Background
Patients with acute kidney injury (AKI) have increased mortality, prolonged hospital stay and accelerated progression to chronic kidney disease (CKD) [
1]. AKI is associated with high morbidity and mortality despite advances in modern medical care [
2,
3]. Early and exact detection of potentially significant AKI is important in clinical practice and could lead to timely management [
4]. Serum creatinine has traditionally served as a surrogate of renal function, despite its limitations as a diagnostic predictor of AKI [
5]. The limitations of serum creatinine include a lack of steady-state conditions in critically ill patients, with the determinants of serum creatinine (rate of production, apparent volume of distribution and rate of elimination) being variable in the ICU setting.
Novel biomarkers can detect renal tubular injury earlier than serum creatinine in the setting of AKI [
6‐
8]. As these markers correlate with renal tubular injury or function, their concentrations in urine over time, alone or in combination, could provide important information on the progression of AKI [
9]. However, most of those patients who develop AKI experience a milder form of AKI (e.g., Kidney Disease: Improving Global Outcomes (KDIGO) stage 1) with transient shifts in serum creatinine, and do not progress to more advanced stages of AKI (KDIGO stage 2 or 3) or require acute dialysis [
10‐
12]. Whereas previous studies have attempted to predict AKI (defined as worsening of KDIGO stage) rather than immediate assessment after potential kidney insult, a panel of biomarkers could be used to predict progression of AKI to advanced AKI or mortality among patients after kidney injury (e.g., cardiovascular surgery) [
10,
13,
14]. Therefore, we validated three urinary markers extensively upregulated in renal proximal tubules in response to ischemia-reperfusion AKI: urinary hemojuvelin (uHJV) [
8], urinary kidney injury molecule-1 (uKIM-1) [
15] and urinary α-glutathione S-transferase (uα-GST) [
16], and one marker of distal tubule damage, urinary π-glutathione S-transferase (uπ-GST) [
16] (see Additional file
1). We further measured urinary neutrophil gelatinase-associated lipocalin (uNGAL), the well-studied and established inflammatory biomarker identified both in proximal and distal renal tubular damage [
17], to validate the prediction of AKI following sequential biomarker measurements after cardiovascular surgery.
Moreover, multiple AKI severity scores have also been derived to predict patient outcome [
18]. We further compared the contribution of the clinical models of Liano’s score (see Additional file
1) [
19], Cleveland Clinic acute renal failure score [
20], and Sequential Organ Failure Assessment (SOFA) score [
21], and combined these with the predictive power of the urinary biomarkers in an effort to identify and validate the best prediction model for advanced AKI and composite outcomes in patients undergoing cardiovascular surgery.
We hypothesized that some combinations of clinical scores and these site-orientated renal biomarkers would provide better diagnostic accuracy than either alone. In addition, these combinations might involve biomarkers that reflect differing aspects of the pathogenesis of advanced AKI in this population.
Methods
Study population
This study was conducted by the biomarker investigation group from the National Taiwan University Study Group on Acute Renal Failure (NSARF) (see Additional file
1), using a multicenter, prospectively constructed database of AKI [
22‐
26]. Patients undergoing cardiovascular surgery (including coronary bypass, valvular operations and aortic aneurysm repair) between August 2009 and December 2014, were enrolled prospectively from a tertiary center in northern Taiwan and two regional hospitals in central and northern Taiwan. Patients with the following conditions were excluded: those who were younger than 18 years of age, those who were diagnosed with AKI under the KDIGO definitions during index hospitalization before surgery, those who had undergone renal replacement therapy, those who had a history of nephrectomy or renal transplantation and estimated glomerular filtration rate (eGFR) < 30 mL/1.73 m
2 at the time of ICU enrollment. The enrollee was required to have a baseline serum creatinine measurement, defined as 7–180 days prior to the index hospital admission.
Ethics and consent
This study was approved by the research ethics review board of National Taiwan University Hospital (201105040RC) along with established written informed consent. This research was carried out in accordance with the approved guidelines. Written informed consent was obtained from all participants before inclusion. This study was conducted in accordance with the Declaration of Helsinki.
Clinical data collection
Medical records of study participants were prospectively reviewed to retrieve hospitalization data, including baseline demographic characteristics (Table
1), intervention procedures, and comorbidity status. Creatinine level with the accompanying eGFR (through the Modification of Diet in Renal Disease (MDRD) Study equation formula) was used for eGFR in this study. Among adults, the MDRD Study equation provides a clinically useful estimate of GFR in patients with stable kidney status [
27]; urine output was recorded at each time point after surgery as detailed in the study protocol. Postoperative inotropic agent use was quantified as per inotropic equivalents (see Additional file
1) [
23,
25,
28]. Surgery-related parameters included surgical methods, aortic clamping time and cardiopulmonary bypass time. Lengths of total admission and ICU admission were also recorded. Disease severity was evaluated as post-surgical SOFA score. The clinical classifications of Liano’s score [
19] and Cleveland Clinic Foundation Acute Renal Failure Scoring System (CCF ARF score) [
20] were also assessed to examine the risk of postoperative renal failure. Clinical and demographic characteristics of these patients were collected at each site on a case report form, which was sent to the coordinating center for entry into the NSARF database [
29‐
31].
Table 1
Summary of baseline and clinical characteristics of the study patients
Patient characteristics |
Age | 62.36 ± 13.64 | 62.32 ± 13.62 | 62.72 ± 14.23 | 0.907 |
Gender (male) | 104 (69.8%) | 95 (72.5%) | 9 (50.0%) | 0.051 |
BMI | 24.87 ± 3.73 | 24.82 ± 3.48 | 25.26 ± 5.33 | 0.735 |
Comorbidities |
Hypertension | 77 (51.7%) | 69 (52.7%) | 8 (44.4%) | 0.512 |
Diabetes mellitus | 36 (24.2%) | 32 (24.4%) | 4 (22.2%) | 0.838 |
COPD | 4 (2.7%) | 4 (3.1%) | 0 (0.0%) | 0.452 |
Liver cirrhosis | 4 (2.7%) | 4 (3.1%) | 0 (0.0%) | 0.452 |
Congestive heart failure | 14 (9.4%) | 14 (10.7%) | 0 (0.0%) | 0.145 |
Malignancy | 5 (3.4%) | 4 (3.1%) | 1 (5.6%) | 0.58 |
Laboratory data at admission |
Preoperative creatinine (mg/dL) | 1.18 ± 0.33 | 1.17 ± 0.31 | 1.26 ± 0.43 | 0.388 |
eGFR (MDRD) (mL/min/1.73 m2) | 63.27 ± 20.59 | 63.86 ± 20.29 | 58.97 ± 22.85 | 0.346 |
eGFR between 30 and 60 mL/min/1.73 m2 | 63 (42.3%) | 56 (42.7%) | 7 (38.9%) | 0.756 |
Hemoglobin (g/dL) | 13.02 ± 2.02 | 13.18 ± 1.96 | 11.83 ± 2.17 | 0.007* |
Albumin (g/dL) | 4.15 ± 0.63 | 4.26 ± 0.47 | 3.40 ± 1.03 | 0.003* |
LVEF < 55% | 35 (23.5%) | 31 (23.7%) | 4 (22.2%) | 0.892 |
LVEF < 35% | 10 (6.7%) | 9 (6.9%) | 1 (5.6%) | 0.834 |
Perioperative condition |
Inotropic equivalents | 5.83 ± 6.61 | 4.82 ± 4.45 | 13.10 ± 12.81 | 0.014* |
Presence of CPB | 93 (62.4%) | 81 (61.8%) | 12 (66.7%) | 0.691 |
CPB time (min) | 106.57 ± 98.87 | 102.23 ± 97.43 | 143.27 ± 111.96 | 0.200 |
Presence of cross clamp | 72 (48.3%) | 63 (48.1%) | 9 (50.0%) | 0.879 |
Clamp time (min) | 57.41 ± 65.11 | 55.66 ± 65.18 | 78.20 ± 63.82 | 0.295 |
Operative method |
CABG | 77 (51.7%) | 70 (53.4%) | 7 (38.9%) | 0.247 |
Valve | 64 (43.0%) | 55 (42.0%) | 9 (50.0%) | 0.519 |
Aorta | 20 (13.4%) | 16 (12.2%) | 4 (22.2%) | 0.243 |
Post-surgery |
SOFA score | 6.14 ± 2.96 | 5.84 ± 2.69 | 8.47 ± 3.92 | 0.015* |
Respiratory | 0.67 ± 0.75 | 0.66 ± 0.73 | 0.77 ± 1.01 | 0.608 |
Coagulation | 0.43 ± 0.60 | 0.40 ± 0.57 | 0.69 ± 0.85 | 0.093 |
Liver | 0.87 ± 0.65 | 0.85 ± 0.64 | 1.00 ± 0.71 | 0.432 |
Cardiovascular | 0.28 ± 0.55 | 0.24 ± 0.48 | 0.62 ± 0.96 | 0.191 |
Central nervous system | 3.08 ± 1.20 | 3.08 ± 1.21 | 3.08 ± 1.12 | 0.997 |
Renal function | 0.72 ± 0.79 | 0.58 ± 0.66 | 2.07 ± 0.64 | < 0.001* |
Cleveland score | 3.80 ± 1.62 | 3.79 ± 1.56 | 3.89 ± 2.03 | 0.802 |
Liano’s score | −0.60 ± 0.84 | −0.61 ± 0.84 | −0.47 ± 0.87 | 0.506 |
Length of admission (days) | 18.69 ± 37.62 | 17.56 ± 38.51 | 30.75 ± 24.41 | 0.247 |
Length of ICU admission (days) | 3.55 ± 3.30 | 3.43 ± 2.84 | 7.64 ± 5.35 | 0.027* |
90-day mortality | 16 (10.7%) | 6 (4.6%) | 10 (55.6%) | < 0.001* |
Sample collection
Fresh urine samples were obtained in the ICU at 0, 3, 6, 12 and 24 h after completion of surgery. Other laboratory examinations were performed as indicated clinically. Urine specimens were collected by a standardized procedure, centrifuged within 1 h and the sediments discarded. The urine samples, collected in separate polypropylene tubes containing sodium azide, were stored at − 80 °C until required. Each specimen was centrifuged (800 g at 4 °C for 5 min) and the supernatant was collected for ELISA.
Biomarker Measurements
The urinary HJV, KIM-1 and NGAL levels were measured by a human hemojuvelin ELISA kit (USCN Life Science, Inc., Wuhan, China), human urinary TIM-1/KIM-1/HAVCR Quantikine ELISA kit (R&D Systems, USA) and a human lipocalin-2/NGAL ELISA kit (R&D Systems), respectively. All of the results were expressed in nanograms per milliliter. The lower limit of detection for HJV, KIM-1 and NGAL was 0.156, 0.046 and 0.2 ng/mL, respectively. The urinary α-GST and π-GST levels were determined using Human Alpha and Pi GST EIA Test Kits (EKF Diagnostics). The results were expressed in micrograms per liter. The inter-assay and intra-assay coefficient of variation for α-GST was 6.3% and 2.7%, respectively, and for π-GST, it was 8.6% and 3.1%, respectively. The lower limit of detection for α-GST and for π-GST was 0.3 μg/L. Assays were completed as described by the manufacturer’s protocol, and each measurement was performed in duplicate. Urinary creatinine levels were measured using the Jaffe assay, with standardization to isotope dilution mass spectrometry (IDMS)-traceable reference. Technicians performing the biomarker measurements were blinded to each patient’s clinical information. All biomarkers were measured from frozen aliquots that did not undergo any additional freeze–thaw cycles.
Outcome definitions
The clinical primary endpoint was defined as the development of advanced AKI (stage 2 or 3, specified by the KDIGO criteria [
32]), with both urine and creatinine criteria applied. The secondary outcomes were 90-day mortality after hospital discharge and composite outcomes defined as 90-day mortality after hospital discharge or advanced AKI, whichever occurred earlier.
Statistical analysis
All analyses were performed using SPSS software, version 20 (IBM, Armonk, NY, USA), R software, version 3.2.2 (Free Software Foundation, Inc., Boston, MA, USA), and MedCalc Statistical Software, version 15.11.3 (MedCalc Software bvba, Ostend, Belgium;
https://www.medcalc.org; 2015). The two-sample
t test or Mann–Whitney rank sum test was used as appropriate to compare continuous variables; for categorical variables, the chi-square (χ2) or Fisher’s exact test was applied. Friedman two-way analysis of variance (ANOVA) was used to assess the overall difference in HJV, KIM-1, NGAL, α-GST and π-GST between the “no AKI or stage 1 AKI” and the “stage 2 or 3 AKI” groups at 0, 3, 6, 12 and 24 h after cardiovascular surgery [
33]. We normalized biomarker levels with urine creatinine concentrations and analyzed them at each time point [
34]. We fitted logistic regression models to examine the association between each marker (urinary NGAL, HJV, KIM-1, α-GST and π-GST) and advanced AKI, and generated an area under the receiver-operating characteristic (ROC) curve to assess the predictive accuracy of each marker. Power analysis was conducted based on the prior knowledge that the ratio of cases to controls was 4:1. We required 105 patients (21 with severe AKI and 84 without severe AKI) to achieve power of 0.8, with type I error of 0.05. This was based on the preliminary knowledge that the discriminatory power of urinary HJV to predict severe AKI was 0.7 [
35]. We used a generalized additive model (GAM) (with spline) incorporating the subject-specific (longitudinal) random effects and adjusted for other clinical parameters to predict the outcomes [
36,
37]. Simple and multiple generalized additive models (GAMs) were fitted to detect nonlinear effects of continuous covariates and identify appropriate cutoff point(s) for discretizing continuous covariates, if necessary, during the stepwise variable selection procedure. We defined the optimal cutoff value as odd equals to zero [
38].
To assess the additive prediction ability of each biomarker compared with traditional clinical predictors, we identified clinical risk prediction models (including Liano’s score, CCF ARF score and SOFA score) that were based on area under the curve (AUC) analysis and then added each biomarker individually to this clinical model. We estimated the risk score function using a logistic regression model including clinical and biomarker variables as covariates. The Hosmer-Lemeshow logistic regression model test for goodness of fit was used to assess the calibration between the current model and the expected model. Finally, we calculated the net reclassification improvement (NRI) and integrated discrimination improvement (IDI) to estimate overall improvement in reclassification with urinary biomarkers adding to clinical variables [
39]. We reclassified the patients who developed advanced AKI using 0–12%, 12–30% and > 30% for the risk categories. A
p value <0.05 was considered significant.
Hierarchical clustering analysis was performed using the cluster program and the results were visualized using the Treeview program [
40]. The biomarkers were arranged in such a way that the most similar expression profiles were placed next to each other. In the color scheme, strong positive staining is indicated as a red cube, weak positive staining as a black cube and negative staining as a green cube. Absence of staining data is indicated with a grey cube.
Discussion
In this prospective multicenter observational study, we have shown that the combined biomarker panel of creatinine-normalized uHJV and uKIM-1 at 3 h after cardiovascular surgery had the strongest predictive ability for the development of advanced AKI and composite outcomes. This predictive ability was increased further after combination with Liano’s score. The combination of clinical models and biologic biomarkers could better anticipate the future development of advanced AKI in the early postoperative period, suggesting future clinical outcomes.
The combination of uHJV and uKIM-1, both markers of structural kidney injury, provided added value to the traditional clinical AKI score in forecasting the risk of advanced AKI. Consequently, combination approaches could account for different time courses of biomarker release or could reflect different pathophysiological mechanisms [
41]. HJV, a glycophosphatidylinositol (GPI)-linked membrane protein, is highly expressed in liver and skeletal muscles. The molecular weight of HJV is 42 kDa for the soluble form (sHJV) [
42], and can be passed through glomerular filtration and reabsorbed by the renal tubules [
8]. Increased iron content in the kidney and urine is observed in human and animal models of AKI [
43], and increased iron load can induce renal tubular cell injury [
44]. There is evidence that the increased expression of the hemojuvelin-hepcidin-ferroportin pathway is an intrinsic response to iron overload conditions during AKI. Therefore, uHJV has the potential to be an early AKI biomarker in response to iron homeostasis during AKI, which may explain the temporal relationship between uHJV and its predictive capacity. In this study, we showed that the novel biomarker of HJV can predict AKI after cardiac surgery. This finding reinforced our previous finding, in which high urinary HJV levels were observed in patients after cardiac surgery and rhabdomyolysis-related AKI. Although HJV is highly expressed in liver, the post-surgery liver function was not significantly different between the no AKI/stage 1 AKI and the stage 2/3 AKI groups This result suggested higher HJV concentration in patients with advanced AKI was mainly from kidney injury. According to our animal model, the elevation of urinary HJV in acute tubular necrosis should be originally from filtration after renal tubular destruction rather than the liver (see Additional file
1) [
8].
Kidney injury molecule-1 (KIM-1), a 38.7-kDa transmembrane protein with immunoglobulin-like and mucin domains in its ectodomain, is a novel biomarker for renal proximal tubule injury and could play a role in tubulo-interstitial damage [
45]. In addition to being a biomarker of AKI, KIM-1 could also play a role in renal recovery and tubular regeneration after AKI [
46]. However, because increased urinary KIM-1 concentration can indicate either injury or the repair response to injury, KIM-1 by itself may not be a suitable marker for early prediction of AKI whereas the combination of KIM-1 with other injury markers may be highly useful. In support of this hypothesis, our result was consistent with a study investigating biomarker combinations for prediction of AKI following cardiac surgery [
14], whereby KIM-1 in combination with another biomarker had the best predictive value for severe AKI.
We further showed that the combination of a panel of biomarkers and Liano’s AKI score may be beneficial in providing additional information with respect to the risk of AKI severity and other adverse clinical outcomes. Due to the heterogeneous causes of AKI and underlying comorbidities, the use of biomarkers and clinical disease severity scores to predict renal injury seems reasonable and actually did have better accuracy [
17]. For increased accuracy and added predictive ability beyond the AUC value, the biomarker and clinical score combination model was supported by net reclassification improvement (NRI) and integrated discrimination improvement (IDI) values. The NRI and IDI represent modern, more sensitive statistical methods to reclassify model improvement from the use of a biomarker with an existing clinical model. Urinary HJV and uKIM-1 had the unique ability to improve the NRI for both events and non-events and augmented the AUC value compared with use of a clinical model alone and thus, uHJV and uKIM-1 may serve as an ideal biomarker panel to help early detection of advanced AKI.
Biomarkers predicting advanced AKI will allow future clinical trials in AKI intervention to be targeted to those patients at highest risk of disease progression, which would be far more efficient in terms of costs and resources [
47]. The majority of patients (28/46, 60.9% in this study) who developed AKI after cardiovascular surgery had stage 1 AKI and this spontaneously resolved. The ability to distinguish those patients at highest risk of severe outcomes will be crucial as nephrologists and intensivists conduct more clinical trials in the treatment of AKI [
48]. Given the clinical uncertainty and high complexity of medical decision-making around this issue (e.g., timing and duration of renal replacement therapy, placement of permanent access and follow up), any method that provides a modest improvement in clinical prediction could have clinical importance. We endeavor to find the biomarkers and clinical severity score to determine which patients will develop advanced AKI, and help to improve clinical decision-making - such as earlier or delayed dialysis [
4,
49], guiding decision-making, patient counseling and facilitating enrollee selection in interventional trials of AKI.
There were several limitations to our study. First, because of the limited sample size, the statistical significance may be improved after recruiting more patient samples. Second, only those patients undergoing cardiovascular surgery were included in this study and consequently other serious causes of AKI, such as sepsis or drug-induced AKI, are not represented. The results of our study could add to the value of AKI prediction after generalized cardiovascular surgery, and further large prospective studies are necessary to validate our results. Third, about 3% of urinary biomarker data were lost due to anuria at 12 and 24 h post-surgery. In addition, we did not obtain urine samples during surgery. Tubular injury may occur 30 min after commencing cardiopulmonary bypass in cardiac surgery with a peak immediately following completion of cardiopulmonary bypass [
50]. However, our results clearly showed these markers had their peak level at 3 h after surgery and could predict advanced AKI. For clinical practice, urine collection after surgery was more feasible than collection during surgery. Finally, other urinary biomarkers, such as IL-18 and L-fatty acid binding protein, which also have promising results for predicting AKI, were not examined [
51,
52].
Our study also has several strengths. First, this was a large multicenter prospective study and included a homogenous (post-cardiovascular surgery) patient population. Second, our study contained five biomarkers (including markers of proximal and distal tubular injury) and tested urine samples obtained in the ICU at 0, 3, 6, 12 and 24 h after completion of surgery, which allowed exploration of the dynamic changes in their urinary concentrations and provided more comprehensive and credible results than previous studies [
12,
53,
54]. Finally, this is, to our best knowledge, the first study using panels of urinary biomarkers in combination with clinical risk scores to focus on prediction of advanced AKI after cardiovascular surgery. Urinary biomarkers can lead to more accurate diagnosis of AKI and can further be used for recruitment of more homogenous patient populations when implementing a clinical trial [
55].
Acknowledgements
This study was supported by Taiwan National Science Council (101-2314-B-002-085-MY3, 102-2314-B-002-140-MY2, 104-2314-B-002-125-MY3) and NTUH 106-FTN20/106-P02/UN106-014 106-S3582/105-P05/VN105-04/105-S3061/VN104-07/104-S2718. This work was also supported by the Ministry of Science and Technology (MOST) of the Republic of China (Taiwan) (grant number, MOST 106-2321-B-182-002). We also express our sincere gratitude to all the staff of the Taiwan Clinical Trial Consortium, TCTC.