Impact of Early Renal Recovery on Survival After Cardiac Surgery ...

CARDIOTHORACIC ANESTHESIOLOGY: The Annals of Thoracic Surgery CME Program is located online at http://cme.ctsnetjournals.org. To take the CME activity related to this article, you must have either an STS member or an individual non-member subscription to the journal.

ADULT CARDIAC

Impact of Early Renal Recovery on Survival After Cardiac Surgery-Associated Acute Kidney Injury Madhav Swaminathan, MD, Christopher C.C. Hudson, FRCPC, Barbara G. Phillips-Bute, PhD, Uptal D. Patel, MD, Joseph P. Mathew, MD, Mark F. Newman, MD, Carmelo A. Milano, MD, Andrew D. Shaw, MB, FRCA, and Mark Stafford-Smith, MD, FRCPC Division of Cardiothoracic Anesthesiology, Department of Anesthesiology; Department of Medicine and the Duke Clinical Research Institute, and Department of Surgery, Duke University Medical Center, Durham, North Carolina

Background: Cardiac surgery-associated acute kidney injury (CSA-AKI) is a major postoperative complication. Although some early recovery is common, its effect on long-term outcomes is unclear. We tested the hypothesis that early renal recovery after CSA-AKI is independently associated with improved long-term survival. Methods: Data were examined for 10,275 consecutive patients undergoing isolated coronary artery bypass grafting from 1996 to 2005. Patients with CSA-AKI were identified, defined as a peak postoperative creatinine level exceeding 50% above baseline. Renal recovery was characterized using postoperative creatinine values. The recovery variable with the strongest association with 1-year survival was selected and validated internally. The independent association of early renal recovery with long-term survival during a 10-year follow-up was assessed with Cox proportional hazards modeling.

Results: CSA-AKI occurred in 1113 patients (10.8%). The renal recovery variable with the strongest association with 1-year survival was the percentage decrease in creatinine 24 hours after its peak value (PD24; C index, 0.72; p ⴝ 0.002). Cox proportional hazards analysis showed a significant negative association between PD24 and long-term mortality (0.82 hazard ratio for each 10% change). Conclusions: Early recovery of renal function is associated with improved long-term survival after CSA-AKI. This variable is clinically useful because it occurs immediately after the peak creatinine level and simultaneously helps define the severity of AKI and the magnitude of recovery. Given the high risk of death associated with postoperative AKI, early renal recovery seems to offer a distinct survival benefit and may represent an important therapeutic focus. (Ann Thorac Surg 2010;89:1098 –105) © 2010 by The Society of Thoracic Surgeons

C

anisms may enhance recovery of kidney function after an acute insult. The recovery paradigm is important because it represents a potential focus for interventions that may modify the effect of AKI on death. Although the predictive value of CSA-AKI on long-term survival is known [8], the modifying influence of early recovery of renal function is unclear. We therefore tested the hypothesis that early recovery of renal function after AKI is independently associated with improved long-term survival in patients undergoing cardiac operations.

ardiac surgery-associated acute kidney injury (CSAAKI) remains a major postoperative complication that is associated with a high risk of short-term and long-term death [1–3]. Although epidemiologic studies have used variable metrics to define CSA-AKI, these are almost universally based on perioperative changes in serum creatinine values. Despite abundant data on the prognostic importance of AKI, there is limited information on early recovery of renal function after the creatinine level has reached its postoperative peak. The kidney is remarkable in its ability to almost completely recover from a total loss of function. Investigators have reported favorable outcomes when kidneys demonstrate some recovery of function after acute failure [4, 5]; however, recovery of renal function has been studied mainly in terms of freedom from dialysis [4 –7]. The effect of early recovery from smaller— but still significant— decrements in renal function remains largely unknown. Emerging data suggest that endogenous repair mech-

Material and Methods The study was approved by the Duke University Institutional Review Board as a retrospective, observational cohort study. The requirement for informed consent was waived. A study protocol with the specified hypothesis was submitted before data retrieval and analysis.

Accepted for publication Dec 9, 2009.

Data Sources

Address correspondence to Dr Swaminathan, Box 3094 Anesthesiology, Duke University Medical Center, Durham, NC 27710; e-mail: [email protected].

Detailed clinical, laboratory, and outcomes data were obtained from prospectively entered databases for consecutive adult patients undergoing coronary artery by-

© 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc

0003-4975/10/$36.00 doi:10.1016/j.athoracsur.2009.12.018

pass grafting (CABG) operations at Duke University Medical Center during the 10 years between January 1996 and December 2005. Demographic and procedural data were obtained from the Duke Databank for Cardiovascular Diseases, a large, quality-assured data repository for patients undergoing cardiovascular procedures maintained by the Duke Clinical Research Institute, Durham, North Carolina. Serum creatinine values for the perioperative period were obtained from the Duke Clinical Pathology Laboratory. Follow-up was conducted by the Duke Clinical Research Institute Follow-up Services Group and has been previously described [9]. Briefly, this group is responsible for collecting annual follow-up survival data and nonfatal end-point information for the Duke Databank for Cardiovascular Diseases. The annual surveys collect data 6 months after an index visit and yearly thereafter. Follow-up is 95% complete for mortality, and patients who are lost to follow-up (2%) or who have asked to be withdrawn (3%) are submitted for an annual search of the National Death Index. Cause of death is assigned after agreement from independent reviews by an adjudication committee. Specific outcomes for this study were obtained for up to 10 years after the initial operation in all patients and included all-cause mortality. For patients who had more than one surgical revascularization during the follow-up period, only the data pertaining to the first operation was retained.

Participant Selection Patients undergoing isolated, elective CABG procedures were selected for this study. To identify a cohort with defined AKI, only those with a peak postoperative creatinine level exceeding 50% of the preoperative (baseline) value were included. Patients requiring renal replacement therapy (RRT) before or after the operation during the index hospitalization were excluded to eliminate the confounding effect of RRT on creatinine values. Those who died before hospital discharge were excluded because they may not have had time to demonstrate a peak in creatinine levels or recovery from AKI. From this subset of patients, those with peak creatinine levels occurring after postoperative day 5 were excluded because they may have had ongoing injury or an additional kidney injury that could compromise functional recovery. Our purpose was to identify a homogenous cohort with CSA-AKI defined to be temporally related to a known event—the surgical procedure— but not confounded by postoperative influences on renal recovery such as in-hospital death, dialysis, or a peak in creatinine after day 5.

Renal Function Variables The initial cohort of patients selected from the Duke Databank for Cardiovascular Diseases was cross-referenced with the laboratory database to retrieve all creatinine values from 3 days before the operation to 10 days after. Preoperative (baseline) creatinine was defined as the preoperative value closest to the operation. For multiple daily creatinine values, the first

CREATININE VALUES.

SWAMINATHAN ET AL RENAL RECOVERY AND POST-CABG SURVIVAL

1099

value for the day was selected for analysis. Peak postoperative creatinine was defined as the first day of the highest in-hospital creatinine value within the first 5 days after the operation. MISSING VALUES. Creatinine values are often less frequently measured as the hospital stay increases and the patient becomes more stable. Patients with missing creatinine values within the first 3 postoperative days were excluded from the analysis. Starting on postoperative day 4, missing creatinine values were imputed from the previous day’s creatinine value. DEFINITION OF AKI. Criteria suggested by the Acute Kidney Injury Network (AKIN) [10] were used to define AKI as an increase of 50% or more in the peak postoperative creatinine level above baseline. In addition, for assessment of distribution of renal recovery according to severity of injury, AKI was also categorized into the AKINrecommended stages based on the percentage change in peak postoperative creatinine above baseline: stage 1, 50% to 100%; stage 2, 100% to 200%; and stage 3, more than 200%. Creatinine values after AKI were used to construct several recovery variables (Table 1). While renal recovery area represented creatinine levels after the peak value and accounted for duration of recovery, the 24- and 48-hour post-peak creatinine values represented early decline in creatinine and early renal recovery. The 24and 48-hour values of change were characterized as absolute and percentage of peak postoperative creatinine to indicate magnitude of early recovery.

Statistical Analyses We characterized early renal recovery using seven different definitions (Table 1). To determine the most clinically relevant recovery variable, we tested the association of each variable with 1-year mortality in a series of multivariable logistic regression models. Each model was adjusted for baseline creatinine, peak postoperative creatinine, and risk score based on the European System for Cardiac Operative Risk Evaluation (EuroSCORE), a validated method of mortality risk prediction for patients undergoing cardiac operations [11, 12]. The covariables were selected a priori and were included in each model regardless of significance so that all models would be comparable. We used the C index of the models to select the most predictive characterization of renal recovery. VALIDATION. To validate the selection of the variable, a bootstrap analysis was performed in which each of the seven multivariable models was fit to 1000 bootstrap samples of the data set. For each bootstrap sample, patient records were randomly sampled from the data set with replacement. Point estimates for the odds ratios and the C indices were obtained for each bootstrap sample. Goodness of fit was assessed for each model with the Hosmer-Lemeshow test. VARIABLE SELECTION.

Survival Analysis After identifying the renal recovery variable with greatest predictive ability, we tested the primary hypothesis that early recovery of renal function after AKI is associated

ADULT CARDIAC

Ann Thorac Surg 2010;89:1098 –105

1100


Ann Thorac Surg 2010;89:1098 –105

Table 1. Creatinine-Based Variables and Definitions of Early Renal Recovery Variable

ADULT CARDIAC

Abbreviation

Definition

Creatinine variables Pre-op creatinine Peak post-op creatinine Discharge creatinine Post-peak 24-h creatinine Post-peak 48-h creatinine

CrPre CrMax DcCr p24Cr p48Cr

Earliest Cr value pre-op Highest in-hospital Cr value post-op Cr value at discharge or at post-op day 10, whichever was earlier Cr value on first day after CrMax Cr value on second day after CrMax

Renal recovery variables Renal recovery area Absolute decrease, mg/dL Percentage decrease Peak decrease 24, mg/dL Percentage decrease 24 Peak decrease 48, mg/dL Percentage decrease 48

RRA AD PD PkD24 PD24 PkD48 PD48

Area under curve from CrMax to DcCr CrMax ⫺ DcCr [(CrMax ⫺ DcCr)/CrMax] ⫻ 100 CrMax ⫺ p24Cr [(CrMax ⫺ p24Cr)/CrMax] ⫻ 100 CrMax ⫺ p48Cr [(CrMax ⫺ p48Cr)/CrMax] ⫻ 100

with improved long-term survival. A survival curve was constructed for the entire follow-up period using the Kaplan-Meier method, with start time as the date of the operation and time to death as the principal outcome. Further investigation with Cox proportional hazards regression analysis tested the association of the selected recovery variable and long-term survival adjusting for preoperative body weight, EuroSCORE, preoperative creatinine, peak postoperative creatinine, and cardiopulmonary bypass time. Patients who did not die within the study follow-up period were considered censored at the point of the last follow-up. Proportional hazards assumptions were tested with an interaction between the difference between the peak creatinine and the value at the next 24 hours as a percentage of the peak value (PD24) and the log of the study follow-up time. Significance was assessed at a two-tailed value of p ⬍ 0.05. All statistical analyses were conducted using SAS 9.1.3 software (SAS Institute Inc, Cary, NC). Odds ratios (OR) and hazard ratios (HR) are presented with 95% confidence intervals (CI).

Results Between January 1996 and December 2005, 10,275 consecutive elective, isolated CABG procedures were performed. From this population, 1113 patients who met inclusion criteria for CSA-AKI constituted the principal study cohort. According to exclusion criteria, 146 patients died in-hospital postoperatively, 502 received perioperative RRT (444 preoperatively and 58 more postoperatively), the postoperative creatinine change in 8481 patients did not exceed 50%, and the postoperative creatinine level in 33 patients peaked after postoperative day 5. Survival status was determined for 1111 patients with a median follow-up time of 40 months. Characteristics of the CSA-AKI study population, including key renal function variables, are reported in Table 2. AKI was at stage 1 in 799 patients, stage 2 in 264, and stage 3 in 50.

The 1-year mortality rate was 8% (n ⫽ 91). Multivariable logistic regression models for 1-year mortality were developed for the seven renal recovery variables (Table Table 2. Perioperative Characteristics of Study Population, Including Key Renal Function Variables Variable Age, y Female sex White race Weight, kg Diabetes Hypertension History of Myocardial infarction Stroke Smoking Congestive heart failure Preoperative aspirin use CPB time, min Cross-clamp time, min Hyperlipidemia Ejection fraction COPD Peripheral vascular disease Creatinine variables, mg/dL Pre-op creatinine Peak post-op creatinine Discharge creatinine Post-peak 24-hour creatinine Post-peak 48-hour creatinine

Valuea Mean (SD) or %

Range

65.9 (10.4) 30.8 81.8 87.5 (20.1) 39.4 75.0

... ... ... ... ... ...

46 10.9 45.0 15.9 51.5 107.2 (55.8) 65.0 (28.0) 52.5 0.508 (0.159) 11.7 16.3

... ... ... ... ... ... ... ... ... ... ...

1.07 (0.32) 2.06 (0.87) 1.28 (0.47) 1.69 (0.80) 1.46 (0.68)

0.5–3.2 0.8–7.8 0.5–5.0 0.3–7.3 0.3–6.7

a

Data are presented as mean values with standard deviation in parentheses or percentage of population where appropriate. CPB ⫽ cardiopulmonary bypass; nary disease.

COPD ⫽ chronic obstructive pulmo-

Ann Thorac Surg 2010;89:1098 –105


1101

Variablea Renal recovery area Peak decrease 24 Peak decrease 48 Absolute decrease Percentage decrease 24 Percentage decrease 48 Percentage decrease a

␤ Coefficient

OR (95% CI)

p Value

C Index

Boot-strapped C Index

0.118 ⫺0.86 ⫺0.45 ⫺0.70 ⫺0.03 ⫺0.02 ⫺0.009

1.13 (1.011–1.25) 0.42 (0.18–0.99) 0.64 (0.34–1.19) 0.49 (0.26–0.93) 0.97 (0.95–0.98) 0.98 (0.97–1.001) 0.99 (0.98–1.01)

0.03 0.04 0.15 0.03 0.002 0.06 0.27

0.707 0.715 0.706 0.706 0.724 0.707 0.703

0.713 0.721 0.713 0.712 0.729 0.713 0.709

Definitions of variables are provided in Table 1.

CI ⫽ confidence interval;

OR ⫽ odds ratio.

3). The PD24 (Table 1) was the recovery variable that showed the strongest negative association with 1-year mortality (OR estimate per unit change, 0.97; 95% CI, 0.95 to 0.99; p ⫽ 0.002) with the highest C index (0.72). Peak postoperative creatinine (p ⫽ 0.01) and EuroSCORE index (p ⬍ 0.0001) were also significantly associated with 1-year mortality in this model. Bootstrap analysis showed that in 84% of samples, the recovery variable with the strongest negative association with 1-year survival was PD24 (C index, 0.729). The Hosmer-Lemeshow test revealed that values for all models were non-significant, with no differences between observed and expected frequencies of events, thereby indicating adequate goodness of fit. The distribution of PD24 for the entire CSAAKI population and among the three AKI stages is shown in Figure 1. Survival data using the Kaplan-Meier method demonstrated a median survival time of 40 months for the cohort (Fig 2), with a maximum follow-up of 135 months. During the follow-up, 316 patients died, and the 797 patients who did not die were censored at the time of the last follow-up. Patients at risk at each 12-month point are shown in Figure 2. After adjustment for EuroSCORE, preoperative creatinine, peak postoperative creatinine, weight, and cardiopulmonary bypass time, the Cox proportional hazards regression model showed a significant negative association between PD24 and long-term mortality (HR for a 10% unit change, 0.82; 95% CI, 0.74 to 0.90; Table 4). Assumptions of proportional hazards were satisfied (Wald ␹2, 2.29; p ⫽ 0.13) indicating that we did not reject the null hypothesis.

Comment The present study confirmed our hypothesis that early recovery of renal function after AKI is associated with improved long-term survival in patients undergoing cardiac operations. Because strategies to prevent CSA-AKI have been largely unsuccessful, early recovery of function may therefore represent a new focus for intervention that alters the risk of long-term death associated with AKI. Recovery of renal function is not well characterized, and its long-term effects are even less clear. In a study

primarily designed to assess the effect of CSA-AKI on 30-day mortality, Lassnigg and colleagues [13] found that patients with a negative change in postoperative creatinine had the lowest mortality rate. Although their study did not examine recovery from AKI, they showed that a negative change in creatinine was associated with improved survival. Recently, Bihorac and colleagues [14] reported that patients with partial or complete recovery after postoperative AKI were at a lower risk of long-term death compared with those that had no recovery. Their study defined recovery using a ratio of discharge to baseline creatinine and included patients undergoing cardiovascular operations. In a meta-analysis of renal recovery after AKI, Schmitt and colleagues [15] reported that age was a major factor associated with a lower likelihood of recovery of renal function. They acknowledged their study was limited by variable definitions of AKI, renal recovery, and time to renal recovery, and insufficient data on baseline creatinine and comorbidity scores. Other studies of AKI have used a relatively firm definition of renal recovery—that of freedom from dialysis [16 –20]. The use of this variable as a definition of renal recovery, however, limits data to only those who have severe AKI requiring RRT and does not provide information on the larger group of patients who have smaller but important decrements in renal function. Our study, in contrast, adds significantly to current knowledge on early renal recovery in a large, homogenous population of CABG patients with standard definitions of CSA-AKI based on detailed creatinine data. These data were used to develop specific characterizations of renal recovery that are clinically meaningful and were validated in their association with outcome. Functional recovery of the kidneys is well documented, and the capacity of the kidney to use endogenous mechanisms for repair and regeneration has been recognized [21–23]. Although the precise mechanisms of renal cellular repair remain unclear, the concept of endogenous repair after an acute insult is vitally important. The ability of organs to repair themselves after an injury represents a survival response, and a greater ability to recover from an insult may be reflected in improved overall survival. The renal recovery variable in our study

ADULT CARDIAC

Table 3. Multivariable Logistic Regression Model of the Independent Association of Renal Recovery Variables With 1-year Mortality

1102


Ann Thorac Surg 2010;89:1098 –105

ADULT CARDIAC Fig 1. Frequency distribution is shown of early renal recovery, defined as the percentage decrease in creatinine 24 hours after its peak value (PD24) in the (A) entire study population, and in the subsets of patients grouped according to severity of acute kidney injury (AKI) based on standard criteria of (B) stage I, (C) stage II, and (D) stage III.

with the strongest negative association with 1-year mortality was PD24, reflecting the importance of the magnitude of early recovery of renal function. In contrast, other variables based on more gradual recovery did not influence survival after AKI to the same degree. Our findings suggest that when the delicate balance between ongoing injury and early repair tilts in favor of repair, the magnitude of this early reversal of injury significantly alters the risk of death. Mechanisms that control the magnitude of early recovery after a renal insult may therefore have a major effect in survival after AKI. The principal strengths of our study are the use of a large cohort of CABG patients with standardized definitions of AKI from a quality-controlled database with robust follow-up methods. However, our study also has certain limitations associated with the study design and measurement of renal function. Our findings apply to a relatively homogenous CABG population limited by our strict selection criteria and may not be generalizable across the entire spectrum of patients encountered in clinical practice. However, our findings encourage the

search for the value of early renal recovery after AKI in other similar populations. Although the study was retrospective and data collection was not hypothesis-driven, data were prospectively collected and continuously validated to enable outcomes measurement. Serum creatinine as a biomarker of renal function has well-known limitations. More sensitive biomarkers of renal injury have been identified, including interleukin18, neutrophil gelatinase-associated lipocalin, kidney injury molecule-1, and cystatin C [24]. Although these biomarkers are indeed sensitive to subtle changes in kidney function, they have not yet been validated as independent markers of adverse long-term outcomes in contrast to creatinine, probably because decreases in creatinine occur well after renal function has already declined, indicating a clinically relevant injury. Moreover, other biomarkers have also not yet been investigated as indicators of renal recovery, and creatinine remains the current standard on which clinical management is based. An important contribution of this study is that it further characterizes patterns of renal response to peri-


1103

ADULT CARDIAC

Ann Thorac Surg 2010;89:1098 –105

Fig 2. Separate survival curves are shown for long-term survival according to early renal recovery, as defined by the percentage decrease in creatinine 24 hours after its peak value (PD24) of 9% (black line), 18% (dashed gray line), and 26% (solid gray line), representing the 25th, 50th, and 75th percentiles, respectively.

operative insult independent from current definitions of AKI while significantly supplementing their ability to predict the risk of death. Also notable is that this information on early serum creatinine decline is often immediately available to intensivists as the peak is recognized (presuming daily evaluation). To demonstrate the importance of this concept, we inserted hypothetic serum creatinine values typically seen perioperatively in our primary model (Fig 3). The following example illustrates the significance of early renal recovery. Patients A and B have identical baseline serum creatinine values of 1 mg/dL and sustain similar AKI, reflected by peak postoperative creatinine values of 3.0 mg/dL (an example of stage 3 AKI). Without additional data, assuming a mean EuroSCORE and an approximately average creatinine decline over the subsequent 24 hours (PD24) of 20% to 2.4 mg/dL, the predicted mortality risk at 1 year for both patients would Table 4. Cox Proportional Hazards Regression Model Including Early Recovery of Renal Functiona and Other Variables Associated With Long-Term Postoperative Mortality Variable PD24 Peak post-op creatinine Pre-op creatinine EuroSCORE CPB time Weight

HR (95% CI)

p Value

0.82 (0.74–0.90) 1.23 (1.07–1.40) 1.47 (1.03–2.10) 1.15 (1.11–1.19) 1.00 (0.99–1.00) 0.99 (0.98–0.99)

⬍0.001 0.003 0.04 ⬍0.001 0.16 ⬍0.001

a Defined as the percentage decrease in the creatinine level 24 hours after its peak value (PD24)

CI ⫽ confidence interval; CPB ⫽ cardiopulmonary bypass; EuroSCORE ⫽ European System for Cardiac Operative Risk Evaluation; HR ⫽ hazard ratio; PD24 ⫽ see footnote a.

Fig 3. The effect of early renal recovery on the mortality risk is shown by inserting hypothetical values for preoperative creatinine (1.0 mg/dL),with peak postoperative creatinine levels of 1.5, 2.0, and 3.0 mg/dL as examples of acute kidney injury (AKI) stages I (diamonds), II (squares), and III (triangles), respectively, and mean European System for Cardiac Operative Risk Evaluation (EuroSCORE) inserted into the primary model. Predicted 1-year mortality was calculated for each AKI stage based on three different values (10%, 20%, and 30%) of PD24, defined as percentage decrease in creatinine 24 hours after its peak value.

be 7.9%. However, if by the next day the serum creatinine in patient A declines instead to 2.1 mg/dL, whereas in patient B the decline is to 2.7 mg/dL, reflecting PD24 values of 30% and 10%, respectively, their predicted mortality risks are significantly different. The accelerated renal recovery reflected in patient A is associated with a reduced mortality risk of 5.7%, whereas the slower recovery for patient B portends an almost twofold higher mortality risk of 10.8%. Thus, in our study more rapid recovery of renal filtration after the maximum insult, although reflecting a net sum of ongoing injury and a potentially more robust renal recovery profile, translates into a clinically important survival advantage 1 year after the operation. For cardiac surgeons and intensivists, the post-peak decline is a clinically useful value in that it improves the ability to predict long-term mortality risk based on the magnitude of early recovery. We could then use this value to improve postoperative surveillance if patients do not show sufficient early recovery. The next steps could include elucidation of modifiable perioperative factors that influence early renal recovery and assessment of interventions that target such factors, including mechanisms of renal repair and regeneration. Indeed, given the limited success of interventions that prevent CSA-AKI, improvement in early renal recovery may represent a novel and valid therapeutic goal. In summary, our study found that early recovery of renal function is associated with improved long-term survival after postoperative AKI in CABG patients. Given the high mortality risk associated with postoperative AKI, early renal recovery seems to offer a distinct survival benefit and may represent a novel therapeutic target.

1104


Markers of early renal recovery should also be incorporated into studies that examine the effect of AKI on adverse outcomes. ADULT CARDIAC

References 1. Loef BG, Epema AH, Smilde TD, et al. Immediate postoperative renal function deterioration in cardiac surgical patients predicts in-hospital mortality and long-term survival. J Am Soc Nephrol 2005;16:195–200. 2. Cooper WA, O’Brien SM, Thourani VH, et al. Impact of renal dysfunction on outcomes of coronary artery bypass surgery: results from the Society of Thoracic Surgeons National Adult Cardiac Database. Circulation 2006;113:1063–70. 3. Shaw A, Swaminathan M, Stafford-Smith M. Cardiac surgery-associated acute kidney injury: putting together the pieces of the puzzle. Nephron Physiol 2008;109:p55– 60. 4. Bagshaw SM, Mortis G, Godinez-Luna T, Doig CJ, Laupland KB. Renal recovery after severe acute renal failure. Int J Artif Organs 2006;29:1023–30. 5. Theriault J, Agharazzi M, Dumont M, et al. Atheroembolic renal failure requiring dialysis: potential for renal recovery? A review of 43 cases. Nephron Clin Pract 2003;94:c11– 8. 6. Albright RC Jr, Smelser JM, McCarthy JT, et al. Patient survival and renal recovery in acute renal failure: randomized comparison of cellulose acetate and polysulfone membrane dialyzers. Mayo Clin Proc 2000;75:1141–7. 7. Bagshaw SM, Laupland KB, Doig CJ, et al. Prognosis for long-term survival and renal recovery in critically ill patients with severe acute renal failure: a population-based study. Crit Care 2005;9:R700 –9. 8. Brown JR, Cochran RP, MacKenzie TA, et al. Long-term survival after cardiac surgery is predicted by estimated glomerular filtration rate. Ann Thorac Surg 2008;86:4 –11. 9. Swaminathan M, Morris RW, De Meyts DD, et al. Deterioration of regional wall motion immediately after coronary artery bypass graft surgery is associated with long-term major adverse cardiac events. Anesthesiology 2007;107:739 – 45. 10. 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 2007;11:R31. 11. Nashef SA, Roques F, Michel P, et al. European system for cardiac operative risk evaluation (EuroSCORE). Eur J Cardiothorac Surg 1999;16:9 –13.

Ann Thorac Surg 2010;89:1098 –105

12. Biancari F, Kangasniemi OP, Luukkonen J, et al. EuroSCORE predicts immediate and late outcome after coronary artery bypass surgery. Ann Thorac Surg 2006;82:57– 61. 13. Lassnigg A, Schmid ER, Hiesmayr M, et al. Impact of minimal increases in serum creatinine on outcome in patients after cardiothoracic surgery: do we have to revise current definitions of acute renal failure? Crit Care Med 2008;36:1129 –37. 14. Bihorac A, Yavas S, Subbiah S, et al. Long-term risk of mortality and acute kidney injury during hospitalization after major surgery. Ann Surg 2009;249:851– 8. 15. Schmitt R, Coca S, Kanbay M, et al. Recovery of kidney function after acute kidney injury in the elderly: a systematic review and meta-analysis. Am J Kidney Dis 2008;52:262–71. 16. Bell M, Granath F, Schon S, Ekbom A, Martling CR. Continuous renal replacement therapy is associated with less chronic renal failure than intermittent haemodialysis after acute renal failure. Intensive Care Med 2007;33:773– 80. 17. Cosentino F, Chaff C, Piedmonte M. Risk factors influencing survival in ICU acute renal failure. Nephrol Dial Transplant 1994;9(suppl 4):179 – 82. 18. Silvester W, Bellomo R, Cole L. Epidemiology, management, and outcome of severe acute renal failure of critical illness in Australia. Crit Care Med 2001;29:1910 –5. 19. Spurney RF, Fulkerson WJ, Schwab SJ. Acute renal failure in critically ill patients: prognosis for recovery of kidney function after prolonged dialysis support. Crit Care Med 1991; 19:8 –11. 20. Uchino S, Bellomo R, Morimatsu H, et al. Continuous renal replacement therapy: a worldwide practice survey. The beginning and ending supportive therapy for the kidney (B.E.S.T. kidney) investigators. Intensive Care Med 2007;33: 1563–70. 21. Kale S, Karihaloo A, Clark PR, et al. Bone marrow stem cells contribute to repair of the ischemically injured renal tubule. J Clin Invest 2003;112:42–9. 22. Ishibe S, Cantley LG. Epithelial-mesenchymal-epithelial cycling in kidney repair. Curr Opin Nephrol Hypertens 2008; 17:379 – 85. 23. Gupta S, Verfaillie C, Chmielewski D, Kim Y, Rosenberg ME. A role for extrarenal cells in the regeneration following acute renal failure. Kidney Int 2002;62:1285–90. 24. Stafford-Smith M, Patel UD, Phillips-Bute BG, Shaw AD, Swaminathan M. Acute kidney injury and chronic kidney disease after cardiac surgery. Adv Chronic Kidney Dis 2008;15:257–77.

INVITED COMMENTARY Cardiac surgery associated with acute kidney injury (AKI) is defined as a 0.3 mg/dL or 50% increase in serum creatinine (Cr) from baseline [1], and is one of the most common and complicated adverse events after cardiac surgery [2]. In this paper, Swaminathan and colleagues [3] tackle the associations between early recovery from AKI and long-term survival among patients undergoing elective isolated coronary artery bypass graft (CABG) surgery. Patients with a baseline Cr ⬎ 2.0 (mg/dL), dialysis before or after CABG, or death prior to discharge were excluded. The cohort included only patients that had AKI develop after cardiac surgery; all others were excluded for the purposes of this analysis. There are several calculations in the analysis readers should review on how to evaluate early renal recovery. These included the highest peak postoperative Cr (CrMax), peak Cr at 24 hours after the CrMax (p24Cr) and 48 hours after the CrMax (p48Cr), and discharge Cr © 2010 by The Society of Thoracic Surgeons Published by Elsevier Inc

(DcCr), which were calculated for patients with available Cr data. The percent decrease in Cr at 24 hours after surgery (PD24) was calculated by the following equation: 关共CrMax-p24Cr兲 ⁄ CrMax兴 ⫻ 100%. In determining how to interpret the results, the cohort was restricted to elective patients who had AKI develop after cardiac surgery, but not in renal failure before or after the surgery, and who were alive at discharge. The PD24 early recovery measure was identified as the best predictor of reducing the likelihood of 1-year mortality. In Table 3, the continuous form of PD24 was used (odds ratio, 0.97; 95% confidence interval, 0.95 to 0.98), suggesting for each 1% drop in Cr at 24 hours after the max Cr, the likelihood of 1-year mortality is reduced by 3%. In Table 4, PD24 was divided into deciles; the adjusted hazard ratio of 0.82 (95% confidence interval, 0.74 to 0.90) suggests that for each 10% drop in Cr at 24 hours after the CrMax, the likelihood of 1-year mortality is reduced by 0003-4975/10/$36.00 doi:10.1016/j.athoracsur.2009.12.049