Test Characteristics of Urinary Biomarkers Depend on Quantitation Method in Acute Kidney Injury

CLINICAL RESEARCH www.jasn.org Test Characteristics of Urinary Biomarkers Depend on Quantitation Method in Acute Kidney Injury Azrina Md Ralib,* John W. Pickering,* Geoffrey M. Shaw,* Prasad Devarajan, Charles L. Edelstein, Joseph V. Bonventre, and Zoltan H. Endre* *Christchurch Kidney Research Group, Department of Medicine, University of Otago, Christchurch, New Zealand; Department of Nephrology and Hypertension, Cincinnati Children s Hospital Medical Center, University of Cincinnati College of Medicine, Cincinnati, Ohio; Division of Nephrology, University of Colorado, Denver, Colorado; Renal Division, Department of Medicine, Brigham and Women s Hospital, Harvard Medical School, Boston, Massachusetts; and Department of Nephrology, Prince of Wales Clinical School, University of New South Wales, Sydney, Australia ABSTRACT The concentration of urine influences the concentration of urinary biomarkers of AKI. Whether normalization to urinary creatinine concentration, as commonly performed to quantitate albuminuria, is the best method to account for variations in urinary biomarker concentration among patients in the intensive care unit is unknown. Here, we compared the diagnostic and prognostic performance of three methods of biomarker quantitation: absolute concentration, biomarker normalized to urinary creatinine concentration, and biomarker excretion rate. We measured urinary concentrations of alkaline phosphatase, g-glutamyl transpeptidase, cystatin C, neutrophil gelatinase associated lipocalin, kidney injury molecule 1, and IL-18 in 528 patients on admission and after 12 and 24 hours. Absolute concentration best diagnosed AKI on admission, but normalized concentrations best predicted death, dialysis, or subsequent development of AKI. Excretion rate on admission did not diagnose or predict outcomes better than either absolute or normalized concentration. Estimated 24-hour biomarker excretion associated with AKI severity, and for neutrophil gelatinase associated lipocalin and cystatin C, with poorer survival. In summary, normalization to urinary creatinine concentration improves the prediction of incipient AKI and outcome but provides no advantage in diagnosing established AKI. The ideal method for quantitating biomarkers of urinary AKI depends on the outcome of interest. J Am Soc Nephrol 23: 322 333, 2012. doi: 10.1681/ASN.2011040325 The concentration of urinary biomarkers of AKI is influenced by variation in urinary concentration within and between individuals. 1 In diabetes mellitus, the urinary albumin:creatinine ratio is an accepted method of accounting for variation in albumin concentration arising from variations in urine flow rate due to hydration, diuresis, or concentration changes induced by antidiuretic hormone or tubular injury. 2,3 Consequently, many studies investigating urinary biomarker utility for the detection of AKI have normalized biomarkers to creatinine. 4 6 The accuracy of this method is compromised by tubular secretion of creatinine and variations in urine creatinine excretion in non-steady state when the GFR changes. 1,3,7 Creatinine excretion may also vary because of diurnal creatinine production, physical activity, emotional stress, diet, muscle mass (hence age, sex, and body weight variation), and disease state. 1,8,9 Arecent theoretical analysis showed that normalizing to creatinine may briefly amplify the biomarker signal soon after a reduction in GFR. 1 To date, relatively few studies have evaluated biomarker performance by both absolute and normalized concentration, and there is no consensus on how data should be Received April 1, 2011. Accepted September 15, 2011. Published online ahead of print. Publication date available at www.jasn.org. Correspondence: Dr. Zoltan Endre, Department of Nephrology, Prince of Wales Hospital, High Street, Randwick, Sydney, NSW 2031, Australia. Email: Maureen.Sullivan@sesiahs.health.nsw.gov.au Copyright 2012 by the American Society of Nephrology 322 ISSN : 1046-6673/2302-322 J Am Soc Nephrol 23: 322 333, 2012

www.jasn.org CLINICAL RESEARCH reported. This hinders the comparison of biomarkers between trials and adds uncertainty regarding how biomarkers should be utilized in clinical practice. Alternatives to normalizing biomarker concentration to creatinine include using the absolute concentration or quantifying the excretion rate. Intuitively, the excretion rate may also account for variation in water reabsorption and urine flow rate. In addition, total biomarker excretion in AKI might more accurately reflect the mass of injured tubular cells, a function of both severity and duration, parameters associated with long-term mortality. 10 12 We compared the three biomarker quantitation methods, the absolute and normalized concentrations, and the excretion rate in the diagnosis of AKI, prediction of AKI, death, and the need for renal replacement therapy (RRT) in patients admitted to two general intensive care units (ICUs). In addition, we assessed the association of estimated total biomarker excretion over 24 hours to AKI severity and mortality. RESULTS Demographic Profile Of the 528 recruited patients, 484 had 4-hour creatinine clearance measurements on ICU admission from which urine output volumes could be obtained for the calculation of biomarker excretion rates. These patients comprised the analyzed cohort (Table 1). Of patients with missing clearances, 30 were a subcohort of cardiothoracic surgical patients in whom no clearance on admission to ICU postoperatively was planned, five had collection errors, and nine missed samples. Patients with AKI at anytime within 48 hours of admission were older and had higher baseline plasma creatinine, severity scores, and lengths of stay in the ICU. AKI was more common in those with abdominal aortic aneurysm rupture and repair (P,0.0001) and in sepsis (P=0.002), and was less common in those with neurologic events (P=0.001). Of the 484 patients, we measured the following on admission: g-glutamyl transpeptidase (GGT) and alkaline phosphatase (AP) in 484, IL-18 and kidney injury molecule-1 (KIM-1) in 481, cystatin C (CysC) in 480, and neutrophil gelatinase associated lipocalin (NGAL) in 449 patients. Compared with those with NGAL measurement, patients with missing NGAL data had more AKI on admission (P,0.05) but no differences in dialysis, subsequent AKI postadmission (AKIN48 or RIFLE24), or mortality. In this study, AKIN48 refers to the Acute Kidney Injury Network (AKIN) definition at any time within 48 hours. RIFLE24 refers to sustained AKI as defined by RIFLE (Risk, Injury, Failure, Loss, ESRD) criteria that occur for.24 hours at any time within 7 days. Because there were no differences in biomarker performance with or without the inclusion of the erythropoietin (EPO) cohort, patients triaged to receive EPO were included to increase power. 13 Table 2 presents data from all patients for the absolute and normalized biomarker concentration, excretion rate on admission to the ICU, and estimated total excretion over 24 hours (24-hour excretion) of all biomarkers. There were significant differences between the absolute and normalized concentrations and the excretion of CysC, NGAL, IL-18, and KIM-1 between those with and without AKI on admission. Biomarker Quantitation Method Comparison Figure 1 depicts the difference between the area under the receiver-operator characteristic (ROC) curve for the normalized concentration (referent value) and the area under the curve (AUC) for the absolute concentration and the excretion rate for each outcome measure. For a diagnosis of AKI on admission, absolute concentration performed better than the normalized concentration or excretion rate (Figure 1 and Table 3). The AUCs for absolute concentration were greater than for both normalized concentration and excretion rate for all biomarkers (P#0.003) except KIM-1. For prediction of 7-day and 365-day mortality, normalizing to creatinine produced the highest AUCs. Normalized concentrations of AP, GGT, and NGAL better predicted RRT than their excretion rates (P#0.04). Medians (interquartile range) of urine creatinine concentrations were greater in those who had AKI on admission compared with those without AKI (7.4 [3.5 10.8] versus 4.9 [2.2 8.7] mmol/l, P,0.0001). Patients with AKI also had lower urine flow rates compared with those without AKI (0.99 [0.54 1.8] versus 1.44 [0.80 2.8] ml/min, P,0.0001). AP and GGT were not diagnostic of AKI or prognostic of outcome (Table 3). Nevertheless, the pattern of the AUC differences between the absolute and normalized concentrations and the excretion rate of AP and GGTwere consistent with the other biomarkers. The time courses of biomarker excretion rates from putative time of insult were similar to those of normalized concentrations as previously published (Supplemental Figure 1). 14 In the cohort of patients without AKI on admission (n=339), there were no differences in performance between absolute and normalized concentrations in the prediction of AKI within 48 hours (AKIN48) or sustained AKI within 7 days (RIFLE24). However, the normalized concentrations had higher AUCs than the excretion rates (P#0.01) (Figure 1). The creatinine excretion rate on admission was lower in those who developed AKI within 48 hours compared with those who did not (6.264.0 versus 8.164.6 mmol/min, P=0.001). The excretion rate increased in this cohort over the first 48 hours (repeated measures ANOVA, P=0.003) (Figure 2). Forty-eight hours later, urine creatinine excretion rates were similar in both groups (8.065.3 versus 9.165.4 mmol/min, P=0.16). Estimated Total Biomarker Excretion over 24 Hours The 24-hour excretion of CysC, NGAL, IL-18, and KIM-1 increased with the increasing severity of injury stage (maximum AKIN) reached within 48 hours (one-way ANOVA, P,0.0001) (Figure 3). Differences between successive AKIN stages of increasing severity were demonstrated for NGAL J Am Soc Nephrol 23: 322 333, 2012 Biomarker Quantitation Method 323

CLINICAL RESEARCH www.jasn.org Table 1. Demographic profile Parameter Cohort (N=484) No AKIN48 (n=266) alone (post hoc Fisher s least significant difference analysis, P#0.02). The 24-hour excretion of CysC, NGAL, and IL-18 was higher in those with sepsis than those without (P,0.0001), and the association with AKI severity was maintained in the sepsis cohort (Supplemental Figures 2 and 3). AKIN48 (n=218) Age (yr) 60617 57617 63617 0.001 Male sex 294 (60.7) 158 (59.4) 136 (62.4) 0.50 Weight (kg) 79619 77618 82620 0.01 APACHE II score 1866 1766 1966,0.0001 SOFA score 6.362.8 5.562.5 7.362.9,0.0001 Baseline plasma 72 (60 90) 70 (60 86) 80 (60 98) 0.01 creatinine (mmol/l) a AKI on admission 145 (30) stage 1 112 stage 2 25 stage 3 8 AKI within 48 h (including 218 (45) on admission) stage 1 146 stage 2 44 stage 3 28 Renal replacement therapy 12 (2.5) 0 (0) 12 (100),0.0001 Mortality at 7 d 49 (10.1) 21 (7.9) 28 (12.8) 0.07 Mortality at 365 d 119 (24.6) 63 (23.4) 56 (25.7) 0.61 Length of ICU stay (h) 78 (44 180) 65 (40 157) 95 (49 194) 0.001 Sepsis 97 (20.0) 40 (15.0) 57 (26.1) 0.002 Primary diagnosis class abdominal aortic aneurysm 22 (4.5) 4 (1.5) 18 (8.3),0.0001 rupture and repair abdominal surgery 51 (10.5) 31 (11.7) 20 (9.2) or inflammation burns 5 (1.0) 3 (1.1) 2 (0.9) cardiac arrest or failure 63 (13.0) 38 (14.3) 25 (11.5) cardiac surgery 64 (13.2) 32 (12.0) 32 (14.7) collapse (unknown cause) 3 (0.6) 1 (0.4) 2 (0.9) neurologic event 71 (14.7) 52 (19.5) 19 (8.7) respiratory failure 63 (13.0) 36 (13.5) 27 (12.4) or thoracic surgery sepsis 97 (20.0) 40 (15.0) 57 (26.1) trauma 42 (8.7) 26 (9.8) 16 (7.3) other 3 (0.6) 3 (1.1) 0 (0) Baseline comorbidities hypertension 163 (33.7) 80 (30.1) 83 (38.1) 0.06 congestive cardiac failure 17 (3.5) 6 (2.3) 11 (5.0) 0.10 ischemic heart disease 110 (22.7) 54 (20.3) 56 (25.7) 0.16 chronic obstructive 54 (11.2) 24 (9.0) 30 (13.8) 0.10 airways disease asthma 34 (7.0) 17 (6.4) 17 (7.8) 0.55 cerebral vascular accident 30 (6.2) 16 (6.0) 14 (6.4) 0.85 diabetes mellitus 52 (10.7) 24 (9.0) 28 (12.8) 0.18 Data expressed as mean 6 SD, n (%), or median (lower quartile upper quartile). a Baseline plasma creatinine was determined from a chart review as previously described. 13 Creatinine measurements before ICU admission were used if available (n=229). When the measurements were not available, the lowest creatinine at follow-up (n=38), last creatinine in the ICU (n=138), or creatinine on admission to the ICU (n=79) was used. P Patient survival was assessed by the extent of biomarker excretion over 24 hours ranked by tertiles (Figure 4). Only NGAL demonstrated a significant association between 24-hour excretion and survival over 365 days (log-rank test, P=0.04). We adjusted for age, sex, sepsis, EPO therapy, and Acute Physiology, Age, Chronic Health Evaluation II (APACHE II) and Sequential Organ Failure Assessment (SOFA) severity scores. After adjustment, patients with greater NGAL excretion (higher tertile, 24-hour excretion.184 mg) had a higher mortality at 365 days compared with those with lower NGAL excretion (lower tertile, 24-hourexcretionof,40 mg) (hazard ratio [HR], 2.10; 95% confidence interval [95% CI], 1.21 3.65; P=0.009) (Table 4). The 24-hour excretion of NGAL predicted 30-day and 365-day mortality, whereas CysC predicted 30-day mortality only (Table 4). To determine the added contribution of the excretion data to known clinical predictors of mortality, NGAL and CysC were added separately to a multivariable logistic regression that included age, sex, sepsis, APACHE II and SOFA scores, EPO therapy, and AKIN48. NGAL remained independently predictive of 365-day mortality (odds ratio, 1.43; 95% CI, 1.03 1.99), along with age and APACHE II score (Table 5). However, the inclusion of NGAL only marginally improved the AUC of the base model from 0.66 (0.61 0.72) to 0.68 (0.62 0.74) (P=0.31). For prediction of 30-day mortality, the addition of NGAL increased the AUC from 0.61 (0.53 0.69) to 0.66 (0.58 0.74) (P=0.035), and the addition of CysC increased the AUC from 0.60 (0.52 0.68) to 0.64 (0.57 0.72) (P=0.042). DISCUSSION Individual variation in water handling by surviving nephrons may compromise the utility of urinary biomarker concentration in the detection and prediction of AKI and mortality. Normalizing to urine creatinine and measuring the excretion rate can be used to account for variations in water reabsorption. Waikar et al. 1 recently suggested that the excretion rate may provide a better measure of biomarker performance. This is the first study to address this hypothesis directly. For all injury biomarkers, absolute 324 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 322 333, 2012

www.jasn.org CLINICAL RESEARCH Table 2. Urinary biomarkers on ICU admission Urinary Biomarker All Patients AKI on Admission No AKI on Admission P AP (n=484) (n=145) (n=339) normalized (U/mmol Cr) 0.87 (0.47 1.7) 0.71 (0.35 1.8) 0.94 (0.51 1.7) 0.06 concentration (U/L) 4.1 (2.0 9.0) 5.0 (2.1 11) 4.0 (2.0 8.3) 0.22 ER ( 310 23 U/min) 6.1 (3.1 11) 4.8 (2.1 12) 6.3 (3.7 11) 0.08 24-h excretion (U) 8.4 (5.0 14) 6.9 (4.1 13) 8.7 (5.4 14) 0.01 GGT (n=484) (n=145) (n=339) normalized (U/mmol Cr) 13 (7.1 26) 17 (7.6 44) 11 (7.0 22) 0.001 concentration (U/L) 74 (28 160) 115 (51 290) 58 (53 130),0.0001 ER (310 23 U/min) 85 (44 180) 110 (48 240) 78 (42 150) 0.01 24-h excretion (U) 120 (75 210) 130 (72 260) 120 (78 200) 0.38 CysC (n=480) (n=144) (n=336) normalized (mg/mmol Cr) 0.02 (0.01 0.19) 0.06 (0.01 0.61) 0.02 (0.01 0.10),0.0001 concentration (mg/l) 0.12 (0.05 1.0) 0.44 (0.09 2.6) 0.09 (0.04 0.37),0.0001 ER (mg/min) 0.16 (0.06 1.13) 0.42 (0.09 3.6) 0.12 (0.06 0.59),0.0001 24-h excretion (mg) 0.23 (0.10 2.0) 0.88 (0.19 5.9) 0.17 (0.08 0.72),0.0001 NGAL (n=449) (n=129) (n=320) normalized (mg/mmol Cr) 7.7 (2.3 41) 20 (5.1 99) 5.6 (1.8 24),0.0001 concentration (ng/ml) 47 (11 260) 149 (40 500) 26 (6.7 140),0.0001 ER (ng/min) 56 (15 300) 170 (33 760) 38 (13 150),0.0001 24-h excretion (mg) 78 (2 310) 220 (69 740) 56 (21 190),0.0001 IL-18 (n=481) (n=144) (n=337) normalized (ng/mmol Cr) 0.54 (0.001 35) 9.7 (0.001 84) 0.001 (0.001 23),0.0001 concentration (pg/ml) 4.4 (0.05 240) 110 (0.05 600) 0.05 (0.05 156),0.0001 ER (pg/min) 1.2 (0.1 260) 84 (0.1 600) 0.1 (0.1 160),0.0001 24-h excretion (ng) 68 (0.1 300) 115 (2.3 620) 36 (0.14 240),0.0001 KIM-1 (n=481) (n=145) (n=337) normalized (ng/mmol Cr) 86 (38 220) 140 (59 380) 74 (31 180),0.0001 concentration (pg/ml) 500 (130 1300) 870 (350 3500) 360 (100 1000),0.0001 ER (pg/min) 580 (240 1600) 953 (375 2994) 470 (200 1100),0.0001 24-h excretion (ng) 1800 (710 3800) 2700 (1200 4900) 1500 (560 3200),0.0001 Urine creatinine (mmol/l) 5.5 (2.6 9.5) 7.4 (3.5 10.8) 4.9 (2.2 8.7),0.0001 Data are expressed as median (lower quartile upper quartile). Cr, creatinine; concentration, absolute biomarker concentration; normalized, normalized biomarker concentration; ER, biomarker excretion rate; 24-h excretion, estimated total biomarker excretion over 24 hours. concentration performed best in the diagnosis of AKI, whereas the normalized concentration performed best for the prediction of death, dialysis, and subsequent development of AKI. The excretion rate on admission was not a better diagnostic or prognostic biomarker than the absolute or normalized concentration. Biomarker Quantitation Method Comparison For all biomarkers, normalizing to creatinine or measuring the excretion rate diagnosed AKI on ICU admission more poorly than the absolute concentration, which may be explained as follows. In evolving AKI, the creatinine excretion rate decreases because the GFR decreases. As plasma creatinine increases, creatinine excretion rates increase, asymptoting toward the original rate before the GFR decreases (Figure 5). The data support this explanation. Excluding patients with obvious AKI on admission, creatinine excretion on admission was lower in those who developed AKI within 48 hours compared with those who did not. Creatinine excretion then increased over the first 48 hours to the level at admission, similarly to those who had not developed AKI at all (Figure 2). In theory, an acute loss of GFR will result in an immediate decrease in filtered creatinine but should have no effect on induced (e.g., IL-18 and KIM-1) or preformed (AP and GGT) urinary biomarker excretion. However, in the case of NGAL, final urinary concentration will be the complex sum of both distal tubular induction and release as well as glomerular filtration and impaired proximal reabsorption. 16 Consequently, normalization to urine creatinine will amplify urinary biomarker signal immediately after GFR reduction, enhancing the prediction of incipient AKI (AKIN48 or RIFLE24; Figure 1); however, because creatinine excretion soon returns to baseline (Figure 2), amplification is likely to be minimal by the time AKI is detected by an elevated plasma creatinine. 1 On the other hand, an immediate decrease in filtered creatinine will be paralleled by an immediate decrease in filtration of other freely filtered biomarkers (e.g., CysC), effectively negating the amplification of such biomarkers. A high urinary concentration of such biomarkers (e.g., CysC and NGAL) at this time point indicates the presence proximal tubular injury which impairs reabsorption of these biomarkers. 16 J Am Soc Nephrol 23: 322 333, 2012 Biomarker Quantitation Method 325

CLINICAL RESEARCH www.jasn.org Figure 1. The difference in AUCs of the absolute concentration and excretion rate with the normalized concentration (referent value) on admission to the ICU for multiple outcomes. Horizontal lines depict the AUC 95% CI. AKIN48, $0.3 mg/dl or $50% increase in plasma creatinine above baseline at any time within 48 hours; RIFLE24 (sustained AKI), $50% increase in plasma creatinine above baseline for.24 hours during 7 days. # Calculated for the cohort without AKI on admission (n=339). *P,0.05 for the difference between the AUC of absolute concentration or excretion rate and the normalized concentration. The urine creatinine concentration was higher in those who had AKI on admission compared with those without AKI. Consequently, when biomarker concentrations are normalized, the relative differences between the non-aki and AKI cohorts are reduced, resulting in fewer true positives and a reduced AUC (Figure 6). A higher urine creatinine concentration could be due to increased tubular secretion of creatinine 7 or, more likely, to increased reabsorption of water (decreased free water clearance) because of an increased antidiuretic hormone activity. Whereas the biomarker concentration was higher in the AKI than the non-aki cohort, the urinary flow rate was lower, consistent with the second alternative. Therefore, the ratios of the AKI to non-aki biomarker excretion rate and of the normalized concentration were lower than the ratio of the absolute concentration; this combination reduced the AUCs. Normalized concentration also best predicted hard outcomes such as mortality and the need for dialysis. Without the bias introduced by the presence of AKI on one side of the ROC equation (e.g., among survivors compared with nonsurvivors, in which the ratio of AKI to non-aki patients was similar), there was improvement in predictive performance with respecttosubsequentaki,mortality,andneedforrrt.this most likely resulted from the normalization accounting for the kidney concentrating capacity in the denominator (urine creatinine) and capturing cellular injury in the numerator (biomarker concentration). In the prediction of the development of subsequent AKI (AKIN48 or RIFLE24), the AUC for all biomarkers was greater for the normalized concentration compared with the absolute concentration or excretion rate, irrespective of whether the AUC was significantly.0.5. This may be due to decreased excretion of urine creatinine as a result of reduced GFR, with resulting amplification of the signal of these biomarkers. 1 Normalization to urine creatinine amplifies the biomarker signal when there is a change of GFR, irrespective of whether the biomarker is itself increased by injury. Although it may potentially be clinically useful as an amplified signal, using normalized values alone (i.e., without knowing the absolute concentration) conceals the mechanism of biomarker signal increase. We suggest reporting both absolute and 326 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 322 333, 2012

www.jasn.org CLINICAL RESEARCH Table 3. Biomarker performance (AUC) in patients with and without a AKI on admission Urinary Biomarker AKI on ICU Admission AKIN48 a RIFLE24 a RRT Mortality at 7 d Mortality at 365 d AP n 145 73 22 12 49 119 normalized 0.45 (0.39 0.50) 0.53 (0.46 0.61) 0.66 (0.54 0.78) 0.60 (0.42 0.77) 0.62 (0.54 0.71) 0.59 (0.53 0.65) concentration 0.54 (0.48 0.59) 0.49 (0.42 0.56) 0.55 (0.43 0.68) 0.50 (0.33 0.67) 0.55 (0.46 0.64) 0.57 (0.51 0.63) ER 0.45 (0.39 0.50) 0.44 (0.36 0.51) 0.46 (0.33 0.59) 0.49 (0.33 0.66) 0.54 (0.46 0.63) 0.54 (0.48 0.60) GGT n 145 73 22 12 49 119 normalized 0.59 (0.54 0.65) 0.55 (0.48 0.62) 0.59 (0.49 0.69) 0.55 (0.38 0.72) 0.65 (0.56 0.73) 0.56 (0.50 0.62) concentration 0.66 (0.60 0.71) 0.49 (0.41 0.57) 0.50 (0.37 0.62) 0.49 (0.32 0.65) 0.57 (0.48 0.66) 0.54 (0.48 0.60) ER 0.59 (0.53 0.64) 0.44 (0.36 0.51) 0.35 (0.23 0.47) 0.44 (0.28 0.59) 0.56 (0.47 0.64) 0.50 (0.44 0.56) CysC n 144 72 22 12 49 119 normalized 0.59 (0.53 0.65) 0.57 (0.50 0.65) 0.72 (0.60 0.84) 0.88 (0.75 1.00) 0.66 (0.57 0.75) 0.57 (0.51 0.63) concentration 0.68 (0.62 0.73) 0.53 (0.45 0.60) 0.65 (0.52 0.77) 0.86 (0.72 0.99) 0.62 (0.53 0.71) 0.57 (0.51 0.63) ER 0.59 (0.54 0.65) 0.52 (0.44 0.59) 0.61 (0.48 0.74) 0.85 (0.71 0.99) 0.63 (0.54 0.71) 0.55 (0.49 0.61) NGAL n 129 67 20 10 47 111 normalized 0.66 (0.61 0.72) 0.54 (0.46 0.62) 0.69 (0.57 0.80) 0.84 (0.68 0.99) 0.66 (0.58 0.75) 0.60 (0.54 0.66) concentration 0.71 (0.66 0.77) 0.51 (0.43 0.60) 0.62 (0.49 0.74) 0.82 (0.65 0.98) 0.62 (0.53 0.71) 0.58 (0.53 0.65) ER 0.67 (0.61 0.72) 0.50 (0.42 0.58) 0.60 (0.48 0.73) 0.80 (0.63 0.96) 0.64 (0.55 0.73) 0.58 (0.52 0.64) IL-18 n 145 73 22 12 49 119 normalized 0.61 (0.56 0.67) 0.55 (0.47 0.63) 0.75 (0.63 0.87) 0.80 (0.65 0.95) 0.67 (0.58 0.75) 0.56 (0.50 0.62) concentration 0.63 (0.57 0.69) 0.52 (0.44 0.59) 0.70 (0.57 0.82) 0.80 (0.65 0.95) 0.64 (0.55 0.73) 0.55 (0.49 0.62) ER 0.60 (0.54 0.65) 0.51 (0.43 0.58) 0.68 (0.55 0.81) 0.81 (0.66 0.96) 0.65 (0.56 0.74) 0.55 (0.49 0.61) KIM-1 n 145 73 22 12 49 119 normalized 0.66 (0.60 0.71) 0.54 (0.46 0.61) 0.61 (0.48 0.74) 0.69 (0.52 0.86) 0.56 (0.47 0.64) 0.58 (0.52 0.64) concentration 0.68 (0.63 0.73) 0.51 (0.44 0.59) 0.56 (0.43 0.68) 0.65 (0.47 0.82) 0.50 (0.42 0.59) 0.56 (0.50 0.62) ER 0.65 (0.59 0.70) 0.47 (0.39 0.54) 0.46 (0.34 0.58) 0.61 (0.44 0.78) 0.48 (0.40 0.57) 0.54 (0.48 0.60) Data represent AUC of the ROC curve with 95% confidence intervals. AKI on ICU admission, AKIN definition on ICU admission; AKIN48, AKIN definition at any time within 48 hours; RIFLE24, sustained AKI as defined by RIFLE criteria that occur for.24 hours at any time within 7 days; normalized, normalized concentration; concentration, absolute concentration; ER, excretion rate. a In the cohort without AKI on admission. normalized concentrations to facilitate understanding of the biomarker data. Biomarker prediction of RIFLE24 was generally better than for AKIN48. RIFLE24 represents, on average, more severe AKI than AKIN48. This is because in the majority of patients, the change in GFR is greater and more sustained because an increase of 50% requires a larger change in GFR than an increase of 0.3 mg/dl if baseline plasma creatinine is.0.6 mg/dl (87.4% of patients in this cohort). In addition, AKIN diagnosis requires only a single time point, whereas RIFLE24 requires at least two time points 24 hours apart. The pattern of relative performance of absolute concentration and excretion rate to normalized concentration was similar for all biomarkers, even when the biomarkers were not significantly diagnostic or prognostic. The individual performance of these biomarkers in this cohort was discussed in detail in our recent publication. 14 The generally poor performance of these markers remains an important issue that must be addressed. This poor performance compared with earlier studies can be explained by the heterogeneity of the cohort with respect to time, cause of injury, and baseline renal function. 14 Normalizing to urine creatinine or measuring excretion rate does not turn a poor biomarker into a good one, or a good biomarker into a better one. Given the small differences observed with each approach, the issue of whether to normalize seems academic, at least with these markers. Assuming that an ideal biomarker is available, the question arises as to what properties or temporal patterns of expression such a biomarker needs, given the pattern of creatinine excretion in injury. There is no easy answer to this question. The more we understand about biomarker performance in heterogeneous groups, the harder it is to provide a simple solution, such as normalization. An ideal biomarker is probably one that changes measurably in real time with the degree of injury. However, if we use new high-sensitivity troponins as an analogous example, the absence of a baseline in most patients means that serial measurements remain essential to exclude moderate myocardial injury. We speculate that the same conclusion will be reached with biomarkers of injury in AKI. Nevertheless, this study demonstrates that the optimal method of assessing urinary AKI J Am Soc Nephrol 23: 322 333, 2012 Biomarker Quantitation Method 327

CLINICAL RESEARCH www.jasn.org mortality, normalization to urine creatinine was best. Although the choice whether to normalize to urinary creatinine may be made according to whether a clinician is attempting to diagnose existing AKI or prognose future AKI, it is likely that both measurements would be made for many patients in whom the timing of injury is unknown (e.g., sepsis patients in the ICU). Figure 2. Creatinine excretion rate over time from ICU admission. After exclusion of patients with AKI on admission, creatinine excretion rate on admission was lower in those who developed AKI within 48 hours compared with those who did not (*P=0.001). The excretion rate later increased over the first 48 hours (repeated measures ANOVA, P=0.003). Forty-eight hours later, the excretion rates were similar in both groups (P=0.16). biomarkers depends on the outcome being assessed. For diagnosis of AKI on admission to ICU, absolute concentration was the best method. However, for prediction of AKI, or for Estimated Total Biomarker Excretion over 24 hours The 24-hour excretion of CysC, NGAL, IL-18, and KIM-1 was positively associated with AKI severity. Differences between successive AKIN stages of increasing severity were demonstrated only for NGAL. Presumably, total excretion of NGAL is a reflection of the mass of injured tubular cells in AKI. Increasing concentration of urinary NGAL has been previously associated with severity of AKI in critically ill children and cardiac surgery patients. 17 19 The added value of calculating 24-hour excretion is that it integrates change over time, which captures the severity and duration of injury. In addition to severity of injury, inclusion of duration of injury increased prediction of long-term mortality in diabetes and postcardiac surgery patients. 10 12 Because it captures both severity and duration of injury, we postulated that 24-hour excretion was predictive of long-term mortality. Of the six biomarkers, only 24-hour excretion of NGAL was associated with long-term mortality. Patients with a higher 24-hour excretion of NGAL (above a threshold of.184 mg) were 2.1 times more likely to die within 365 days compared with those with a lower excretion. NGAL also remained independently Figure 3. Estimated total biomarker excretion over 24 hours compared with maximum AKIN severity stage within 48 hours. The boxes show the median and interquartile ranges, whereas the whiskers show the 10th 90th percentile in each stage. One-way ANOVA significance levels: (A) AP, P=0.04; (B) GGT, P=0.048; (C) NGAL, P,0.0001; (D) CysC, P,0.0001; (E) IL-18, P,0.0001; and (F) KIM-1, P,0.0001. *Different from previous stage with post hoc Fisher s least squares difference analysis (P,0.05). 328 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 322 333, 2012

www.jasn.org CLINICAL RESEARCH Figure 4. Kaplan Meier survival plots of tertiles of estimated total biomarker excretion over 24 hours. Log-rank test: (A) AP, P=0.74; (B) GGT, P=0.52; (C) NGAL, P=0.04; (D) CysC, P=0.48; (E) IL-18, P=0.30; and (F) KIM-1, P=0.45. Tertiles of Cys-C and IL-18 were not equal due to more than one third of patients having concentrations below the level of detectability (lower tertile of CysC [n=168] and IL-18 [n=195]). The n in each graph denotes the number of patients who survived at 365 days. Table 4. Prediction of mortality in highest versus lowest estimated 24-hour excretion tertile Biomarker Mortality at 30 d Mortality at 365 d HR (95% CI) a P HR (95% CI) P AP 1.83 (0.99 3.39) 0.05 1.54 (0.92 2.56) 0.10 GGT 0.54 (0.15 1.96) 0.35 0.75 (0.47 1.20) 0.75 CysC 3.32 (1.61 6.81) 0.001 0.81 (0.47 1.38) 0.43 NGAL 2.71 (1.27 5.76) 0.01 2.10 (1.21 3.65) 0.009 KIM-1 0.82 (0.43 1.53) 0.52 0.72 (0.44 1.19) 0.20 IL-18 1.37 (0.76 2.46) 0.29 1.20 (0.74 1.95) 0.45 a HR of the highest tertile of 24-hour excretion compared with the lowest tertile, calculated with Cox regression model, and adjusted for age, sex, sepsis, APACHE II and SOFA scores, and EPO treatment. predictive of long-term mortality. To our knowledge, this is the first large study to examine AKI biomarkers predicting long-term outcomes at 365 days. For a more proximate mortality, both NGAL and CysC were associated with poorer survival at 30 days. Furthermore, the addition of NGAL and CysC to other mortality risk factors improved the predictive performance. If these results are supported in larger multicenter studies and across a broad demographic base, then 24-hour excretion of CysC and NGAL may be useful as outcome measures in AKI intervention trials. This is analogous to the use of the relative average value of creatinine (RAVC) metric of functional change in randomized trials in preference to categorical metrics such as AKIN or RIFLE. 20 The RAVC captures both Table 5. Multivariable logistic regression for prediction of 365-day mortality after inclusion of estimated 24-hour NGAL excretion Variable Odds Ratio (95% CI) P Age 1.02 (1.00 1.03) 0.04 Sex 1.10 (0.69 1.78) 0.68 Sepsis 1.60 (0.81 3.15) 0.17 APACHE II score 1.08 (1.04 1.13),0.0001 SOFA score 0.95 (0.85 1.05) 0.29 EPO treatment 0.72 (0.39 1.30) 0.28 Serum creatinine increase 1.13 (0.68 1.87) 0.63.0.3 mg/dl within 48 h Estimated 24-h excretion of NGAL (Log 10 ) 1.43 (1.03 1.99) 0.03 extent and duration of functional change in a single variable. A mild correlation between 24-hour excretion with RAVC at 24 hours was observed, which further supports a relationship between the extent of injury and functional loss (Supplemental Table 1). The analogy is that biomarker excretion can integrate duration and severity of injury, whereas RAVC integrates both duration and severity of functional change. Sepsis is the leading contributing factor to AKI in the ICU. 21,22 Because sepsis seems to influence urinary CysC, 23 IL-18, 24,25 and NGAL, 26,27 we investigated the association between severity of AKI in sepsis and nonsepsis. The 24-hour excretion of CysC, J Am Soc Nephrol 23: 322 333, 2012 Biomarker Quantitation Method 329

CLINICAL RESEARCH www.jasn.org Figure 5. Schematic of the effect of an abrupt decrease in function in a 70-kg male patient with a production rate of creatinine of 1 mg/min and an initial GFR of 100 ml/min. (A) Plasma creatinine increases slowly. (B) GFR decreases suddenly. (C) Urine creatinine excretion rate initially decreases and then recovers to the same rate as before when plasma creatinine increases. This is because the creatinine excretion rate is the product of plasma creatinine and creatinine clearance (approximated by GFR). Based in part on Figure 2 in Moran and Myers. 15 Adapted by permission from Macmillan Publishers Ltd: [KIDNEY INTERNATIONAL] (Ref. 15), copyright 1985. IL-18, and NGAL was greater in the sepsis cohort than in the nonsepsis cohort, and these biomarkers remained associated with AKI and AKI severity in the sepsis cohort. In the presence of sepsis and inflammation, plasma concentrations of IL-18 and NGAL increased due to release from activated macrophages, Kupffer cells, activated neutrophils, and injured epithelia. 28 30 It is possible that increased plasma concentrations of IL-18 and NGAL lead to increased urinary concentrations. We previously showed that increased urinary CysC concentrations are independently associated with sepsis in this cohort. 23 Furthermore, increased urinary NGAL and CysC in sepsis could result from reduced reabsorption due to competition with albumin for megalin transport at the proximal tubule. 16,31,32 Study Limitations This study has several limitations. First, NGAL measurements were not available in 7% of patients. Because there was a higher proportion of patients with AKI on admission in this 7% than in the remaining 93% of patients, some caution is required in comparing the performance of NGAL in the diagnosis of AKI with the other biomarkers. Second, the urinary flow rate measurements are potentially inaccurate. Urine flows over 4 hours were recorded at three separate time points. Averaging urine flow over a longer period would limit the observation of rapid changes in biomarker excretion rate, whereas collection over a shorter time period might enhance the effect of fluid loading ordiuresison urine flow rate and hence confound the interpretation of the biomarker excretion rate. Third, because the excretion rate is calculated from the urine flow rate, it is already a corrected term and accounts for variation in water reabsorption and urine flow rate. This form of normalization is perhaps best visualized as the time taken to acquire sufficient volume to measure flow (4 hours in our case) so it is not a snapshot in time like the absolute concentration or the normalized concentration. Fourth, serum creatinine based definitions of AKI were used in this study despite known limitations of creatinine. The association of 24-hour excretion with AKI severity might be better demonstrated by histology (extent of structural injury) or by inulin clearance (extent of functional loss). Fifth, the choice of baseline plasma creatinine is known to affect determination of AKI outcomes. 33 In a subcohort of patients with known baselines (available from pre- ICU admission data), the relative differences in biomarker AUCs between normalized, absolute, and excretion rate generally follow the same patterns as for the entire cohort (Supplemental Table 2). Sixth, in 79 patients with no pre- ICU baseline creatinine, the lowest creatinine was on admission to the ICU and was used as a baseline, which required these patients to be classified as non-aki on admission. It is possible that some of these patients were misclassified. Finally, we only measured the biomarkers within the first 24 hours. Estimating the total excretion over longer periods of time (e.g., over 48 72 hours and with more frequent sampling) may provide a clearer association with long-term outcome. Normalizing to urinary creatinine improved the performance of urinary AKI biomarkers in predicting the development of AKI, death, and need for dialysis. The normalized urinary biomarker concentration thus remains the method of choice for early detection of incipient AKI. However, when AKI had already been established long enough to increase plasma creatinine, absolute concentration alone was a better marker of injury than the normalized concentration. The 4-hour excretion rate did not improve performance in the diagnosis of AKI; however, estimated 24-hour biomarker excretion was strongly associated with AKI severity and, for NGAL and CysC, with survival. These associations with severity and survival suggest that the 24-hour excretion may be a useful surrogate outcome measurement in clinical trials of AKI. The ideal method for standardizing urinary AKI biomarkers depends on the outcome being assessed. However, observational studies should present results on the basis of both absolute and normalized concentrations and, where possible, excretion rates. 330 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 322 333, 2012

www.jasn.org CLINICAL RESEARCH Figure 6. The effect of normalizing to urinary creatinine when the average urinary creatinine concentration was higher in the AKI than the non-aki cohort. (A) NGAL absolute and normalized concentrations (log scale) for AKI and non-aki cohorts. The horizontal lines are the median, lower, and upper quartiles in each group. For the absolute concentration, the upper quartile for the non-aki group was slightly lower than the median for the AKI group (dotted line). After normalization, the upper quartile for the non-aki group was between the median and upper quartile of the AKI group. The increased overlap of non-aki and AKI values results in the lower AUC (B). CONCISE METHODS This was a retrospective analysis of urinary biomarker data from the two-center Early Intervention in Acute Renal Failure randomized, controlled trial of high-dose EPO for AKI prevention in the ICU. 13,14 This study was approved by the multiregional ethics committee of New Zealand (MEC/ 050020029) and registered under the Australian and New Zealand Clinical Trials Registry (ACTRN012606000032550; http://www.actr. org.au). Urinary concentrations of AP, GGT, CysC, NGAL, IL-18, KIM-1, and creatinine were measured on admission to the ICU (time 0) and at 12 and 24 hours postadmission (Figure 7). Plasma creatinine was measured daily for up to 7 days. AP, GGT, CysC, and creatinine concentrations were assayed immediately as described previously. 13,23 NGAL, KIM-1, and IL-18 samples were stored at 280 C until batch analysis using a NGAL ELISA Kit 036 (AntibodyShop, Grusbakken, Denmark), 17 microsphere-based xmap technology (Luminex, Austin, TX, USA) with polyclonal antibodies raised against the human KIM-1 ectodomain, 34,35 and a human IL-18 ELISA kit (Medical and Biologic Laboratories, Nagoya, Japan), 36 respectively. The normalized biomarker concentrations were derived by dividing the biomarker concentration by the urinary creatinine concentration (millimoles per liter). The 4-hour urine creatinine collections were collected from 0 4 hours, 12 16 hours, and 24 28 hours (Figure 7). An average urine hourly flow rate for each interval was calculated. Urinary biomarker excretion rates were derived by multiplication of urinary biomarker concentrations at times 0, 12, and 24 hours by the average urinary flow rate over each subsequent 4-hour period. The excretion rate between these periods was linearly interpolated. The total biomarker excretion over 24 hours (24-hour excretion) for each biomarker (i.e., integration of excretion rate with respect to time) was calculated using the trapezoidal rule (the sum of shaded areas A and B, Figure 7). Ameasuredbaselineplasmacreatininewas determined from a chart review as previously described. 13 Briefly, creatinine measurements before ICU admission were used if available (n=229). When the measurements were not available, the lowest creatinine at follow-up (n=38), last creatinine in the ICU (n=138), or creatinine on admission to the ICU (n=79) was used. AKI status and maximum severity stage reached were determined according to the AKIN definition ($0.3 mg/dl or $50% increase in plasma creatinine above baseline) on admission to the ICU (AKI on ICU admission) and any time within 48 hours (AKIN48). 37 Sustained AKI was defined by the RIFLE criteria ($50% increase in plasma creatinine for.24 hours at any time within 7 days) J Am Soc Nephrol 23: 322 333, 2012 Biomarker Quantitation Method 331

CLINICAL RESEARCH www.jasn.org ACKNOWLEDGMENTS Figure 7. Time course for urine spot sampling and 4-hour collection. The average excretion rate was calculated from these three time points. Estimated total excretion over 24 hours was calculated using the trapezoidal rule (the sum of shaded areas A and B). (RIFLE24). 38 Other outcomes were as follows: the need for RRTwithin 7 days and death within 7, 30, and 365 days. The performance of absolute and normalized concentrations and excretion rate on admission to the ICU in the diagnosis or prediction of outcome was assessed by comparison of the AUC of ROC for each parameter using the DeLong method. 39,40 For the assessment of prediction of AKI outcomes (AKIN48 and RIFLE24), the cohort was restricted to patients without AKI on admission (n=339). The association between the estimated total biomarker excretion over the first 24 hours with severity of injury and 365-day mortality was also determined. Severity of injury was defined by both a categorical variable (maximum AKIN stage 37 within 48 hours of ICU admission inclusive of those with AKI on admission) and a continuous variable (relative average value of creatinine or RAVC). 20 A sensitivity analysis of patients stratified according to the presence of sepsis was conducted. Sepsis was defined by the ICU physicians and required two or more systemic inflammatory response syndrome criteria, as well as suspected or confirmed bacterial or viral infection. 41 Statistical Analyses Statistical analysis was performed using PASW software (version 18.0; IBM, Somers, NY) and PRISM 5.0 software (GraphPad, La Jolla, CA). 13 All of the CIs presented are 95%. The 24-hour excretion was log-transformed before conducting ANOVA for association with AKIN severity stages. Post hoc Fisher s least significant difference analyses were performed for all significant associations with ANOVA (P,0.05). Associations of 24-hour excretion with RAVC were analyzed by Spearman s correlation. Tertiles of 24-hour excretion for each biomarker were determined for Kaplan Meier and Cox regression survival analyses. HRs for each tertile were calculated relative to the lowest tertile (HR =1), and were adjusted for age, sex, sepsis, APACHE II and SOFA scores, and EPO therapy. Predictions of mortality using multivariable logistic regression were calculated using a base model that included age, sex, sepsis, APACHE II and SOFA scores, EPO therapy, and AKIN48 (increase in creatinine of.0.3 mg/dl from baseline within 48 hours). Comparison of the AUC using the base model with and without inclusion of biomarker excretion was performed with the DeLong method. 40 We gratefully acknowledge the dedication of John Dean, Jill Robinson, Robyn Hutchison, and the nursing staff of Canterbury and Dunedin hospitals and the Canterbury Health Laboratory. The Early Intervention in Acute Renal Failure trial was supported by the Health Research Council of New Zealand Grant 05/131 (Early Intervention in Acute Renal Failure). A.M.R. is the recipient of a Ministry of Higher Education Malaysia and International Islamic University Malaysia scholarship. J.W.P. is the recipient of a Fast Start grant from the Marsden Fund, administered by the Royal Society of New Zealand, and was supported by an Australian-New Zealand Society of Nephrologists infrastructure and enabling grant. Part of this manuscript was presented in poster form at the World Congress of Nephrology, April 8 12, 2011, Vancouver, Canada. DISCLOSURES P.D. is a co-inventor on patents involving NGAL as a biomarker of chronic and acute kidneydisease, C.L.E. is a co-inventor on patents involving IL-18, and J.V.B. is a co-inventor on patents involving KIM-1. REFERENCES 1. Waikar SS, Sabbisetti VS, Bonventre JV: Normalization of urinary biomarkers to creatinine during changes in glomerular filtration rate. Kidney Int 78: 486 494, 2010 2. Levey AS, Eckardt KU, Tsukamoto Y, Levin A, Coresh J, Rossert J, De Zeeuw D, Hostetter TH, Lameire N, Eknoyan G: Definition and classification of chronic kidney disease: A position statement from Kidney Disease: Improving Global Outcomes (KDIGO). Kidney Int 67: 2089 2100, 2005 3. Goldstein SL: Urinary kidney injury biomarkers and urine creatinine normalization: A false premise or not? Kidney Int 78: 433 435, 2010 4. Westhuyzen J, Endre ZH, Reece G, Reith DM, Saltissi D, Morgan TJ: Measurement of tubular enzymuria facilitates early detection of acute renal impairment in the intensive care unit. Nephrol Dial Transplant 18: 543 551, 2003 5. Mishra J, Dent C, Tarabishi R, Mitsnefes MM, Ma Q, Kelly C, Ruff SM, Zahedi K, Shao M, Bean J, Mori K, Barasch JM, Devarajan P: Neutrophil gelatinase-associated lipocalin (NGAL) as a biomarker for acute renal injury after cardiac surgery. Lancet 365: 1231 1238, 2005 6. Nickolas TL, O Rourke MJ, Yang J, Sise ME, Canetta PA, Barasch N, Buchen C, Khan F, Mori K, Giglio J, Devarajan P, Barasch J: Sensitivity and specificity of a single emergency department measurement of urinary neutrophil gelatinase-associated lipocalin for diagnosing acute kidney injury. AnnInternMed148: 810 819, 2008 7. Perrone RD, Madias NE, Levey AS: Serum creatinine as an index of renal function: New insights into old concepts. Clin Chem 38: 1933 1953, 1992 8. Greenblatt DJ, Ransil BJ, Harmatz JS, Smith TW, Duhme DW, Koch- Weser J: Variability of 24-hour urinary creatinine excretion by normal subjects. J Clin Pharmacol 16: 321 328, 1976 9. Heymsfield SB, Arteaga C, McManus C, Smith J, Moffitt S: Measurement of muscle mass in humans: Validity of the 24-hour urinary creatinine method. Am J Clin Nutr 37: 478 494, 1983 10. Coca SG, King JT Jr, Rosenthal RA, Perkal MF, Parikh CR: The duration of postoperative acute kidney injury is an additional parameter predicting long-term survival in diabetic veterans. Kidney Int 78:926 933, 2010 332 Journal of the American Society of Nephrology J Am Soc Nephrol 23: 322 333, 2012

www.jasn.org CLINICAL RESEARCH 11. Uchino S, Bellomo R, Bagshaw SM, Goldsmith D: Transient azotaemia is associated with a high risk of death in hospitalized patients. Nephrol Dial Transplant 25: 1833 1839, 2010 12. Brown JR, Kramer RS, Coca SG, Parikh CR: Duration of acute kidney injury impacts long-term survival after cardiac surgery. Ann Thorac Surg 90: 1142 1148, 2010 13. Endre ZH, Walker RJ, Pickering JW, Shaw GM, Frampton CM, Henderson SJ, Hutchison R, Mehrtens JE, Robinson JM, Schollum JBW, Westhuyzen J, Celi LA, McGinley RJ, Campbell IJ, George PM: Early intervention with erythropoietin does not affect the outcome of acute kidney injury (the EARLYARF trial). Kidney Int 77: 1020 1030, 2010 14. Endre ZH, Pickering JW, Walker RJ, Devarajan P, Edelstein CL, Bonventre JV, Frampton CM, Bennett MR, Ma Q, Sabbisetti VS, Vaidya VS, Walcher AM, Shaw GM, Henderson SJ, Nejat M, Schollum JB, George PM: Improved performance of urinary biomarkers of acute kidney injury in the critically ill by stratification for injury duration and baseline renal function. Kidney Int 79: 1119 1130, 2011 15. Moran SM, Myers BD: Course of acute renal failure studied by a model of creatinine kinetics. Kidney Int 27: 928 937, 1985 16. Nejat M, Hill JV, Pickering JW, Edelstein CL, Devarajan P, Endre ZH: Albuminuria increases cystatin C excretion: implications for urinary biomarkers [published online ahead of print May 5, 2011]. Nephrol Dial Transplant doi:10.1093/ndt/gfr222 17. Bennett M, Dent CL, Ma Q, Dastrala S, Grenier F, Workman R, Syed H, Ali S, Barasch J, Devarajan P: Urine NGAL predicts severity of acute kidney injury after cardiac surgery: A prospective study. Clin J Am Soc Nephrol 3: 665 673, 2008 18. Zappitelli M, Washburn KK, Arikan AA, Loftis L, Ma Q, Devarajan P, Parikh CR, Goldstein SL: Urine neutrophil gelatinase-associated lipocalin is an early marker of acute kidney injury in critically ill children: A prospective cohort study. Crit Care 11: R84, 2007 19. Du Y, Zappitelli M, Mian A, Bennett M, Ma Q, Devarajan P, Mehta R, Goldstein SL: Urinary biomarkers to detect acute kidney injury in the pediatric emergency center. Pediatr Nephrol 26: 267 274, 2011 20. Pickering JW, Frampton CM, Endre ZH: Evaluation of trial outcomes in acute kidney injury by creatinine modeling. Clin J Am Soc Nephrol 4: 1705 1715, 2009 21. Bagshaw SM, Uchino S, 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 Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators: Septic acute kidney injury in critically ill patients: Clinical characteristics and outcomes. Clin J Am Soc Nephrol 2: 431 439, 2007 22. 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 Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators: Acute renal failure in critically ill patients: A multinational, multicenter study. JAMA 294: 813 818, 2005 23. Nejat M, Pickering JW, Walker RJ, Westhuyzen J, Shaw GM, Frampton CM, Endre ZH: Urinary cystatin C is diagnostic of acute kidney injury and sepsis, and predicts mortality in the intensive care unit. Crit Care 14: R85, 2010 24. Siew ED, Ikizler TA, Gebretsadik T, Shintani A, Wickersham N, Bossert F, Peterson JF, Parikh CR, May AK, Ware LB: Elevated urinary IL-18 levels at the time of ICU admission predict adverse clinical outcomes. Clin J Am Soc Nephrol 5: 1497 1505, 2010 25. Washburn KK, Zappitelli M, Arikan AA, Loftis L, Yalavarthy R, Parikh CR, Edelstein CL, Goldstein SL: Urinary interleukin-18 is an acute kidney injury biomarker in critically ill children. Nephrol Dial Transplant 23: 566 572, 2008 26. Mårtensson J, Bell M, Oldner A, Xu S, Venge P, Martling C-R: Neutrophil gelatinase-associated lipocalin in adult septic patients with and without acute kidney injury. Intensive Care Med 36: 1333 1340, 2010 27. Bagshaw SM, Bennett M, Haase M, Haase-Fielitz A, Egi M, Morimatsu H, D amico G, Goldsmith D, Devarajan P, Bellomo R: Plasma and urine neutrophil gelatinase-associated lipocalin in septic versus non-septic acute kidney injury in critical illness. Intensive Care Med 36: 452 461, 2010 28. Oberholzer A, Steckholzer U, Kurimoto M, Trentz O, Ertel W: Interleukin-18 plasma levels are increased in patients with sepsis compared to severely injured patients. Shock 16: 411 414, 2001 29. Schmidt-Ott KM, Mori K, Li JY, Kalandadze A, Cohen DJ, Devarajan P, Barasch J: Dual action of neutrophil gelatinase-associated lipocalin. JAmSocNephrol18: 407 413, 2007 30. Shapiro NI, Trzeciak S, Hollander JE, Birkhahn R, Otero R, Osborn TM, Moretti E, Nguyen HB, Gunnerson KJ, Milzman D, Gaieski DF, Goyal M, Cairns CB, Ngo L, Rivers EP: A prospective, multicenter derivation of a biomarker panel to assess risk of organ dysfunction, shock, and death in emergency department patients with suspected sepsis. Crit Care Med 37: 96 104, 2009 31. Mori K, Lee HT, Rapoport D, Drexler IR, Foster K, Yang J, Schmidt-Ott KM, Chen X, Li JY, Weiss S, Mishra J, Cheema FH, Markowitz G, Suganami T, Sawai K, Mukoyama M, Kunis C, D Agati V, Devarajan P, Barasch J: Endocytic delivery of lipocalin-siderophore-iron complex rescues the kidney from ischemia-reperfusion injury. J Clin Invest 115: 610 621, 2005 32. Bolignano D, Donato V, Coppolino G, Campo S, Buemi A, Lacquaniti A, Buemi M: Neutrophil gelatinase-associated lipocalin (NGAL) as a marker of kidney damage. Am J Kidney Dis 52: 595 605, 2008 33. Pickering JW, Endre ZH: Back-calculating baseline creatinine with MDRD misclassifies acute kidney injury in the intensive care unit. Clin J Am Soc Nephrol 5: 1165 1173, 2010 34. Vaidya VS, Waikar SS, Ferguson MA, Collings FB, Sunderland K, Gioules C, Bradwin G, Matsouaka R, Betensky RA, Curhan GC, Bonventre JV: Urinary biomarkers for sensitive and specific detection of acute kidney injury in humans. Clin Transl Sci 1: 200 208, 2008 35. Liangos O, Perianayagam MC, Vaidya VS, Han WK, Wald R, Tighiouart H, MacKinnon RW, Li L, Balakrishnan VS, Pereira BJG, Bonventre JV, Jaber BL: Urinary N-acetyl-beta-(D)-glucosaminidase activity and kidney injury molecule-1 level are associated with adverse outcomes in acute renal failure. J Am Soc Nephrol 18: 904 912, 2007 36. Shibata M, Hirota M, Nozawa F, Okabe A, Kurimoto M, Ogawa M: Increased concentrations of plasma IL-18 in patients with hepatic dysfunction after hepatectomy. Cytokine 12: 1526 1530, 2000 37. Mehta RL, Kellum JA, Shah SV, Molitoris BA, Ronco C, Warnock DG, Levin A Acute Kidney Injury Network: Acute Kidney Injury Network: Report of an initiative to improve outcomes in acute kidney injury. Crit Care 11: R31, 2007 38. Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P Acute Dialysis Quality Initiative workgroup: Acute renal failure - definition, outcome measures, animal models, fluid therapy and information technology needs: The Second International Consensus Conference of the Acute Dialysis Quality Initiative (ADQI) Group. Crit Care 8: R204 R212, 2004 39. Hanley JA, McNeil BJ: The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 143: 29 36, 1982 40. DeLong ER, DeLong DM, Clarke-Pearson DL: Comparing the areas under two or more correlated receiver operating characteristic curves: A nonparametric approach. Biometrics 44: 837 845, 1988 41. Levy MM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, Cohen J, Opal SM, Vincent J-L, Ramsay G SCCM/ESICM/ACCP/ATS/SIS: 2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis Definitions Conference. Crit Care Med 31: 1250 1256, 2003 This article contains supplemental material online at http://jasn.asnjournals. org/lookup/suppl/doi:10.1681/asn.2011040325/-/dcsupplemental. J Am Soc Nephrol 23: 322 333, 2012 Biomarker Quantitation Method 333