Abstract
Rationale: Measured at intensive care unit admission (ICU), the predictive value of neutrophil gelatinase-associated lipocalin (NGAL) for severe acute kidney injury (AKI) is unclear.
Objectives: To assess the ability of plasma and urine NGAL to predict severe AKI in adult critically ill patients.
Methods: Prospective-cohort study consisting of 632 consecutive patients.
Measurements and Main Results: Samples were analyzed by Triage immunoassay for NGAL expression. The primary outcome measure was occurrence of AKI based on Risk-Injury-Failure (RIFLE) classification during the first week of ICU stay. A total of 171 (27%) patients developed AKI. Of these 67, 48, and 56 were classified as RIFLE R, I, and F, respectively. Plasma and urine NGAL values at ICU admission were significantly related to AKI severity. The areas under the receiver operating characteristic curves for plasma and urine NGAL were for RIFLE R (0.77 ± 0.05 and 0.80 ± 0.04, respectively), RIFLE I (0.80 ± 0.06 and 0.85 ± 0.04, respectively), and RIFLE F (0.86 ± 0.06 and 0.88 ± 0.04, respectively) and comparable with those of admission estimated glomerular filtration rate (eGFR) (0.84 ± 0.04, 0.87 ± 0.04, and 0.92 ± 0.04, respectively). Plasma and urine NGAL significantly contributed to the accuracy of the “most efficient clinical model” with the best four variables including eGFR, improving the area under the curve for RIFLE F prediction to 0.96 ± 0.02 and 0.95 ± 0.01. Serial NGAL measurements did not provide additional information for the prediction of RIFLE F.
Conclusions: NGAL measured at ICU admission predicts the development of severe AKI similarly to serum creatinine–derived eGFR. However, NGAL adds significant accuracy to this prediction in combination with eGFR alone or with other clinical parameters and has an interesting predictive value in patients with normal serum creatinine.
There is discrepancy in the literature regarding the usefulness of neutrophil gelatinase-associated lipocalin (NGAL) as a biomarker to predict acute kidney injury in patients in intensive care.
This study suggests that NGAL could be of value alone and in combination with estimated glomerular filtration rate and other clinical variables. Moreover, this study suggests that serial measurements are useful for predicting severe acute kidney injury.
Recent experimental (6–8) and clinical (9–11) studies have identified biomarkers that may serve as early indicators of AKI. Of these, neutrophil gelatinase-associated lipocalin (NGAL) seems to be the most promising. NGAL is a 25-kD protein that is covalently bound to gelatinase and is secreted from human neutrophils (12). It is generally expressed at low concentrations in various organs containing epithelial tissues, including the kidney. When acute tubular damage occurs it is rapidly expressed at high concentrations in both plasma and urine (7–9, 13).
The first clinical validation was performed in pediatric cardiac surgery patients (9). In this study NGAL measured 2 hours after surgery was an excellent predictor of AKI, whereas serum creatinine (SCr) did not start to rise until 24 to 72 hours after surgery. However, in settings in which the initiation of renal injury is unclear, such as in cases of sepsis, trauma, and acute and critical illness, the predictive value of plasma NGAL (pNGAL) and urine NGAL (uNGAL) is less certain (13–19). Furthermore, whether NGAL on its own or in combination with clinical parameters can be of additional value for the prediction of severe AKI has yet to be determined.
We therefore conducted a prospective study in a large cohort of adult patients in the intensive care unit (ICU) to assess the predictive value of pNGAL and uNGAL levels at the time of admission with regard to the development of severe AKI during the early days of ICU treatment and their extended contribution in early diagnosis beyond estimated glomerular filtration rate (eGFR). None of the results of this current study have been previously reported in abstract form.
A detailed method session is given in the online supplement.
The institutional review board of Erasmus University Medical Center, Rotterdam, The Netherlands, approved the study. All consecutive admitted patients between September 2007 and April 2008 were eligible for enrollment. Exclusion criteria included age under 18 years, refusal of consent, nephrectomy, chronic kidney disease (CKD), ESRD, and renal transplantation. Deferred consent was used, and written informed consent was obtained from all participants or their heath care proxy (20).
After admission, plasma and urine samples were collected (T = 0) and thereafter at 4, 8, 24, 36, 48, 60, and 72 hours. Missing admission (T = 0) samples were replaced by first collection values at either 4 or 8 hours after admission. pNGAL and uNGAL were measured on the Triage NGAL Test point-of-care fluorescence immunoassay in a laboratory, masked to patient clinical data (Biosite, Inc, San Diego, CA). The Triage NGAL test has been validated against an NGAL ELISA assay (see online supplement) (21).
SCr was measured at admission and thereafter daily at 6:00 a.m. The eGFR was calculated using the Modification of Diet in Renal Disease Study Equation (MDRD) (see online supplement) (22). Baseline SCr was defined as the steady state level 4 weeks before admission. If not available, the admission value was used as a surrogate baseline. Other variables included age, sex, body mass index (BMI), temperature, pH, bicarbonate, potassium, blood-urea-nitrogen (BUN) content, white blood cell (WBC) count, C-reactive protein (CRP), and lactate. For disease severity assessment, the Acute Physiology and Chronic Health Evaluation score (APACHE II) and the sequential organ failure assessment score (SOFA) were used. Furthermore, the cumulative urine output, initiation of renal replacement therapy (RRT), ICU days, ICU mortality, and hospital mortality were recorded. The primary outcome variable was AKI occurring within 7 days after ICU admission according to the Risk-Injury-Failure (RIFLE) classification (23). The RIFLE classification is based on the rise in SCr compared with a baseline value. Risk (RIFLE R) represents a 1.5–2 times increase, injury (RIFLE I) a 2–3 times increase, and failure (RIFLE F) a more than 3 times increase.
MATLAB version 7.5.0 and SPSS version 16.0 were used. The relationships between AKI and NGAL levels were assessed using the Mann-Whitney U test and the chi-square test. Continuous variables were described by medians and interquartile ranges. Receiver operating characteristic (ROC) curves with their area under the curve (AUC) with two times its standard error was calculated. Univariable and multivariable logistic regression analyses were used to assess the predictive value of NGAL in combination with clinical parameters. Statistical significance was assessed by estimating the standard error of its coefficient and conducting a Wald test of the null hypothesis. Stepwise forward likelihood ratio regression was used to determine the model's most efficient predictors. Goodness of fit was assessed using the Hosmer-Lemeshow test. The net reclassification improvement was calculated. All reported P values are two-tailed, and P values less than 0.05 were considered statistically significant.
Biosite Incorporated (San Diego, CA) provided biomarker measurements and statistical support. They had no role in study design, data collection, or writing of the manuscript. The first author had full access to all data and had final responsibility to submit for publication.
Of the 700 consecutive patients who were screened for inclusion in the study, 68 (9.8%) were excluded because of refusal of consent (n = 6), nephrectomy (n = 6), CKD, ESRD, kidney transplantation (n = 25), or missing admission data (n = 31). Thus, 632 (90.2%) patients were included in the analysis. Patient characteristics are shown in Table 1.
Variable | Non-AKI (n = 461) | RIFLE R (n = 67) | RIFLE I (n = 48) | RIFLE F (n = 56) | P Value |
|---|---|---|---|---|---|
| Age, yr | 58 (43,68) | 59 (45,70) | 61.5 (53, 75) | 62 (50, 68) | NS |
| Male, n (%) | 264 (57) | 46 (69) | 29 (60) | 30 (54) | NS |
| BMI, kg/m2 | 24.5 (22.5, 27.2) | 25.5 (22.5, 27.4) | 25.5 (22.9, 28.6) | 25.3 (22.1, 28.1) | NS |
| Temperature | 36.9 (36.2, 37.6) | 37 (36.2, 37.6) | 36.6 (35.8, 37.7) | 36.9 (36.3, 37.8) | NS |
| SCr, mg/dl | 0.75 (0.61, 0.91) | 1.10 (0.82, 1.39) | 1.30 (0.82, 1.64) | 2.09 (1.31, 2.86) | <0.0001 |
| eGFR, ml/min/1.73 m2 | 104 (84, 129) | 70 (50, 97) | 54 (41, 92) | 32 (21, 50) | <0.0001 |
| Plasma NGAL, ng/ml | 153 (85, 233) | 268 (145, 397) | 353 (169, 531) | 680 (332, 1195) | <0.0001 |
| Urine NGAL, ng/ml | 75 (37, 206) | 323 (74, 963) | 523 (199, 2640) | 2,013 (564, 4124) | <0.0001 |
| pH | 7.39 (7.34, 7.44) | 7.35 (7.29, 7.42) | 7.33 (7.27, 7.41) | 7.31 (7.26, 7.40) | <0.0001 |
| HCO3−, mmol/L | 22 (20.1, 24.3) | 21 (18.2, 23.7) | 19.9 (16, 23.7) | 18 (13.4, 20.9) | <0.0001 |
| K, mmol/L | 3.9 (3.6, 4.3) | 4.1 (3.7, 4.6) | 4.3 (3.6, 4.5) | 4.3 (3.9, 4.9) | <0.0001 |
| BUN, mmol/L | 5.5 (4.2, 7.3) | 8.6 (5.1, 12.1) | 8.8 (5.8, 17.1) | 14.1 (8.4, 26.6) | <0.0001 |
| White blood cell count, 109/ml | 11.4 (8.4, 15.1) | 10 (6.9, 14.8) | 11.8 (6.9, 16.2) | 11 (6.3, 17.6) | NS |
| CRP, mmol/L | 12 (3, 68) | 72 (8, 158) | 25 (6, 134) | 118 (36, 198) | <0.0001 |
| Lactate, mmol/L | 1.5 (1, 2.4) | 2.2 (1.4, 3.2) | 2.3 (1.3, 4.6) | 2.3 (1.2, 4.2) | <0.0001 |
| Apache II score | 16 (13, 22) | 19 (15, 28) | 24 (20, 29) | 25 (22, 28) | <0.0001 |
| SOFA score | 4 (2, 6) | 7 (4, 9) | 8 (6, 11) | 11 (8, 13) | <0.0001 |
| UP, ml/kg/h | 1.1 (0.8, 1.7) | 1 (0.7, 1.4) | 0.8 (0.6, 1.3) | 0.5 (0.2, 0.9) | <0.0001 |
| RRT, n (%) | 0 (0) | 0 (0) | 0 (0) | 28 (50) | <0.0001 |
| ICU mortality, n (%) | 49 (8) | 10 (15) | 9 (19) | 26 (46) | <0.0001 |
| Hospital mortality, n (%) | 71 (11) | 20 (30) | 16 (33) | 30 (54) | <0.0001 |
| Diagnostic group, n (%) | |||||
| Postoperative | 166 (36) | 15 (22) | 6 (13) | 5 (9) | <0.0001 |
| Medical | 99 (22) | 13 (19) | 15 (31) | 11 (20) | NS |
| Neurologic | 88 (19) | 5 (8) | 1 (2) | 1 (2) | <0.0001 |
| Neurotrauma | 27 (6) | 2 (3) | 0 (0) | 1 (2) | NS |
| Multitrauma | 26 (6) | 6 (9) | 4 (8) | 1 (2) | NS |
| LTX | 19 (4) | 8 (12) | 1 (2) | 1 (2) | NS |
| Sepsis | 14 (3) | 6 (9) | 8 (17) | 15 (27) | <0.0001 |
| CPR | 11 (2) | 6 (9) | 7 (15) | 3 (5) | <0.0001 |
| Hemorrhagic shock | 9 (2) | 4 (6) | 3 (6) | 3 (5) | NS |
| MOF | 1 (0) | 2 (3) | 3 (6) | 15 (27) | <0.0001 |
AKI occurred in 171 patients (27%). Of those patients, 67 developed RIFLE R, 48 patients developed RIFLE I, and 56 patients developed RIFLE F. The time to reach a SCr increase of more than 50% compared with baseline for the first time (= RIFLE R) was T = 0 in 58.5%, T = 24 in 24%, T = 48 in 6.4%, and T = 72 in 5.8% of the patients. Thus, 94.7% of the patients reached “first AKI” within 72 hours after ICU admission. Twenty-eight (50%) of the patients with AKI in the RIFLE F class received RRT (4.4% of the overall patient cohort).
Baseline characteristics in all RIFLE classes were compared with subjects who did not develop AKI. There were no differences with respect to age, sex, or BMI. Patients with AKI had higher APACHE II and SOFA scores than patients without AKI (Table 1). Furthermore, there were positive correlations between the severity of kidney injury and length of stay, ICU mortality, and hospital mortality (Table 1). The incidence of AKI was higher in patients admitted after cardiopulmonary resuscitation was performed, and in patients with sepsis or multiorgan failure syndrome (P < 0.0001) (Table 1).
Patients' pNGAL and uNGAL concentrations at the time of ICU admission were significantly related to their RIFLE scores (P < 0.0001) (Table 1, Figure 1). The pNGAL test performance for predicting the severity of AKI in the entire cohort showed an AUC of 0.77 ± 0.05 for RIFLE R and above, 0.80 ± 0.06 for RIFLE I and above, and 0.86 ± 0.06 for RIFLE F. Similar analysis for uNGAL revealed AUCs of 0.80 ± 0.04 (RIFLE R), 0.85 ± 0.04 (RIFLE I), and 0.88 ± 0.04 (RIFLE F) (Figures 2A and 2B). The differences between the plasma and urine AUCs were not significant. The AUC and ROC curves for eGFR predicting AKI stratified for RIFLE stage are shown in Figure 2C. Comparing the performance of eGFR with pNGAL and uNGAL showed that only pNGAL predicting R and above or I and above were significantly different compared with the corresponding AUCs of eGFR (P = 0.015 and P = 0.039).
Figure 1. Admission plasma (A) and urine (B) neutrophil gelatinase-associated lipocalin (NGAL) concentrations stratified by Risk-Injury-Failure (RIFLE) classification. An exploratory Mann-Whitney U test of adjacent categories, including nonacute kidney injury versus R, R versus I, and I versus F, resulted in P values of < 0.0001, 0.10, and 0.0005, respectively, for plasma NGAL and < 0.0001, 0.028, and 0.001, respectively, for urine NGAL. AKI = acute lung injury.
[More] [Minimize]
Figure 2. Receiver operating characteristic curve analysis for the ability of admission plasma (A) and urine (B) neutrophil gelatinase-associated lipocalin, estimated glomerular filtration rate (C) to predict acute kidney injury, stratified by Risk-Injury-Failure (RIFLE) classification. (D) Both plasma and urine neutrophil gelatinase-associated lipocalin's predictive properties for RIFLE I or worse in patients with estimated glomerular filtration rate above 60 ml/min/1.73 m2. AUC values ± 2 SE are presented parenthetically after the RIFLE classification. eGFR = estimated glomerular filtration rate; MDRD = Modification of Diet in Renal Disease Study Equation; NGAL = neutrophil gelatinase-associated lipocalin; SENS = sensitivity; SPEC = specificity.
[More] [Minimize]Table 2 lists the calculated sensitivities at fixed specificities of 50%, 70%, and 90% (derived by visual inspection of the ROC curves) with the corresponding cut-off concentrations of pNGAL and uNGAL for the prediction of RIFLE F.
Sensitivity | Specificity | Positive Predictive Value | Negative Predictive Value | |
|---|---|---|---|---|
| Cutoffs for pNGAL | ||||
| 168 ng/ml | 0.91 | 0.50 | 0.15 | 0.98 |
| 245 ng/ml | 0.82 | 0.70 | 0.21 | 0.98 |
| 417 ng/ml | 0.70 | 0.90 | 0.40 | 0.97 |
| Cutoffs for uNGAL | ||||
| 94 ng/ml | 0.98 | 0.50 | 0.16 | 1.00 |
| 247 ng/ml | 0.89 | 0.70 | 0.22 | 0.98 |
| 1,310 ng/ml | 0.55 | 0.90 | 0.35 | 0.95 |
To determine the potential additional contribution of NGAL as a biomarker predicting AKI before SCr has started to rise and consequently eGFR has started to decline a subset analyses was performed in patients with apparently normal renal function (n = 498) at the time of ICU admission (i.e., excluding patients with an eGFR <60 ml/min/1.73 m2). ROC analysis demonstrated that in patients who did not show any increase in SCr yet at ICU admission, pNGAL and uNGAL had diagnostic superiority over SCr and eGFR for predicting severe AKI (RIFLE I and F). The AUCs for pNGAL and uNGAL were respectively 0.75 ± 0.10 and 0.79 ± 0.10 compared with 0.65 ± 0.10 and 0.67 ± 0.10 for SCr and eGFR, respectively (Figure 2D).
Adding pNGAL and uNGAL to eGFR in a multivariable logistic regression model improved the prediction significantly (P < 0.001). To determine the added contribution of NGAL to eGFR and other available clinical variables at ICU admission for predicting the occurrence of RIFLE F within the first week of patients' ICU stay, additional logistic regression analysis was performed (Table 3). The available clinical predictors included age, BMI, temperature, diagnosis of sepsis, pH, bicarbonate, potassium, BUN, WBC count, CRP, and lactate. Adding NGAL to eGFR and clinical variables improved the prediction significantly for pNGAL (P = 0.014) and almost significantly for uNGAL (P = 0.092).
Plasma NGAL | Urine NGAL | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RC | RC | |||||||||||||
| Variable | OR | (B) | (SE) | P Value | OR | (B) | (SE) | P Value | ||||||
| NGAL, ng/ml | 1.83 | 0.6 | (0.25) | 0.017 | 1.49 | 0.40 | (0.23) | 0.088 | ||||||
| eGFR, ml/min/1.73 m2 | 0.95 | −0.05 | (0.01) | 0.000 | 0.94 | −0.06 | (0.01) | 0.000 | ||||||
| Age, yr | 0.98 | −0.02 | (0.02) | 0.240 | 0.97 | −0.03 | (0.02) | 0.131 | ||||||
| BMI, kg/m2 | 0.93 | −0.07 | (0.06) | 0.240 | 0.91 | −0.09 | (0.06) | 0.144 | ||||||
| Temp, °C | 0.68 | −0.39 | (0.16) | 0.016 | 0.68 | −0.39 | (0.16) | 0.015 | ||||||
| Sepsis | 10.52 | 2.35 | (0.70) | 0.001 | 14.16 | 2.65 | (0.66) | 0.000 | ||||||
| PH | 2.07 | 0.73 | (3.35) | 0.828 | 3.79 | 1.33 | (3.57) | 0.709 | ||||||
| HCO3−, mmol/L | 1.01 | 0.01 | (0.06) | 0.822 | 1.01 | 0.01 | (0.06) | 0.819 | ||||||
| K, mmol/L | 2.10 | 0.74 | (0.34) | 0.028 | 1.93 | 0.66 | (0.35) | 0.057 | ||||||
| BUN, mmol/L | 0.99 | −0.01 | (0.04) | 0.816 | 1.00 | 0.00 | (0.04) | 0.972 | ||||||
| WBC,109/ml | 0.93 | −0.07 | (0.03) | 0.025 | 0.94 | −0.06 | (0.03) | 0.025 | ||||||
| CRP, mmol/L | 1.00 | 0.00 | (0.00) | 0.259 | 1.00 | 0.00 | (0.00) | 0.326 | ||||||
| Lactate, mmol/L | 0.87 | −0.14 | (0.11) | 0.224 | 0.88 | −0.13 | (0.11) | 0.237 | ||||||
| Total | 0.014 | 0.092 | ||||||||||||
Using a stepwise forward likelihood ratio logistic regression NGAL, eGFR, diagnosis of sepsis, WBC count, and temperature on admission made the most efficient clinical model for the prediction of RIFLE F for plasma out of the available variables in this study. For urine the most efficient model comprised NGAL, eGFR, diagnosis of sepsis, and WBC count (Table 4). Adding NGAL changed the model's AUCs from 0.95 ± 0.02 to 0.96 ± 0.02 for pNGAL and from 0.94 ± 0.02 to 0.95 ± 0.01 for uNGAL. Furthermore, we assessed the ability of pNGAL and uNGAL to “reclassify” the degree of risk for RIFLE F within 7 days as assessed by the model. Subjects were categorized into prespecified “low-risk,” “medium-risk,” and “high-risk” groups using cut-offs of less than 30%, 30–60%, and greater than 60%, respectively. We compared the proportions of reclassified subjects across these three risk groups when NGAL was added to the clinical model for plasma and urine (see online supplement for detailed reclassification table). For five patients with RIFLE F reclassification was more accurate when the model with all four variables for pNGAL was used and for two patients it became less accurate. Among the subjects without RIFLE F, nine were correctly reclassified in a lower risk category, whereas three were incorrectly reclassified to be at higher risk. The same analysis was performed for uNGAL (see online supplement). The generated net reclassification improvement for pNGAL and uNGAL added to the clinical prediction model was 8.5% (P = 0.087) and 2.3% (P = 0.370), respectively.
Plasma NGAL | Urine NGAL | |||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| RC | RC | |||||||||||||
| Variable | OR | (B) | (SE) | P Value | OR | (B) | (SE) | P Value | ||||||
| NGAL, ng/ml | 1.71 | 0.54 | (0.21) | 0.010 | 1.42 | 0.36 | (0.17) | 0.039 | ||||||
| eGFR, ml/min/1.73 m2 | 0.95 | −0.05 | (0.01) | 0.000 | 0.95 | −0.06 | (0.01) | 0.000 | ||||||
| Sepsis | 9.94 | 2.30 | (0.59) | 0.000 | 9.15 | 2.21 | (0.53) | 0.000 | ||||||
| WBC, 109/ml | 0.95 | −0.06 | (0.03) | 0.057 | 0.95 | −0.05 | (0.02) | 0.051 | ||||||
| Temp, °C | 0.78 | −0.25 | (0.13) | 0.061 | ||||||||||
| Total | 0.000 | 0.000 | ||||||||||||
The progression of mean pNGAL and uNGAL concentrations stratified by RIFLE classification over time is shown in the online supplement (see Figures E1 and E2). To determine if serial sampling could be of additional value for the prediction of RIFLE F pNGAL and uNGAL values and those of the other predictors at T = 0 and T = 24 (temperature, pH, bicarbonate, potassium, BUN, WBC, CRP, and lactate) were used for multivariable logistic regression analysis. In addition, age, BMI, diagnosis of sepsis, eGFR MDRD T = 0, the 24-hour urine production, the 24-hour cumulative fluid balance, APACHE II, and SOFA score were added. All subjects with established RIFLE F or missing data in the first 24 hours were excluded, leaving 429 patients for the plasma and 411 for the urine analysis. With stepwise forward likelihood ratio logistic regression the most efficient predictors were pNGAL T = 24 (P = 0.000) and CRP T = 0 (P = 0.024) for the pNGAL model. Adding pNGAL T = 24 changed the model's AUC from 0.63 ± 0.04 to 0.91 ± 0.03, underlining that pNGAL T = 24 is a very strong predictor for RIFLE F. For urine, the model showed NGAL T = 24 (P = 0.001), temperature T = 0 (P = 0.02), APACHE T = 24 (P = 0.011), urine production T = 24 (P = 0.009), pH T = 24 (P = 0.005), and potassium T = 0 (P= 0.055) as most efficient predictors. Adding uNGAL T = 24 changed the model's AUC from 0.84 ± 0.06 to 0.93 ± 0.04.
Assessment of both pNGAL and uNGAL values' difference scores in the logistic regression analysis showed that the temporal changes were not relevant, pointing out that the NGAL value measured closer to the end point “RIFLE F” was the strongest predictor.
Analyzing further contribution of serial measurements over the succeeding time points was not possible because of the significant reduction in sample size with the diminished availability of equal measurements. Furthermore, because the difference in serial measurements in the first 24 hours did not add to the prediction of RIFLE F, it is not expected that the results will be different when analyzing subsequent time points.
In patients with sepsis (n = 14) who did not develop AKI, uNGAL levels were significantly higher than those of patients in the other diagnostic groups. The median NGAL value was 1,264.1 ng/ml (650.3, 4,124) (Figure 3). In the group of 14 patients with septic non-AKI, one received renal drainage because of obstructive hydronephrosis, one had a positive urine culture with Acinetobacter species, one had a positive WBC and nitrite count in the urine sediment without a positive culture already under antibiotic treatment, and one patient had a proved renal abscess with Escherichia coli. After we adjusted the uNGAL analysis removing those patients and patients who died within 48 hours after admission to the ICU, uNGAL levels were still significantly higher among patients with a diagnosis of sepsis than among patients in the other diagnostic groups (P = 0.0005).
Figure 3. Admission plasma (A) and urine (B) neutrophil gelatinase-associated lipocalin values for patients without acute kidney injury stratified by diagnostic groups. CPR = cardiopulmonary resuscitation; NGAL = neutrophil gelatinase-associated lipocalin.
[More] [Minimize]In the entire cohort both NGAL plasma and urine values were predictive of RRT initiation within the first week of ICU admission (respectively, AUC 0.88 ± 0.06 and AUC 0.89 ± 0.04). However, SCr and eGFR reached similar performances (respectively, AUC 0.90 ± 0.05 and 0.91 ± 0.05). Both pNGAL and uNGAL have a minor role in predicting hospital mortality with very modest performances (AUC 0.63 ± 0.06 and AUC 0.64 ± 0.06).
The present study shows that pNGAL and uNGAL levels at time of ICU admission predict the development of severe AKI and the initiation of RRT in critically ill patients within the first 7 days of their ICU stay. Furthermore, adding NGAL values to a model with eGFR alone or to the most efficient clinical model with available parameters improves the prediction significantly. Using serial NGAL measurements did not provide additional accuracy in the prediction of RIFLE F. Finally, patients with sepsis but no AKI have significantly higher urinary NGAL values compared with other patients without AKI.
NGAL fulfills a central role in regulating epithelial neogenesis, and in iron chelation and delivery after ischemic or toxic insults to the renal tubular epithelium (24, 25). After kidney injury, NGAL is rapidly expressed on the apical epithelial membranes of the distal nephron. NGAL is excreted in the urine through exocytosis and has local bacteriostatic and proapoptotic effects (26, 27). PNGAL is easily filtered by the glomerulus and reabsorbed in the apical membranes of the proximal tubules. Reabsorbtion is mediated by megalin-cubulin dependent endocytosis with a very high affinity. The delivered iron is needed in processes activating and repressing iron-responsive genes that are vital to the regeneration processes that occur after damage is inflicted to these cells. Under normal circumstances the estimated half life of pNGAL is approximately 10 minutes, with urinary loss less than 0.2% (28, 29). PNGAL and uNGAL concentrations increase by 10- to 100-fold during the 2 hours that follow tubular injury (7–9), whereas SCr does not start to rise until 24 to 72 hours after the initial renal insult (9, 16, 30).
Because we are interested in the possible prevention of (further) kidney injury in patients who are critically ill, AKI was assessed only during the first week of each patient's ICU stay to link the condition of the patient at the time of admission and the initial resuscitation efforts to the development of AKI.
In this study, we found that pNGAL and uNGAL measured at the time of admission were good predictors of AKI. The test performance of both pNGAL and uNGAL increased as the severity of the functional damage to the kidney's increased; the AUCs ranged from 0.77 (RIFLE R) to 0.86 (RIFLE F) for pNGAL and from 0.80 (RIFLE R) to 0.88 (RIFLE F) for uNGAL.
Previous studies in pediatric patients in the ICU with sepsis and septic shock (14) and in a group of adult critically ill patients (17) have studied the predictive accuracy of pNGAL and uNGAL reporting AUCs of 0.68 and 0.64 for sustained AKI. Both Zappitelli and coworkers (16) (pediatric population) and Cruz and coworkers (19) (adult population) observed AUC's for prediction of RIFLE R or worse AKI by NGAL that were comparable with those observed in the present study. Constantin and coworkers (18) and Nickolas and coworkers (13) reported very high AUCs for the ability of pNGAL and uNGAL to predict AKI in critically ill adult and emergency department patients (0.92 and 0.95, respectively). Several explanations exist for the observed variability of NGAL's test performance in these studies, in which the timing of renal insult was not strictly identified.
First, in the current study NGAL measurement was performed immediately after ICU admission and patients were monitored for the occurrence of AKI for the next 7 days. The timing of NGAL measurement in the previously mentioned studies ranged from 48 hours after the initiation of mechanical ventilation (up to 3 d after admission) to within 24 hours of ICU admission to the first possible moment on ICU admission. With the rapid changes in pNGAL and uNGAL concentrations, the slow changes in SCr concentrations, the reversibility of the early phases in the continuum of AKI, and the effects of intensive resuscitation in the golden hours after ICU admittance, timing of measurement has effects on the NGAL concentrations measured in relation to the changes in SCr (31). Therefore the time at which NGAL levels are measured clearly influences their test performance.
Second, the number of patients with AKI in a given study and their RIFLE class distribution also influences test results (32). Because of the large sample size in this study and the fairly equal patient distribution between RIFLE categories, we were able to analyze the ability of NGAL to predict more severe AKI end points, such as RIFLE F. In contrast, Wheeler and coworkers (14) used very unusual criteria for AKI, making it impossible to compare their results with those of other studies. The AKI cohort in the study performed by Siew and coworkers (17) was comprised of patients with less severe stages of AKI (median uNGAL 127 ng/ml interquartile range [IQR]: 32–623 and median SCr 1.5 mg/dl IQR: 1–2.2 at enrollment) resulting in low performance characteristics of NGAL (AUC = 0.71; 95% confidence interval, 0.63–0.78). Nickolas and coworkers (13) reported that NGAL was an excellent predictor of AKI (AUC = 0.95; 95% confidence interval, 0.88–1) in an emergency department setting. However, the mean SCr and fractional sodium excretion of this entire AKI subgroup at the time of study inclusion were 5.6 mg/dl (SD = 5.5) and 6.9% (SD 9.1), respectively, indicating that severe loss of renal function had already occurred in most of these patients. Accordingly, test results generated in patients with established AKI should not be used for the comparison with those in a cohort of developing AKI.
Third, AKI and its severity defined by RIFLE are dependent on how baseline SCr values are determined and will contribute to different outcomes between studies. In our study the first available SCr value was used as a surrogate baseline when a patient's historical data were not available. This undoubtedly has resulted in an underestimation of attained RIFLE stage in some of these patients. Furthermore, with the artificial definition of AKI using three set severity stages the issue of timing may simply be definitional.
This study adds to the current literature because it showed that NGAL significantly improves the diagnostic accuracy for severe AKI adding it to MDRD eGFR calculated at ICU admission, even in patients having an apparently normal eGFR at admission. Especially in these patients this could be of value because their AKI is not yet reflected in an increase in SCr. Patients in the ICU are typically diagnosed with AKI several days after the onset of their illness or injury, resulting in a delay in the discontinuation or dose adjustment of nephrotoxic medications or continued use of procedures that could cause further renal damage. Whether NGAL levels have the potential to influence clinical decision making in the ICU should be the topic for further randomized studies that should be performed before using NGAL measurements in clinical practice.
These studies may include applying more intensive resuscitation, avoiding nephrotoxic drugs, or implementation of a more timely initiation of RRT in patients with elevated NGAL levels (33). In addition, recent animal studies examining interventions to reverse AKI have been promising, implying that it may be possible to reverse AKI in humans if it is treated early (29, 34–39). Second, this study adds to current knowledge because we defined a most efficient clinical model in the prediction of AKI using available data at the time of ICU admission, improving the predictive accuracy for RIFLE F significantly with NGAL above eGFR and clinical predictors. The predictive accuracy of eGFR on its own was roughly comparable with that of pNGAL or uNGAL. However, we should take into account that SCr is used to define the end point RIFLE F and is likewise used to calculate eGFR, which is incorporation bias. Therefore, it is somewhat biased to compare NGAL's performance with the ability of SCr to predict itself. Furthermore, in this study, the point of first AKI was satisfied in many patients at the time of ICU admission (58.5%). As such, it is to be expected that in many of the AKI cases, SCr would already be elevated at the time of admission. AKI that was present at the time of ICU admission was determined by retrospective collection of baseline SCr values from the patient records before admission. However, in clinical practice, a prior baseline SCr is more often not available at ICU admission and as such it is not possible to correctly determine the end point of AKI compared with CKD. Furthermore, we should also take into account that NGAL is a direct injury marker that is unfortunately compared with a gold standard AKI diagnosis that is based on a functional marker (SCr), which has major imperfections on its own (40). In this context it is indispensable to emphasize the importance of (injury) biomarker combinations to achieve more accurate predictions irrespective of SCr.
Third, we showed that temporal changes in NGAL measurements do not provide additional information for the prediction of RIFLE F. And finally, we found that patients with sepsis without AKI had markedly increased uNGAL concentrations, whereas there were no significant differences between groups with regard to the pNGAL values. A possible explanation for our results lies in the two-compartment model theory of NGAL (which applies to an animal model under relatively normal conditions) (27) and the fact that AKI is an inflammatory disease (41). In patients with AKI, human Toll-like receptor 2 (TLR2) stimulates tubular epithelial apoptosis (42) and NGAL expression (43). Bacterial pathogens produce lipoproteins and activate cytokine networks by inducing the expression of multiple proinflammatory genes. Lipoproteins also have strong affinity for TLRs that trigger an innate immune response. Therefore, it could be postulated that these circulating ligands that are linked to tubular epithelial TLR activation are responsible for the increased uNGAL concentrations that we observed in patients who had sepsis but showed no increases in their SCr levels (44). However, a very recent study in patients with sepsis, septic shock, and systemic inflammatory response syndrome has reported contradictory findings (45). A possible explanation for this difference is the variability of the subject inclusion time (up to 48 h after ICU admission). Intensive resuscitation and the administration of antibiotics may have already occurred before study inclusion, therefore most likely inducing rapid changes of uNGAL values.
In conclusion, the present study shows that both pNGAL and uNGAL levels at ICU admission are good predictors of severe AKI and significantly add to the prediction of AKI using eGFR and to a model with clinical parameters. Because the study population reflects a mixed group of diagnoses that are present in most ICUs these findings could have major clinical implications regarding optimization of therapy in patients at risk for AKI. Our findings could also facilitate studies of the effectiveness of early therapeutic and supportive interventions in patients with established AKI.
The authors thank Ken Kupfer, Brian Noland, Kristina Little, and Gillian Parker from Inverness Medical Innovations, Inc., for their support of this project. They express appreciation to the nurse coordinator, Wil Mol, for her contribution and to the patients and their families for their participation.
| 1. | Liangos O, Wald R, O'Bell JW, Price L, Pereira BJ, Jaber BL. Epidemiology and outcomes of acute renal failure in hospitalized patients: a national survey. Clin J Am Soc Nephrol 2006;1:43–51. |
| 2. | Metnitz PG, Krenn CG, Steltzer H, Lang T, Ploder J, Lenz K, Le Gall JR, Druml W. Effect of acute renal failure requiring renal replacement therapy on outcome in critically ill patients. Crit Care Med 2002;30:2051–2058. |
| 3. | Uchino S, Kellum JA, Bellomo R, Doig GS, Morimatsu H, Morgera S, Schetz M, Tan I, Bouman C, Macedo E, et al.; Beginning and Ending Supportive Therapy for the Kidney (BEST Kidney) Investigators. Acute renal failure in critically ill patients: a multinational, multicenter study. JAMA 2005;294:813–818. |
| 4. | Tan SS, Hakkaart-van Roijen L, Al MJ, Bouwmans CA, Hoogendoorn ME, Spronk PE, Bakker J. A microcosting study of intensive care unit stay in the Netherlands. J Intensive Care Med 2008;23:250–257. |
| 5. | Ishani A, Xue JL, Himmelfarb J, Eggers PW, Kimmel PL, Molitoris BA, Collins AJ. Acute kidney injury increases risk of ESRD among elderly. J Am Soc Nephrol 2009;20:223–228. |
| 6. | Brian Reeves W, Kwon O, Ramesh G. Netrin-1 and kidney injury. II. Netrin-1 is an early biomarker of acute kidney injury. Am J Physiol Renal Physiol 2008;294:F731–F738. |
| 7. | Mishra J, Ma Q, Prada A, Mitsnefes M, Zahedi K, Yang J, Barasch J, Devarajan P. Identification of neutrophil gelatinase-associated lipocalin as a novel early urinary biomarker for ischemic renal injury. J Am Soc Nephrol 2003;14:2534–2543. |
| 8. | Mishra J, Mori K, Ma Q, Kelly C, Barasch J, Devarajan P. Neutrophil gelatinase-associated lipocalin: a novel early urinary biomarker for cisplatin nephrotoxicity. Am J Nephrol 2004;24:307–315. |
| 9. | Mishra J, Dent C, Tarabishi R, Mitsnefes MM, Ma Q, Kelly C, Ruff SM, Zahedi K, Shao M, Bean J, et al. Neutrophil gelatinase-associated lipocalin (NGAL) as a biomarker for acute renal injury after cardiac surgery. Lancet 2005;365:1231–1238. |
| 10. | Parikh CR, Abrahan E, Ancukiewicz M, Edelstein CL. Urine IL-18 is an early diagnostic marker for acute kidney injury and predicts mortality in the intensive care unit. J Am Soc Nephrol 2005;16:3046–3052. |
| 11. | Portilla D, Dent C, Sugaya T, Nagothu KK, Kundi I, Moore P, Noiri E, Devarajan P. Liver fatty acid-binding protein as a biomarker of acute kidney injury after cardiac surgery. Kidney Int 2008;73:465–472. |
| 12. | Kjeldsen L, Johnsen AH, Sengeløv H, Borregaard N. Isolation and primary structure of NGAL, a novel protein associated with human neutrophil gelatinase. J Biol Chem 1993;268:10425–10432. |
| 13. | Nickolas TL, O'Rourke MJ, Yang J, Sise ME, Canetta PA, Barasch N, Buchen C, Khan F, Mori K, Giglio J, et al. Sensitivity and specificity of a single emergency department measurement of urinary neutrophil gelatinase-associated lipocalin for diagnosing acute kidney injury. Ann Intern Med 2008;148:810–819. |
| 14. | Wheeler DS, Devarajan P, Ma Q, Harmon K, Monaco M, Cvijanovich N, Wong HR. Serum neutrophil gelatinase-associated lipocalin (NGAL) as a marker of acute kidney injury in critically ill children with septic shock. Crit Care Med 2008;36:1297–1303. |
| 15. | Makris K, Markou N, Evodia E, Dimopoulou E, Drakopoulos I, Ntetsika K, Rizos D, Baltopoulos G, Haliassos A. Urinary neutrophil gelatinase-associated lipocalin (NGAL) as an early marker of acute kidney injury in critically ill multiple trauma patients. Clin Chem Lab Med 2009;47:79–82. |
| 16. | 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 2007;11:R84. |
| 17. | Siew ED, Ware LB, Gebretsadik T, Shintani A, Moons KG, Wickersham N, Bossert F, Ikizler TA. Urine neutrophil gelatinase-associated lipocalin moderately predicts acute kidney injury in critically ill adults. J Am Soc Nephrol 2009;20:1823–1832. |
| 18. | Constantin JM, Futier E, Perbet S, Roszyk L, Lautrette A, Gillart T, Guerin R, Jabaudon M, Souweine B, Bazin JE, et al. Plasma neutrophil gelatinase-associated lipocalin is an early marker of acute kidney injury in adult critically ill patients: a prospective study. J Crit Care 2010;25:176.e1–6. |
| 19. | Cruz DN, de Cal M, Garzotto F, Perazella MA, Lentini P, Corradi V, Piccinni P, Ronco C. Plasma neutrophil gelatinase-associated lipocalin is an early biomarker for acute kidney injury in an adult ICU population. Intensive Care Med 2010;36:444–451. |
| 20. | Jansen TC, Kompanje EJ, Bakker J. Deferred proxy consent in emergency critical care research: ethically valid and practically feasible. Crit Care Med 2009;37(Suppl. 1):S65–S68. |
| 21. | Dent CL, Ma Q, Dastrala S, Bennett M, Mitsnefes MM, Barasch J, Devarajan P. Plasma neutrophil gelatinase-associated lipocalin predicts acute kidney injury, morbidity and mortality after pediatric cardiac surgery: a prospective uncontrolled cohort study. Crit Care 2007;11:R127. |
| 22. | Levey AS, Bosch JP, Lewis JB, Greene T, Rogers N, Roth D. A more accurate method to estimate glomerular filtration rate from serum creatinine: a new prediction equation. Modification of Diet in Renal Disease Study Group. Ann Intern Med 1999;130:461–470. |
| 23. | Bellomo R, Ronco C, Kellum JA, Mehta RL, Palevsky P. 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 2004;8:R204–R212. |
| 24. | Yang J, Goetz D, Li JY, Wang W, Mori K, Setlik D, Du T, Erdjument-Bromage H, Tempst P, Strong R, et al. An iron delivery pathway mediated by a lipocalin. Mol Cell 2002;10:1045–1056. |
| 25. | Gwira JA, Wei F, Ishiibe S, Ueland JM, Barasch J, Cantley LG. Expression of neutrophil gelatinase-associated lipocalin regulates epithelial morphogenesis in vitro. J Biol Chem 2005;280:7875–7882. |
| 26. | Schmidt-Ott KM, Mori K, Kalandadze A, Li JY, Paragas N, Nicholas T, Devarajan P, Barasch J. Neutrophil gelatinase-associated lipocalin-mediated iron traffic in kidney epithelia. Curr Opin Nephrol Hypertens 2006;15:442–449. |
| 27. | Schmidt-Ott KM, Mori K, Li JY, Kalandadze A, Cohen DJ, Devarajan P, Barasch J. Dual action of neutrophil gelatinase-associated lipocalin. J Am Soc Nephrol 2007;18:407–413. |
| 28. | Axelsson L, Bergenfeldt M, Ohlsson K. Studies of the release and turnover of a human neutrophil lipocalin. Scand J Clin Lab Invest 1995;55:577–588. |
| 29. | Mori K, Lee HT, Rapoport D, Drexler IR, Foster K, Yang J, Schmidt-Ott KM, Chen X, Li JY, Weiss S, et al. Endocytic delivery of lipocalin-siderophore-iron complex rescues the kidney from ischemia-reperfusion injury. J Clin Invest 2005;115:610–621. |
| 30. | Haase-Fielitz A, Bellomo R, Devarajan P, Story D, Matalanis G, Dragun D, Haase M. Novel and conventional serum biomarkers predicting acute kidney injury in adult cardiac surgery: a prospective cohort study. Crit Care Med 2009;37:553–560. |
| 31. | Molitoris BA. Transitioning to therapy in ischemic acute renal failure. J Am Soc Nephrol 2003;14:265–267. |
| 32. | Haase-Fielitz A, Bellomo R, Devarajan P, Bennett M, Story D, Matalanis G, Frei U, Dragun D, Haase M. The predictive performance of plasma neutrophil gelatinase-associated lipocalin (NGAL) increases with grade of acute kidney injury. Nephrol Dial Transplant 2009;24:3349–3354. |
| 33. | Ronco C. N-GAL: diagnosing AKI as soon as possible. Crit Care 2007;11:173. |
| 34. | Hoglen NC, Chen LS, Fisher CD, Hirakawa BP, Groessl T, Contreras PC. Characterization of IDN-6556 (3-[2-(2-tert-butyl-phenylaminooxalyl)-amino]-propionylamino]-4-oxo-5-(2,3, 5,6-tetrafluoro-phenoxy)-pentanoic acid): a liver-targeted caspase inhibitor. J Pharmacol Exp Ther 2004;309:634–640. |
| 35. | Doi K, Suzuki Y, Nakao A, Fujita T, Noiri E. Radical scavenger edaravone developed for clinical use ameliorates ischemia/reperfusion injury in rat kidney. Kidney Int 2004;65:1714–1723. |
| 36. | Mishra J, Mori K, Ma Q, Kelly C, Yang J, Mitsnefes M, Barasch J, Devarajan P. Amelioration of ischemic acute renal injury by neutrophil gelatinase-associated lipocalin. J Am Soc Nephrol 2004;15:3073–3082. |
| 37. | Jerkić M, Miloradović Z, Jovović D, Mihailović-Stanojević N, Elena JV, Nastić-Mirić D, Grujić-Adanja G, Rodríguez-Barbero A, Marković-Lipkovski J, Vojvodić SB, et al. Relative roles of endothelin-1 and angiotensin II in experimental post-ischaemic acute renal failure. Nephrol Dial Transplant 2004;19:83–94. |
| 38. | Gong H, Wang W, Kwon TH, Jonassen T, Li C, Ring T, FrøkiAEr J, Nielsen S. EPO and alpha-MSH prevent ischemia/reperfusion-induced down-regulation of AQPs and sodium transporters in rat kidney. Kidney Int 2004;66:683–695. |
| 39. | Gueler F, Rong S, Park JK, Fiebeler A, Menne J, Elger M, Mueller DN, Hampich F, Dechend R, Kunter U, et al. Postischemic acute renal failure is reduced by short-term statin treatment in a rat model. J Am Soc Nephrol 2002;13:2288–2298. |
| 40. | Waikar SS, Betensky RA, Bonventre JV. Creatinine as the gold standard for kidney injury biomarker studies? Nephrol Dial Transplant 2009;24:3263–3265. |
| 41. | Bonventre JV, Zuk A. Ischemic acute renal failure: an inflammatory disease? Kidney Int 2004;66:480–485. |
| 42. | Aliprantis AO, Yang RB, Mark MR, Suggett S, Devaux B, Radolf JD, Klimpel GR, Godowski P, Zychlinsky A. Cell activation and apoptosis by bacterial lipoproteins through toll-like receptor-2. Science 1999;285:736–739. |
| 43. | Cowland JB, Sørensen OE, Sehested M, Borregaard N. Neutrophil gelatinase-associated lipocalin is up-regulated in human epithelial cells by IL-1 beta, but not by TNF-alpha. J Immunol 2003;171:6630–6639. |
| 44. | Flo TH, Smith KD, Sato S, Rodriguez DJ, Holmes MA, Strong RK, Akira S, Aderem A. Lipocalin 2 mediates an innate immune response to bacterial infection by sequestrating iron. Nature 2004;432:917–921. |
| 45. | Martensson, J, Bell M, Oldner A, Xu S, Venge P, Martling CR. Neutrophil gelatinase-associated lipocalin in adult septic patients with and without acute kidney injury. Intensive Care Med 2010;36:1333–1340. |
