Background
Acute kidney injury (AKI) is associated with adverse outcomes in critically ill children, such as longer hospital stay and increased mortality, [
1‐
3]. Research on AKI has focused on identifying early biomarkers capable of detecting kidney injury before the rise in serum creatinine (sCr) being observed in recent decades, which could optimize and improve patient outcomes [
4‐
8]. Previous studies have demonstrated that the diagnosis of clinical AKI is delayed by at least 24-48 hours compared to diagnosis of the condition identified by using novel renal biomarkers of tubular injury, such as neutrophil gelatinase-associated lipocalin (NGAL) [
9,
10] and cystatin C (CysC) [
10,
11]. Therefore, questions arise regarding whether AKI could be diagnosed in the absence of the classic signs of clinical AKI and whether the condition of subclinical AKI, identified by damage or functional biomarkers in the absence of oliguria or increased sCr levels, is clinically significant. Recent studies suggest that subclinical AKI, which precedes clinical AKI and is defined as tubular injury in the absence of oliguria or increased sCr levels, is still AKI [
12‐
15] and has an increased risk of adverse outcomes in adult patients [
15]. However, whether subclinical AKI may occur frequently in critically ill neonates and children and whether the development of subclinical AKI in this population is associated with negative clinical outcomes remains to be elucidated.
Cystatin C, a 13-kDa protein, is normally filtered freely and is completely reabsorbed and catabolized in the proximal tubule and therefore is not normally found in urine in significant amounts. Increased levels of CysC in urine may reflect renal tubular injury and impairment [
16,
17]. Urinary CysC (uCysC) has been shown to be diagnostic of AKI in patients in the intensive care unit (ICU) [
11,
18,
19]. Our previous studies demonstrated that uCysC is associated with AKI and is independently predictive of AKI in critically ill neonates [
20,
21]. Increases in uCysC levels predict AKI 24–48 hours before the occurrence of diagnostic sCr increases in response to renal injury [
10,
11], suggesting that uCysC may identify subclinical kidney injury, in which kidney damage occurs without the development of decreased kidney function. However, whether uCysC could provide a method of detecting subclinical AKI in patients with apparently normal sCr is unknown. It also remains unclear whether uCysC is associated with increased mortality in critically ill neonates and children with and without AKI. The aims of the study were to investigate the association between uCysC level and AKI and mortality in critically ill neonates and children, to evaluate whether uCysC could detect subclinical AKI in patients with apparently normal sCr, and to determine whether uCysC-positive subclinical AKI was associated with adverse outcomes in this population.
Methods
Subject selection
All patients admitted to the neonatal and pediatric intensive care unit (NICU and PICU, respectively) from July to October 2016 were eligible for this prospective study. A priori exclusion criteria were known congenital abnormality of the kidney and failure to collect urine samples before discharge from the ICU or death. This study was approved by the Institutional Review Board at the Children’s Hospital of Soochow University and performed in accordance with the Declaration of Helsinki. Informed parental consent was obtained at enrollment.
Clinical data collection
We reviewed the medical records of eligible patients. Clinical status, medication, and therapies were recorded daily until ICU discharge or death. Sepsis, multiple organ dysfunction syndrome (MODS), shock, and disseminated intravascular coagulation (DIC) that developed during the ICU stay were diagnosed by the treating physicians, according to the criteria described previously [
22,
23].
Assessment of illness severity
The score for neonatal acute physiology (SNAP) and the score of the pediatric risk of mortality III (PRISM III), which were calculated on the day of NICU or PICU admission according to methods described in the original study [
24,
25] and in our previous studies [
22,
26], were used as a measure to assess illness severity.
Diagnosis of AKI
The diagnosis of neonatal AKI that developed during the first week after admission was based on (1) the neonatal AKI Kidney Disease: Improving Global Outcome (KDIGO) classification, i.e., sCr rise ≥ 0.3 mg/dL (26.5 μmol/L) within 48 h, sCr rise ≥ 1.5 times the reference sCr within 7 days, or urine output ≤ 0 .5 mL/kg/h for 6–12 h (the reference sCr was defined as the lowest previous sCr value) [
27] or (2) sCr > 1.5 mg/dL (132.6 μmol/L) sustained at least 48 h [
20]. The KDIGO criteria for sCr level and urine output were applied to define pediatric AKI that developed during the first week after PICU admission [
28]. Baseline sCr was defined as the lowest level obtained within 3 months prior to ICU admission [
1]. When the baseline sCr measurement was unavailable, the sCr value at the time of enrollment was used. For patients with elevated sCr ≥ 1.2 mg/dL (106.1 μmol/L) at PICU admission, the lowest sCr value within 2 weeks while in the PICU was considered the baseline sCr, which was adapted from a previous study [
29] and was in accordance with our previous study [
22]. When the two criteria resulted in different KDIGO stages, the higher stage was chosen. Severity of AKI was characterized by KDIGO staging. Severe AKI was defined as KDIGO stage 2 or 3 [
1].
Outcomes
The primary outcome was ICU mortality defined as all-cause mortality occurring during the ICU stay, including death resulting from withdrawal of therapy. Secondary outcomes were the duration of mechanical ventilation (MV) and the ICU length of stay.
Measurement of urinary CysC
Urine samples were collected every 48–72 hours during the first week after the first collection on the first day of being admitted to the ICU and immediately frozen and stored at -80°C. For the measurement of CysC, the urine sample was centrifuged at 1500 g at 4 °C for 10 min, and the supernatants were aliquoted and measured using an automatic biochemical analyzer (Hitachi 7600, Tokyo, Japan) and a latex enhanced immunoturbidimetry assay, as previously described [
20,
21]. The detection limit for CysC was 10 ng/mL. The intra-assay and inter-assay coefficients of variation were ≤ 10%, corresponding to that reported by the manufacturer. Urinary CysC value was expressed in nanograms per milligram of urinary Cr (ng/mg uCr). The uCr level from the same aliquoted sample was measured automatically using the sarcosine oxidase method.
Statistical analysis
Statistical analyses were performed using SPSS statistics. We first checked the assumptions of normality and homogeneity of variance. Skewed distributions are described by the median and interquartile range (IQR) and were compared using the Mann-Whitney U test or the Kruskal-Wallis H(K) test. Categorical variables are presented as counts and percentages, and Fisher’s exact test or the chi-square test was applied for comparison, as appropriate. Stepwise multivariate linear regression analyses were performed to investigate factors potentially associated with the levels of uCysC. Univariate and multivariate logistic regression analyses were performed to calculate the odds ratio (OR) and adjusted OR (AOR) with a 95% confidence interval (CI) to assess the association between variables and AKI and mortality. To analyze the predictive power, the receiver operating characteristic (ROC) curves were generated, the area under the ROC curve (AUC) was determined, and the nonparametric method of Delong was used to compare differences between AUCs. The sensitivity, specificity, positive likelihood ratio (LR+), and negative likelihood ratio (LR-) were calculated using Sigma Plot 10.0 software. Univariate and multivariate linear regression analyses were performed to investigate the association between uCysC and the duration of MV and the length of ICU stay. Continuous variables with skewed distributions were log-transformed for linear regression analysis. The two-sided alpha level was set at 0.05.
Discussion
Our data indicate that subclinical AKI may occur without detectable loss of kidney function and predict worse clinical outcomes, and that urinary CysC is capable of detecting subclinical AKI in critically ill neonates and children.
Urinary CysC was associated with AKI in critically ill neonates and children in this study. The findings further strengthen the evidence for a relationship between elevated uCysC and AKI [
11,
19,
20,
30]. Clinical studies investigating the association between uCysC and mortality have been limited mainly to adult patients [
18,
19]. Studies in general critically ill adults demonstrate that uCysC is an independent predictor of mortality [
18,
19]. In infants undergoing cardiac surgery with cardiopulmonary bypass, elevated uCysC in the early postoperative period is associated with a greater risk of death or need for renal replacement therapy [
31]. Compared to the previous study in infants undergoing cardiac surgery, our study included a general NICU and PICU population. Nevertheless, our results are in line with the previous study and indicate that uCysC is a sensitive biomarker for predicting mortality in critically ill neonates and children. For every 1% increase in uCysC (10,000 ng/mg uCr), the odds of mortality increased by 26%. We further demonstrated that the association between uCysC and mortality was independent of AKI and illness severity. Urinary CysC-positive critically ill neonates and children, with or without clinical AKI diagnosed by the KDIGO criteria, developed worse clinical outcomes.
The major finding in this study was that 25.5% of patients who had an increased risk of adverse outcomes were identified based on increased uCysC, in the absence of oliguria or increased sCr (uCysC+/AKI−). It is well-known that increased sCr is not detected until the glomerular filtration rate is reduced by > 50% [
15,
32]. AKI can be diagnosed early by damage or functional biomarkers preceding loss of filtration function [
12,
13,
15,
33]. Increased uCysC has been demonstrated to reflect renal tubular injury and impairment and predict AKI 24–48 h before sCr [
11]. Our data indicate that this substantial group of critically ill patients with positive uCysC who do not fulfill current KIDGO criteria for AKI are likely to have subclinical AKI, and uCysC can detect subclinical AKI when tubular damage has occurred without oliguria or appreciable increase in sCr.
To our knowledge, this study is the first prospective investigation to determine the association between subclinical AKI and clinical outcomes in critically ill patients. The development of subclinical AKI in critically ill neonates and children is associated with negative clinical outcomes. Our data agree with a previous report conducted in adult patients with cardiorenal syndrome type 1, which was a retrospective analysis of pooled data from ten prospective observational studies of neutrophil gelatinase-associated lipocalin (NGAL) and demonstrated that in the absence of a diagnostic increases in sCr, NGAL can be used to identify patients with subclinical AKI who have an increased risk of adverse outcomes [
15]. It is known that sCr fails to identify likely AKI in some patients who are at an increased risk of death [
9,
15]. The patients with uCysC-positive subclinical AKI were at greater risk of longer ICU stay and mortality compared to those with uCysC(−)/AKI(−), suggesting that the concept of subclinical AKI as a diagnosis with promise for improving the management of AKI is clinically relevant and that critically ill neonates and children with subclinical AKI might benefit from early diagnosis and treatment.
In this study, we also found that 8.4% of the patients had no biomarker evidence of tubular injury but had loss of kidney function (uCysC−/AKI+), suggesting that this subgroup of patients who had functional change without damage might have pre-renal azotemia, which is considered a volume-responsive, reversible alteration in kidney function [
8,
34]. These patients had a fourfold increased risk of mortality compared to patients negative for both (uCysC−/AKI−). A further study is warranted to explore whether the combined use of uCysC with sCr and/or urine output might differentiate between AKI and pre-renal azotemia.
Moreover, the patients with uCysC(+)/AKI(+) had the greatest risk of mortality, which is consistent with findings from the multicenter pooled analysis of the adult studies [
15] and supports the concept that subclinical tubular injury precedes detectable decreased kidney function and that when both features occur together, clinical outcomes are worse. Interestingly, our data confirm the potential classification of patients according to the conceptual framework based on a combination of damage and functional biomarkers. The framework is proposed by the Acute Dialysis Quality Initiative (ADQI) conference and provides a potent approach for assessing patients with AKI for diagnosis, staging, differential diagnosis, and prognosis in order to enable clinicians and researchers to most effectively utilize AKI biomarkers [
8].
There are a number of limitations to our study. First, the study was not a multicenter study. Nevertheless, the relatively large sample size, with a prospective study design and a protocol with frequent uCysC measurements, provided adequate power for assessing the association between uCysC and mortality. Second, the diagnosis of neonatal AKI based on sCr and urine output is troublesome [
27,
35‐
37]. Serum Cr reflects maternal levels during the first 2–3 days of life and declines progressively with increasing postnatal age to reach a stable neonatal level by 2 weeks of life in neonates with more than 26 weeks of gestational age [
38,
39]. The total body water content and the urine output in neonates are normally greater than those in other populations, suggesting that urine output ≤ 0.5 mL/kg/h is a nonsensitive marker of neonatal AKI [
37]. Therefore, the diagnosis and staging of AKI based on neonatal AKI KDIGO may have underestimated the incidence and grade of AKI in neonates [
1,
40]. Moreover, the results from the NICU population seem to be a large driver of the overall study results, which would bias the study towards positive results. Nevertheless, the PICU analyses strengthen the fact that there truly is a signal in the results because the older children do not have this issue to the same extent but still seem to have uCysC-positive subclinical AKI that increases mortality risk. Third, the majority of critically ill children did not have baseline sCr measurement within 3 months prior to PICU admission. The incidence of AKI may be underestimated when sCr at PICU admission is used as a baseline; therefore, the lowest sCr within 2 weeks in the PICU was used as a baseline for patients with elevated sCr ≥ 1.2 mg/dL (106.1 μmol/L) at admission. Although it has not been validated in critically ill children, a previous study suggests that AKI incidence is best estimated by choosing the lowest sCr value within the first week in the ICU as a baseline sCr [
29]. Fourth, we did not measure uCysC daily as part of the study, which may underestimate the incidence and grade of AKI and limit our ability to determine the exact time at which the levels of uCysC rise prior to a sCr increase. Fifth, there is no specific cutoff value for uCysC to identify AKI and mortality in critically ill neonates and children. We set the uCysC level of 1260 ng/mg uCr, which was the optimal cutoff value for uCysC to predict mortality, as the threshold for identifying tubular injury. The discrepancy in cutoff points between this study and our previous study [
20] or between critically ill neonates and children in this study may be related to the fact that we evaluated the predictive accuracy of uCysC in a general and heterogeneous NICU and PICU population in this study. The level of urinary CysC was significantly affected by postnatal age [
41]. Sixth, the concept of uCysC-positive subclinical AKI has not been validated in different case mixes, especially in patients with higher scores of illness severity. Further multicenter prospective studies are necessary to confirm our findings and define age-related reference values of uCysC for neonates and children.