In severe acute kidney injury, a higher serum creatinine is ...

8 downloads 105 Views 81KB Size Report
27 Jun 2007 ... Jorge Cerda´1, Magdalena Cerda´2, Patricia Kilcullen1 and Jayne Prendergast1. 1Deptartment of .... tion coefficient showed that larger BW gains between admission and .... Curr Opin Crit Care 2006; 12: 531–537. Table 2.

Nephrol Dial Transplant (2007) 22: 2781–2784 doi:10.1093/ndt/gfm395 Advance Access publication 27 June 2007

Hypothesis See

In severe acute kidney injury, a higher serum creatinine is paradoxically associated with better patient survival Jorge Cerda´1, Magdalena Cerda´2, Patricia Kilcullen1 and Jayne Prendergast1 1

Deptartment of Medicine, St Peter’s Hospital, Albany, NY and 2School of Public Health, University of Michigan, Ann Arbor, MI, USA

Abstract Lack of precise, reliable and consistent measures of kidney dysfunction in acute kidney injury (AKI) causes uncertainty in the definition and management of this important condition and interferes with treatment standardization. Serum creatinine (SCr) remains a key determinant in the management of renal dysfunction. In disparate populations, previous authors suggested a paradoxical association between higher SCr and better survival. We set out to analyse the association between SCr at start of continuous renal replacement therapy (CRRT) and survival, and to postulate possible mechanisms for this association. We hypothesized that in this setting, the association of higher SCr with better survival may be determined by better nutrition, lesser volume overload or pre-existing chronic kidney disease (CKD). In multivariable logistic regression analysis utilizing multiple imputation parameter estimates, a higher SCr on admission and initiation of CRRT was monotonically associated with better survival (OR 1.438, 95% CI 1.034–1.999) controlling for selected covariates. Nutrition and volume adjustments did not affect the significance of SCr. Adjustment of the model by degree of admission CDK (MDRD formula) and severity of disease (Liano scores) respectively decreased or abolished the significance of SCr levels. In univariate analysis, larger weight gains and lower urine outputs were correlated with lower SCr. In this population of critically ill, virtually anuric patients with AKI, possible explanations of this counterintuitive association include first, that a higher SCr at start of CRRT is related to pre-existing CKD. CKD patients may require a lesser burden of disease to reach the point where CRRT is needed, and therefore have a better survival. Correspondence and offprint requests to: Dr Jorge Cerda´, Nephrology CDRP, 62 Hackett Boulevard, Albany, New York 12209, United States. Email: [email protected]

Inversely, a lower SCr may be an indication of fluid overload, associated with worse survival. Our findings did not support a role of nutrition or muscle mass for this association. All these possibilities are worthy of thorough investigation, as findings will likely result in important changes in patient outcome. Keywords: acute kidney injury; chronic kidney disease; creatinine; fluid overload; glomerular filtration rate; nutrition; outcome prediction

Introduction One of the barriers in early detection of renal failure is the lack of a precise, reliable and consistent measure of kidney function [1]. Until recently, more than 30 different definitions of acute kidney injury (AKI) have been used in the literature. This lack of a common reference point has created confusion and made comparisons difficult [2]. Recently published consensus criteria for the definition of acute renal failure/acute kidney injury have led to significant changes in how we think about this disorder. Serum Creatinine (SCr), glomerular filtration rate (GFR) and diuresis-based RIFLE criteria providing a uniform definition of acute kidney injury are increasingly used in the literature [3,4] and have been shown to accurately predict survival [5]. These new criteria may reduce the wellknown uncertainty about AKI definition and the lack of treatment standardization [6]. In spite of its well-known limitations, serum creatinine continues to be an integral part of these criteria and is a main determinant in the measurement and management of renal dysfunction: calibrated SCr values, adjusted by gender, age and race by the MDRD formula have been recommended as the best estimation of GFR [7]. International recommendations favor the reporting of creatinine-based estimates of

ß The Author [2007]. Published by Oxford University Press on behalf of ERA-EDTA. All rights reserved. For Permissions, please email: [email protected]


GFR using these formulae [8], and in the estimation of GFR prior to procedures potentially injurious to the kidney such as intravenous contrast studies [9]. Inter-laboratory variation of up to 20% [1,10] add to the uncertainty in the interpretation of SCr as a measure of GFR. Serum creatinine levels are dependent upon age, gender and race [11], muscle mass and nutritional status [1,12], dietary protein intake [13] and fluid volume status. Interpretation of SCr values is especially difficult in patients with variable degrees of fluid overload, because available measures of body fluid volume are extremely inaccurate in critically ill AKI patients [12,14,15]. Serum creatinine levels are also dependent upon GFR, tubular secretion of creatinine, and variations in tubular secretion, such as those induced by drugs. In spite of these limitations, it is well known that relatively ‘small’ increments in SCr are associated with significant increases in the risk of death [16]. For the time being, SCr levels are one of the main determinants worldwide in the decision to initiate renal replacement therapies [3,4]. In an apparent paradox, in disparate AKI populations and contexts, Mehta and Chertow [17], Uchino [18], and Paganini [19,20] have shown that a higher SCr is associated with better patient survival. In these studies, the association was described but not further analysed. We set out to evaluate the association between SCr and survival and to postulate possible mechanisms for this association. We hypothesized that in this setting, the association of higher SCr with better survival may be determined by better nutrition, lesser volume overload or pre-existing chronic kidney disease (CKD). We are using our findings as an opportunity to flag an issue deserving of further analysis, as it affects the interpretation of a variable widely used to make important treatment decisions in the management of critically ill patients with acute kidney injury.

Data The original sample consisted of a homogeneous population of critically ill patients with AKI requiring CRRT (Table 1). In order to address missing data concerns, we first conducted multiple imputation of missing observations on respondent variables through the Sequential Regression Imputation Method, using IVEWARE software [21,22]. We produced five imputed datasets, which we used to conduct the main analyses. We analysed the main determinants of hospital patient survival by simple and multivariable logistic regression analysis (Table 2). In multivariable analysis, Model 1 included SCr and relevant variables including number of pressors, sepsis (defined as in [24]) and multiple organ failure (defined as in [25]), volume

Hypothesis Table 1. Patient characteristics Number of patients Age Percent male Patient weight (Kg)

134 67 (11.5) 63.4 On admission 88.7 (0.7) On start CRRT 97.6 (24.4) Type of patient Medical (%) 43 Surgical (%) 22 Cardiac surgery (%) 35 Sepsis (%) 53.8 Number of OSF Three (%) 48.5 Four (%) 29.1 Number of vasopressors One (%) 30 Two (%) 47 Serum creatinine (mg/dl) On admission 2.39 (24.6) On start CRRT 3.66 (1.7) Serum albumin (g/dl) On admission 2.6 (0.7) On start CRRT 2.2 (0.5) Survival to hospital discharge (%) 29.5 OSF, Organ system failures.  P < 0.001.

excess or deficit, weight change between admission and CRRT and serum albumin at start of CRRT. Model 2 incorporated calculation of admission GFR (MDRD formula [7]) and Model 3 further incorporated calculated Liano scores [26] at initiation of CRRT, to adjust for severity of disease. We used the PROC MIANALYZE procedure in SAS to combine the logistic regression model estimates from the five imputed data sets: parameter estimates were averaged over the set of analyses, while standard errors were computed using the average of the squared standard errors over the set of analyses and the between analysis parameter estimate variation [23]. Simple correlation analysis using Pearson’s correlation coefficient showed that larger BW gains between admission and start of CRRT correlated with lower SCr at start of CRRT (P ¼ 0.03), and lower urine output correlated with lower SCr at start of CRRT (P ¼ 0.0006), both correlations suggesting that overhydration led to lower serum creatinine at the start of CRRT. Other correlations were nonsignificant. Multivariable logistic regression (Table 2) indicated that a higher SCr on initiation of CRRT was monotonically associated with better survival controlling for selected covariates (Model 1: OR 1.438, 95% CI 1.034–1.999). Adjustment for nutritional status and volume did not affect the significance of SCr. Adjustment of the model by preadmission CKD (Model 2) and severity of disease (Liano scores, Model 3), respectively decreased or annulled the significance of SCr levels. Higher Liano scores were strongly associated with worse outcomes (OR 0.08, 95% CI 0.009–0.686) In this pilot preliminary study, sample size limited further analysis of the mechanisms of the association, a study currently in course.



Table 2. Estimated odds of survival by patient characteristics Parameter

SCr start CRRT BUN Type 1 PT Type 2 PT Sepsis # OSF # Pressors Vol status  Weight Alb CRRT MDRD Liano

Model 1

Model 2

Model 3










1.438 0.994 2.261 7.038 0.356 0.722 0.516 0.983 0.975 0.731

1.034 0.980 0.566 1.879 0.123 0.430 0.275 0.818 0.946 0.290

1.999 1.009 9.037 26.355 1.036 1.212 0.968 1.181 1.005 1.841

1.386 0.995 2.342 7.280 0.352 0.732 0.520 0.985 0.976 0.732 0.994

0.981 0.980 0.583 1.948 0.120 0.436 0.278 0.818 0.947 0.285 0.975

1.958 1.009 9.411 27.205 1.030 1.227 1.013 1.186 0.975 1.881 1.006

1.302 0.998 2.248 6.651 0.431 0.827 0.567 1.020 0.980 0.716 0.992 0.080

0.910 0.983 0.527 1.657 0.144 0.475 0.300 0.841 0.951 0.267 0.973 0.009

1.862 1.014 9.597 26.690 1.288 1.439 1.070 1.237 1.010 1.922 1.012 0.686

SCr start CRRT, serum creatinine on start CRRT (mg/dl); BUN, blood urea nitrogen at start of CRRT (mg/dl); Type of Pt: 1, noncardiothoracic surgery; 2, Cardiothoracic surgery patients; OSF, organ system failures; #Pressors, number of vasopressors; Volume status, Estimated volume excess/deficit; Weight, change weight (Kg) between admission and start of CRRT; Alb CRRT, serum albumin at start CRRT (g/dl); MDRD, estimated GFR (ml/min/1.73 m2); Liano, Liano scores.

Implications and future directions In this population of critically ill, virtually anuric patients with AKI, a higher SCr on initiation of CRRT was consistently associated with better survival. Possible explanations for this apparently paradoxical and counterintuitive association include: First, a lower serum creatinine at start of CRRT might indicate fluid overload. As fluid overload has been shown to be associated with worse prognosis in AKI [27] and ARDS [28], volume expansion might explain why a lower creatinine associates with worse survival. Second, it is conceivable that patients with pre-existing CKD may require less ‘burden of disease’ (i.e. a lesser number of organ system failures, sepsis) to reach the point when dialysis becomes necessary. As these patients would be exposed to lesser disease severity, survival might be better than that of patients with initially normal renal function, who require extreme degrees of sickness to reach severe AKI and dialysis. Our multivariable analysis results support this contention, as the significance of the association between SCr and survival decreases when admission GFR is added to the model, and disappears when the model is adjusted for overall severity of disease. Alternatively, AKI superimposed on CKD might have a different prognosis, an issue presently unknown and worthy of further research. Third, it is conceivable that fluid overloaded AKI patients with a small muscle mass may not experience rapid increases in SCr, and hence suffer from late recognition of the degree of renal dysfunction, late initiation of dialysis, and worse outcomes [29]. Evidence shows early dialysis is better [30,31] as demonstrated in numerous recent outcome studies, largely involving convective CRRT [4]. Although these studies provide a clinical rationale for the increasingly common clinical practice of earlier initiation of renal

replacement therapies, the inaccuracy of currently available markers makes the decision difficult. As in other studies, lack of rigid prospective criteria to initiate CRRT impeded such analysis in our sample. Fourth, a higher creatinine is an indicator of better nutrition and muscle mass [32], and therefore might be a surrogate indicator of better health status and survival. Analysis of clinical variables associated with nutrition failed to demonstrate such an association in our sample, but the question is clearly worthy of further analysis. From these findings; we conclude that in AKI, SCr is a late and inaccurate indicator of renal dysfunction: better markers are being developed but their correlation with renal function and survival is still not established. It is possible that patients with underlying CKD may have a lower ‘burden of disease’, receive earlier RRT or, alternatively, acute-on-chronic renal dysfunction may have a different prognosis than AKI. Additionally, changes in SCr may be obscured by malnutrition and fluid overload. Such lateness in measurement may presumably have severe consequences in patient outcome, by perpetuating fluid overload and/or delaying the initiation of renal replacement therapies. All of these possibilities are worthy of thorough investigation, as they will likely result in significant changes in patient outcomes. Conflict of interest statement. The results presented in this article have not been published previously in whole or part, except in abstract form: Cerda J, Cerda M, Prendergast J, Kilcullen P: Higher serum creatinine predicts better survival in acute kidney injury. JASN 2006;17:164A.

References 1. Anavekar N, Bais R, Carney S et al. Chronic kidney disease and automatic reporting of estimated glomerular filtration rate: a position statement. Clin Biochem Rev 2005; 26: 81–86

2784 2. Bellomo R, Kellum JA, Ronco C. Defining and classifying acute renal failure: from advocacy to consensus and validation of the RIFLE criteria. Intensive Care Med 2007; 33: 409–413 3. Hoste EA, Kellum JA. Acute kidney injury: epidemiology and diagnostic criteria. Curr Opin Crit Care 2006; 12: 531–537 4. Clark WR, Letteri JJ, Uchino S, Bellomo R, Ronco C. Recent clinical advances in the management of critically ill patients with acute renal failure. Blood Purif 2006; 24: 487–498 5. Uchino S, Bellomo R, Goldsmith D, Bates S, Ronco C. An assessment of the RIFLE criteria for acute renal failure in hospitalized patients. Crit Care Med 2006; 34: 1913–1917 6. Ricci Z, Ronco C, D’Amico G et al. Practice patterns in the management of acute renal failure in the critically ill patient: an international survey. Nephrol Dial Transplant 2006; 21: 690–696 7. Levey AS, Eckardt KU, Tsukamoto Y et al. Definition and classification of chronic kidney disease: a position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int 2005; 67: 2089–2100 8. Lamb EJ, Tomson CR, Roderick PJ. Estimating kidney function in adults using formulae. Ann Clin Biochem 2005; 42: 321–345 9. Lameire N, Adam A, Becker CR et al. Baseline renal function screening. Am J Cardiol 2006; 98: 21K–26K 10. Van Biesen W, Vanholder R, Veys N et al. The importance of standardization of creatinine in the implementation of guidelines and recommendations for CKD: implications for CKD management programmes. Nephrol Dial Transplant 2006; 21: 77–83 11. Johnson CA, Levey AS, Coresh J, Levin A, Lau J, Eknoyan G. Clinical practice guidelines for chronic kidney disease in adults: Part II. Glomerular filtration rate, proteinuria, and other markers. Am Fam Physician 2004; 70: 1091–1097 12. Ikizler TA, Himmelfarb J. Muscle wasting in kidney disease: Let’s get physical. J Am Soc Nephrol 2006; 17: 2097–2098 13. Preiss DJ, Godber IM, Gunn IR. How to measure renal function in clinical practice: eating cooked meat alters serum creatinine concentration and eGFR significantly. BMJ 2006; 333: 1072 14. Ikizler TA, Sezer MT, Flakoll PJ et al. Urea space and total body water measurements by stable isotopes in patients with acute renal failure. Kidney Int 2004; 65: 725–732 15. Himmelfarb J, Evanson J, Hakim RM, Freedman S, Shyr Y, Ikizler TA. Urea volume of distribution exceeds total body water in patients with acute renal failure. Kidney Int 2002; 61: 317–323 16. Levy EM, Viscoli CM, Horwitz RI. The effect of acute renal failure on mortality. A cohort analysis. JAMA 1996; 275: 1489–1494 17. Mehta RL, Pascual MT, Gruta CG, Zhuang S, Chertow GM. Refining predictive models in critically ill patients with acute renal failure. J Am Soc Nephrol 2002; 13: 1350–1357

Hypothesis 18. Uchino S, Bellomo R, Morimatsu H et al. External validation of severity scoring systems for acute renal failure using a multinational database. Crit Care Med 2005; 33: 1961–1967 19. Paganini EP, Halstenberg WK, Goormastic M. Risk modeling in acute renal failure requiring dialysis: the introduction of a new model. Clin Nephrol 1996; 46: 206–211 20. Paganini EP, Larive B, Kanagasundaram NS. Severity scores and outcomes with acute renal failure in the ICU setting. Contrib Nephrol 2001; 132: 181–195 21. Raghunathan T, Van Hoewyk J, Solenberger P. A multivariate technique for multiply imputing missing values using a sequence of regression models. Survey Methodology 2001; 27 22. Raghunathan T, Van Hoewyk J, Solenberger P. IVEware: Imputation and Variance Estimation Software User Guide. Ann Arbor, MI: University of Michigan; 2002 23. SAS. Chapter 10: The MIANALYZE Procedure. SAS OnlineDoc: Version 8: SAS Institute Inc.; 2007. 24. Bone RC, Balk RA, Cerra FB et al. Definitions for sepsis and organ failure and guidelines for the use of innovative therapies in sepsis. The ACCP/SCCM Consensus Conference Committee. American College of Chest Physicians/Society of Critical Care Medicine. Chest 1992; 101: 1644–1655 25. Chertow GM, Christiansen CL, Cleary PD, Munro C, Lazarus JM. Prognostic stratification in critically ill patients with acute renal failure requiring dialysis. Arch Intern Med 1995; 155: 1505–1511 26. Liano F, Pascual J. Outcomes in acute renal failure. Semin Nephrol 1998; 18: 541–550 27. Goldstein SL, Somers MJ, Baum MA et al. Pediatric patients with multi-organ dysfunction syndrome receiving continuous renal replacement therapy. Kidney Int 2005; 67: 653–658 28. Wiedemann HP, Wheeler AP, Bernard GR et al. Comparison of two fluid-management strategies in acute lung injury. N Engl J Med 2006; 354: 2564–2575 29. Chertow GM, Lazarus JM, Paganini EP, Allgren RL, Lafayette RA, Sayegh MH. Predictors of mortality and the provision of dialysis in patients with acute tubular necrosis. The Auriculin Anaritide Acute Renal Failure Study Group. J Am Soc Nephrol 1998; 9: 692–698 30. Gettings LG, Reynolds HN, Scalea T. Outcome in posttraumatic acute renal failure when continuous renal replacement therapy is applied early vs. late. Intensive Care Med 1999; 25: 805–813 31. Schiffl H. Daily haemodialysis in acute renal failure. Old wine in a new bottle? Minerva Urol Nefrol 2004; 56: 265–277 32. Shlipak MG, Wassel Fyr CL, Chertow GM et al. Cystatin C and mortality risk in the elderly: the health, aging, and body composition study. J Am Soc Nephrol 2006; 17: 254–261 Received for publication: 18.3.07 Accepted in revised form: 25.5.07

Suggest Documents