1 Original Paper Received: June 27, 2013 Accepted: November 29, 2013 Published online: January 11, 2014 Transient versus Persistent Acute Kidney Injury and the Diagnostic Performance of Fractional Excretion of Urea in Critically Ill Patients K.A. Wlodzimirow a A. Abu-Hanna a A.A.N.M. Royakkers d P.E. Spronk e L.S. Hofstra f M.A. Kuiper g M.J. Schultz b, c C.S.C. Bouman b a Department of Medical Informatics, b Department of Intensive Care and c Laboratory of Experimental Intensive Care and Anesthesiology (LEICA), Academic Medical Center, University of Amsterdam, Amsterdam, d Department of Anesthesiology, Zaans Medical Center, Zaandam, e Department of Intensive Care, Lukas Hospital, Gelre Hospitals, Apeldoorn, f Department of Intensive Care, Scheper Hospital, Emmen, and g Department of Intensive Care, Medical Center Leeuwarden, Leeuwarden, The Netherlands Key Words Acute kidney injury Transient Persistent Fractional excretion of urea Intensive care Abstract Aims: To evaluate the performance of fractional excretion of urea (FeU) for differentiating transient (T) from persistent (P) acute kidney injury (AKI) and to assess performance of FeU in predicting AKI in patients admitted to the ICU. Methods: We performed secondary analysis of a multicenter prospective observational cohort study on the predictive performance of biological markers for AKI in critically ill patients. AKI was diagnosed according to RIFLE staging. Results: Of 150 patients, 51 and 41 patients were classified as having T- AKI and P-AKI, respectively. The diagnostic performance for FeU to discriminate T-AKI from P-AKI on the day of AKI was poor (AUC-ROC = 0.61; 95% CI: ). The diagnostic performance of FeU to predict AKI 1 and 2 days prior to AKI was poor as well (AUC-ROC = 0.61; 95% CI: , and 0.58; 95% CI: , respectively). Conclusions: FeU does not seem to be helpful in differentiating T- from P-AKI in critically ill patients and it is a poor predictor of AKI S. Karger AG, Basel /14/ $39.50/ S. Karger AG, Basel Introduction Acute kidney injury (AKI) remains associated with high morbidity and mortality [1, 2]. The incidence among critically ill patients reaches 35%, with an in-hospital mortality rate above 50% when AKI develops as a part of multiple organ dysfunction . Early recognition of AKI and identification of those at high risk for worsening may ultimately improve outcomes [4, 5]. The RIFLE (Risk, Injury, Failure, Loss, End-Stage Kidney Disease) and AKIN (Acute Kidney Injury Network criteria) methods for AKI in critically ill patients, based on changes in serum creatinine (SCr) and urine output (UO) were a step forward in AKI diagnosing. Both methods, however, have limited predictive power [6, 7]. Discriminating transient (T-) from persistent (P-) AKI could have advantages [8, 9], e.g. in optimization of treatment such as limiting fluids or starting of renal replacement therapy (RRT) . Fractional excretion of urea (FeU) is suggested to have discriminative power with respect to T-AKI and P-AKI , and unlike the fractional excretion of sodium it is not affected by concomitant diuretic use. Little is known, Kama Wlodzimirow Academic Medical Center, University of Amsterdam Department of Medical Informatics, Suite J1B-127 PO box 22700, NL 1100 DE Amsterdam (The Netherlands) amc.uva.nl
2 however, about its utility in critically ill patients [12, 13]. We hypothesized that (1) FeU differentiates T- from P- AKI on the first day of AKI, and (2) FeU can predict AKI in the days preceding AKI according to RIFLE. M e t h o d s The study was conducted as part of a multicenter prospective observational cohort study on cystatin C as a predictor for AKI. Detailed methods have been described previously . Briefly, the study included adult patients who were expected to need mechanical ventilation for at least 48 h and/or have an ICU stay of at least 72 h. Exclusion criteria were chronic RRT, preadmission treatment with corticosteroids (>5 mg prednisone, or equivalent, per day), treatment with plasmapheresis during ICU admission and participation in another clinical trial. Patients data were documented upon ICU admittance. Blood and urine sampling were performed on the day of admission, next day and alternating days until RRT was started or ICU discharge. The UO was measured hourly. Patients were scored daily for the presence of AKI based on the RIFLE criteria . Baseline SCr was defined as the lower of the premorbid and ICU admission values. When premorbid SCr (within 1 year prior to hospital admission) was unknown, it was estimated using the Modification of Diet in Renal Disease (MDRD) equation assuming a GFR of 75 ml/min/1.73 m 2. The first day of AKI was defined as the day of the first RIFLE event and was termed day 0. The 2 days prior to this day were termed day 2 and day 1, respectively. Patients were classified as having P-AKI when AKI was sustained for more than 3 days according to RIFLE or when the patient died with AKI within 3 days [11, 15]. The RIFLE class cannot be determined once continuous venovenous hemofiltration (CVVH) is started; therefore, we classified patients who received CVVH within 3 days after day 0 as P-AKI. Patients were classified as having T-AKI Table 1. Baseline characteristics No AKI (n = 58) P-AKI (n = 41) T-AKI (n = 51) p Age, years 59 ± ± 9 70 ± 14 < Men 37 (64) 26 (63) 34 (65) 0.99 BMI 25 ± 4 27 ± 7 26 ± APACHE II 18 ± 9 26 ± ± SAPS II 37 ± ± ± 12 < Admission type 0.60 Surgical 23 (40) 20 (49) 21 (41) Medical 35 (60) 21 (51) 30 (59) Comorbidity Hypertension 15 (26) 21 (51) 17 (33) 0.04 Diabetes mellitus 6 (10) 8 (20) 6 (12) 0.42 Chronic renal failure 1 2 (3) 7 (17) Admission diagnosis 0.25 Cardiovascular failure 2 (3) 4 (10) 4 (8) Cerebrovascular event 2 (3) 0 0 Hemorrhagic shock 7 (12) 5 (12.5) 3 (6) Multiple trauma 4 (7) 1 (2.5) 2 (4) Peripheral vascular surgery 1 (2) 3 (7.5) 3 (6) Cardiopulmonary surgery 1 (2) 1 (2.5) 1 (2) Respiratory failure 23 (39) 6 (15) 19 (36) Septic shock 18 (31) 21 (51) 19 (37) Renal function at admission UO, ml/24 h 2,437 ± 1,583 1,397 ± 974 2,211 ± 1, SCr, μmol/l 75 ± ± ± 41 < Serum urea, mmol/l 8 ± 4 16 ± 10 9 ± 5 < ICU stay, days 5 (3 8) 7 (4 14) 6 (4 11.5) 0.04 Outcome RRT 0 13 (32) 1 (2) < ICU mortality 0 14 (34) 2 (4) < Hospital mortality 4 (7) 19 (46) 6 (12) < Values represent means ± SD, n (%), or medians (IQR). p represents comparisons across the three patient groups. 1 GFR <45 ml/min. Diagnostic Performance of FeU in Critically Ill Patients 9
3 when AKI resolved within 3 days without initiation of CVVH. Recovery from AKI was defined as no AKI according to the RI- FLE criteria. From urea and creatinine values in serum and urine we calculated the FeU percentage as follows: [urinary urea/serum urea]/ [urinary creatinine/scr] 100. Table 2. Characteristics on the first day of AKI diagnosis P-AKI (n = 41) T-AKI (n = 51) Renal function UO, ml/24 h 1,630 ± 1,761 2,335 ± 1, SCr, μmol/l 186 ± ± 42 < Serum urea, mmol/l 16 ± ± 7 < FeU, % 31 ± ± AKI severity using RIFLE classification < Risk 14 (34) 46 (90) Injury 15 (37) 4 (8) Failure 12 (29) 1 (2) RIFLE criteria < SCr 21 (51) 26 (51) UO 6 (15) 24 (47) Both 14 (34) 1 (2) Values represent means ± SD or n (%). p S t a t i s t i c s Patients baseline characteristics are presented as n (%), means ± SD or medians (IQR). Data were compared using a t test or Mann- Whitney U test, where appropriate. Categorical variables were compared using a χ 2 or Fisher s exact test. A Kruskal-Wallis test was used to compare continuous variables across the three groups. p < 0.05 was considered to be significant. A logistic regression analysis was performed to investigate the association with mortality. An ROC curve was created to determine the diagnostic performance of FeU to distinguish T-AKI from P-AKI. Test characteristics were calculated. The optimal cutoff point of the ROC curve was defined as the value that maximized the sum of sensitivity and specificity. ROC curves were plotted to determine the predictive value of FeU in the days before AKI diagnosis. Statistical analyses were performed in the statistical environment R version . R e s u l t s One hundred and fifty patients out of 151 from the original study were included in the present analysis. One patient was excluded because of initiation of CVVH for intoxication. Fifty-nine patients never met the criteria for AKI, 57 patients presented with AKI on ICU admission and 35 patients developed AKI after ICU admission. In these latter patients, AKI was diagnosed after a median of 1 day (IQR 1 2). Of the 57 patients included with AKI at ICU admission, 30, 15 and 12 patients were classified as AKI risk, -injury and -failure, respectively. Patient characteristics are summarized in table 1. Fifty-one out of 92 (55%) AKI patients were classified as having T-AKI and 41 (45%) were classified as having P-AKI ( table 2 ). Patients with P-AKI had more severe AKI on day 0 compared with T-AKI (p < ). In T-AKI patients, AKI diagnosis was based on UO criteria only in 46% of the patients, compared to 15% of P-AKI patients (p = 0.002). The type of AKI was significantly associated with mortality: patients with P-AKI had times (95% CI: ) the odds to die than those with T-AKI. After adjusting for age, gender, BMI, admission type, Acute Physiology and Chronic Health Evaluation (APACHE) II score, Simplified Acute Physiology Score (SAPS), hypertension, diabetes and chronic renal failure, only APACHE II and SAPS were confounders. After adjustment for APACHE II, the OR decreased to 8.52 (95% Table 3. Performance of FeU for detecting patients with P-AKI among patients with AKI FeU <38% FeU <35% FeU <40% Sensitivity, % 0.34 ( ) 0.42 ( ) 0.24 ( ) Specificity, % 0.56 ( ) 0.49 ( ) 0.56 ( ) Positive predictive value 0.41 ( ) 0.42 ( ) 0.32 ( ) Negative predictive value 0.49 ( ) 0.49 ( ) 0.45 ( ) Positive likelihood ratio 0.77 ( ) 0.82 ( ) 0.54 ( ) Negative likelihood ratio 1.18 ( ) 1.19 ( ) 1.37 ( ) AUC 0.61 ( ) 95% CI in parentheses. 10 Wlodzimirow et al.
4 FeU (%) Color version available online Sensitivity AUC = 0.61 AUC = Day AKI Non-AKI False-positive rate Day 2 Day Fig. 1. Time course of FeU from 2 days prior to AKI for the patients developing AKI after entry and from entry for the non-aki group. Values are means and the standard error of the means. Fig. 2. ROC demonstrating the performance of FeU for the prediction of AKI on days 2 and 1 prior to AKI. AUC is depicted. CI: ; p < 0.01). After adjustment for SAPS, the OR reduced to 6.95 (95% CI: ; p = 0.02). The diagnostic performance of FeU to discriminate T- AKI from P-AKI was poor (AUC = 0.61; 95% CI: ). Test characteristics of FeU at various cutoff levels are shown in table 3. The optimal cutoff point of the ROC curve corresponded to 38%. Figure 1 compares the trend of FeU starting from 2 days prior to AKI for the patients developing AKI after entry with the non-aki trend. FeU levels did not differ significantly between the groups on days 2, 1 and 0. The diagnostic performance of FeU to predict AKI on the 2 days preceding AKI was poor (AU- ROC = 0.58; 95% CI: , and 0.61; 95% CI: , on days 2 and 1, respectively; fig. 2 ). Discussion Diagnostic Performance of FeU in Critically Ill Patients In this secondary analysis of a multicenter prospective observational cohort study in an ICU population, FeU measured on the first day of AKI diagnosis performed poorly in discriminating T-AKI from P-AKI. In addition, FeU on the 2 days prior to AKI diagnosis did not predict AKI. Two our knowledge only two other studies have investigated the performance of FeU in distinguishing P-AKI from T-AKI in ICU patients [11, 15]. Our results are concordant to a recent multicenter study in a heterogeneous ICU population by Darmon et al. , which concluded that FeU is not helpful in differentiating T-AKI from P- AKI. In contrast, Dewitte et al.  in a single-center study concluded that an FeU of less than 40% was a sensitive and specific index for differentiating T-AKI from P-AKI. The different findings among the studies are probably caused by differences in study design, AKI definition and case mix. Darmon et al.  included only the patients with AKI on the day of ICU admission, while Dewitte et al.  and the present study also included patients developing AKI after ICU admission. In the present study the definition of recovery of oliguria (>0.5 ml/kg/h) was independent of the use of diuretics, while in the other two studies UO normalization had to be in the absence of diuretics. All three studies used SCr of 150% from baseline in their AKI definition; however, Dewitte et al.  used the admission SCr as the baseline while the other two studies used the MDRDbased baseline when premorbid SCr was unavailable . Our findings also support the results of the study by Pepin et al.  concluding that FeU cannot be used for differentiating T-AKI from P-AKI. That study was somewhat different than our study and the studies by Darmon et al.  and Dewitte et al.  because AKI was not defined according to the RIFLE criteria and included not only ICU patients. 11
5 FeU has been reported to be less effective in patients with infection, as cytokines interfere with the urea transporters in the kidney and colon . Unfortunately, our dataset did not allow for a subgroup analysis in patients without sepsis because of small patient numbers and because patients in the original study were not scored for the presence of sepsis. The admission diagnosis of our patients was extracted from the Dutch National Intensive Care Evaluation (NICE) registry. A high percentage of patients (39%) were referred to the ICU because of septic shock failure; however, 27% of patients had respiratory failure as the referral diagnosis, and it is very likely that some patients in this group also had sepsis. Notably, Darmon et al.  performed a subgroup analysis of the sepsis patients and showed no difference compared with the results of the whole study group. Moreover, in a very recent study comparing the urine biochemistry in septic and non-septic AKI, FeU was not significantly different between groups . Notably, in our study the cohort with P-AKI had significantly more patients with CKD (17.5%) as compared to the T-AKI cohort (no CKD). This may affect results because patients with CKD may have abnormal tubular function. However, subanalysis in patients without CKD showed that diagnostic performance of FeU remained poor both to predict AKI (AUROC = 0.57; 95% CI: , and 0.61; 95% CI: , on days 2 and 1, respectively) and to distinguish P-AKI from T-AKI (AUC = 0.62; 95% CI: ). In the present study, the P-AKI type was associated with mortality even after adjusting for severity of illness. This is not surprising as patients with P-AKI had a higher degree of AKI according to RIFLE, and large epidemiologic studies have shown an independent and stepwise increase in mortality as AKI severity increases [3, 21]. In addition, patients with T-AKI more frequently had AKI based on RIFLE urine criteria only compared with P-AKI (46 vs. 15%, p < 0.003). This finding supports the findings of previous studies showing that the use of RIFLE without the UO criteria was associated with less AKI and a higher mortality in comparison with the use of RIFLE with both SCr and UO criteria [22, 23]. Of note, in our study 1 patient classified as having T- AKI received CVVH. This patient developed T-AKI in the first week of ICU admission. After his initial recovery, CVVH was started during a new period of AKI in the second week of ICU admission. During the study, 6 patients died within 3 days of developing AKI and were classified as P-AKI because they all died with AKI. Classifying the deceased as P-AKI may 12 be questionable and we therefore performed additional analysis. Exclusion of the 6 patients who were classified as P-AKI based on death within 3 days of AKI did not change the results; the diagnostic performance of FeU remained poor both to discriminate T-AKI from P-AKI (AUC = 0.60; 95% CI: ) and to predict AKI on the 2 days preceding AKI (AUROC = 0.55; 95% CI: , and 0.58; 95% CI: , on days 2 and 1, respectively). In the present study we used the RIFLE classification because it was used in our previous analysis. However, the most recent guideline, Kidney Disease Improving Global Guidelines (KDIGO), differs from the RIFLE classification, particularly in stage 1 by adding a 0.3 mg/dl ( 26.5 mmol/l) increase of SCr within 48 h and in stage 3 adding initiation of RRT. We converted our data according to KDIGO guidelines and it resulted in 2 patients changing from T-AKI to P-AKI, and 1 patient had AKI 1 day earlier compared to the RIFLE classification. We performed additional analysis classifying patients according to KDIGO and we concluded that the diagnostic performance of FeU to discriminate T-AKI from P-AKI was poor (AUC = 0.60; 95% CI: ), and to predict AKI on the 2 days preceding AKI was poor as well (AUROC = 0.60; 95% CI: , and 0.61; 95% CI: , on days 2 and 1, respectively). Our study has several limitations. First, the study protocol was designed to evaluate the performance of urinary and systemic cystatin C to predict AKI, and in this secondary analysis we retrospectively assessed the performance of FeU to distinguish P-AKI from T-AKI. In the original protocol, specimens were collected on alternate days after the first 2 days. As a result, FeU on day 0 was not always available (28%) in patients who developed AKI after ICU admission. The poor performance of FeU to predict AKI in this subgroup may therefore be related to low statistical power. Second, patients were assigned retrospectively to a diagnosis of P-AKI or T-AKI based on renal function evolution. This evolution, however, may have been caused by factors occurring after day 0. Third, the daily fluid balance was not recorded in the original study and thus we could not analyze the association of fluid balance on AKI type and/or mortality. Fourth, we did not record whether our patients had received diuretics before sampling and therefore we could not study the effects of diuretics. In previous studies, however, the exclusion of patients with diuretic use did not improve FeU performance in distinguishing T-AKI from P-AKI [15, 18]. Wlodzimirow et al.
6 In summary, FeU is not helpful in differentiating T- AKI from P-AKI in a heterogeneous ICU population and FeU has no predictive value for AKI in the 2 days preceding AKI. Disclosure Statement The authors declare that they have no competing interests. References 1 Uchino S, Kellum JA, Bellomo R, Doig GS, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Ronco C: Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005; 294: Wald R, Quinn RR, Luo J, Li P, Scales DC, Mamdani MM, Ray JG: Chronic dialysis and death among survivors of acute kidney injury requiring dialysis. JAMA 2009; 302: Ostermann M, Chang RW: Acute kidney injury in the intensive care unit according to RIFLE. Crit Care Med 2007; 35: Devarajan P: Review: neutrophil gelatinaseassociated lipocalin: a troponin-like biomarker for human acute kidney injury. Nephrology (Carlton) 2010; 15: Lee JB, Forni LG: Recently published papers: acute kidney injury diagnosis and treatment. Crit Care 2009; 13: 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: Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin A: Acute Kidney Injury Network: report of an initiative to improve outcomes in acute kidney injury. Crit Care 2007; 11:R31. 8 Gotfried J, Wiesen J, Raina R, Nally JV Jr: Finding the cause of acute kidney injury: which index of fractional excretion is better? Cleve Clin J Med 2012; 79: Sutton TA, Fisher CJ, Molitoris BA: Microvascular endothelial injury and dysfunction during ischemic acute renal failure. Kidney Int 2002; 62: Labib M, Khalid R, Khan A, Khan S: Volume management in the critically ill patient with acute kidney injury. Crit Care Res Pract 2013; 2013: Dewitte A, Biais M, Petit L, Cochard JF, Hilbert G, Combe C, Sztark F: Fractional excretion of urea as a diagnostic index in acute kidney injury in intensive care patients. J Crit Care 2012; 27: Carvounis CP, Nisar S, Guro-Razuman S: Significance of the fractional excretion of urea in the differential diagnosis of acute renal failure. Kidney Int 2002; 62: Kaplan AA, Kohn OF: Fractional excretion of urea as a guide to renal dysfunction. Am J Nephrol 1992; 12: Royakkers AA, Korevaar JC, van Suijlen JD, Hofstra LS, Kuiper MA, Spronk PE, Schultz MJ, Bouman CS: Serum and urine cystatin C are poor biomarkers for acute kidney injury and renal replacement therapy. Intensive Care Med 2011; 37: Darmon M, Vincent F, Dellamonica J, Schortgen F, Gonzalez F, Das V, Zeni F, Brochard L, Bernardin G, Cohen Y, Schlemmer B: Diagnostic performance of fractional excretion of urea in the evaluation of critically ill patients with acute kidney injury: a multicenter cohort study. Crit Care 2011; 15:R R Development Core Team: R: A Language and Environment for Statistical Computing. Vienna, R Foundation for Statistical Computing, Bagshaw SM, Uchino S, Cruz D, Bellomo R, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, Gibney N, Tolwani A, Oudemans-van Straaten HM, Ronco C, Kellum JA: A comparison of observed versus estimated baseline creatinine for determination of RIFLE class in patients with acute kidney injury. Nephrol Dial Transplant 2009; 24: Pepin MN, Bouchard J, Legault L, Ethier J: Diagnostic performance of fractional excretion of urea and fractional excretion of sodium in the evaluations of patients with acute kidney injury with or without diuretic treatment. Am J Kidney Dis 2007; 50: Schmidt C, Hocherl K, Bucher M: Regulation of renal glucose transporters during severe inflammation. Am J Physiol Renal Physiol 2007; 292:F804 F Bagshaw SM, Bennett M, Devarajan P, Bellomo R: Urine biochemistry in septic and nonseptic acute kidney injury: a prospective observational study. J Crit Care 2013; 28: Hoste EA, Clermont G, Kersten A, Venkataraman R, Angus DC, De BD, Kellum JA: RI- FLE criteria for acute kidney injury are associated with hospital mortality in critically ill patients: a cohort analysis. Crit Care 2006; 10:R Lopes JA, Jorge S: Comparison of RIFLE with and without urine output criteria for acute kidney injury in critically ill patients: a task still not concluded! Crit Care 2013; 17: Wlodzimirow KA, Abu-Hanna A, Slabbekoorn M, Chamuleau RA, Schultz MJ, Bouman CS: A comparison of RIFLE with and without urine output criteria for acute kidney injury in critically ill patients. Crit Care 2012; 16:R200. Diagnostic Performance of FeU in Critically Ill Patients 13