) (Table 2). The overall and absolute mean differences of duration of CPB and 95% CIs for each study are reported in Fig 2. In all 9 studies, the patient cohort with AKI reported longer CPB times compared with the cohort that did not develop AKI-CPB (control). The absolute mean differences between the cohorts (AKICPB and control) for individual studies ranged from 3 minutes (95% CI, ⫺13.36 to 19.36) to 41 minutes (95% CI, 26.5755.43). This difference was statistically significant in 7 of the 9 studies and was not significant in the 2 smallest studies.9,12

Random-effects analysis of the difference in means demonstrated significant evidence for longer CPB times for subjects with AKI. Using the pooled random-effects model, the mean time difference was 23.18 minutes (95% CI, 16.70-29.66; z ⫽ 7.012; corresponding 2-tailed p ⬍ 0.0001). This overall difference was 25.65 minutes (95% CI, 22.86-28.45) with fixedeffects meta-analysis (z ⫽ 17.97; p ⬍ 0.0001). There was evidence of heterogeneity among studies (Q ⫽ 34.234, p ⬍ 0.0001) for the mean CPB time differences between the AKI and no-AKI groups. AKI-CPB and Overall Mortality There was variability in the reporting of overall mortality in the different studies (eg, intensive care unit mortality v 28-day mortality). Mortality data were available in 4 of those 9 studies for patients undergoing RRT versus controls and in 5 studies for patients who developed AKI versus controls. The risk of mortality was higher and statistically significant in the AKI cohort compared with the controls in 5 of the 9 studies.8,10,11,13,15 The risk of death was significantly higher within the cohort of patients with postoperative AKI if they needed any form of RRT during their hospital course (Table 3). To summarize the reported data from the final cohort of studies, multivariate analyses in 3 of the final selected studies also showed a significantly increased risk of AKI-CPB in patients who received blood transfusions (⬎3 U packed cells) while on CPB.10,11,13 Emergency surgery and the perioperative use of an intra-aortic balloon pump were independent risk factors for AKI-CPB.10,11 Preoperative renal dysfunction was one of the most consistent predictors of AKI-CPB.8,11,13,16 All 9 studies showed an increased risk of AKI-CPB in patients with advanced age compared with the control group. Publication Bias

Fig 2. Forest plot of differences in mean cardiopulmonary bypass times between the cohort with and the cohort without acute kidney injury. Studies are in sequence by total sample size. Individual estimates from each study and overall fixed-effects estimates are presented with associated 95% confidence intervals.

In this study, a funnel plot and the classic fail-safe N were used to test for the presence of publication bias. In the funnel plot (Fig 3), differences in means for individual studies are plotted against their precision (1-standard error) for each study. The fixed-effect overall results of the study are represented by the vertical solid line. This plot shows that there are studies on the lower ends of the 2 sides of the overall effect, which

POSTOPERATIVE AKI AND DURATION OF CPB

67

Table 3. Mortality Data in the 3 Patient Cohorts No-AKI (Control) Cohort Study, Year Published

al,15

Santos et 2004 Provenchère et al,10 2003 Landoni et al,13 2007 Bove et al,11 2004 Conlon et al,8 1999

AKI ⫹ RRT Cohort

AKI Cohort

Deaths

N

Mortality (%)

Deaths

N

Mortality (%)

2 9 20 132 44

187 537 2,187 4,897 2,464

1.07 1.68 0.91 2.70 1.79

9 30 16 79 30

36 112 45 171 214

5.56 8.04 44.44 77.19 20.56

Deaths

N

Mortality (%)

7

11

63.64

14 60 5

24 94 19

58.33 63.83 26.32

Abbreviations: AKI, acute kidney injury; CPB, cardiopulmonary bypass; RRT, renal replacement therapy.

indicates that publication bias is unlikely. Similar results are observed for the funnel plot of the random-effects model. The Rosenthal fail-safe N18 is calculated as 541, suggesting that 541 studies with nonsignificant results would need to be included in addition to the present 9 studies so the overall p value of the meta-analysis would exceed 0.05. Hence, a publication bias was very unlikely in the present meta-analysis comparing differences in CPB times. DISCUSSION

The spectrum of CPB-induced pathophysiologic changes includes the systemic inflammatory response, changes in renal vasomotor tone, destruction of red cells, pigment nephropathy, loss of pulsatile flow, activation of complement and coagula-

tion pathways, and generation of microemboli (fibrin, platelet aggregates, cellular debris, fat, and air). Other established patient and surgical risk factors include advanced age, preoperative renal insufficiency, ejection fraction ⬍40%, excessive hemodilution on CPB, use of circulatory support devices (eg, intra-aortic balloon pumps), type of cardiac surgery (cardiac valve and combined valve-coronary artery bypass grafting procedures), length of CPB, and emergency cardiac surgery.1,19 Previous studies have hinted that there likely exists a linear relation between the duration of CPB and the development of AKI. Mangano et al reported that CPB lasting ⱖ180 minutes was an independent risk factor for postoperative renal dysfunction.20 This time window has been revisited by many investi-

Fig 3. Funnel plot of differences is mean cardiopulmonary bypass times between patients with and those without acute kidney injury (renal failure v no renal failure) and precision of the study. Differences in cardiopulmonary bypass times for each study are presented on the x-axis against the precision of a study on the y-axis. The overall fixed-effects difference in means is represented by the vertical solid line.

68

KUMAR ET AL

gators since the original publication and with the revision of criteria for AKI. Recently, Salis et al reported that a CPB time of 115 minutes was an independent risk factor for AKI-CPB.21 The duration of CPB depends on several factors, including the severity of illness of the patient, complexity of the surgical procedure, an individual surgeon’s experience with the particular procedure, and hemodynamic changes in the perioperative period. Research protocols randomizing patients to a prespecified CPB duration group during surgery is not practical. This fundamentally underlies the complexity in studying the role of duration of CPB alone by randomized, controlled trials. The results of this meta-analysis suggest that an increased duration on CPB is a risk factor for developing AKI-CPB. This metaanalysis was not designed to define a safe time window on CPB or determine a time on CPB beyond which the risk of AKI-CPB increases significantly. However, this meta-analysis confirmed what is known from previous studies, that AKI-CPB is associated with a significant increase in postoperative mortality, especially if these patients require RRT.3 The risk of long-term mortality is increased up to 10 years even when the serum creatinine returns to baseline levels in patients who develop AKI-CPB.2,22 This emphasizes the need to recognize changes in serum creatinine and address the modifiable factors during hospitalization to decrease overall postoperative mortality after cardiac surgery. This study has notable limitations. The final set of studies ranged from prospective observational studies to retrospective chart reviews. There were no randomized, controlled trials in the final cohort of studies. Quality assessment of the included studies was performed using the MOOSE criteria,17 which attempt to address some of the shortcomings of using

nonrandomized, controlled trials in a meta-analysis. There were limited data on transfusion triggers, temperature management on CPB, and criteria for initiating RRT in the final cohort of studies. There were no data on AKI-specific biomarkers available in the final cohort of studies. Hence, findings were limited to what was reported in the final cohort of studies. There were limited data on the severity of illness of the patients in the different studies. Only 3 of the 9 studies stratified patients by risk; however, they used the American Society of Anesthesiologists classification, which is not designed to validate outcomes after cardiac surgery. The risk scoring system for cardiac surgical patients commonly uses the European System for Cardiac Operative Risk Evaluation (EuroSCORE) or the Society of Thoracic Surgery score23,24 as a severity-of-illness score. Hence, in these meta-analyses, the severity of illness of the patient populations in the different studies cannot be compared directly. Despite the limitations, meta-analyses of nonrandomized, controlled studies remain an important adjunct to systematic reviews, especially when generalizations have been made based on a collection of existing studies. CONCLUSIONS

Longer duration of CPB is associated with an increased risk of developing postoperative AKI. ACKNOWLEDGMENTS The authors thank Drs Provonchère, Landoni, and Conlon for their willingness to discuss the project and provide specific anonymous data (Drs Landoni and Provonchère) from their previously published studies.

REFERENCES 1. Kumar AB, Suneja M: Cardiopulmonary bypass-associated acute kidney injury. Anesthesiology 114:964-970, 2011 2. Lassnigg A, Schmidlin D, Mouhieddine M, et al: Minimal changes of serum creatinine predict prognosis in patients after cardiothoracic surgery: A prospective cohort study. J Am Soc Nephrol 15:1597-1605, 2004 3. Chertow GM, Levy EM, Hammermeister KE, et al: Independent association between acute renal failure and mortality following cardiac surgery. Am J Med 104:343-348, 1998 4. Thakar CV, Yared JP, Worley S, et al: Renal dysfunction and serious infections after open-heart surgery. Kidney Int 64:239-246, 2003 5. Mehta RL, Kellum JA, Shah SV, et al: Acute Kidney Injury Network: Report of an initiative to improve outcomes in acute kidney injury. Crit Care 11:R31, 2007 6. Kellum JA, Levin N, Bouman C, et al: Developing a consensus classification system for acute renal failure. Curr Opin Crit Care 8:509514, 2002 7. Haase M, Bellomo R, Matalanis G, et al: A comparison of the RIFLE and Acute Kidney Injury Network classifications for cardiac surgery associated acute kidney injury: A prospective cohort study. J Thorac Cardiovasc Surg 138:1370-1376, 2009 8. Conlon PJ, Stafford-Smith M, White WD, et al: Acute renal failure following cardiac surgery. Nephrol Dial Transplant 14:11581162, 1999 9. Fischer UM, Weissenberger WK, Warters RD, et al: Impact of cardiopulmonary bypass management on postcardiac surgery renal function. Perfusion 17:401-406, 2002

10. Provenchère S, Plantefève G, Hufnagel G, et al: Renal dysfunction after cardiac surgery with normothermic cardiopulmonary bypass: Incidence, risk factors, and effect on clinical outcome. Anesth Analg 96:1258-1264, 2003 11. Bove T, Calabrò MG, Landoni G, et al: The incidence and risk of acute renal failure after cardiac surgery. J Cardiothorac Vasc Anesth 18:442-445, 2004 12. Tuttle KR, Worrall NK, Dahlstrom LR, et al: Predictors of ARF after cardiac surgical procedures. Am J Kidney Dis 41:76-83, 2003 13. Landoni G, Bove T, Crivellari M, et al: Acute renal failure after isolated CABG surgery: Six years of experience. Minerva Anestesiol 73:559-565, 2007 14. Sirvinskas E, Andrejaitiene J, Raliene L, et al: Cardiopulmonary bypass management and acute renal failure: Risk factors and prognosis. Perfusion 23:323-327, 2008 15. Santos FO, Silveira MA, Maia RB, et al: Acute renal failure after coronary artery bypass surgery with extracorporeal circulation—Incidence, risk factors, and mortality. Arq Bras Cardiol 83:145-154, 2004 16. Kincaid EH, Ashburn DA, Hoyle JR, et al: Does the combination of aprotinin and angiotensin-converting enzyme inhibitor cause renal failure after cardiac surgery? Ann Thorac Surg 80:1388-1393, 2005 17. Stroup DF, Berlin JA, Morton SC, et al: Meta-analysis of observational studies in epidemiology: A proposal for reporting. Metaanalysis Of Observational Studies in Epidemiology (MOOSE) group. JAMA 283:2008-2012, 2000 18. Borenstein M, Hedges LV, Higgins JPT, et al: Introduction to Meta Analysis. UK, John Wiley & Sons, 2009, pp 277-292

POSTOPERATIVE AKI AND DURATION OF CPB

19. Okusa MD: The inflammatory cascade in acute ischemic renal failure. Nephron 90:133-138, 2002 20. Mangano CM, Diamondstone LS, Ramsay JG, et al: Renal dysfunction after myocardial revascularization: Risk factors, adverse outcomes, and hospital resource utilization. The Multicenter Study of Perioperative Ischemia Research Group. Ann Intern Med 128:194-203, 1998 21. Salis S, Mazzanti VV, Merli G, et al: Cardiopulmonary bypass duration is an independent predictor of morbidity and mortality after cardiac surgery. J Cardiothorac Vasc Anesth 22:814-822, 2008

69

22. Hobson CE, Yavas S, Segal MS, et al: Acute kidney injury is associated with increased long-term mortality after cardiothoracic surgery. Circulation 12:2444-2453, 2009 23. Nashef SA, Roques F, Hammill BG, et al: Validation of European System for Cardiac Operative Risk Evaluation (EuroSCORE) in North American cardiac surgery. Eur J Cardiothorac Surg 22:101-105, 2002 24. Edwards FH, Clark RE, Schwartz M: Coronary artery bypass grafting: The Society of Thoracic Surgeons national database experience. Ann Thorac Surg 57:12-19, 1994