Cardiac Intensive Care
Postoperative Neutrophil Gelatinase–Associated Lipocalin Predicts Acute Kidney Injury After Pediatric Cardiac Surgery* Andrés José Alcaraz, PhD1; Maite Augusta Gil-Ruiz, MD, PhD1; Ana Castillo, MD2; Jorge López, MD1; Carlos Romero, MD3; Sarah Nicole Fernández, MD1; Ángel Carrillo, PhD1
Objective: We investigated the temporal pattern and predictive value of neutrophil gelatinase–associated lipocalin for early identification of acute kidney injury in children undergoing cardiac surgery. Design: Prospective observational cohort study. Setting: One PICU in a tertiary medical center in Madrid, Spain. Patients: One hundred six children older than 15 days and younger than 16 years undergoing surgery for congenital cardiac lesions. Interventions: None. Measurements and Main Results: Urine samples were obtained before and at intervals after surgery. Acute kidney injury was defined according to pediatric Risk, Injury, Failure, Loss, and Endstage kidney disease criteria. The temporal pattern of both urine neutrophil gelatinase–associated lipocalin absolute concentration elevation and normalized to urine creatinine concentration was correlated with the development of acute kidney injury and other clinical outcomes. We evaluated the predictive ability of both urine neutrophil gelatinase–associated lipocalin and urine neutrophil gelatinase–associated lipocalin/creatinine by area under the curve, when added to a clinical predictive model. Data from 106 pediatric patients were analyzed. Acute kidney injury occurred in 42 patients (39.6%). Urine neutrophil gelatinase–associated lipocalin significantly increased in patients with acute kidney injury at 1, 3, and 15 hours postoperatively. Urine neutrophil gelatinase– associated lipocalin and urine neutrophil gelatinase–associated lipocalin/creatinine correlated with surgical variables and clinical outcomes. Acute kidney injury prediction improved when urine neutrophil gelatinase–associated lipocalin was added to a clinical *See also p. 173. 1 Pediatric Intensive Care Unit, Gregorio Marañón General University Hospital, Madrid, Spain. 2 Pediatric Intensive Care Unit, Albacete General University Hospital, Albacete, Spain. 3 Biochemistry and Laboratory Service, Albacete General University Hospital, Albacete, Spain. The authors disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2013 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies DOI: 10.1097/PCC.0000000000000034
Pediatric Critical Care Medicine
model (area under the curve increased at 1 hr from 0.85 to 0.91 and at 3 hr to 0.92). Neither the urine neutrophil gelatinase–associated lipocalin nor the urine neutrophil gelatinase–associated lipocalin/creatinine values were significantly different between patients with prerenal and sustained acute kidney injury. Conclusions: Urine neutrophil gelatinase–associated lipocalin is a predictive biomarker for acute kidney injury after pediatric cardiac surgery, and it may permit earlier intervention that improves outcome of acute kidney injury. Urine neutrophil gelatinase–associated lipocalin normalized to urine creatinine improves the prediction of acute kidney injury severity but offers no advantage in acute kidney injury diagnosis. (Pediatr Crit Care Med 2014; 15:121–130) Key Words: acute kidney injury; biomarker; cardiac surgery; children; creatinine; critical care; neutrophil gelatinase–associated lipocalin
A
cute kidney injury (AKI) is a frequent complication after pediatric cardiac surgery (1–3) and is associated with major adverse outcomes, including longer stay in ICUs, longer duration of mechanical ventilation (MV), and higher mortality even in the milder degrees of AKI (4). In clinical practice, the diagnosis of AKI is based on the increased concentration of serum creatinine (SCr), which is a late marker, is unreliable in the acute phase of renal damage, and is also affected by several nonrenal factors (5–8). Retrospective studies suggest that even minor degrees of postoperative AKI, as manifest by only 0.2–0.3 rise in SCr from baseline, predict a significant increase in short-term mortality (1, 4, 9). The failure of intervention trials to mitigate AKI after cardiac surgery has been attributed, in part, to delayed diagnosis. The application of new technologies like functional genomics and proteomics to human and animal models of AKI has identified several novel genes and soluble gene products overexpressed by renal cells during an ischemic insult that are emerging as promising biomarkers for the early detection of kidney injury (5, 10). Recent studies have focused on the evaluation and validation of biomarkers for early diagnosis of AKI in different clinical scenarios (10, 11). www.pccmjournal.org
121
Alcaraz et al
Neutrophil gelatinase–associated lipocalin (NGAL), a 25-kDa protein, is produced in the renal tubules in response to ischemic and nephrotoxic injury of kidney and is secreted through the urine (12–14). Multiple studies have validated NGAL as a biomarker of ischemic AKI in different populations (13–18), but studies have yielded a wide range of accuracy, with variable optimal cutoff values for the diagnosis of AKI (19–22). Urine NGAL (uNGAL) determination could favor the diagnosis of AKI and enable early intervention and an improved outcome. We conducted a large, prospective, cohort study of pediatric patients to evaluate the usefulness of uNGAL determination as a marker for early AKI diagnosis in children undergoing cardiac surgery for congenital heart diseases and determine the predictive value when added to a clinical predictive model. Besides, we intended to establish an optimal cutoff value of uNGAL concentration for the early diagnosis of AKI in our population. As a secondary objective, we analyze the relationship between early postoperative uNGAL concentration and clinical outcomes.
MATERIALS AND METHODS Study Design We carried out a prospective, observational, and analytical study including all patients older than 15 days and younger than 16 years undergoing surgery for congenital cardiac lesions from January to December 2010 during 1-year period and admitted to the PICU of a tertiary hospital. Patients undergoing heart transplantation and those with preexisting renal dysfunction (reduced kidney function) (SCr 2 times age-adjusted normal range) were excluded. This study was approved by the Institutional Review Board of Gregorio Marañón Children’s Hospital Medical Center. Written informed consent was obtained before enrollment from the legal guardian of each patient. Variable Description We collected preoperative characteristics: anthropometric data (age, weight, height, and body surface area), diagnosis (including single-ventricle physiology), and type of surgery performed in every patient. Relevant background data such as prematurity, syndromes (Down syndrome), and previous cardiac surgeries were also collected. Intraoperative surgical variables related to aggression including cardiopulmonary bypass (CPB) time, the aortic crossclamp use and duration, the minimum temperature on bypass, and the total circulatory arrest (TCA) use were also recorded. We used the Risk Adjustment for Congenital Heart Surgery (RACHS)-1 consensus-based scoring system to categorize the complexity of surgery (23). Outcome Definitions The following clinical outcome data were recorded: MV use and its duration, PICU length of stay (LOS), hospital LOS, requirement of renal replacement therapy, and death. The presence or absence of postoperative low cardiac output and 122
www.pccmjournal.org
use of inotropic drugs were also registered. The definition of a low cardiac output state was identical to criteria published by Hoffman et al (24). The primary outcome was AKI development, defined as a 50% increase in SCr from preoperative baseline values, and AKI severity was based on the pediatricmodified RIFLE criteria (pRIFLE) (25). Preoperative creatinine values (obtained within 72 hr before surgery) were used for evaluation of the baseline renal function, and postoperative SCr was routinely measured immediately after surgery and at least daily in the postoperative period. We calculated pRIFLE scores by using the estimated glomerular filtration rate (GFR), calculated from the estimated creatinine clearance with the Schwartz formula (26). The duration of AKI was defined as the number of days SCr was 50% above baseline. We considered severe AKI when injury or failure criteria were fulfilled. “Early” AKI was established as the renal dysfunction taking place in the first postoperative 72 hours (27), possibly related to surgery and CPB procedures, and “late” AKI as the one occurring after the fourth postoperatory day, which can be secondary to low cardiac output or other complications, such as sepsis or drug toxicity. Prerenal AKI was defined as transient AKI that resolves in less than 48 hours (11, 19). Protocol of the Study SCr was determined by a modified Jaffé method standardized against isotope dilution mass spectometry (Roche Diagnostics, Basel, Switzerland) in blood samples. Laboratory investigators were blinded to clinical outcomes. Urine samples for baseline measurements were obtained after bladder catheterization placed in operatory room. When the patients arrived to the PICU, the urine collected in the operatory room was measured and separated. We collected the urine output over the first hour and took a sample for analysis (1 hr). Subsequently, we repeated the process in the third hour (3 hr) and in the 15th hour (15 hr). Samples for uNGAL determination were immediately sent to the laboratory and centrifuged at 2,000 × g for 5 minutes. An aliquot from the supernatant was stored in polypropylene tubes without additives at –80°C until NGAL analysis. Urine samples were just thawed once on the day of analysis. uNGAL was measured by means of a standardized clinical platform (ARCHITECT, Abbott Diagnostics, Abbott Park, IL) (29, 30). The Architect NGAL assay is a two-step sandwich immunoassay using Chemiluminescent Microparticle technology. The reported functional sensitivity of the assay is less than 2 ng/mL (20% coefficient of variation [CV]) and total CV less than 5%. In addition to NGAL, we determined protein, creatinine, urea, glucose, sodium, potassium, chloride, and osmolality of each urine sample. Urinary excretion of NGAL was also normalized to urine creatinine to control the possible dilution on uNGAL concentration (31). NGAL results were analyzed as absolute values of urine concentration (ng/mL) and as uNGAL to urine creatinine ratio (uNGAL/cr) (µg/g). February 2014 • Volume 15 • Number 2
Cardiac Intensive Care
Statistical Methods Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS, Chicago, IL) and MedCalc (MedCalc, Mariakerke, Belgium). Continuous variables are expressed as median (P25–P75) and categorical variables as frequencies (%). Mann-Whitney U test and Kruskal-Wallis test were used to compare continuous variables and Pearson chi-square to compare categorical variables (or Fisher exact test as indicated). Statistical significance was defined as p value of less than or equal to 0.05. The covariate analysis for the repeated-measures data on each subject was performed by using the general linear model repeated-measures procedure with simple contrasts, in which the reference category was the first category (baseline). Spearman correlation coefficients were calculated to evaluate the possible association between uNGAL and uNGAL/cr ratio at each postoperative time point and surgical and clinical outcome variables. In addition, we performed Spearman partial correlations, adjusting for CPB time, RACHS-1 score, and age, to determine whether uNGAL or uNGAL/cr was independently correlated with the clinical outcome variables. Univariable logistic regression was used to assess the discriminative ability of NGAL concentration and of the NGAL/cr ratio to predict AKI, and receiver-operating characteristic (ROC) curves were generated at each postoperative time point. We calculate the area under the curve (AUC) including 95% CIs. We selected the most favorable cutoff values (according to their sensitivity and specificity). In a secondary analysis, we examined the ability of NGAL and ratio NGAL/cr to predict severe AKI (fulfilled criterion I or F of the pRIFLE classification). Multivariable logistic regression analysis was used to assess the predictors of AKI and the performance of the predictive models. Potential predictor variables included age, sex, CPB time, TCA use, RACHS-1 score, and previous cardiac surgeries. In addition, postoperative uNGAL concentrations and ratio uNGAL/cr were included individually as potential predictors. A clinical model containing clinical variables (age, CPB time, total circulatory arrest use, and RACHS-1 score) was first determined using backward elimination. The uNGAL and uNGAL/cr ratios at each earlier postoperative time point were added individually to the clinical model, and we calculated the AUC to evaluate the improvement in the model performance. The AUC were compared using nonparametric tests (32).
RESULTS During this period, 114 patients underwent cardiac surgery, five were heart transplants. We enrolled 109 patients and excluded three of them (2.75%), because of insufficient volume samples. One of these patients presented severe oliguria and kidney failure associated with a postoperative low cardiac output state requiring mechanical circulatory support. In the other two patients, the insufficient samples were due to errors in the sample took place. No patient presented preexisting renal dysfunction. Finally, 106 children were included for analysis (25 mo [6–72 mo] old). Forty-six of the patients (43.4%) were younger than 12 months, and 58% were boys. Demographic, surgical, Pediatric Critical Care Medicine
and clinical characteristics of enrolled patients are shown in Table 1. CPB was used in most of surgeries (96%) and was followed by a modified ultrafiltration protocol in these cases. MV in the postoperative period was used in 72 children, lasting more than 24 hours in 43 cases. The duration of stay at PICU was 6 days (4–10 d), and three children died. Preoperative baseline SCr values were available in all patients. Postoperative AKI according to pRIFLE classification was presented in 42 patients (39.6%). Among these patients with AKI, 23 (55%) fulfilled the R criteria, 11 (26%) fulfilled the I criteria, and eight (19%) fulfilled the F criteria. In 36 patients (34%), AKI was initiated prior to the 72 postoperative hours, considered as early AKI. These patients with early AKI were stratified by severity according to pRIFLE criteria: 22 patients (61%) developed pRIFLE-R, seven (19%) developed pRIFLE-I, and seven (19%) developed pRIFLE-F. Prerenal AKI (AKI duration < 48 hr) occurred in 22 of the patients with early AKI (61%). Six patients (5.7%) required continuous renal replacement therapies in the postoperative period, and in four of them, renal replacement was initiated in the first 72 postoperative hours. Patients’ outcome according to the postoperative AKI development and univariate and bivariate analyses comparing patients with or without AKI are presented in Table 1. Patients in whom early AKI developed had significantly higher baseline estimated GFR, but no differences in baseline SCr values nor in their age were found, although we found higher percentage of male patients. Patients with AKI had higher RACHS-1 score, longer CPB and cross-clamping times, and lower minimum temperature on bypass. They had longer MV time and had longer PICU and hospital stays. uNGAL and Postoperative AKI Results uNGAL concentrations were low in all patients before surgery, as shown in Table 2. We observed that these baseline uNGAL concentrations were similar in AKI and non-AKI groups of patients. uNGAL concentrations increased reached the peak level markedly at 1 and 3 hours postoperatively and reached the peak level during the whole study period (Fig. 1A). Interestingly, at these times, the urine values of NGAL were markedly higher in patients with AKI compared with the other group of patients without AKI. Subsequently, uNGAL concentrations declined (Fig. 2) in the 15 hours, although patients with AKI still demonstrated significantly higher urinary NGAL levels than patients without AKI (Fig. 1B). Analyzing the uNGAL/cr ratio, we found similar significant results, as shown in Table 2 and Figure 1B. However, the increase in SCr by 50% or greater from baseline was delayed by at least 1 day after PICU admission, as shown in Figure 2. There was a significant difference in the uNGAL concentrations or the uNGAL/cr ratio in each of the pRIFLE categories and the patients without AKI at all postoperative time points (data not shown). However, there was no significant correlation between uNGAL concentrations and the severity of AKI www.pccmjournal.org
123
Alcaraz et al
Table 1. Demographic, Surgical, and Clinical Characteristics of the All Patients and According to the Presence or Absence of Early Postoperative Acute Kidney Injury Variable
All (n = 106)
Early AKI (n = 36)
Non-AKI (n = 70)
p
Preoperative variables Age (mo)
25.0 (6.0–72.0)
60.0 (6.0–82.8)
17.0 (5.8–60.0)
0.166
Weight (kg)
11.2 (5.8–19.6)
15.1 (6.0–22.1)
10.0 (5.7–18.1)
0.136
84 (63–114)
106 (63–121)
77 (63–112)
0.198
0.46 (0.32–0.75)
0.128
Height (cm) Body area (m ) 2
0.52 (0.32–0.78)
0.69 (0.32–0.85)
Male gender (%)
62 (58.5)
26 (72.2)
36 (51.4)
0.040
Previous surgery
46 (43.4)
20 (55.6)
26 (37.1)
0.070
Baseline estimated glomerular filtration rate (mL/min/1.73 m2)
134 (102–155)
141 (126–174)
124 (96–153)
0.003
Baseline serum creatinine (mg/dL)
0.34 (0.27–0.44)
0.32 (0.25–0.44)
0.36 (0.28–0.44)
0.284
Variables related to surgery CPB use
102 (96.2)
CPB time (min)
116 (79–158)
Cross-clamp use
77 (72.6)
Cross-clamp time (min)
51 (31–86)
Total circulatory arrest use
21 (19.8)
Minimum temperature on bypass (°C)
30 (23–32)
Ventricle unique physiology
36 (100) 173 (130–250) 22 (66.7) 101 (61–151) 9 (27.3)
66 (94.3) 97 (68–125) 51 (76.1) 42 (26–63)
0.297 < 0.001 0.317 < 0.001
10 (14.9)
0.139
28 (16–30)
30 (28–32)
0.001
35 (33.0)
16 (44.4)
19 (27.1)
0.073
Left ventricular hypoplastic syndrome
16 (15.1)
8 (22.2)
8 (11.4)
0.142
RACHS-1 score
3.0 (2.0–3.0)
3.0 (2.3–4.0)
2.0 (1.8–3.0)
RACHS-1 score
< 0.001
1
18
1
17
2
33
8
25
3
37
16
21
4
14
7
7
5
4
4
0
RACHS-1 ≥ 4
< 0.001
18 (17.0)
11 (30.6)
7 (10.0)
0.008
3 (2.8)
1 (2.8)
2 (2.9)
1.000
Ventilated
72 (67.9)
29 (80.6)
43 (61.4)
0.046
Hours ventilated
16 (0–96)
48 (15–192)
5 (0–72)
0.002
PICU LOS (d)
6.0 (4.0–10.0)
7.0 (5.0–17.0)
5.0 (3.2–7.8)
0.001
Hospital LOS (d)
15 (8–24)
23 (14–37)
12 (7–19)
Postoperative variables Death
< 0.001
AKI = acute kidney injury, CPB = cardiopulmonary bypass, RACHS = Risk Adjustment for Congenital Heart Surgery, LOS = length of stay. Continuous variables presented as median (P25–P75) and categorical variables as frequencies (%).
according to pRIFLE criteria. In this way, urinary concentrations of NGAL at 1 and 3 hours in patients with pRIFLE-R were similar to those in patients with pRIFLE-I or pRIFLEF. However, the uNGAL/cr ratios at 15 hours were higher in patients with pRIFLE-I (404 [247–4,925] vs 59 [27–88] pg/g, 124
www.pccmjournal.org
p = 0.010) or pRIFLE-F (329 [153–979] vs 59 [27–88] pg/g, p = 0.011) than those in patients with pRIFLE-R (Fig. 3). uNGAL levels were not significantly different in patients with prerenal AKI compared with those in patients with longer lasting AKI. Similar results were found when uNGAL/cr ratio February 2014 • Volume 15 • Number 2
Cardiac Intensive Care
Table 2. Median Values (P25–P75) of Urine Neutrophil Gelatinase–Associated Lipocalin (ng/mL) and Urine Neutrophil Gelatinase–Associated Lipocalin/Creatinine Ratios (μg/g cr) in Children With and Without Early Postoperative Acute Kidney Injury Variables
p
All (n = 106)
No AKI (n = 70)
Early AKI (n = 36)
4.9 (1.8–19.8)
5.4 (2.6–22.1)
4.4 (1.6–17.8)
uNGAL 1 hr
56.4 (6.3–321.2)
15.5 (3.7–94.2)
350.0 (119.5–896.5)
< 0.001
uNGAL 3 hr
17.6 (6.1–166.8)
10.0 (4.80–32.6)
254.4 (78.1–707.3)
< 0.001
uNGAL 15 hr
12.5 (7.2–26.2)
9.4 (4.7–18.1)
21.6 (14.6–66.2)
< 0.001
uNGAL/cr preoperative
10.6 (3.4–44.1)
11.3 (3.7–45.2)
9.3 (3.8–32.1)
uNGAL preoperative
uNGAL/cr 1 hr
505.4 (64.3–3,326.0)
110.0 (36.9–740.1)
uNGAL/cr 3 hr
119.0 (29.9–1,106.1)
57.4 (22.6–212.5)
uNGAL/cr 15 hr
34.0 (13.4–150.6)
22.1 (11.6–68.0)
0.560
0.517
3,952.5 (1,137.3–9,771.7)
< 0.001
1,642.5 (324.0–6,925.7)
< 0.001
88.0 (37.8–417.5)
< 0.001
AKI = acute kidney injury, uNGAL = urine neutrophil gelatinase–associated lipocalin, uNGAL/cr = urine neutrophil gelatinase–associated lipocalin/creatinine.
was analyzed, except for values of ratio uNGAL/cr at 15 hours, which showed higher values for prolonged AKI (247 [113–718] vs 61 [27–158] pg/g, p = 0.032).
found that the uNGAL/cr ratio at 3 hours was independently correlated with a longer MV time (r = 0.420, p < 0.001) and with longer PICU LOS (r = 0.295, p = 0.006).
Associations Between uNGAL and Clinical Characteristics uNGAL at 1 and 3 hours after admission correlated with longer CPB time, longer cross-clamping time, and higher RACHS-1 score. uNGAL/cr ratio in this early time point also correlated with longer CPB time, longer cross-clamping time, and higher RACHS-1 score, but in addition, these values correlated with lower minimum temperature on bypass and with clinical outcome variables (longer all of them ventilation time, PICU stay, and hospital stay). Spearman correlation coefficients for uNGAL and uNGAL/cr ratio with surgical and clinical variables are shown in Table 3. We performed Spearman partial correlations, adjusting for CPB time, RACHS-1 score, and age, to determine whether uNGAL/cr ratio was independently correlated with the clinical outcome variables. Therefore, we
Predictive Ability of uNGAL and uNGAL/cr ROC curves for each early postoperative time point of uNGAL and uNGAL/cr ratio (1 and 3 hr after PICU admission) of the univariable logistic regression analysis are presented in Figure 4. We found that uNGAL values were predictive of AKI at each postoperative time point. Besides, when the uNGAL/cr ratio was analyzed, we found a similar discrimination (all p
Postoperative Neutrophil Gelatinase–Associated Lipocalin Predicts Acute Kidney Injury After Pediatric Cardiac Surgery* Andrés José Alcaraz, PhD1; Maite Augusta Gil-Ruiz, MD, PhD1; Ana Castillo, MD2; Jorge López, MD1; Carlos Romero, MD3; Sarah Nicole Fernández, MD1; Ángel Carrillo, PhD1
Objective: We investigated the temporal pattern and predictive value of neutrophil gelatinase–associated lipocalin for early identification of acute kidney injury in children undergoing cardiac surgery. Design: Prospective observational cohort study. Setting: One PICU in a tertiary medical center in Madrid, Spain. Patients: One hundred six children older than 15 days and younger than 16 years undergoing surgery for congenital cardiac lesions. Interventions: None. Measurements and Main Results: Urine samples were obtained before and at intervals after surgery. Acute kidney injury was defined according to pediatric Risk, Injury, Failure, Loss, and Endstage kidney disease criteria. The temporal pattern of both urine neutrophil gelatinase–associated lipocalin absolute concentration elevation and normalized to urine creatinine concentration was correlated with the development of acute kidney injury and other clinical outcomes. We evaluated the predictive ability of both urine neutrophil gelatinase–associated lipocalin and urine neutrophil gelatinase–associated lipocalin/creatinine by area under the curve, when added to a clinical predictive model. Data from 106 pediatric patients were analyzed. Acute kidney injury occurred in 42 patients (39.6%). Urine neutrophil gelatinase–associated lipocalin significantly increased in patients with acute kidney injury at 1, 3, and 15 hours postoperatively. Urine neutrophil gelatinase– associated lipocalin and urine neutrophil gelatinase–associated lipocalin/creatinine correlated with surgical variables and clinical outcomes. Acute kidney injury prediction improved when urine neutrophil gelatinase–associated lipocalin was added to a clinical *See also p. 173. 1 Pediatric Intensive Care Unit, Gregorio Marañón General University Hospital, Madrid, Spain. 2 Pediatric Intensive Care Unit, Albacete General University Hospital, Albacete, Spain. 3 Biochemistry and Laboratory Service, Albacete General University Hospital, Albacete, Spain. The authors disclosed that they do not have any potential conflicts of interest. For information regarding this article, E-mail: [email protected] Copyright © 2013 by the Society of Critical Care Medicine and the World Federation of Pediatric Intensive and Critical Care Societies DOI: 10.1097/PCC.0000000000000034
Pediatric Critical Care Medicine
model (area under the curve increased at 1 hr from 0.85 to 0.91 and at 3 hr to 0.92). Neither the urine neutrophil gelatinase–associated lipocalin nor the urine neutrophil gelatinase–associated lipocalin/creatinine values were significantly different between patients with prerenal and sustained acute kidney injury. Conclusions: Urine neutrophil gelatinase–associated lipocalin is a predictive biomarker for acute kidney injury after pediatric cardiac surgery, and it may permit earlier intervention that improves outcome of acute kidney injury. Urine neutrophil gelatinase–associated lipocalin normalized to urine creatinine improves the prediction of acute kidney injury severity but offers no advantage in acute kidney injury diagnosis. (Pediatr Crit Care Med 2014; 15:121–130) Key Words: acute kidney injury; biomarker; cardiac surgery; children; creatinine; critical care; neutrophil gelatinase–associated lipocalin
A
cute kidney injury (AKI) is a frequent complication after pediatric cardiac surgery (1–3) and is associated with major adverse outcomes, including longer stay in ICUs, longer duration of mechanical ventilation (MV), and higher mortality even in the milder degrees of AKI (4). In clinical practice, the diagnosis of AKI is based on the increased concentration of serum creatinine (SCr), which is a late marker, is unreliable in the acute phase of renal damage, and is also affected by several nonrenal factors (5–8). Retrospective studies suggest that even minor degrees of postoperative AKI, as manifest by only 0.2–0.3 rise in SCr from baseline, predict a significant increase in short-term mortality (1, 4, 9). The failure of intervention trials to mitigate AKI after cardiac surgery has been attributed, in part, to delayed diagnosis. The application of new technologies like functional genomics and proteomics to human and animal models of AKI has identified several novel genes and soluble gene products overexpressed by renal cells during an ischemic insult that are emerging as promising biomarkers for the early detection of kidney injury (5, 10). Recent studies have focused on the evaluation and validation of biomarkers for early diagnosis of AKI in different clinical scenarios (10, 11). www.pccmjournal.org
121
Alcaraz et al
Neutrophil gelatinase–associated lipocalin (NGAL), a 25-kDa protein, is produced in the renal tubules in response to ischemic and nephrotoxic injury of kidney and is secreted through the urine (12–14). Multiple studies have validated NGAL as a biomarker of ischemic AKI in different populations (13–18), but studies have yielded a wide range of accuracy, with variable optimal cutoff values for the diagnosis of AKI (19–22). Urine NGAL (uNGAL) determination could favor the diagnosis of AKI and enable early intervention and an improved outcome. We conducted a large, prospective, cohort study of pediatric patients to evaluate the usefulness of uNGAL determination as a marker for early AKI diagnosis in children undergoing cardiac surgery for congenital heart diseases and determine the predictive value when added to a clinical predictive model. Besides, we intended to establish an optimal cutoff value of uNGAL concentration for the early diagnosis of AKI in our population. As a secondary objective, we analyze the relationship between early postoperative uNGAL concentration and clinical outcomes.
MATERIALS AND METHODS Study Design We carried out a prospective, observational, and analytical study including all patients older than 15 days and younger than 16 years undergoing surgery for congenital cardiac lesions from January to December 2010 during 1-year period and admitted to the PICU of a tertiary hospital. Patients undergoing heart transplantation and those with preexisting renal dysfunction (reduced kidney function) (SCr 2 times age-adjusted normal range) were excluded. This study was approved by the Institutional Review Board of Gregorio Marañón Children’s Hospital Medical Center. Written informed consent was obtained before enrollment from the legal guardian of each patient. Variable Description We collected preoperative characteristics: anthropometric data (age, weight, height, and body surface area), diagnosis (including single-ventricle physiology), and type of surgery performed in every patient. Relevant background data such as prematurity, syndromes (Down syndrome), and previous cardiac surgeries were also collected. Intraoperative surgical variables related to aggression including cardiopulmonary bypass (CPB) time, the aortic crossclamp use and duration, the minimum temperature on bypass, and the total circulatory arrest (TCA) use were also recorded. We used the Risk Adjustment for Congenital Heart Surgery (RACHS)-1 consensus-based scoring system to categorize the complexity of surgery (23). Outcome Definitions The following clinical outcome data were recorded: MV use and its duration, PICU length of stay (LOS), hospital LOS, requirement of renal replacement therapy, and death. The presence or absence of postoperative low cardiac output and 122
www.pccmjournal.org
use of inotropic drugs were also registered. The definition of a low cardiac output state was identical to criteria published by Hoffman et al (24). The primary outcome was AKI development, defined as a 50% increase in SCr from preoperative baseline values, and AKI severity was based on the pediatricmodified RIFLE criteria (pRIFLE) (25). Preoperative creatinine values (obtained within 72 hr before surgery) were used for evaluation of the baseline renal function, and postoperative SCr was routinely measured immediately after surgery and at least daily in the postoperative period. We calculated pRIFLE scores by using the estimated glomerular filtration rate (GFR), calculated from the estimated creatinine clearance with the Schwartz formula (26). The duration of AKI was defined as the number of days SCr was 50% above baseline. We considered severe AKI when injury or failure criteria were fulfilled. “Early” AKI was established as the renal dysfunction taking place in the first postoperative 72 hours (27), possibly related to surgery and CPB procedures, and “late” AKI as the one occurring after the fourth postoperatory day, which can be secondary to low cardiac output or other complications, such as sepsis or drug toxicity. Prerenal AKI was defined as transient AKI that resolves in less than 48 hours (11, 19). Protocol of the Study SCr was determined by a modified Jaffé method standardized against isotope dilution mass spectometry (Roche Diagnostics, Basel, Switzerland) in blood samples. Laboratory investigators were blinded to clinical outcomes. Urine samples for baseline measurements were obtained after bladder catheterization placed in operatory room. When the patients arrived to the PICU, the urine collected in the operatory room was measured and separated. We collected the urine output over the first hour and took a sample for analysis (1 hr). Subsequently, we repeated the process in the third hour (3 hr) and in the 15th hour (15 hr). Samples for uNGAL determination were immediately sent to the laboratory and centrifuged at 2,000 × g for 5 minutes. An aliquot from the supernatant was stored in polypropylene tubes without additives at –80°C until NGAL analysis. Urine samples were just thawed once on the day of analysis. uNGAL was measured by means of a standardized clinical platform (ARCHITECT, Abbott Diagnostics, Abbott Park, IL) (29, 30). The Architect NGAL assay is a two-step sandwich immunoassay using Chemiluminescent Microparticle technology. The reported functional sensitivity of the assay is less than 2 ng/mL (20% coefficient of variation [CV]) and total CV less than 5%. In addition to NGAL, we determined protein, creatinine, urea, glucose, sodium, potassium, chloride, and osmolality of each urine sample. Urinary excretion of NGAL was also normalized to urine creatinine to control the possible dilution on uNGAL concentration (31). NGAL results were analyzed as absolute values of urine concentration (ng/mL) and as uNGAL to urine creatinine ratio (uNGAL/cr) (µg/g). February 2014 • Volume 15 • Number 2
Cardiac Intensive Care
Statistical Methods Statistical analysis was performed using the Statistical Package for the Social Sciences (SPSS, Chicago, IL) and MedCalc (MedCalc, Mariakerke, Belgium). Continuous variables are expressed as median (P25–P75) and categorical variables as frequencies (%). Mann-Whitney U test and Kruskal-Wallis test were used to compare continuous variables and Pearson chi-square to compare categorical variables (or Fisher exact test as indicated). Statistical significance was defined as p value of less than or equal to 0.05. The covariate analysis for the repeated-measures data on each subject was performed by using the general linear model repeated-measures procedure with simple contrasts, in which the reference category was the first category (baseline). Spearman correlation coefficients were calculated to evaluate the possible association between uNGAL and uNGAL/cr ratio at each postoperative time point and surgical and clinical outcome variables. In addition, we performed Spearman partial correlations, adjusting for CPB time, RACHS-1 score, and age, to determine whether uNGAL or uNGAL/cr was independently correlated with the clinical outcome variables. Univariable logistic regression was used to assess the discriminative ability of NGAL concentration and of the NGAL/cr ratio to predict AKI, and receiver-operating characteristic (ROC) curves were generated at each postoperative time point. We calculate the area under the curve (AUC) including 95% CIs. We selected the most favorable cutoff values (according to their sensitivity and specificity). In a secondary analysis, we examined the ability of NGAL and ratio NGAL/cr to predict severe AKI (fulfilled criterion I or F of the pRIFLE classification). Multivariable logistic regression analysis was used to assess the predictors of AKI and the performance of the predictive models. Potential predictor variables included age, sex, CPB time, TCA use, RACHS-1 score, and previous cardiac surgeries. In addition, postoperative uNGAL concentrations and ratio uNGAL/cr were included individually as potential predictors. A clinical model containing clinical variables (age, CPB time, total circulatory arrest use, and RACHS-1 score) was first determined using backward elimination. The uNGAL and uNGAL/cr ratios at each earlier postoperative time point were added individually to the clinical model, and we calculated the AUC to evaluate the improvement in the model performance. The AUC were compared using nonparametric tests (32).
RESULTS During this period, 114 patients underwent cardiac surgery, five were heart transplants. We enrolled 109 patients and excluded three of them (2.75%), because of insufficient volume samples. One of these patients presented severe oliguria and kidney failure associated with a postoperative low cardiac output state requiring mechanical circulatory support. In the other two patients, the insufficient samples were due to errors in the sample took place. No patient presented preexisting renal dysfunction. Finally, 106 children were included for analysis (25 mo [6–72 mo] old). Forty-six of the patients (43.4%) were younger than 12 months, and 58% were boys. Demographic, surgical, Pediatric Critical Care Medicine
and clinical characteristics of enrolled patients are shown in Table 1. CPB was used in most of surgeries (96%) and was followed by a modified ultrafiltration protocol in these cases. MV in the postoperative period was used in 72 children, lasting more than 24 hours in 43 cases. The duration of stay at PICU was 6 days (4–10 d), and three children died. Preoperative baseline SCr values were available in all patients. Postoperative AKI according to pRIFLE classification was presented in 42 patients (39.6%). Among these patients with AKI, 23 (55%) fulfilled the R criteria, 11 (26%) fulfilled the I criteria, and eight (19%) fulfilled the F criteria. In 36 patients (34%), AKI was initiated prior to the 72 postoperative hours, considered as early AKI. These patients with early AKI were stratified by severity according to pRIFLE criteria: 22 patients (61%) developed pRIFLE-R, seven (19%) developed pRIFLE-I, and seven (19%) developed pRIFLE-F. Prerenal AKI (AKI duration < 48 hr) occurred in 22 of the patients with early AKI (61%). Six patients (5.7%) required continuous renal replacement therapies in the postoperative period, and in four of them, renal replacement was initiated in the first 72 postoperative hours. Patients’ outcome according to the postoperative AKI development and univariate and bivariate analyses comparing patients with or without AKI are presented in Table 1. Patients in whom early AKI developed had significantly higher baseline estimated GFR, but no differences in baseline SCr values nor in their age were found, although we found higher percentage of male patients. Patients with AKI had higher RACHS-1 score, longer CPB and cross-clamping times, and lower minimum temperature on bypass. They had longer MV time and had longer PICU and hospital stays. uNGAL and Postoperative AKI Results uNGAL concentrations were low in all patients before surgery, as shown in Table 2. We observed that these baseline uNGAL concentrations were similar in AKI and non-AKI groups of patients. uNGAL concentrations increased reached the peak level markedly at 1 and 3 hours postoperatively and reached the peak level during the whole study period (Fig. 1A). Interestingly, at these times, the urine values of NGAL were markedly higher in patients with AKI compared with the other group of patients without AKI. Subsequently, uNGAL concentrations declined (Fig. 2) in the 15 hours, although patients with AKI still demonstrated significantly higher urinary NGAL levels than patients without AKI (Fig. 1B). Analyzing the uNGAL/cr ratio, we found similar significant results, as shown in Table 2 and Figure 1B. However, the increase in SCr by 50% or greater from baseline was delayed by at least 1 day after PICU admission, as shown in Figure 2. There was a significant difference in the uNGAL concentrations or the uNGAL/cr ratio in each of the pRIFLE categories and the patients without AKI at all postoperative time points (data not shown). However, there was no significant correlation between uNGAL concentrations and the severity of AKI www.pccmjournal.org
123
Alcaraz et al
Table 1. Demographic, Surgical, and Clinical Characteristics of the All Patients and According to the Presence or Absence of Early Postoperative Acute Kidney Injury Variable
All (n = 106)
Early AKI (n = 36)
Non-AKI (n = 70)
p
Preoperative variables Age (mo)
25.0 (6.0–72.0)
60.0 (6.0–82.8)
17.0 (5.8–60.0)
0.166
Weight (kg)
11.2 (5.8–19.6)
15.1 (6.0–22.1)
10.0 (5.7–18.1)
0.136
84 (63–114)
106 (63–121)
77 (63–112)
0.198
0.46 (0.32–0.75)
0.128
Height (cm) Body area (m ) 2
0.52 (0.32–0.78)
0.69 (0.32–0.85)
Male gender (%)
62 (58.5)
26 (72.2)
36 (51.4)
0.040
Previous surgery
46 (43.4)
20 (55.6)
26 (37.1)
0.070
Baseline estimated glomerular filtration rate (mL/min/1.73 m2)
134 (102–155)
141 (126–174)
124 (96–153)
0.003
Baseline serum creatinine (mg/dL)
0.34 (0.27–0.44)
0.32 (0.25–0.44)
0.36 (0.28–0.44)
0.284
Variables related to surgery CPB use
102 (96.2)
CPB time (min)
116 (79–158)
Cross-clamp use
77 (72.6)
Cross-clamp time (min)
51 (31–86)
Total circulatory arrest use
21 (19.8)
Minimum temperature on bypass (°C)
30 (23–32)
Ventricle unique physiology
36 (100) 173 (130–250) 22 (66.7) 101 (61–151) 9 (27.3)
66 (94.3) 97 (68–125) 51 (76.1) 42 (26–63)
0.297 < 0.001 0.317 < 0.001
10 (14.9)
0.139
28 (16–30)
30 (28–32)
0.001
35 (33.0)
16 (44.4)
19 (27.1)
0.073
Left ventricular hypoplastic syndrome
16 (15.1)
8 (22.2)
8 (11.4)
0.142
RACHS-1 score
3.0 (2.0–3.0)
3.0 (2.3–4.0)
2.0 (1.8–3.0)
RACHS-1 score
< 0.001
1
18
1
17
2
33
8
25
3
37
16
21
4
14
7
7
5
4
4
0
RACHS-1 ≥ 4
< 0.001
18 (17.0)
11 (30.6)
7 (10.0)
0.008
3 (2.8)
1 (2.8)
2 (2.9)
1.000
Ventilated
72 (67.9)
29 (80.6)
43 (61.4)
0.046
Hours ventilated
16 (0–96)
48 (15–192)
5 (0–72)
0.002
PICU LOS (d)
6.0 (4.0–10.0)
7.0 (5.0–17.0)
5.0 (3.2–7.8)
0.001
Hospital LOS (d)
15 (8–24)
23 (14–37)
12 (7–19)
Postoperative variables Death
< 0.001
AKI = acute kidney injury, CPB = cardiopulmonary bypass, RACHS = Risk Adjustment for Congenital Heart Surgery, LOS = length of stay. Continuous variables presented as median (P25–P75) and categorical variables as frequencies (%).
according to pRIFLE criteria. In this way, urinary concentrations of NGAL at 1 and 3 hours in patients with pRIFLE-R were similar to those in patients with pRIFLE-I or pRIFLEF. However, the uNGAL/cr ratios at 15 hours were higher in patients with pRIFLE-I (404 [247–4,925] vs 59 [27–88] pg/g, 124
www.pccmjournal.org
p = 0.010) or pRIFLE-F (329 [153–979] vs 59 [27–88] pg/g, p = 0.011) than those in patients with pRIFLE-R (Fig. 3). uNGAL levels were not significantly different in patients with prerenal AKI compared with those in patients with longer lasting AKI. Similar results were found when uNGAL/cr ratio February 2014 • Volume 15 • Number 2
Cardiac Intensive Care
Table 2. Median Values (P25–P75) of Urine Neutrophil Gelatinase–Associated Lipocalin (ng/mL) and Urine Neutrophil Gelatinase–Associated Lipocalin/Creatinine Ratios (μg/g cr) in Children With and Without Early Postoperative Acute Kidney Injury Variables
p
All (n = 106)
No AKI (n = 70)
Early AKI (n = 36)
4.9 (1.8–19.8)
5.4 (2.6–22.1)
4.4 (1.6–17.8)
uNGAL 1 hr
56.4 (6.3–321.2)
15.5 (3.7–94.2)
350.0 (119.5–896.5)
< 0.001
uNGAL 3 hr
17.6 (6.1–166.8)
10.0 (4.80–32.6)
254.4 (78.1–707.3)
< 0.001
uNGAL 15 hr
12.5 (7.2–26.2)
9.4 (4.7–18.1)
21.6 (14.6–66.2)
< 0.001
uNGAL/cr preoperative
10.6 (3.4–44.1)
11.3 (3.7–45.2)
9.3 (3.8–32.1)
uNGAL preoperative
uNGAL/cr 1 hr
505.4 (64.3–3,326.0)
110.0 (36.9–740.1)
uNGAL/cr 3 hr
119.0 (29.9–1,106.1)
57.4 (22.6–212.5)
uNGAL/cr 15 hr
34.0 (13.4–150.6)
22.1 (11.6–68.0)
0.560
0.517
3,952.5 (1,137.3–9,771.7)
< 0.001
1,642.5 (324.0–6,925.7)
< 0.001
88.0 (37.8–417.5)
< 0.001
AKI = acute kidney injury, uNGAL = urine neutrophil gelatinase–associated lipocalin, uNGAL/cr = urine neutrophil gelatinase–associated lipocalin/creatinine.
was analyzed, except for values of ratio uNGAL/cr at 15 hours, which showed higher values for prolonged AKI (247 [113–718] vs 61 [27–158] pg/g, p = 0.032).
found that the uNGAL/cr ratio at 3 hours was independently correlated with a longer MV time (r = 0.420, p < 0.001) and with longer PICU LOS (r = 0.295, p = 0.006).
Associations Between uNGAL and Clinical Characteristics uNGAL at 1 and 3 hours after admission correlated with longer CPB time, longer cross-clamping time, and higher RACHS-1 score. uNGAL/cr ratio in this early time point also correlated with longer CPB time, longer cross-clamping time, and higher RACHS-1 score, but in addition, these values correlated with lower minimum temperature on bypass and with clinical outcome variables (longer all of them ventilation time, PICU stay, and hospital stay). Spearman correlation coefficients for uNGAL and uNGAL/cr ratio with surgical and clinical variables are shown in Table 3. We performed Spearman partial correlations, adjusting for CPB time, RACHS-1 score, and age, to determine whether uNGAL/cr ratio was independently correlated with the clinical outcome variables. Therefore, we
Predictive Ability of uNGAL and uNGAL/cr ROC curves for each early postoperative time point of uNGAL and uNGAL/cr ratio (1 and 3 hr after PICU admission) of the univariable logistic regression analysis are presented in Figure 4. We found that uNGAL values were predictive of AKI at each postoperative time point. Besides, when the uNGAL/cr ratio was analyzed, we found a similar discrimination (all p
