Urine NGAL Predicts Severity of Acute Kidney Injury After Cardiac Surgery: A Prospective Study

Patients and Study Design

This investigation was approved by the institutional review board of the Cincinnati Children's Hospital Medical Center. All children undergoing elective CPB for surgical correction or palliation of congenital heart lesions between January 2004 and June 2006 were prospectively enrolled. We obtained written informed consent from the legal guardian of every participant before enrollment. Exclusion criteria included preexisting renal insufficiency, diabetes mellitus, peripheral vascular disease, and use of nephrotoxic drugs before or during the study period. Preexisting renal insufficiency was defined as a serum creatinine level that was greater than the 90th percentile for the child's age and gender. Patients with a history of potential nephrotoxin use (including aminoglycosides, angiotensin inhibitors, and nonsteroidal anti-inflammatory drugs) during the preoperative day or the first two postoperative days were excluded because of potential confounding effects on urinary NGAL measurements.

We used the Risk Adjustment for Congenital Heart Surgery 1 consensus-based scoring system to categorize the complexity of surgery (33) (Table 1). This method of risk stratification has emerged as a widely accepted tool for the evaluation of differences in outcomes of surgery for congenital heart disease.

To obviate postoperative volume depletion and prerenal azotemia, all subjects received at least 80% of their maintenance fluid requirements during the first 24 h after surgery and 100% maintenance subsequently. We obtained spot urine samples at baseline and at frequent intervals after initiation of CPB. Urine samples were centrifuged at 2000 × g for 5 min, and the supernatants stored in aliquots at −80°C. Serum creatinine was measured by the hospital clinical laboratory at baseline and routinely monitored at least twice daily during the first 2 d after CPB and at least daily after the third postoperative day.

The primary outcome variable was the development of AKI, defined as a 50% or greater increase in serum creatinine from baseline. Other outcomes included percent change in serum creatinine, days in AKI, dialysis requirement, length of hospital stay, and mortality. Other variables we obtained included age, gender, ethnic origin, CPB time, previous heart surgery, and urine output.

In a pilot cross-sectional study, we measured NGAL concentrations with 136 urine samples (NGAL range, 0.3 to 815 ng/ml) and 6 calibration standards (NGAL range, 0 to 1000 ng/ml) to determine the correlation between the two assay methods described below. In a subsequent prospective study, serial urine samples from 196 children undergoing CPB were assayed for NGAL using the standardized clinical platform (ARCHITECT® analyzer, Abbott Diagnostics) to assess its ability to predict AKI and other adverse outcomes.

NGAL Analysis by Standardized Clinical Platform (ARCHITECT®)

The ARCHITECT® NGAL assay is currently under development. High affinity antibodies were generated toward distinct epitopes on NGAL and assay standards were prepared using human recombinant NGAL. The assay is a two-step (sandwich) assay using Chemiluminescent Microparticle Immunoassay technology. The ARCHITECT® assay had a functional sensitivity <2 ng/ml (20% coefficient of variation [CV], 95% confidence) and total CVs <5.0%.

NGAL Analysis by ELISA

The urine NGAL ELISA was performed using a commercially available assay (NGAL ELISA Kit 036; AntibodyShop, Grusbakken, Denmark) that specifically detects human NGAL. The assay was performed as per the manufacturer's protocol. Briefly, 100 μl of NGAL standards or diluted samples was applied on to the precoated microwells in duplicates. Microwells were then incubated for 1 h at room temperature and then washed with washing buffer. In succession, biotinylated NGAL antibody and HRP-streptavidin were incubated in the wells for 1 h each with shaking at 200 × g. TMB substrate was added for 10 min in the dark before adding stop solution. Finally, NGAL concentration was measured at 450 nm wavelength in each well with reference reading at 620 nm in blank wells. The intra-assay CV were 2.1% (range, 1.3% to 4.0%) and 3.0% (range, 1.2% to 4.0%) in urine and plasma, respectively. Interassay variation was 9.1% (range, 6.8% to 18.1%) and 8.2% (range, 2.2% to 11.2%) in urine and plasma, respectively. Urine creatinine was measured using quantitative colorimetric Microplate Assay Kit (Oxford Biomedical Research, Oxford, MA) to standardize urinary NGAL for changes in urine concentration. Urinary NGAL excretion is presented as the amount of urinary NGAL in ng per ml urine as well as in ng per mg of urine creatinine to correct for differences in NGAL due to urine dilution. All measurements were made in triplicate. The laboratory investigators were blinded to the sample sources and clinical outcomes until the end of the study.

Statistical Analysis

Statistical analysis for Table 2 was performed using SAS, version 9.2. Either a two-sample t test or Mann-Whitney rank sum test was used for continuous variables and χ² or Fisher's exact test was used for categorical variables. The associations between NGAL and change in creatinine (Delta creat), duration of AKI (days), and hospital length of stay were assessed by Spearman Rank Order correlation analysis. Correlations between NGAL and AKI outcome, dialysis requirement, and death were performed using a logistic correlation. Multivariate stepwise multiple logistic regression analysis was undertaken to assess predictors of adverse clinical outcomes (duration of AKI and hospital length of stay). Potential independent predictor variables included age, gender, CPB time, and urinary NGAL values at 2, 4, and 6 h after CPB. The area under the curve (AUC) was calculated from a standard receiver-operating characteristic plot. Spearman Rank Order correlations were performed using the website www.wessa.net/rankcorr.wasp (v1.1.21-r4), and the logistic correlations were performed using JMP (v5.0.1a). An AUC of 0.5 is no better than expected by chance, whereas a value of 1.0 signifies a perfect biomarker. A P ≤ 0.05 was considered statistically significant.

Verification of the Standardized Clinical Platform (ARCHITECT®)

The cross-sectional pilot study was designed to verify the standardized clinical platform (ARCHITECT®) against the research-based NGAL ELISA assay. As shown in Figure 1, NGAL concentrations in 136 urine samples from subjects undergoing CPB (NGAL range, 0.3 to 815 ng/ml) and 6 calibration standards (NGAL range, 0 to 1000 ng/ml) determined by the two assays were highly correlated (r = 0.99), as shown in Figure 1. The slope difference observed between the assays is in large part due to differences in assay standardization between the assays. For commercialization, the ARCHITECT® assay will use highly purified, recombinant NGAL expressed in a mammalian system.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.

Correlation between urine NGAL measurements obtained by ARCHITECT® platform and research-based NGAL ELISA assay (AntibodyShop).

NGAL as a Predictor of AKI and Other Adverse Outcomes

In a subsequent prospective study, serial urine samples from 196 children who met the inclusion and exclusion criteria were assayed for NGAL using the ARCHITECT® platform to assess its ability to predict AKI and other adverse outcomes. Ninety-nine patients (51%) met the criteria for AKI within a 3-d period. However, the increase in serum creatinine by 50% or greater from baseline was delayed by 1 to 3 d after CPB, as shown in Figure 2. Of these 99 patients, the increase in serum creatinine was noted in 7 patients (7.1%) at 24 h after CPB, in 60 patients (60.1%) at 48 h, and in 32 patients (32.1%) at 72 h. The increase in serum creatinine was sustained for at least 4 d after CPB, suggesting the presence of intrinsic AKI rather than a prerenal etiology. Based on this primary outcome, we classified subjects into those with and without AKI. No differences were noted between the two groups with respect to age, gender, race, urine output, hypertension, baseline use of angiotensin converting enzyme inhibitors, or baseline serum creatinine levels (Table 2). In patients who developed AKI, the duration of CPB was significantly longer, the complexity of surgery was significantly greater by Risk Adjustment for Congenital Heart Surgery 1 scoring, and the clinical outcomes were significantly worse. The serum creatinine rose by a greater percentage in the AKI group, and both length of hospitalization and mortality rate were significantly higher (Table 2). Four patients required dialysis, and there were three deaths, all in the AKI group.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

Serum creatinine measurements obtained at various time points post-CPB. AKI, acute kidney injury, defined as a 50% increase in serum creatinine from baseline. Values are mean ± SEM and are shown in Table 2. *P < 0.05 comparing AKI versus no AKI groups.

By ARCHITECT® assay, the absolute urine NGAL measurements at baseline were comparable in the two groups (Table 3). In the non-AKI group, there was a small but statistically significant increase in urine NGAL at 2 and 4 h after CPB, which normalized back to baseline levels at later time points. In marked contrast, in subjects who subsequently developed AKI, there was a robust 15-fold increase in urine NGAL at 2 h after CPB, which was even further accentuated at the 4 h (25-fold increase) and 6 h (26-fold increase) time points (Table 3; Figure 3). In the AKI group, a statistically significant increase in urine NGAL from baseline was apparent up to 48 h after CPB. Similar findings were recorded when the urine NGAL measurements were corrected for urinary creatinine concentrations (Table 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

Urine NGAL measurements obtained by ARCHITECT® assay at various time points post-CPB. AKI, acute kidney injury, defined as a 50% increase in serum creatinine from baseline. Values are mean ± SEM and are shown in Table 3. *P < 0.05 comparing AKI versus no AKI groups.

To assess the relationship between urinary NGAL and AKI severity, we compared NGAL values in the entire AKI cohort with those that developed early AKI (≥50% increase in serum creatinine at 24 h after CPB) or required dialysis. In both scenarios, urinary NGAL levels were found to be significantly higher in subjects who had either an earlier rise in serum creatinine or needed renal replacement therapy (Table 5). This difference was seen in the 2 h urine NGAL levels and persisted in the 4 and 6 h NGAL measurements.

To assess independent predictors for the development of AKI in the entire cohort, multivariate logistic regression was performed. All variables that were found by univariate analysis to display a P < 0.1 were entered into the model. The most powerful independent predictors of AKI were the 6 h urine NGAL measurements in ng/ml (R² = 0.68, P < 0.0001) and in ng/mg creatinine (R² = 0.75, P < 0.0001). Age, gender, and race were not independent predictors of AKI.

To test the hypothesis that urine NGAL levels measured soon after CPB could be used to predict eventual clinical outcomes, the Spearman Rank Order correlation analysis or logistic regression analysis was performed as appropriate. The results are illustrated in Table 6. The 2, 4, and 6 h urine NGAL levels strongly correlated with percent change in serum creatinine, duration of AKI, and length of hospital stay (all P < 0.0001). The 2 h urine NGAL levels also correlated with subsequent dialysis requirement (r = 0.48, P = 0.01) as well as mortality (r = 0.53, P = 0.01).

To assess the utility of urine NGAL as a predictor of adverse clinical outcomes, a multivariate analysis was undertaken with duration of AKI and hospital length of stay as outcome variables and age, gender, CPB time, and NGAL levels as input variables. The 2 h NGAL value was found to be a strong independent predictor of duration of AKI (P < 0.0001). The 4 h NGAL measurement was a strong indicator of both duration of AKI (P < 0.0001) and length of stay (P = 0.0007). Similarly, the 6 h NGAL measurement was a powerful indicator of both duration of AKI (P < 0.0001) and length of stay (P < 0.0001).

To assess the utility of urine NGAL measurements at varying cutoff values to predict AKI, conventional receiver-operating characteristic curves were generated, and the AUCs calculated. Table 7 lists the derived sensitivities and specificities at different cutoff concentrations, and the calculated AUCs at each time point are shown in Table 3. For urine NGAL at 2 h after CPB, sensitivity and specificity were optimal at the 100 ng/ml cutoff, with an AUC of 0.93 for the prediction of AKI, indicative of an excellent biomarker. Using a similar approach, the 2 h urine NGAL was also an excellent biomarker for the prediction of dialysis requirement (AUC = 0.86) and mortality (AUC = 0.91). However, the number of events was small (n = 4 for dialysis requirement and n = 3 for mortality).